The Routledge International Handbook of Creative Cognition (Routledge International Handbooks) [1 ed.] 0367443783, 9780367443788

Citation preview

THE ROUTLEDGE INTERNATIONAL HANDBOOK OF CREATIVE COGNITION

The Routledge International Handbook of Creative Cognition is an authoritative reference work that offers a ­well-​­balanced overview of current scholarship across the full breadth of the rapidly expanding field of creative cognition. It contains 43 chapters written by ­world-​­leading researchers, covering foundational issues and concepts as well as ­state-­​­­of-­​­­the-​­a rt research developments. The handbook draws extensively on contemporary work exploring the cognitive representations and processes associated with creativity, whether studied in the laboratory or as it arises in ­real-​­world practice in domains such as education, art, science, entrepreneurship, design, and technological innovation. Chapters also examine the sociocognitive and cultural aspects of creativity in teams and organisations, while additionally capturing the latest research on the cognitive neuroscience of creativity. Providing a compelling synopsis of emerging trends and debates in the field of creative cognition and positioning these in relation to established findings and theories, this text provides a clear sense of the way in which new research is challenging traditional viewpoints. It is an essential reading for researchers in the field of creative cognition as well as advanced students wishing to learn more about the latest developments in this important and rapidly growing area of enquiry. Linden J. Ball is Professor of Cognitive Psychology at the University of Central Lancashire, Preston, UK. He is interested in the role of m ­ eta-​­cognitive monitoring and control in thinking, reasoning, ­problem-​­solving, and creativity. He is the ­Editor-­​­­in-​­Chief of the Journal of Cognitive Psychology and Associate Editor of Thinking & Reasoning. He is also the Editor of the Current Issues in Thinking & Reasoning book series (­Routledge) and ­Co-​­Editor of the Routledge International Handbook of Thinking & Reasoning (­2018). Frédéric ­Vallée-​­Tourangeau is Professor of Psychology at Kingston University, London, UK. Recent projects on creative ­problem-​­solving have drawn inspiration from William James and Bruno Latour. He ­co-​­edited (­w ith Stephen Cowley) Cognition Beyond the Brain (­Second edition, 2017), and he also edited Insight: On the Origins of New Ideas ( ­Routledge, 2018). He is the author of Systemic Creative Cognition: Bruno Latour for Creativity Researchers ( ­Routledge, 2023).

THE ROUTLEDGE INTERNATIONAL HANDBOOK OF CREATIVE COGNITION

Edited by Linden J. Ball and Frédéric ­Vallée-​­Tourangeau

Designed cover image: © Getty Images First published 2024 by Routledge 4 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 605 Third Avenue, New York, NY 10158 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2024 selection and editorial matter, Linden J. Ball and Frédéric ­Vallée-​ ­Tourangeau; individual chapters, the contributors The right of Linden J. Ball and Frédéric ­Vallée-​­Tourangeau to be identified as the authors of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks and are used only for identification and explanation without intent to infringe. British Library ­Cataloguing-­​­­in-​­Publication Data A catalogue record for this book is available from the British Library Library of Congress ­Cataloging-­​­­in-​­Publication Data Names: Ball, Linden J., 1963– editor. | Vallee-Tourangeau, Frederic, editor. Title: The Routledge international handbook of creative cognition / edited by Linden J Ball and Frédéric Vallée-Tourangeau. Description: Abingdon, Oxon; New York, NY: Routledge, 2024. | Series: Routledge international handbooks | Includes bibliographical references and index. | Identifiers: LCCN 2023006469 (print) | LCCN 2023006470 (ebook) | ISBN 9780367443788 (hardback) | ISBN 9780367503468 (paperback) | ISBN 9781003009351 (ebook) Subjects: LCSH: Creative ability. | Creation (Literary, artistic, etc.) | Cognition. Classification: LCC BF408.R723 2024 (print) | LCC BF408 (ebook) | DDC 153.3/5—dc23/eng/20230224 LC record available at https://lccn.loc.gov/2023006469 LC ebook record available at https://lccn.loc.gov/2023006470 ISBN: ­978-­​­­0 -­​­­367-­​­­4 4378-​­8 (­hbk) ISBN: ­978-­​­­0 -­​­­367-­​­­50346-​­8 (­pbk) ISBN: ­978-­​­­1-­​­­0 03-­​­­0 0935-​­1 (­ebk) DOI: 10.4324/­9781003009351 Typeset in Bembo by codeMantra

To Robert Weisberg, who taught us to distrust hagiographies of creativity, and to Ken Gilhooly, a generous and inspirational mentor.

CONTENTS

List of Contributors xii Preface xxiv PA RT I

Reflections on the Fundamental Contents and Mechanisms of Creative Cognition 1 Divergent Thinking as Creative Cognition Mark A. Runco 2 Measuring Creativity with the Consensual Assessment Technique (­CAT) Karl K. Jeffries

1 3 17

3 Constraints and Creativity: Classifying, Balancing, and Managing Constraints 32 Balder Onarheim and Dagný Valgeirsdóttir 4 The Role of Serendipity in Creative Cognition Wendy Ross and Selene Arfini

46

5 The Two Faces of Curiosity in Creative Cognition: Curiosity1, Curiosity2 (­and Their Interaction) Janet Metcalfe and William James Jacobs

65

6 Thinking Wide and Narrow: A ‘­Cultural Creative Cognition’ Approach to Possibility Thinking Vlad P. Gl ăveanu

80

vii

Contents

7 Analogy and the Transfer of Creative Insights Thomas C. Ormerod

95

8 Creative Cognition: From Ideation to Innovation Nathaniel Barr, Lucas Klein, Michael J. McNamara and Kelly Peters

109

9 Insight in the Kinenoetic Field Frédéric ­Vallée-​­Tourangeau

127

PA RT I I

Reflections on the Nature of Creative Cognition as Revealed through Traditional Methodologies 10 Idea Generation and Associative Memory Richard W. Hass 11 Creatively Searching Through Semantic Memory Structure: A Short Integrative Review Yoed N. Kenett

141 143

160

12 Mental Imagery and Creative Cognition David G. Pearson

180

13 Incubation Ken Gilhooly

199

14 Of Night and Light and the ­Half-​­Light: The Role of Multidimensions of Emotion and Tolerance of Uncertainty in Creative Flow Genevieve M. Cseh

215

15 ­Problem-​­Solving, Collaboration, and Creativity Jennifer Wiley and Olga Goldenberg

233

16 Metaphoric Creativity as Embodied Performance in Social Interaction Thomas Wiben Jensen

254

17 Analyzing Changing Patterns of External Reference Use from Informal Lab Group Presentations to Formal Colloquia Christian D. Schunn, Lelyn D. Saner and Susannah B. F. Paletz

viii

271

Contents

18 Measuring Judgment and ­Decision-​­Making in Developmental Samples: Assessment of the Generation of Cognitively Sophisticated Responses and Implications for the Study of Creative Cognition Jala Rizeq and Maggie E. Toplak 19 The Phenomenology of Insight: The Aha! Experience Amory H. Danek

284 308

20 Reconcilable Differences: Working Memory Capacity Both Supports and Hinders Insight Marci S. DeCaro and Charles A. Van Stockum, Jr.

332

21 Comparing Theoretical and Computational Models of Insight: Investigating Cognitive and Phenomenological Perspectives Margaret E. Webb

351

22 Collaborative ­Meta-​­Reasoning in Creative Contexts: Advancing an Understanding of Collaborative Monitoring and Control in Creative Teams 372 Beth H. Richardson, Linden J. Ball, Bo T. Christensen and John E. Marsh PA RT I I I

Reflections on the Nature of Creative Cognition as Revealed through Cognitive Neuroscience Approaches

397

23 An Archaeological Perspective on Creative Cognition Frederick L. Coolidge

399

24 The Role of Semantic Versus Episodic Memory in Creative Cognition Halima Ahmed, Kata ­Pauly-​­Takacs and Anna Abraham

415

25 Network Neuroscience of D ­ omain-​­General and ­Domain-​­Specific Creativity 433 Roger E. Beaty, Hannah M. Merseal and Daniel C. Zeitlen 26 A Closer Look at Transitions between the Generative and Evaluative Phases of Creative Thought Andre Zamani, Caitlin Mills, Manesh Girn and Kalina Christoff 27 Markers of Insight Carola Salvi

453 475

ix

Contents

28 A Cognitive Neuroscience Perspective on Insight as a Memory Process: Searching for the Solution Maxi Becker, Roberto Cabeza and Jasmin M. Kizilirmak

491

29 A Cognitive Neuroscience Perspective on Insight as a Memory Process: Encoding the Solution Jasmin M. Kizilirmak and Maxi Becker

511

PA RT I V

Reflections on Creative Cognition from Pedagogical, Organisational, Archaeological and ­Post-​­Phenomenological Perspectives

533

30 Creativity in Education Mary A. Pei and Jonathan A. Plucker

535

31 Creative Learning: A Pedagogical Perspective Ronald A. Beghetto and Ananí M. Vasquez

551

32 Tool Use and Creativity Chris Baber

570

33 Art Through Material Engagement…and Vice Versa Paul Louis March and Lambros Malafouris

585

34 Creativity as a Discursive Construct Darryl Hocking

605

35 Collaborative Creativity: ­Information-​­Driven Coordination Dynamics and Prediction in Movement and Musical Improvisation Travis J. Wiltshire and Merle T. Fairhurst 36 Common Creativity Karenleigh A. Overmann

624 646

PA RT V

Reflections on Creative Cognition in Domains Involving Creativity and Innovation 663 37 Team Cognition and Team Creativity Roni ­Reiter-​­Palmon and Payge Japp

665

38 Creative Cognition in Engineering and Technology David H. Cropley

684

x

Contents

39 Creative Cognition and Entrepreneurship Bo T. Christensen

698

40 Creative Cognition in Advertising Wangbing Shen, Linden J. Ball and Beth H. Richardson

709

41 The Creative Generation and Appreciation of Artistic Artifacts in the Visual Domain Oshin Vartanian

728

42 Individual Innovation and Creativity: An Interdisciplinary Mapping of Creative Cognition Gaëlle ­Vallée-​­Tourangeau, Evy Sakellariou and Fanni Szigetvari

741

PA RT V I

Reflections on the Paradoxical Misalignment Between Findings that Derive from In Vivo Versus In Vitro Research on Creative Cognition 43 A Quandary in the Study of Creativity: Conflicting Findings from Case Studies versus the Laboratory Robert W. Weisberg

763 765

Index 795

xi

CONTRIBUTORS

Anna Abraham is the E. Paul Torrance Professor and Director of the Torrance Center for Creativity and Talent Development at the University of Georgia (­USA). She investigates the psychological and neurophysiological mechanisms underlying creativity and other aspects of the human imagination, including the ­reality-​­fiction distinction, mental time travel, social and ­self-​­referential cognition, and mental state reasoning. She is the author of the 2018 book, The Neuroscience of Creativity (­Cambridge University Press), and the editor of the multidisciplinary volume, The Cambridge Handbook of the Imagination (­2020). Halima Ahmed, PhD, is a Lecturer in Psychology at Leeds Beckett University, UK. Her research focusses on the cognitive underpinnings of creativity. More broadly, she is interested in ­age-​­related cognitive changes, ­long-​­term memory processes, and creativity. Selene Arfini is a Research Fellow and Professor of Philosophy of Science at the Department of H Section of the University of Pavia. Her research aims at ­ umanities – Philosophy ​­ founding a cognitively oriented epistemology of ignorance, and to this goal, she has recently ­co-​­edited Embodied, Extended, Ignorant Minds: New Studies on the Nature of ­Not-​­Knowing (­Springer 2022), a Special Issue for Synthese entitled Knowing the Unknown: Philosophical Perspectives on Ignorance (­2021), and authored Ignorant Cognition. A Philosophical Investigation of the Cognitive Features of ­Not-​­Knowing (­Springer, 2019). Chris Baber is a Chair of Pervasive and Ubiquitous Computing in the School of Computer Science at the University of Birmingham, UK. His research concerns human interaction with technology, specifically in terms of how humans form teams with intelligent technology and in terms of ­sensor-​­based ­human-​­technology interaction. He has published over 100 papers in international journals as well as over 400 conference contributions and half a dozen books. His research has been funded by the UK Ministry of Defence, RCUK, the European Union, and various industries. Linden J. Ball  is a Professor of Cognitive Psychology at the University of Central Lan­ eta-​­cognitive monitoring and control cashire, Preston, UK. He is interested in the role of m in thinking, reasoning, ­problem-​­solving, and creativity. Linden is the ­Editor-­​­­in-​­Chief of the xii

Contributors

Journal of Cognitive Psychology (­since 2017) and Associate Editor of Thinking & Reasoning (­since 2011). He is also the Editor (­since 2011) of the Current Issues in Thinking & Reasoning book series (­Routledge) and the ­Co-​­Editor of the Routledge International Handbook of Thinking & Reasoning (­2018). Nathaniel Barr, PhD, is a Professor of Creativity and Creative Thinking at Sheridan College and a Senior Innovation Advisor at BEworks. He has w ­ ide-​­ranging interests in the cognitive and behavioural sciences, including creativity and innovation, the intersection of thinking and technology, misinformation and belief, and applied behavioural science. Roger E. Beaty is the Dr. Frances Keesler Graham Early Career Professor and Director of the Cognitive Neuroscience of Creativity Lab at Pennsylvania State University. He completed his doctoral degree at UNC at Greensboro and his postdoctoral training at Harvard University. He received the Berlyne Award for his early career contributions to creativity research from Division 10 of the American Psychological Association. His research on creativity neuroscience and measurement has been funded by grants from the US National Science Foundation and the John Templeton Foundation. Dr. Beaty is an Associate Editor for the Creativity Research Journal, and he serves on the Executive Committee of the Society for the Neuroscience of Creativity. Maxi Becker is currently a Postdoctoral Researcher in Cognitive Neuroscience at Humboldt University of Berlin. Her main research focus is neural mechanisms of insight. Working with Roberto Cabeza, she investigates the diverse interactions between insight and memory. Dr. Becker holds a Master’s in Psychology as well as in History, Philosophy, and Sociology of Science, and a PhD in Cognitive Neuroscience. Ronald A. Beghetto is the Pinnacle West Presidential Chair and Professor in the Mary Lou Fulton Teachers College at Arizona State University. His research focusses on creative thought and action, uncertainty, and the possible in educational settings. He is a fellow of the American Psychological Association (­Division 10) and the International Society for the Study of Creativity and Innovation. Roberto Cabeza  is a Professor of Psychology and Neuroscience and a Member of the Center for Cognitive Neuroscience at Duke University, and he is a Professor of Cognitive Neuroscience of Memory and Aging at Humboldt ­University-​­Berlin. He investigates the neural mechanisms of memory in young and older adults using behavioural, brain imaging, and brain stimulation techniques. Dr. Cabeza graduated from the University of Buenos Aires, obtained a PhD at the University of Tsukuba ( ­Japan), and had postdoctoral training at the Rotman Research Institute (­University of Toronto). He is a member of the Governing Board of the Cognitive Neuroscience Society and was a member of the Advisory Board of the Max Planck Institute for Human Development, Berlin. Dr. Cabeza received the Young Investigator Award from the Cognitive Neuroscience Society and is an elected Fellow of the Society of Experimental Psychologists. Bo T. Christensen is a Professor of Design Research at Copenhagen Business School. A cognitive psychologist by training, his research centres on supporting and developing creativity, design, and entrepreneurship in educational settings as well as in professional practice. He has pioneered the combined use of video ethnography and protocol analyses. He xiii

Contributors

has twice been awarded the Design Studies Award by the Design Research Society for the best paper published in the journal Design Studies – the ​­ top international scientific journal focussing on design. Kalina Christoff is a Professor of Psychology at the University of British Columbia. Her work focusses on the cognitive neuroscience of human thought, from spontaneous thought phenomena such as ­m ind-​­wandering, daydreaming, and creativity, to ­goal-​­directed thought, including deliberate reasoning and ­problem-​­solving. Her work also examines the neurocognitive mechanisms of introspection, ­meta-​­cognition, meditation, and different forms of ­self-​­experience and ­self-​­regulation. Frederick L. Coolidge  is a Professor at the University of Colorado, Colorado Springs. His research interests include cognitive archaeology, brain evolution, and personality  & neuropsychological assessment. He was appointed as a Senior Visiting Scholar at Oxford University (­Keble College), in March 2015, and a Guest Professor at the Indian Institute of Technology Gandhinagar, ­2020–​­2024. David H. Cropley, PhD, is the University of South Australia’s Professor of Engineering Innovation. His research spans creativity in education, organisational innovation capacity, and the nexus of creative p­ roblem-​­solving and engineering. Dr Cropley is the author/­­co-​ ­author of ten books, including Core Capabilities for Industry 4.­0  – ​­Foundations of the ­Cyber-​ ­Psychology of Engineering (­W bv Media, 2021) and Femina Problematis ­Solvendis – ­​­­Problem-​­Solving Woman: A History of the Creativity of Women (­Springer, 2020). He was also a scientific consultant and ­on-​­screen expert for the Australian ABC TV Documentaries Redesign My Brain (­2013), Life at 9 (­2014), and Redesign My Brain, Series 2 (­2015). Genevieve M. Cseh  is a Senior Lecturer at Buckinghamshire New University (­BNU) in High Wycombe, UK. She is the current Course ­Co-​­Leader of BNU’s MSc in Applied Positive Psychology (­M APP) programme. Her main areas of research interest include the intersection between psychology and the arts, peak states (­especially “­flow”), and how people emotionally grapple with and handle the inherent paradoxes, ambiguity, uncertainty, and dialectics of the creative process. Amory H. Danek, PhD, is a Postdoctoral Researcher and Lecturer at the University of Heidelberg, Germany, with a research focus on human p­ roblem-​­solving, creativity, memory, and emotions. In her work, she addresses questions pertaining to the insight memory advantage, the phenomenology of Aha! Experiences, and false insights. She is Associate Editor for Thinking & Reasoning, a member of the Editorial Board of the Journal of Cognitive Psychology, and has published numerous papers and book chapters on insight ­problem-​­solving, together with talks and other conference contributions. Marci S. DeCaro is an Associate Professor at the University of Louisville where she directs the Learning and Performance Lab. Dr. DeCaro conducts a research on memory and ­problem-​­solving in both laboratory and educational settings. Merle T. Fairhurst is a Professor for Social Affective Touch at the TU Dresden, working within the Cluster for the Tactile Internet with Human in the Loop. She is a cognitive neuroscientist with strong interdisciplinary ties that facilitate crosstalk with philosophers and xiv

Contributors

engineers. She studies the interaction between sensory signals that allow us to make sense of the world around us and to successfully interact with others. Her projects range from trying to understand what makes touch special to identifying factors that make interacting in a group different from interacting in pairs. As a classical singer, she is passionate about the special cases of sensory perception in music and art. And, as a mother of five, she strongly believes in promoting women in academia. Ken Gilhooly is an Emeritus Professor of Psychology at the University of Hertfordshire. His research has concerned higher mental processes such as concept learning, reasoning, ­decision-​­making, expertise  & ageing effects, p­ roblem-​­solving, creativity, and incubation. This work has attracted regular support from UK Research Councils, the European Union, and major charities and has led to numerous publications in the form of journal articles, book chapters, textbooks, and monographs. Currently, he is focussing on attempting to clarify basic concepts in the areas of p­ roblem-​­solving and creativity. Manesh Girn is a PhD student at McGill University and will be joining the University of California San Francisco as a Postdoctoral Researcher in August 2023. His primary research focusses are on the psychological and ­large-​­scale brain effects of psychedelic drugs in humans as measured with a combination of functional neuroimaging and behavioural paradigms. Vlad P. Glăveanu,  PhD, is a Full Professor of Psychology in the School of Psychology at Dublin City University and Professor II at the Centre for the Science of Learning and Technology, University of Bergen. He specializes in the study of creativity, culture, imagination, wonder, social activism, and human possibility. His latest books include The Possible: A Sociocultural Theory (­Oxford University Press, 2020) and Wonder: The Extraordinary Power of an Ordinary Experience (­Bloomsbury, 2020). He is a president of the Possibility Studies Network and editor of the Palgrave Encyclopedia of the Possible (­Palgrave, 2023) and the Sage journal Possibility Studies and Society. Olga Goldenberg is an Associate Professor at Columbia College Chicago, teaching coursework in Psychology. Her scholarship focusses on understanding individual and small group processes and performance in the domain of creativity and innovation. She investigates how cognitive, emotional, motivational, and social factors influence the creative process and outcomes. In her current research project, Olga explores the relationship between Social and Emotional Learning and creativity among college students. She has published her work in scholarly journals such as Small Group Research and Discourse Processes. She holds a PhD in Social Psychology from the University of Illinois at Chicago. Richard W. Hass  is an Associate Professor and a Program Director of the Population Health Science PhD and Health Data Science Master’s programs at Thomas Jefferson University. In 2019, he earned the APA Division 10 Berlyne Award in recognition of his contributions to creativity and cognition research. His current work focusses on psychometric, cognitive, and data science approaches to population health. Darryl Hocking is a Senior Lecturer at Auckland University of Technology. His primary research areas are discourse, genre, metaphor, and corpus analysis, with a particular focus on the interactional genres and communicative practices in art and design settings and how these impact creative activity. On this subject, he has authored the books Communicating xv

Contributors

Creativity: The Discursive Facilitation of Creative Activity in Arts (­Palgrave Macmillan) and The Impact of Everyday Language Change on the Practices of Visual Artists (­Cambridge University Press). William James Jacobs is a Professor of Psychology in the College of Science at the University of Arizona, where he is also a Fellow in Sports Medicine and Director of the Anxiety Research Group. His broad research interests relate to learning rule governance, context (­spatial and social), cognitive mapping, executive function, stress, fear, anxiety, trauma, anxiety disorders, clinical neuropsychology, cognitive neuroscience, and evolutionary psychology. Payge Japp is an MA/­PhD student in the Industrial Organizational Psychology problem at the University of Nebraska at Omaha (­U NO). She earned a BS in Psychology in 2019 from UNO. Payge’s research interests include individual and team creativity, malevolent creativity, and creative ­problem-​­solving. Karl K. Jeffries is a creativity coach who specializes in supporting students and professional designers to attain consistent peak performance, stay fresh, and avoid creativity fatigue. Before his move to creativity coaching, he was a Senior Lecturer in Design at the University of Central Lancashire and Course Leader for the MA Creative Thinking (­Distance Learning) until 2019. His previous research has focussed on the relationship between competencies and creativity within various domains, such as the ­audio-​­visual industry, games design, and graphic design. His current research is focussed on the Consensual Assessment Technique (­CAT) and its application to design creativity. Thomas Wiben Jensen is an Associate Professor at the University of Southern Denmark. His research interests include cognitive approaches to metaphor and metonymy, metaphor in therapy, multimodal metaphor, embodied interaction, and the relationship between language and emotion. He has published two monographs in Danish on cognition, emotion, and constructivism and has published a number of research articles in various international journals such as Frontiers in Psychology, Cognitive Semiotics, Metaphor and Symbol, and Psychology of Language and Communication. Yoed N. Kenett is an Assistant Professor at the Faculty of Industrial Engineering and Management at the ­Technion – ​­Israel Institute of Technology. Yoed studies the complexity of ­h igh-​­level cognition, such as creativity, knowledge, and associative thinking, at the cognitive and neural levels in typical and clinical populations. Jasmin M. Kizilirmak is currently working as a Lecturer at the Institute for Psychology, University of Hildesheim, Germany, and as a senior researcher at the German Centre for Higher Education Research and Science Studies, Hannover, Germany. Her research focus is on ­long-​­term memory and learning via insightful ­problem-​­solving as well as on scientific career development. Lucas Klein, MSc, is a PhD candidate at McMaster University, where he studies the cognitive neuroscience of music performance. Lucas developed an interest in improvisation, flow states, and group dynamics from experience playing music in bands and team sports. His research focusses on selective auditory attention in group musical improvisation. xvi

Contributors

Lambros Malafouris is a Professor of Cognitive and Anthropological Archaeology at the Institute of Archaeology and Tutorial Fellow in Archaeology and Anthropology at Hertford College, University of Oxford. He is the author of How Things Shape the Mind: A Theory of Material Engagement (­MIT Press, 2013). He is also a principal investigator of HANDMADE: Understanding Creative Gesture in Pottery Making funded through a European Research Council Consolidator Grant (­No. 771997; European Union Horizon, 2020 programme). Paul Louis March is an artist who works mainly with clay. Before that, he was a chartered clinical psychologist specialising in brain injury rehabilitation. Alongside art, he is currently completing his DPhil in Archaeology (­Keble College, Oxford). Paul’s research combines the theory of Material Engagement and the practice of sculpting in clay into a procedure for exploring the process of creative ideation as a materially enacted phenomenon. As an artist, Paul exhibits regularly in Switzerland and France. John E. Marsh is a Reader in Cognitive Psychology at the University of Central Lancashire, Preston, UK. He has extensive expertise in relation to the effects of distraction and attention on cognitive task performance, such as with problems that involve creative thinking. John has received research funding from the Swedish Research Council, the British Academy, and the Leverhulme Trust. He serves as Chair of Team 4 (“­Effects of Noise on Performance and Behavior”) for the International Commission of the Biological Effects of Noise (­ICBEN). He is also an Associate Editor for the Journal of Cognitive Psychology (­since 2017). Michael J. McNamara is a Professor of Creativity at Sheridan College and Project D ­ irector of the Community Ideas Factory, an a­ cademic-​­community collaboration that seeks to foster social innovation by connecting the creativity resources of the College with the needs and efforts of the ­non-​­profit sector. As a Political Scientist, his interests include creativity and collaborative ­decision-​­making, creativity and community development, and creativity and innovation in the 21st century. Hannah M. Merseal is a PhD student in Psychology at Pennsylvania State University. Her research spans the neurocognitive mechanisms underlying music improvisation, equitable creativity assessment, and the use of l­arge-​­scale creativity studies to inform policy decisions in r­eal-​­world settings. She has received the Sonophilia Foundation’s 2021 Outstanding Young Scientists in Creativity Research award and APA Division 10’s 2021 Community Presentation Award. She also served as guest associate editor of Translational Issues in Psychological Science’s special issue on creativity research and is on the student editorial review boards for Creativity Research Journal and Psychology of Aesthetics, Creativity, and the Arts. Janet Metcalfe  is a Professor of Psychology at Columbia University in the city of New York. Her general areas of research relate to ­meta-​­cognition, the evolution of ­self-​­reflective consciousness, study time allocation, and judgements of learning. Her current research focusses particularly on how people know what they know, that is, their m ­ eta-​­cognitive abilities, and whether they use this evolutionarily unique ability efficaciously, such as for effective ­self-​­control. Caitlin Mills is an Assistant Professor at the University of Minnesota. Her primary research interests are at the intersection of cognitive psychology, computer science, and education. xvii

Contributors

She is particularly interested in m ­ ind-​­wandering, boredom, and creativity: their neural correlates, relationships to affect, and relationships with learning. Balder Onarheim is the CEO of PlatoScience Neurostimulation and a former Associate Professor at the Technical University of Denmark. Balder’s work focusses on cognitive training and rehabilitation and the use of neuromodulation techniques for supporting clinicians within mental healthcare. Thomas C. Ormerod is a Professor of Psychology at the University of Sussex, UK. He is a cognitive psychologist with research interests in human thinking and expertise. He has published over 100 ­peer-​­reviewed journal articles on expertise, systems design, and human ­decision-​­making and has managed over £10m external research funding, with a focus on creative expertise, ­problem-​­solving, and d­ ecision-​­making. His current theoretical work focusses on the development of computational models of insight during ­problem-​­solving, while his applied focus is on investigative expertise in the criminal justice system. He was elected a Fellow of the British Psychological Society in 2013. His current role at Sussex is as Director of the Applied Behavioural Science Unit (­w ww.appliedbehaviouralscience. co.uk), which provides psychology as a service to industry, commerce, government, and ­non-​­governmental organisations. Karenleigh A. Overmann is an Associate Professor of Anthropology (­Adjunct) and Director of the Center for Cognitive Archaeology at the University of Colorado, Colorado Springs (­UCCS). In June 2020, she completed two years of postdoctoral research at the University of Bergen (­MSCA individual fellowship, EU project 785793), and she was a visiting scholar at the University of Pittsburgh from September 2020 to June 2021. She earned her doctorate in Archaeology from the University of Oxford in 2016 as a Clarendon scholar. Susannah B. F. Paletz is an Associate Professor at the University of Maryland, College Park, College of Information Studies. Her research focusses on the intersections of teams, culture, creativity, and technology as well as on applied psychology in the national interest. She is on the editorial board of Small Group Research and has published in Human Factors, Behavior Research Methods, Cognition, the Journal of Organizational Behavior, and other outlets. She received her PhD and MA in Social/­Personality Psychology from UC Berkeley and her BA from Wesleyan University, where she received high honours in two majors, Psychology and Science in Society. Her past positions include being a civil servant research psychologist at NASA Ames Research Center, a Research Associate and postdoc at the University of Pittsburgh Learning Research and Development Center, and a Research Scientist at the Center for Advanced Study of Language at the University of Maryland. Kata ­Pauly-​­Takacs, PhD, is a Senior Lecturer in Cognitive Psychology at Leeds Beckett University, UK. Her research focusses on the cognitive neuropsychology of human memory, where she often adopts the s­ingle-​­case approach to appreciate the idiosyncratic memory failures associated with brain injury. She is also interested in human memory more broadly, considering the relationship between memory and associated cognitions such as ­meta-​­memory, creativity, and future thinking. David G. Pearson  is a Professor of Cognition and Cognitive Neuroscience at Anglia Ruskin University, Cambridge, UK. His research examines the cognitive processes involved xviii

Contributors

during memory, mental imagery, and v­ isuo-​­spatial thinking, with a particular focus on applications in the fields of clinical and environmental psychology. Mary A. Pei is a PhD candidate at the Johns Hopkins University School of Education and the ­2022–​­2023 recipient of the Johns Hopkins School of Education’s Innovation in Research Fellowship for her ­three-​­study, ­m ixed-​­methods dissertation examining talent identification tests. Her research uses linguistics and semiotics to examine creativity, talent development, and ­A sian-​­American experiences in education. Kelly Peters is the c­ o-​­founder and former CEO of BEworks. Recognised with the Lifetime Achievement Award in 2022 by Consulting Magazine, her career spans over thirty years of work in innovation and strategy for major corporations. She is a published author, speaker, and university lecturer. Jonathan A. Plucker  is the Julian C. Stanley Professor of Talent Development at Johns Hopkins University, where he works in the Center for Talented Youth and School of Education. His work focusses on talent development, creativity, and education policy. Roni ­Reiter-​­Palmon, PhD, is a Distinguished Professor of Industrial/­Organizational (­I/­ O) Psychology and the Director of the I/­O Psychology Graduate Program at the University of Nebraska at Omaha (­U NO). She is also the Director of Innovation for the Center for Collaboration Science, an interdisciplinary program at UNO. She received her PhD in I/­O Psychology from George Mason University. Her research focusses on creativity and innovation in the workplace, team creativity, development of teamwork and creative p­ roblem-​­solving skills, and leading creative individuals and teams. She is the ­President-​­Elect of Division 10 of the American Psychological Association (­Creativity). She is also a Fellow of Divisions 10 and 14 (­Creativity and SIOP) of the APA and has won the system wide research award from the University of Nebraska system in 2017. Beth H. Richardson is a Senior Lecturer in Experimental Social Psychology at the University of Central Lancashire, Preston, UK. She is interested in the role of language in team communication and the factors that lead to communication success and failure in a variety of ­real-​­world contexts, including ones that require p­ erspective-​­taking such as negotiation and joint ­problem-​­solving. Beth is currently developing a theoretical model of collaborative ­meta-​­reasoning to capture how groups monitor and control their ongoing reasoning in relation to the attainment of cooperative goals such as achieving creative outputs. Jala Rizeq,  PhD, is a Lecturer in Clinical Psychology within the School of Health and Wellbeing at the University of Glasgow. Her doctoral research was focussed on rational thinking and individual differences in contaminated mindware. Currently, she is interested in studying vulnerability pathways for serious mental health difficulties as a function of stressful life events. In her programme of research, she aims to integrate decision theory into mental health research. Wendy Ross is a Senior Lecturer at London Metropolitan University. Her research is concerned with the systematic investigation of serendipitous cognition. She is a ­co-​­chair of the Serendipity Society and ­co-​­editor of both The Art of Serendipity (­Palgrave, 2022) and Serendipity Science (­Springer, 2023). In 2021, she was awarded the Frank X Barron Award by the xix

Contributors

American Psychological Association for outstanding contribution by a research student to the study of creativity, aesthetics, and the arts. Mark A. Runco (­w ww.markrunco.com) is the Past President of the APA’s Division 10 and Editor of the Journal of Creativity. He is the Editor Emeritus of the Creativity Research Journal and has ­co-​­edited three editions of the Encyclopedia of Creativity. His Creativity textbook has been translated into over one dozen languages. (­The 3rd edition is due out early in 2023.) Mark is currently the Director of Creativity Research and Programming at Southern Oregon University. In May, he is to receive his second Lifetime Achievement Award for his work on creativity. For the past six years, he has been ­Co-​­Executive Director of the Annual Creativity Conference at SOU (­w ww.soucreativityconference.com). Evy Sakellariou  is an Associate Professor of Creativity and Innovation Management at Kingston Business School, Kingston University London and a Fellow of the Institute of Innovation and Knowledge Exchange, the UK professional body for innovators. Her previous work experience involves the leadership of multinational innovation marketing functions in the United Kingdom and overseas. Her research focusses on creativity and its role in organizational innovation processes, user innovation, and strategic foresight, and it has been published at World Elite and international leading academic journals and conference proceedings. Carola Salvi is a Professor in the Department of Psychology and Social Sciences at the John Cabot University ( ­JCU) of Rome and Associate Faculty at the Department of Psychiatry and Behavioral Sciences at the University of Texas at Austin (­Dell Med). Professor Salvi’s research focusses on the neural mechanisms underlying insight ­problem-​­solving, creativity, and cognitive flexibility. Professor Salvi’s theoretical questions expand from cognitive ­neuro-​­and behavioural sciences to social psychology and psychophysiology. In parallel to her scientific career, in 2021, Salvi started working on the theme of domestic violence, creating an art collection to raise awareness and funds to support this cause. In 2022, her paintings were exposed at the Witte Museum in San Antonio, TX, and became part of a permanent art collection at the Law School of the University of Texas at Austin. Lelyn D. Saner  is an Associate Behavior Scientist with the consulting firm Booz Allen Hamilton. His primary interest is in how people use information representations for situation awareness, critical d­ ecision-​­making, and ­problem-​­solving in operational environments. He has conducted research on the cognitive processes involved in scientific reasoning, dynamic behavioural ­decision-​­making, and collaboration, as well as the design and use of data analytics. He has used ethnographic techniques, tailored scenarios, and interactive simulations to assess and improve human performance in language analysis, intelligence analysis, and cyberspace defence. He received his MS in Human Factors/­Applied Cognition from George Mason University and his PhD in Cognitive Psychology from the University of Pittsburgh. Christian D. Schunn  is a Senior Scientist at the Learning Research and Development Center and a Professor of Psychology, Learning Sciences and Policy, and Intelligent Systems at the University of Pittsburgh. His current research interests include the psychology of engineering design, w ­ eb-​­based peer interaction and instruction, STEM learning, and engagement and learning. He is currently the Chair of the Executive of the International Society for Design & Development in Education, as well as a Fellow of several scientific societies xx

Contributors

(­A AAS, APA, APS). He has served on two National Academy of Engineering committees: ­K-​­12 Engineering Education and ­K-​­12 Engineering Education Standards. Finally, he has launched a startup called Peerceptiv that is based upon his research on t­ echnology-​­based peer assessment in high school and college settings. Wangbing Shen is a Professor in the Public Administration School at Hohai University, China. He is interested in the interaction between emotion and cognition in thinking, creativity, and p­ ro-​­social behaviours and conducts both controlled laboratory experiments as well as naturalistic studies in environmental psychology, marketing, and advertising. He has published 50+ publications in these domains, and he has also hosted awards from the National Natural and Social Sciences Foundation of China. Wangbing is a member of the editorial boards of the Journal of Cognitive Psychology and Applied Cognitive Psychology. Fanni Szigetvari  is a Behavioural Science Lab Manager at Kingston Business School, Kingston University London. She is also a current PhD student and previously worked as a Research Assistant at Kingston Business School. She holds a BA degree in Psychology from Karoli Gaspar University, Budapest (­2017), and she completed her MSc in Occupational and Business Psychology at Kingston University London (­2019). Maggie E. Toplak,  PhD, is an Associate Professor in the ­Clinical-​­Developmental Area of the Department of Psychology at York University, Canada. The focus of her research is on judgement, d­ ecision-​­making, and rational thinking, including their associations with individual differences in cognitive abilities. Her research has been informed by using participants across the lifespan (­including children, youth, and adults) and with special populations (­including youth with ADHD). Dagný Valgeirsdóttir is an Assistant Professor at DTU Skylab, the innovation hub of the Technical University of Denmark. Dagny conducts her work under the program initiative ‘­Technology Leaving No One Behind’, where she challenges engineering students to develop sustainable solutions using creative constraints and ­co-​­creation to enable an inclusive mindset. Frédéric ­Vallée-​­Tourangeau is a Professor of Psychology at Kingston University London. Recent projects on creative p­ roblem-​­solving have drawn inspiration from William James and Bruno Latour. He ­co-​­edited, with Stephen Cowley, Cognition Beyond the Brain (­Second Edition, Springer, 2017); Fred edited Insight: On the Origins of New Ideas (­Routledge, 2018). His upcoming book is titled Systemic Creative Cognition: Bruno Latour for Creativity Researchers ( ­Routledge, 2023). Gaëlle ­Vallée-​­Tourangeau is a Professor of Behavioural Science and Director of Research and Enterprise at Kingston Business School, Kingston University London. She studies people’s choices and decisions as well as how to change behaviour. She has applied behavioural sciences to answer questions such as what motivates healthcare workers to decide to get (­or not to get) vaccinated, what motivates knowledge workers to decide to support an application for funding, or how can we boost creative and insightful solutions to problems. Her research has been funded by the Fyssen Foundation (­2004), the French National Research Agency (­2008), the Leverhulme Trust (­2011), ­Sanofi-​­Pasteur (­2013), and the Wellcome Trust (­2019). xxi

Contributors

Charles A. Van Stockum, Jr. is a Doctoral Student at the University of Louisville, where he conducts research on cognitive flexibility and executive control in p­ roblem-​­solving. Oshin Vartanian  is a Defence Scientist at Defence Research and Development Canada and an Adjunct Associate Professor in the Department of Psychology at the University of Toronto. He received his PhD in experimental psychology from the University of Maine. He is the ­Co-​­Editor of Psychology of Aesthetics, Creativity, and the Arts and past Editor of Empirical Studies of the Arts. His ­co-​­edited volumes include Neuroaesthetics (­Baywood Publishing Company), Neuroscience of Creativity (­The MIT Press), Neuroscience of Decision Making ( ­Psychology Press), The Cambridge Handbook of the Neuroscience of Creativity (­Cambridge University Press), and most recently The Oxford Handbook of Empirical Aesthetics (­Oxford University Press). His main areas of interest include the cognitive and neural bases of aesthetics and creativity. Ananí M. Vasquez is a Doctoral Candidate in the Learning, Literacies, and Technologies program at Arizona State University. She is a former elementary teacher and teacher coach who combines her experiences in general, bilingual, gifted, and special education(­s) to envision an inclusive education. Ananí draws on creativity theory, disability studies in education, the neurodiversity paradigm, process philosophy, and a­ rts-​­based inquiry while working with others towards ­post-​­oppositional educational transformation. She is a ­co-​­editor of Writing and the Articulation of ­Post-​­Qualitative Research ( ­Routledge). Margaret E. Webb,  PhD, is an Honorary Research Fellow at the Melbourne School of Psychological Sciences, University of Melbourne. An insight enthusiast, all areas of insight are of interest to her. However, current areas of particular focus are individual biases towards feelings of insight in p­ roblem-​­solving and diversifying the task base used in insight research. Robert W. Weisberg is a Professor of Psychology and Neuroscience at Temple University, Philadelphia, Pennsylvania. His area of interest is creative thinking, the cognitive processes involved in the intentional production of novelty: solutions to problems, works of art, scientific discoveries, and inventions. He has published numerous papers and books investigating cognitive mechanisms underlying ­problem-​­solving and creative thinking through laboratory studies as well as by examining r­eal-​­world creative thinking through case studies of people such as Edison, Picasso, Frank Lloyd Wright, and jazz great Charlie Parker. In those studies, attempts are made to apply scientific methods to historical “­d ata” to derive conclusions concerning how the creative process functions at the highest levels. Jennifer Wiley is a Professor in the Department of Psychology at the University of Illinois, Chicago. She earned a PhD in Cognitive Psychology from the University of Pittsburgh, where she studied at the Learning Research and Development Center. Much of her work lies at the intersection of cognition and education. Her research interests include text comprehension, ­meta-​­comprehension, and p­ roblem-​­solving. She investigates conditions that enable individuals to solve problems creatively or more effectively and the impact of collaboration on learning and ­problem-​­solving. She also studies conditions that support effective comprehension and accurate ­meta-​­comprehension and interventions that improve learning from text. Travis J. Wiltshire is an Assistant Professor in the Department of Cognitive Science and Artificial Intelligence, Tilburg University. He has contributed over 40 scholarly works and xxii

Contributors

is a (­co)­principal investigator on research projects investigating various aspects of coordination, complex systems, cognition, and h ­ uman-​­system interaction. He completed his PhD in Modelling and Simulation from the University of Central Florida in 2015. Andre Zamani is a Master’s Student at the University of British Columbia. His main areas of research interest include spontaneous thought, thought dynamics, ­self-​­experience, and functional magnetic resonance imaging (­f MRI) methods. In his current work, Andre uses a neurophenomenological approach combining m ­ editation-​­assisted experience sampling and fMRI to study the conscious onset of spontaneous thoughts. Daniel C. Zeitlen is a PhD student in Psychology at Pennsylvania State University, working in the Cognitive Neuroscience of Creativity Lab under the mentorship of Dr. Roger Beaty. Daniel received his BA in Psychology and Sociology from the University of North Carolina at Asheville, where he conducted interdisciplinary research on music cognition and perception, and his MS in Psychology from the Pennsylvania State University. Daniel primarily uses cognitive, neural, and social methods to study relationships among creativity, memory, and emotion and is interested in understanding both how and why people think in creative ways (­including the mechanisms, motivations, and effects of creativity).

xxiii

PREFACE

In deciding to compile the present handbook, our aim was to provide an authoritative, international reference work to offer researchers and students a w ­ ell-​­balanced overview of current scholarship across the full breadth of the rapidly expanding field of creative cognition. We believe that we have achieved this ­a im – ​­but only because we have been privileged to be able to draw upon the extensive expertise of so many key researchers working in the field of creative cognition. Most of our contributors have already established themselves as world leaders in this field, but the handbook also includes valuable contributions from m ­ id-​­career researchers who have a reputation for exceptional research as well as some rising stars who are still at an early career stage but who are already being noticed for their outstanding work. Collectively, it has been a great pleasure for us to interact with such an inspiring group of authors, and we thank them all for their dedicated work in producing ­h igh-​­quality chapters while also being patient with us as we progressed with the compilation of such a major volume of work. The resulting handbook surveys the current state of the art in research on creative cognition, reviewing both empirical evidence and theoretical developments. The handbook draws extensively on contemporary work exploring the cognitive representations and processes associated with creativity, whether studied in the laboratory or as it arises in r­eal-​­world practice in domains such as education, art, design, science, business, and technological innovation. The handbook also extends to an examination of the sociocognitive and cultural aspects of creativity in teams and organisations while additionally capturing the latest research on the cognitive neuroscience of creativity. Overall, the handbook overviews key, emerging trends and debates in the field of creative cognition and positions these in relation to established findings and theories to give a clear sense of the way in which new ideas are challenging traditional viewpoints. The handbook is organized thematically into six parts, and we summarise in more detail the chapters that appear in each of these parts so as to give a flavour of the way in which they advance an understanding of creative cognition. First, however, we would like to take the opportunity to extend our gratitude to many people who helped to make this handbook a reality. Again, we thank the authors for their dedicated work in producing excellent chapters, all of which required at least one further iteration to address feedback and suggestions for improvement. Many of our contributors also acted as peer reviewers of each other’s xxiv

Preface

chapters, which we believe has enabled further enhancements to be made to the presented research. We additionally thank our wonderful editorial colleagues at Routledge for their support and assistance throughout the development of this volume: Ceri McLardy, Cloe Holland, Alex Howard, Khyati Sanger, and Emilie Coin. We are now pleased to be bringing this endeavour to its fruition, and we very much hope that readers will enjoy the research that is discussed in the following pages.

Part I: Reflections on the Fundamental Contents and Mechanisms of Creative Cognition As befits a handbook of creative cognition, many of the chapters in the present volume address the underpinning representational contents and thought processes that are associated with ideational cognition, namely the generation of new ideas. The chapters in Part I of the handbook provide readers with a useful grounding in some of the fundamental definitional, conceptual, and measurement issues and challenges that arise in the analysis and understanding of creative cognition. Mark Runco’s chapter reviews research on divergent thinking in terms of process models and research practices that measure it. He offers an interesting warning (­reminiscent of Perkins’s 1981 lesson at the end of The mind’s best work): It’s best to avoid creativity as a noun and rather aim to trace and understand the constellation of factors that fashion creative behaviour. Karl Jeffries reviews the consensual assessment technique (­CAT), which is often used to evaluate the creative products that arise from divergent thinking. He offers some methodological recommendations to restore the shine to this ­once-​­called “­gold standard” technique, possibly tarnished through the heterogeneous manner in which it has been deployed. His chapter delivers a unique, important, and highly rigorous insight into the past, present, and “­potential” future of the CAT in creativity assessment. Jeffries calls upon the creativity research community to come together and negotiate a set of parameters within which the CAT could be consistently applied (­a “­CAT Accord”), which would enhance comparability across studies and improve replicability. Balder Onarheim and Dagný Valgeirsdóttir’s reflections on the contextual factors that enact creative behaviour invite us to abandon the notion that an environment free of constraints is ideal for the manifestation of creativity. If creative behaviour is situated, then constraints are participatory. Onarheim and Valgeirsdóttir’s taxonomy of constraints is particularly helpful in better understanding the situations (­a nd their “­constraindness”) that enhance or mitigate the expression of creative behaviour. The situated character of creative behaviour calls for the granular analysis of the factors external to the agent, enjoining a more systemic characterisation of creative cognition. Wendy Ross and Selene Arfini also focus on the essential role of the wider context in creative cognition, with their examination of how the external environment (­including human networks) can trigger the production of creative ideas and solutions. Ross and Arfini develop a principled framework within which the constitutive role of “­serendipity” in creativity can be acknowledged and measured. In capitalising upon the concept of serendipity as an explanatory device, Ross and Arfini argue that it describes an essential feature of creative cognition that, to be fully understood, requires a perspective on mind that embraces both its distributed and extended nature. The chapter by Janet Metcalfe and William Jacobs and the one by Vlad P. Glăveanu offer a novel and useful taxonomy of different types of curiosity and thinking, respectively. Metcalfe and Jacobs argue that curiosity falls into two broad types. Curiosity1 is a feature of focussed ­problem-​­solving that seeks to achieve a normative solution. Curiosity2, however, is xxv

Preface

exploratory and playful, a form of interactive instrumentality particularly attuned to contingent, unexpected, and serendipitous discoveries. In turn, Glăveanu invites a distinction between “­possibility” and “­actuality” thinking. His chapter focusses primarily on the former, a system of thought formatted along four dimensions: what could be, what is to come, what could have been, and what is not/­never could be. The chapter by Tom Ormerod maps out the useful intersection of research on analogical reasoning and insight p­ roblem-​­solving that has, hitherto, remained largely unexplored. The chapter argues that insight into a source problem is critical for successful analogical transfer, yet at the same time, analogical transfer can offer a powerful mechanism to enable people to make effective use of the knowledge that they have gained through insight. It seems clear that further exploration of the fascinating interplay between insight and analogy represents a rich vein for future research. Nathaniel Barr, Lucas Klein, Michael McNamara, and Kelly Peters offer an interesting contrast between ideational cognition and innovative cognition. Launched from the premise that “­creativity is not enough”, they underscore the importance of idea development and implementation. Their chapter provides a good review of the mechanics of idea generation (­w ith some pertinent reflections on insight), and readers of the handbook will appreciate the clarity with which the challenges of innovative cognition research are described as well as how the authors detail a number of interesting c­ ross-​­disciplinary collaborations that leverage the concepts and methods of cognitive psychology in innovation research. Innovative cognition invites a distinction between “­d iffusion” and “­translation”, a distinction initially articulated in Latour (­1987). In a diffusion model, an initial mentally evinced idea survives unscathed through its various expressions and representations across different media and people. A translation model, however, stresses that with each articulation, with each formulation wrought through different media, the idea is translated into a slightly different one. These translations carry a cost and leave material traces that guide and constrain their next expression. In his chapter, Frédéric ­Vallée-​­Tourangeau forefronts the material expression of an idea, embodied in the physical model of a problem’s solution. He enjoins ­problem-​­solving researchers to design procedures wherein participants can interact with and construct a physical model of the solution to a vexing problem. ­Vallée-​­Tourangeau argues that the model and its transformation over time and space capture an agent’s knowledge and strategies, and that a granular analysis of the model’s temporal polymorphy reveals the origin of a new idea.

Part II: Reflections on the Nature of Creative Cognition as Revealed through Traditional Methodologies Part II of the handbook is devoted to providing a deep and broad coverage of our contemporary understanding of the nature of creative cognition as revealed through various traditional research methodologies that include experimental, psychometric, ecological, qualitative, and case study approaches. It will be seen how these different methodologies function in a complementary manner to provide a convergent understanding of creative cognition. Richard Hass offers an ­up-­​­­to-​­date, articulate, and astute consideration of how search in associative memory supports idea generation in divergent thinking, as assessed using the Alternative Uses Task (­AUT). Hass reviews evidence demonstrating that processing during the AUT is related to semantic memory search and retrieval. Readers of the handbook will appreciate the frank and open appraisal of the strengths and weaknesses of research on divergent thinking that draws upon theories of semantic memory search and retrieval. Yoed Kenett’s xxvi

Preface

chapter also addresses the links between semantic memory and creative cognition. He develops the important thesis that creativity involves search processes that are both facilitated and constrained by the structure of semantic memory. The chapter moves from an initial consideration of the application of computational models in studying semantic memory in creative thinking and the nature of creative search processes through to a consideration of the costs and benefits of creative search that arise from a rich semantic memory structure. David Pearson focusses on the role of mental imagery in creativity. His chapter opens with the anecdotal accounts (­Kekulé, Mozart, Einstein, and Hawking) of mental imagery in creativity and discovery, with nicely toned critical reflections. The review of creative mental synthesis as well as the brief window on working memory set the stage for the discussion of the role of internal and external representations in creative cognition. The chapter closes with a review of the benefits of imagery on creative cognition. Ken Gilhooly reviews incubation as a psychological process and as a research procedure in a chapter that also includes a substantial review of sleep as a form of incubation. Gilhooly additionally offers a constructive commentary on anecdotal accounts of incubation and related proverbial wisdom, illustrating how at times research draws from folk notions of incubation as well as identifying their shortcomings. The chapter opens with a useful definition of routine and n ­ on-​­routine problems, the latter solved only through restructuring that evinces a new idea, which in turn unveils the solution. Wallas’ model is reviewed with the helpful reminder that he considered a stage before illumination, namely “­intimation”. The chapter outlines the main candidate explanations for incubation effects, including the o ­ pportunistic-​­assimilation model and the author’s Goal + Associative Network Interaction (­GANI) model. Genevieve Cseh reviews efforts aimed at understanding the relationship between emotion/­ a ffect and creativity, offering excellent reflections on flow, mixed/­ paradoxical emotions, threat, and challenge in creativity. The dialectical perspective suggests a dynamic, temporal dimension to the interaction between emotion and creativity. The review ranges across many different types of creative behaviour and cognition, elicited through many different tasks. Task complexity, temporality, materiality, and expertise are all dimensions that may inform the nature of the emotion dialectic at the heart of this chapter. Jennifer Wiley and Olga Goldenberg’s chapter outlines the history and rationale of brainstorming research, the logic of the comparison with nominal groups, and hence how the effectiveness of brainstorming is assessed. The chapter reviews process loss and how this might be mitigated through virtual interfaces and alternative idea generation and sharing techniques. A detailed analysis of the p­ roblem-​­solving tasks employed in collaborative thinking research offers an interesting topography of the opportunities and constraints in group creative ideation, identifying fruitful avenues for research. Thomas Wiben Jensen looks at language in action and how creativity is enacted in metaphor and metonymy. His chapter helps us understand how work on embodied cognition, as illustrated by research on metaphor, ironically strengthened rather than weakened a dualist cognitive science. Jensen offers a detailed qualitative analysis of conversational interactions in two ­settings – ​­in a kindergarten and a therapy ­session – ​­and shows how metaphoric or metonymic meaning emerges from embodied enactment rather than just a linguistic process. Christian Schunn, Lelyn Saner, and Susannah B. F. Paletz present an interesting and unique angle on the nature of scientific narratives as presented in lab meetings and colloquia. The authors focus on external references and the d­ omain-​­distance of such references, which prove to be informative in revealing important differences across contexts, phases, and speaker roles. The chapter reports original data and analyses of the narrative employed in research presentations with the aim of addressing two research questions: (­i) to examine xxvii

Preface

the rate of “­external references” in colloquia and ­lab-​­group meetings; and (­ii) to examine the relative “­d istance” (­­near-​­domain or f­ ar-​­domain) of such external references in colloquia and ­lab-​­g roup meetings. Jala Rizeq and Maggie E. Toplak’s contribution draws interesting parallels between the research questions and methodologies employed in judgment and ­decision-​­making research with those in ­creative-​­cognition research. The concepts that structure the authors’ ­narrative – namely, cognitive sophistication, miserly processing, pro​­ cessing requirements, knowledge requirements, and contaminated ­m indware – ​­map out a landscape that could also be fruitfully explored by creative cognition researchers. Amory Danek’s chapter opens with fascinating scholarship on the Würzburg institute and the work of Karl Bühler at the turn of the 20th century, who first documented the phenomenology of the “­A ha!” experience: “­A peculiar yet pleasurable experience that occurs after a sudden insight”. Bühler also sought to explain the mental mechanics that produced the experience and expressed them in terms of associations among remote concepts. Danek’s chapter offers a detailed analysis of the phenomenology of Aha! and describes the dimensions along which it can be measured. Marci DeCaro and Charles Van Stockum, Jr. address the conflicting results relating to the association between working memory and insight and tackle the complex issues carefully, systematically, and astutely so as to provide considerable theoretical clarification. By the end of the chapter, the reader is well positioned to appreciate the likely reasons for inconsistent findings and have a full understanding of the implications of existing data for theories of insight ­problem-​­solving. Margaret E. Webb captures the growth and breadth of research efforts on insight. Her chapter offers an excellent review of the models developed to account for insight, as well as some ­meta-​­reflections on the nature of the models themselves (­d iagrammatic, verbal, s­emi-​­formal, and computational). Some of these models also aim to capture the phenomenology of the experience. Beth Richardson, Linden J. Ball, Bo Christensen, and John Marsh take a step away from the core “­­object-​­level” thought processes that are involved in creative cognition in order to focus on the m ­ eta-​­cognitive processes that function to monitor and control ongoing creative ­problem-​­solving. These authors note that although existing research has gone some way towards addressing such “­­meta-​­reasoning” processes at an individual level, there is as yet only a limited understanding of ­meta-​­reasoning at a collaborative level. This relative dearth of research on collaborative ­meta-​­reasoning presents a major theoretical challenge in the field of creative cognition, where much creative activity in ­real-​­world domains such as design, entrepreneurship, and scientific discovery is t­eam-​­based rather than driven by individuals. In their chapter, Richardson and colleagues aim to motivate future research by mapping out the cornerstones of a “­collaborative m ­ eta-​­reasoning framework”, which is of key relevance to ­team-​­based creativity.

Part III: Reflections on the Nature of Creative Cognition as Revealed through Cognitive Neuroscience Approaches Part III peers into the brain and focusses on studies of creative cognition that involve neuroscience approaches such as structural and functional analyses of brain systems underpinning creative cognition as well as research that has examined biomarkers of creativity. The chapters illustrate the productive interplay between efforts to map the functional architecture of creative processes and the exploration and identification of their neural correlates. However, the stage is first set with some reflections from cognitive archaeology. Frederick Coolidge anchors his narrative in terms of three major cognitive leaps, namely the Acheulian handaxe (­1.8 million years ago), artistic (­or “­­over-​­determined”) handaxes (­500,000 years ago), and the explosive wealth of artefacts xxviii

Preface

that mark the start of the upper Paleolithic (­50,000 years ago). The remit and stakes of cognitive archaeology are particularly nicely illustrated in Coolidge’s chapter. The three “­cognitive Rubicons” and their conjectured neural correlates are clearly outlined, along with interesting commentaries on creativity as manifest in these tools and artefacts. There is also some helpful pedagogy on offer about the cerebellum (­and cerebrum), tools, hominins, and hominin species. Halima Ahmed, Kata ­Pauly-​­Takacs, and Anna Abraham’s chapter offers a synoptic review of the role of declarative memory in creative ideation. There is an excellent pedagogical element here concerning semantic and episodic memory and a clear outline of how memory organization and retrieval processes are implicated in creative cognition. The constructive nature of the episodic memory system may be implicated in processes of creative idea generation. Roger Beaty, Hannah Merseal, and Daniel Zeitlen offer a clear, u ­ p-­​­­to-​­date, and informative review of network neuroscience research and findings in creative cognition. The authors identify the neurocognitive mechanisms that underpin creative cognition, including the functional contributions of the default network and executive control network. Their review and analysis are structured in a way that captures the sophisticated interplay between these brain networks across ­domain-​­general and ­domain-​­specific creative contexts. Readers of the handbook will welcome the authors’ perspective on individual differences in relation to brain connectivity and creativity, which also extends to the role of expertise in ­domain-​­specific creative performance (­e.g., in musical improvisation). Andre Zamani, Caitlin Mills, Manesh Girn, and Kalina Christoff focus on the generative and evaluative modes of thought. There is much neuroscience research that undergirds the distinction, but the authors focus primarily on the transition and dynamic interplay between these two modes of thought. They review models of transitions and propose neurocognitive mechanisms. Their chapter concludes with a series of open questions concerning the nature of these transitions. Carola Salvi’s chapter reviews the neuroscience of Aha! moments, offering an interesting typology of “­m arkers of insight”. Such Aha! moments can be measured in terms of internal attention allocation, increased activity in the right anterior temporal lobe, the ­reward-​­dopamine system, and pupil dilation. Insight may originate through the involvement of subcortical areas responsible for learning, alertness, and emotions. Maxi Becker, Roberto Cabeza, and Jasmin Kizilirmak cast ­insight – captured through the solution ​­ of Remote Associate ­Problems  – ​­as the product of memory processes in the search for a solution. Their chapter provides an excellent review of these processes and their neural correlates. The authors offer a clear synopsis of the recent literature, summarised diagrammatically, that provides a clear map of the processes as they unfold in time (­a nd neural space). In turn, Jasmin Kizilirmak and Maxi Becker’s chapter explores the role of insight in the formation of new memories. They discuss the cognitive, affective, and neurocognitive candidate mechanisms that may underlie learning in insight p­ roblem-​­solving. The chapter reviews how insight is implicated in learning with clearly structured ­reflections – ​­in terms of the behavioural evidence and neural ­correlates – ​­on the generation effect, the affective component, novelty and surprise, and prior learning.

Part IV: Reflections on Creative Cognition from Pedagogical, Organisational, Archaeological and ­Post-​­Phenomenological Perspectives In Part IV, the handbook moves beyond research on creative cognition that has relied on traditional methodologies as well as cognitive neuroscience approaches so as to capture concepts deriving from pedagogical, organisational, archaeological, and p­ ost-​­phenomenological research. The 21st-​­century workforce should be equipped with flexible ­problem-​­solving skills, xxix

Preface

which poses an interesting and important challenge to systems of education. Mary A. Pei and Jonathan Plucker’s chapter helps us appreciate the magnitude of this challenge and the tensions within pedagogic practice, which relies on standardised evaluation as well as stable and predictable classroom environments. The authors review four models of creativity and offer reflections on how these models intersect with classroom exigencies. The chapter closes with actionable recommendations applicable to both students and teachers. Creative learning is the focus of Ronald Beghetto and Ananí Vasquez’s chapter. They cast creative learning as an emergent product of a broad range of intrapsychological and interpsychological factors that can only be captured by an equally expansive, pluralistic, and open research methodology. Readers are introduced to the concept of “­actionable uncertainty” and the importance of curating creative artefacts. This curating process offers an interesting distributed and systemic perspective on learning with important pedagogical implications. Chris Baber’s chapter clearly takes creativity out of the mind, encouraging a process ontology perspective that offers much traction, which in turn cues a different set of questions concerning its manifestation (­coupled with new methodologies and research avenues). From Baber’s perspective, tools are not mere passive instruments that physically implement an idea in an agent’s mind. Tools are also performative; they actively contribute to the articulation, or instauration, of an idea; they are “­part of an autopoietic system”. The author also forefronts the ­object – ​­qua creative ­product – in ​­ its different interim p­ hases – ​­and its evolution during production. For Baber “­the artwork becomes a perceptual object”, which engages the agent in a “­constant dialogue”. Creativity emerges through a contingent, t­ool-​­mediated process. Paul Louis March and Lambros Malafouris invite us to reflect on the development of a large sculptural installation from the perspective of “­Material Engagement Theory”. Clay is thus an important actant in the development of the artwork. The artwork’s developmental trajectory is cast in terms of systemic forces, namely, as the result of dynamic imbroglios of human and n ­ on-​­human actants involved in the protracted coalescing of the sculptural installation. And in the process, important questions are raised and answers sketched about agency, intentionality, and creativity. As the authors memorably put it, the artwork “­learns itself into existence”. A praxiographic study of creativity, or more specifically, how researchers, through a broad spectrum of methods, enact creativity, would likely illustrate how such methods perform it. That is, if we abandon assumptions about the antecedent singularity of reality (­waiting to be discovered if only the right method was employed) and adopt in turn a perspective that casts methods as participatory, shaping reality, then research methods themselves become an object of scrutiny to better appreciate how creativity is reified. On that note, Darryl Hocking looks at the discursive construction of creativity. He reviews a number of ­d iscourse-​ ­oriented studies that treat creativity as a phenomenon produced through social and historical discursive practices. Travis Wiltshire and Merle T. Fairhurst enact the phenomenon of collaborative creativity through a mathematical characterisation of coordination dynamics in terms of transfer entropy and prediction decay. They use extant datasets from a dyadic mirror game and study on musical improvisation. These analyses make manifest the nature of unidirectional and bidirectional influences as a function of task and expertise. Karenleigh Overmann’s chapter casts the invention and development of ­w riting –​ ­c irca 4th millennium BCE in Mesopotamia and ­Egypt – ​­as a collaborative and collective process distributed over people and generations. The creative innovations revealed in the development of these new forms of encoding and communication are best understood not in terms of individual breakthroughs but rather in terms of the distributed and collaborative use of early writing systems such that by the 2nd millennium BCE literacy emerges as a xxx

Preface

cognitive feature of homo sapiens. Overmann shuns an individualistic model of creativity and underscores the collective dynamics through which scripts and literacy c­ o-​­evolved.

Part V: Reflections on Creative Cognition in Domains Involving Creativity and Innovation In Part V, the handbook considers creative cognition in ­real-​­world domains such as art, advertising, engineering, technology, and entrepreneurship. This part of the handbook opens with a chapter by Roni ­Reiter-​­Palmon and Payge Japp on team cognition and creativity. Their review is structured in terms of: (­i) problem identification and construction; (­ii) idea generation or brainstorming; and (­iii) idea evaluation and selection. The social and conversational dynamics of teamwork offer a fascinating window onto what has been traditionally investigated in terms of individuated ­problem-​­solving agents. David Cropley’s chapter takes readers on a tour of creative cognition in engineering and technology. The chapter opens with a helpful nomenclature of innovations (­stagnation, replication, incrementation, and disruption) and a concrete illustration of its application with the development of the velocipede. The research reviewed here intersects constructively with concepts such as divergent and convergent thinking. Cropley also offers an interesting and useful distinction between discovery and invention. Bo Christensen’s chapter explores entrepreneurial cognition. Creative outcomes abound in the processes involved in the recognition of opportunities, the generation of new business ideas, and the strategies for business p­ roblem-​­solving. Christensen demonstrates how many general principles from creative cognition play out in entrepreneurial cognition in nuanced ways that are attuned to the dynamic context that entrepreneurs find themselves in, which is fraught with uncertainty and where limited means and resources are available to achieve successful outcomes. Christensen’s chapter is followed by an interesting review and exploration of creativity as showcased in advertising, with Wangbing Shen, Linden J. Ball, and Beth H. Richardson offering a granular review of the manner in which creative advertising takes shape from the initial stage of advertising planning to the final stage of marketing communication. The chapter also reviews studies that explore creative advertising effectiveness in terms of levels of attention capture, understanding, and memorability. Oshin Vartanian’s chapter deals with the underpinning neuroscience of creativity and the appreciation of visual artistic artefacts. Vartanian’s key argument is that although the creation and appreciation of art are supported by common systems, there are also some differences in terms of the structures that support ideation, sensory processing, and semantic processing. Moreover, the evidence reviewed is broadly consistent with the notion that art serves as the communicative medium for the transmission of meaning from the creator to the consumer through the engagement of psychological processes to meet specific goals. This position is nicely captured by Tinio’s (­2013) “­M irror Model of Art”– ​­that the aesthetic experience mirrors (­in reverse order) the three stages of the ­art-​­making process, namely initialisation, expansion and adaptation, and finalising. Part V of the handbook closes with an interdisciplinary scoping review of creative cognition conducted by Gaëlle ­Vallée-​­Tourangeau, Evy Sakellariou, and Fanni Szigetvari. The chapter is particularly useful for the broad constituency of readers of this handbook, outlining as it does the concerns and methods employed in innovation management research and experimental cognitive psychology. ­Vallée-​­Tourangeau and colleagues structure their review in terms of five d­ imensions – ​­topos, demos, logos, pathos, and e­ thos – ​­together with the subsequent addition of praxis. This structure helps develop a highly productive synoptic xxxi

Preface

and contrastive analysis of creative cognition research across disciplines. The systematic mapping protocol provided here is rigorous and transparent, and the method itself makes a very useful contribution to the handbook in and of itself, in addition to the resulting findings.

Part VI: Reflections on the Paradoxical Misalignment Between Findings that Derive from In Vivo Versus In Vitro Research on Creative Cognition The handbook closes with a chapter by Robert Weisberg, who presents a fascinating and important critical perspective on the prevalent framework of in vitro research that dominates studies of creative cognition. This framework is so ubiquitous and paradigmatic that it has become invisible to both practitioners and consumers of creativity research conducted in the lab. A key contribution of Weisberg’s chapter is the light that it sheds on this prevailing paradigm, bringing it out of the invisible shadow that it casts on research. The underlying principle of this perspective on creativity is that a new idea arises from the association between distant and hitherto unrelated ideas. The origin of this ­perspective – ​­that creativity builds up from remote ­a ssociations – ​­is nicely traced and illustrated in the first third of the chapter: from Poincaré to Mednick and beyond. After briefly reviewing eight case studies (­v iz. Picasso’s Guernica, Watson and Crick’s discovery of DNA, the Wright Brother’s invention, Warren and Marshall’s bacterial theory of ulcers, Wag Dodge’s escape fire, Frank Lloyd Wright’s cantilevered Fallingwater, da Vinci’s aerial screw, and Pollock’s dripped paintings), the author acknowledges that creativity unfolds in time and space, but in all cases, the author illustrates convincingly how near associations rather than remote ones appear to be implicated in the origin of the innovation. This section of the chapter then equips the reader to appreciate the creativity paradox: the manner with which creativity researchers seek to mobilise creativity in the laboratory implicates remote associations while the granular analysis of in vivo work does not. Weisberg offers different ways out of this paradox in the final section of his chapter, scaffolded primarily in terms of his own model of creative ideation. In closing our handbook with Weisberg’s erudite contribution to the literature, which challenges as it does the accepted orthodoxy in the study of creative cognition, we hope to inspire researchers to investigate the apparent misalignment that exists between findings deriving from in vivo versus in vitro research so as to ensure progress towards a satisfactory rapprochement. Frédéric ­Vallée-​­Tourangeau Kingston University London, UK Linden J. Ball University of Central Lancashire Preston, UK December 2022

References Latour, B. (­1987). Science in action: How to follow scientists and engineers through society. Harvard University Press. Perkins, D. N. (­1981). The mind’s best work. Harvard University Press. Tinio, P. P. L. (­2013). From artistic creation to aesthetic reception: The mirror model of art. Psychology of Aesthetics, Creativity, and the Arts, 7, ­265–​­275. https://­doi.org/­10.1037/­a0030872

xxxii

PART I

Reflections on the Fundamental Contents and Mechanisms of Creative Cognition

1 DIVERGENT THINKING AS CREATIVE COGNITION Mark A. Runco

Introduction Divergent thinking (­DT) tests have been used for decades to estimate the potential for creative thinking. They are reliable and have selective evidence for validity. This chapter reviews the theory behind DT and summarizes the evidence for it. It also explores what DT says about creative cognition. The original theoretical basis of DT is summarized, as are the theoretical and methodological advances that have been presented since the original conception of DT. This includes the theory of cognitive hyperspace and the computer models that are informing advances in the automated scoring of DT tests. The use of DT testing for predicting actual creative behavior is discussed, as is research using DT tests that get at particular p­ roblem-​­finding skills. The limitations of DT are included in this chapter, as are various implications for our understanding of creative cognition. DT is not synonymous with creative thinking. It is often involved in creative thinking, and DT tests have been refined such that they provide reliable information about one kind of creative cognition. Certain conditions must be met when doing research with DT or construct validity suffers. Under the wrong conditions, the results of DT testing are not indicative of creative thinking. Indeed, there is a right way to collect DT data and a wrong way. This chapter identifies each and reviews the theories that explain what parts of creative cognition are in fact tied to DT. These objectives for this chapter require that some of the original work on DT be reviewed (­e.g., Guilford, 1968), but this chapter also summarizes some of the very recent work using DT tests and measures.

Context There is a good reason to avoid the noun “­creativity.” There are just too many things that fit under the creative umbrella. It is much better, especially in scholarly discourse, to use the adjective and thus specify the creative process, creative achievement, creative product, creative economy, creative capacity, or the like. This applies to the present chapter because the focus is specific and the interest is in creative cognition. Creative cognition plays a central role in creative problem solving, creative achievement, and perhaps all creative behavior, but it is not ­a ll-​­important. That means that the focus here on creative cognition does not provide a DOI: 10.4324/9781003009351-2

3

Mark A. Runco

Motivation

Knowledge

(Intrinsic and Extrinsic)

(Procedural and Declarative)

Problem Finding

Ideation

(Identification and Definition)

(Fluency, Originality,

Evaluation/Decision-Making (Interpersonal and Intrapersonal)

Flexibility)

­Figure 1.1 The ­two-​­tier model of the creative process (­Runco & Chand, 1992). Not all directions of effect (­a rrows) are included. Recursion, where the individual has moved through the process but then returns to an earlier stage, is also not depicted but is likely Source: Copyright 2021 Mark A. Runco.

­Figure 1.2 Hierarchical framework for the study and understanding of creativity. Persuasion (­Simonton, 1988) and Potential (­Runco, 2007) were added to the original 4Ps (­i n italics, from Rhodes, 1961) when it was ­re-​­organized as a hierarchy

comprehensive perspective on “­creativity.” Creative behavior probably depends on creative cognition but also depends on motivation, attitude, setting and climate, and personality. The ­two-​­tier model of the creative process conveys this idea (­see ­Figure 1.1), as does the idea that “­creativity” is a syndrome or complex (­and not a unitary thing) (­MacKinnon, 1965; Mumford & Gustafson, 1988). Another way of putting DT into context is to consider the 4P theory and its most recent revision, which is a hierarchy (­see F ­ igure  1.2). The 4P framework dates back to Rhodes (­1961) and included creative product, creative process, creative press (­or place), and creative 4

Divergent Thinking as Creative Cognition

person. Two decades later, this was extended to include persuasion, the idea being that creative people change the way other people think (­Simonton, 1988). Creative ideas and creative people are persuasive. That meant there was a fifth P, but then a sixth was proposed. It was creative potential. When the idea of potential was added, the entire framework was revamped such that all approaches to creativity were subsumed under one of two superordinate categories, creative potential or creative performance. The former is latent and does not guarantee actual creative behavior. It requires support and fulfillment. The latter is manifest and can be objectively measured. This is relevant because creative cognition is indicative of creative potential. A person may have the capacity for creative cognition but not use it. Still, DT is so often studied because it is a kind of creative cognition and, as we will see, lends itself to objective study. This will be apparent throughout this chapter as empirical research is summarized.

Background There is a surprisingly long history of theory and research about DT. Typically, the influential work of Guilford (­1950) is cited as a starting point, but the history is actually much older than that. Why is the older work ignored? One reason is that the term divergent thinking was not used until after Guilford (­1950). He actually referred to divergent production because, in his view, DT was just one part of the structure of the intellect. Several reviews (­Plucker et al., in press; Runco, 1991) have identified research using o ­ pen-​­ended tasks and examining ideation going back much further than 1950. Some of it is well over 100 years old. Even the w ­ ell-​­known work of Alfred Binet can be cited here (­Binet  & Simon, 1905). He included an ­open-​­ended task as one part of his tests of “­mental ability.” This was to be modified by Terman in the USA and became the first IQ test, the S­ tanford-​­Binet. Binet was interested in intelligence and not creativity. That was typical way back then. As a matter of fact, the first significant issue to be explored in the research on creativity was how it was related to, and perhaps distinct from, general intelligence. There was great debate, with some people thinking that creativity was distinct and others believing that creativity was just a particular kind of intelligence. In this latter view, creative abilities were subsumed under general intellectual abilities. IQ tests were supposed to measure general abilities, also known as g, which is why Binet included an ­open-​­ended task that was similar to DT tasks. ­Open-​­ended tasks allow examinees to generate a large number of ideas (­rather than just one). Apparently, Binet held the view that all intellectual tasks, including those requiring one correct answer as well as those allowing multiple responses, were indicative of “­g.” Guilford did quite a bit to get creativity research brought into the sciences. He gave the presidential address to the American Psychological Association, the published version being the 1950 article titled “­Creativity.” This remains one of the most important papers published on the topic. Indeed, a special issue of the Creativity Research Journal summarized the lasting impact of that influential article by G ­ uilford – ​­50 years after it was published (­Plucker, 2001). One of the important messages of the 1950 paper was that creativity was distinct from intelligence. As a matter of fact, Guilford believed that there were 180 different kinds of cognitive ability. His program of research was focused on a theory and model that described each of these distinct abilities. This structure of the intellect model is usually depicted as a ­three-​­dimensional cube. One dimension is all about the products of “­the intellect,” which explains why Guilford used the term divergent production. Divergent production was one way that individuals produce ideas and other mental outcomes. Guilford was clear that divergent processes were orthogonal to processes that led to convergent products. This distinction 5

Mark A. Runco

provided a way to think about how things like IQ and general intelligence do, in fact, differ from creative processes. The former require convergent thinking, and the latter, DT. Guilford developed a number of tests for both DT and convergent processes. Several of his tests (­e.g., Consequences) are still used today. There is still a bit of research on the structure of intellect (­Bachelor & Michael, 1991; Meeker, 1969), but most of the research is specifically on DT rather than the broader structure of intellect.

Distinctiveness of Divergent Thinking The psychometric research examining how DT (­or any test of creative potential) is distinct from IQ and general intelligence is concerned with discriminant validity. As noted above, such distinctiveness was the central issue early on, and as such it received a huge amount of attention for several decades (­e.g., Getzels & Jackson, 1962; Wallach & Kogan, 1965; Wallach & Wing, 1969). There is still research examining the discriminant validity of DT, and in fact so much has been published that there are now m ­ eta-​­analyses of the ­intelligence–​­creativity relationship (­e.g., Gerwig et al., 2021; Kim, 2005). A general interpretation of the research is that there is good discriminant validity and separation between intelligence and DT. Like all forms of validity, this is not an either/­or situation. Validity is always a matter of degree (­which is indicated by a correlation coefficient). In the case of intelligence and DT, the degree depends on the sample of examinees and the tests used (­Runco & Albert, 1986). The sample is very important because the relationship seems to vary at different levels of intelligence. This is described by the threshold theory, wherein there is a l­evel – ​­the t­hreshold – ​­of general intelligence below which an individual cannot really be highly original on a DT test, but above that threshold, individuals can either be original and creative or not. Guilford himself had data to support this view, which he called the triangular theory. That is because when he plotted DT as a function of intelligence, the scatterplot formed a triangle. Examination of these scatterplots is useful. The interpretation may be simplified by comparing quadrants of the scatterplot, thus allowing a comparison of four groups: high intelligence/­h igh creativity, high intelligence/­low creativity, low intelligence/­low creativity, and low intelligence/­h igh creativity. There is virtually no one in the low intelligence/­ high creativity quadrant of data, as predicted by the threshold theory. Support for the discriminant validity of DT meant that research on it was justified. If discriminant validity was lacking, there would be no reason to do the research. That would have meant that DT was dependent on intelligence, and thus all we would need to do was study intelligence, and we would understand DT. There are also educational implications of the support for the discriminant validity of DT tests. That is because an educator who holds the view that creative talent is dependent on general intelligence is likely to assume that students who display only a low level of intelligence are also uncreative. The educator may assume an interdependence that does not exist. This educator is likely to miss the creativity of many students, including those who are only moderately intelligent in an academic sense (­where convergent thinking and general intelligence are extremely important) but are also highly creative. Given the separation of creativity and DT from intelligence, there are students in this group, and their creative potentials should be recognized and supported.

Dimensions of Divergent Thinking Another form of discriminant validity must be checked with DT tests because of their dimensionality. More specifically, DT tests can be scored for several indices, including ideational 6

Divergent Thinking as Creative Cognition

originality, flexibility, and fluency. They are sometimes scored for elaboration, but this is rare compared to the other three indices. This second question of discriminant validity concerns the overlap or separation of the various indices. The most important index from DT tests is, no doubt, originality. That is because originality is a prerequisite for creativity, according to the standard definition of creativity (­reviewed by Runco  & Jaeger, 2012). This definition is standard in that it is prevalent in the theories of and research on creativity. Originality alone is not sufficient, however, and an idea that is only original does not qualify as creative. The second part of the standard definition is effectiveness. Just as originality may be called novelty or rarity, effectiveness may be called utility, fit, or usefulness. Debates about other criteria (­e.g., value, surprise, and aesthetic value) that may be involved in creative efforts and products can be found in Acar et al. (­2016), Martin and Wilson (­2017), Harrington (­2018), Kharkhurin (­2014), Simonton (­2012), and Weisberg (­2018). None of these is routinely used with DT tests; they are instead occasionally attached to creativity more broadly. DT tests are so often used because there are good methods for objectively determining the originality of ideas. The traditional approach involves collecting data from a sample (­a s representative as possible) and then organizing ideas into a lexicon. This can be done alphabetically, although that requires some treatment of the data (­e.g., articles such as “­the” removed from how ideas were presented by examinees). Identical or nearly identical ideas (­e.g., automobile and car) are combined, and the number of examinees who gave each idea is determined. If an idea was only given by one examinee, it is unique and earns maximal originality points. If an idea is rare but not unique (­e.g., given by ­1–​­2% of the sample), it earns originality points but not the same number of points as a unique idea. Ideas that are common (­e.g., given by a large portion of the sample) receive no originality points (­but do count toward an individual’s fluency score, which is just the number of ideas given, regardless of their originality or quality). The method scoring of originality, just described, may reflect subjectivity if humans are involved and asked to decide how the various ideas differ and which ideas are unique. This is not a huge problem because i­nter-​­rater assessments usually reflect perfect agreement more than 70% of the time, and often agreement is in the high 80% range. Still, humans need not play a role in scoring originality other than submitting the responses obtained from DT tests to a computer system. An algorithm can be used instead of a human to organize the ideas and determine, on a lexical level, which are unique and which are common. More will be said about computerized testing and the scoring of tests below. Obviously, sample size is an issue when uniqueness is used, given that it is easier to be unique in a small sample than it is in a large sample, but then again, the percentages used as cutoffs can be adjusted to take that into account. There is also the possibility that fluency scores confound originality scores (­Hocevar, 1979), but there are ways around this too (­e.g., regressing fluency on originality or simply using a ratio of originality to fluency, as in Runco et al., 1987). Importantly, when fluency is statistically controlled, the unique variance of originality is reliable, at least in some samples (­Dumas & Runco, 2018; Runco & Albert, 1985), which means that originality does have discriminant validity. This may also suggest something about creative cognition and the underlying processes for fluency, originality, and flexibility. The discriminant validity of the various indices has also been experimentally supported by giving explicit instructions for one index and then determining what impact those instructions have on the other indices. If originality is related to flexibility, for example, facilitating flexible ideation with explicit instructions should also increase originality (­Runco & Okuda, 1991). By and large, this is not the case. More will be said about explicit instructions below. 7

Mark A. Runco

Automated, algorithmic methods for scoring DT tests are now available (­Beketayev & Runco, 2016; Hass, 2020; Kenett et al., 2014), and these may actually replace the traditional methods. Before looking at those newer methods, something more should be said about ideational flexibility. This will help when we turn to the computerized scoring specifically for flexibility. Something should also be said about judges’ rating of ideas from DT tests. That will help when the outstanding objectivity of the computerized scoring is discussed below.

Ideational Flexibility Ideas from DT tests should be scored for flexibility. This will provide useful information and, like originality, is easy to connect to the processes that may be used when thinking creatively. Flexibility scores have the same kind of discriminant validity (­e.g., are correlated with but not dependent on fluency). Flexibility is defined in terms of the variety and diversity of ideas. Usually, conceptual categories are used to score flexibility. Thus, if an individual is asked the question, “­Name all the things you can think of that move on wheels,” and an individual responds “­tow truck, automobile, pickup truck, ­semi-​­truck, U ­ -​­haul truck, and race car,” all of these ideas are fitting but they all represent one conceptual (­vehicular) category. Another individual may respond to the same question with “­motorcycles, bicycles, skateboards, airplanes, drivers, wheelchairs, luggage, and trains.” This individual has a higher flexibility score because his or her ideas represent a larger range of conceptual categories. The algorithmic method described by Beketayev and Runco (­2016), cited above, provides a very useful flexibility score. More information is given in below. One recent investigation of flexibility illustrates how DT research has been used to study the neurochemical bases of creativity. Mastria, Agnoli, Zanon, Acar, Runco, and Corraza (­2021) proposed that “­E EG alpha synchronization, especially in posterior parietal cortical regions of the right hemisphere, is indicative of high internal processing demands that are typically involved in DT. During the course of DT, as ideation proceeds, ideas tend to become more creative, being more likely to be drawn from new conceptual categories through the use of the cognitive mechanism of flexibility.” Their empirical results confirmed that, in the lower alpha band (­­8–​­10 Hz), Whereas clustering showed synchronization typically lateralized in the right posterior parietal areas, switching induced posterior parietal synchronization over both right and left hemispheres. These findings indicate that the two distinct cognitive mechanisms subsuming flexibility (­switching and clustering) are associated with a different hemispheric modulation of lower alpha activity, as switching, in comparison to clustering, is related to higher power in the lower alpha band over the left hemisphere. What is especially relevant is that “­switching in comparison to clustering may thus require a larger investment of cognitive resources due to the exploratory process of moving from one semantic conceptual category to another in the course of creative ideation.” The idea of switching categories or staying within a c­ ategory – ​­something proposed as least as early as 1985 (­Runco, 1985) – ​­is too often overlooked when scoring DT tests. There are a number of reviews of the neuroscience of creativity (­e.g., Dietrich, 2015). Yoruk and Runco (­2014) reviewed all available research specifically on the neuroscience of DT, but that topic of research has grown quite a bit in the past few years. Flexibility may be used in a kind of profile, so each examinee has three scores: fluency, originality, and flexibility. This approach makes a great deal of sense, given (­a) the data 8

Divergent Thinking as Creative Cognition

available and (­b) the theory of DT (­Guilford, 1968; Mednick, 1962). Occasionally, a composite has been calculated using all DT indices (­Torrance, 1974), but this does not make much sense cognitively, and it raises questions about how to optimize the composite and weigh the three indices. Other times fluency alone is used, but this makes the least amount of sense, given the discriminant validity mentioned above and the fact that originality is a prerequisite for creative thinking and fluency is not. Admittedly, this view of originality applies most clearly to Western cultures and samples (­K harkhurin, 2014; Tan, 2016).

Other Scoring Methods The tests of DT are very rarely scored for creativity, and they are rarely scored for the effectiveness that is a part of the standard definition. One exception is the research of Runco and Charles (­1993). They asked judges to rate the creativity and the appropriateness or ideational pools. They chose appropriateness instead of effectiveness per se because the former is easier to operationalize with DT tests. Simplifying some, a DT question might ask examinees to “­list things that are square,” and highly appropriate responses will include things that are two dimensional with four equal sides, just to give one example. The ideational pools used in this investigation contained each examinee’s entire output. This is quite different from the typical method of examining and scoring individual ideas. There, each idea is evaluated to determine if it is original or not. One idea is scored, and then the next, and the next, and so on, and an individual’s originality score is some composite, perhaps a total or average of each of those scores for each individual idea. The method of ideational pools was based on the rationale that it might be useful to examine all ideas (­from any one examinee) all at once. This would provide more information about an individual examinee than the traditional method, which focuses on one idea at a time. The problem is that, when rating ideational pools, statistical infrequency cannot be used. Instead, originality or any other dimension of DT must be based on judgments from raters. This lowers the objectivity of the scoring, but it can be used to obtain (­judgmental) data about creativity per se and effectiveness, or at least appropriateness. The finding that stands out from the first study of ideational pools was that appropriateness ratings were inversely related to creativity ratings. This might suggest that judges do not take both originality and effectiveness into account when looking for creativity, but Runco and Charles (­1993) emphasized the definition of appropriateness used in the investigation. As described above, it was fairly constrained and may not have done justice to the various possible interpretations of “­effectiveness.” Judges may provide reliable ratings of DT, but it is a different kind of reliability than that obtained when more objective methods are used. A larger concern when judges are involved in scoring DT tests is that the ratings may not generalize to groups who are not involved in the research. In fact, the adequate ­inter-​­rater reliabilities reported when judges are involved may be misleading. They may indicate that there is agreement among the judges who are involved, but what of other judges, especially those who may be found in the natural environment? In one study where judges rated art portfolios, differences were found between students who offered ratings, the art teachers of the students, and professional artists (­Runco, 1989). The point is that differences among groups of judges should be expected; generalization is dubious when ratings from judges are used. Most important is that there is no real reason to use judgments at all when scoring DT tests. That is because there are objective alternatives that require little or no judgment. The traditional method, relying on lexicons and statistical infrequency, requires almost no human judgment, and lately, those algorithmic methods have demonstrated their usefulness (­Beketayev & Runco, 2016; Hass, 2020; Kenett, Anaki, & Faust, 2014). 9

Mark A. Runco

The method that was used by Kenett et al. (­2014) was deployed to r­ e-​­examine Mednick’s (­1962) associative theory. Kenett et al. wanted to test the idea that creative people have more flexible and detailed associative networks. Kenett et al. were also interested in the differences between individuals with flat versus steep associative networks. The former think with broad associations, and the latter with very few common associations. More broadly, the original theory of associative thinking held that “­creative individuals appear to have a richer and more flexible associative network than less creative individuals. Thus, creative individuals are characterized by “­flat” (­broader associations) instead of “­steep” (­few, common associations) associational hierarchies.” Kenett et al. examined these individual differences computationally. They described the “­core notion” of their innovative method as follows: “­concepts in the [associative] network are related to each other by their association ­correlations—​ ­overlap of similar associative responses (“­a ssociation clouds”).” They used decision trees to identify high and low creative groups and then compared them in terms of semantic memory networks. The findings confirmed that individuals in the low creativity group were more rigid and less flexible. There are obvious implications of this work for creative cognition. It suggests, for example, that flexibility is a characteristic of the creative process. An algorithmic method used with ideas from DT tasks also underscores the importance of flexibility. Beketayev and Runco (­2 016, ­p. 210) developed the ­semantics-​­based algorithmic (­SBA) method, which is fully automated. This means that individuals work on DT tasks on a mobile device or computer, and their responses (­the ideas) are immediately added to existing semantic norms and the semantic networks are immediately modified. Beketayev and Runco compared scores from the SBA method with the traditional DT score (­i.e., fluency, originality, and flexibility). Data were collected using the “­M any Uses Test” of DT. Results confirmed the key hypothesis, which was that the flexibility scores from the SBA would be correlated with the flexibility scores from the traditional method. Indeed, the correlation was moderately high .74. Beketayev and Runco suggested that the SBA method be used for DT testing, given that it is c­ ost-​­efficient and takes little time. Again, it is fully automated. This would be particularly helpful with large samples. Importantly, the SBA scores were not correlated with GPA, which means that they had good discriminant validity. Generalizations from this demonstration may be limited because only one DT test was used. There are several other computerized methods. A very good overview was presented by Hass (­2 020).

Concerns When Using Human Judges A bit more should be said about the topic of judges. That is especially true because what was already summarized involved interpersonal judgments, but there is also relevant work on intrapersonal judgments. This is not really a part of the scoring of DT tests but instead targets another component of the creative process. It complements the generation of ideas that is nicely assessed with DT tests (­see ­Figure 1.2). Research on intrapersonal judgments of ideas (­Runco, 1989; Runco & Vega, 1990) starts by collecting ideas with DT tests. In a second phase of the research, highly original ideas are identified. These are then put on a Likert scale. That also contains unoriginal ideas (­i.e., those that made up a sizeable portion of the original sample listed when they took the DT tests). That way, the new measure, used to assess intrapersonal judgments, has a range of ideas, both original and unoriginal. Intrapersonal judgmental accuracy can be evaluated by asking the original sample of respondents (­who took the initial DT tests) to rate the

10

Divergent Thinking as Creative Cognition

originality of the ideas found on the new Likert scale. Do these raters recognize an original idea when they see it? Even if it is their own idea (­from Phase 1 of the research)? Empirical results indicate that they often do not recognize an original idea when they see it. As a matter of fact, interpersonal judgments can also be assessed with this method. This is important because it means that teachers or parents, for example, can be asked to rate the originality of children’s ideas. This line of research has demonstrated that i­ntra-​­and interpersonal judgments of ideas can be reliably assessed. Results also indicate that people are not very good at judging their own ideas! In one investigation, participants recognized original ideas less than 50% of the time. There was a correlation between DT and intrapersonal judgmental accuracy, which makes sense in that a person who often thinks of an original idea probably has a good amount of practice judging originality. There was a negligible correlation between judgmental accuracy and a standardized measure of critical (­convergent) thinking, which implies that it is one thing to judge the originality of an idea but something else entirely to judge for correctness.

Cognitive Hyperspace The research on judgmental accuracy and the algorithmic scoring procedures, as just summarized, represent methodological advances, but there are also theoretical advances. Guilford’s (­1950, 1968) theory of DT is cited quite frequently, but even it has been extended. One extension describes the cognitive hyperspace that might be used when thinking divergently. This theory was partly a reaction to the observation that what is labeled DT may not be very divergent at all. Consider the fact that an individual can generate all kinds of ideas, many of which could be original, but the process may be more linear than divergent. Consider the associative description of an individual, when faced with an o ­ pen-​­ended problem, generating one idea or solution. That idea is then somehow connected to another idea. Explanations why certain ideas lead to other ideas have been proposed for a long ­t ime – ​­since before psychology was a ­science – ​­and include previous relationships among ideas, personal experience, functionality, and even acoustic similarity of the ideas. In any case, one idea leads to another, and then another. This chain of associations may lead to remote associates and original ideas, but there may not be much divergence along the way. The ideas could all be in the same conceptual direction, and the originality is more about the distance covered in the ideation than divergence per se. Acar and Runco (­2016) described an ideational process that is much more divergent. It begins with an individual producing one idea. The linear, n ­ on-​­divergent process may lead to another idea and another. At some point in the associative process, the individual may think of something that is conceptually very different from preceding ideas. In fact, it could be a dramatically different idea, as if the individual’s thinking has taken a perpendicular turn rather than continuing down the same conceptual path. This would reflect a bit of divergence. The individual may follow this new conceptual direction with a few ideas, but then could do the same thing and take another conceptually perpendicular turn in his or her thinking. This could happen again and again and again because, cognitively, humans are not limited in the same way as they are in the physical world, where there are only three dimensions. Cognitive hyperspace is possible, just as it is in mathematics and certain theories of astrophysics. Acar and Runco described this process, with a large number of dramatic turns from one conceptual category to a very different conceptual category, as a kind of hyperspace. They offered initial evidence for it, with one operationalization of it correlating with originality.

11

Mark A. Runco

Predictions from DT Test Scores There are probably at least as many studies of the predictive validity of DT tests as there are of their discriminant validity. Like the investigations of discriminant validity, there has been a kind of evolution rather than a smooth accumulation of findings. Many early studies of predictive validity offered only marginal support. Standing back, it seems that too much was expected of the DT tests. I say that because when correlations between DT test scores and certain criteria of actual creative behavior were reported in the .­30–​­.35 range, a number of researchers concluded that DT tests were not useful. What they were overlooking was (­a) there is a criterion problem in all studies of creativity and in DT research and (­b) the criteria used in the early studies were not really fitting. Those criteria reflected creative behavior in a broad sense (­e.g., accomplishment in some particular domain, such as art, math, or science), and such behavior surely involves more than ideation. DT tests provide data about ideation, but if the criterion reflects more than ideation, it makes little sense to expect high correlations. The criterion problem was explored in detail by Paek (­2020; Paek & Runco, 2018). Sometimes, a predictive validity correlation of .­30–​­.35 is impressive. Consider, for example, the 4­ 0-​­and 5­ 0-​­year ­follow-​­up reports from the Torrance longitudinal project (­Cramond et  al., 2005; Runco et  al., 2011). Torrance started collecting data with his own tests of creative thinking in 1958. He followed the same individuals (­initially school children) and collected ­follow-​­up data, usually at around ­ten-​­year intervals. ­Follow-​­up data were also collected after his death. The latest f­ollow-​­up reported correlations of slightly over .30, and these were based on the baseline data and then ­follow-​­up data collected 50 years later. In other words, DT test scores from testing around 1960 were predictive, at about .30, of certain creative activities, even 50 years later. Higher correlations have been reported. Runco (­1986), for example, reported a predictive validity coefficient of approximately .55. This is probably a reflection of the fact that he used canonical correlational methods, which allowed a predictive composite to be optimized so the association with the criteria was maximized. This composite contained not only the three typical DT scores (­fluency, originality, and flexibility) but also their interactions. Work in progress is replicating the 1986 study. Given concerns about the need for fitting criteria, a new criterion measure was developed. It focuses on ideation, which of course means that it is perfectly aligned with DT tests. The new criterion is a ­self-​­report that asks respondents how often they have thought about (­or generated ideas about) certain situations. An example is, “­how often have you had ideas for a new product or service?” Another is, “­how often have you had ideas about alternative titles for books or movies?” A number of studies have employed this ­self-​­report, and correlations with DT tests are indeed higher than those found when less fitting criteria are employed (­Runco, Plucker, & Lim, 2000). One investigation using the new ideational measure [known as the Runco Ideational Behavior Scale (­R IBS)] tested the possibility that it contains several factors. The hypothesis here was that it would be possible to separate ideation that reflects thinking about products from ideation that reflects process (­Chand O’Neal, Paek, & Runco, 2015). This distinction is well recognized in the creativity literature (­R hodes, 1961; Runco, 2007). Results indicated that product and process ideation could be separated in a ­ ne-​­factor solution was also meaning­two-​­factor solution from a confirmatory analysis. A o ful. Guilford’s (­1968) theory would predict more than one DT factor, but most research on DT assumes just one. Before turning to conclusions, one other investigation using DT as a predictor should be summarized. It is an example of how DT may be useful in a range of studies of creativity, 12

Divergent Thinking as Creative Cognition

including those involving health (­M raz  & Runco, 1994; Runco  & Mraz, 1993). It also demonstrated how DT is involved in more than just problem solving and can also be used to assess ­problem-​­finding skills. The investigation to be summarized used a measure of suicide ideation (­i.e., thinking about suicide) as a criterion. Predictors included both problem finding and problem solving, each scored for the traditional fluency, originality, and flexibility indices. Importantly, suicide ideation is not the same thing as a suicide attempt. Suicide ideation simply refers to thoughts about suicide, sans action. Many people think about suicide, but most take no action. Apparently, what is troubling is when an individual has suicide ideation but is also depressed and, worse yet, makes a plan. But suicide ideation is not in and of itself a clinical condition. It is subclinical yet worth studying. The general rationale was that the DT and suicide ideation both involve ideation. The unique aspect of this investigation was that interactions among ideational indicators were tested. To be precise, one measure of DT allowed examinees to generate problems. The idea here is that plenty of evidence points to some sort of problem finding as playing a role in some creative achievements, and it is possible to get at something very similar with ­open-​­ended tasks. These ­problem-​­generation DT tasks do not really get at problem finding; the examinee is given a situation and asked to make a list of problems. This sort of thing would be different from problem discovery or problem identification, where an individual actually finds new problems (­a nd nothing is presented). In the problem generation DT tasks, an examinee might be asked to list problems that are involved with relationships, or problems that are involved with a living situation, or problems at work. These situations are ­open-​­ended, fairly realistic, and they can be scored for originality, fluency, and flexibility. In the study of suicide ideation, examinees also received traditional DT tasks. These are more indicative of p­ roblem-​­solving ability and, of course, can be scored for the same fluency, originality, and flexibility. The most interesting finding was that an interaction of problem generation with problem solving was highly predictive of suicide ideation. As a matter of fact, the equation with the interaction terms was more accurate than depression in predicting suicidal ideation. This is fairly remarkable because depression is thought to be an accurate predictor of suicide ideation, but the interaction explained variance above and beyond that explained by depression. Importantly, the interaction reflected a particular set of indices from the DT tests. It was an interaction between high fluency when generating problems and low flexibility when solving problems. This of course makes sense because it implies that an individual perceives all kinds of problems but, at the same time, does not see many different options for solving those problems. Low flexibility can be viewed as a kind of cognitive rigidity, which is a negative correlate of creative cognition.

Conclusions This chapter reviewed the theory and empirical research on DT. Various connections to creative cognition (­e.g., rigidity versus flexibility, associative theory, and cognitive hyperspace) were made, as were ties to neuroscience and psychometrics. The psychometric studies offer reasonable support for DT tests. Still, it is best not to view DT tests as creativity tests. Indeed, there really is no such thing as a creativity test. That is because no test takes into account and assesses everything that is relevant to authentic, r­ eal-​­world creativity. Tests always focus on one or a very few parts of what has been called the creativity syndrome (­MacKinnon, 1965; Mumford  & Gustafson, 1988). Additionally, no test is perfect. This is why all tests must be examined in terms of their reliability and validity. These things tell us how much 13

Mark A. Runco

measurement error exists, and this is a way of quantifying the imperfections of a test. Even more to the point, all tests are samples. These things are reminders of the point made early in this chapter, namely, that DT tests are useful estimates of creative p­ otential – ​­but it is the potential to be creative and not creativity per se. The research summarized in this chapter confirms that the results from DT tests, if the tests are administered and scored correctly, provide useful information about one form of creative cognition. The term potential is vital because DT tests show what a person can do, at least in terms of ideational fluency, originality, and flexibility, but this does not guarantee what that person will do. Anyone interpreting DT test results must keep this in mind. Anyone administering DT tests may need to make a decision. To explain that, we can think back to Guilford’s (­1968) work and contrast it with that of Torrance (­1995). As Cramond (­1999) summarized it, Guilford was very interested in what people actually did. His interest was in spontaneous creative thinking that actually occurs in the natural environment. Torrance, on the other hand, was interested in what people could do if the conditions were right. For this reason, Torrance gave DT tests with directions to examinees that were like those mentioned earlier in this chapter from the research on explicit instructions. He came right out and asked the examinees to be creative. Guilford, on the other hand, often did not mention creativity in the directions because he felt that people in the natural environment do not receive such guidance. Research after Guilford and Torrance compared the two types of test administration with predictable results (­Harrington, 1975; Runco et al., 2005a, 2005b). When told to be original, for example, originality scores are high, but when merely asked to give as many ideas as they can, fluency is especially high and originality is lower than under the explicit testing conditions. This is a fairly brief treatment of the topic of DT. The key points about DT were covered herein, but the literature on DT is quite extensive (­Acar & Runco, 2019; Runco, 2020). Even in this brief review, it should be clear that DT theory has held up well over the years, though there have been revisions and extensions. A large amount of data have been collected. These point to individual differences, neuroanatomical bases, differences among particular DT tests and assessment conditions, scoring methods, including those involving a computer, and the usefulness of DT as a predictor. Theories and methods of DT have evolved nicely, but it is very likely that this evolution is not yet complete.

References Acar, S., & Runco, M. A. (­2015). Thinking in multiple directions: Hyperspace categories in divergent thinking. Psychology of Art, Creativity, and Aesthetics, 9, ­41–​­53. Bachelor, P., & Michael, W. B. (­1991). Higher order factors of creativity within Guilford’s ­structure-­​­­of-​ i­ ntellect model: A ­re-​­analysis of a ­fi fty-​­three variable data base. Creativity Research Journal, 4, ­157–​­175. Beketayev, K., & Runco, M. A. (­2016). Scoring divergent thinking tests with a s­ emantics-​­based algorithm. Europe’s Journal of Psychology, 12(­2), ­210–​­220. https://­doi.org/­10.5964/­ejop.v12i2.1127 Binet, A., & Simon, T. (­1905). The development of intelligence in children. L’Annee Psychologique, 11, ­163–​­191. Chand O’Neal, I., Runco, M. A., & Paek, S.-​­H. (­2015). Comparison of competing theories about ideation and creativity. Creativity: ­T heories-­​­­Research-​­Application, 2, ­145–​­165. Cramond, B. (­1993). The torrance tests of creative thinking: From design through establishment of predictive validity. In R. F. Subotnik & K. D. Arnold (­Eds.), Beyond Terman: Contemporary longitudinal studies of giftedness and talent (­­pp. ­229–​­254). Norwood, NJ: Ablex. Cramond, B. (­1999). The Torrance Tests of Creative Thinking: From design through establishment of predictive validity. In A. Fishkin, B. Cramond & P. ­Olszewski-​­Kubilius (­Eds.), Creativity in youth: Research and methods. Cresskill, NJ: Hampton Press. Deitrich, A. (­2015). How creativity happens in the brain. New York: Palgrave McMillan.

14

Divergent Thinking as Creative Cognition Dumas, D., & Runco, M. A. (­2018). Objectively scoring divergent thinking tests for originality: A ­re-​­a nalysis and extension. Creativity Research Journal, 30, ­466–​­468. Gerwig, A., Miroshnik, K., Forthmann, B., Benedek, M., Karwowski, M.,  & Holling, H. (­2021). ­ eta-​­analytic update. Journal of The relationship between intelligence and divergent t­ hinking—​­A m Intelligence, 9, ­1–​­28. https://­doi.org/­10.3390/­jintelligence9020023. Getzels, J. A., & Jackson, P. W. (­1962). Creativity and intelligence: Explorations with gifted students. New York: Wiley. Guilford, J. P. (­1950). Creativity. American Psychologist, 5, ­4 44–​­454. Guilford, J. P. (­1968). Intelligence, creativity, and their educational implications. San Diego, CA: EDITS/ ­Robert Knapp. Harrington, D. M. (­1975). Effects of explicit instructions to “­be creative” on the psychological meaning of divergent thinking test scores. Journal of Personality, 43, ­434–​­454. Harrington, D. M. (­2018). On the usefulness of “­value” in the definition of creativity: A commentary. Creativity Research Journal, 30(­1), ­118–​­121. DOI: 10.1080/­10400419.2018.1411432 Hass, R. (­2020). Computerized creativity testing and scoring. In M. A. Runco  & S. R. Pritzker (­Eds.), Encyclopedia of Creativity, 3rd edition. (­­pp.  ­94–​­99). Oxford: Elsevier. https://­doi. org/­10.1016/­­B978-­​­­0 -­​­­12-­​­­809324-​­5.­23810-​­8 Kenett, Y. N., Anaki, D., & Faust, M. (­2014). Investigating the structure of semantic networks in high and low creative persons. Frontiers of Human Neuroscience, 8, 407. Kharkhurin, A. V. (­2014). Creativity. 4in1: F ­ our-​­criterion construct of creativity. Creativity Research Journal, 26, 3­ 38–​­352. Kim, K.-​­H. (­2005). Can only intelligent people be creative? Journal of Advanced Academics. https://­ journals.sagepub.com/­doi/­10.4219/­­jsge-­​­­2005- ​­473 MacKinnon, D. (­1965). Personality and the realization of creative potential. American Psychologist, 20, ­273–​­281. Martin, L., & Wilson, N. (­2017). Defining creativity with discovery. Creativity Research Journal, 29(­4), ­417–​­425. DOI: 10.1080/­10400419.2017.1376543 Mastria, S., Agnoli, S., Zanon, M., Acar, S., Runco, M. A., & Corraza, G. E. (­2021). Clustering and switching in divergent thinking: Neurophysiological correlates underlying flexibility during idea generation. Neuropsychologia, 158, ­1–​­11. https://­doi.org/­10.1016/­j.neuropsychologia.2021.107890 Mednick, S. (­1962). The associative basis of the creative process. Psychological Review, 69, ­220–​­232. Meeker, M. N. (­1969). The structure of intellect: Its interpretation and uses. New York: Merrill. Mraz, W.,  & Runco, M. A. (­1994). Suicide ideation and creative problem solving. Suicide and Life Threatening Behavior, 24, ­38–​­47. Mumford, M. D., & Gustafson, S. B. (­1988). Creativity syndrome: Integration, application, and innovation. Psychological Bulletin, 103, 2­ 7–​­43. Paek, S.-​­H. (­2020). Criterion problem. In M. A. Runco & S. R. Pritzker (­Eds.), Encyclopedia of Creativity, 3rd edition (­­pp. ­281–​­285). Oxford: Elsevier. https://­doi.org/­10.1016/­­B978-­​­­0 -­​­­12-­​­­809324-​­5.­23809-​­1 Paek, S.-​­H., & Runco, M. A. (­2018). A latent profile analysis of the ­criterion-​­related validity of a divergent thinking test. Creativity Research Journal, 30, ­212–​­223. Plucker, J. A. (­2001). Introduction to the special issue: Commemorating Guilford’s 1950 presidential address. Creativity Research Journal, 13, 3­ –​­4, 247. https://­doi.org/­10.1207/­S15326934CRJ1334_02 Plucker, J. A., Meyer, M. S., & Liu, P. (­in press). Divergent thinking: Early views. In M. A. Runco & S. Acar (­Eds.), Handbook of creativity assessment. Northampton, MA: Elgar. Rhodes, M. (­1961). An analysis of creativity. Phi Delta Kappan, 42, ­305–​­310. Runco, M. A. (­1985). Reliability and convergent validity of ideational flexibility as a function of academic achievement. Perceptual and Motor Skills, 61, ­1075–​­1081. Runco, M. A. (­1989a). The creativity of children’s art. Child Study Journal, 19, ­177–​­189. Runco, M. A. (­1989b). Parents’ and teachers’ ratings of the creativity of children. Journal of Social Behavior and Personality, 4, ­73–​­83. Runco, M. A. (­Ed.) (­1991). Divergent thinking. Norwood, NJ: Ablex Publishing Corporation. Runco, M. A. (­2007). A hierarchical framework for the study of creativity. New Horizons in Education, 55(­3), ­1–​­9. Runco, M. A. (­2020). Divergent thinking. In M. A. Runco & S. R. Pritzker (­Eds.), Encyclopedia of creativity, 3rd edition (­­pp. ­351–​­356). Oxford: Elsevier. https://­doi.org/­10.1016/­­B978-­​­­0 -­​­­12-­​­­809324-​­5.­23824-​­8 Runco, M. A., & Acar, S. (­2019). Divergent thinking. In J. C. Kaufman & R. J. Sternberg (­Eds.), The Cambridge handbook of creativity (­­pp. ­224–​­253). New York: Cambridge University Press.

15

Mark A. Runco Runco, M. A., Alabbasi, A. M. A., & Paek, S.-​­H. (­2016). Which test of divergent thinking is best? Creativity: ­T heories-­​­­Research-​­Applications, 3, ­4 –​­18. Runco, M. A.,  & Albert, R. S. (­1985). The reliability and validity of ideational originality in the divergent thinking of academically gifted and nongifted children. Educational and Psychological Measurement, 45, ­483–​­501. Runco, M. A.,  & Albert, R. S. (­1986). The threshold hypothesis regarding creativity and intelligence: An empirical test with gifted and nongifted children. Creative Child and Adult Quarterly, 11, ­212–​­218. Runco, M. A., & Chand, I. (1995). Cognition and creativity. Educational Psychology Review, 7, 243–267. Runco, M. A., & Charles, R. (­1993). Judgments of originality and appropriateness as predictors of creativity. Personality and Individual Differences, 15, 5­ 37–​­546. Runco, M. A., & Jaeger, G. J. (­2012). The standard definition of creativity. Creativity Research Journal, 24(­1), ­92–​­96. DOI: 10.1080/­10400419.2012.650092 Runco, M. A., Illies, J. J., & Eisenman, R. (­2005a). Creativity, originality, and appropriateness: What do explicit instructions tell us about their relationships? Journal of Creative Behavior, 39, ­137–​­148. ­ eiter-​­Palmon, R. (­2005b). Explicit instructions to be creative and origRunco, M. A., Illies, J. J., & R inal: A comparison of strategies and criteria as targets with three types of divergent thinking tests. Korean Journal of Thinking and Problem Solving, 15, ­5 –​­15. Runco, M. A., Millar, G., Acar, S., & Cramond, B. (­2011). Torrance Tests of Creative Thinking as predictors of personal and public achievement: A 5­ 0-​­year ­follow-​­up. Creativity Research Journal, 22, ­361–​­368. Runco, M. A., & Okuda, S. M. (­1991). The instructional enhancement of the ideational originality and flexibility scores of divergent thinking tests. Applied Cognitive Psychology, 5, ­435–​­4 41. Runco, M. A., Okuda, S. M., & Thurston, B. J. (­1987). The psychometric properties of four systems for scoring divergent thinking tests. Journal of Psychoeducational Assessment, 5, ­149–​­156. Runco, M. A., Plucker, J. A., & Lim, W. (­2000). Development and psychometric integrity of a measure of ideational behavior. Creativity Research Journal, 13, ­393–​­400. Runco, M. A., & Vega, L. (­1990). Evaluating the creativity of children’s ideas. Journal of Social Behavior and Personality, 5, ­439–​­452. Simonton, D. K. (­1988). Creativity, leadership, and chance. In R. J. Sternberg (­Ed.), The nature of creativity: Contemporary psychological perspectives (­­pp. ­386–​­426). New York: Cambridge University Press. Simonton, D. K. (­2012). Taking the U.S. patent office criteria seriously: A quantitative ­three-​­criterion creativity definition and its implications. Creativity Research Journal, 24, ­96–​­107. Tan, C. (­2016). Creativity and confucius. Journal of Genius and Eminence, 1, ­79–​­84. Torrance, E. P. (­1974). The torrance tests of creative thinking. Bensenville, IL: Scholastic Testing Services. Torrance, E. P. (­1995). Why fly? Cresskill, NJ: Hampton Press. Wallach, M. A., & Kogan, N. (­1965). Modes of thinking in young children. New York: Holt Rinehart Winston. Wallach, M. A., & Wing, C. (­1969). The talented student. New York: Holt Rinehart & Winston. Weisberg, R. W. (­2018). Response to Harrington on the definition of creativity. Creativity Research Journal, 30(­4), ­461–​­465. DOI: 10.1080/­10400419.2018.1537386 Yoruk, S., & Runco, M. A. (­2014). The neuroscience of divergent thinking. Activitas Nervosa Superior, 56, ­1–​­16. https://­doi.org/­10.1007/­BF03379602, http://­rdcu.be/­A 5rL

16

2 MEASURING CREATIVITY WITH THE CONSENSUAL ASSESSMENT TECHNIQUE (­CAT) Karl K. Jeffries

Introduction On first encountering the study of creativity, many people are amazed that such a field exists. Even more remarkable is that such scholarship has considerable roots and can be traced back through over 60 years of dedicated focus within art, design, and psychology research. Thousands of research studies have explored links between creativity and other vital constructs: for example, educational attainment, skill development, mental health, employee motivation, and economic value, to name a few. The impact of such findings to inform government policy and economic development, as well as capture the attention of mass media, is demonstrable. Nevertheless, the quality of the research underpinnings is dependent on the credibility of the measures used, specifically, in these cases, in the measurement of creativity. While a popular response to the assessment of creativity can be incredulity, to the dedicated scholar of creativity, such understandable apprehension is at odds with theoretical and empirical findings within the field. For example, creativity researchers have focussed on one or more of the following aspects: the creative person, the creative process, the creative product, and the creative press (­i.e., the cultural and physical environment). Additionally, the influence of social recognition has been a longstanding debate within the field. Kaufman and Beghetto’s (­2009) “­Four C Model” of creativity (­­m ini-​­c, ­l ittle-​­c, ­Pro-​­c, and B ­ ig-​­c) highlights that studies have explored expressions of everyday creativity through to Nobel ­Prize-​­w inning achievements. Taken together, such theoretical models clarify the similarities and differences in approaches to the study of creativity: where some researchers have focussed upon the creative processes of psychology students, others have focussed upon the organisational environments of ­award-​­winning design agencies, and so on. Each approach has value, though the methods and theoretical assumptions about creativity in each context can be quite different. As a result, like many academic disciplines, there are schools of thought and preferences for some types of creativity assessment methods over others, and one method, used by both popular culture and academic study, has been the use of experts to assess creativity outputs. In its most basic form, the use of expert judges is found in various forms of entertainment (­from talent shows to professional competitions). In more formal settings, the consensus of domain judges remains a prevalent method for assessment in design education and industry DOI: 10.4324/9781003009351-3

17

Karl K. Jeffries

a­ wards – ​­of which creativity is either an implicit or explicit expectation. This chapter will be relevant to studies of creativity that require the assessment of a creative product within their research design, for example, studies that need to assess the creativity of graphic designs, musical scores, or engineering solutions and where judges will assess the creativity of those outputs.

Assessing Creativity with Domain Experts, and the Consensual Assessment Technique Since the 1950s, a considerable number of methods have been used to assess creativity. Of these, the consensus of domain expert judges has remained a common method for research studies. Design research, for ­instance – ​­even when focussed on the creative ­process – ​­has been premised on creative achievement among peers. MacKinnon’s (­1962) creativity research on architects; Roy’s (­1993) studies of product designers, like James Dyson; Lawson’s (­1994) studies of successful architects; or Cross’s (­2001) studies of outstanding designers are all examples of using expert judges to assess creativity. Many other fields of research have borrowed and developed methods and procedures, and one method, known as the Consensual Assessment Technique (­CAT), has become a prominent feature of creativity research. As a method, the CAT has been advocated as the “­gold standard” (­Baer & McKool, 2009) in creativity assessment. Notwithstanding debates about its precise implementation (­Cseh & Jeffries, 2019), it is considered to be a versatile and reliable measure of creativity. The CAT procedure was operationalised by Amabile (­1982), and since then, CAT usage has grown exponentially ( ­Jeffries, 2012). The CAT has been widely used across many disciplines and settings (­A mabile, 1979; Amabile, Goldfarb & Brackfleld, 1990; Baer, Kaufman & Gentile, 2004; Batey & Furnham, 2009; Christiaans & Venselaar, 2005; Hickey, 2001; Jeffries, 2017; Jung, Grazioplene, Caprihan, Chavez  & Haier, 2010; Kaufman, Baer, Cole,  & Sexton, 2008). For example, it has been used at all educational levels (­from kindergarten to higher education) and across numerous professions, both those traditionally associated with the creative industries and sciences and those in less apparent domains, such as the military (­McClary, 2009).

What is the CAT? Amabile (­1982, 1996) argued that the concept of what is and is not creative is largely shared among a domain’s experts as tacit knowledge and that creativity should therefore be assessed by consensus between domain experts. If a satisfactory agreement were reached, this would define the assessed product as creative, relative to the other products within a sample, within a particular context of time and place. This approach not only eschews and to some degree resolves the longstanding creativity definition debate (­for an example of this debate, see Runco & Jaeger, 2012), but simultaneously quantifies and operationalises creativity assessment for scientific research. By emphasising the subjective element in creativity assessment, the CAT contrasts with other measures of creativity, like the Torrance Test for Creative Thinking (­TTCT), for example. To appreciate this distinction, it is helpful to give some background history regarding the development of the CAT. Although a significant body of work had been completed on creativity, the majority of studies, before the 1980s, had focussed on the psychology of individual creativity. Amabile (­1982) argued that you could not truly understand creativity without taking the social context of creativity into account: relationships with others, 18

Measuring Creativity with the Consensual Assessment Technique

particular environments, and externally imposed working constraints. When Amabile began her enquiry into the social dimension of creativity, the research design was a concern: many of the accepted research methods were not ideal for social studies on creativity. There were several reasons for this. The TTCT (­Torrance, 1974) and other divergent thinking tests are about quantifying those aspects that set people apart; they aim to define and quantify the micro and macro factors that distinguish one individual from another. In contrast, researching the social context of creativity requires the need to define and quantify group characteristics beneficial for comparing a control group against a test group. Thus, there was a need to minimise individual differences and test hypotheses about how one group may react differently to another, given changes in their social environment. The limitations of previous research methodologies meant a revised method to assess creativity needed to be evolved and proven to be valid. This context was the original basis to develop the CAT. Suppose we accept the standard definition of creativity (­Runco & Jaeger, 2012) that creativity produces work that is both new (­originality) and useful (­effectiveness). In that case, we could state that our initial criteria for assessing creativity could centre on assessing how new and useful the final product is. From such a position, we are left with several questions: what are the appropriate assessment criteria for new and useful? How should assessors evaluate new and useful, and how do they do this with transparency and objectivity. Amabile (­1982) argued that objective methods did not currently exist (­a nd may never exist) on which to assess creativity in this way. Moreover, the judgements required to assess creativity “….can ultimately only be subjective” (­­p. 1001). And as such, with an appropriate group of judges, “….is something that people can recognise when they see it” (­­p. 1001). From this came an operational definition of creativity, upon which the CAT is based. That is, ….A product or response is creative to the extent that appropriate observers independently agree it is creative. Appropriate observers are those familiar with the domain in which the product was created or the response articulated. (­­p. 1001) By basing the criteria on the judges’ subjective opinions, this also negates the need for explicit criteria. As long as the judges agree, then that is enough; they may not be able to articulate precisely why a product has a certain level of creativity, but if they independently agree that it does, then this shall form the basis for evaluation. It is worth highlighting that, for these reasons, the CAT is considered to be a more theoretically neutral measure of creativity; theoretically neutral in that the researcher is not defining what those criteria may be. Judges are free to use whatever tacit or implicit criteria they may have evolved regarding creativity within their domain. In this respect, the CAT is quite distinct not only from the TTCT but also from other measures of creativity that define the criteria judges should apply when rating creativity, for example, novelty, surprise, or usefulness. This theoretical neutrality is one reason creativity researchers have advocated the CAT as a “­Gold Standard” of creativity assessment (­K aufman, Plucker & Baer, 2008). The initial CAT was also premised on Amabile’s guidance, which included that judges should: • • •

Be experts in their domain; Rate creativity independently (­i.e., without training or specific criteria); Rate creativity relative to and within a specific sample and context; 19

Karl K. Jeffries

•

Be asked to rate factors other than overall creativity (­e.g., technical execution and aesthetic appeal) to ensure discriminant validity of the creativity measure.

While these guidelines are specific to a degree, they also leave room for interpretation and expansion. Using this working definition, the CAT moved away from the notion of objectivity in creativity assessment and towards subjectivity. Given such a radical departure from assessment norms at the time, research studies were developed to evaluate whether such levels of agreement existed and to what degree they were reliable and consistent. Over five years, Amabile and her colleagues conducted eight studies using the CAT (­A mabile, 1996). With a wide range of groups represented from primary, secondary, and undergraduate education, the total number of students engaged in the research was 423. These groups either took part in a study to assess visual or verbal creativity. A range of assessors recruited from academia, working practice, and education judged this work. One hundred and ­t wenty-​­five judges participated in the research; each judge was free to use his or her subjective definition of creativity with which to assess the work. From these studies, it was concluded that high levels of judge agreement existed on the creativity rating. Furthermore, judges were able to distinguish “­creativity” from other aspects of the work, such as aesthetic appeal and technical execution. With this validation in place, the CAT was used to research the social impact on creativity through several experimental studies (­e.g., Amabile, 1979; Amabile, Hennessey, & Grossman, 1986; Hennessey, Amabile, & Martinage, 1989; Amabile, Goldfarb, & Brackfleld, 1990). As the CAT research method developed, creativity researchers (­Baer, Kaufman, & Gentile, 2004) extended the CAT to less stringent experimental conditions than were initially utilised. Since then, many other researchers have adapted the CAT for their studies, and the method has continued to be applied to a wide variety of different domains.

The Problem The CAT is unusual as a measure within creativity research, in part because it is no longer “­owned” by anyone. While the CAT can be attributed to Amabile’s earlier work, the use of domain experts for assessing creative outputs began before the CAT. Through numerous other researchers, the CAT has significantly evolved since its introduction in the early 1980s. In this way, the CAT and its methodology have become owned by a community of scholars and researchers, with no one individual having a veto over its development. This relatively egalitarian feature of the CAT is both a strength and a source of challenge. Likewise, a further strength of the CAT is its (­seemingly) simple method and its adaptability to a wide variety of domains. It is precisely this deceptive simplicity and adaptability, however, that are also its weaknesses and barriers to future advancement. As with any scientific tool, methodological consistency is paramount for integrity, replication, and comparison across findings. Problematically, the way Amabile’s broad guidelines are interpreted and implemented in practice by different researchers shows a wide variation (­a s will be explored later in this chapter). Additional guidance ­exists – ​­Kaufman, Plucker and Baer dedicated a complete chapter of their book Essentials of creativity assessment to the CAT, as did Hennessey, Amabile, and Mueller (­2011) for the Encyclopaedia of creativity – ​­but there remains substantial variation across research studies concerning the CAT method. With the continued growth of the CAT in creativity research, from an international perspective, such growth has a negative impact when the method is interpreted in different ways, thus

20

Measuring Creativity with the Consensual Assessment Technique

changing the reliability and validity of research findings. Options for ­cross-​­domain comparison, ultimately, become obscured or limited. For the field to develop, it urgently needs to harmonise the protocols that creativity researchers have used to date. As other creativity researchers have acknowledged, over the past two decades, discussions across a variety of publications have occurred without much progress or room for c­ ompromise – ​­a seemingly simple matter of guidance on the research protocol, but one that is reaching an impasse without a strategic change. One resolution is to accept that such an endeavour has gone beyond the singular efforts of any given scholar. The sections to follow detail further about the challenges outlined above and are present in the CAT literature. The chapter then proposes a solution to bridge these differences by adopting an innovative approach to explore new ideas around international harmonisation for the CAT. While researchers await such guidance, the final section of the chapter offers a checklist of five key issues for researchers to consider with recommendations on practical implementation.

Procedural Differences and Distinctions At its core, the CAT has four fundamental components: a task, a set of creative outputs, several judges, and a rating protocol. Together, these form a system, and each component can influence the others. In this respect, nuanced relationships likely exist that are still to be fully understood. Notwithstanding, for the moment, dissection is a useful approach to understand the challenges of individual aspects in isolation. As a procedure, the CAT can be defined in three broad phases. The first phase is to gather a set of outputs that respond to a task (­e.g., write a poem about nature). In doing so, the researcher has both defined the task and acquired a range of creative outputs. Challenges can occur due to the complexity of the task and the variation in quality among the outputs. For example, it is possible to set a complex task and gather a range of outputs that are of much the same quality. Alternately, a task can be easily understood, and the range of outputs is highly diverse in quality. In terms of judging these scenarios, it is reasonable to consider that the latter is more straightforward than the former. This difference is likely to impact the level of consensus among judges. Phase two requires rating the outputs with suitable judges. A single judge would rate individual outputs on creativity, and the process is repeated with several other judges. Challenges at this stage relate to how many judges are required, the number of creative outputs each judge will rate, the wording of the instructions given to judges, and even the rating scale used; each of these can have an impact on the CAT’s validity as a measure of creativity. The third phase is to analyse judges’ ratings and compute the level of consensus. If consensus is adequate, the creative outputs can be arranged on a spectrum from lower to h ­ igher-​­scoring outputs. While statistical tests, like Cronbach’s alpha, have been common to compute consensus for the CAT, there are alternative tests and, likewise, a substantial debate about their merits. As individual researchers, it is tempting to give answers to the questions and challenges posed at each of these three phases. Indeed, often it is required. To undertake a study using the CAT, one must make decisions on implementation that either directly or indirectly determine the methodological procedure and thus impact the underpinning theoretical position. Practical choices are required and can be made with or without full theoretical awareness. In this regard, the cited studies in this chapter are not meant to single out individual researchers or to suggest good or poor practice. They serve merely as examples of the variety of methodological considerations facing all researchers.

21

Karl K. Jeffries

The subsections to follow will explore specific examples of these variations, namely: the rating scale; the number of judges; the number of items rated; the instructions given to judges; and the issue of relative assessments. Such a list is not meant to be exhaustive. Cseh and Jeffries’s (­2019) review of the CAT raised further considerations, each of importance. Those selected for this chapter aim to cover a range of issues, from those amenable to relatively practical compromise to those that are more theoretically challenging. As will be argued towards the end of this chapter, compromise around these issues is required. There must be compromise between the individual researcher and the development of the domain, between fieldwork practicalities and theoretical ideals. If this compromise were straightforward, international harmonisation of the CAT would have already occurred; it has not and requires a change of strategy. What that strategy entails will be explored later in this chapter.

Rating Scale Within the CAT literature, a wide range of rating scales have been used. Most range between a ­three-​­and ­ten-​­point scale. For example, Kwon, Bromback, and Kudrowitz (­2017) used a ­three-​­point scale and noted that, as the judges were required to rate hundreds of ideas, the use of a ­three-​­point scale simplified this task. Jauk, Benedek, Dunst, and Neubauer (­2013) used a ­four-​­point scale in their study of creativity and intelligence. Kaufman, Evans, and Baer (­2010) used a ­five-​­point scale, as did Sosik, Kahai, and Avolio (­1998). Kaufman, Baer, Cole, and Sexton (­2008) and Jeffries (­2017) each used ­six-​­point scales, though they asked participants to categorise items into three categories (­low, medium, and high creativity) first. Daly, Seifert, Yilmaz, and Gonzalez (­2016) used a s­even-​­point scale. Harvey (­2013) used a ­n ine-​­point scale. Yuan and Lee (­2014) used a ­ten-​­point scale, as did Christiaans (­2002). The question to consider is: why is such a wide variation in rating scales required? This could be related to each research study’s design. Researcher preferences are possible, as is a theoretical stance on ­Likert-​­type rating scales. Much debate, for example, has occurred about the pros and cons of ­even-​­versus ­odd-​­numbered scales (­K rosnick & Presser, 2010). Preston and Colman (­2000) consider the granularity of a rating scale is optimal when between five and seven points.

Number of Judges The number of judges used varies greatly from study to study. For example, Daly et  al. (­2016) asked two judges to rate the creativity of 439 designs. Valgeirsdottir, Onarheim, and Gabrielsen (­2015), in contrast, gathered the assessments of 134 general public judges to rate the creativity of two mobile phone products. Hickey’s (­2001) study used five different expertise groups to rate children’s musical compositions. The number of judges ranged from 24 ­second-​­grade children to three composers. To mitigate this difference in group size, three judges per group were used to calculate ­inter-​­rater reliability. The impact of doing so considerably reduced i­nter-​­rater agreement in several of the groups. Given this, is it possible, or desirable, to standardise the number of judges required for a CAT study? Interpretation by researchers on the number of judges required ranges from over 100 to 2, with every number in between. Differing research conditions and variables partly explain this variation, but not completely. Silvia et al.’s (­2008) perspective is nicely summed up as “­One is clearly not enough; 20 seems like overkill” (­­p. 81). Kaufman, Plucker, and Baer (­2008) have suggested that “­for most purposes, ­5 –​­10 judges is an adequate number” (­­p. 74). 22

Measuring Creativity with the Consensual Assessment Technique

Number of Items and Rater Fatigue While the rating scale and number of judges each have their complexities, rater fatigue appears to highlight a greater challenge for the harmonisation of CAT procedures, namely, how many items can a judge assess at one time? The issue is complex, but two themes are considered here: first is the sheer quantity of items; second is the complexity of those items. Given the research literature within cognitive psychology about choice overload and cognitive load (­M iller, 1956; Scheibehenne, Greifeneder, & Todd, 2010), judges will fatigue at some point. In this situation, CAT reliability and validity could be significantly different when a judge is rating only a few items versus being asked to rate a pool of 100+ data. CAT good practice to randomise the order of items shown to judges is likely to mitigate the influence of fatigue to an extent but not entirely. In the CAT literature, the number of items each judge rates at a time shows a substantial range, from single figures to hundreds and possibly thousands. For example, Karwowski et al. (­2016, Study 8) reused the data gathered from Jauk et al. (­2013); both studies explored the relationship between creativity and intelligence. In the original dataset, four student judges rated each participant’s response on a ­four-​­point scale using a method “­similar to the consensual assessment technique proposed by Amabile, 1982” (­K arwowski et al., 2016, ­p. 7). In terms of the number of items rated, each judge would appear to have assessed 297 participants’ creativity outputs with an average of 12 responses in total, namely, 3,564 ratings per judge. In terms of complexity, the type of domain being studied (­e.g., reading 100 short stories versus reviewing 100 photographs) is a factor in rater fatigue. A judge’s expertise and familiarity with the task will also play a part, but there is likely to be a direct link here with time. For example, artist judges (­A mabile, 1982, Study 1) who spent four hours rating artworks resulted in lower consensus levels than judges who only spent half an hour rating works.

Instructions to Judges In its practical application, CAT guidance states that when creativity is assessed in a new domain (­i.e., one that has not been studied with a particular task), researchers should ask judges to rate additional constructs. Most commonly, technical execution and aesthetic appeal are used (­A mabile, 1982; Hennessey, 1994) to check whether creativity ratings are distinct from these criteria for the sake of construct validity. Indeed, in Amabile’s (­1982) work, the extent to which creativity may be isolated from such factors was a formative part of her paper. Hennessey’s (­1994) discussion of the CAT, however, acknowledged that for some domains, the distinction between technical execution and aesthetic appeal might be less transparent, and creativity is likely to correlate with these aspects. This is the case: Jeffries, Zamenopoulos, and Green (­2018), for example, detail high correlations between technical execution and aesthetic appeal on a graphic design task. The broader implication is that when rating creativity, certain domains are sensitive to instructions around technical execution and aesthetic appeal. As can be seen within the CAT literature, some researchers have created instructions that directly ask judges to discount technical execution and aesthetic appeal from their creativity rating (­Baer, 1993). Other researchers have instructed judges to rate creativity alongside technical execution and aesthetic appeal (­A mabile, 1982; Christiaans & Venselaar, 2005; Valgeirsdottir, Onarheim, & Gabrielsen, 2015). Some only do this the first time they undertake a new CAT task (­Hennessey, 1994; Kaufman, Plucker,  & Baer, 2008); some do not. Thus, instructions to 23

Karl K. Jeffries

judges on technical execution and aesthetic appeal range from explicit through to implied, implied only once, or not mentioned at all.

Relative Assessments Of the issues highlighted in this chapter, the most challenging are those raised by relative assessments. These highlight a tension between practical implementation and theoretical purity: between how researchers may want to use and adapt the CAT for a study and the limitations and restrictions suggested by its theoretical underpinnings. Amabile’s (­1996) guidelines stated that the CAT should be conducted in a relativistic fashion (­i.e., comparing items within a sample to one another) rather than on an absolute or broader criterion. For example, the difference between comparing a university art student’s work to other university art students’ work from the same study sample as opposed to comparing it to the works of Pablo Picasso or Frida Kahlo. From a theoretical perspective, this distinction would appear to be core to what a CAT is: a relative form of creativity assessment within a given sample. The CAT aims to quantify the subjective and tacit knowledge of a set of judges for a sample of works within the context of a specific moment in time and culture. Due to the “­fuzzy” nature of such an assessment, the CAT protocol attempts to balance procedural rigour while minimising the theoretical bias that may result from doing so. It does this by controlling confounding variables that impact judges’ assessments. Put another way, if judges subjectively know creativity when they “­see it”, part of that recognition is the relationship to the other works they are assessing within a sample (­e.g., this poem is more creative than this poem). Recognition of creativity can also be formed by a judge’s knowledge of previous creative outputs based on a similar task. It can also be formed by how they themselves would have undertaken the task. In these two latter scenarios, comparison outside of the sample may, in most cases, set a more challenging benchmark for creativity. The CAT protocol, for this reason, attempts to harmonise the basis for comparison by instructing judges to assess works relative to the other works they are given. That is, to reduce making comparisons outside of the sample. Doing so implies that some works will be relatively higher or lower in creativity than each other, and this is reflected by judges using the full range of the rating scale. On this basis, fruitful data can be gathered to evaluate the level of ­inter-​­rater agreement. As an aspect that could be considered vital to the CAT procedure, it is not necessarily noted in methodology reports how judges are instructed to compare the works and if they are explicitly told to make relative judgements within the sample of the presented works.

The Need for International Harmonisation Each of the five issues above highlights examples of the variability in how researchers have interpreted the CAT protocol. As stated previously, while a degree of variability is understandable and to be expected, when variability is too broad, this is a problem. The CAT has offered creativity researchers a high degree of adaptability, applicable across a broad range of domains; however, wide variability impacts the reliability and validity of research findings and limits options for comparison and replication. For the field to develop, it needs to harmonise the CAT protocols that creativity researchers have used to date. Thankfully, all the knowledge and expertise are available to achieve this. These circumstances have not occurred overnight, nor are they fleeting; they are the result of respective researchers pushing at the boundaries of their disciplines for decades. What appears to be missing is a fresh strategy to find compromise. The section to follow outlines a solution to bridge these differences 24

Measuring Creativity with the Consensual Assessment Technique

in variation by adopting an innovative approach to international harmonisation for the CAT. For the purposes of this chapter, this initiative will be described as the CAT Accord.

The CAT Accord Ultimately, harmonising protocols requires engaging the creativity research community to explore what harmonisation means for both practice and theory. A crucial step would be the creation of a green paper and policy advice document on assessing creativity with experts to offer the creativity research community a series of recommendations on harmonisation. The purpose of an accord would be to stimulate international debate and focus discussion on the merits of compromise and agreement. What marks this approach as different from the discussion and debate that has occurred to date would be the intention to explore compromise around specific issues. To do this, the CAT Accord would gather a mediated panel of 12 researchers for a t­ wo-​ ­d ay ­m icro-​­symposium. It is worth stating at this point that, given the importance of creativity to many domains, it is possible to study creativity and, yet, for a researcher to remain siloed within their primary discipline, be this art and design, business, education, or psychology. Indeed, given the pressures of academic silos, it is quite reasonable for scholars to hold to these identities, even when they have become, possibly without knowing it, creativity researchers. As a result, where research methods have been developed and widely used by some branches of the international creativity research community, for a variety of reasons, similar methods have been developed in isolation and outside of interaction. It would be necessary for the CAT Accord to gather as broad a representation as possible, within the constraints of the ­12-​­person panel. To support the panel, a sustained period of community engagement would be required during the six months leading up to the mediated ­m icro-​­symposium. The purpose of this would be to explore creativity assessment research practice and stimulate debate among creativity researchers and other stakeholders to influence the priorities and agenda for the CAT Accord. To mediate a successful discussion and compromise among the panel, a professional mediator would be required to facilitate the group. This is a crucial distinction from the previous debate to date. Arguably, mediation, and the techniques it has developed, is different from the more conventional use of a group facilitator, or chair. As this will be the fi ­ rst-​­time a discussion on opportunities for international harmonisation will have taken place, it is prudent to give the CAT Accord the full benefits that mediation offers. In doing so, this is a novel approach to a ­m icro-​­symposium (­though one that is becoming more sought in business contexts). The use of mediation strategies would enable the panel to explore discussion in a conducive environment, with a twofold benefit. Mediation strategies themselves aim to mitigate conflict from the outset. Where areas of contentious debate may arise, a professional mediator is ideally placed to navigate the panel towards resolution and insight. Both of these features are suited to the aims of the CAT Accord. Additionally, if harmonisation cannot be achieved on specific topics, then the CAT Accord would be the catalyst for more focussed research and study that aims to bridge a specific divide. It is through the aspiration for international harmonisation that such areas of further study can reveal themselves and their importance. The CAT Accord could offer a framework to house other methods for assessing creativity with experts and begin to define a taxonomy of methods that are distinct from each other. Through mediation and goodwill, should a substantial compromise be possible, the achievement of the CAT Accord would mark a significant step towards international 25

Karl K. Jeffries

harmonisation of the CAT. As a tangible and considered set of suggestions, it may resolve matters of variation and establish guidance acceptable to the broader creativity research community. This would be an ideal outcome. The likelihood is that its publication would foster further debate. Not all will agree with the guidance, but hopefully all can see the value of an approach towards compromise and convergence and the best efforts of the panel and our respective peers.

Practical Advice for Researchers Using the CAT While researchers await such guidance, what can be done now to move towards international harmonisation? The final section of the chapter offers a checklist of five key issues for researchers to consider with recommendations on implementation. Each builds on the issues identified in this chapter. As argued above, the ideal basis for recommendations would be a collective response through the development of the CAT Accord. What follows here, therefore, is intended as an informed estimate of where the CAT Accord may be able to find a compromise. With the best of intentions, it is offered as guidance to those new to using the CAT and, for established practitioners, to consider to what extent they could align their current practice towards these recommendations. It also serves to offer questions for researchers to ask themselves to evaluate the quality of the CAT method within a study. For example, consider the merits of a study that has two judges to one that has eight, 100s of items to rate or less than 100, uses a ­three-​­point scale in contrast to a ­ten-​­point scale, gives no opportunity to make relative assessments, and one that directly asks judges to make relative assessments. This is not to shame or call out such practices; there can be research design justification for such choices. What is essential is to understand the rationale and rigour behind those choices.

Rating Scale Regardless of the number of judges, items, or instructions to judges, future CAT studies are likely to require a rating scale. How granular should a CAT rating scale be? Too few rating points within a scale and consensus may be easier to achieve; too many rating points and ratings may become a more challenging task for the judges to complete and impact consensus. As a basis for guidance, Preston and Colman (­2000) have suggested an optimal scale is likely to be between five and seven points. Likewise, Krosnick and Presser (­2010) highlight the debate between ­even-​­versus ­odd-​­numbered scales. The pros and cons of each revolve around whether the CAT should force respondents to decide if work tends towards lower or higher creativity as opposed to allowing neutral or middling creativity ratings. As a position for compromise, the suggestion is for the CAT rating scale to be either five or six. A step further would be that, as the CAT requires judges to make distinctions between lower and higher levels of creativity within a sample, an argument can be made that a neutral position is somewhat counterproductive to the CAT’s purpose: to force distinctions and check for consensus under these conditions. If that argument finds merit, then for international harmonisation, future CAT studies would apply a ­six-​­point scale.

Creativity Only Instruction to Judges Like rating scale harmonisation, the instructions given to judges are not contingent on the number of items or judges. Each future study that uses the CAT, like past CAT studies, will have research design considerations that shape the instructions given to judges. These 26

Measuring Creativity with the Consensual Assessment Technique

are probably specific to that study and should remain so. At a certain point, however, as the purpose of the CAT is to assess creativity, then instructions to rate creativity will occur. How these instructions are worded and phrased would seem beneficial to international harmonisation. As wording already exists within the CAT literature on instructing judges, the building blocks for harmonisation are in place. Options tend to distinguish themselves in how they respond to the issue of confounding factors, notably technical execution and aesthetic appeal. CAT research has shown that some domains are more sensitive to the influence of technical execution and aesthetic appeal than others (­Christiaans, 2002; Jeffries, Zamenopoulos, & Green, 2018; Wojtczuk & Bonnardel, 2012). Given this, the argument presented here is to select CAT instructions that can accommodate sensitive domains. The rationale is that by establishing a baseline at this level, then all other (­less sensitive) domains may be accommodated. In this respect, Baer’s (­1993) wording (­adapted by Jeffries, 2017) explicitly instructs judges only to use creativity as the criterion to assess the works. This form of wording asks judges to tacitly reduce the background “­noise” in creativity rating that is present from technical execution and aesthetic appeal. Isolating creativity may be a challenge, but the evidence is that judges can achieve this. What is required is that the instruction make this clear to do so. The wording would be as follows: There is only one criterion in rating these works: creativity. We realise that creativity probably overlaps other criteria one might consider ( ­for example, aesthetic appeal, or technical execution) but we ask you to rate the works solely on the basis of their creativity. For future CAT studies, the word “­works” may not fit all circumstances. In this case, “­works” could be substituted for a word more appropriate for the domain (­a rtwork, poems, stories, plans).

Number of Judges The CAT’s purpose is to gather consensus on creativity for a given set of works. The question then is, how many judges are required to achieve an appropriate level of consensus? To an extent, this will depend on the statistical measure of consensus used (­the standard has been the use of Cronbach’s alpha, with ­inter-​­rater reliability at or above .7). On this basis, Kaufman, Plucker, and Baer’s (­2008) suggestion was that between five and ten judges are adequate in most circumstances. This seems like sensible guidance to reiterate and emphasise for international harmonisation. Whatever statistical test will be used to measure CAT consensus, future researchers will not want to inflate consensus by having too many judges, nor undermine consensus with too few. Five to ten judges offer a reasonable level of practicality and rigour for recruiting domain expert judges. If a more defined threshold is feasible for international harmonisation, then opting for either seven or eight domain expert judges (­a nd splitting the difference) offers a more straightforward message.

Number of Items Specific guidance on the number of items to be rated is a considerable challenge for international harmonisation, especially when research design and statistical power require a large number of items. This is an issue interconnected with the complexity of the task being 27

Karl K. Jeffries

judged, the total amount of time required, and the nature of the assessment (­i.e., is this assessment relative to the other works in the sample or not?). Simply put, fatigue will begin when the time and effort are too great. Moreover, this poses several ethical considerations about what is reasonable to ask of judge participants for a study. With these points foremost, perhaps some thresholds are more reasonable than others. As argued above, this is a debate best mediated among peers through the CAT Accord. As a basis for guidance, it seems reasonable to suggest time limitations for each judge’s assessment. An upper limit of 90 minutes would define how many items could be rated. Depending on the complexity of the task, 90 minutes may still place a considerable burden on judges, and a balance between rater fatigue and the research design requirement needs to be found.

Relative Assessment The relative assessment of creativity can be argued to be core to CAT theory. It traces back to the CAT as a method to explore the social influences on creativity (­e.g., comparing a control group to an experimental group). However, no known CAT studies have directly examined how many items judges can effectively compare to one another at a time in terms of relative creativity. In this respect, relative assessment is linked to the number of items and rater fatigue issues discussed above. Relative assessment is, understandably, a divisive issue for the CAT protocol. For studies that require 100s of items to be rated, it is difficult to see how relative assessment can be implemented and how rater fatigue can be addressed. At present, a simple solution to bridge this theoretical and practical difference is elusive. The guidance below, then, is related to future CAT studies that use relative assessment. These are based on an adapted version of Kaufman, Baer et al.’s (­2008) study (­cited as an exemplar of CAT instructions in Kaufman, Plucker, & Baer, 2008), with the addition by Jeffries (­2017) highlighted in bold type. They offer a form of wording that emphasises relative assessment. The instruction is worded as follows: There is no need to explain or defend your ratings in any way; we ask only that you use your own sense of which is more or less creative (­relative to the other works provided). The sentence above is only part of the instruction offered by Kaufman, Plucker, and Baer (­2008) and Jeffries (­2017). The full set of instructions offers procedures for judges that further reinforce relative assessment. However, in the spirit of compromise towards international harmonisation, these have not been included in the guidance above.

Conclusion Creativity research offers academia and industry alike a means to reflect on creative practice. To find new, unexpected associations and to challenge myths that surround and undermine creativity in the variety of forms it takes. For many, the arguments are well established: the ability to adapt, innovate, and respond creatively to our future will be even more critical in the face of global technological, economic, and ecological challenges. Creativity research has much to offer this future, primarily when based on credible findings and exemplary research design. To maintain this position, the methods and tools of research need to be refined, updated, and developed. This chapter has explored the measurement of creativity through the use of the CAT. The purpose was twofold: first, to offer a substantial introduction to those unfamiliar with 28

Measuring Creativity with the Consensual Assessment Technique

the CAT; and second, to spark the needed debate and more ­in-​­depth investigation among the community of past, current, and especially future CAT researchers about best practices for the CAT’s implementation. As illustrated in this chapter, several issues have not been fully considered and settled about the CAT procedure to date. This has resulted in various CAT and C ­ AT-​­like procedures and, as a result, some inconsistent outcomes. Such examples highlight the necessity for creativity researchers worldwide to harmonise the consistency and transparency of CAT procedures going forward. International harmonisation is fraught and, to date, has been elusive. What is required is a fresh strategy to find compromise, which is described in this chapter as the CAT Accord. The CAT Accord would create the conditions for a professionally mediated panel of researchers to address the issues highlighted here. The aim is to offer the creativity research community a series of recommendations on international harmonisation. While researchers await such guidance, this chapter offers a set of recommendations on the future use of the CAT. Applied in its entirety, the guidance is summarised as follows. A panel of seven or eight domain experts would rate creative works on a ­six-​­point scale. Individual judges would complete their assessment of creativity within 90 minutes. Each judge would be instructed as follows: There is only one criterion in rating these works: creativity. We realise that creativity probably overlaps other criteria one might consider ( ­for example, aesthetic ­appeal, or technical execution) but we ask you to rate the works solely on the basis of their creativity. There is no need to explain or defend your ratings in any way; we ask only that you use your own sense of which is more or less creative (­relative to the other works provided).

References Amabile, T. M. (­1979). Effects of external evaluation on artistic creativity. Journal of Personality and Social Psychology, 37, ­221–​­233. Amabile, T. M. (­1982). Social psychology of creativity: A consensual assessment technique. Journal of Personality and Social Psychology, 43(­5), ­997–​­1013. Amabile, T. M. (­1996). Creativity in context: Update to the social psychology of creativity. Boulder, CO: Westview Press. Amabile, T. M., Goldfarb, P., & Brackfleld, S. C. (­1990). Social influences on creativity: Evaluation, coaction, and surveillance. Creativity Research Journal, 3, ­6 –​­21. Amabile, T. M., Hennessey, B. A.,  & Grossman, B. S. (­1986). Social influences on creativity: The effects of ­contracted-​­for reward. Journal of Personality and Social Psychology, 50, ­14–​­23. Baer, J. (­1993). Creativity and divergent thinking: A ­task-​­specific approach. Hillsdale, NJ: Lawrence Erlbaum Associates, Inc. Baer, J., Kaufman, J. C., & Gentile, C. A. (­2004). Extension of the consensual assessment technique to nonparallel creative products. Creativity Research Journal, 16(­1), 1­ 13–​­117. Baer, J., & McKool, S. S. (­2009). Assessing creativity using the consensual assessment technique. In C. Schreiner (­Ed.), Handbook of Research on Assessment Technologies, Methods, and Applications in Higher Education (­­pp. ­65–​­77). Hershey, NY: IGI Global. Batey, M., & Furnham, A. (­2009). The relationship between creativity, schizotypy and intelligence. Individual Differences Research, 7(­4), ­272–​­284. Christiaans, H. (­2002). Creativity as a design criterion. Creativity Research Journal, 14, ­41–​­54 Christiaans, H. H. C. M., & Venselaar, K. (­2005). Creativity in design engineering and the role of knowledge: Modelling the expert. International Journal of Technology and Design Education, 15(­3), 2­ 17–​­236. Cross, N. (­2001). Strategic knowledge exercised by outstanding designers. In J. Gero & K. Hori (­Eds.), Strategic knowledge and concept formation III (­­pp. ­17–​­30). Sydney: University of Sydney.

29

Karl K. Jeffries Cseh, G. M., & Jeffries, K. K. (­2019). A scattered CAT: A critical evaluation of the consensual assessment technique for creativity research. Psychology of Aesthetics, Creativity, and the Arts, 13(­2), ­159–​­166. Daly, S. R., Seifert, C. M., Yilmaz, S., & Gonzalez, R. (­2016). Comparing ideation techniques for beginning designers. Journal of Mechanical Design, 138(­10), ­101–​­108. Harvey, S. (­2013). A different perspective: The multiple effects of deep level diversity on group creativity. Journal of Experimental Social Psychology, 49(­5), ­822–​­832. Hennessey, B. A. (­1994). The consensual assessment technique: An examination of the relationship between ratings of product and process creativity. Creativity Research Journal, 7(­2), ­193–​­208. Hennessey, B. A., Amabile, T. M., & Martinage, M. (­1989). Immunizing children against the negative effects of reward. Contemporary Educational Psychology, 14, ­212–​­227. Hennessey, B. A., Amabile, T. M., & Mueller, J. S. (­2011). Consensual assessment. In M. A. Runco & S. R. Pritzker (­Eds.), Encyclopedia of creativity (­2nd ed., Vol. 1., ­pp. ­253–​­260). San Diego, CA: Academic Press. Hickey, M. (­2001). An application of Amabile’s consensual assessment technique for rating the creativity of children’s musical compositions. Journal of Research in Music Education, 49(­3), ­234–​­244. Jauk, E., Benedek, M., Dunst, B., & Neubauer, A. C. (­2013). The relationship between intelligence and creativity: New support for the threshold hypothesis by means of empirical breakpoint detection. Intelligence, 41(­4), ­212–​­221. Jeffries, K. K. (­2012). Amabile‘­s consensual assessment technique: Why has it not been used more in design creativity research? Proceedings of the 2nd International Conference on Design Creativity (­ICDC2012), 1, ­211–​­220. Jeffries, K. K. (­2017). A CAT with caveats: Is the consensual assessment technique a reliable measure of graphic design creativity? International Journal of Design Creativity and Innovation, 5(­­1–​­2), ­16–​­28. Jeffries, K. K., Zamenopoulos, T., & Green, A. J. K. (­2018). Design creativity, technical execution and aesthetic appeal: A CAT with caveats (­Part 2). International Journal of Design Creativity and Innovation, 6(­­1–​­2), ­66–​­79. Jung, R. E., Grazioplene, R., Caprihan, A., Chavez, R. S., & Haier, R. J. (­2010). White matter integrity, creativity, and psychopathology: disentangling constructs with diffusion tensor imaging. PloS One, 5(­3), e9818. Karwowski, M., Dul, J., Gralewski, J., Jauk, E., Jankowska, D. M., Gajda, A.,  …  & Benedek, M. (­2016). Is creativity without intelligence possible? A necessary condition analysis. Intelligence, 57, ­105–​­117. Kaufman, J. C., Baer, J., Cole, J. C., & Sexton, J. D. (­2008). A comparison of expert and nonexpert raters using the consensual assessment technique. Creativity Research Journal, 20(­2), ­171–​­178. Kaufman, J. C.,  & Beghetto, R. A. (­2009). Beyond big and little: The four c model of creativity. Review of General Psychology, 13(­1), 1. Kaufman, J. C., Evans, M. L., & Baer, J. (­2010). The American idol effect: Are students good judges of their creativity across domains? Empirical Studies of the Arts, 28(­1), ­3 –​­17. Kaufman, J. C., Plucker, J. A., & Baer, J. (­2008). Essentials of creativity assessment (­Vol. 53). New York, NY: John Wiley & Sons. Krosnick, J. A., & Presser, S. (­2010). Question and questionnaire design. In P. V. Marsden & J. D. Wright (­Eds.), Handbook of survey research (­2nd ed., ­pp. ­263–​­313). Bingley: Emerald Group Publishing Ltd. Kwon, J., Bromback, L.,  & Kudrowitz, B. (­2017). Divergent thinking ability + interest = creative ideas: Exploring the relationship between cognitive creativity assessments and product design idea generation. In Proceedings of the 29th International Conference on Design Theory and Methodology (­DTM) (­­pp. ­1–​­5). Cleveland, OH: ASME. Lawson, B. (­1994). Design in mind. Architectural Press. MacKinnon, D. W. (­1962). The nature and nurture of creative talent. American Psychologist, 17(­7 ), 484. McClary, R. B. (­2009). An investigation into the relationship between tolerance of ambiguity and creativity among military officers. Unpublished Doctoral Dissertation. Kansas State University, Manhattan, KS. Miller, G. A. (­1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(­2), ­81–​­97. Preston, C. C.,  & Colman, A. M. (­2000). Optimal number of response categories in rating scales: Reliability, validity, discriminating power, and respondent preferences. Acta Psychologica, 104(­1), ­1–​­15.

30

Measuring Creativity with the Consensual Assessment Technique Roy, R. (­1993). Case studies of creativity in innovative product development. Design Studies, 14(­4), ­423–​­4 43. Runco, M. A., & Jaeger, G. J. (­2012). The standard definition of creativity. Creativity Research Journal, 24(­1), ­92–​­96. Scheibehenne, B., Greifeneder, R., & Todd, P. M. (­2010). Can there ever be too many options? A ­meta-​­analytic review of choice overload. Journal of Consumer Research, 37(­3), ­409–​­425. Silvia, P. J., Winterstein, B. P., Willse, J. T., Barona, C. M., Cram, J. T., Hess, K. I., … & Richard, C. A. (­2008). Assessing creativity with divergent thinking tasks: Exploring the reliability and validity of new subjective scoring methods. Psychology of Aesthetics, Creativity, and the Arts, 2(­2), ­68–​­85. Sosik, J. J., Kahai, S. S., & Avolio, B. J. (­1998). Transformational leadership and dimensions of creativity: Motivating idea generation in c­ omputer-​­mediated groups. Creativity Research Journal, 11(­2), ­111–​­121. Torrance, E. P. (­1974). Torrance tests for creativity: Norms and technical manual. Bensenville, IL: Scholastic Testing Service, Inc. Valgeirsdottir, D., Onarheim, B., & Gabrielsen, G. (­2015). Product creativity assessment of innovations: Considering the creative process. International Journal of Design Creativity and Innovation, 3(­2), ­95–​­106. Wojtczuk, A., & Bonnardel, N. (­2012). Differences in creative design assessment. Proceedings of the 2nd International Conference on Design Creativity (­ICDC2012), Glasgow, United Kingdom. Yuan, X., & Lee, J.-​­H. (­2014). A quantitative approach for assessment of creativity in product design. Advanced Engineering Informatics, 28(­4), ­528–​­541.

31

3 CONSTRAINTS AND CREATIVITY Classifying, Balancing, and Managing Constraints Balder Onarheim and Dagný Valgeirsdóttir Constraints as Fundamental to Creativity The idea that creativity has the best conditions when free from constraints is persistent, but as argued by Elster (­2000), in human life, there is no such thing as an unconstrained situation. Elster argues that daydreaming is the closest we can get to a c­ onstraint-​­free environment. In wakefulness, humans will always be under a certain level of constraint, whether arising from our cognition, knowledge, ambitions, and others’ expectations or wishes, or from more formal constraints like cost and time, to fundamental constraints such as a material constraint or simply natural forces like gravity. Imagine the most unconstrained creative context possible, and there will still be a myriad of constraints present. Creativity, in practice, will always exist in an environment of constraints (­or “­press”), shaping the creative person and process and the requirements of what the creative product must be. Moreover, contradictory to common belief, constraints are not the enemy of c­ reativity –​ ­they can also be enablers. Famously, Google was guided by the slogan “­Creativity loves constraints”, and history is full of powerful examples of highly creative outcomes from situations where creativity has been severely constrained, either by external factors, as in the Apollo 13 rescue mission, or by famous prison escapes, such as the escape from the supposedly inescapable Alcatraz in 1962, or by strict ­self-​­imposed constraints, such as the nine rules of the Dogme 95 film tradition, or the conscious avoidance of the letter “­E” in the ­300-​­page novel La Disparition by French author Georges Perec. In creativity research, as in other related disciplines, the notion of constraints, restraints, criteria, press, and rules has been a recurring topic. Guilford (­1950) briefly addressed the important relation between restraints and creativity, asserting that creative work must be performed under some degree of evaluative restraint and that too much restraint can be “­fatal to the birth of new ideas” (­­p. 453), but creativity can only be assessed with the relevant restraints. Since the early mentions of restraints in the 1950s, there have been a number of different terms and conceptualizations of the rules or limits on creativity, but in the last two decades, the term “­constraints” has emerged as the most common umbrella term to describe any type of “­restriction on freedom” for creativity. A crucial distinction should be emphasized regarding research on constraints and research on “­barriers” to creativity, which focuses mainly on factors that are shown to have a strong 32

DOI: 10.4324/9781003009351-4

Constraints and Creativity

limiting effect on creative performance (­e.g., Davis, 2011), such as cultural and perceptual barriers, fear, habits, and certain types of personality. “­Person constraints” have also been discussed: personality traits can create such strong barriers that cognitive ­creativity  – ​­the ability to think in novel w ­ ays – ​­becomes “­useless” in itself unless one constrains oneself with risk willingness, to fight for unconventional ideas, or to intrinsically motivate oneself to enable the development of useful application of those novel ideas (­Sternberg & Kaufman, 2010). Constraints can furthermore be understood in relation to the proposed “­standard definition” of creativity, with creativity being defined as the combination of novelty and usefulness (­Runco & Jaeger, 2012; Stein, 1953). The definition implies the necessity of constraints for creativity, as both the “­novelty” and the “­usefulness” requirements are constraints in themselves. Moreover, as stated by Sternberg and Kaufman, “­what makes a person or a product creative is the flair of originality constrained by usefulness, and the benefit of usefulness constrained by originality” (­Sternberg & Kaufman, 2010, ­p. 481). Finally, both are dependent on constraints in what it means to be novel and what it means to be useful, and what can be limiting to those. As such, constraints are inherent to the definition of creativity itself.

Early Development within Creativity Research While the word constraint might have negative ­connotations – ​­as a limitation or ­restriction –​ ­in the academic field, it has long since been dispelled as a myth that constraints are negative for creativity, and within the research referenced here, the term is used for factors that can both enhance or limit creativity. Since the address by Guilford, constraints, restraints, and requirements have been mentioned by key researchers in relation to definitions of theories and frameworks. One such example is Mednick’s (­1962) usage of “­requirements” in his landmark Associative Theory of the Creative Process. Another example is where Amabile (­1982) discusses the necessity of defining specific “­criteria” implicit in an operationalized definition of creativity. Furthermore, constraints play a crucial role within the four Ps of creativity (­R hodes, 1961): Person, Process, Product, and Press, where the Press can be understood as the total sum of constraints. This is just to point out a few known examples from creativity research where constraints are implied. Nevertheless, as constraints are a part of the construct of creativity itself, a number of authors in the second half of the 19th century repeated that, without restraint, there can be no creativity. Still, the empirical efforts for understanding the nature of the relationship were limited during the period. However, after the turn of the 21st century, research efforts gradually increased, and more studies started to focus specifically on constraints and their relationship with creativity. Stokes (­2007, 2009) investigated the importance of actively using constraints as a way of fostering creativity. Joyce (­2009) pioneered work on both clearer conceptualizations as well as contributing with studies of the impact of concrete constraints such as time and cost and the notion of “­both too much and too little constraint being negative for creativity”. A definition of constraints when related to creativity was proposed by Onarheim (­2012b, ­p. 16): “­Constraints are explicit or tacit factors governing what the creative actor/­s must, should, can and cannot do; and what the creative output must, should, can and cannot be”. These contributions were part of a growing tradition that emerged in the field with a second wave of more conceptual work on types and distinctions of constraints, with a number of experimental studies investigating types of constraints in relation to their effect on creativity. An increase in PhD projects since the turn of the century on the subject (­e.g., Biskjaer, 2013; 33

Balder Onarheim and Dagný Valgeirsdóttir

Joyce, 2009; Onarheim, 2012b) is further indicative of a continuous increase in the interest and ­in-​­depth study of the relationship between constraints and creativity. The overarching research themes seem to be: (­1) classification of constraints (­understanding and analyzing constraints on creativity); (­2) balancing constraints (­balancing between being ­over-​­or ­under-​­constrained); and (­3) constraint management (­developing approaches to strategically use constraints to enhance creativity). These three themes will be the main topics of this chapter.

Classification of Constraints As research interest in constraints and creativity began to increase, a number of researchers started developing general approaches to analyze, group, and understand constraints. When looking across these various conceptualizations, it can be overwhelming, as multiple terms are used for seemingly similar, and at times overlapping, aspects of constraints: clusters, categories, types, distinctions, continuums, and so forth. Here, we will propose “­Classes of Constraints”, a h ­ igh-​­level model for classifying constraints. The classification is helpful as a way to organize the current conceptualizations from constraint research and for analyzing and working with constraints for creative practitioners. The classification is, among others, inspired by the t­hree-​­dimensional model suggested by Lawson (­2004), and consists of four classes representing different elements inherent to constraints: (­1) Origin: Where does the constraint itself originate from; (­2) Target: What or who is the constraint targeting or specifically constraining; (­3) Type: What is the constraint about; and (­4) Nature: How should the constraint itself be interpreted. While none of the four classes is absolute or exhaustive, they can function as a way of organizing and understanding the relationship between various previous works that, at times, can otherwise seem contradictory, and for operationalizing constraints in creative work. The Classes of Constraints can be seen as a grammar of constraints and can be directly compared to the concept of grammar in linguistic sentence analysis, classifying words using terms such as subjective, objective, adjective, and the like (­­Figure 3.1). To operationalize the classification, a simplified example will be used: in a global engineering company, an engineering team has been provided with a design brief, requesting the team to propose an updated design for the company’s b­ est-​­selling product. In the brief, one of the design specifications is that the manufacturing division has requested that the main material (­PVC plastic) in the product remains the same due to the existing investments in manufacturing equipment and worker skills. While this specific, simplified example is taken from engineering, the Classes of Constraints model should apply equally to other domains, from law to arts.

Origin The first of the four classes concerns the Origin of the constraint, that is, where is the constraint originating from, or what is the source of the constraint? For instance, a constraint that is imposed on a creative product from an external source, for example, a design specification from the manufacturing division requesting a given material to be used in the ­product – ​­the manufacturing division would be the Origin. Studies and distinctions such as those of Amabile (1996) and Stokes (2005), focusing on who or where the constraint came from, would be included in the Origin class. In the literature, there is normally a general distinction made between “­internal and external” constraints, but there are inconsistencies in how “­internal” is defined, and it depends on where the line is drawn. Stokes drew it at 34

Constraints and Creativity

CONSTRAINT Type Origin E.g. Laws of physics Legal regulations Organisational Tradition Team Person Cognition

E.g. Time Material Weight Color Manufacturing

Nature E.g. Absolute Suggestive Unclear Flexible Anticulation lmportance

Target E.g. Country Organisation Team Person Product Process

­Figure 3.1 Classes of constraints visualized

the person, which is internal to the creative person or external from outside the person; with respect to the example given at the beginning of this paragraph, the constraint would be classified as external (­Stokes, 2005). However, Cromwell and Amabile (­2017) view it from an organizational perspective and therefore define the organization, including the creative person as internal and the outside of the organization as external, meaning that the constraint from the example would be classified as internal. Elster (­2000) generalizes the internal/­external distinction by defining the limit as the “­context” – ​­did the constraint arise from the creative person (­­self-​­imposed) or outside of the context (­imposed). This “­internal/­external” distinction of the origin of constraints can be directly translated to most creative contexts: constraints from within the actor (­­self-​­imposed constraints, like ambitions) or from the actors’ immediate surroundings (­imposed constraints; e.g., from the social context, organization, and domain). Thus, according to Elster’s distinction, the example constraints from the design specifications would be classified as an imposed constraint. From this, it becomes apparent that, although all of the distinctions and nuances outlined above are useful and important, looking across them, there can be overlapping conceptualizations. Therefore, one can take a step back and look at the common denominator as Origin, as a distinctive aspect with any constraint: where does it originate from?

Target Independent from the origin of a constraint, the Target of constraints is concerned with what the constraints are aimed at, or who/­what they are generally ­constraining  – ​­for instance, constraints on the creative person, or constraints on the product itself. In the above example, a constraint that is imposed on a creative product from an external source would be the design specification from the manufacturing division defining a given material to be used in the product. In this example, the manufacturing division would be the Origin and the product would be the Target. 35

Balder Onarheim and Dagný Valgeirsdóttir

The literature describing a distinction between constraints on the process versus those regarding the product would belong to the Target level. These typically focus on the difference between constraints on the creative process (­e.g., time, skills, tools, and deadlines) and the creative product (­e.g., sales cost, material, size, and style) (­Onarheim, 2012a). Acar, Tarakci, and van Knippenberg (­2019) use these clusters of constraints and add a third one, that of input constraints, which contains any constraints being set when initiating the creative process. This is again related to the type of creative process, if it is a p­ roblem-​­solving process or a Geneplore process (­Cromwell & Amabile, 2017) – because if there is no clearly ​­ defined problem, specific input constraints might not be in place. Another, slightly different tripartition is offered by Sternberg and Kaufman (­2010), where they differentiate constraints by three of the 4Ps of creativity (­a s defined by Rhodes, 1961): the constraints present for the Person, the Product, and the Process. Still, none of these ways of distinguishing between the Target for constraints is definite, and in practice, it can sometimes be difficult to clearly define in which of the above categories a given constraint “­belongs”. But generalized, the Target of constraints is always something specific, such as being the person or team, their process, or the product they are working toward.

Type Unrelated to which class is used to distinguish constraints in relation to their Origin or Target, any constraint will be of a given Type. Here, Type is used to describe specifically what the constraint is about, so in addition to the Target itself (­e.g., the product or the process), the constraint is about “­something” for the Target, such as the cost, the time, the resources, the cognition, the function, the material, the weight, and the like. Looking back to the example, a constraint that is imposed on a creative product from an external source, such as design specifications (­the constraints) from the manufacturing division, defining a given material to be used in the product, would have the manufacturing division as the Origin, the product as the Target, and the material as the Type of constraint. The Type class of constraints is important in relation to empirical work, where there has been significant research done on different types of constraints, for example, time constraints (­Baer & Oldham, 2006). Additional research efforts on types of constraints include, but are not limited to, the influence of lighting on constraints (­Steidle & Werth, 2013), the influence of financial constraints (­Scopelliti et al., 2014), the influence of applying certain format constraints on writing processes (­Haught, 2015), and the influence of various task constraints (­Doms & Weiss, 2017; Rosso, 2014) and material constraints (­Noguchi, 1999). In line with the argumentation for the classes of Origin and Target, here it will be argued that in order to declutter the various different forms of constraints that have been studied, the investigations can all be deemed to relate in essence to some Type of constraint. Thus, by classifying them all as Type constraints, the core and aim of these studies become clearer.

Nature In addition to the Origin (­where is the constraint from), the Target (­who or what is the constraint aimed at), and the Type (­exactly what is the constraint about), it can be helpful to understand constraints in relation to the Nature, or the properties, of the constraint itself. Going back to the example, a constraint that is imposed on a creative product from an external source, such as design specifications from the manufacturing division, defining a given 36

Constraints and Creativity

material to be used in the product, would have the manufacturing division as the Origin, the product as the Target, the material as the Type, and the format in which it is given (­formally written in a specification document) as part of its Nature. Furthermore, the constraint could be either flexible or absolute, or important or unimportant. These are examples of what the literature discusses as “­continua”: a constraint is not necessarily either formalized or informal, but it has some degree of formalization, thus the constraint exists on a point of the continuum of flexibility. At the same time, the constraint can also exist on other continua, such as degree of flexibility and degree of importance. The Nature of a constraint is defined by these different continua, and any constraint can be understood as simultaneously existing on numerous different continua regarding the properties of the constraint itself. Another example could be a cost constraint (­in the Type class) that is either informal in Nature, as informal advice by a manager, or highly formalized in Nature, as a specifically formulated restriction on a spending account. These two, the flexibility and the formalization of the constraint, are examples of what have been discussed by Onarheim and Wiltschnig (­2010; see also Onarheim, 2012b) as continua for constraints, which represent ways of grading a given constraint in terms of its nature between two extremes on a continuum. While these continua are not exhaustive, there are certain types of continua that are reoccurring in the constraints and creativity literature and summarized in Onarheim (­2012b) as: • • • • •

​­ Articulation (­or formalization)  – The extent to which the constraint is articulated and/­or formulated. Abstraction (­or “­hardness”)  – ​­The level of details provided in the description of the constraint that also defines to what extent the fulfillment of the constraint is measurable. ­ omplexity – ​­The level of difficulty in complying with the constraint. C F ­ lexibility – ​­The extent to which the constraint is fixed in its current form or can be adapted. I­ mportance – ​­The importance of taking the constraint into account and related consequences of not complying with the constraint.

For some of these continua, like “­Importance”, in practice, they are rather considered extremes (­need to have versus nice to have) or segmented into categories (­e.g., must have, should have, and could have). Furthermore, certain ends of such continua have been addressed as distinctive and particularity important for creative work, such as: • • •

Primary generators: Highly important and complex constraints (­Darke, 1979). Crucial constraints: Highly complex constraints selected by the actors as particularly crucial to the process (­Onarheim, 2012a). Tacit constraints: Constraints unconsciously held by the participants (­Onarheim, 2012b).

In the context of our example, where the manufacturing division has defined a given material to be used in the final product, the Nature of the constraint would be understood as whether the constraint is flexible (­can the designers challenge it?), formalized (­has it been exactly defined in the written design brief?), and at a level of abstraction (­does it describe a general type of material or a highly specific subtype of material to use?). Since it is only a “­request” and not, for instance, a requirement, it can be assumed that the constraint is negotiable, but as it relates to financial aspects of the production, it is still a crucial constraint. Considering the complexity and diversity of current ways to understand and categorize constraints as outlined above, future creativity and constraint research would benefit from a 37

Balder Onarheim and Dagný Valgeirsdóttir

shared ­h igh-​­level classification of constraints. By broadening the perspective and looking at the different current categorizations through the lens of their core commonalities rather than contradictions, the above Classes of Constraints provides a way to organize, understand, and compare current contributions.

Balancing Constraints in Creativity Having established that neither creativity nor human life can exist without constraints, the key consideration in creativity research is how these constraints influence creativity. The question of whether constraints “­are good or bad for creativity” is null and void, as constraints in themselves can be both, and it is rather a question of “­how much” or “­the right amount” of constraints for fostering creativity. Any given constraint can potentially have both negative and positive impacts on creativity, if too much or l­ittle – ​­so the key notion becomes the question of “­balancing constraints” to find the right (­number of ) ­constraints –​ ­because when having the right constraints, to the right degree, they become enablers for creativity. For instance, a study by Doms and Weiss (­2017) found that adaptive individuals, while often being related to lower levels of creativity, could indeed generate enhanced creative output, but only when exposed to higher levels of constraints. Studies focusing on “­balancing constraints” represent a noticeable theme that has emerged in the last decade in terms of how constraints are studied in creativity ­research – ​­recognizing the duality of constraints in terms of when they act as inhibitors and when they are enhancers. Notable studies into how and when in the creative process constraints are beneficial, as well as when they are not, have been conducted by Haught-Tromp (2017), Lombardo and Kvålshaugen (2014), Onarheim (2012a, 2012b), Onarheim and Biskjaer (2013), Rosso (2014, 2017), and Stokes (2007, 2009, 2014), among others. The consensus seems to be that the relationship between creativity and constraints is both dual and dynamic, and that it differs when and where in the process there are too many or too few constraints and what effect they have on creativity. However, when, how, and to what extent they can and should be balanced still remains somewhat of an enigma.

Constrainedness The concept of constrainedness in relation to creativity was first developed by Onarheim (­2012b). Constrainedness is an expression for the total sum of constraints at any given point in a creative process, seeking to capture the total picture of the constraints on creativity. It is not an exact or measurable concept, but a theoretical concept that provides a way to describe all constraints rather than each constraint by itself. It is also a dynamic concept, as each constraint making up the total constrainedness can change at any point throughout a process, subsequently influencing the constrainedness. Building on the notion of constrainedness, Onarheim and Biskjaer (­2013) introduced “­The Sweet Spot Model of Creativity”, operationalizing the concept together with “­­under-​­ and ­over-​­constrained” in a conceptual model. The model describes the relationship with the creative potential of individuals as shaped like a normal distribution “­bell curve”, with the top of the bell curve being the sweet spot, that is, when the creative person is using their creative potential at its highest (­­Figure 3.2). When a person feels ­under-​­constrained, with too few constraints, their perceived creative potential is lower; their creative potential is also affected with too many constraints. The model does not address a potential relationship between the perceived creative potential 38

Constraints and Creativity

­Figure 3.2 The sweet spot model of creativity

and the actual creative production, but other researchers have found similar patterns for the value of creative production. For instance, Joyce (­2009), who tested three levels of time constraints, found the highest creative value in the m ­ id-​­case scenario.

Dynamic Nature of Constraints In addition to holding the potential to have both a positive and a negative influence on creativity, constraints are not necessarily constant as a function of time, as any constraint might change over time as the creative process progresses. For instance, certain constraints, such as a cost constraint in a product development process, are not always fixed and can be changed during a project due to external reasons such as organizational budget ­decisions – ​­or requests from the design team. Requirements change, timelines change, and people and ambitions shift; thus, the total constrainedness that creativity is subject to is dynamic. Certain constraints cannot be defined at the beginning of a creative process and will more often emerge toward the end of creative processes, while other ­constraints – ​­like gravity and material ­constraints – ​­are constant in most creative processes. Still, while not fixed, different circumstances can have a general higher or lower sum of constraints, taking freestyle poetry and nanoparticle engineering as two radically different examples of constrainedness. Thus, constraints, or the sum of constraints and subsequently level of constrainedness, can only be understood as a momentary concept that is influenced by the circumstances and the creative actor and will change throughout any creative process. It is important to note a distinction between “­t ime” and “­t iming” when it comes to constraints. Time constraints are a Type constraint, referring to the specific amount of time allocated to a task. The notion of timing of constraints, however, relates more to the balancing of constraints perspective, namely when a specific constraint is introduced or discovered in a process. As an example, what are the effects of introducing late constraints (­Nature constraint) and not introducing all constraints from the beginning of a creative process? One study found that it could actually be beneficial to productivity in a creative task to receive some constraints later on in the process (­Onarheim & Valgeirsdottir, 2017). Medeiros et al. (­2017) have found that the timing of constraints is “­everything” in creative problem solving and 39

Balder Onarheim and Dagný Valgeirsdóttir

actually propose the opposite, namely that having all constraints prior to problem identification proved beneficial to creative performance. Thus, constraints are also reoccurring in various studies of problem solving, both related to ­ill-​­structured and ­well-​­structured problems (­how well defined is the constraint when initiating a p­ roblem-​­solving process?) and to the notion of “­problem and solution spaces” (­Dorst & Cross, 2001). In the latter, constraints can be understood as the factors that define the conceptual space of both the problem and the solution, and being the factors influencing the dynamic shifts in a problem/­solution space ­co-​­evolution (­Wiltschnig et al., 2013).

Constraint Management In this section, an overview will be provided in terms of the current understanding of constraints and the options that have been proposed as beneficial to consider for enhancing creativity, which is ultimately always affected by constraints to some extent. Constraints can be understood from different perspectives as highlighted in previous sections, for example distinguishing between process constraints and product constraints (­Rosso, 2014) or making a tripartition separating input, process, and output constraints (­Acar et al., 2019). From a more general perspective, the focus can be shifted from such classifications of constraints toward the creative process itself and its dynamic interplay with constraints. Here, “­constraint management” will be considered an appropriate term to describe the actions that can be taken, both by individuals and teams and organizations, in order to facilitate constraints. Understanding the way in which constraints can influence the creative task at hand is key to managing ­constraints – ​­subsequently being able to approach the task more deliberately, deploying tactics where appropriate throughout the process. Therefore, it is proposed here that constraint management is an important application of theoretical findings in practice, not only for individuals when managing their creative processes but also for teams when facilitating team creativity as well as for organizations when managing for creativity (­Onarheim, 2012a). Therefore, to enable effective constraint management, the more that is understood about constraints and the more consciously that is applied in practice, the greater the chance of using constraints consciously to enhance creativity rather than hinder it.

Constraints Operationalized in Methods Constraints are being used in practical settings all the time when ­well-​­known creativity methods are applied in daily work, as methods can be understood by considering them as constraint management tools. This might, however, be an atypical way to understand constraints or methods, but when looking at creativity and design methods through the lens of constraint management, it becomes quite apparent that they are at their core just that: different types of constraints are imposed, or sometimes removed, to manipulate the way actors think, reason, generate ideas, and behave to a certain extent. For instance, looking at the Negative Brainstorming method where the actors are asked to ideate on negative ways to solve a problem before turning the final solution into a positive (­Silva, 2010), then the actors are given a set of steps to follow, each constraining the way they think and reason. And in that instance, the constraint of generating negative ideas has been regarded as changing participants’ cognitive flexibility and enhancing the chances of an unexpected output. Another example is De Bono’s (­1956) six Thinking Hats, where the hats themselves, each representing different thinking styles, are constraining in nature, putting constraints on the 40

Constraints and Creativity

participants and how they should think. Therefore, in that sense, constraint management has been practiced for decades to enhance creativity through various creative methods.

Constraint Management and Metacognition Taking an even further step back, constraint management can be related to metacognition (­F lavell, 1979), as it requires the individual not only to possess a certain amount of knowledge about different classes of constraints and how they can potentially impact cognition in project and process work but also to consciously monitor the creative process and respond when appropriate through constraint management (­Valgeirsdottir, 2017b; Valgeirsdottir & Onarheim, 2017a). Thus, constraint knowledge alone is not sufficient in itself; in order to effectively manage constraints, it is also necessary to monitor the level of constrainedness imposed on the process and the individuals working in it. Due to the dynamic nature of constraints, they cannot be defined as fixed at any level of the creative process and can vary at any given point of the process. Therefore, the idea of a level of constrainedness is important in this regard because it defines the total sum of constraints (­Onarheim & Biskjaer, 2014). This relates back to the important discussion concerning the dual nature of constraints. Is the individual ­under-​­or ­over-​­constrained, or are they working in their “­sweet spot” (­Onarheim, 2012a)? It is apparent from the more recent trend in constraint research that the topic of balancing constraints is gaining interest (­e.g., ­Haught-​­Tromp, 2017; Lombardo & Kvålshaugen, 2014; Onarheim, 2012a; Rosso, 2017; Stokes, 2014), which is a positive development as the more understanding that is established on the nature of that aspect of constraints, the better it is for managing them in practice. In relation to constrainedness and constraint management, some practical guidance based on empirical evidence has been suggested by Onarheim and Biskjær (­2014) to actively manage individuals into the “­sweet spot”. If an individual consciously monitors their work and ­ nder-​­constrained, it is suggested that the individual can apply determines that they are u some ­self-​­imposed constraints, for example, by applying a time constraint and deciding on a deadline or trying to make the task at hand more concrete by applying methods. These activities, or constraint management, have been observed in experienced creative professionals (­e.g., Onarheim, 2012b). Constrainedness can be regarded in relation to motivation, as in when a person feels u ­ nder-​­constrained, there might not be sufficient pressure on an individual to feel motivated to act on a task; the same principle applies if there are too many constraints on the task and the creative person feels o ­ ver-​­constrained. That might undermine their intrinsic motivation and move them out of their “­sweet spot”. One such approach to manage the constraints is “­constraint removal” (­Onarheim, 2012a), which involves temporarily removing a constraint while continuing to work on the task and subsequently adding it at a later given point and adapting the solution to the constraint that was removed. This might lead to a different solution than if the ­over-​­constrained person had kept on working on the task with all constraints present.

Perceived Versus Actual Constraints One useful distinction to make in practice when operationalizing knowledge derived from research on constraints and creativity is differentiating between perceived constraints and actual constraints. Analyzing imposed constraints at the beginning of a creative task or process can prove useful for further managing the creative work. Perceived constraints are constraints that are not necessarily correct or relevant to the task at hand; however, they inadvertently 41

Balder Onarheim and Dagný Valgeirsdóttir

can influence the creative task as long as the creative actor perceives them to be in p­ lace –​ ­even though they might not be relevant. Perceived constraints can, for instance, be any type of assumption the creative actor holds, rooted in, for example, habits, culture, cognitive biases, and incorrect or f­olk-​­psychological assumptions about their own work practices. Actual constraints are, however, the opposite: the constraints that actually do influence the task at hand, such as ­resource-​­related constraints or time constraints. Therefore, it is a useful distinction to make, especially when working in a team, as the perceived constraints could be different among members, meaning that constraints that are not necessarily relevant could be inadvertently inhibiting the process. By discussing constraints at the start of a creative task, using methods where such perceived constraints are mapped at the beginning of a new project, and actively monitoring them throughout the process, this can prove u ­ seful – ​­acting as a constraint management tool.

Conclusions and Future Research The first years of constraint research in relation to creativity were attributed to understanding and establishing the concept and subsequently forming the consensus that constraints were not to be seen as negative but actually necessary for creativity. Constraints are inherent in the concept of creativity, but the dynamics of their relationship, namely how and when constraints are inhibiting creativity and how and when they can enable it, are still not fully understood. This chapter outlines three main research themes that have emerged in current research efforts and continued work under each theme will contribute to evolving the understanding of the dynamic relationship between constraints and creativity. The first theme concerns the classification of constraints, where there is still a need for a coherent model for the classification of constraints on creativity. The Classes of Constraints model is proposed in this chapter, arguing that the current contributions can be understood based on a distinction between the Origin of a constraint, the intended Target of the constraint, the Type of constraint, and the Nature of the constraint. These four classes act as an approach to organize, understand, and compare already existing categorizations of constraints. Reaching a consensus in that regard will enable more concrete research on the effects of different types of constraints on the creative person and their creative output and make it easier to look across different contributions. Research efforts have already been allocated to researching the effects of types and categorizations of constraints, but without ­ igh-​­level taxonomy, these contributions are not easily aligned. Thus, future research efah forts should continue working towards a shared h ­ igh-​­level taxonomy for existing and new contributions. The second theme concerns research intended to advance the conceptualization of balancing constraints: “­what is too much and what is too little”, exactly what needs to be balanced, and how it should be balanced. What are the crucial constraints to be aware of in order to be able to balance them appropriately to enhance creativity? Falling under this theme is also research into the number of constraints as a function of time and the consequences for creativity of expanding or limiting constrainedness throughout a process. Such results could help demystify the enigma that is currently surrounding this complex phenomenon. An especially important notion in relation to balancing constraints is that of constrainedness. For instance, to advance the understanding of how constrainedness influences the creative process, not just at a point in the process but the total sum of constraints in a process as a whole. Finally, it would be valuable to attain a deeper understanding of the relationship between perceived potential for creativity, perceived constrainedness, and actual creative production. 42

Constraints and Creativity

This can be achieved by comparing constraints, levels of constrainedness, and perceived potential for creativity to the actual creativity of the output to investigate the relationship between these four components. And finally, the third theme concerns constraint management, that is, applying results from the abovementioned empirical work to manage constraints with the purpose of enhancing creativity. Further developing such an understanding will contribute to both theoretical and practical implications, which is important for academic progress. Metacognition has been shown to be a positive enabler for creativity (­Valgeirsdottir & Onarheim, 2017), and constraint management can be regarded as a practical application of knowledge gained from literature with the added metacognitive layer of monitoring constraints and constrainedness for enhancing creativity. Based on all the research outlined in this chapter, the most relevant directions for continued empirical research into constraints on creativity are seen as further developing the understanding of the consequences of timing of constraints, the concrete effects of balancing constraints, differences between individual perceptions of being ­under-​­or ­over-​­constrained and the actual production, and finally the concrete impact of the four Classes of Constraints on creative performance and output. Improving the understanding of these relationships will contribute to further developing current and future approaches to manage constraints to enhance creativity. There is no doubt that constraints have an intricate relationship with creativity, and while research has come a long way, there is still substantial potential for further advancing our understanding of the nature of this relationship. As constraints play a fundamental role in both limiting and enhancing creativity, progress toward a stronger theory of constraints will ultimately contribute to advancing creativity research as a whole.

References Acar, O. A., Tarakci, M.  & van Knippenberg, D. (­2019) ‘­Creativity and Innovation under Constraints: A C ­ ross-​­Disciplinary Integrative Review’, Journal of Management, 45(­1), p­ p. ­96–​­121. doi: 10.1177/­0149206318805832. Amabile, T. M. (­1982) ‘­Social Psychology of Creativity : A Consensual Assessment Technique’, Journal of Personality and Social Psychology, 43(­5), ­pp. ­997–​­1013. Amabile, T. M. (­1996) Creativity in Context. Boulder, CO: Westwood Press. Baer, M. & Oldham, G. R. (­2006) ‘­The Curvilinear Relation Between Experienced Creative Time Pressure and Creativity: Moderating Effects of Openness to Experience and Support for Creativity’, Journal of Applied Psychology. American Psychological Association, 91(­4), ­p. 963. Biskjaer, M. M. (­2013) ­Self-​­Imposed Creativity Constraints. PhD Dissertation. Aarhus University, Aarhus, Denmark. Cromwell, J. & Amabile, T. M. (­2017) ‘­Toward Resolving the Paradox of Creativity and Constraints in Organizations: A Taxonomic Approach’, Academy of Management Proceedings, 2017(­1), ­p. 15067. doi: 10.5465/­a mbpp.2017.15067abstract. Darke, J. (­1979) ‘­The Primary Generator and the Design Process’, Design Studies, 1(­1), ­pp. ­36–​­4 4. doi: 10.1016/­­0142-​­694X(­79)­­90027-​­9. Davis, G. A. (­2011) Barriers to creativity and creative attitudes. In M. A. Runco  & S. R. Pritzker (­Eds.), Encyclopedia of Creativity. 2nd edn (­­ pp.  ­ 115–​­ 121). Amsterdam: Elsevier Inc. doi: 10.1016/­­b978-­​­­0 -­​­­12-­​­­375038-​­9.­0 0021-​­2 . De Bono, E. (­1956) Six Thinking Hats. Cambridge: Little, Brown and Company. Doms, H. & Weiss, M. (­2017) ‘­W hen adaption promotes creativity: the role of task constraints and intuiting’, In Academy of Management Proceedings, Vol. 2017, No. 1 (­­p. 13595). Briarcliff Manor, NY: Academy of Management.. Dorst, K. & Cross, N. (­2001) ‘­Creativity in the Design Process: ­Co-​­evolution of ­Problem-​­-​­Solution’, Design Studies, 22(­5), ­pp. ­425–​­437.

43

Balder Onarheim and Dagný Valgeirsdóttir Elster, J. (­2000) Ulysses Unbound: Studies in Rationality, Precommitment, and Constraints, Ulysses Unbound. Cambridge: Cambridge University Press. Flavell, J. H. (­1979) ‘­Metacognition and Cognitive Monitoring: A New Area of ­Cognitive–​­Developmental Inquiry’, American Psychologist, 34(­10), ­pp. ­906–​­911. doi: 10.1037/­­0 003-​­066x.34.10.906. Guilford, J. P. (­1950) ‘­Creativity’, American Psychologist, 5, ­pp. ­4 44–​­454. Haught, C. (­2015) ‘­The Role of Constraints in Creative Sentence Production’, Creativity Research Journal, 27(­2), ­pp. ­160–​­166. doi: 10.1080/­10400419.2015.1030308. ­Haught-​­Tromp, C. (­2017) ‘­The Green Eggs and Ham Hypothesis: How Constraints Facilitate Creativity’, Psychology of Aesthetics, Creativity, and the Arts, 11(­1), ­pp. ­10–​­17. doi: 10.1037/­aca0000061. Joyce, C. K. (­2009) The Blank Page: Effects of Constraint on Creativity. Berkeley: University of California. Lawson, B. (­2004) What Designers Know. Oxford, England: Architectural Press. ­ onstraint-​­Shattering Practices and Creative Action in OrgaLombardo, S. & Kvålshaugen, R. (­2014) C nizations. Organization Studies, 35(­4), 5­ 87–​­611. Medeiros, K. E. et al. (­2017) Timing Is Everything: Examining the Role of Constraints throughout the Creative Process. Psychology of Aesthetics, Creativity, and the Arts, 12(­4), 4­ 71–​­488. doi: 10.1037/­ aca0000148. Mednick, S. (­1962) ‘­The Associative Basis of the Creative Process’, Psychological Review, 69(­3), ­pp. ­220–​ ­232. doi: 10.1037/­h0048850. Noguchi, H. (­1999) ‘­How Do Material Constraints Affect Design Creativity?’, C&C ’99 Proceedings of the 3rd Conference on Creativity & Cognition, ­pp. ­82–​­87. doi: 10.1002/­hep.24074. Onarheim, B. (­2012a) ‘­Creativity from Constraints in Engineering Design: Lessons Learned at Coloplast’, Journal of Engineering Design, 23(­4), p­ p. ­323–​­336. doi: 10.1080/­09544828.2011.631904. Onarheim, B. (­2012b) Creativity under Constraints: Creativity as Balancing ‘­Constrainedness’. Copenhagen: Copenhagen Business School. Onarheim, B. & Biskjaer, M. M. (­2013) An Introduction to Creativity Constraints’. In ISPIM Conference Proceedings, p­ . 1. Onarheim, B. & Biskjaer, M. M. (­2014) ‘­Balancing Constraints and the Sweet Spot as Coming Topics for Creativity Research’, Creativity in Design: Understanding, Capturing, Supporting, 1(­1), ­1–​­18. ­ reativity – ​­Testing the Impact of “­Late Onarheim, B. & Valgeirsdottir, D. (­2017) Constraints and C Constraints.” In 24th Innovation and Product Development Management Conference. Reykjavik, Iceland: European Institute for Advanced Studies in Management. Onarheim, B.  & Wiltschnig, S. (­2010) ‘­Opening and Constraining: Constraints and Their Role in Creative Processes’, 1st DESIRE Network Conference on Creativity and Innovation in Design: DESIRE2010, (­August), p­ p. ­83–​­89, Aarhus, Denmark. Rhodes, M. (­1961) ‘­A n Analysis of Creativity’, The Phi Delta Kappan. JSTOR, 42(­7 ), p­ p. ­305–​­310. Rosso, B. D. (­2014) ‘­Creativity and Constraints: Exploring the Role of Constraints in the Creative Processes of Research and Development Teams’, Organization Studies, 35(­4), ­pp. ­551–​­585. doi: 10.1177/ ­0170840613517600. Rosso, B. (­2017) ‘­Freedom in Constraint: Understanding How Constraints Enhance and Inhibit R&D Team Creativity’, Academy of Management Proceedings, 2016(­1), ­p. 18167. doi: 10.5465/­a mbpp.2016. 18167abstract. Runco, M. A. & Jaeger, G. J. (­2012) ‘­The Standard Definition of Creativity’, Creativity Research Journal, 24(­1), ­pp. ­92–​­96. doi: 10.1080/­10400419.2012.650092. Sauder d.studio. (­n.d.) Assumption Dumption. http://­dstudio.ubc.ca/­research/­toolkit/­­temporary-​­techniques/­­ new-­​­­6 -­​­­toolkittechniques-­​­­5 -­​­­assumption-​­dumption/ Scopelliti, I. et al. (­2014) ‘­How Do Financial Constraints Affect Creativity?’, Journal of Product Innovation Management, 31(­5), ­pp. ­880–​­893. doi: 10.1111/­jpim.12129. Silva, P. A. (­2010) BadIdeas 3.0: A Method for Creativity in Innovation and Design. Proceedings of the 1st DESIRE Network Conference on Creativity and Innovation in Design, p­ p. ­154–​­162, Aarhus, Denmark. Steidle, A. & Werth, L. (­2013) ‘­Freedom from Constraints: Darkness and Dim Illumination Promote Creativity’, Journal of Environmental Psychology, 35, ­pp. ­67–​­80. doi: 10.1016/­j.jenvp.2013.05.003. Stein, M. I. (­1953) ‘­Creativity and Culture’, The Journal of Psychology, 36(­2), ­pp. ­311–​­322. Sternberg, R. J. & Kaufman, J. C. (­2010) ‘­Constraints on Creativity: Obvious and Not so Obvious’, In J. C. Kaufman & R. J. Sternberg (­Eds.), The Cambridge Handbook of Creativity (­­pp. ­467–​­482). New York: Cambridge University Press. Stokes, P. D. (­2005) Creativity from Constraint. New York: Springer.

44

Constraints and Creativity Stokes, P. D. (­2007) ‘­Using Constraints to Generate and Sustain Novelty’, Psychology of Aesthetics, ­Creativity, and the Arts, ­pp. ­107–​­113. doi: 10.1037/­­1931-​­3896.1.2.107. Stokes, P. D. (­2009) ‘­Using Constraints to Create Novelty: A Case Study’, Psychology of Aesthetics, ­Creativity, and the Arts, 3(­3), ­pp. ­174–​­180. doi: 10.1037/­a0014970. Stokes, P. D. (­2014) ‘­Crossing Disciplines: A ­Constraint-​­Based Model of the Creative/­Innovative Process’, Journal of Product Innovation Management, 31(­2), ­pp. ­247–​­258. doi: 10.1111/­jpim.12093. Valgeirsdottir, D. (­2017b) Enhancing Creativity: Metacognitive Training for Innovation Practitioners. Technical University of Denmark Valgeirsdottir, D.  & Onarheim, B. (­2017a) ‘­Metacognition in Creativity: Process Awareness Used to Facilitate the Creative Process’, in B. T. Christensen, L. J. Ball & K. Halskov. (­Eds.), Analysing Design Thinking: Studies of ­Cross-​­Cultural ­Co-​­Creation (­­pp. ­215–​­228). Leiden: CRC Press/­Taylor & Francis. Wiltschnig, S., Christensen, B. T. & Ball, L. J. (­2013) ‘­Collaborative ­Problem-​­Solution ­Co-​­Evolution in Creative Design’, Design Studies, 34(­5), p­ p. ­515–​­542. doi: 10.1016/­j.destud.2013.01.002.

45

4 THE ROLE OF SERENDIPITY IN CREATIVE COGNITION Wendy Ross and Selene Arfini

The Role of Serendipity in Creative Cognition In general, creativity as a broad concept refers to the generation of something novel and useful (­Runco & Jaeger, 2012), while creative cognition is narrower and indicates the production of a new thought rather than a tangible object. This disembodied and ­non-​­material “­thought” is assumed to not only inspire the necessarily material expression of creativity but to also be separate from it both physically and temporally; this follows a hylomorphic model of creativity where the will of the creator is imposed on inert matter (­Ingold, 2010). While there is a growing literature of distributed perspectives on creativity (­such as Glăveanu, 2014) alongside the ­well-​­established systems perspective (­Csikszentmihalyi, 1998, 2014), the moment of creative ­cognition – ​­the initial spark, if you ­w ill – ​­is usually considered to be a matter for individualist investigation. At heart, such investigations are tasked with the following problem, succinctly summarized by Ohlsson (­1992, ­p. 1): The main puzzle of creative cognition is that it can produce novel concepts, beliefs, problem solutions and products that are not in anyone’s prior experiences. How is this possible? Where does the novelty come from? We suggest that this puzzle stems directly from a view that describes all cognition (­not just creative processes) as a mainly internal phenomenon, without relevant embodied, extended, and distributed features. On this view, creative cognition can only be understood in either terms of individual (­and often exceptional) differences, aptitudes and skills, or a spontaneous, ­hard-­​­­to-​­track emergence. Such an ontological stance (­which has its roots in the computational approach in cognitive science, e.g., Boden, 2008; Caiani, 2016) has inspired a research program that has tied itself up in knots trying to answer how novelty can arise from mundanity without recourse to metaphysically improbable explanations. We propose rather creativity is better explained by an open cognitive system that is meaningfully distributed over the internal resources of the cognizers and the material and social world in which they find themselves embedded. This alternative view opens up the research field to encompass interesting, relevant, and operationalizable questions that derive from an externalist account of



46

DOI: 10.4324/9781003009351-5

The Role of Serendipity in Creative Cognition

creative cognition but that have been overlooked in the directed search for creative thought as an internal phenomenon. In this chapter, we aim to actively explore this gap in the literature and, by so doing, start to unpick Ohlsson’s puzzle. We start by recognizing that, given that creativity requires action in the material world to become manifest (­Glăveanu, 2020; Malafouris, 2014), models of creative cognition should also consider the nature of that engagement with the world. We suggest that novelty arises from shifting and dynamic entangled states, in which objects and people form transient systems with different centers of agency. We will argue that, alongside intentional and meaningful couplings, these ­soft-​­assembled cognitive systems also develop through unplanned action, or serendipity.

Theoretical Background and Goals Developments in cognitive research have already challenged the standard internalist view on different phenomena such as perception (­Gibson, 2014), memory (­Clark & Chalmers, 1998), and language (­Love, 2004; Pinker, 2003). While still mainly grounded in individualist approaches, creative cognition research has also started to embrace the view of creativity as emergent from groups of people, suggesting a parallel between team cognitive processes and individual processes (­Ball & Christensen, 2020; Clapp, 2021; ­Reiter-​­Palmon, 2021). More radical perspectives, which cast cognition as a systemic phenomenon (­­Vallée-​­Tourangeau & ­Vallée-​­Tourangeau, 2020), suggest that creative cognition can best be seen as arising from ­soft-​­assembled cognitive ecosystems extended across not only people but also materials and things. Creative cognition in such programs is seen as stemming from the interaction between a person and the material, ­thing-​­saturated environment in which they are embedded (­Ross, 2022; V ­ allée-​­Tourangeau & March, 2020). Such approaches fundamentally reject a modular account of creativity, wherein internal processes can be isolated from the external environment in which the moment of thinking arises. In this view, thought is not anterior to the creative product but is generated through creative action and indeed is inseparable from that action (­Malafouris, 2014). The argument we put forward in this chapter regarding the central role of serendipity in creative cognition both supports and extends this theoretical framework. Once we move from understanding creative cognition in terms of internal and computational models, new insights into how creative processes unfold can be generated. We suggest that the role of accidents and agents’ interactions with chance, ignorance, and environmental uncertainty are an integral and yet overlooked part of the creative cognitive ecosystem. Indeed, we argue that central to the discussion of systemic cognition (­G. ­Vallée-​ ­Tourangeau & ­Vallée-​­Tourangeau, 2017) is the concept of serendipity: the moment when the material and the mental are combined in unplanned and unintentional action and cognitive agency shifts. In serendipitous moments, the environment is neither simply an influence or scaffold recruited in the service of the explicitly delineated human mind, nor is it the generator of accidental and random bisociations that are taken up by a passive human agent. Instead, serendipity forces us to dissolve these boundaries of agency and distribute epistemic credit across both internal and external forces. Along the iterative, h ­ etero-​­scalar pathway of creativity, serendipity describes moments when creativity is necessarily emergent from the properties of both person and environment. We further suggest that the phenomenon of serendipity undermines the legitimacy of an internalist model to describe all creative cognition. Thus, in the first part of the chapter,

47

Wendy Ross and Selene Arfini

we discuss emerging perspectives in psychology and philosophy that have started to explore the extended, distributed, and emerging properties of creative reasoning. Through adopting this externalist framework, we show how an understanding of the emergence of new ideas comes naturally from the definition of the cognizer as part of an open cognitive ecosystem rather than a closed one. We move on to discuss the concept of serendipity, describing it as more related to the expression of creative cognition than to the exploitation of a lucky event. The focus on the role of serendipity will permit us to analyze and discuss, not only features pertaining to the creative agent, her community, and environment, but also the kinds of inferential reasoning that she can employ to create new and valuable products. Finally, in the third part of the article, we will exploit the concept of serendipity as an explanatory device. By discussing serendipity as an emergent, distributed, and embodied phenomenon, we will argue that it encloses essential features of creative cognition, which needs a distributed and extended mind perspective to be properly appreciated. In particular, putting emphasis on the role of serendipity will allow us to reframe some issues regarding epistemic credit and the value of ­non-​­agentic luck.

Existing Research in Creative Cognition Not surprisingly, many different cognitive processes come under the umbrella of “­creative cognition.” Nevertheless, we will focus the argument here on the moment of insight because it is at once the most empirically intangible and the most recognizable under a folk view of discovery. Furthermore, it is the aspect of creativity that is most phenomenologically akin to serendipity, and a detailed analysis of the relationship between the two is likely to be fruitful for both research fields.

­Insight-​­Problem Solving: Beyond ­Test-​­Tube Creativity Tracking something idiosyncratic like creative cognition is not easy (­ Abraham, 2013). Research in a psychologist’s laboratory is easiest to conduct with ­ well-​­ structured and ­k nowledge-​­lean tasks, but the phenomenon of spontaneous creative cognition is not simply elicited nor manipulated within such a setting. Psychologists then tend to rely on experimental tasks that are presumed to generate either the feeling of insight or at least an insightful process. Traditionally, this class of tasks, ­so-​­called insight problems, contrasts with analytical problems. In analytical problems, while the answer itself is unknown, the process of getting to the answer is clear and known. There is no need for creative cognition. Take mental arithmetic: While the problem solver may not know the solution, they will know the most efficient process to get it. These steps have already been predetermined, either culturally or through personal experience. Indeed, the accuracy of the solution is often not clear in itself but is predicated on an accurate following of a culturally transmitted pathway to solution. Trust arises not from the nature of the answer but from the process that led to that answer. For some insight problems, the answer is given, but the method of reaching the answer is unclear. Take, for example, the “­Triangle of Coins” problem (­­Figure 4.1), which requires participants to move from one triangle direction to its opposite in only three moves. The end state is clear and easily recognizable, and the answer, when found, functions as its own accuracy check. However, in many more insight problems, the problem solvers are in a double state of ignorance: They know neither the correct solution nor how to reach it. 48

The Role of Serendipity in Creative Cognition

­Figure 4.1 The triangle of coins problem and one possible solution

Take, for example, stumpers (­e.g. Bar Hillel, 2021), which require the problem solver to solve riddles such as Ryan has a box set of books, arranged in order from Vol. 1 on the left to Vol. 5 on the right. Each book is 150 pages long. A bookworm had eaten its way in a straight line from the first page of Vol. 1 to the last page of Vol. 2. How many pages did it damage?.1 These problems do not have a clear, already known process. and neither is the answer offered as part of the question. The resolution of these types of insight problems cannot be generated through prior experience, whether individual or cultural, and requires creativity through the recognition or invention of a novel process. More often than not, these tasks are presented as abstracted, linguistic riddles, what we would term “­­second-​­order” problems (­­Vallée-​­Tourangeau & March, 2020). ­Second-​­order problems are those that do not afford a direct interaction with the problem representation but rather proceed on the basis of abstracted mental representations. At times, participants are immobilized in a functional Magnetic Resonance Imaging scanner; other times, they are presented with pictures on a computer screen. Occasionally, they are given models of the problems (­e.g., Fleck & Weisberg, 2013), but this is often not possible with the type of problems presented when they are dependent on linguistic stimuli. Unsurprisingly, such setups provide much evidence for internal mental processes, but they also restrict the investigation to the mental arena. As one of us has argued elsewhere (­Ross, 2022), a theory of serendipity requires an understanding of the manner through which material agency is manifest and particularly a detailed theory of accidents. Environmentally generated accidents occur rarely in materially impoverished situations; however, when insight problems are presented in fi ­ rst-​­order environments, the role of accidents in generating problem solutions is often reported. This is clearly seen in Steffensen et al. (­2016). The researchers presented problem solvers with an insight problem that masquerades as an arithmetic problem: How to arrange 17 animals in four enclosures so that there is an odd number of animals in each enclosure. The problem, however, is not a mathematical one, and the solution requires overlapping the animals’ pens. The ­problem-​­solving trajectory of one participant given model animals and pipe cleaners to play with was analyzed in detail, revealing that the pivotal moment in the cognitive trajectory arose after “­a serendipity” (­­p. 92): an accidental overlap of the enclosures. Accidents have been reported anecdotally in other studies that allow participants to interact with movable objects. In these instances, knowledge is not generated through a priori thinking and consideration but rather reflects unplanned and unintended changes in the environment. Fleck and Weisberg (­2013) refer to this environmentally driven cognition as “­­d ata-​­d riven restructuring”: 49

Wendy Ross and Selene Arfini

­ ata-​­driven restructuring included instances when the individual changed his or D her representation of the problem in response to something he or she saw from the physical configuration of the problem […]. Observations occurred as the participant was attempting to construct or implement another ­heuristic-​­based solution. (­F leck & Weisberg, 2013, ­p. 452) Fioratou and Cowley (­2009) describe a version of the cheap necklace problem 2 presented with actual link chains and observe that six of the 21 solvers (­a lmost a third) solved the problem through the exploitation of an accident. Similarly, Chuderski et al. (­2020, ­p. 18) suggest that In a matchstick algebra problem,3 it is arguably easier to arrive at the solution by accident or trial and error, for instance by realizing as a result of a random movement of a stick that it could act as a negative sign. These accidental revelations undermine a linear, disembodied, and unsituated model of creative cognition. If creative solutions can be generated through an accident in the controlled environment of the laboratory, then theories of the creative process should engage with the nature of accidental solutions, not, we suggest, as a replacement for existing theories but rather as a necessary expansion of the investigative field. Beyond the notes from the psychologists’ laboratories, the material agency readily acknowledged in qualitative reports (­e.g., Glăveanu et al., 2013) is also in part generated by accidents. For example, accidents are seen as a key disruptive mechanism in combinatory play (­Sjöholm, 2014), and accidental visual overlaps can generate ideas for works of art (­Vallée-Tourangeau & March, 2020). Additionally, the role of accidents in the creative process was also emphasized as a key emergent theme by Sawyer (­2018) in his qualitative work on the creative processes of Master of Fine Art (­MFA) students: “­a rtists described a moment in their creative process when an unexpected accident, originating in the external environment, contributed to their creative process” (­­p. 134; see Ross & ­Vallée-​­Tourangeau, 2021). However, just as with laboratory research, such reports are rarely pursued, and the implications for models of creative cognition are underexplored. To some extent, psychologists (­and certainly cognitive psychologists) are somewhat bound by disciplinary allegiances and overarching theoretical frameworks that preclude a close examination of these moments, so we turn now from the psychological literature to situate our argument in a growing tradition in philosophy, especially in the recent development of theories regarding ampliative (­Grosholz, 2016; Psillos, 1996; Woods, 2010) and ­model-​­based reasoning (­Davis & Hamscher, 1988; Ifenthaler & Seel, 2013; Magnani, 1999; Nersessian, 1999). Ampliative reasoning is a general term that defines those inferences that may expand the agents’ knowledge, in contrast with deductive inferences that merely preserve what the agent already knows. This distinction is a traditional assumption of both logic as the study of inferential relations (­H intikka & Sandu, 2007), and epistemology as the study of knowledge acquisition and maintenance (­­Nepomuceno-​­Fernández et  al., 2014).4 Since philosophical discussions of externalism preceded the contemporary use of it in cognitive science,5 the idea that some forms of reasoning may depend on the use of external resources and scaffolding did not get in the way of their philosophical and logical analysis. What, instead, slowed the research regarding creative cognition was the traditional focus on classical, deductive, and ­k nowledge-​­preserving logic. Indeed, it is only recently that philosophical and epistemological analysis has ­fore-​­fronted the research on ampliative reasoning, and in particular abductive 50

The Role of Serendipity in Creative Cognition

inferences (­which define defeasible, hypothetical, and explanatory reasoning), applying that also to the study of creativity and creative cognition (­Ervas et  al., 2018; Magnani, 2017; Meheus, 1999; Meheus & Nickles, 2009). Since philosophical discussion regarding the role of external resources in creative reasoning has fruitfully advanced their theoretical understanding, we suggest that a multidisciplinary and pluralistic stance is necessary to understand something as complex as creative cognition.

Ampliative Reasoning, Abduction, and Creativity in Science As mentioned above, the epistemological and logical focus on deductive and ­k nowledge-​ ­preserving inferences has, in part, delayed the philosophical study of creative reasoning ( ­Boden, 2008).6 Simply, this topic requires arguments that relate to how agents make discoveries and how innovation emerges, which do not readily lend themselves to the rigid concept of knowledge as a stable phenomenon. Rather, understanding creative processes requires research on ampliative inferences, the role of the environment in cognitive processes, and an analysis that conceives knowledge as tentative, temporary, and revisable. In brief, the study of creative reasoning in philosophy required the same framework from which externalism and theories on distributed cognition arose and kept growing. Indeed, when the topic of creativity did finally attract epistemologists’ attention, externalist perspectives had already permeated the literature, so the questions around “­how do creative processes emerge?” were almost indifferent as to “­where” they happen. Rather, to examine how new knowledge is generated, theorists turned to abduction as a key theme and mechanism for understanding the generation of novel thoughts. The term “­abduction” refers to ­hypothesis-​­led reasoning rather than classical ­k nowledge-​ ­led reasoning (­Park, 2017). H ­ ypothesis-​­led reasoning describes a situation in which the agent selects an answer for a particular problem from an array of possible alternatives (­which is usually described as selective abduction) or creates a provisional solution for it by evaluating its appropriateness in specific circumstances (­which is instead presented as creative a­ bduction –​ S­ churz, 2008). Crucially, in both these scenarios, there is seldom a clear answer sustained by a fixed external authority, so the form of knowledge it generates is different from traditionally examined forms. In this way, it reflects true novelty and creativity, the same as the one creativity researchers are keen to pinpoint. When we consider abduction as the inferential root of creativity, different puzzling features of creative cognition emerge and claim our attention. First, it requires an engagement with the outside world through ampliative inferential processes (­Atã & Queiroz, 2014; Viola, 2016): The word abduction, coined by the pragmatist philosopher Charles Sanders Peirce, is a way to translate Aristotle’s term Apagoghé, which means “­to lead away” (­Magnani, 2015). The original form expresses the need to have external feedback and engage the outside world to perform ampliative reasoning, which distances it from deductive inferences based on stable, fixed knowledge and internal processes. Second, it affirms the ignorance (­or uncertainty)-​­preserving feature of creative processes (­Bertolotti et  al., 2016; Magnani, 2013; Magnani et  al., 2016; Woods, 2007). This stems from the idea that abduction is ­hypothesis-​­based, and it requires an epistemological leap from certainty to a possible but not sure solution. It strays away from deductive and k­ nowledge-​ ­preserving inferences in the sense that it offers temporary and ­context-​­based solutions instead of necessary ones. In that sense, ignorance (­a s the agent’s uncertainty) stays within the theoretical framework of the inference. Agents can evaluate hypotheses only if alternatives are actually or potentially available, but they cannot assume to have considered or assessed them all because they are necessarily ignorant of those that are not available. 51

Wendy Ross and Selene Arfini

Third, it raises the issue of the fallibilist nature of ordinary reasoning (­a concept that embeds all Peircean philosophy, to which Margolis, 1998 made an encompassing review) and underscores the provisional nature of the knowledge we create or select on the ground of our hypotheses (­Bam et al., 1979; Haack, 1979; Haack & Kolenda, 1977). A creative abduction emerges because the cognitive agent does not find an appropriate solution for a problem in the array of ones at hand (­or if there are no available solutions at hand), Of course, this does not mean that the creative solution that the agent ends up adopting would necessarily be the only one that could emerge in that circumstance. More than that, discussing the nature of abduction always implies a fallibilist take on the nature of knowledge, which defines our beliefs as always potentially open to revisions or valid objections. Since solutions are, in principle, provisional and revisable, acquiring knowledge is more of a negotiation with the external framework that permits the inference than a necessary consequence of an inference. All these features provided a rich framework of study for the epistemological understanding of creative reasoning and determined the study of what is now commonly referred to as “­abductive cognition” (­Magnani, 2009; Park, 2017; Shank, 1998). This term refers to cognitive processes that hold the logical structure of abductive reasoning but relate to the embodied features of the agents who perform them (­a s visual, perceptual, and manipulative abductions), whether they are human or animal agents. Borrowing from this tradition supports the view of creative cognition as more complex and necessarily extended in the environment, a view that encompasses cognitive processes that rely on bodily functions and skills, on the support and manipulation of external artifacts, and on the construction and manipulation of external models (­Bhatta et al., 1994; Meheus & Nickles, 2009; Pennisi & Falzone, 2020). Thus, by now in the epistemological literature, unlike the psychological literature, it is firmly accepted that a systemic reliance upon and epistemic exchange with the external environment are given features of those cognitive processes that aim to create, set, and evaluate new ideas both in scientific contexts and n ­ on-​­specialized frameworks. For this reason, the focus of the investigations has shifted to how these forms of reliance and exchange with the environment emerge and can be prodded. This shift has directed attention to the idea of chance (­a s the convergence of opportunity and ­accident – ​­McBurney & Ohsawa, 2003) and the talent for recognizing it, which is commonly referred to as sagacity (­especially, as we will see, in studies of serendipity). We argue that evidence from psychology, philosophy, and cognitive science has started to converge on the necessity of external and ­context-​­related events in the development of creative processes. In this renewed context of research, the concepts of chance, accident, and sagacity (­or the “­prepared mind”) emerge as important points of focus, and, from them, the previously obscure notion of serendipity begins to get deserved space and attention.

Serendipity: The Chance of Creativity The word “­serendipity” was coined by Horace Walpole in 1754 to describe a combination of “­sagacity and accident” that leads to happy discoveries (­Silver, 2015). Notwithstanding its folk appeal, scholars have only recently initiated a systematic investigation into its underlying mechanisms, and a precise definition is yet to be distilled (­see Merton & Barber, 2004 for a detailed discussion of the shifting meaning of the word). The nature of serendipity as a scientific notion is far more heterogenous and disparate than its popular understanding. Indeed, since the word is nebulous enough, it often ends up as a “­hook,” put to the service of distilling the concerns of its users, rather than as a research topic in its own right (­Napolitano, 52

The Role of Serendipity in Creative Cognition

2013), and so only in recent years has a systemic research program begun to properly focus on its underlying mechanisms. Even so, the bipartite nature of Walpole’s definition invites us to turn our attention to the importance of the circumstances in which serendipity occurs. The reference to accident, or chance, implies a direct effect of external and unexpected conditions on the agent’s reasoning and behavior. For this reason, connecting serendipity to creative processes is not a conceptual stretch; the role the environment plays has begun to emerge as a necessary element in understanding this complex phenomenon. Taking the two analyses together, in the next subsections, we will contend that, rather than an unpredictable and obscure process, we should see serendipity as a complex but understandable mechanism, which is, at times, responsible for new thoughts.

What Exactly Is Serendipity? Broadly, the concept relies on the theory of planned and unplanned actions. A serendipitous occurrence is essentially disruptive and encompasses breaks, accidents, and ruptures in planned activities. It is also “­at the intersection of chance and wisdom” (­Copeland, 2019, ­p. 2) and emerges in the interaction between a person and an uncertain and changing environment. In brief, serendipity describes how an unsought finding combined with the right person yields an unexpected and happy conclusion. The words “­unsought” and “­unexpected” should be regarded as loaded with cognitive and epistemic meaning. Serendipitous occurrences depend on the impact that unplanned events, encounters, and features have on the agents’ choices, actions, and reasoning. As such, it is a vexingly contingent phenomenon, but one that has been granted credit for several innovations and scientific discoveries and is likely to be responsible for more than has been reported (­Campanario, 1996; Cunha et al., 2010; Yaqub, 2018). Serendipity collapses the Cartesian distinction between mind and material and is qualitatively different from luck, which is necessarily ­non-​­agentic (­Coffman, 2009). Indeed, drawing on long accepted case studies,7 the argument becomes clearer that both the person and luck are necessary. For instance, no take on serendipity would be complete without recounting the story of Fleming’s accidental discovery of ­penicillin – the ​­ sloppy but brilliant scientist who stumbled across penicillin in a moldy petri dish (­Henderson, 1997). Other famous cases include the discoveries of pulsar stars (­Fabian et al., 2010) or helicobacter pylori in the stomach (­Copeland, 2019) and the invention of ­Post-​­it® Notes (­Van Andel, 1994), Velcro (­Cunha et al., 2010), and the microwave oven (­Kotkov et al., 2016). On a more mundane level, serendipity is an ongoing theme in the story of daily scientific discovery. These stories do highlight the importance of taking serendipity in innovation seriously: The relationship between all its constitutive elements: the agent, the environment, and whatever happens ­in-​­between. The notion of sagacity (­a lso the “­prepared mind,” “­w isdom,” and “­readiness”) directs our attention toward those agents who perform a serendipitous discovery, their actions, reasoning, and temperaments. This is further reinforced by a research methodology that views serendipity as an experience mediated by a retrospective report rather than an objectively occurring event (­Ross & ­Vallée-​­Tourangeau, 2021). Since scholars usually view serendipity as a rare occurrence, the idea of “­sagacity” leads them to examine the agents’ qualities and actions, deeming them as necessarily different, singular, or more remarkable than those of others. In this way, research into sagacity mirrors an individual differences approach to creative cognition that seeks the locus for creativity in the exceptional qualities of the person. 53

Wendy Ross and Selene Arfini

This focus has directed the studies on serendipity until now, at times minimizing considerations of the impact of the material environment on agents’ performances. The external triggers located in the environment are seen as distinct from the sagacity that exploits them rather than integral to that moment.

Planning the Unexpected: The Agent, the Environment, and the I­ n-​­Between Central to our argument is that a key characteristic of all types of serendipity is that they arise from unplanned activity. Serendipity happens through unintentional action and diverts the creative pathway down an unanticipated route. It is necessarily linked to and emerges from an exploitation of the unexpected, which makes it particularly relevant to creative cognition, especially under the externalist framework outlined above. From the perspective of the environment, the shifting cogs of a complex, materially laden backdrop will necessarily yield moments of luck in the form of accidents, but these opportunities are inert until they are enacted by an agent. From the perspective of that person, it requires an act of noticing this chance and the personal characteristics8 (­sagacity, a prepared mind) to exploit the opportunity. Thus, serendipity is truly emergent and systemic; internal as well as external processes are necessary, and neither is sufficient on its own (­Ross, 2023). As we have noted above, despite some evidence that there is a weak link between individual differences and serendipity (­­McCay-​­Peet et al., 2015), in an echo of the focus of creative cognition research, much research on serendipity assumes an internalist stance. This is not unreasonable. It is the addition of sagacity, which raises serendipity from accident, luck, or random chance, and so traditionally this has been the focus of much theorizing. For many writers, these internal characteristics provide the element of control of serendipity (­Foster & Ford, 2003), and in wider folk psychology, we often refer to “­m aking your own luck.” Indeed, de Rond (­2014a) suggests that the only proper analysis of serendipity is to focus on human agency, suggesting that it is the capability of being able to match two disparate events. ­McCay-​­Peet et al.’s (­2015) process model (­­Figure 4.2) has more room for the environment as a trigger for serendipity as part of a process. According to this model, a serendipitous moment, to be called so, must often be followed up and worked on with varying levels of granularity. Only when the valuable outcome appears, which combines both the practical result and the feeling that accompanies it, can it be perceived as a serendipitous occurrence. Perhaps the most detailed taxonomy of serendipitous environments comes from Björneborn (­2017), who sees serendipity as an affordance or “­usage potential” (­­p. 3) rather than an experience or capability. He identifies three key affordances for serendipity that, in turn, encompass ten ­sub-​­affordances. Although he also identifies personal factors, he devotes the majority of his framework to these affordances of diversifiability, traversability, and sensoriability. An environment high in diversifiability facilitates serendipity because it allows for more accidental bisociations and juxtapositions. Traversability refers to how easy it is to move and access diverse content. The accidents need to be enacted in an environment that supports this. Finally, sensoriability describes the way an environment appeals to all the senses and deals with how the environment can make different resources stand out for different senses and facilitate noticing. So, a detailed examination of the environments that facilitate serendipity is possible. These environments can arise naturally or can be reinforced deliberately by design. Although there appears to be a paradox in planning for serendipity because serendipity arises from unplanned action, environments can be designed to create more opportunities for accidents to arise (­Makri & Race, 2016). The designers can plan for serendipity, but it is the users who must 54

The Role of Serendipity in Creative Cognition

­Figure 4.2  Serendipitous process Source: Adapted from McCay-Peet & Toms, 2015.

uncover things in an unplanned manner (­Björneborn, 2017). Thus, we return to our original definition: Serendipity is the outcome of unplanned action directed by material changes in the e­ nvironment – designing for serendipity simply brings the changes in the environment ​­ under control.

Serendipity in Creative Cognition Language and concepts overlap in the literature on serendipity and insight, especially when they refer to the phenomenological feeling that accompanies either insight or a moment of serendipity. Indeed, authors often use the same ­phrasing – ​­a ‘eureka moment’ or ‘aha’. Makri and Blandford (­2012) suggest that serendipity in their model (­­Figure 4.3) requires ‘insight’, and later, Makri argues that the shift in serendipity is the same shift found in innovation and creativity (­Race & Makri, 2016). In organizational research, attention has been paid to the link between serendipity and innovation (­Cunha et al., 2010; Olma, 2016), and there is a long history of the role of serendipity in scientific creativity and discovery (­Copeland, 2019; Thagard, 2012). We suggest that this overlap may point to a fundamental similarity in underlying mechanisms. Furthermore, it is also related to the epistemological leap that agents need to make when performing creative abductions. In all three, something happens that changes the agent and her understanding of the problem space and, in turn, generates novelty. In terms of serendipity and creative cognition, we have seen that in many cases, the change is externally 55

Wendy Ross and Selene Arfini

­Figure 4.3 The process of serendipity Source: Adapted from Makri & Blandford, 2012.

generated and unplanned, and thus, we argue for an equal role for the environment in the process of creative thought. Through this lens, it becomes clearer how what happens in that black box is not just insight but also “­outsight” (­Vallée-Tourangeau & March, 2020). In the following section, we will therefore review some of the clearest ways in which reciprocities can support research in both fields.

The Prepared Mind and a Theory of Hints The famous comment by Louis Pasteur in a lecture given in 1854 that “­le hazard ne favorise que les esprits preparés”9 is often used as a framework to understand the interaction of human agents with luck. The notion of the prepared mind is a useful aid to distinguish between serendipity, which requires both accident and sagacity, and n ­ on-​­agentic luck (­A rfini et al., 2020; Busch & Barkema, 2020). Seifert et al. (­1994) explored the nature of this preparedness and argued convincingly for the importance of considering the external world as a mechanism for alleviating the impasse that is often encountered when problem solvers can no longer proceed. In this way, the prepared mind moves beyond a consideration of which traits can make someone more or less likely to generate a creative thought to considering it as a cognitive state. It is not implausible that this state would share features with the ­more-​ ­researched state of incubation. In this case, research in creative cognition can inform research in serendipity by moving the discussion away from exceptional characteristics to clear mental states that are already regularly evoked in the psychologist’s laboratory. Another way creative cognition research can support our understanding of serendipity is through the research program on hints. An early example of this was Maier (­1931), who asked participants to solve the t­ wo-​­string problem. Briefly, this problem requires participants to uncover a way to join two strings hanging from the ceiling but set too far apart to grasp simultaneously. The solution requires the problem solvers to set one rope swinging (­w ith the help of pliers as a pendulum weight). In the experiment of interest, an experimenter set one rope swinging as if by accident. This hint increases the solution rate as it nudges participants toward the idea (­Ball & Litchfield, 2013). The environment here pushes the problem solver toward a valid hypothesis and provides direction. The nature of these environmental hints and their scaffolding effect are explicitly addressed by Kirsh (­2009), who offers a “­theory of hints” as a way of establishing the interaction between environmental cues; his analysis, however, raises more questions than he answers by also suggesting that “­a more theoretically motivated explanation of how these cues trigger candidate generation is needed” (­­p. 293). We wholeheartedly agree that such an explanation is necessary. We suggest that a serious engagement with the literature on serendipity can enhance this understanding. ­McCay-​­Peet 56

The Role of Serendipity in Creative Cognition

and Toms’ (­2015) external “­triggers” could easily be recast as the “­h ints” that facilitate ­problem-​­solving, and the profiling of environmental affordances described by Björneborn (­2017) will allow researchers to understand the relations between hints as they naturally arise and the problem solver. Under laboratory conditions, the hints are ­experimenter-​­generated and controlled. Beyond the laboratory, these hints are the fluctuations of environmental chance, the interaction with which is serendipity. Thus, while serendipity forces us to reassess the ontological locus of a new thought, theoretical and experimental research on creative cognition can extend our understanding of the cognitive processes of serendipity.

Restructuring This reciprocity continues as we consider the restructuring framework in insight ­problem-​ ­solving (­Ohlsson, 1992). This framework suggests that insight requires the unhelpful problem representation to shift to a more productive one. Makri and Blandford’s process model itself quite neatly echoes theories of the processing of creative cognition such as that outlined by Danek to explore ­non-​­monotonic (­a lso abductive) ­problem-​­solving (­Danek, 2018). When Foster and Ford (­2003, ­p. 333) write of serendipity “­t aking the researcher in a new direction, in which the problem conception or solution is reconfigured in some way,” there are palpable echoes of the restructuring (­which motivated also the research on radical forms of abduction as the t­ rans-​­paradigmatic type; Hendricks & Faye, 1999). The role of accident here may be an important explanatory factor for insight, but, when it is recognized, it tends to be subsumed into existing models rather than investigated further. So, for example, as discussed above, Fleck and Weisberg’s (­2013) ­data-​­driven restructuring describes restructuring triggered by unintended actions over objects, which fits neatly into a theoretical perspective that accepts such mechanisms. However, observations such as ­these –​ t­hat there are times when the environment yields a solution to the ­problem  – ​­are ­under-​ i­nvestigated precisely because the theoretical and methodological allegiances mean that they are simply dismissed. That is, if the aim of ­problem-​­solving research is to uncover the internal algorithms and processing that lead to a solution, then the role of accidents can shed little light on this. If we are interested in ­problem-​­solving as it unfolds, however, and not as a reflection of unstable internal states, then it behoves us to take all possible explanatory factors seriously. Insight research is turning toward viewing insight as having two dimensions: the affective and the cognitive (­Danek et al., 2020). The cognitive is seen as a rerepresentation. The affective is normally measured across five dimensions: the feeling of aha, the feeling of being stuck, the feeling of certainty in your answer, the feeling of happiness, and a feeling of surprise. Surprise, indeed, is also a fundamental topic in studies on abductive and ampliative reasoning, since it is considered the origin of the explanatory and ­meaning-​­making processes (­i.e., the “­trigger” of abduction; Pierce, 1931). This comes from the intuitive idea that surprise clearly marks a dislocation from the original plan (­g iven that it could not emerge from something that is preplanned and prespecified).

Accidents, Epistemic Credit, Sense, and Sensibility It is our contention that the role of accident is necessary to fully understand the interaction between world and person. Instead of a hylomorphic model of discovery that takes the environment as a passive recipient of an innovator’s poking, the model of serendipity posits a world that pokes back, or indeed, even pokes first.10 Take, for example, the astronomer Vera 57

Wendy Ross and Selene Arfini

Rubin’s description of her discovery that stars belonging to a galaxy do not all rotate in the same direction. Her words clearly echo the phenomenology of insight, but it was an insight wrought by the combination of skill and the physical manipulation of objects in the world: I sit in front of this very exotic TV screen next to a computer, but it gives me the images of these spectra very carefully and I can play with them. And I don’t know, one day I just decided that I had to understand what this complexity was that I was looking at and I made sketches on a piece of entry and suddenly I understood it all. I have no other way of describing it. It was exquisitely clear. (­Csikszentmihalyi, 1996, ­p. 4) Kirsh and Maglio (­1994) would divide actions as either pragmatic or epistemic. They suggest that pragmatic actions advance the problem solver toward a physical goal, while epistemic actions advance her epistemically. The actions described by Rubin in the quotation above cannot be easily filed under these categories; however, unplanned actions such as those described by Rubin over the images on the screen fulfill neither function. Rather, these unplanned actions have an exaptative function, that is, their original p­ urpose  – ​­fiddling or ­playing – ​­changes the environmental affordances, which then change the nature of the actions, and so the result arises in a recursive process emerging from unplanned changes wrought over the environment. The actions changed the epistemic state of the agent in an unplanned manner, which led the agent to change the status of the affordances she could perceive and exploit. Exaptative actions of this kind are necessarily abductive. However, all types of luck, from accidents to fortunate coincidences, are problematic in the case of creativity. Creativity in the research literature and in popular understanding has acquired the status of an epistemic and character virtue (­Blackburn, 2017; Kieran, 2017), whether that be a virtue associated with a unique talent or with the level of perseverance and hard work. If creativity is considered a virtue, then it must be agential and not overly dependent on chance. That is, “­schizophrenic word salads” (­Weisberg, 2010, ­p. 237) are examples of something that is not creative, even if it is novel, and arguably, as useful as some other forms of literature. This instinct to draw a line between true creativity and unintentional products underlies the anxieties generated by ­non-​­intentional algorithms and puts accidentally created works of great beauty outside of the bounds of creativity research. It also reflects a mind/­ world dualism that often marks mainstream considerations of our actions in the world. We suggest that serendipity, by overcoming the rigid dualism between agent and environment, helps resolve the anxieties around epistemic credit. Its blended and relational status gives serendipity its power as a framework for understanding our interactions with chance and accident. A serious consideration of serendipity does not require us to devolve epistemic credit to the environment, even though it cautions against awarding it solely on agentic action. Without deep interaction between the agents, their epistemic and cognitive frameworks, and accidents, we would not have serendipitous encounters. Creativity, we argue, works the same and should not be reduced to any of its constitutive elements.

Concluding Thoughts In this paper, we argued that taking serendipity seriously invites us to move away from describing creative cognition as the internal spark that necessarily precedes engagement with the material form of that creativity. While this description does encapsulate parts of the creative process, such a narrow view ultimately leads research efforts into solving the wrong sort 58

The Role of Serendipity in Creative Cognition

of puzzle. Rather, focusing on o ­ bject-​­thought mutualities (­­Vallée-​­Tourangeau, this volume) allows us to expand our explanatory mechanisms to incorporate the role of human networks and the active nature of the environment into our investigations regarding the development and emergence of creative cognition. Studying creative phenomena by discussing the role of serendipity requires an accompanying methodological shift. Laboratory problems that are inescapably mental are replaced by ­ fi rst-​­ order, materially embedded problems that allow space for accidents (­­ Vallée-​ ­Tourangeau & March, 2020). This also requires us to use finely grained microgenetic methods that capture the recursive and distributed nature of creative agency (­­Neves-​­Pereira, 2019; Ross & ­Vallée-​­Tourangeau, 2021; Tanggaard & Beghetto, 2015). The theory of creativity outlined here also poses a strong challenge to the established scientific practice, which requires both replicability and the possibility to predict events (­Cummins, 2000; ­Neves-​ ­Pereira, 2019). Predictive causality is hard to infer from accidents. Of course, the externalist reframing does not answer all our research questions. But we argue that it leads to a more fertile field of study. The field requires us to clarify various related issues; for example, is there room for chance and creativity in heavily structured research (­a nd the f­ollow-​­up “­If there is not, should we allow it?”) and how serendipity unfolds in collaborative creative efforts, including both collaboration across people and also material collaborations across people and things (­Ross et al., 2020). Notwithstanding these potential new directions of the research, nothing says that others could not, obviously, serendipitously emerge in their wake. Authors’ Note: We thank Samantha Copeland and the editors for thoughtful comments on a previous draft.

Notes 1 The answer is 0; when books are lined up on a shelf the first page of the first volume touches the last page of the second volume. 2 The cheap necklace problem requires participants to make a complete closed loop (­necklace) out of 12 links of chain, with the starting point being four smaller, ­three-​­link chains. A cost constraint (­2 cents to break a link, and 3 cents to join a link) is imposed. The correct solution involves breaking all three links of one of the ­three-​­link chains, and using the individual links to connect the three remaining ­three-​­l ink chains together. 3 A matchstick algebra problem presents the problem solver with an equation expressed in roman numerals which is not correct such as VI = VII + I which requires her to move one matchstick to make the equation correct, in this instance VII = VI + I. 4 Aristotelian syllogisms provide the most known deductive examples (­Corcoran, 1974; Smiley, 1973). The logical structure of the sentence “­A ll men are mortal, Socrates is a man, therefore Socrates is mortal” allows us to say that if the premises of this sentence are true, the conclusion must also be true. Ampliative reasoning in logical terms does not preserve truth because it is based on a weaker relation between premises and conclusion. A generalization from a set of observations is an ampliative argument (­or an inductive one) because it allows us to say, for example, that all crows are black (­conclusion), after having observed various crows and having found all of them black (­premises). Ampliative reasoning defines all ­non-​­deductive inferences because they may expand the knowledge of the agent, even if they do not always lead to a true conclusion, since they ​­ example, fallacious reasoning and biases have also this are based on hypotheses and g­ uesses – for ampliative, if fallible, nature (­Ippoliti, 2015). 5 The distinction often presented in philosophy of mind between semantic externalism (­Kallestrup, 2013) and active externalism (­Clark & Chalmers, 1998) defines them respectively as the traditional versus the novel account, because it is assumed that externalist perspectives in philosophy of language and epistemology were already discussed before the emergence of their cognitive/­mental counterpart.

59

Wendy Ross and Selene Arfini 6 This does not mean that in the history of philosophy the topic of creativity was not discussed: Plato, Aristotle, Kant, Schopenhauer, and Nietzsche famously discussed in length of creation in arts and in the artists’ and geniuses’ mind. Still, the topic was shushed away by early 20th century’s philosophical currents and perspectives that were interested in the study of scientific progress, reasoning, and cognitive processes, until it remerged in the late decades of the century; for a discussion of this strange theoretical fluctuation cf. (­Paul & Kaufman, 2014). 7 These tales populate the world’s history (­w ith varying degrees of anecdotal and apocryphal relish). Of course, such stories are mostly embellishments, full of survivorship bias, which erase the network effort necessary to solidify moments of serendipity (­Copeland, 2018). As such it would be tempting to disregard them, however, they mimic stories from Wallas, Poincaré and Köhler which are foundational to the research programme in creative cognition and have yielded fruitful insights. 8 It is unclear at this moment whether these characteristics denote a trait or a state. It may be that as the research field unfolds the sagacity of the prepared mind is understood less as a psychometric trait and more as an attitude. 9 Multiple translations of this abound (­see Van Andel, 1994). At its heart though it refers to the nature of the mind which is prepared to take advantage of chance occurrences: In the field of discovery, luck will only favour the prepared mind. 10 We are indebted to conversations with Frédéric ­Vallée-​­Tourangeau and Paul March for this idea.

References Abraham, A. (­2013). The promises and perils of the neuroscience of creativity. Frontiers in Human Neuroscience, 7, ­1–​­9. https://­doi.org/­10.3389/­f nhum.2013.00246 Arfini, S., Bertolotti, T., & Magnani, L. (­2020). The antinomies of serendipity how to cognitively frame serendipity for scientific discoveries. Topoi, 39, ­939–​­948. https://­doi.org/­10.1007/­­s11245-­​­­018-­​­­9571-​­3 Atã, P., & Queiroz, J. (­2014). Icon and abduction: Situatedness in Peircean cognitive semiotics. In L. Magnani (­Ed.), ­Model-​­Based Reasoning in Science and Technology. Studies in Applied Philosophy, Epistemology and Rational Ethics (­­pp. ­301–​­333). Springer Ball, L. J.,  & Christensen, B. T. (­2020). How sticky notes support cognitive and ­socio-​­cognitive processes in the generation and exploration of creative ideas. In B. T. Christensen, K. Halskov,  & C. N. Klokmose (­Eds.), Sticky Creativity (­­pp. ­19–​­51). Academic Press. https://­doi.org/­10.1016/­­B978­​­­0 -­​­­12-­​­­816566-​­9.­0 0002-​­1 Ball, L. J., & Litchfield, D. (­2013). Interactivity and embodied cues in problem solving, learning and insight: Further contributions to a “­theory of hints”. In S. J. Cowley & F. Vallée-Tourangeau (­Eds.), Cognition beyond the Brain (­­pp. ­232–​­239). Springer. Bam, V., Cook, T., & Lincourt, J. (­1979). Hypothetical fallibilism in Peirce and Jevons. Transactions of the Charles S. Peirce Society, 15(­2), ­132–​­157. https://­w ww.jstor.org/­stable/­40319495 ­Bar-​­Hillel, M. (­2021). Stumpers: An annotated compendium. Thinking & Reasoning, 27(­4), ­536–​­566. Bertolotti, T., Arfini, S., & Magnani, L. (­2016). Abduction: From the ignorance problem to the ignorance virtue. FLAP, 3(­1), ­153–​­173. ­ odel-​­based approach. Bhatta, S., Goel, A., & Prabhakar, S. (­1994). Innovation in analogical design: A m In J. S. Gero & F. Sudweeks (­Eds.), Artificial Intelligence in Design ’94 (­­pp. ­57–​­74). Springer Netherlands. https://­doi.org/­10.1007/­­978-­​­­94- ­​­­011- ­​­­0928- ​­4 _4 Björneborn, L. (­2017). Three key affordances for serendipity: Toward a framework connecting environmental and personal factors in serendipitous encounters. Journal of Documentation, 73, ­1053–​ ­1081. https://­doi.org/­10.1108/­­J D-­​­­07-­​­­2016-​­0 097 Blackburn, S. (­2017). Creativity and ­no-­​­­so-​­dumb luck. In S. Elliot Paul & S. B. Kaufman (­Eds.), The Philosophy of Creativity: New Essays (­­pp. ­147–​­156). Oxford University Press. Boden, M. A. (­2008). Mind as Machine: A History of Cognitive Science. Oxford University Press. Busch, C., & Barkema, H. (­2020). Planned luck: How incubators can facilitate serendipity for nascent entrepreneurs through fostering network embeddedness. Entrepreneurship Theory and Practice, 46(­4), ­884–​­919. http://­doi.org/­10.1177/­1042258720915798. Caiani, S. Z. (­2016). Corporeità e Cognizione: La Filosofia Della Mente Incorporata (­Prima edizione Le Monnier Università). Le Monnier università. Campanario, J. M. (­1996). Usingcitation classics to study the incidence of serendipity in scientific discovery. Scientometrics, 37(­1), ­3 –​­24. https://­doi.org/­10.1007/ ­BF02093482

60

The Role of Serendipity in Creative Cognition Chuderski, A., Jastrzębski, J.,  & Kucwaj, H. (­2020). How physical interaction with insight problems affects solution rates, hint use, and cognitive load. British Journal of Psychology. https://­doi. org/­10.1111/­bjop.12442 ­ hapter ­6 —​­Don’t call it collaboration! Reframing creative success in teams from Clapp, E. P. (­2021). C the perspective of participatory creativity. In A. S. McKay, R. R ­ eiter-​­Palmon & J. C. Kaufman (­Eds.), Creative Success in Teams (­­pp. ­101–​­121). Academic Press. https://­doi.org/­10.1016/­­B978-­​­­0 -­​­­12-­​­­ 819993-​­0.­0 0006-​­0 Clark, A., & Chalmers, D. (­1998). The extended mind. Analysis, 58(­1), ­7–​­19. https://­doi.org/­doi:10.1093/­ analys/­58.1.7 Coffman, E. J. (­2009). Does luck exclude control? Australasian Journal of Philosophy, 87(­3), ­499–​­504. https://­doi.org/­10.1080/­0 0048400802674677 Copeland, S. (­2018). “­F leming leapt on the unusual like a weasel on a vole”: Challenging the paradigms of discovery in science. Perspectives on Science, 26(­6), ­694–​­721. Copeland, S. M. (­2019). On serendipity in science: Discovery at the intersection of chance and wisdom. Synthese, 196, ­2385–​­2406. https://­doi.org/­10.1007/­­s11229- ­​­­017-­​­­1544-​­3 Corcoran, J. (­1974). Aristotelian syllogisms: Valid arguments or true universalized conditionals? Mind, 83(­330), ­278–​­281. Csikszentmihalyi, M. (­1996). Creativity: The Psychology of Discovery and Invention. Harper Collins. Csikszentmihalyi, M. (­1998). Implications of a systems perspective for the study of creativity. In R. J. Sternberg (­Ed.), Handbook of Creativity (­1st ed., ­pp. ­313–​­336). Cambridge University Press. Csikszentmihalyi, M. (­2014). The Systems Model of Creativity. Springer. Cummins, R. (­2000). ‘­How does it work?’ versus ‘­W hat are the laws?’: Two conceptions of psychological explanation. In F. C. Keil & R. A. Wilson (­Eds.), Explanation and Cognition ( ­­pp. ­118–​­144). MIT Press. Cunha, M. P., Clegg, S. R., & Mendonça, S. (­2010). On serendipity and organizing. European Management Journal, 28, ­319–​­330. https://­doi.org/­10.1016/­j.emj.2010.07.001 Danek, A. H. (­2018). Magic tricks, sudden restructuring and the aha! experience. In F. V ­ allée-​ ­Tourangeau (­Ed.), Insight: On the Origin of New Ideas (­­pp. ­51–​­79). Routledge. Danek, A. H., Williams, J., & Wiley, J. (­2020). Closing the gap: Connecting sudden representational change to the subjective Aha! experience in insightful problem solving. Psychological Research, 84(­1), ­111–​­119. Davis, R., & Hamscher, W. (­1988). ­Model-​­based reasoning: Troubleshooting. In H. E. Shrobe and the American Association for Artificial Intelligence (­Eds.) Exploring Artificial Intelligence (­­pp. ­297–​­346). Elsevier. de Rond, M. (­2014). The structure of serendipity. Culture and Organization, 20(­5), ­342–​­358. https://­doi. org/­10.1080/­14759551.2014.967451 Ervas, F., Ledda, A., Ojha, A., Pierro, G. A., & Indurkhya, B. (­2018). Creative argumentation: When and why people commit the metaphoric fallacy. Frontiers in Psychology, 9, 1815. Grosholz, E. R. (­2016). Starry Reckoning: Reference and Analysis in Mathematics and Cosmology. Springer. Fabian, A. C., de Rond, M., & Morley, I. (­2010). Serendipity in astronomy. In M. de Rond & I. Morley (­Eds.), Serendipity: Fortune and the Prepared Mind (­­pp. ­22–​­73). Cambridge University Press Fioratou, E., & Cowley, S. J. (­2009). Insightful thinking: Cognitive dynamics and material artifacts. Pragmatics & Cognition, 17(­3), ­549–​­572. https://­doi.org/­10.1075/­pc.17.3.04fio Fleck, J. I., & Weisberg, R. W. (­2013). Insight versus analysis: Evidence for diverse methods in problem­ solving. Journal of Cognitive Psychology, 25(­4), ­436–​­463. https://­doi.org/­10.1080/­20445911.2013. 779248 Foster, A. E., & Ford, N. (­2003). Serendipity and information seeking: An empirical study. Journal of Documentation, 59, ­321–​­340. https://­doi.org/­10.1108/­0 0220410310472518 Gibson, J. J. (­2014). The Ecological Approach to Visual Perception: Classic Edition. Psychology Press. Glăveanu, V. P. (­2014). Distributed Creativity: Thinking Outside the Box of the Creative Individual. Springer. Glăveanu, V. P. (­2020). A sociocultural theory of creativity: Bridging the social, the material, and the psychological. Review of General Psychology, 24(­4), ­335–​­354. 108926802096176. https://­doi. org/­10.1177/­1089268020961763 Glăveanu, V. P., Lubart, T., Bonnardel, N., Botella, M., Biaisi, P.-​­M. de, ­Desainte-​­Catherine, M., Georgsdottir, A., Guillou, K., Kurtag, G., Mouchiroud, C., Storme, M., Wojtczuk, A., & Zenasni, F. (­2013). Creativity as action: Findings from five creative domains. Frontiers in Psychology, 4, 176, ­1–​­14. https://­doi.org/­10.3389/­f psyg.2013.00176

61

Wendy Ross and Selene Arfini Haack, S. (­1979). Fallibilism and necessity. Synthese, 41(­1), 3­ 7–​­63. https://­w ww.jstor.org/­stable/­20115371 Haack, S., & Kolenda, K. (­1977). Two fallibilists in search of the truth. Proceedings of the Aristotelian Society, Supplementary Volumes, 51, ­63–​­104. Henderson, J. W. (­1997). The yellow brick road to penicillin: A story of serendipity. Mayo Clinic Proceedings, 72(­7 ), ­683– ​­687. https://­doi.org/­10.4065/­72.7.683 Hendricks, V. F.,  & Faye, J. (­1999). Abducting explanation. In L. Magnani, N. Nersessian,  & P. Thagard (­Eds.), ­Model-​­Based Reasoning in Scientific Discovery (­­pp. ­271–​­292).Springer. Hintikka, J., & Sandu, G. (­2007). What is logic? In D. Jacquette (­Ed.) Philosophy of Logic (­­pp. ­13–​­39). Elsevier. Ifenthaler, D., & Seel, N. M. (­2013). ­Model-​­based reasoning. Computers & Education, 64, ­131–​­142. Ingold, T. (­2010). The textility of making. Cambridge Journal of Economics, 34(­1), ­91–​­102. https://­doi. org/­10.1093/­cje/­bep042 Ippoliti, E. (­Ed.) (­2015). Heuristic reasoning. Springer. Kallestrup, J. (­2013). Semantic Externalism. Routledge. Kieran, M. (­2017). Creativity as virtue of character. In E. S. Paul & S. B. Kaufman (­Eds.), The Philosophy of Creativity: New Essays (­­pp. ­125–​­144). Oxford University Press. Kirsh, D. (­2009). Problem solving and situated cognition. In P. Robbins & M. Ayeded (­Eds.), The Cambridge Handbook of Situated Cognition (­­pp. ­264–​­306). Cambridge University Press. Kirsh, D., & Maglio, P. (­1994). On distinguishing epistemic from pragmatic action. Cognitive Science, 18, ­513–​­549. Kotkov, D., Wang, S.,  & Veijalainen, J. (­2016). A survey of serendipity in recommender systems. ­Knowledge-​­Based Systems, 111, ­180–​­192. https://­doi.org/­10.1016/­j.knosys.2016.08.014 Love, N. (­2004). Cognition and the language myth. Language Sciences, 26(­6), ­525–​­544. https://­doi. org/­10.1016/­j.langsci.2004.09.003 Magnani, L. (­1999). ­Model-​­based creative abduction. In L. Magnani, N. Nersessian, & P. Thagard (­Eds.), ­Model-​­Based Reasoning in Scientific Discovery (­­pp. ­219–​­238). Springer. ­ co-​­Cognitive Dimensions of Hypothetical Magnani, L. (­2009). Abductive Cognition: The Epistemological and E Reasoning (­Vol. 3). Springer Science & Business Media. Magnani, L. (­2013). Is abduction ­ignorance-​­preserving? Conventions, models and fictions in science. Logic Journal of the IGPL, 21(­6), ­882–​­914. https://­doi.org /­10.1093/­jigpal/­jzt012. Magnani, L. (­2015). The ­eco-​­cognitive model of abduction: Ἀπαγωγή now: Naturalizing the logic of abduction. Journal of Applied Logic, 13(­3), ­285–​­315. https://­doi.org/­10.1016/­j.jal.2015.04.003 Magnani, L. (­2017). Abductive Structure of Scientific Creativity. Springer. Magnani, L., Arfini, S.,  & Bertolotti, T. (­2016). Of habit and abduction: Preserving ignorance or attaining knowledge? In D. E. West and M. Anderson (­Eds.), Consensus on Peirce’s Concept of Habit (­­pp. ­361–​­377). Springer. Maier, N. R. F. (­1931). Reasoning in humans: II. The solution of a problem and its appearance in consciousness. Journal of Comparative Psychology, 12, ­181–​­194. https://­doi.org/­10.1037/­h0071361 ​­ 1: A process Makri, S.,  & Blandford, A. (­2012). Coming across information s­erendipitously  – Part model. Journal of Documentation, 68, ­684–​­705. https://­doi.org/­10.1108/­0 0220411211256030 Makri, S., & Race, T. M. (­2016). Serendipity in current digital information environments. In Tammera Race & S. Makri (­Eds.), Accidental Information Discovery: Cultivating Serendipity in the Digital Age (­­pp. ­53–​­80). Chandos. Malafouris, L. (­2014). Creative thinging: The feeling of and for clay. Pragmatics  & Cognition, 22(­1), ­140–​­158. https://­doi.org/­10.1075/­pc.22.1.08mal Margolis, J. (­1998). Peirce’s fallibilism. Transactions of the Charles S. Peirce Society, 34(­3), ­535–​­569. McBurney, P., & Ohsawa, Y. (­2003). Chance Discovery. ­Springer-​­Verlag. ­McCay-​­Peet, L., Toms, E. G., & Kelloway, E. K. (­2015). Examination of relationships among serendipity, the environment, and individual differences. Information Processing and Management, 51(­4), ­391–​­412. https://­doi.org/­10.1016/­j.ipm.2015.02.004 Meheus, J. (­1999). Deductive and ampliative adaptive logics as tools in the study of creativity. Foundations of Science, 4(­3), 3­ 25–​­336. Meheus, J., & Nickles, T. (­2009). Models of Discovery and Creativity. Springer. Merton, R., & Barber, E. (­2004). The Travels and Adventures of Rerendipity: A Study in Sociological Semantics and the Sociology of Science. Princeton University press.

62

The Role of Serendipity in Creative Cognition Napolitano, C. M. (­2013). More than just a simple twist of fate: Serendipitous relations in developmental science. Human Development, 56, ­291–​­318. https://­doi.org/­10.1159/­0 00355022 ­Nepomuceno-​­Fernández, Á., ­Soler-​­Toscano, F., & ­Velázquez-​­Quesada, F. R. (­2014). The fundamental problem of contemporary epistemology. Teorema: Revista Internacional de Filosofía, 23(­2), ­89–​­103. Nersessian, N. J. (­1999). ­Model-​­based reasoning in conceptual change. In L. Magnani, N. Nersessian & P. Thagard (­Eds.) ­Model-​­based reasoning in scientific discovery (­­pp. ­5 –​­22).Springer. ­Neves-​­Pereira, M. S. (­2019). Microgenetic analysis and creativity: Analyzing psychological change processes. In I. Lebuda & V. P. Glăveanu (­Eds.), The Palgrave Handbook of Social Creativity Research (­­pp. ­145–​­161). Springer International Publishing. https://­doi.org/­10.1007/­­978-­​­­3 -­​­­319-­​­­95498-​­1_10 Ohlsson, S. (­1992). ­Information-​­processing explanations of insight and related phenomena. In M. T. Keane & K. J. Gilhooly (­Eds.), Advances in the Psychology of Thinking (­­pp. ­1–​­44). Harvester Wheatsheaf. Olma, S. (­2016). In Defence of Serendipity. Repeater. Park, W. (­2017). Abduction in Context. Springer. Pasteur, L. (­1854). Oeuvres de Pasteur / Réunies par Pasteur ­Vallery-​­Radot. https://­upload.wikimedia.org/­ wikipedia/­commons/­6/­62/­Louis_Pasteur_Universit%C3%A9_de_Lille_18541857_dans_les_champs_ de_l%27observation_le_hasard_ne_favorise_que_les_esprits_pr%C3%A9par%C3%A9s.pdf Paul, E. S., & Kaufman, S. B. (­Eds.) (­2014). The Philosophy of Creativity: New Essays. Oxford University Press. Pennisi, A., & Falzone, A. (­2020). Presentation. In A. Pennisi & A. Falzone (­Eds.), The Extended Theory of Cognitive Creativity: Interdisciplinary Approaches to Performativity (­­pp. ­1–​­5). Springer International Publishing. https://­doi.org/­10.1007/­­978-­​­­3 - ­​­­030-­​­­22090-​­7_1 Pierce, C. S. (­1931). Collected Papers, Vols. 1 ­ –​­6, Hartshorne and Weiss, P. Cambridge: Harvard University Press. Pinker, S. (­2003). Language as an adaptation to the cognitive niche. In M. H. Christiansen & S. Kirby (­Eds.), Studies in the Evolution of Language. Language Evolution (­­pp. ­16–​­37). Oxford University Press. Psillos, S. (­ 1996). Scientific realism and the ‘­ pessimistic induction’.  Philosophy of Science,  63(­S3), ­S306–​­S314. Race, T, & Makri, S. (­2016). Making room for serendipity. In Tammera Race & S. Makri (­Eds.), Accidental Information Discovery: Cultivating Serendipity in the Digital Age (­­pp. ­15–​­26). Chandos. ­Reiter-​­Palmon, R. (­2021). ­Chapter ­3 -​­leading for team creativity: Managing people and processes. ­ eiter-​­Palmon & J. C. Kaufman (­Eds.), Creative Success in Teams (­­pp. ­33–​­54). In A. S. McKay, R. R Academic Press. https://­doi.org/­10.1016/ ­­B978- ­​­­0 -­​­­12- ­​­­819993- ​­0.­0 0003-​­5 Ross, W. (­2022). Heteroscalar serendipity and the importance of accidents. In W. Ross & S. Copeland (­Eds.), The Art of Serendipity (­­pp. ­75–​­99). Palgrave MacMillan. Ross, W. (­2023). Serendipity. In V. P. Glǎveanu (­Ed.), The Palgrave Encyclopedia of the Possible. Palgrave MacMillan. Ross, W., & ­Vallée-​­Tourangeau, F. (­2021). Microserendipity in the creative process. Journal of Creative Behavior, 55(­3), ­661– ​­672. https://­doi.org/­10.1002/­jocb.478 Ross, W., ­Vallée-​­Tourangeau, F., & Glǎveanu, V. P. (­2020). Collaboration. In M. A. Runco & S. R. Pritzker (­Eds.), Encyclopedia of Creativity. Runco, M. A., & Jaeger, G. J. (­2012). The standard definition of creativity. Creativity Research Journal, 24, ­92–​­96. https://­doi.org/­10.1080/­10400419.2012.650092 Sawyer, R. K. (­2018). How artists create: An empirical study of MFA painting students. The Journal of Creative Behavior, 52, ­127–​­141. https://­doi.org/­10.1002/­jocb.136 Schurz, G. (­2008). Patterns of abduction. Synthese, 164(­2), ­201–​­234. Seifert, C. M., Meyer, D. E., Davidson, N., Patalano, A. L., & Yaniv, I. (­1994). Demystification of cognitive insight: Opportunitsic assimilation and the p­ repared-​­mind hypothesis. In R. J. Sternberg & J. Davidson (­Eds.), The Nature of Insight (­­pp. ­65–​­124). MIT Press. Shank, G. (­1998). The extraordinary ordinary powers of abductive reasoning. Theory  & Psychology, 8(­6), ­841– ​­860. https://­doi.org/­10.1177/­0959354398086007 Silver, S. (­2015). The prehistory of serendipity, from bacon to walpole. Isis, 106(­2), ­235–​­256. https://­ doi.org/­10.1086/­681977 Sjöholm, J. (­2014). The art studio as archive: Tracing the geography of artistic potentiality, progress and production. Cultural Geographies, 21, ­505–​­514. https://­doi.org/­10.1177/­1474474012473060 Smiley, T. J. (­1973). What is a syllogism? Journal of Philosophical Logic, 2(­1), ­136–​­154.

63

Wendy Ross and Selene Arfini Steffensen, S. V., ­Vallée-​­Tourangeau, F.,  & ­Vallée-​­Tourangeau, G. (­2016). Cognitive events in a ­problem-​­solving task: A qualitative method for investigating interactivity in the 17 Animals problem. Journal of Cognitive Psychology, 28(­1), ­79–​­105. https://­doi.org/­10.1080/­20445911.2015.1095193 Tanggaard, L., & Beghetto, R. A. (­2015). Ideational pathways: Toward a new approach for studying the life of ideas. Creativity. ­T heories – ­​­­Research -​­ Applications, 2(­2), ­129–​­144. https://­doi.org/­10.1515/ ­­ctra-­​­­2015-​­0 017 Thagard, P. (­2012). Creative combination of representations: Scientific discovery and technological invention. In R. Proctor & E. J. Capaldi (­Eds.), Psychology of Science: Implicit and Explicit Processes (­­pp. ­389–​­404).Oxford University Press. Vallée-Tourangeau, F., & March, P. L. (­2020). Insight out: Making creativity visible. The Journal of Creative Behavior, 54(­4), ­824– ​­842. https://­doi.org/­10.1002/­jocb.409 ­Vallée-​­Tourangeau, F., & ­Vallée-​­Tourangeau, G. (­2020). Mapping systemic resources in problem solving. New Ideas in Psychology, 59, ­1–​­10. https://­doi.org/­10.1016/­j.newideapsych.2020.100812 ­Vallée-​­Tourangeau, G., & V ­ allée-​­Tourangeau, F. (­2017). Cognition beyond the classical information processing model: Cognitive interactivity and the systemic thinking model (­SysTM). In Stephen J. Cowley & F. ­Vallée-​­Tourangeau (­Eds.), Cognition Beyond the Brain (­­pp. ­133–​­155). Springer International Publishing. https://­doi.org/­10.1007/­­978-­​­­3 -­​­­319- ­​­­49115- ​­8 Van Andel, P. (­1994). Anatomy of the unsought finding. The British Journal for the Philosophy of Science, 45, ­631– ​­648. http://­w ww.jstor.org/­stable/­687687 Viola, T. (­2016). Peirce on abduction and embodiment. In R. Madzia & M. Jung (­Eds.), Pragmatism and Embodied Cognitive Science (­­pp. ­251–​­268). De Gruyter. https://­doi.org/­10.1515/­­9783110480238-​­013 Weisberg, R. (­2010). The study of creativity: From genius to cognitive science. International Journal of Cultural Policy, 16(­3), ­235–​­253. https://­doi.org/­10.1080/­10286630903111639 Woods, J. (­2007). Ignorance and semantic tableaux: Aliseda on abduction. Theoria. Revista de Teoría, Historia y Fundamentos de La Ciencia, 22(­3), ­305–​­318. http://­doi.org/­10.1387/­theoria.451 Woods, J. (­2010). Defeasible reasoning. In C. Reed & C. W. Tindale (­Eds.), Dialectics, Dialogue and Argumentation. An Examination of Douglas Walton’s Theories of Reasoning (­­pp.  ­239–​­262). College Publications. Yaqub, O. (­2018). Serendipity: Towards a taxonomy and a theory. Research Policy, 47, 1­ 69–​­179. https://­ doi.org/­10.1016/­j.respol.2017.10.007

64

5 THE TWO FACES OF CURIOSITY IN CREATIVE COGNITION Curiosity1, Curiosity2 (­and Their Interaction) Janet Metcalfe and William James Jacobs

Introduction The Oxford English Dictionary defines curiosity as ‘­a strong desire to know.’ In this chapter, we delineate two distinctly different ways in which that term has been used and argue that what has often been taken as a singular impulse is, in fact, underpinned by two discrete constructs. Our focus on teasing apart differences between these constructs, which are both associated with the term ‘­curiosity,’ echoes a similar growing concern in several areas of science, including developmental psychology (­e.g., Chu  & Schulz, 2020) and machine reinforcement learning modeling (­e.g., Sutton  & Barto, 2018; Wilson et  al., 2014). Early research on curiosity by Berlyne (­1950, 1954, 1962) also pointed to the possibility of two different kinds of curiosity. He labeled the desire to find out the answer to something, in which a specific target is sought, epistemic curiosity. We call such ­goal-​­directed curiosity, in which the problem solver is seeking out or tracking a particular answer, Curiosity1. Berlyne contrasted epistemic curiosity to more discursive information seeking or novelty s­ eeking – ​­a kind of curiosity that he called perceptual curiosity. Roughly, Berlyne’s perceptual curiosity corresponds to what we here call Curiosity2, although we propose that the latter should not be defined by its being sensory or perceptual rather than cognitive.1 Instead, we posit that it is better characterized by its ostensibly undirected, ­goal-​­indifferent nature. The desire to find out the answer to a puzzle or question of interest, a desire that gets stronger and more motivating as the individual feels that he or she is approaching the solution, then, is Curiosity1. The inclination to cast about intellectually, to randomly or quasi-​­ ­ randomly investigate new territory, apparently uncompelled and unhindered by ­goal-​­directed impulses, is Curiosity2. Both, more or less, fit the dictionary definition of ­curiosity – ​­a desire to k­ now – ​­but these two forces can work in opposite directions. The first is drawn almost magnetically to the object of one’s desire. The ­a s-­​­­yet-​­unknown answer to the problem seems to beckon the individual in a particular direction, with the feeling of curiosity increasing with intuited proximity. It disavows distractions or detours. The second is all about detours and distractions and disdain for the relentless pursuit of goals. It is play, and it is mindless of goals (­or else it would not be play). These two diametrically opposed inclinations have both been called curiosity, and though different from one another, both appear to be needed for creative cognition. DOI: 10.4324/9781003009351-6

65

Janet Metcalfe and William James Jacobs

After delineating some of the behavioral characteristics, memorial consequences, and brain correlates of each of these two faces of curiosity, we will propose that, in many creative efforts, both of these two modes are necessary for generating truly creative solutions. We suggest that Curiosity1 and Curiosity2 interact in a synergistic manner. Although they are broadly applicable in many domains of learning and discovery, for illustration of their interaction, we will examine the processes involved in solving insight problems and stumpers.

Curiosity1 Most laboratory research on curiosity that has been carried out in a­ dult-​­based cognitive neuroscience has investigated Curiosity1. Participants are provided with questions or puzzles, to which there are answers, and the research focus is on the internal processes, the cognitive, metacognitive, and neural correlates, the environmental conditions, the consequences, and the personality characteristics associated with people either being curious and engaged or bored and indifferent about knowing the answers to those particular problems. Most frequently, the questions that are used are trivia questions: sometimes quirky trivia questions (­e.g., ‘­W hat is the only lizard that has a voice?’: Fastrich et al., 2018), but also more mundane general information questions (­e.g., ‘­W ho was the second President of the United States?’: Bloom et al., 2018). People are typically asked to come up with the answer if they can and also to indicate how curious they are to know the ­a nswers – ​­sometimes by simply stating their curiosity, but sometimes by indicating how much they would pay to know, or by choosing particular items for study over others.

Experimental Findings An influential study of this type and one that triggered much f­ollow-​­up research was thestudy using functional Magnetic Resonance Imaging (­fMRI) and pupillary response conducted by Kang et al. (­2009). In a series of experiments, participants were presented with a set of trivia questions and tried to think of (­or, in some experiments, state) the answer to each. They then rated their curiosity or desire to have the answer given to them and their confidence that they had produced the correct answer to each question. Then the correct answer was given. Later, participants were tested for their answers. The results showed that people’s curiosity was low when they were either very sure that they had the right answer or when their confidence judgments were very low. Curiosity ratings were high at middle levels of confidence; that is, curiosity was characterized by an inverted U ­ -​­shaped function in relation to confidence. A pupillary response was associated with high but not low levels of curiosity. Brain regions associated with reward expectation circuitry (­left caudate, bilateral prefrontal cortex including the inferior frontal gyrus, putamen, and globus pallidus) showed greater activation for high versus low curiosity items during the p­ re-​­feedback phase of the ­experiment – the ​­ time when people were experiencing curiosity. The neural correlates of curiosity, including the activation of the reward circuitry observed by Kang et al. (­2009), have held up very well. An association between feelings of curiosity, in this Curiosity1 paradigm, and anticipatory reward c­ ircuitry – ​­the dopamine ­system – ​­has been found in many studies (­­Bromberg-​­Martin & Hikosaka, 2009; Gruber et al., 2014; Gruber & Ranganath, 2019; Lau et al., 2020; Marvin & Shohamy, 2016). Later memory for the answers to those questions on which participants had been wrong in the initial phase of the experiment was higher when curiosity had been high. Areas associated 66

Two Faces of Curiosity in Creative Cognition

with explicit memory (­e.g., parahippocampal gyrus, hippocampus) were activated before and after answer presentation, particularly for items for which the person’s first response was incorrect rather than correct and about which they were curious rather than not curious. Behaviorally, memory was enhanced for those answers about which people were curious. This is a ubiquitous finding in the literature (­Bloom et al., 2018; Fandakova & Gruber, 2020; Galli et al., 2018; Grossnickle, 2016; Gruber et al., 2014; Lau et al., 2020; McGillivray et al., 2015; Stare et al., 2018; Von Stumm et al., 2011; Wade & Kidd, 2019). Furthermore, the memorial consequences of curiosity are not restricted to memory for the answer itself. In a particularly interesting study, Gruber et al. (2014, and see Murphy et al., 2021) interposed a picture of an unrelated face while people were considering it but before they were given the answer to a puzzle, that is, while they were still curious. They found that memory for the face was enhanced when people had been curious about the solution to the puzzle as compared to when they had not been especially ­curious – ​­suggesting that curiosity itself heightens attention and that this enhanced engagement can improve memory even for unrelated information. The state of Curiosity1 appears to be a special state that has memory repercussions. Often, the receipt of the solution to the problem about which the individual was curious is thought to be rewarding/­pleasurable (­Litman, 2005). Its receipt is associated with dopamine system activation (­­Bromberg-​­Martin & Hikosaka, 2009; Kang et al., 2009; Lau et al., 2020; and see Daw & Tobler, 2013; Delgado, 2007). However, sometimes the receipt of the solution can be disappointing (­Loewenstein, 2007). Researchers agree, though, that once the answer is known, curiosity is extinguished (­Berlyne, 1962; Loewenstein, 1994).

Relation to TOTs The ­t ip-­​­­of-­​­­the-​­tongue (­TOT) state is associated with curiosity (­Metcalfe et al., 2017). TOTs are expressed when the person is in a phenomenological state of almost knowing or of being on the brink of knowing. Curiosity1 seems to be related to an internally experienced goal gradient in which the cues that indicate how close the individual is to the reward are metacognitive rather than environmental or learned. Although Kang et  al. (­2009) specifically asked about TOT states, in their original study, the data were too sparse to allow analyses. To attain more TOT data, Metcalfe et al. (­2017) conducted speeded response studies in which the participants had to answer the TOT and curiosity questions very quickly (­before TOTs could be resolved), and these studies allowed examination of the relation between TOT states and curiosity. They found that being in a TOT state is strongly related to curiosity. Regardless of whether participants’ answers were correct, or whether they involved errors of commission or omission, if they reported being in a TOT state, they were approximately twice as curious to know the answer as they were if they reported that they were not in a TOT state. Brooks et al. (­2021) also showed that the feeling of knowing was associated with curiosity. Similarly, in a related study, Litman et al. (­2005) found that when participants gave judgments that they were in a TOT state, they were more curious than when their judgments indicated that they did not know. TOTs are defined as the subjective feeling of being extremely close to having the answer (­Brown, 1991; Schwartz, 1999; Schwartz & Cleary, 2016; Schwartz & Metcalfe, 2011). As ­ -​ such, the association between TOTs and curiosity seems inconsistent with the inverted U ­shaped confidence/­curiosity function found by Kang et al. (­2009), in which intermediate levels of confidence judgments were most associated with curiosity. However, this apparent discrepancy between the finding that people are extremely curious when in a TOT state 67

Janet Metcalfe and William James Jacobs

and the finding that medium confidence is associated with curiosity may be more apparent than real. In the Kang et al. (­2009) study, people made their curiosity judgments without indicating whether they had retrieved or already knew the answer. The reported data did not eliminate or separate out items that were known. As noted above, researchers of curiosity agree that once the answer is known, curiosity is sated and, as such, falls to a very low level (­e.g., Berlyne, 1962; Loewenstein, 1994). The very high levels of confidence in Kang et al.’s study included both the items that people knew and knew that they knew (­for which their curiosity should have been very low) and the items that they ‘­a lmost knew’ or were on the verge of knowing (­i.e., TOT items, for which curiosity should have been very high). The averaging of these two types of items (­a long with a noisy known/­unknown boundary) is sufficient to account for the inverted ­U-​­shaped function that was observed. Without this averaging, a s­calloped-​­shaped curiosity/­confidence function might be expected if curiosity increases monotonically from not knowing, to internally perceiving some knowledge, to the TOT state, and then falls sharply once the individual knows the answer. In a ­follow-​­up to the earlier TOT findings, Bloom et  al. (­2018) conducted an e­ vent-​ ­related potential (­ERP) investigation. They found that, following the presentation of the answer as feedback, participants exhibited a strong P3 and late positivity ERP tracing that was (­a) distinctive to having just been in the curious/­TOT state and that (­b) predicted later enhanced memory performance on those items. These findings are consistent with Kang et al.’s (­2009) ­curiosity-​­related ­post-​­feedback fMRI findings implicating enhanced memory encoding for the answers to questions about which the individual was curious. Curiosity1, then, appears to be similar to reward/­­goal-​­gradient situations, where excitation is most prominent when the person feels they are close to the goal. Such goal gradients have been extensively investigated in the animal learning literature. The difference between the human curiosity situation and the animal g­ oal-​­gradient/­­primary-​­reward situation is that the cues that humans use to indicate how close they are to the goal, or to the answer to the problem, are metacognitive. There are many examples in the literature of increased interest and excitement as n ­ on-​ ­human animals approach what they think will be a reward. Hull (­1932, 1934), for example, observed that rodents trained to run down an alleyway to obtain access to a bit of food run slowly at the beginning of the alley but accelerate their running rate as they approach the goal. He labeled the phenomenon the ­goal-​­gradient effect. Pavlov and Anrep (­1927) observed a similar phenomenon in dogs. Pavlov sounded a metronome and, after a fixed temporal interval, delivered a bit of food to the dog’s mouth. When the sound of the metronome extended for some period (­e.g., 45 seconds), ­well-​­trained dogs initially produced little or no saliva but produced increasing quantities as the time for food delivery approached. This pattern occurred both when the metronome was present for the entire 45 seconds (­long delay conditioning) and when the metronome sounded briefly and then went silent for the remainder of the 45 second interval (­t race conditioning). Ferster and Skinner (­1957) also described a ­goal-­​ ­­gradient-​­like effect in rodents and other species. W ­ ell-​­trained subjects, who were required to wait for a fixed temporal interval before a response produced brief access to food, responded slowly at the beginning of the interval but emitted an accelerated rate of responding toward the end of the interval (­see Kirkpatrick & Church, 2003). These data converge on the finding that many animals expend an increasing amount of effort as they approach a known goal. Similar patterns occur in humans with primary rewards. For example, people enrolled in a ‘­Buy 10 Coffees; Get One Free’ reward program buy coffee more frequently as they approach the ­Free-​­Coffee reward (­K ivetz et al., 2006). Importantly, even the illusion of progress toward a goal produces a ­goal-​­gradient behavioral pattern (­see Cryder et al., 2013; Kivetz et al., 68

Two Faces of Curiosity in Creative Cognition

2006). In each of these examples, it is as if the organism is becoming increasingly excited as the perceived proximity to a spatial, temporal, or illusory goal increases.

Implications for Learning Interestingly, many researchers have proposed that learning itself is optimized by focusing on those items that are closest to the goal of learning. Piaget (­1954), for example, outlined the constraints of assimilation and a­ ccommodation – the ​­ mechanisms that, in his framework, power the child to new cognitive levels. The new constructs and schemata need to be close enough to those that the child has already mastered to allow accommodation; if they are too far away, the child does not adapt. Similarly, Vygotsky (­1978) detailed a ‘­zone of proximal development’ that was just outside what the child already knew. It was within this zone that scaffolding was most effective in enhancing learning. Atkinson (­1972) detailed that there is a transition zone in which items are almost known. These are the items that are most open to learning. Being in this zone is crucial for learning. Research from our lab has pointed to a similar idea. It appears that there is a region of proximal ­learning – ​­a region in which people have not fully mastered the t­ o-­​­­be-​­learned materials, yet in which those materials are not overly ­d ifficult – ​­in which people choose to focus their study and learning. For instance, n ­ on-​­expert children and adults tend to focus study on ­a s-­​­­yet-​­unlearned materials that are easy rather than difficult (­Metcalfe, 2002; Metcalfe & Kornell, 2003, 2005; Son  & Metcalfe, 2000). With expertise, the easy items become too easy (­a s they become known) and the focus shifts to more difficult items (­i.e., to the easiest ­a s-­​­­yet-​­unknown items). When given a choice about which items they would like to find out about, people choose the items that they perceive they are closest to knowing but have not yet mastered, rather than items that they perceive as being farther away from learning (­Kornell & Metcalfe, 2006). Their choices are consistent with the idea that their curiosity follows a goal gradient that increases with proximity to the goal of learning. Studying materials in their own region of proximal learning reduces mind wandering (­Xu & Metcalfe, 2016), as would be expected when people are curious rather than either bored or frustrated. Furthermore, when people are allowed to study as dictated by their own choice of what to learn, as opposed to being given the items for study that they either did not choose themselves or items to which they had assigned low rather than high judgments of learning, their learning is less effective as measured by performance on a subsequent test (­Kornell & Metcalfe, 2006). Thus, people appear to know, metacognitively, which items will repay their study efforts: they are the items that, metacognitively, are almost but not quite ­k nown – the ​­ very items that elicit Curiosity1. Bjork (­2017) has posited that some kinds of difficulties are desirable for learning. This is undoubtedly true. Learning situations that are too ­easy – ​­as occur when people already know the answer, for ­example – ​­do not help learning. But the ‘­desirable difficulties’ framework leaves unanswered the question of which particular difficulties are or are not desirable. The region of proximal learning/­Curiosity1 framework seems to provide a partial answer. Those items that are on the threshold of being known, but are not yet fully mastered, appear to be those that offer the greatest learning returns. Once curiosity is quenched, however, further study is no longer sought. The region of proximal learning framework, then, may help define desirability. It appears to be ­co-​­extensive with the experience of Curiosity1. Finally, the notion of gradient descent is seen in machine learning (­reinforcement learning) models (­a lthough the incentive/­motivational consequences of being near to the goal are not so clear in Artificial Intelligence applications). In those models, the ‘­agent’ traverses 69

Janet Metcalfe and William James Jacobs

the space by means of a gradient descent algorithm. In a w ­ ell-​­trained network, the path smoothly takes the agent closer and closer until it finally reaches the goal. Essentially, the job of learning or training in such models is to sculpt out a landscape whereby the agent can get to the goal easily and efficiently by taking the highest probability decision choices, that is, by smooth and unobstructed gradient descent.

Education The Curiosity1 paradigm is highly relevant for pedagogy. Often teachers will pose puzzles to students in the hopes of engaging their interest and c­ uriosity – ​­which is thought by the teachers (­a nd affirmed by the data on Curioisty1) to facilitate learning because the students are more engaged, and that engagement matters for memory. In ­Curiosity1-​­driven learning, there is a question, and there is an ­a s-­​­­yet-​­unknown but knowable answer. If the teacher can arrange the materials and the situation so that the individual feels himself or herself to be close to the answers, they should feel Curious1. And because they are in a Curious1 state, they should be motivated to devote time and energy to finding out the answer. Once they get the answer, they will feel pleasure (­which, if it happens enough, may even translate into a love of learning). In addition, they will remember the specific answers well. As such, this paradigm seems an apt formula for engaging students and enhancing learning. There is, however, another sense in which the term curiosity is used, and it is the opposite of this. We call it Curiosity2. It involves taking the path less traveled.

Curiosity2 One way to approach Curiosity2 is to consider the behavior of a person who is high on it. One of the best exemplars in Western literature of a Curious2 person is Odysseus, the Greek hero who is legendary for his unquenchable exploratory spirit and his cleverness at overcoming apparently insurmountable obstacles (­Homer, 1999, tr. Fagles). Although, after winning the Trojan War, Odysseus purports to have the goal of getting back to Ithaca to be with his beloved, ­long-​­suffering, and faithful wife Penelope, his Curiosity2 is anything but helpful in fostering his attainment of this goal. Indeed, the distractions brought on by his wandering not only do not facilitate goal acquisition, but they delay him for 20 years. His d­ alliances –​ ­from the temptations of the lotus flowers, to battling the Cyclops, enjoying the Sirens’ song while tied to the mast, passing between the Scylla and Charybdis, and his s­ even-​­year detour on the island of Ogygia with the beautiful witch ­Calypso – ​­offer more pleasure for him than his actual return. Even the successful attainment of the nominal goal proves unsatisfying. He sets off for new adventures almost immediately upon reaching Ithaca. His drive to reach the goal of returning home, then, is a poor second to the adventure of the voyage i­tself –​ ­exploring strange new worlds and experiencing every wonder that might present itself, for good or for ill. What we here call Curiosity2, then, is the dominant meaning of the term ‘­curiosity’ in popular depictions. This is the curiosity that beset humankind with the biting of a fruit from the Tree of Knowledge even at the risk of suffering and death, resulting in dissatisfaction and even boredom with the perfect and bountiful garden. It is the kind of curiosity underlying Pandora’s need to open the box containing Zeus’ cruel wedding gift. We see it in the imaginary voyages of Star Trek and the real wanderings of the Mars Rover Curiosity. It is this kind of ­ eorge – ​­depicted in so many beloved curiosity that gets Margret and H. A. Rey’s monkey, G children’s ­books – ​­into trouble. It was this kind of curiosity that killed the cat. 2 We argue, 70

Two Faces of Curiosity in Creative Cognition

here, that although this kind of curiosity, which we call simply Curiosity2, is not the same as Curiosity1, it, too, is necessary for higher creativity. Curiosity2, however, is not restricted to the popular domain. In the machine reinforcement learning models there is a ­well-​­known conflict between exploitation and exploration (­Wilson et al., 2014; Wilson et al., 2021, cf., Peterson & Verstynen, 2020). The latter relates to Curiosity2. (­The former aligns more closely with Curiosity1.) The conflict arises because in determining which choices or actions to take, there is a ­so-​­called ‘­g reedy’ action: the one with the highest known value. However, there are also n ­ on-​­greedy choices that have lower known values. So long as the ultimate values of each choice are known, the greedy action is the preferred choice. However, under uncertainty, it may sometimes be beneficial for the agent to opt for a ­non-​­greedy action, that is, for one that does not have the highest known value but which might enable the agent to improve estimates overall and hence end up leading to a better solution. Exploitation basically refers to taking the greedy action or the pathway that leads most directly to the rewarding goal. It has the ­best-​­known chance of allowing the agent to attain the g­ oal – it ​­ has the steepest downhill slope. But taking the greedy action can risk the possibility that there is an even better outcome that would result if the ­non-​­greedy option were chosen, and that better option remains unrecognized. Exploration involves opting for a ­non-​­greedy choice. According to Sutton (­2017): ­ on-​­greedy If you have many plays yet to make, then it may be better to explore the n actions and discover which of them are better than the greedy action. Reward is lower in the short run, during exploration, but higher in the long run because after you have discovered the better actions, you can exploit them. This view suggests that exploration, or what we here call Curiosity2, has the purpose of ultimately optimizing goal attainment or ‘­rewards.’ For humans, such exploration need not be directed at proximal rewards. It may, nevertheless, have unanticipated favorable consequences because the enrichment of the person’s knowledge base may serendipitously enhance later problem solving and creativity (­Austin, 1979). For example, Nadel (­2021) has proposed that exploration is triggered when a person or animal encounters something unexpected. The results of the exploration are used to update and enrich the animals’ schemata, including their cognitive maps. This updating might have beneficial consequences, but such consequences were not the goal of the animal or person at the time. But regardless of whether the exploratory path is taken instrumentally or not, people and sometimes ­a nimals – ​­and particularly those whom we call curious people and animals (­Fowler, 1965; Litman, 2008) – o ​­ ften take ­non-​­g reedy, ­low-​­probability actions. Exploration often ostensibly leads away from the goal instead of toward it. People do this even when the probabilities or known rewards on that path are much smaller and much more risky. This ‘­explore’ option is Curiosity2. It has distinct brain consequences (­Cohen et al., 2007). And, we argue, it differs from Curiosity1, which is more in line with exploitation. Much of the psychological research investigating what we here call Curiosity2 has been subsumed under investigations of exploration, play (­m ainly children’s), or novelty seeking. We will touch on each of these below. With the exception of novelty seeking, which is often formulated as being for the purpose of knowing (­see, for example, Fitzgibbon et al., 2020; Gottlieb et al., 2013), exploration and play are often not framed in terms of the individual’s ‘­strongly wanting to know.’ They may simply be ­f un – ​­with no compelling imperative of increasing knowledge implied at all. Indeed, in neither exploration, play, nor novelty seeking is 71

Janet Metcalfe and William James Jacobs

there a discrete goal, a particular answer, or a primary reinforcer being sought, even though the experiences of exploring, playing, or seeking novelty may be pleasurable (­see Litman, 2005).

Exploration As early as 1930, Tolman and his students were investigating learning in animals. At the time, the dominant view was that learning only occurred as a result of reinforcement (­e.g., Thorndike, 1927). Tolman, though, suspected that animals might also just explore for no reason at all and learn implicitly (­w ithout explicit rewards being provided to them). In one of the first demonstrations, Tolman and Honzik (­1930) placed hungry rats (­call them ‘­E’ for ‘­explorer’ rats) in a complex maze and permitted them to simply wander around the maze for an hour on each of ten days. An alternative, equally hungry control group of ‘­rewarded’ ‘­R’ rats received reinforcement after successfully getting to the target location on every ­trial  – ​­spending the same amount of time in the maze but learning by g­ oal-​­seeking and getting a reward each time for doing it. The R rats, clearly, should have been the better learners. On the eleventh day, the researchers placed a small bit of food in the goal box at the end of the maze and placed R rats or E rats in the start box and waited to see what would happen. Upon discovering the food, the E ­rats – ​­those that had previously explored the maze without ­reinforcement – ​­found their way to the goal box more quickly than the R rats that had been trained to run in the maze for reward. The unrewarded time spent exploring had beneficial effects in the long run. There was no evidence that the E rats had explored for the purpose of getting the reward, whose existence was not even known to them at the time of the exploratory behavior. The rats apparently learned the layout of the maze with no obvious ‘­reinforcement’ (­Epstein et al., 2017; Qiu et al., 2020). This phenomenon has come to be known as latent learning (­or behaviorally silent learning) and may be one of the ­long-​­term benefits of Curiosity2. Humans, too, of course, remember the events that happen to them ­ eeded – ​­even and the things that they did, and they can use those remembrances later if n when they were not initially engaging in actions for the purpose of attaining a reward.

Play Children are often extolled for their curiosity. And yet they seem, in many ways, to be just meandering. Essentially, Curiosity2, in the guise of play, is unconstrained by time pressures, by the need for a concrete solution, by the desire to get reward, or by the imperative to learn new skills or information or to in any way better themselves. Chu and Shultz (­2020) suggest that despite the many possible beneficial reasons for children to engage in p­ lay – ​­that it will enhance later learning ( ­Jirout & Klahr, 2012), that it is practicing adult skills, from mechanical to mothering, in order to make those behaviors more fluent later (­MacDonald, 2007), that it will result in improved school grades (­Pellegrini, 1993; Ramstetter et al., 2010), that it develops physical skills (­Pellegrini & Smith, 1998), that it advertises fitness (­Chu & Schulz, 2020; Panksepp, 1981; Pellis & Iwaniuk, 2000), that it improves social bonding (­De Waal, 1986; Whiten  & De Waal, 2018), that it helps to develop cognitive maps (­Nadel, 2021), that it improves causal perception and learning (­Buchsbaum et al., 2012; Gopnik & Walker, 2013), or even that it will eventually enhance serendipity (­Austin, 1979) – ​­the most likely possibility is that play has no function at all. Indeed, they (­Chu & Schulz, 2020) argue that we should take their claim about a lack of purpose seriously. Panksepp and Biven (­2012) and

72

Two Faces of Curiosity in Creative Cognition

Pellis and Pellis (­2007) concur in the view that play is without ulterior purpose. By this view, any ­survival-​­related benefit (­or cost) from play comes about by happenstance. In particular, though, the goal of ­single-​­mindedly solving a specific problem is antithetical to play.

Novelty Seeking Curiosity2 can also be framed as novelty seeking as opposed to g­ oal-​­seeking. Novelty seeking refers to the tendency of humans and animals to explore novel and unfamiliar stimuli and environments. There is abundant evidence that people and animals alike seek novelty. Fantz (­1965), for instance, showed that starting at two months of age, babies prefer a changing (­or novel) sequence of visual display over a constant sequence. Infants look longer at events that violate their expectations of the world (­Spelke et al., 1992). Indeed, the infant preference for novelty is so pervasive that it has been used since Fantz’ seminal experiments to investigate preverbal infants’ knowledge (­Spelke, 1985). Visual novelty seeking has been studied intensively in the context of curiosity (­see, for example, Gottlieb et  al., 2013; Gottlieb  & Oudeyer, 2018). And while the ocular motor system is an excellent model for the study of Curiosity2, novelty seeking is not restricted to the visual system. When children and adults alike seek novelty, they engage sensory, motor, cognitive, emotional, and other systems as well. The reverse of novelty responding and novelty seeking is habituation to repetitive or already learned stimuli. The experience attendant upon novelty appears to be rewarding in its own right and is related to the dopamine system (­Glimcher, 2011; Hazy et  al., 2010). Furthermore, dopamine modulates novelty seeking in explore/­exploit choices, enhancing n ­ ovelty-​­driven value (­Costa et al., 2014). While novelty seeking appears to be innate, if it is excessive, it can be dysfunctional. Costa et al. (­2014) suggest that excessive novelty seeking, such as is seen in impulsivity and behavioral addictions, might result from increases in dopamine due to diminished reuptake (­cf. Berridge, 1996; Panksepp & Biven, 2012, for discussion of alternate ­dopamine-​­related mechanisms).

Synthesis Köhler, in The Mentality of Apes (­1925) notes that …we do not speak of being intelligent, when human beings or animals attain their objective by a direct unquestionable route which without doubt arise naturally out of their organization. But what seems to us ‘­intelligence’ tends to be called into play when circumstances block a course which seems obvious to us, leaving open a roundabout path which the human being or animal takes, so meeting the situation. (­­p. 4) In addition to physical routes or paths, such as those Köhler used in his demonstrations (­a nd see Tolman, 1932), this idea that there are two components to creative problem ­solving –​ ­the direct and obvious route and the detour that needs to be taken if the obvious route is ­obstructed – ​­occurs in insight problems (­Danek, 2018; Metcalfe, 1986; Metcalfe & Wiebe, 1987), and stumpers (­­Bar-​­Hillel et al., 2018), and perhaps in the new understanding that may come about in scientific revolutions (­cf., Brannigan, 1981; Kuhn, 1962). Köhler describes a dog in a problem situation in which food is dropped close to a wire fence so that she was separated from it only by the wire:

73

Janet Metcalfe and William James Jacobs

She stood seemingly helpless, as if the very nearness of the object and her concentration thereon... blocked the ‘­idea’ of a wide circle round the fence; she pushed again and again with her nose at the wire fence, and did not budge from the spot. (­1925, ­p. 14) The dog seemed to be exhibiting a surfeit of Curiosity1. In contrast, a toddler of one year and three m ­ onths – just ​­ barely walking on her o ­ wn – ​­put in a similar situation with a desired object on the other side of a barrier can also engage Curiosity2. She “…first pushes towards the object, i.e., against the partition, then looks round slowly, lets her eyes run along the blind alley, suddenly laughs joyfully, and in one movement is off on a trot around the corner to the objective” (­Köhler, 1925, ­p. 14, and see Kabadayi et al., 2018). Magic tricks (­Danek, 2018), stumpers (­­Bar-​­Hillel et  al., 2018), and insight problems (­Danek et al., 2014; Hedne et al., 2016; Ollinger & Knoblich, 2009) all exhibit this structure. Consider a classic insight problem in which the participant is told to imagine that she is a gardener and is given four sacred trees to plant with the only constraint being that they all need to be equally distant from one another. The problem is set up as if there were a direct path to the solution. The solver, no doubt, thinks about the trees and a nice symmetrical, ritualistic arrangement on the flat surface of most gardens. Most people think this is an easy problem when they are at this stage of solving (­Metcalfe, 1986; Metcalfe & Wiebe, 1987), since it is intentionally contrived to induce the participant to think within a routine and fairly simplistic schematic problem space, but one that will not lead to the solution. Because of this devious construction of the problem, people are induced to be in a state of Curiosity1: they think they almost have the solution. The problem may not even be perceived as being a problem on first reading. Furthermore, many participants may make an outright error at this point and arrange the trees in a square, neglecting to notice that the diagonals count, as do the edges, and that the lengths of these two are different. But once this difficulty or barrier is noticed and it is seen that the direct solution will not work, then it is a problem. Gestalt psychologists, Duncker (­1945) and Luchins (­1942), describe this state, when the problem is recognized but not yet solved, as functional fixedness due to einstellung, that is, a rigid mindset. A schematically bound mindset is problematic, though, only when it does not work. Most of the time, people can go directly to their goals, and Curiosity1 will be all that is needed. In the present case, though, once the blocker is acknowledged, the person or animal needs to back away from the beckoning goal and explore other p­ ossibilities – ​­as the dog could not but the toddler could. It is at this point that Curiosity2 needs to come into play. One of the most interesting insights to emerge from widespread efforts to develop artificial autonomous agents that can solve difficult problems such as face or voice recognition or play chess is that taking the most direct route to the ­goal – ​­simple gradient descent and the relentless pursuit of the ­solution – ​­is often less effective in the long run than playful Odyssean dalliance (­i.e., taking the ­non-​­greedy option). Sutton and Barto (­2018) described AI pioneer Selfridge’s ‘­r un and twiddle’ learning rule as ‘­keep going in the same way if things are getting better, otherwise move around.’ The explore/­exploit distinction has proven to be highly effective in models such as these, which encounter a problem if they only follow the steeper gradient (­or the higher probability pathway), guided by the exploit ­option – ​­namely, that they can get stuck in local minima. If this happens, they cannot reach an optimal solution. To overcome this limitation, the model needs to sample the explore option, which can shake the procedure out of the trap it is caught in. Play is often thought to be the domain of children and perhaps scientists and artists. But having the procedures in place for exploration may be beneficial in situations in which any 74

Two Faces of Curiosity in Creative Cognition

person senses that they can solve a problem and are close to getting the answer, and hence is highly motivated, as per Curiosity1, but the ‘­r un’ choice has led to a barrier. ‘­Twiddle’ is needed. Even with both Curiosity2 and Curiosity1 options available, though, there is no guarantee, in real scientific discovery as well as in daily life, that there is an ultimate or even a better solution. The individual may have the sense that they almost have the solution and be driven hard by Curiosity1. They may, nevertheless, hit an impasse. They may appropriately apply all of their knowledge gained from play situations and engage Curiosity2 exploration. But even with all this, they may still not solve the problem. Furthermore, this can happen to people who are both curious and knowledgeable; even brilliant scientists can be unsuccessful. Albert Einstein, for instance, thought that while quantum theory provided a useful description of part of the universe, he doubted its ability to explain all of physics (­Becker, 2019). Einstein could not accept the level of uncertainty in the universe implied by quantum theory. He was apparently driven by Curiosity1 to find a better and more correct theory that would overcome these concerns. He felt the solution was imminent, and he clearly knew how to explore all options that were available to him. He expressed his view that he had not yet found the correct solution in a letter to Max Born, in 1926, as ‘­The [Old One] does not play dice with the universe.’ Accordingly, Einstein spent most of the final 30 years of his life trying to formulate a better theory to replace quantum mechanics. He was unsuccessful.

Conclusion The term curiosity has been applied to two distinctly different constructs that, while both interesting and necessary for human creativity, sometimes work in directions opposite to one another. The first kind of curiosity, called Curiosity1, which is the kind most studied in the lab, is intensely g­ oal-​­directed and driven by the feeling state, or metacognition, that a solution is imminent. The phenomenology is of being hot on the trail of a breakthrough. Curiosity1 may be implicated in the ­sometimes-​­dogged perseverance exhibited by creative individuals in pursuit of their goal. In contrast, what we here call Curiosity2 is not g­ oal-​ ­seeking at all. Instead, if anything, it is novelty seeking. It entails ­free-​­form and undirected mental exploration, unconstrained by any notion of explicit goal or reward. It is ­play-​­like and embraces the wonder of childhood exploration, uncontaminated by needs, wants, or goals other than, perhaps, the joy of being stimulated by novelty. We have argued that both kinds of curiosity are vital to the creative processes of human beings, but that they differ from one ­another  – ​­served by different motivations and constraints and underpinned by different mental processes. Conflating them, as if they were guided by the same cognitive principles, may hamstring understanding of the creative process. Although the two have broad applicability to many situations, we suggest that their interaction is particularly salient in insight problem solving. Despite their distinctively different roles, we suggest that Curiosity1 and Curiosity2 work hand in hand in truly creative discovery.

Notes 1 The distinction made here between Curiosity1 and Curioisty2 is unrelated to the speed of thinking or to Kahneman’s (­2011) System 1 and 2. 2 As noted by the editor, it is possible that Curiosity1 is also implicated in cat killing: “­Say, it was curious in its pursuit of food and something happened in the course of that ­goal-​­attaining ­behavior – ​­this would be an example of Curiosity1 killing the cat, no?”

75

Janet Metcalfe and William James Jacobs

References Atkinson, R. C. (­l972). Optimizing the learning of a ­second-​­language vocabulary. Journal of Experimental Psychology, 96, ­124–​­129. Austin, J. H. (­1979). The varieties of chance in scientific research. Medical Hypotheses, 5, ­731–​­742. ­Bar-​­Hillel, M. Noah, T.,  & Frederick, S. (­2018). Learning psychology from riddles: The case of stumpers. Judgment and Decision Making, 13, ­112–​­122. Becker, A. (­2019). What Is Real? The Unfinished Quest for the Meaning of Quantum Physics. Basic Books. Berlyne, D. E. (­1950). Novelty and curiosity as determinants of exploratory behavior. British Journal of Psychology, 41, ­68–​­80. Berlyne, D. E. (­1954). A theory of human curiosity. British Journal of Psychology, 45, ­180–​­91. Berlyne, D. E. (­1962). Uncertainty and epistemic curiosity. British Journal of Psychology, 53, ­27–​­34. Berridge, K. C. (­1996). Food reward: Brain substrates of wanting and liking. Neuroscience and Biobehavioral Review, 20, ­1–​­25. Bjork, R. A. (­2017). Creating desirable difficulties to enhance learning. In I. Wallace & L. Kirkman (­Eds.), Best of the Best: Progress (­­pp. ­81–​­85). Crown House Publishing. Bloom, P. A., Friedman, D., Xu, J., Vuorre, M., & Metcalfe, J. (­2018). ­Tip–­​­­of-­​­­the-​­tongue states predict enhanced feedback processing and subsequent memory. Consciousness and Cognition, 62, ­206–​­217. Brannigan, A. (­1981). The Social Basis of Scientific Discoveries. CUP Archive. ­Bromberg-​­Martin, E. S., & Hikosaka, O. (­2009). Midbrain dopamine neurons signal preference for advance information about upcoming rewards. Neuron, 63, ­119–​­128. Brooks, G., Yang, H., & Köhler, S. (­2021). ­Feeling-­​­­of-​­knowing experiences breed curiosity. Memory, 29, ­153–​­167. Brown, A. S. (­1991). A review of the ­t ip-­​­­of-­​­­the-​­tongue experience. Psychological Bulletin, 109, ­204–​­223. Buchsbaum, D., Bridgers, S., Skolnick, W. D., & Gopnik, A. (­2012). The power of possibility: Causal learning, counterfactual reasoning, and pretend play. Philosophical Transactions of the Royal Society B, 367, ­2202–​­2212. Chu, J., & Schulz, L. (­2020). Play, curiosity, and cognition. Annual Review of Developmental Psychology, 2, ­317–​­343. Cohen, J. D., McClure, S. M., & Yu, A. J. (­2007). Should I stay or should I go? How the human brain manages the ­trade-​­off between exploitation and exploration. Philosophical Transactions of the Royal Society B: Biological Sciences, 362, ­933–​­944. Costa, V. D., Tran, V. L., Turchi, J., & Averbeck, B. B. (­2014). Dopamine modulates novelty seeking behavior during decision making. Behavioral Neuroscience, 128, ­556–​­566. Cryder, C. E., Loewenstein, G., & Seltman, H. (­2013). Goal gradient in helping behavior. Journal of Experimental Social Psychology, 49, ­1078–​­1083. Danek, A. H. (­2018). Magic tricks, sudden restructuring, and the Aha! experience: A new model of nonmonotonic problem solving. Insight, 1, ­51–​­78. Danek, A. H., Fraps, T., Von Mueller, A., Grothe, B.,  & Öllinger, M. (­2014). Working wonders? Investigating insight with magic tricks. Cognition, 130, ­174–​­185. Daw, N. D., & Tobler, P. N. (­2013). Value learning through reinforcement: The basics of dopamine and reinforcement learning. In P. W. Glimcher, & E. Fehr (­Eds.), Neuroeconomics: Decision Making and the Brain: Second Edition (­­pp. ­283–​­298). Elsevier Inc. Delgado, M. R. (­2007). Reward-related responses in the human striatum. Annals of the New York Academy of Sciences, 1104, ­70–​­88. de Waal, F. B. (­1986). The integration of dominance and social bonding in primates. The Quarterly Review of Biology, 61(­4), ­459–​­479. Duncker, K. (­1945). On p­ roblem-​­solving (­L . S. Lees, Trans.). Psychological Monographs, 58, ­1–​­113. Epstein, R. A., Patai, E. Z., Julian, J. B., & Spiers, H. J. (­2017). The cognitive map in humans: Spatial navigation and beyond. Nature Neuroscience, 20, ­1504–​­1513. Fandakova, Y., & Gruber, M. J. (­2020). States of curiosity and interest enhance memory differently in adolescents than in children. Developmental Science, 23, ­1–​­15. Fantz, R. L. (­1965). Visual perception from birth as shown by pattern selectivity. Annals of the New York Academy of Sciences, 118, ­793–​­814. Fastrich, G. M., Kerr, T., Castel, A. D., & Murayama, K. (­2018). The role of interest in memory for trivia questions: An investigation with a ­large-​­scale database. Motivation Science, 4, ­227–​­250. Ferster, C. B., & Skinner, B. F. (­1957). Schedules of Reinforcement. ­Appleton- ­​­­Century- ​­Crofts.

76

Two Faces of Curiosity in Creative Cognition Fitzgibbon, L., Lau, J. K. L., & Murayama, K. (­2020). The seductive lure of curiosity: Information as a motivationally salient reward. Current Opinion in Behavioral Sciences, 35, ­21–​­27. Fowler, H. (­1965). Curiosity and Exploratory Behavior. Macmillan. Galli, G., Sirota, M., Gruber, M. J., Ivanof, B. E., Ganeshalingam, J., Materassi, M., Thorpe, A., Loaiza, V., Cappelletti, M., & Craik, F. I. M. (­2018). Learning facts during aging: The benefits of curiosity. Experimental Aging Research, 44, ­1–​­18. Glimcher, P. (­2011). Understanding dopamine and reinforcement learning: The dopamine reward prediction error hypothesis. Proceedings of the National Academy of Science, USA, 108, ­15647–​­15654. Gopnik, A., & Walker, C. M. (­2013). Considering counterfactuals: The relationship between causal learning and pretend play. American Journal of Play, 6, ­15–​­22. Gottlieb, J., Oudeyer, P.-​­Y.,  & Baranes, A. (­2013). ­Information-​­seeking, curiosity, and attention: Computational and neural mechanisms, Trends in Cognitive Sciences, 17, ­585–​­593. Gottlieb, J., & Oudeyer, P.-​­Y. (­2018). Towards a neuroscience of active sampling and curiosity. Nature Reviews Neuroscience, 19, ­758–​­770. Grossnickle, E. M. (­2016). Disentangling curiosity: Dimensionality, definitions, and distinctions from interest in educational contexts. Educational Psychological Review, 28, ­23– ​­60. Gruber, M. J., Gelman, B. D., & Ranganath, C. (­2014). States of curiosity modulate h ­ ippocampus-​ d­ ependent learning via the dopaminergic circuit. Neuron, 84, ­486–​­496. ­ ippocampus-​­dependent memory: Gruber, M. J., & Ranganath, C. (­2019). How curiosity enhances h The prediction, curiosity and exploration (­PACE) framework. Trends in Cognitive Sciences, 23, ­1014–​­1025. Hazy, T. E., Frank, M. J., & O’Reilly, R. C. (­2010). Neural mechanisms of acquired phasic dopamine responses in learning. Neuroscience Biobehavior Review, 34, ­701–​­720. Hedne, M. R., Norman, E., & Metcalfe, J. (­2016). Intuitive feelings of warmth and confidence in insight and noninsight problem solving of magic tricks. Frontiers in Psychology, Section Cognitive Science, 7, 1314. https://­doi.org/­10.3389/­f psyg.2016.01314 Homer (­1999). The Odyssey (­Robert Fagles, Trans.). Penguin Group. Hull, C. L. (­1932). The ­goal-​­gradient hypothesis and maze learning. Psychological Review, 39, ­25–​­43. Hull, C. L. (­1934). The rat’s ­speed-­​­­of-​­locomotion gradient in the approach to food. Journal of Comparative Psychology, 17, ­393–​­422. Jirout, J. & Klahr, D. (­2012). Children’s scientific curiosity: In search of an operational definition of an elusive concept. Developmental Review, 32, ­125–​­160. Kabadayi, C., Bobrowicz, K., & Osvath, M. (­2018). The detour paradigm in animal cognition. Animal Cognition, 21, ­21–​­35. https://­doi.org/­10.1007/­­s10071-­​­­017-­​­­1152-​­0 Kahneman, D. (­2011). Thinking Fast and Slow. Farrar, Straus and Giroux. Kang, M. J., Hsu, M., Krajbich, I. M., Loewenstein, G., McClure, S. M., Wang, J. T., & Camerer, C. F. (­2009). The wick in the candle of learning: Epistemic curiosity activates reward circuitry and enhances memory. Psychological Science, 20, ­963–​­973. Kirkpatrick, K., & Church, R. M. (­2003). Tracking of the expected time to reinforcement in temporal conditioning procedures. Learning & Behavior, 31, ­3 –​­21. Kivetz, R., Urminsky, O., & Zheng, Y. (­2006). The ­goal-​­gradient hypothesis resurrected: Purchase acceleration, illusionary goal progress, and customer retention. Journal of Marketing Research, 43, ­39–​­58. Köhler, W. (­1925). The Mentality of Apes (­2nd rev. ed.) (­E. Winter, Trans.). Harcourt, Brace & Company, Inc. Kornell, N., & Metcalfe, J. (­2006). Study efficacy and the region of proximal learning framework. Journal of Experimental Psychology: Learning, Memory, and Cognition, 32, ­609– ​­622. Kuhn, T. S. (­1962). The Structure of Scientific Revolutions. University of Chicago Press. Lau, J. K. L., Ozono, H., Kuratomi, K., Komiya, A., & Murayama, K. (­2020). Shared striatal activity in decisions to satisfy curiosity and hunger at the risk of electric shocks. Nature Human Behavior, 4, ­531–​­543. Litman, J. A. (­2005). Curiosity and the pleasures of learning: Wanting and liking new information. Cognition and Emotion, 19, ­793–​­814. Litman, J. A. (­2008). Interest and deprivation factors of epistemic curiosity. Personality and Individual Differences, 44, ­1585–​­1595. Litman, J. A., Hutchins, T. L., & Russon, R. K. (­2005). Epistemic curiosity, f­eeling-­​­­of-​­knowing, and exploratory behaviour. Cognition and Emotion, 19, ­559–​­582.

77

Janet Metcalfe and William James Jacobs Loewenstein, G. (­1994). The psychology of curiosity: A review and reinterpretation. Psychological Bulletin, 6, ­75–​­98. Loewenstein, G. (­2007). Exotic Preferences. Oxford University Press. Luchins, A. S. (­1942). Mechanization in problem solving: The effect of Einstellung. Psychological Monographs, 54, ­1–​­95. MacDonald, K. (­2007). ­Cross-​­cultural comparison of learning in human hunting. Human Nature, 18, ­386–​­402. Marvin, C. B., & Shohamy, D. (­2016). Curiosity and reward: Valence predicts choice and information predictions errors enhance learning. Journal of Experimental Psychology: General, 145, ­266–​­272. McGillivray, S., Murayama, K., & Castel, A. D. (­2015). Thirst for knowledge: The effects of curiosity and interest on memory in younger and older adults. Psychology and Aging, 30, ­835–​­841. Metcalfe, J. (­1986). Premonitions of insight predict impending error. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, ­623–​­634. Metcalfe, J. (­2002). Is study time allocated selectively to a region of proximal learning? Journal of Experimental Psychology: General, 131, ­349–​­363. Metcalfe, J., & Kornell, N. (­2003). The dynamics of learning and allocation of study time to a region of proximal learning. Journal of Experimental Psychology: General, 132, ­530–​­542. Metcalfe, J.,  & Kornell, N. (­2005). A region of proximal learning model of study time allocation. Journal of Memory and Language, 52, ­463–​­477. Metcalfe, J., Schwartz, B. L., & Bloom, P. A. (­2017). The ­t ip-­​­­of-­​­­the-​­tongue (­TOT) state and curiosity. Cognitive Research: Principles and Implications, 2, ­31–​­39. Metcalfe, J., & Wiebe, D. (­1987). Intuition in insight and noninsight problem solving. Memory & Cognition, 15, ­238–​­246. Murphy, C., Dehmelt, V., Yonelinas, A. P., Ranganath, C., & Gruber, M. J. (­2021). Temporal proximity to the elicitation of curiosity is key for enhancing memory for incidental information. Learning & Memory, 2, ­34–​­39. Nadel, L. (­2021). Some implications of postnatal hippocampal development. Hippocampus. online first version. doi: 10.1002/­h ipo.23369 Öllinger, M.,  & Knoblich, G. (­2009). Psychological research on insight problem solving. In H. Atmanspacher & H. Primas (­Eds.), Recasting Reality (­­pp. ­275–​­300). Springer. Panksepp J. (­1981). The ontogeny of play in rats. Developmental Psychobiology, 14, ­327–​­332. Panksepp, J., & Biven, L. (­2012). The Archaeology of Mind: Neuroevolutionary Origins of Human Emotions (­Norton series on interpersonal neurobiology). WW Norton & Company. Pavlov, I. P., & Anrep, G. V. (­1927). Conditioned Reflexes: An Investigation of the Physiological Activity of the Cerebral Cortex. Oxford University Press. Pellegrini, A. D. (­1993). Boys’ rough-​­a nd-​­t umble play, social competence and group composition. British Journal of Developmental Psychology, 11, ­237–​­248. Pellegrini, A., & Smith, P. K. (­1998). Physical activity play: The nature and function of a neglected aspect of play. Child Development, 69, ­557–​­598. Pellis, S. M., & Iwaniuk, A. N. (­2000). ­Adult–​­adult play in primates: Comparative analyses of its origin, distribution and evolution. Ethology, 106, ­1083–​­1104. ­ ough-­​­­and-​­tumble play and the development of the social brain. Pellis, S. M., & Pellis, V. C. (­2007). R Current Directions in Psychological Science, 16, ­95–​­98. Peterson, E. J., & Verstynen, T. D. (­2020). Curiosity eliminates the ­exploration-​­exploitation dilemma. Memory, 1, ­1–​­32. Piaget, J. (­1954). The Construction of Reality in the Child. Routledge (­Taylor & Francis, 1999 Edition). Qiu, X., Wen, L., Wu, C., Yang, Z., Wang, Q., Li, H.,  & Wang, D. (­2020). Impact of learning methods on spatial knowledge acquisition. Frontiers in Psychology, 11, ­1–​­17. https://­doi.org/­10.3389/ ­f psyg.2020.01322 Ramstetter, C. L., Murray, R., & Garner, A. S. (­2010). The crucial role of recess in schools. Journal of School Health, 80, 5­ 17–​­526. Schwartz, B. L. (­1999). Sparkling at the end of the tongue: The etiology and phenomenology of t­ip-­​ ­­of-­​­­the-​­tongue states. Psychonomic Bulletin & Review, 6, ­379–​­393. Schwartz, B. L., & Cleary, A. M. (­2016). ­Tip-­​­­of-­​­­the-​­tongue states, déjà vu and other metacognitive oddities. In J. Dunlosky & S. Tauber (­Eds.), Oxford Handbook of Metamemory (­­pp. ­95–​­108). Oxford University Press.

78

Two Faces of Curiosity in Creative Cognition Schwartz, B. L., & Metcalfe, J. (­2011). ­Tip-­​­­of-­​­­the-​­tongue (­TOT) states: Retrieval, behavior, and experience. Memory & Cognition, 39, ­737–​­749. Son, L. K., & Metcalfe, J. (­2000). Metacognitive and control strategies in s­ tudy-​­time allocation. Journal of Experimental Psychology: Learning, Memory, and Cognition, 26, ­204–​­221. Spelke, E. S. (­1985). ­Preferential-​­looking methods as tools for the study of cognition in infancy. In G. Gottlieb & N. A. Krasnegor (­Eds.), Measurement of Audition and Vision in the First Year of Postnatal Life: A Methodological Overview (­­pp. ­323–​­363). Ablex. Spelke, E. S., Breinlinger, K., Macomber, J., & Jacobson, K. (­1992). Origins of knowledge. Psychological Review, 99, ­605–​­632. Stare, C. J., Gruber, M. J., Nadel, L., Ranganath, C.,  & Gómez, R. L. (­2018). ­Curiosity-​­driven memory enhancement persists over time but does not benefit from p­ ost-​­learning sleep. Cognitive Neuroscience, 9, ­100–​­115. Sutton, R. S. (­2017). Retrieved from http://­w ww.incompleteideas.net/­sutton/­book/­fi rst/­2/­node2. html Sutton, R. S., & Barto, A. G. (­2018). Reinforcement Learning: An Introduction (­2nd ed.). MIT Press. Thorndike, E. L. (­1927). The law of effect. The American Journal of Psychology, 39, ­212–​­222. Tolman, E. C. (­1932). Purposive Behavior in Animals and Men. University of California Press. Tolman, E. C., & Honzik, C. H. (­1930). Introduction and removal of reward, and maze performance in rats. University of California Publications in Psychology, 4, ­257–​­275. von Stumm, S., Hell, B., & ­Chamorro-​­Premuzic, T. (­2011). The hungry mind: Intellectual curiosity is the third pillar of academic performance. Perspectives on Psychological Science, 6, ­574–​­588. Vygotsky, L. S. (­1978). Mind in Society. Harvard University Press. Wade, S., & Kidd, C. (­2019). The role of prior knowledge and curiosity in learning. Psychonomic Bulletin & Review, 26, ­1377–​­1387. Whiten, A., & de Waal, E. (­2018). The pervasive role of social learning in primate lifetime development. Behavioral Ecological Sociobiology, 72, ­1–​­16. Wilson, R. C., Bonawitz, E., Costa, V. D., & Ebitz, R. B. (­2021). Balancing exploration and exploitation with information and randomization. Current Opinion in Behavioral Sciences, 38, ­49–​­56. Wilson, R. C., Geana, A., White, J. M., Ludvig, E. A., & Cohen, J. D. (­2014). Humans use directed and random exploration to solve the ­explore–​­exploit dilemma. Journal of Experimental Psychology: General, 143, ­2074–​­2081. Xu, J. & Metcalfe, J. (­2016). Studying in the region of proximal learning reduces mind wandering. Memory & Cognition, 44, ­681–​­695.

79

6 THINKING WIDE AND NARROW A ‘­Cultural Creative Cognition’ Approach to Possibility Thinking Vlad P. Glăveanu

It is a common and relatable occurrence being caught up in daydreaming at exactly the time when our attention should focus on the task at hand. Whether we disconnect out of boredom, because it’s more exciting to think about something else, or simply because we can’t keep concentrating any longer, the result is similar: we are transported from an experience of the ‘­here and now’ to one of the ‘­then and there’, from what is actual and immediate to what is possible, probable and even impossible. In other words, we bracket the world ‘­a s is’ in order to indulge, for a shorter or longer while, in fantasies about how things could, might or will never be (­Singer, 2014). But this is no indulgence in the negative sense of the word. Grasping for experiences that lie beyond one’s current situation is a fundamentally human attribute, closely related to our capacity to use signs and symbols as a way of placing a certain distance between sensorial inputs and mental representations, between reflex and deliberate actions, between the world as is and the world as imagined (­Valsiner, 2007). The key to understanding this propensity, however, is the fact that reality and imagination are not disconnected from each other but rather stand in close union (­Vygotsky, 2004). To return to the example of daydreaming, the triggers, contents and outputs of this phenomenon all point back to ‘­the real world’. And yet, phenomenologically, episodes of reverie mark a distinct kind of experience akin to a widening of our mental horizon, to enjoying a certain degree of freedom and ­open-​­endedness dissimilar to any narrow focus on what is the case. In other words, while this chapter focusses primarily on thinking about the possible, it needs to be emphasised that most human thoughts is ‘­hybrid’ or ­non-​­binary in nature, revealing actuality and possibility as two sides of the same coin. This chapter is dedicated to the exploration of precisely this qualitatively different mode of thought that I tentatively call ‘­possibility thinking’ (­PT). This notion, used in the past to capture a set of e­ ducation-​­focussed skills and practices (­Craft, 2003; Craft, Cremin, Burnard & Chappell, 2008), is reformulated here as a broader principle organising our (­creative) cognition and, with it, our relation to ourselves, to others and to the shared world we inhabit. In order to distinguish exactly what makes PT different from other modes of thought, we need to contrast it with its complement (­not opposite), which is ‘­actuality thinking’ (­AT). The latter focusses our attention on what is the case and keeps us within the boundaries of what we remember to have happened, what we consider to exist or be true, and what is likely to happen next. The former doesn’t deny these constraints on our thinking and 

80

DOI: 10.4324/9781003009351-7

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

action but expands them into the sphere of what could be or could have been the case, what is improbable or could never happen. Daydreaming is one example of PT being placed in the foreground and AT in the background. And it is also a good illustration of how these processes of ­fore-​­and backgrounding can occur in quick succession (­leading to blended experiences of actuality and possibility) or, at times, how they generate shifts of longer duration (­e.g., lengthy episodes of daydreaming interrupted only by outside events). My focus here, independent of duration, is to examine how foregrounding the possible transforms our experience of and relation to what is actual and why its ­processes – ​­e.g., pretend, anticipatory, counterfactual and utopian ­thinking  – ​­are deeply interconnected. What I do not intend, however, is to reify the dichotomy between PT and AT as independent modes of thought. As the foregrounding/­backgrounding metaphor suggests, there is always actuality in our exploration of the possible, and there are new possibilities to be found when paying attention to what is. The ‘­then and there’ of PT is experienced in AT’s ‘­here and now’, and the latter both grounds and lifts the former. I approach all the issues above from the standpoint of creativity research and, in particular, ‘­cultural creative cognition’ (­CCC), a subfield based on the assumption that creative thinking is as much a psychological as it is a social, cultural and material/­embodied phenomenon (­for intersections between creativity and culture, see Glăveanu, 2014, 2016). This particular theoretical lens stands in contrast to traditional ‘­creative cognition’ (­CC) approaches (­e.g., Finke, Ward & Smith, 1992; Ward, Smith & Finke, 1999) – ​­although there are some overlaps between the two, key among them being a recognition of the role played by cognition in creative e­ xpression – ​­in which the inputs, processing and outputs of creativity are all i­ntra-​ ­psychological. Social and material interactions as well as cultural contexts are traditionally considered in CC as external influences that come to shape mental processes of generating and exploring ideas. CCC reframes this dynamic by recognising creative ideation as an embodied act that plays out in specific social and material settings and is framed by broader institutions, values and practices. In this way, what appear as universal, ­person-​­bound processes in CC are conceptualised in local and distributed terms within CCC, and what is neatly categorised as person, process and product in the former becomes a holistic study of experience in the latter (­e.g., Dewey, 1934). How is this relevant for PT? First of all, while not all episodes of PT lead to creative outcomes and there is a clear and important part to be played by AT in creativity (­Cropley, 2006), it is the case that all PT shares, more or less explicitly, the basic qualities of a creative experience, which are ­open-​­endedness or a lack of predetermined goals, ­pluri-​­perspectivism or the ability to bring in multiple perspectives, ­non-​­linearity, describing the ups and downs of creative work, and orientation towards the future, especially towards multiple, open futures (­Glăveanu & Beghetto, 2020). Second, the creative nature of PT, at a cognitive level, is explained here in transactional, not i­ntra-​­individual, terms, pointing to how the possible emerges in our lives as a consequence of our embodied, social and cultural existence rather than despite of it. Last but not least, expressions of PT (­a s well as AT) are all viewed as contextual in nature, depending, at once, on person and environment and especially on their ­co-​­evolution. In fact, while this is not the focus of the present chapter, a CCC approach can be useful for theorising AT as much as it is for examining PT in the sense that a focus on the actual can b­ e – ​­and often i­s – creatively productive (­consider, for example, moments of ​­ creative flow). The chapter will proceed by analysing cases in which PT has been theorised as a mode of ­thought – in ​­ narratives, play and w ­ onder-​­based explorations of the world. Then it will propose a concrete typology of PT built on four pillars (­pretence, anticipation, counterfactuals 81

Vlad P. Glăveanu

and utopia) and argue why they are distinct yet interdependent phenomena in neurological, evolutionary and developmental terms. A brief review of the research for each of the four instances of PT will be offered. In the end, the discussion will return to the relation between PT and AT and their connection to other, more popular distinctions, such as that between automatic, or System 1, and controlled, or System 2 thinking. Conclusions will be drawn concerning the expansion of CCC into a broader theory concerning the role, mechanisms and value of thinking both ‘­w ide’ and ‘­narrow’.

A Different Kind of Thought There are multiple strands of scholarship in the history of psychology and philosophy that speak to the difference between what I call here possibility and actuality thinking and, in particular, to the special status of the former for mundane thought and action. In his ­well-​ ­k nown book ‘­Actual Minds, Possible Worlds’, Bruner (­1985, p­ . 97) advanced a bold argument about the existence of ‘­two irreducible modes of cognitive functioning’ that help us organise our experience, filter perception and construct reality. These are the paradigmatic (­­logico-​­scientific) mode and the narrative mode. The first is based on the categorisation and conceptualisation of the actual as well as on establishing logical relations between these categories and conceptions, for example, following the principles of consistency and noncontradiction. It is a type of thinking that anchors us firmly into ‘­what is’ and tries to describe it in objective terms. While some might argue that one can never achieve such a ‘­v iew from nowhere’ (­Nagel, 1989), the strength of paradigmatic thought lies in the fact that it neatly structures the world around us. Conversely, narrative thinking embraces the messiness and ­non-​­linearity of human existence and organises it on entirely different principles. Instead of ­cause-­​­­and-​­effect relations, the logic of narrative is that of the temporal unfolding of meaningful action. To think in terms of stories rather than ­pre-​­existing categories is thus more ‘­f reeing’ and more adapted to everyday interactions and communication between people. Most of all, the narrative mode of thought opens up our cognition to a wider range of possible worlds (­outside of those that are statistically or logically likely) and infuses them with actors, storylines and consequences. It introduces us to the realm of what could happen or could have happened without having truth or truthfulness as a primary concern. Which is not to say that narrative thinking is fanciful or deceitful. The stories we construct with its help are judged based on whether they are believable or not, whether they are consistent with other narrative accounts and whether they are generative of deeper and essentially different understandings of the world ‘­a s is’. Moreover, all stories are open to a variety of different interpretations, which is not the case (­or, at least, not intended to be) with paradigmatic outcomes. It is not surprising, then, to discover that narratives are a primary carrier of PT as they stand in close connection with the imagination (­for a discussion of the notion of narrative imagination, see Andrews, 2014; Brockmeier, 2009; Hardy, 1975). Stories are also an integral part of play episodes, from an early age, and their main role is to both organise and solidify a certain experience (­e.g., that of doctor and patient) while, at the same time, keeping it flexible, ambiguous and open to a multitude of endings (­a s anyone who has witnessed children at play can testify). Playfulness is, in this context, a state of mind that individuals of all ages adopt when they allow themselves to consider the world different than it is. This makes playful approaches to reality another expression of PT (­Craft, McConnon, & Matthews, 2012). What makes this thinking stand out compared to ­non-​­play situations? Its capacity to essentially accelerate development. In his work on play, Vygotsky (­1978, ­p. 102) made this observation: 82

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

Play creates a zone of proximal development of the child. In play a child always behaves beyond his average age, above his daily behaviour; in play, it is as though he were a head taller than himself. As in the focus of a magnifying glass play contains all developmental tendencies in a condensed form and is itself a major source of development. There is considerable empirical evidence that Vygotsky’s intuition was, in fact, correct: when faced with the same kind of situation or task, children do demonstrate more sophisticated ways of thinking and acting in play than they do outside of it (­see Bodrova & Leong, 2015). The question is, why do we notice such a powerful transformation? To understand this, we need to return to the notion of PT and its assumed capacity to reveal the world as malleable and ­pluri-​­perspectival. When relating to the world in this manner, children adopt, practice and exchange different positions, expand their repertoire of knowledge and skills and practice their agency (­Gillespie, 2006). Play offers also opportunities to not only learn about but also deconstruct everyday life and question the processes and outcomes of AT. This kind of ­de-​­familiarisation of the familiar is specific to yet another phenomenon situated at the heart of P ­ T – ​­the experience of wonder. Wondering is a mundane occurrence, even if we often undergo ‘­m ilder’ versions of it in which we question some not all of our understandings. To experience wonder means, in this context, to ­become – ​­more or less s­uddenly – ​­aware of a wider field of possibility and motivated to explore it, not for the sake of gaining knowledge that is certain and final, but provisional and ready for ­re-​­interpretation (­Glăveanu, 2020a). This capacity to withstand uncertainty and doubt and consider them productive states of mind is yet another defining feature of PT (­see also Beghetto, 2017). It is what Heidegger poetically called ‘­dwelling in the unknown’ (­Heidegger, 1962), accepting not knowing as an opportunity to engage with the openness of being that describes each human existence. This fundamental openness is itself something we come to recognise when we are faced with macro or micro ruptures in our understanding of the world and our sense of self. Therefore, our propensity to wonder doesn’t make PT a more enjoyable thinking system, but it certainly contributes to it being highly rewarding for our intellectual life. This is why, in Antiquity, Socrates famously placed this phenomenon at the origin of philosophy or the love of wisdom (­Plato (­1973/­ca. 369 B.C.). Just like wondering, PT doesn’t result in the certainties specific to AT, but what it does as a mode of thought is keep us alert, curious and, above all, humble. A final observation about the cases of narratives, play and wonder is that they colour the experience of individuals and yet depend on a wider social, material and cultural context. Most of all, they are ­culture-​­enabled in the sense that, without the signs and tools made available by living in a shared, human society, it would be impossible to link events to each other into a storyline, to pretend that things are not what they are or to question the very nature of reality. All of these actions are creative or, at the very least, potentially creative. They result in novel and meaningful ideas about oneself, others, and society, and still, the cognitive processes underlying them cannot be reduced to the operation of individual minds; their understanding requires a consideration of mind, body and world together. It requires grounding possibility (­and actuality) thinking within CCC rather than CC alone.

Possibility and Actuality Thinking It is a bold conceptual move to ­formulate – ​­yet ­another – ​­binary system of thinking. Some key characteristics of PT have already been outlined above, but a much more systematic description of it is required and, at the same time, a more consistent reflection on the meaning 83

Vlad P. Glăveanu

and value of AT (­which is not, simply, everything that PT is not). To substantiate PT and AT as modes of thought and, more specifically, as distinct modes of thought, the following claims should be evidenced: 1 Widespread manifestation of ­PT-​­and ­AT-​­associated processes in the general population, including across cultures; 2 Differentiated levels of achievement in the case of both thinking systems; 3 Neurological differences in the way information associated with possibility and actuality is processed, including distinct pathways for each; 4 Clear yet differentiated evolutionary roles for PT and AT; 5 Early onset in development as well as distinct pathways of development; 6 Specific sets of operations and ­mechanisms  – ​­cognitive, ­affective-​­motivational and ­behavioural – ​­that are associated with specific antecedents and consequences; 7 The capacity to prime or otherwise scaffold the PT and AT modes of thought. Some of these criteria are easier to address than others, especially since evidence is still accumulating, mainly concerning neurological, developmental and evolutionary data. Other types of information are intrinsically difficult to gather, for instance, how widespread the types of thinking referred to here actually are; in this regard, estimates can be based on the definitions of PT and AT. And this brings us to the core issue of operations and ­mechanisms –​ ­a s long as these begin to be clarified, the other types of evidence can be gathered in a more systematic manner. The remainder of this chapter will be dedicated, primarily, to uncovering these underlying processes, particularly in relation to PT, and reviewing theories and research associated with them. Concretely, PT is discussed here in terms of four main s­ub-​­processes or pillars (­see also ­Table 6.1). These refer to thinking about ‘­what could be’, ‘­what is to come’, ‘­what could have been’ and ‘­what is not or can never be’. For each, we can distinguish a main temporal focus, a core psychological phenomenon, a process of thinking (­in lay terms) and a corresponding facet of creative experience (­a s outlined in Glăveanu & Beghetto, 2020). For example, thinking about ‘­what could be’ focusses the person on the present and, specifically, on developing alternatives to it. This is the essence of pretence or our capacity to imagine reality different than it ­is – ​­an exercise in ‘­­as-​­if thinking’. Finally, as a creative experience, ‘­what could be’ foregrounds ­pluri-​­perspectivism by revealing the world as multiple, malleable and open to being ­re-​­imagined. ‘­W hat is to come’ refers, as the name suggests, to thinking about the ­f uture – not ​­ what is likely to happen in the next moment, but various scenarios for the short, medium and long term. The psychological process at the heart of this is anticipation, the ­Table 6.1  Dimensions of Possibility Thinking Relation to What Is

Main Temporal Focus

Psychological Phenomenon

Type of Thinking

Facet of Creative Experience

What could be What is to come What could have been What is not/­ never can be

Present Future Past

Pretence Anticipation Counterfactuals

­A s-​­i f thinking ­W hat-​­i f thinking ­A s-​­else thinking

­Pluri-​­perspectivism Future orientation Nonlinearity

­Cross-​­temporal

Utopia

­W hat-​­else thinking

­Open-​­endedness

84

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

capacity to ask ‘­what if ’ this or that came to be. Naturally, the dimension of creative experience highlighted here is future orientation. ‘­W hat could have been’, in contrast, focusses not on the past but on the past as it never was. This is the domain of counterfactual thinking or of conceiving alternatives for past events; metaphorically, ‘­­as-​­else thinking’ helps us relate to what was as if it were something else. Last but not least, ‘­what is not’ or ‘­never can be’ reveals imaginations of things that are absent or altogether impossible. It reflects utopian or ‘­­what-​­else’ thinking, in which the task is to represent the world in ways that are explicitly unrealistic yet productive or interesting. The dominant creative experience feature is o ­ pen-​ ­endedness, given the fact that what is not and cannot be is, by definition, more indeterminate than what is. PT is not a new term. In fact, it has been widely popularised in the psychology of creativity and education by Anna Craft and her collaborators. The earliest operationalisations of this notion relate it to ­question-​­posing, play, immersion, innovation, ­r isk-​­taking, being imaginative and able to s­elf-​­determine (­Burnard, Craft, Cremin, Duffy, Hanson, Keene, Haynes & Burns, 2006). Typologies of children’s questions were later proposed (­Chappell, Craft, Burnard & Cremin, 2008), including a focus on both question posing and question responding, and evidence of peer collaboration in PT was found (­Craft, Cremin, Burnard, Dragovic  & Chappell, 2012). Considerable research within this tradition has focussed on pedagogical strategies that can enhance PT, such as provoking possibilities, allowing time and space, being in the moment, making interventions and mentoring in partnership (­Craft, Mcconnon & Matthews, 2012). As we can notice from the brief description above, some of the key elements discussed at the beginning of this p­ aper – ​­storytelling, playfulness and ­question-​­triggered ­wonder – ​­are all found in educational studies of PT, including a strong emphasis on the social and cultural nature of c­ hild–​­child and ­child–​­adult interactions. However, the focus remains on childhood and education. The present chapter builds on these understandings and tries to articulate them into a broader frame. This frame necessarily includes thinking about ‘­actuality’. As noted from the start, possibility and actuality are not opposites; in fact, the possible builds on, connects us to and transforms our relationship with the real (­Glăveanu, 2020b). At the same time, AT constitutes its own mode of thought, vis a vis PT, that is marked by a focus on ‘­what is the case’ (­a s compared to ‘­what could be’), ‘­what will happen next’ (­a s compared to ‘­what is to come’), ‘­what has happened’ (­a s compared to ‘­what could have been’) and ‘­what is real’ (­a s compared to ‘­what is not/­never can be’). In practice, these distinctions often collapse in the enactment of thought. AT shares characteristics with Bruner’s paradigmatic thinking (­Bruner, 1985), Piaget’s concrete thought (­Piaget, 1926) and Schroyens’ thinking about what is true (­Schroyens, 2010). Which is not to say that logical thought is exclusively the domain of AT. Observing a child or adult dealing with a practical problem shows the necessary involvement of PT and AT, just as being creative is based on both divergent and convergent thinking and their integration (­de Vries & Lubart, 2019). Last but not least, if PT and its complement, AT, are to be established as modes of thought, then their different pillars or facets should be internally consistent. Given the special focus on PT in this chapter, this means that we should have grounds to expect that pretence, anticipation, counterfactual and utopian thinking are intrinsically related to each other. The main supposition is that thinking about what is possible, in the past, present and/­or future, is what substantiates the four pillars of PT. There is indeed evidence that pretence and anticipation ­co-​­occur in play episodes (­Hunleth, 2019), that we can anticipate counterfactuals (­Hetts, Boninger, Armor, Gleicher & Nathanson, 2000), that counterfactual forms of worldmaking are close to utopia (­H illgren, Light & Strange, 2020) and that anticipation meets utopian 85

Vlad P. Glăveanu

thinking (­e.g., in ‘­real utopias’; Wright, 2010). To understand whether the mechanisms specific for pretence, anticipation, counterfactual and utopian thinking converge on the notion of possibility, we need, however, to examine them closer and analytically separate from each other, which is the kind of exercise we engage in next.

What Could Be One of the most direct expressions of PT has to do with conceiving alternatives to the real. This enables us not only to take some (­symbolic) distance from ‘­what is’ (­Valsiner, 2007), but also to develop a more flexible relationship to it, one in which reality itself becomes multiple and malleable. More than this, PT as a mode of thought allows us to temporarily suspend ‘­what is’ in favour of ‘­what could be’. This is the essential feature of pretence or the capacity to relate to aspects of reality as if they were different than they are. It is an evolved capacity that appears relatively early in human development, within pretend play (­Weisberg, 2015), and continues throughout the life course whenever we get immersed in fictional worlds (­Frigg, 2010), engage in role play (­Landreth, 2012) or even use irony as a way of mocking something or someone (­Currie, 2006). These acts of ­m ake-​­believe or ‘­­as-​­if thinking’ free us from the constraints of actuality without denying or forgetting the latter. When children pretend play, they are not confused as to what objects ‘­really are’ or how causality works (­Harris, 2000); on the contrary, they reimagine certain natural relations while leaving other causal links intact. This is why even the freest episodes of play are themselves r­ ule-​­bound, yet these rules are the product of ‘­­as-​­if ’ rather than ‘­­as-​­is’ thought. AT is not displaced by PT but built upon and expanded to a wider view of the world. An interesting attempt at explaining how exactly we widen our horizon of possibility through pretend play is found in Vygotsky’s work (­see Vygotsky, 1967). He postulated that, during the preschool years, in episodes of play, the ‘­field of vision’ is gradually placed under the control of the ‘­field of meaning’. In other words, the child goes from acting on the physical properties of objects, as perceived, to acting based on interpretations of what these ob­ ell-​­known example, a stick becomes a horse, and, even though jects could become. In his w it is visually perceived as a stick, the child is happy to ride it and experience the emotions he or she would if the stick were a real horse. Vygotsky also helpfully argued against a more Piagetian view of play in which this is a transitional stage in our development whose main purpose is to be superseded by formal intelligence (­­Sutton-​­Smith, 1966). Conceiving of possibilities and enacting them is a constant of the human intellect for as whimsical as these possibilities might be. And, in fact, there is a large body of accumulated evidence when it comes to the benefits of pretence that range from its close ties with creativity (­Carruthers, 2002) and the acquisition of knowledge (­Sutherland & Friedman, 2012) to organisational success (­Hjorth, Strati, Drakopoulou Dodd  & Weik, 2018) and the success of therapeutic practices (­Schaefer & Drewes, 2011). These different contexts all testify to the power of PT to transform our experience of what is real and infuse it with alternatives, hope and playful ­re-​­creations. Building on Vygotskian ideas, it becomes clear that the dimension of creative experience best captured by pretence, ‘­­as-​­if thinking’, is ­pluri-​­perspectivism. Indeed, to think about what could be involves the development of new perspectives on what is (­Glăveanu, 2020b). These perspectives are not merely mental schemas or simulations (­Goldman, 2002); they are born out of and inform our actions with objects in a shared social and cultural world. As such, adopting more than one perspective on the world means being able to think and act in more ways than one. Pretence and, more broadly, PT expand our repertoire of ­action-​­relations 86

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

to our environment and, in doing so, help us make the familiar unfamiliar and question ­taken-­​­­for-​­granted assumptions about ourselves, other people and our surroundings. This dynamic is situated primarily in the present moment, even if it engages past knowledge and can construct visions of the future. In the end, pretence shares a fundamental connection with anticipation, in as much as it orients us towards what is to come, with counterfactuals, given that it proposes a counterfactual reading of ‘­what is’, and to utopian thinking through its power to critique reality and, sometimes, reconstruct it for us anew.

What Is to Come The future is one of the most common ‘­domains’ of PT, as expected. And yet, not every way of thinking about the future qualifies as p­ ossibility-​­enhancing. As mentioned before, a focus on what is likely to happen (­i.e., what will ‘­realistically’ happen) anchors our thought more in AT than PT. It is primarily anticipating multiple rather than singular futures that characterises the latter. This follows, in many ways, the distinction between forecast and foresight (­Van Lente, 2012), with the former usually based on quantitative estimations while the latter engaging abductive inferences. This distinction is fundamental for anticipation studies (­Poli, 2017), an emerging interdisciplinary field concerned with the way in which we construct ­action-​­based imaginations of the future. This pillar of PT corresponds, in the model proposed here, to thinking about ‘­what is to come’ and, more broadly, to ‘­­what-​­if ’ types of thought directed towards the future. This is an important observation given that ‘­­what-​­if ’ questions can be asked also in relation to the present (­connecting it to the ‘­­as-​­if ’ of pretence) and the past (­relating it to counterfactuals and ‘­­as-​­else’ thinking). In this case, the focus is on imagining ‘­what if scenario X or Y were to happen’ and, in this way, on generating alternatives and picturing their consequences. There are certainly many benefits associated with being able to anticipate; key among them is the fact that we can adapt better to complex and ­ever-​­changing environments (­K rogh, 2018). Second, this process makes PT an important system for planning and d­ ecision-​­making (­Williams & Ward, 2007). More broadly, anticipation substantiates agentic and creative ac­ ell-​ tion (­Hastrup, 2005). Last but not least, the capacity to anticipate contributes to our w ­being and psychological health (­Luo, Chen, Qi, You  & Huang, 2018) through building expectations about positive future events. Anticipations are not always positive or uplifting, however, and they can sometimes fill a person with dread rather than hope. But, even in those cases, it is the possibility of anticipating multiple courses of action that matters. Being certain about what will come, positive or negative, constrains possibilities rather than extends them, and, while there is a kind of convergence that is beneficial for quickly deciding on a response, it doesn’t allow for a reflective consideration of alternatives that is at the heart of PT. In terms of creative experience, anticipation connects, unsurprisingly, to future orientation even though, it should be noted, it is not the only process to do so. In fact, all PT processes share an intrinsic connection to the future in the sense that they are all prospective (­Seligman, Railton, Baumeister & Sripada, 2016). Pretence constructs multiple perspectives on the present that enable future action. Counterfactuals about the past are engaged in with the aim of learning something useful about what could come. Utopias have an ­a-​­temporal dimension to them, but utopian thinking is clearly aimed at envisioning unlikely futures. In this sense, anticipation is an integral part of all PT processes, one that depends considerably on AT as well. The basis of all predictions is past experience (­Moulton & Kosslyn, 2009), even if anticipation doesn’t use it deterministically. ‘­­What-​­if ’ thinking operates with 87

Vlad P. Glăveanu

multiple scenarios, some of them closer and some further to what is to be expected. In this sense, to anticipate doesn’t mean to think ahead as much as it does to think otherwise about what is ahead. And, when we anticipate, we also change our behaviour accordingly; as Poli (­2017) notes, anticipation is about what we are ready to do, not merely think. It is the action orientation that makes PT a distributed phenomenon, embodied, enacted and extended into the world. While not all our anticipations get to materialise, they do position us towards an open future in ways that have deep consequences for how this future is brought about.

What Could Have Been The past orientation of PT focusses our attention on ‘­what could have been’ or counterfactual thinking (­see Roese, 1997). Counterfactuals, as the name suggests, are perspectives that go against the ‘­facts’ or what has actually happened (­e.g., imagining what life would have been like for you if you were born into a different family or a different country). While we can construct such perspectives in the present (­in many ways, this is the equivalent of pretence; Kavanaugh & Harris, 1999) and sometimes about the future (­Weber, 1996), the clearest ways in which we can diverge from reality are by considering what was (­AT) and conceiving alternatives to it (­PT). Counterfactuals have long been a theme of study within cognitive psychology (­e.g., Boninger, Gleicher & Strathman, 1994; Epstude & Roese, 2008; Roese & Olson, 1995). A lot of this research has focussed on two main issues: (­a) the relation between counterfactuals and positive and negative affect and (­b) the need to demonstrate that counterfactual thinking is, overall, beneficial. The two themes are related to each other and point to a recurrent finding in the literature: that when misfortunes come our way, we tend to engage in counterfactual thinking about alternatives that, on many occasions, intensify our feelings of shame, guilt or even depression. On the other hand, there is another mechanism by which, when we avoid misfortune, we can reflect on how narrowly we escaped it and feel more fortunate, even derive important lessons for the future (­e.g., how to avoid such close calls). This is, above all, considered to be the main advantage of counterfactual ­thinking – ​­the possibility to learn from it. But, in a PT framework, there is much more to this phenomenon than cautionary tales. To imagine what could have been opens up, first and foremost, a space of reflection about the nature and consequences of our actions and of the actions of others. In this space, we get to play with the usual chains of cause and effect, suspending or even reversing them. We experience, thus, a degree of freedom ­v is-­​­­a-​­vis ‘­what is’ and the usual processes of AT (­while remaining anchored within reality and, for instance, the order in which the events occurred; Segura, ­Fernandez-​­Berrocal & Byrne, 2002), a kind of experience that has important consequences for our thinking and action, present and future. Typically, in psychology, counterfactual thinking has been studied in relation to perceptions of control (­Nasco  & Marsh, 1999), false belief (­Drayton, ­Turley-​­Ames  & Guajardo, 2011), fantasy proneness (­Bacon, Walsh & Martin, 2013) and even procrastination (­Sirois, 2004). But what remains to be explored further are precisely its links to our awareness of and sense of the possible. This is why, in our proposed model of PT, the dominant characteristic ­ on-​­linearity. By reflecting on ‘­what of creative experience captured by counterfactuals is n could have been’, we not only come up with alternatives, we also trouble the linearity of time and more easily glide from past to present and future. Counterfactuals show us that even something as ‘­fi xed’ in time as the past can be r­e-​­thought and ‘­played’ in cognitive and emotional terms. The key outcome of these exercises is not simply alternatives to past ­events  – ​­it is a sense that the actual can be reimagined even when it cannot be changed 88

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

through direct action, that the past is as open to our interpretation as much as the future and that linear relations can be troubled and turned ­bi-​­or ­multi-​­directional. I have called this pillar of PT also ‘­­as-​­else thinking’, which is a rather uncommon expression, especially when compared to ‘­­as-​­if ’ and ‘­­what-​­if ’ (­e.g., Kane, 2015). This term tries to capture our ability, through counterfactuals, to consider past events as if they were something else, something that is brought back to the realm of possibility from that of actuality. Of course, this is not to suggest confusing ‘­what might have been’ with ‘­what was’ and becoming trapped in the bubble of ­post-​­truth misinformation that is rampant today, both online and offline (­Rochlin, 2017). Counterfactual thinking and, more broadly, PT neither change the past nor do they aim to. When engaging in such thought, our aim is to reappropriate a large part of our experience (­arguably the largest) in ways that reveal the fundamental openness of time itself and of our being (­Heidegger, 1962).

What Is Not and Might Never be The limits of PT are set not by the a­ ctual – given ​­ the close interplay between PT and AT, mentioned throughout this ­chapter – ​­but by the impossible. And yet, the category of the impossible can and does serve to expand our understanding of the world, including the world as it is. This is because ‘­what is not’ is made up of a set of perspectives above and beyond those produced through pretence, anticipation and counterfactual thinking (­Merchant, 2017). It is only the unthinkable that has nothing to offer any existing set of perspectives, except perhaps the idea of unthinkability itself. In this context, ‘­what is not’ means ‘­what cannot be’, to differentiate it from the other processes discussed above that each engage with absence in their own way: pretence brings into the situation meanings that exist outside of it, anticipation constructs pictures of the ‘­not yet’ and counterfactuals are, by definition, not reflective of reality. The best way to capture the uniqueness of this pillar is to refer to utopias (­Levitas, 1990). Made up of two Greek words, ou or ‘­not’ and topos or ‘­place’, utopia literally means ‘­nowhere’, and it is a term popularised in the 16th century by Thomas More’s novel of the same name. The society described by More was impossible at the time of writing and it remained so, even if socialist ideologies (­U lam, 1965) claimed, in the 20th century, the imaginary existence of a perfect society. Utopias easily turn into dystopias, though, and this is also a big part of the lessons of the last century. Nowhere, like the impossible, stands in a dialogical relation to somewhere and to the possible. Utopian thinking is, in this context, a phenomenon that started to attract considerable attention precisely at a time when many people and communities worldwide find it difficult to adapt to a world that is unequal, fragmented and on the verge of environmental collapse. It might sound highly idealistic, but it is precisely in this day and age that we need utopias more than ever before. In recognition of the fact that some of the grandiose realities of the past are gone, many authors started proposing ­scaled-​­down, more manageable ‘­real’ (­Wright, 2011) or ‘­concrete’ utopias (­A lier, 1992). For the purposes of our discussion, these are just as much ­products – ​­and ­t riggers – of ​­ PT as revolutionary expressions (­Hermand, 1975). In fact, referring to utopias as a way of thinking expands their scope towards all those aspects of the world that can only ever live in our imagination. These are the most distant alternatives to ‘­what is’, so distant that we might not realise, easily, just how impossible they really are. They are often the product of what I call here ‘­­what-​­else thinking’, a mode of thought driven by dissatisfaction with the world ‘­a s is’ and a desire to r­ e-​­create it for the benefit of self, others, or society. What else can be thought of? How else can the world be? 89

Vlad P. Glăveanu

Studies of utopian thinking unsurprisingly come from those domains that reveal the dire need for radical transformation: social action (­Friedmann, 2000), education (­Peters & ­Freeman-​­Moir, 2006), sustainability (­De Geus, 2002), inequality and health (­­Scott-​­Samuel & Smith, 2015) and feminism (­Mellor, 1982). Few people reflect, however, on how utopias help us think about the ­possible – even ​­ when it is the most unlikely possible of them ­a ll – ​­and place us in a different state of mind ­v is-­​­­à-​­vis ‘­t ame’ perspectives that prevent us from dreaming, hoping and striving. This is also why, at the level of creative experience, I associate utopian thinking first and foremost with ­open-​­endedness, namely with the widest horizon made accessible to us by renouncing the use of ‘­what is’ as a measuring stick for all possibility. Utopias have a way of enticing one ‘­in’, but they don’t promise to return their followers in the same state as before. Once we understand, with the help of utopian thinking and PT, that the world we live in can be fundamentally r­ e-​­imagined, we get to discover possibilities for change everywhere and learn to enjoy them, regardless of their odds of success.

Concluding Thoughts on Thinking Wide and Narrow While there is some evidence, as noted above, concerning the relation between the four pillars or dimensions of PT, there is ample scope for future research to unpack how pretence, anticipation, counterfactual and utopian thinking can be primed and fostered, and to what effect. The hypothesis advanced based on existing studies and theories is that there is at least considerable overlap between them, an overlap referred to here as ‘­PT’. More than this, an even stronger claim is that PT and its counterpart (­not polar opposite), AT, are distinct systems of thinking, that they organise much more than our CC but cognition in general. These two modes can be called (­metaphorically) wide and narrow in the sense that PT expands our focus from ‘­what is’ in the situation to ‘­what could be’, ‘­what is to come’, ‘­what could have been’ and/­or ‘­what is not/­w ill never be’. Engaging in thought related to each one of these ­d imensions – ​­and combining them, a common occurrence in everyday life, as noted from the start of this ­chapter – ​­is assumed to offer us a qualitatively different experience of oneself, others and the world. Instead of zooming in on what we perceive or know, we are here invited to develop alternative perspectives that are both ­open-​­ended and multiple. It is this multiplicity and this openness that not only colour but effectively transform our experience in ways that stand in contrast to AT and its emphasis on singular, closed and focussed ideas and solutions. To take a concrete example of this experiential difference, let’s consider Jerome Bruner’s example of a conversation between the young German physicist Werner Heisenberg and his established colleague, Niels Bohr, that took place in Denmark, beside the Kronberg Castle. Bruner cites Gordon Mills, who, in turn, gives an account of what Bohr told Heisenberg on this occasion: Isn’t it strange how this castle changes as soon as one imagines that Hamlet lived here. As scientists we believe that a castle consists only of stones, and admire the way the architect put them together. The stone, the green roof with its patina, the wood carvings in the church, constitute the whole castle. None of this should be changed by the fact that Hamlet lived here, and yet it is changed completely. Suddenly the walls and the ramparts speak a different language. The courtyard becomes an entire world, a dark corner reminds us of the darkness of the human soul, we hear Hamlet’s ‘­To be or not to be.’ Yet all we really know about Hamlet is that this name appears in a ­thirteenth-​­century chronicle. No one can prove that he really lived here. But 90

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking

everyone knows the questions Shakespeare had him ask, the human depths he was made to reveal, and so he too had to be found a place on earth, here in Kronberg. (­Bruner, 1985, ­p. 45) This meditation on the ‘­nature’ of a castle offers a vivid contrast between thinking about ­actuality – ​­the construction as a pile of rocks put together by an a­ rchitect – ​­and thinking about p­ ossibility – ​­about how Hamlet’s story changes our perception of the material construction and invites us to reflect on the nature of death, of history and our relation to them. The fact that Hamlet’s presence is highly uncertain doesn’t impede the work of imagination and its transformation of the experience of walking by a Danish castle. We can still admire the actual stonework, but its sight is changed for us by the possible stories that make ‘­the walls and the ramparts speak a different language’. For Bruner, this is the language or narrative; for us, here, it is that of possibility, and the two have a great deal in common (­Glăveanu, 2020b). By referring to these modes as ‘­w ide’ and ‘­narrow’, I don’t intend to imply that one is superior or more desirable than the other. In fact, narrow thinking (­AT) is what often gets us through the day, safe and sound, able to solve practical problems efficiently by considering their nature and facts. Conversely, wide thinking (­PT) can easily get us into trouble in the sense of constructing too many alternatives that render us unable to act or simply have us ignore real constraints with either devastating or comical effects. We can be reminded here about the story of Thales of Miletus, who, gazing at the night’s sky, full of stars, fell into a hole right in front of him and was made fun of by his companion, a Thracian woman (­Blumenberg, 2015). Presumably, the philosopher was caught up in wonder, a state of mind specific to wide thought, and became oblivious to the actual danger on the ground. These anecdotes might be amusing, but they also hold the risk of dichotomising too sharply the two thinking systems. To solve even the most practical problems, one needs attention to facts and details, as well as the possibility to imagine alternatives, some of them far off from the situation at hand. So, in practice, we mostly experience blended, ‘­­narrow-​­wide’ or ‘­­wide-​­narrow’ thinking, depending on the slight dominance of one over the other. In the end, Thales might have missed a particular hole, allegedly, but he wasn’t unaware of his ­surroundings – ​­in fact, he was focussing his attention on a part of them, in the sky. There has been a lot written in cognitive science about the difference between System 1, fast and easy, and System 2 thinking, requiring more cognitive effort (­Evans, 2003), automatic and controlled thinking (­Bargh, 1989) or thinking fast and slow (­K ahneman, 2011). While there are some differences among dual processing models of cognition, the core of these distinctions has to do with how fast, effortless and essentially biased thinking is when interpreting the world or solving problems. Just like in the case of PT and AT, or wide and narrow thinking, there is no superior mode of thought, and it is widely recognised that automatic and controlled processes are both meant to adapt us to our environment and its various circumstances (­e.g., those moments when we have to decide quickly versus moments when we are asked to deliberate). It might be tempting to map fast into narrow and slow into wide thinking, but, as shown here, these are different organising dimensions of our cognition. AT and PT can be both fast or slow, effortless or effortful. Consider, for example, the difference between a daydream and writing a long essay on alien life. Both of these involve PT (­and, to some extent, AT), but the experience will be different. This holds even more truth when reflecting on the production of a work of art. Within a creative process, instances of fast narrow thinking and slow narrow thinking will be intertwined with fast wide thinking and slow wide thinking. Future empirical research could focus precisely on these intersections and what they tell us about the mind in its relationship to the body, to others, and to the world. 91

Vlad P. Glăveanu

And this is a last point I would like to return to. As mentioned before, what I wanted to advance here was a ‘­CCC’ approach to the notion of PT. And yet, not much has been said about CCC itself. What grants it the (­extra) qualifier of ‘­cultural’ is the fact that it considers creative processes and creative experience as expanded into the environment through physical, social and symbolic means. In other words, it studies creative thinking as embodied, material and cultural cognition rather than ­intra-​­cranial and ­de-​­contextualised. This general approach leaves an important mark on how we understand PT and AT as thought systems. Unlike the assumptions of universality that often accompany dual information processing models (­a critique Bruner, one of the architects of the ­so-​­called cognitive revolution, championed later on), PT and AT are modes of thought that are rooted in the physical and social embeddedness of our existence. Without physical tools, social relations and cultural resources, including language and meanings, there would be no pretence, anticipation, counterfactual or utopian thinking or, at least, not as we know them. AT would be equally impaired, given its closeness to ‘­what is’ the case in a given ­socio-​­material environment. In sum, CC is not the privileged realm of PT but a close interplay between possibility and actuality. This, if nothing else, is the key lesson of the cultural approach to what it means to think, to create and to live as a human being in a world of, with and for others.

References Alier, J. M. (­1992). Ecological economics and concrete utopias. Utopian Studies, 3(­1), ­39–​­52. Andrews, M. (­2014). Narrative imagination and everyday life. Oxford: Oxford University Press. Bacon, A. M., Walsh, C. R., & Martin, L. (­2013). Fantasy proneness and counterfactual thinking. Personality and Individual Differences, 54(­4), ­469–​­473. Bargh, J. A. (­1989). Conditional automaticity: Varieties of automatic influence in social perception and cognition. In J. S. Uleman & J. A. Bargh (­Eds.), Unintended thought (­­pp. ­3 –​­51). New York, NY: Guilford Press. Beghetto, R. A. (­2017). Inviting uncertainty into the classroom. Educational Leadership, 75(­2), ­20–​­25. Blumenberg, H. (­2015). The laughter of the Thracian woman: A protohistory of theory. New York, NY: Bloomsbury Publishing. Bodrova, E.,  & Leong, D. J. (­2015). Standing “­a head taller than himself ”: Vygotskian and ­post-​ ­Vygotskian views on children’s play. In J. E. Johnson, S. G. Eberle, T. S. Henricks & D. Kuschner (­Eds.), The handbook of the study of play (­Vol. 2, p­ p. ­203–​­214). Lanham, MD: Rowman & Littlefield. Boninger, D. S., Gleicher, F., & Strathman, A. (­1994). Counterfactual thinking: From what might have been to what may be. Journal of Personality and Social Psychology, 67(­2), ­297–​­307. Brockmeier, J. (­2009). Reaching for meaning: Human agency and the narrative imagination. Theory & Psychology, 19(­2), ­213–​­233. Bruner, J. (­1985). Actual minds, possible worlds. Cambridge, MA: Harvard University Press. Burnard, P., Craft, A., Cremin, T., Duffy, B., Hanson, R., Keene, J., Haynes, L., & Burns, D. (­2006). Documenting ‘­possibility thinking’: A journey of collaborative enquiry. International Journal of Early Years Education, 143, ­243–​­262. Carruthers, P. (­2002). Human creativity: Its cognitive basis, its evolution, and its connections with childhood pretence. The British Journal for the Philosophy of Science, 53(­2), ­225–​­249. Chappell, K., Craft, A., Burnard, P., & Cremin, T. (­2008). ­Question-​­posing and ­question-​­responding: The heart of possibility thinking in the early years. Early Years, 283, ­267–​­286. Craft, A. (­2003). Creative thinking in the early years of education. Early Years, 23(­2), ­143–​­154. Craft, A., Cremin, T., Burnard, P.,  & Chappell, K. (­2008). Possibility thinking with children in England aged ­3 –​­7. In A. Craft, T. Cremin & P. Burnard (­Eds.), Creative learning ­3 –​­11 and how we document it (­­pp. ­65–​­73). Stoke on Trent: Trentham Books. Craft, A., Cremin, T., Burnard, P., Dragovic, T., & Chappell, K. (­2012). Possibility thinking: Culminative studies of an ­evidence-​­based concept driving creativity? International Journal of Primary Education, ­3 –​­13, 41(­5), ­538–​­556.

92

A ‘Cultural Creative Cognition’ Approach to Possibility Thinking Craft, A., McConnon, L.,  & Matthews, A. (­2012). ­Child-​­initiated play and professional creativity: Enabling ­four-­​­­year-​­olds’ possibility thinking. Thinking Skills and Creativity, 7(­1), 4­ 8–​­61. Cropley, A. (­2006). In praise of convergent thinking. Creativity Research Journal, 18(­3), ­391–​­404. Currie, G. (­2006). Why irony is pretence. In S. Nichols (­Ed.), The architecture of the imagination (­­pp. ­111–​ 3­ 3). Oxford: Oxford University Press. De Geus, M. (­2002). Ecotopia, sustainability, and vision. Organization & Environment, 15(­2), ­187–​­201. de Vries, H. B., & Lubart, T. I. (­2019). Scientific creativity: divergent and convergent thinking and the impact of culture. The Journal of Creative Behavior, 53(­2), ­145–​­155. Dewey, J. (­1934). Art as experience. New York, NY: Perigree. Drayton, S., ­Turley-​­Ames, K. J., & Guajardo, N. R. (­2011). Counterfactual thinking and false belief: The role of executive function. Journal of Experimental Child Psychology, 108(­3), ­532–​­548. Epstude, K., & Roese, N. J. (­2008). The functional theory of counterfactual thinking. Personality and Social Psychology Review, 12(­2), ­168–​­192. Evans, J. S. B. (­2003). In two minds: ­Dual-​­process accounts of reasoning. Trends in Cognitive Sciences, 7(­10), ­454–​­459. Finke, R. A., Ward, T. B., & Smith, S. M. (­1992). Creative cognition: Theory, research, and applications. Cambridge, MA: MIT Press. Friedmann, J. (­2000). The good city: In defense of utopian thinking. International Journal of Urban and Regional Research, 24(­2), ­460–​­472. Frigg, R. (­2010). Models and fiction. Synthese, 172(­2), ­251–​­268. Gillespie, A. (­2006). Games and the development of perspective taking. Human Development, 49(­2), ­87–​­92. Glăveanu, V. P. (­2014). Thinking through creativity and culture: An integrated model. New Jersey, NJ: Transaction Publishers. Glăveanu, V. P. (­Ed.) (­2016). The Palgrave handbook of creativity and culture research. London: Palgrave. Glăveanu, V. P. (­2020a). Wonder: The extraordinary power of an ordinary experience. London: Bloomsbury Press. Glăveanu, V. P. (­2020b). The possible: A sociocultural theory. New York, NY: Oxford University Press. Glăveanu, V. P., & Beghetto, R. A. (­2020). Creative experience: A ­non-​­standard definition of creativity. Creativity Research Journal, early view. Goldman, A. I. (­2002). Simulation theory and mental concepts. In J. Dokic  & J. Proust, J. (­Eds.), Simulation and knowledge of action (­­pp.  ­1–​­19). Amsterdam: John Benjamins Publishing Company, Amsterdam. Hardy, B. (­1975). Tellers and listeners: The narrative imagination. London: University of London/­Athlone Press. Harris, P. L. (­2000). The work of the imagination. Oxford: Blackwell. Hastrup, K. (­2005). Performing the world: Agency, anticipation and creativity. Cambridge Anthropology, 25(­2), ­5 –​­19. Heidegger, M. (­1962). Being and time. New York, NY: Harper. Hermand, J. (­1975). The necessity of utopian thinking. Soundings, 58, ­97–​­111. Hetts, J. J., Boninger, D. S., Armor, D. A., Gleicher, F., & Nathanson, A. (­2000). The influence of anticipated counterfactual regret on behavior. Psychology & Marketing, 17(­4), ­345–​­368. Hillgren, P. A., Light, A., & Strange, M. (­2020). Future public policy and its knowledge base: Shaping worldviews through counterfactual ­world-​­m aking. Policy Design and Practice, 3(­2), ­109–​­122. Hjorth, D., Strati, A., Drakopoulou Dodd, S., & Weik, E. (­2018). Organisational creativity, play and entrepreneurship. Organization Studies, 39(­­2 –​­3), ­155–​­168. Hunleth, J. (­2019). Zambian children’s imaginal caring: On fantasy, play, and anticipation in an epidemic. Cultural Anthropology, 34(­2), ­155–​­186. Kahneman, D. (­2011). Thinking, fast and slow. London: Macmillan. Kane, E. (­2015). ‘­W hat If? As If ’, an approach to action research practice: ­Becoming-​­different in ­school-​­a ge childcare. Educational Action Research, 23(­3), ­350–​­365. Kavanaugh R. D., & Harris, P. L. (­1999). Pretense and counterfactual thought in young children. In L. Balter & C. S. ­Tamis-​­L emonda (­Eds.), Child psychology: A handbook of contemporary issues (­­pp. ­158–​ 1­ 76). Philadelphia, PA: Psychology Press. Krogh, S. (­2018). Anticipation of organizational change. Journal of Organizational Change Management, 31(­6), ­1271–​­1282. Landreth, G. L. (­2012). Play therapy: The art of the relationship. London: Routledge.

93

Vlad P. Glăveanu Levitas, R. (­1990). The concept of Utopia. Syracuse, NY: Syracuse University Press. Luo, Y., Chen, X., Qi, S., You, X., & Huang, X. (­2018). ­Well-​­being and anticipation for future positive events: Evidences from an fMRI study. Frontiers in Psychology, 8, 2199. Mellor, A. K. (­1982). On feminist utopias. Women’s Studies: An Interdisciplinary Journal, 9(­3), ­241–​­262. Merchant, B. (­2017). What we are called to do ‘­when the unthinkable becomes the thinkable and the impossible really happens’. International Journal of Qualitative Studies in Education, 30(­10), ­1027–​­1035. Moulton, S. T., & Kosslyn, S. M. (­2009). Imagining predictions: Mental imagery as mental emulation. Philosophical Transactions of the Royal Society B: Biological Sciences, 364(­1521), ­1273–​­1280. Nagel, T. (­1989). The view from nowhere. Oxford: Oxford University Press. Nasco, S. A., & Marsh, K. L. (­1999). Gaining control through counterfactual thinking. Personality and Social Psychology Bulletin, 25(­5), ­557–​­569. Peters, M. A., & ­Freeman-​­Moir, D. J. (­Eds.) (­2006). Edutopias: New Utopian thinking in education. London: Sense Publishers. Piaget, J. (­1926). Language and thought of the child. London: Kegan Paul, Trench, & Trubner. Plato (­1973/­ca. 369 B.C.). Theaetetus (­McDowell, J. Trans.). Oxford: Clarendon Press. Poli, R. (­2017). Introduction to anticipation studies. Cham: Springer. Rochlin, N. (­2017). Fake news: Belief in ­post-​­truth. Library Hi Tech, 35, ­386–​­392. Roese, N. J. (­1997). Counterfactual thinking. Psychological Bulletin, 121(­1), 1­ 33–​­148. Roese, N. J., & Olson, J. M. (­Eds.) (­1995). What might have been: The social psychology of counterfactual thinking. Mahwah, NJ: Erlbaum. Schaefer, C. E., & Drewes, A. A. (­2011). The therapeutic powers of play and play therapy. In C. E. Schaefer (­Ed.), Foundations of play therapy (­2nd ed., ­pp. ­15–​­25). Hoboken, NJ: Wiley. Schroyens, W. (­2010). A ­meta-​­analytic review of thinking about what is true, possible, and irrelevant in reasoning from or reasoning about conditional propositions. European Journal of Cognitive Psychology, 22(­6), ­897–​­921. ­Scott-​­Samuel, A., & Smith, K. E. (­2015). Fantasy paradigms of health inequalities: Utopian thinking? Social Theory & Health, 13(­3), ­418–​­436. Segura, S., ­Fernandez-​­Berrocal, P.,  & Byrne, R. M. (­2002). Temporal and causal order effects in thinking about what might have been. The Quarterly Journal of Experimental Psychology Section A, 55(­4), ­1295–​­1305. Seligman, M. E., Railton, P., Baumeister, R. F., & Sripada, C. (­2016). Homo prospectus. New York, NY: Oxford University Press. Singer, J. L. (­2014). Daydreaming and fantasy. London: Routledge. Sirois, F. M. (­2004). Procrastination and counterfactual thinking: Avoiding what might have been. British Journal of Social Psychology, 43(­2), ­269–​­286. Sutherland, S. L., & Friedman, O. (­2012). Preschoolers acquire general knowledge by sharing in pretense. Child Development, 83(­3), ­1064–​­1071. ­Sutton-​­Smith, B. (­1966). Piaget on play: A critique. Psychological Review, 73, ­104–​­110. Ulam, A. (­1965). Socialism and utopia. Daedalus, 103, ­382–​­400. Valsiner, J. (­2007). Culture in minds and societies: Foundations of cultural psychology. New Delhi: Sage. Van Lente, H. (­2012). Navigating foresight in a sea of expectations: Lessons from the sociology of expectations. Technology Analysis & Strategic Management, 24(­8), 7­ 69–​­782. Vygotsky, L. S. (­1967). Play and its role in the mental development of the child. Soviet Psychology, 5(­3), ­6 –​­18. Vygotsky, L. (­1978). Mind in society. Cambridge, MA: Cambridge University Press. Vygotsky, L. S. (­2004). Imagination and creativity in childhood. Journal of Russian & East European Psychology, 42(­1), ­7–​­97. Ward, T. B., Smith, S. M., & Finke, R. A. (­1999). Creative cognition. In R. J. Sternberg (­Ed.), Handbook of creativity (­­pp. ­189–​­212). Cambridge: Cambridge University Press. Weber, S. (­1996). Counterfactuals, past and future. In P. E. Tetlock & A. Belkin (­Eds.), Counterfactual thought experiments in world politics (­­pp. ­269–​­288). Princeton, NJ: Princeton University Press. Weisberg, D. S. (­2015). Pretend play. Wiley Interdisciplinary Reviews: Cognitive Science, 6, ­249–​­261. Williams, A. M., & Ward, P. (­2007). Anticipation and decision making: Exploring new horizons. In G. Tenenbaum & R. Eklund (­Eds.), Handbook of sport psychology (­3rd ed., p­ p. ­203–​­223). Hoboken, NJ: John Wiley. Wright, E. O. (­2010). Envisioning real utopias. London: Verso. Wright, E. O. (­2011). Real utopias. Contexts, 10(­2), ­36–​­42.

94

7 ANALOGY AND THE TRANSFER OF CREATIVE INSIGHTS Thomas C. Ormerod

Introduction Early 20th century researchers, notably from the Gestalt school (­e.g., Duncker, 1945; Kohler, 1925; Wertheimer, 1938), identified insight as a unique kind of p­ roblem-​­solving associated with a range of phenomena such as impasse, functional fixedness, incubation, and an experience of an ‘­a ha’ moment upon achieving insight. With the advent of behaviourist and ­information-​­processing accounts of cognition, interest waned somewhat, but over the past 30 years, particularly since Ohlsson’s (­1984a, 1984b) seminal ­re-​­examination of the concept of restructuring as information processing, there has been a resurgence of interest in insight ­problem-​­solving. Much of that wave of research has focussed on the experience of insight (­e.g., Bilalić et al., 2019; Bowden & ­Jung-​­Beeman, 2003; Danek et al., 2014; Webb et al., 2018), on sources of difficulty in insight ­problem-​­solving (­e.g., Jones, 2003; Kershaw & Ohlsson, 2003; Weisberg, & Alba, 1981), and on the conditions under which insight is most likely to arise (­e.g., Patrick & Ahmed, 2014; Sio & Ormerod, 2015; Smith & Blankenship, 1991; ­Vallée-​­Tourangeau et al., 2016). However, uncertainty remains as to where insights themselves come from. Once sources of difficulty are overcome and conditions are optimal, how does a p­ roblem-​­solver choose what to focus upon that might yield an insightful solution? This chapter points to the role that analogy can play in providing a source of insight, and also how insight can mediate the experience of analogy. As previous chapters have shown, many definitions of insight have been adopted, some focussing on phenomenology (­the presence of impasse, ‘aha’, etc.), others focussing on task ­ on-​­insight problems) and still characteristics (­d istinguishing between ­so-​­called insight and n others focussing on processes (­the ‘­special process’ versus ‘­business as usual’ debate). None of these definitions captures what, arguably, might be seen as the broader purpose of insight: to turn the enlightenment generated from current ­problem-​­solving activity into knowledge that can be utilised in future ­problem-​­solving. This chapter argues that this transfer of knowledge, typically described as analogical transfer, is the very essence of insight. Recently, Ormerod et  al. (­2023) have proposed that insight should be defined not in terms of phenomena, tasks, or processes but as the outcome of its achievement, that is, what new knowledge is acquired through p­ roblem-​­solving. They argue that insight might arise because of reaching a solution to a problem or during solution attempts, irrespective of their DOI: 10.4324/9781003009351-8

95

Thomas C. Ormerod

success. Thus, for Ormerod et al., an ‘­insight problem’ is defined as one in which ideas that are new to the solver in the context of solving that problem must be discovered, and insight is the discovery of that new knowledge or idea. Using an analogical transfer paradigm focusses research questions on what is discovered during ­problem-​­solving and addresses questions concerning the value, durability and flexibility of insights acquired during p­ roblem-​­solving. This creates a common link between insight and analogy: they are both about learning for future performance. Just over 40 years ago, Mary Gick and Keith Holyoak opened an exceptionally productive line of research exploring analogical transfer between variants of Duncker’s (­1945/­1951) radiation problem (­Gick & Holyoak, 1980; 1983), widely recognised as having the characteristics of a problem requiring insight to solve. Their work is usually pitched less as a study of insight ­problem-​­solving than as a demonstration of the power of analogy to transform our theoretical understanding of human thought. Despite its focus on solutions to a classic problem seen as evoking insight characteristics, it affirmed the central place of analogy as a learning and ­problem-​­solving mechanism in cognitive architectures such as Anderson’s (­2014) ­ACT-​­R (­Adaptive Character of ­Thought – ​­Rational) framework. As well as having theoretical importance, analogy was seen as key to successful education and training (­e.g., Kolodner, 1997). Welling (­2007) identified analogy as one of the main processes involved in creative thinking, and it appears to play an important role in creative design (­e.g., Ball & Christensen, 2019; Bearman et al., 2007). In addition to understanding how analogical transfer might arise in laboratory conditions, we are beginning to look at how it occurs in everyday situations and how studying situated analogical transfer might give new insights into insight. For example, Monaghan et  al. (­2 015) explored how sleep and incubation affect the analogical transfer of insight ­problem-​­solving skills. In Experiment 1, participants were exposed to a set of source problems that invoke insight characteristics (­e.g., the Four Trees problem). Then, after a ­12-​ ­hour period involving sleep or wake, they attempted target problems structurally related to the source problems but with different surface features. Experiment 2 controlled for time of day by testing participants either in the morning or the evening. Sleep improved analogical transfer, and the effects were not due to improvements in subjective memory or similarity recognition but rather appeared to be caused by structural generalisation across problems. There is current controversy over the effects of sleep on ­problem-​­solving performance (­e.g., Schönauer et al., 2018), and these results need replication, but they point to an interesting ­sub-​­conscious relationship between insight and analogy that is worthy of further exploration. For two decades, analogy was a major topic in cognitive science research, with roughly equal numbers of articles from Google Scholar mentioning analogy and insight in their titles, abstracts or introductions (­approximately 2000 in each case from 1980 to 1995). Since 1996, there have been five times as many articles mentioning insight as analogy, so it seems the halcyon days of analogy research are behind us. In the remainder of this chapter, I examine the significance of analogical transfer and how research into it may have lost its way. I will argue that analogical transfer should be central to the development of the next generation of theories of insight p­ roblem-​­solving. Insight researchers tend to be interested in studying how insight is achieved but pay less attention to what insights might be used for once they have arrived. One thing that analogy research offers is a clear purpose: to tackle future problems. A focus on future rather than current performance should also be a goal of insight research.

96

Analogy and the Transfer of Creative Insights

The Nature of Analogical Transfer and Its Significance for Insight Research An analogy can be defined as a set of correspondences that hold between a ‘­target’ an individual is trying to understand, communicate or solve and a ‘­base’ or ‘­source’ problem that provides knowledge or solutions from experience. We can use a process of analogical transfer to apply the source domain knowledge to enhance our understanding of the target domain knowledge or use a known solution to a source problem to solve a target problem. Arguably, in everyday life, people rarely attempt to solve a new problem for which they have absolutely no relevant experience. Instead, we call on knowledge from experience and so may seek to draw analogies between our current situation and situations we have encountered previously. The study of analogy, from the Greek ana logos or ‘­same logic’, has a long history dating back to Aristotle (­for a summary, see Hoffman et al., 2009). Analogies can be literal similarities (­e.g., a cheese sandwich is analogous to an egg sandwich), in which the objects have shared categories (­e.g., foodstuffs) and attributes (­sandwich fillings), making the identification of relations between objects and their attributes in base and target straightforward. Researchers tend not to be too impressed by theories that explain only how we use literal similarities, since the process does not seem much different from recalling and applying relevant knowledge from memory. Much more impressive is the kind of analogy that draws together two problems that seem at first to be entirely dissimilar superficially but turn out to have common underlying conceptual similarities. For example, the 19th-​­century scientist Rutherford used the solar system as an analogy to determine key features of the structure of the atom. This analogy is impressive, not only for the product (­a major scientific advance), but also for the fact that Rutherford was able to detect the relevance of the solar system’s ‘­source’ problem and apply its solution to the ‘­target’ problem of the structure of the atom. Rutherford was able to transfer the solar system concept and use it to derive new inferences about how the atom must be structured, inferences that were not part of the solar system concept itself, a process of analogical reasoning to derive new information and understanding. Gentner (­2003) argues that analogy is central to human cognitive performance. She claims, “­the great value of analogy… lies in creating a focus on common relational systems and thus lifting a relational pattern away from its object arguments” (­­p. 201). Gentner recognises that similarities between the attributes of objects are easier to identify than relational similarities. According to her, the process of analogy overcomes this problem. Early comparisons focus upon superficial similarities based on objects or attributes, and the continuing application of a comparison process leads individuals to develop an understanding of conceptual similarities based on relations. Analogy plays a critical role in the A ­ CT-​­R framework for modelling human thought (­A nderson, 2014), being a source for learning new procedures. For example, Singley and Anderson (­1985) demonstrate how the acquisition of skills in using a new numeric keyboard is influenced by the positive or negative transfer of previous experience with different keyboards. Gust et al. (­2008) take the view that analogy is central to human cognition a step further, claiming that analogy is the fundamental process underlying all human thought, including deduction, abduction and induction. They point out how some concepts, such as ‘­heat’, are not observable and can only be conceptualised using analogy. One understands how heat is transferred by analogy to water flowing from one vessel to another, an example of generating knowledge by abduction. While some see analogy as a good thing and central to how people think, there are those who think otherwise. For example, while generally arguing for the value of analogies for

97

Thomas C. Ormerod

managers in strategic ­decision-​­making, Gavetti et al. (­2005) illustrate how analogical reasoning can lead to poor ­decision-​­making. They discuss the case of Enron, the multinational energy corporation that failed in 2001 amid allegations of fraud and corruption. Gavetti et al. suggest that Enron executives looked to expand into new markets that shared relational similarities with energy trading markets. One specific example was Enron’s expansion into the supply of broadband Internet. The analogy focussed on relational similarities between broadband and energy markets (­e.g., fragmented demand, rapid change because of technological progress, etc.), and this focus masked important differences, such as wholesale (­energy) versus d­ irect-­​­­to-​­customer (­broadband) supply chains, unproven technology, domination of the market by telecoms companies and so on. The Enron analogy provides an illustration of negative transfer, when applying prior knowledge from a source domain impairs performance in a target domain. Negative transfer arises when there are both superficial and structural similarities and is less likely with only structural similarities. Thus, negative transfer provides an illustration of the paradox of analogy: superficial similarities make it easier to analogise, but they also make it more likely that poor analogies will be chosen.

A Shift Away from Analogical Transfer Research As noted above, the influence of research into analogical transfer has decreased over the past two decades, and I suggest three reasons why this might be the case: the complexities of the frame problem; the problem of superficial versus conceptual reminding; and a lack of empirical evidence for spontaneous analogical transfer. It turns out, variants of these problems are also faced by researchers investigating insight ­problem-​­solving. The frame problem: When Rutherford retrieved knowledge of the solar system from his memory, how did he know that it would be relevant to the structure of the atom? Analogy presents a ­chicken-­​­­and-​­egg problem: in seeking a source that is relevant to a target, you need to have some idea of how to judge whether similarities between them are relevant. But how can you judge similarities before you have found a potentially relevant source? The difficulty of searching for appropriate analogs is an example of the frame problem (­Hayes, 1981), which describes the difficulty of determining from potentially infinite effects of information those that are useful and those that are trivial. You might search through every bit of knowledge you have and end up with nothing that appears useful to the target. Worse, you might see too much of potential relevance and be unable to choose. Worse still, you might find what seems to be a relevant analogy, but it turns out to be an irrelevant distractor that leads to the wrong solution. Equally problematic, you might select the right analogy but misapply it, in the process deciding erroneously that it is not an analogy after all. Insight ­problem-​­solving research is similarly challenged by concerns over the frame problem (­e.g., Ormerod et al., 2022). In solving an insight problem, one faces the same dilemma: how do you know where to look for information that is relevant to the solution among a potentially infinite set of knowledge or ideas available in the internal world (­from memory) or the external world (­the task environment). If you knew where to look, then insight would be trivial because the information you need to solve the problem would be automatically activated at the point it is needed. But if you don’t know where to look because you don’t yet know what information is needed for a solution, how do you know when you have found the relevant ideas for a solution? Superficial versus conceptual reminding: Identifying similarities appears to be driven by recalling objects, events or concepts that share similar associations in memory, and most theories of memory predict that the stronger the association, the more likely an item is to be 98

Analogy and the Transfer of Creative Insights

retrieved. Yet what one wants for an analogy to be useful (­i.e., for it not to deliver what you already know about the target domain) is not usually superficial similarities but conceptual ones like those discovered by Rutherford in his solar s­ ystem-​­atom analogy. As Fodor (­1983) points out, analogical reasoning “depends precisely on the transfer of information among cognitive domains previously assumed to be irrelevant” (­­p. 105). Insight researchers must also understand the roles played by superficial and conceptual problem content in determining ­problem-​­solving performance. The phenomenon of functional fixedness illustrates the concern, in which individuals are unable to see alternative roles for objects that have common uses. Duncker’s (­1945/­1951) Box problem (­often referred to as the ‘­candle problem’) serves as a prototypical example of a problem that requires solvers to see new uses for objects and not to fixate on their usual use. In one of Duncker’s experiments (­­pp. ­88–​­90), participants in one condition were presented with three boxes, one filled with three candles, another with tacks and a third with matches. In a second condition, participants received the same items, but the boxes were filled with items like buttons. In each case, the task was to attach the candles to a wall and light them. Participants in the first condition were much more likely to solve the problem, Duncker argued, because the ‘­storage’ function of the boxes was reduced in salience compared with the condition showing boxes as containers of buttons, etc. The participant numbers in Duncker’s studies were low, but the concept of functional fixedness has been frequently demonstrated elsewhere (­e.g., ­Munoz-​ ­Rubke et al., 2018; Neroni & Crilly, 2021). Common object functions are, in the context of problems requiring insight to solve, the superficial content, while the alternative uses to which they might be put represent conceptual understanding. One contemporary theory of insight, the Representational Change theory of Knoblich et al. (­1999), puts the constraints of prior knowledge, a generalised form of functional fixedness, at the forefront of explaining why insight is difficult to achieve. Interestingly, Chrysikou and Weisberg (­2 005) demonstrate how fixation can inhibit transfer of design concepts, suggesting that a superficial understanding of problem solutions can inhibit ­real-​­world ­problem-​­solving. More encouragingly, they also demonstrate that conceptual transfer can be enhanced by giving participants ­de-​­fixation instructions. Empirical evidence: If analogy is fundamental and automatic, then one might expect there to be overwhelming positive evidence for its occurrence. However, evidence for analogical transfer, particularly spontaneous rather than prompted, is equivocal. For example, while 90% of Gick and Holyoak’s (­1980) participants solved the transfer task when given a hint to use the source problem, only 20% solved the transfer task in the absence of a hint to analogise. One review by Detterman (­1993) claims that there is little evidence that transfer occurs reliably. Similarly, in a systematic analysis of transfer research, Barnett and Ceci (­2002) claim that there is evidence, but it is confounded by the lack of a proper framework for identifying the moderators of transfer, such as whether transfer is to near or far domains (­where distance might be in time, application domain, format of presentation and so on), flexible or rigid, spontaneous or prompted and so on. They question “ ­just how significant, flexible, unprodded, and far does transfer have to be to count?” (­­p. 619). The framework they offer for classifying transfer studies distinguishes between two h ­ igher-​­order dimensions: content (­i.e., what is being transferred) and context (­i.e., when and where knowledge is transferred from and to). They claim the content dimensions encompass issues affecting the spontaneity and generality of analogical transfer, while context determines whether transfer is between near (­i.e., highly similar) or far (­i.e., seemingly very different) domains. Barnett and Ceci suggest that the ultimate demonstration of successful analogical transfer and the one that is key for 99

Thomas C. Ormerod

educational practice would be one in which far transfer from source to target is achieved across all context and content dimensions (­i.e., source and target separated by a long time, in a different physical context, with different knowledge domains and so on). However, in their review of studies, they found none. Insight researchers do not have the same problem regarding the absence of empirical evidence. However, they do face an alternative problem: each theorist tends to select problems to test their theories that are perfect for that purpose but do not generalise. For example, proponents of representational change theory (­e.g., Knoblich et al., 1999) use s­ingle-​­move ­k nowledge-​­rich problems like matchstick algebra problems, which are perfect for demonstrating the constraints of prior knowledge on performance. Proponents of the criterion of satisfactory progress theory (­e.g., MacGregor et al., 2001) use m ­ ulti-​­move ­k nowledge-​­lean problems like the ­n ine-​­dot problem, which are perfect for demonstrating the role of search and ­progress-​­monitoring (­though see Ormerod & MacGregor, 2020, as a test of both representational change and the criterion of satisfactory progress theories using matchstick algebra problems). Theorists have not, until recently, used analogy as a tool to test theories of insight. Indeed, Knoblich et al. (­1999) argued that, in the case of matchstick algebra, “­Once a problem representation has been changed, the change should persist and so should transfer to all relevant subsequent problems. Hence, differences in initial difficulty due to the need to relax constraints should disappear.” (­­p. 1535). However, as Ormerod et al. (­2006) demonstrate with a range of coin problems, solving a source problem with insight does not necessarily lead to transfer to a target problem. Indeed, even r­e-​­solution of the same problem can be problematic. Ormerod et al. also showed that most ­non-​­naïve participants failed to ­re-​­solve the ­n ine-​­dot problem at the first attempt, despite many being aware of the solution shape or the hint to ‘­go outside the square’ of dots. In what follows, I outline how analogy is an important part of achieving insight, that is, how it provides a strategy for solving problems to generate insights. I then look at how insight is a key determinant of successful analogical transfer. The fact that analogy can determine insight and vice versa demonstrates how closely the two topics are interwoven.

Using Analogy to Achieve Insight One of the key messages from Gick and Holyoak’s (­1980, 1983) studies of analogical transfer between variants of Duncker’s radiation problem is that analogy offers a strategy for solving problems that appear to require a degree of insight to solve. The significance of analogy as a strategy extends beyond laboratory studies of puzzle solving. For example, Dunbar (­1995) describes how scientists use analogies to address scientific questions, and he suggests their use is routine rather than occasional, an observation also made by Bearman et al. (­2007) in the management domain. However, Dunbar notes that, as in laboratory studies, in vivo studies of scientists in the field show that most of their analogies are drawn from closely related domains. Whether the analogy use reported by Dunbar would count as a ‘­far transfer’ in Barnett and Ceci’s scheme is moot. Another demonstration of how an analogical reasoning strategy can aid insight comes from Christensen and Shunn (­2005). They used a technique of ‘­reverse analogy’ introduced by Gick and Holyoak (­1980, Experiment 5) to explore insight p­ roblem-​­solving. Participants in their study were initially presented with the target problems and were then presented either with analogous problems that they did not solve but had to rate for difficulty or with 100

Analogy and the Transfer of Creative Insights

distractor problems that had no relationship with the target problems. Participants then attempted the target problems again. The proportions of problems eventually solved that had not been solved prior to the presentation of analogous or distractor problems were .53 and .17 for the a­ nalogous-​­cued and d­ istracter-​­cued problems, respectively. Thus, it appears that participants were spontaneously able to utilise analogies that facilitated insight when they had worked on the target problem prior to the analogs being made available. What is particularly impressive about Christensen and Shunn’s data is that participants were able to benefit from the presence of analogs even though all problems were presented within the same ­45-​­minute trial; in other words, they were able to discriminate between relevant analogs and distractors. Moreover, the analogs created by Christensen and Shunn were constructed to minimise superficial similarity with the target problems. Thus, although the source and target analogs were presented within the same p­ roblem-​­solving session, we begin to see a degree of far transfer emerging in this study. Recently, Ormerod and Gross (­2023) have offered a contender for the badge of far transfer. In our study, we examined the transfer of skills from different domains of expertise for overcoming impasses in insight ­problem-​­solving. In our first study, we compared the performance of two highly experienced expertise groups, financiers (­bankers, analysts, investment managers, etc.) and designers (­architects, graphic designers, design engineers, etc.), in solving verbal problems (­e.g., the 54BC coin p­ roblem – Metcalfe, 1986) or visual problems ​­ ​­ et al., 2002). We found that both expertise groups (­e.g., the e­ ight-​­coins ­problem – Ormerod solved the verbal problems at an equivalent rate (­58% solved by designers, 63% by financiers), but designers solved more visual problems than financiers (­60% vs. 27%, respectively). Although this result demonstrates far transfer of design expertise to insight ­problem-​­solving, it is hardly surprising: one would expect experts in visual p­ roblem-​­solving to be better at solving visual problems. What was revealing about these results was that initially, designers and financiers made the same types of unsuccessful attempts until reaching an impasse. It was after the impasse that their behaviours differed. Designers made more sketches, moved the problem objects around physically more often, changed the orientation of the problem array and made more hand gestures. It appears they were using strategies for exploring the visual problem representation that form part of their repertoire of visual design skills. In a second experiment, we explored what would happen if expert designers were unable to interact physically in the same way with visual problems. Two groups of participants, experienced designers and novices (­undergraduate students taking ­design-​­related degrees), were further assigned to one of three conditions: solving while sitting on their hands (­i.e., unable to draw, gesture or move objects), ­hands-​­free but unable to draw, and unconstrained performance. We found that, while the novices were largely unaffected by the constraints on performance (­46%, 48% and 48% solved for n ­ o-​­hands, ­no-​­drawing and unconstrained conditions, respectively), experts showed an increasing solution advantage with reduced constraints (­40%, 57% and 79% solved for n ­ o-​­hands, ­no-​­drawing and unconstrained conditions, respectively). These results confirm the importance of physical interaction in solving the visual problems but also show that the advantages of physical interaction in this case are limited to their execution as part of an expert skill set. The kinds of analogical transfer that facilitated the solution of problems by invoking the characteristics of insight in the experiments of Gick and Holyoak (­1980, 1983) and of Christensen and Shunn (­2005) are very different from those observed by Ormerod and Gross (­2023). In the former case, what is being transferred is solution knowledge, that is, a schematic understanding of the necessary components of a problem solution (­e.g., the splitting ​­ and convergence solution schema of the radiation p­ roblem – Gick & Holyoak, 1983). In the 101

Thomas C. Ormerod

latter case, what is being transferred is strategic knowledge of how to overcome impasse, that is, a set of procedures for changing the way in which the problem space is searched. This distinction maps onto two major theoretical approaches to explaining insight ­problem-​ ­solving: the Representational Change Theory of Knoblich et al. (­1999), and the Criterion for Satisfactory Progress theory of MacGregor et al. (­2001). The relatively limited success of eliciting spontaneous analogical transfer of solution knowledge is consistent with the concept of Representational Change Theory that prior knowledge typically constrains the ideas that individuals generate to solve problems, constraints that need to be relaxed for a solution to be found. The finding of spontaneous transfer of strategic knowledge by Ormerod and Gross indicates how equipping individuals with strategies for overcoming impasse can enhance insight without invoking p­ roblem-​­specific prior knowledge. It might reasonably be argued that, while transfer of solution knowledge is analogical in nature, transfer of strategic knowledge is more generic than analogical. At the same time, manipulations of strategic rather than solution knowledge appear to offer a more promising avenue for achieving the far transfer sought by Barnett and Ceci (­2002).

Using Insight to Achieve Analogical Transfer The relative lack of evidence for spontaneous analogy in laboratory studies might lead one to conclude that analogical transfer is less important than previously thought, and as suggested above, there does seem to have been something of a move away from researching the topic. Yet when one looks at the role that insight can play in achieving analogical transfer, there is room for some optimism. For example, Needham and Begg (­1991, Expt. 1) investigated the impacts on analogical transfer of p­ roblem-​­oriented processing, in which participants were asked to explain why the solutions to source problems were correct, versus ­memory-​­oriented processing, in which participants were told to memorise solutions for later recall. Of the five problems presented to participants (­including, for example, Maier’s, 1931, t­wo-​­string problem), participants in the ­problem-​­oriented group solved 90% of the target problems compared to 69% for the ­memory-​­oriented group. It appears that explaining the nature of a solution to the source problem helps participants achieve greater insight into that solution, which, in turn, benefits spontaneous analogical transfer. Several other i­nsight-​­oriented ­problem-​­solving activities seem to facilitate analogical transfer. Chen and Deihler (­2000) found that asking participants to create concrete examples of abstract source problem descriptions led to increased levels of spontaneous analogical transfer. Minervino et  al. (­2017) explored the impacts of requiring participants to create their own analogous target problem on transfer between variants of the radiation problem. They found a significant, albeit relatively modest, increase in analogical transfer, with 25% of participants who created their own analogous problem subsequently solving the target problem compared with 10% of those receiving the standard transfer condition. However, 45% of participants who created a fully analogous problem solved the target problem, compared with 18% who generated an incomplete or incorrect analogous problem. What these manipulations have in common is that they are designed to get participants to think more about the problems they are solving, instilling in them a deeper understanding of the structural and conceptual properties of source and target problems. They encourage participants to go beyond merely knowing the solution to the source problem and to create a deeper conceptual, indeed insightful, understanding of the nature of the source and target problems. Other researchers have shown that it is not just getting participants to think that helps promote spontaneous analogical transfer; it also helps to get participants who can think 102

Analogy and the Transfer of Creative Insights

more effectively. Kubricht et al. (­2017) demonstrate how fluid intelligence is a strong predictor of spontaneous analogical transfer between variants of the radiation problem. They also showed that presenting participants with an animation of the source problem facilitated spontaneous analogical transfer, especially for participants with lower levels of fluid intelligence. It seems as if individuals who can think carefully about the source and target problems achieve an insight that enables transfer, and that spelling out the nature of the source and target problems to those less able to think about them also helps. Similarly, Cushen and Wiley (­2018) have shown how the ability to solve compound Remote Associates Tasks is a strong predictor of spontaneous analogical transfer between variants of the radiation problem. They argue that an ability to make remote associations may be conducive to constructing a broad representation of a source problem; essentially, the wider the spread of ideas that are activated during solving a source, the richer its representation will be, and hence the more likely is its transfer to an analog. An alternative way of looking at the importance of insight for analogical transfer is to examine the effects of a failure of insight. For example, Gick and McGarry (­1992) examined analogical transfer with the Mutilated Checkerboard task. You are given a checkerboard and 32 dominoes. Each domino covers exactly two adjacent squares on the board. Thus, the 32 dominoes can cover all 64 squares of the checkerboard. Now suppose two squares are cut off at diagonally opposite corners of the board. If possible, show how you would place 31 dominoes on the board so that all the 62 remaining squares are covered. If you think it is impossible, give a proof of why. They devised two versions of an analog source problem, the ‘­ d inner party’ problem (­essentially swapping dominoes on black and white squares for male and female couples seated at dinner), one of which emphasised parity of gender, the same parity concept being central to the Mutilated Checkerboard solution and the other removing a parity emphasis. They found (­Experiment 2) that participants who solved the ­parity-​­emphasised source problem were less likely to solve the Mutilated Checkerboard transfer problem than participants who solved the p­ arity-​­absent problem, even though the p­ arity-​­emphasised source was much easier to solve than the ­parity-​­absent source. The interpretation of this result offered by Gick and McGarry was that the increased difficulty of the ­parity-​­absent source problem forced participants to create a richer representation of the source problem; in other words, to obtain greater insight into its structural properties. This interpretation was confirmed in a series of experiments by Didierjean and Nogry (­2004), who gave participants a ­post-​­transfer classification task requiring them to sort problem variants according to structural similarity to the Mutilated Checkerboard problem. Participants receiving a ­parity-​­absent source problem classified more problems correctly (­55%) than those receiving a p­ arity-​­emphasised source problem (­24%), suggesting that the greater insight obtained in solving a more difficult source created a richer representation of its structural properties. It seems as if entering a state of impasse, in which participants are unable immediately to solve the source problem, may make the knowledge gained from solving it more useful in tackling a target analog. Ross (­2021) has proposed that impasse can have two natures: being ‘­stuck’, which does not correlate strongly with eventual insight, and being ‘­challenged’, which does. In this sense, impasse can act both as a block to p­ roblem-​­solving and as a motivator that encourages solvers to seek new ways of tackling the problem, a result consistent with Gick and McGarry’s and Didierjean and Nogry’s (­2004) findings. 103

Thomas C. Ormerod

A related issue concerns problem difficulty. For example, Ormerod et al. (­2006) examined transfer between the ­four-​­tree problem (­Metcalfe & Wiebe, 1987) and the ­four-​­coin problem (­Ormerod et al., 2002). The ­four-​­tree problem asks how a gardener can plant four trees so that the trees are equidistant, the f­our-​­coin problem asks how four identical coins can be arranged so that each one touches each of the other three. The problems have superficial similarities (­e.g., the number four) and conceptual similarities (­e.g., the solution to both requires the use of three dimensions). We found an asymmetry in transfer, in which participants solved the f­our-​­coin problem more often (­57%) when preceded by the f­our-​­tree problem than when they solved it first (­30%), but no positive transfer effect was found with the ­four-​­tree problem (­3% vs. 4% solved when it was first vs. second). The ­four-​­tree problem is considerably more difficult for participants than the ­four-​­coin problem, possibly because it is easier to conceive of and implement a physical solution to the latter. Our findings mirror a ­well-​­known finding by Reed et al. (­1974) that problem difficulty mediates analogical transfer, in their case transfer between variants of the Hobbits and Orcs problem. It seems as if preceding a target problem with a relatively easier source problem does not induce the depth of conceptual understanding necessary to support analogical transfer. A negative impact of a failure of insight on analogical transfer speaks to the general hypothesis that the more individuals are able or encouraged to think about the conceptual structure of the source problem, the more likely it is for spontaneous analogical transfer to occur. However, not all thinking has a positive effect. Bearman et al. (­2011), again exploring transfer between variants of the radiation problem, examined the effect of asking participants to evaluate the sufficiency of a source problem’s solution on spontaneous analogical transfer. They introduced a procedure in which participants selected one card from a set of five cards, each of which had written on it a solution to the source problem. Participants in the evaluation conditions were told that each card had an alternative solution and that they should “­evaluate whether the suggestion is a good solution to the problem”; in fact, all cards had identical solutions. Participants in control conditions were told to “­read out loud and summarise the problem and suggested solution”. Across four experiments, we found a negative effect of evaluation on solution rates to the target problem. We suggest that participants in evaluation conditions ­over-​­elaborated the source problem and its solution. When participants were asked to evaluate, they produced much more extraneous information than did participants in other conditions. Thus, it seems possible to overthink as well as think insightfully. So far, the evidence for an effect of insight p­ roblem-​­solving on analogical transfer has focussed on the availability or otherwise of solution knowledge, yet as we saw with the effects of drawing analogies on insight ­problem-​­solving, effects of insight on analogical transfer can come from strategic knowledge as well. For example, Gick and McGarry (­1992) found that inducing failures to solve a source problem that mapped onto the kinds of failures individuals typically make in solving the target problem enhanced analogical transfer. In our own research, we have found that the experience of failure during attempts to solve a problem can facilitate positive analogical transfer (­Ormerod & MacGregor, 2017). In a series of experiments, we examined transfer between variants of the ­n ine-​­ball problem (­g iven nine balls, one of which is fractionally heavier than the others, use a balance scale twice only to find the odd ball). The n ­ ine-​­ball problem is difficult, we argue (­Ormerod et al., 2013), because participants typically, and wrongly, try to maximise the number of balls weighed on the first weighing by placing four versus four or even four versus five balls on the scales. We have argued that they do so to try, erroneously, to maximise the perceived progress towards a solution. The solution requires a ­non-​­maximising first weighing of three versus 104

Analogy and the Transfer of Creative Insights

three balls, which allows a second weighing of one versus one ball. In the experiments, we manipulated problem difficulty by comparing the n ­ ine-​­ball variant against an easier s­even-​ ­ball variant where the maximising three versus three weighing lies on the solution path. We also added an additional constraint: that it costs £1 to weigh each ball, and we allowed participants either £8 or £12 to solve. This constraint is irrelevant to the solution, but it acts to signal after the first weighing that a maximising weigh (­four vs. four) fails because there is no money left to proceed. We predicted that participants would solve the easier s­ even-​­ball variant more often than the n ­ ine-​­ball variant. However, the imposition of limited financial resources would lead to more solutions for the n ­ ine-​­ball variant while having no effect on the ­seven-​­ball variant because it would indicate that a strategy of maximising progress was ineffective for solutions in the former but not the latter. Moreover, we predicted that having the n ­ ine-​­ball variant as the source would lead to greater levels of analogical transfer than the s­ even-​­ball variant. Also, having an £8 limit would lead to more analogical transfer than a £12 limit because running out of money after a first weigh provides a tangible reason to change ­problem-​­solving strategy that is likely to endure to attempts at other tasks. Across four experiments, our predictions were confirmed. In Experiment 1, the n ­ ine-​ ­ball plus £8 source condition gave the highest levels of transfer to an e­ ight-​­ball variant (­a ­near-​­transfer demonstration, to use Barnett & Ceci’s, 2002, terminology). In Experiments 2 and 3, a ­n ine-​­ball plus £8 source condition produced spontaneous analogical transfer to superficially different but conceptually similar target problems (­the ‘­nuclear rods’ and ‘­m ad sheepdog’ problems), a demonstration of far transfer. Perhaps most surprisingly, in Exper­ ine-​­ball plus £8 source problem on solutions iment 4, we demonstrated facilitation of a n to the cheap necklace problem, a target problem that is not structurally analogous to the ­n ine-​­ball problem. The effect of suppressing a strategy to maximise progress induced by the source problem seems to have r­ e-​­directed participants’ solution attempts to the cheap necklace problem. Typically, participants attempt to solve this problem by splitting one link and joining it to another chain, a move type that seems to make immediate progress but that does not yield a solution. In our experiment, participants who had experience with the ­n ine-​­ball problem with an £8 constraint as the source problem were much less likely to make this kind of move attempt and much more likely to discover the correct move (­to break up one chain length completely and use its components to link the remaining chain pieces). Thus, it appears that one can achieve insight into the strategies that are appropriate for solving target problems, whether they are structurally analogous or not, as well as achieving insight into the structural relevance of source problem solutions. In both cases, the ‘­insight’ achieved exemplifies the definition of insight we offer in our opening paragraph: the discovery of information not available at the outset of p­ roblem-​­solving that contributes to an eventual solution. In this sense, the information being discovered is both about what the problem’s ideal representation is and a mechanism for progressing towards that representation.

Conclusions In this chapter, I have examined the relationship between insight and analogy. My contention is that insight research has a lot to benefit from routinely adopting an analogical transfer paradigm to develop and test theories. At its most fundamental level, the value of analogical transfer speaks to what we define as insight. There remains much to be explored about the relation between insight and analogy. The adherence of many studies to the variants of the radiation problem that were used by Gick and Holyoak (­1980, 1983) to spark into life the field of analogical transfer research is an indicator of the relative immaturity of the research 105

Thomas C. Ormerod

domain. We don’t yet know how a range of different problem demands will impact analogies between problems that invoke insight phenomena. So, despite the decades that have intervened since we first started being interested in the radiation problem and its analogs, we have not gotten that far down the road to understanding, across a range of problems, contents and contexts, how individuals ­re-​­use insightful knowledge to tackle new problems. Finally, it seems likely that as we begin to understand the processes that enable insight, we will also begin to understand the processes that enable analogical transfer, and vice versa. For example, Antonietti and Balconi (­2010) have suggested that neuroscientific evidence of cortical excitation during analogical reasoning indicates that analogy might itself be better characterised as a process of sudden insight rather than an incremental mapping process. Whether this view is sustained across problems, manipulations, and neuroscientific measures, remains to be seen. It is clear, however, that the futures of research into insight and analogy are inextricably linked.

References Anderson, J. R. (­2014). Rules of the Mind. New York: Psychology Press. Antonietti, A.,  & Balconi, M. (­2010). Analogical reasoning: An incremental or insightful process? What cognitive and cortical evidence suggests, Cognitive Neuroscience, 1, 1­ 37–​­138. https://­doi. org/­10.1080/­17588921003786606. Ball, L. J., & Christensen, B. T. (­2019). Advancing an understanding of design cognition and design metacognition: Progress and prospects. Design Studies, 65, ­35–​­59. Barnett, S. M., & Ceci, S. J. (­2002). When and where do we apply what we learn? A taxonomy for far transfer. Psychological Bulletin, 128, ­612–​­637. https://­doi.org/­10.1037/­­0 033-​­2909.128.4.612. Bearman, C. R., Ball, L. J., & Ormerod, T. C. (­2007). The structure and function of spontaneous analogising in ­domain-​­based problem solving. Thinking and Reasoning, 13, ­273–​­294. Bearman, C., Ormerod, T. C., Ball, L. J., & Deptula, D. (­2011). Explaining away the negative effects of evaluation on analogical transfer: The perils of premature evaluation. Quarterly Journal of Experimental Psychology, 64, ­942–​­959. Bilalić, M., Graf, M., Vaci, N., & Danek, A. H. (­2019). The temporal dynamics of insight problem ­solving – ​­restructuring might not always be sudden. Thinking & Reasoning, 27(­1), ­1–​­37. https://­doi. org/­10.1080/­13546783.2019.1705912 Bowden, E. M., & ­Jung-​­Beeman, M. (­2003). Aha! Insight experience correlates with solution activation in the right hemisphere. Psychonomic Bulletin & Review, 10(­3), 7­ 30–​­737. https://­doi.org/­10.3758/­ BF03196539 Chen, Z., & Daehler, M. W. (­2000). External and internal instantiation of abstract information facilitates transfer in insight problem solving. Contemporary Educational Psychology, 25, 4­ 23–​­4 49. Christensen, B. T.,  & Schunn, C. D. (­2005). Spontaneous access and analogical incubation effects. Creativity Research Journal, 17, ­207–​­220. Chrysikou, E. G., & Weisberg, R. W. (­2005). Following the wrong footsteps: Fixation effects of pictorial examples in a design p­ roblem-​­solving task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 31, 1134. Cushen, P. J., & Wiley, J. (­2018). Both attentional control and the ability to make remote associations aid spontaneous analogical transfer. Memory & Cognition, 46, ­1398–​­1412. Danek, A. H., Fraps, T., von Müller, A., Grothe, B., & Öllinger, M. (­2014). It’s a kind of magic: What ­self-​­reports can reveal about the phenomenology of insight problem solving. Frontiers in Psychology, 5, 1408, ­1–​­11. Detterman, D. (­1993). The case for the prosecution: Transfer as an epiphenomenon. In D. Detterman & R. Sternberg (­Eds.), Transfer on Trial: Intelligence, Cognition, and Instruction (­­pp.  1­ –​­24). Norwood, NJ: Ablex. Didierjean, A., & Nogry, S. (­2004). Reducing ­structural-​­element salience on a source problem produces later success in analogical transfer: What role does source difficulty play? Memory & Cognition, 32(­7 ), 1­ 053–​­1064.

106

Analogy and the Transfer of Creative Insights Dunbar, K. (­1995). How scientists really reason: Scientific reasoning in ­real-​­world laboratories. In R. J. Sternberg & J. E. Davidson (­Eds.), The Nature of Insight (­­pp. ­365–​­395). Cambridge, MA: MIT Press. Duncker, K. (­1945/­1951). On problem solving. Psychological Monographs, 58(­5), ­1–​­113. https://­doi. org/­10.1037/­h0093599. Fodor, J. A. (­1983). The Modularity of Mind. Cambridge, MA: MIT Press. Gavetti, G., Levinthal, D. A., & Rivkin, J. W. (­2005). Strategy making in novel and complex worlds: The power of analogy. Strategic Management Journal, 26(­8), ­691–​­712. ­ hapter 8 in D. Gentner & S. G ­ oldin-​­Meadow (­Eds.), LanGentner, D. (­2003). Why we’re so smart. C guage in Mind: Advances in the Study of Language and Thought (­­pp. ­195–​­235). Cambridge, MA: MIT Press. Gick, M. L., & Holyoak, K. J. (­1980). Analogical problem solving. Cognitive Psychology, 12, ­306–​­355. Gick, M. L., & Holyoak, K. J. (­1983). Schema induction and analogical transfer. Cognitive Psychology, 15, ­1–​­38. Gick, M. L., & McGarry, S. J. (­1992). Learning from mistakes: Inducing analogous solution failures to a source problem produces later successes in analogical transfer. Journal of Experimental Psychology: Learning, Memory, & Cognition, 18, ­623–​­639. Gust, H., Krumnack, U., Kühnberger, K. U., & Schwering, A. (­2008). Analogical reasoning: A core of cognition. Künstliche Intelligenz, 22, ­8 –​­12. Hayes, P. J. (­1981). The frame problem and related problems in artificial intelligence. In B. L. Webber & N. J.Nilsson (­Eds.), Readings in Artificial Intelligence (­­pp. ­223–​­230). Los Altos, CA: Morgan Kaufmann. Hoffman, R. R., Eskridge, T., & Shelley, C. (­2009). A naturalistic exploration of forms and functions of analogizing. Metaphor and Symbol, 24, ­125–​­154. Jones, G. (­2003). Testing two cognitive theories of insight. Journal of Experimental Psychology: Learning, Memory, and Cognition, 29(­5), ­1017–​­1027. Kershaw, T. C., & Ohlsson, S. (­2003). Multiple causes of difficulty in insight: The case of the n ­ ine-​­dot problem. Journal of Experimental Psychology: Learning, Memory, and Cognition, 30, ­3 –​­13. Knoblich, G., Ohlsson, S., Haider, H., & Rhenius, D. (­1999). Constraint relaxation and chunk decomposition in insight problem solving. Journal of Experimental Psychology: Learning Memory and Cognition, 25, ­1534–​­1555. https://­doi.org/­10.1037/­­0278-​­7393.25.6.1534 Kohler, W. (­1925). The Mentality of Apes (­E. Winter, Trans.). New York, NY: Harcourt, Brace. Kolodner, J. L. (­1997). Educational implications of analogy: A view from ­case-​­based reasoning. American Psychologist, 52, 5­ 7–​­66. Kubricht, J. R., Lu, H.,  & Holyoak, K. J. (­2017). Individual differences in spontaneous analogical transfer. Memory & Cognition, 45, 5­ 76–​­588. MacGregor, J. N., Ormerod, T. C., & Chronicle, E. P. (­2001). ­Information-​­processing and insight: A ­ ine-​­dot and related problems. Journal of Experimental Psyprocess model of performance on the n chology: Learning, Memory and Cognition, 27, ­176–​­201. https://­doi.org/­10.1037/­­0278-​­7393.27.1.176. Maier, N. R. F. (­1931). Reasoning in humans: II. The solution of a problem and its appearance in consciousness. Journal of Comparative  & Physiological Psychology, 12, 1­ 81–​­194. https://­doi.org/­10.1037/ ­h0071361. Metcalfe, J. (­1986). Premonitions of insight predict impending error. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12, 6­ 23–​­634. Metcalfe, J., & Wiebe, D. (­1987). Intuition in insight and ­non-​­insight ­problem-​­solving. Memory and Cognition, 15, ­238–​­246. https://­doi.org/­10.3758/ ­BF03197722. Minervino, R. A., Olguín, V.  & Trench, M. (­2017). Promoting interdomain analogical transfer: When creating a problem helps to solve a problem. Memory & Cognition, 45, 2­ 21–​­232. https://­doi. org/­10.3758/­­s13421-­​­­016-­​­­0 655-​­2 Monaghan, P., Sio ­U-​­N., Lau S. W., Woo, H. K., Linkenauger, S. K.,  & Ormerod, T. C. (­2015). Sleep promotes analogical transfer in problem solving. Cognition, 143, 2­ 5–​­30. http://­d x.doi.org/­ 10.1016/­j.cognition.2015.06.005. ­Munoz-​­Rubke, F., Olson, D., Will, R.,  & James, K. H. (­2018). Functional fixedness in tool use: Learning modality, limitations and individual differences. Acta Psychologica, 190, ­11–​­26. Needham, D. R., & Begg, I. M. (­1991). ­Problem-​­oriented training promotes spontaneous analogical transfer: ­Memory-​­oriented training promotes memory for training. Memory & Cognition, 19, 5­ 43–​­557.

107

Thomas C. Ormerod Neroni, M. A., & Crilly, N. (­2021). How to guard against fixation? Demonstrating individual vulnerability is more effective than warning of general risk. The Journal of Creative Behavior, 55(­2), ­4 47–​­463. Ohlsson, S. (­1984a). Restructuring revisited: I. A summary and critique of the Gestalt theory of problem solving. Scandinavian Journal of Psychology, 25, ­67–​­78. Ohlsson, S. (­1984b). Restructuring revisited: II. An information processing theory of restructuring and insight. Scandinavian Journal of Psychology, 25, ­117–​­129. Ormerod, T. C., Chronicle, E. P., & MacGregor, J. N. (­2006). Asymmetrical analogical transfer in insight problem solving. In R. Sun  & N. Miyake (­Eds.), Proceedings of the ­Twenty-​­Eighth Annual Conference of the Cognitive Science Society (­­pp. ­1899–​­1905). Alpha, NJ: Sheridan Printing. Ormerod, T. C., Fioratou, E., Chronicle, E. P., &MacGregor, J. N. (­2006). The remnants of insight. In ­ wenty-​­Eighth Annual Conference of the Cognitive Science R. Sun & N. Miyake (­Eds.), Proceedings of the T Society (­­pp. ­1893–​­1898). Alpha, NJ: Sheridan Printing. Ormerod, T. C., & Gross, H. (­2023). Expertise overcomes impasse to yield far transfer in insight p­ roblem-​ ­solving. Thinking & Reasoning. Ormerod, T. C., MacGregor, J. N., & Chronicle, E. P. (­2002). Dynamics and constraints in insight problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28, ­791–​­799. Ormerod, T. C., MacGregor, J. N., Chronicle, E. P., Dewald, A. D., & Chu, Y. (­2013). Act first, think later: Effects of maximization, minimization and lookahead on inferential planning in p­ roblem-​ ­solving. Memory & Cognition, 41, ­1096–​­1108. https://­doi.org/­10.3758/­­s13421-­​­­013-­​­­0318-​­5. Ormerod, T. C., & MacGregor, J. N. (­2017). Enabling spontaneous analogy through heuristic change. Cognitive Psychology, 99, ­1–​­16. Ormerod, T. C., & MacGregor, J. N. (­2020). The complementary roles of knowledge and strategy in insight ­problem-​­solving. In Proceedings of the 42nd Annual Meeting of the Cognitive Science Society (­­pp. ­357–​­362). Seattle, WA: Cognitive Science Society. Ormerod, T. C., MacGregor, J. N., Banks, A., & Rusconi, P. (­2023). PRODIGI: A Theory and Computational Implementation of Insight P ­ roblem-​­Solving Through Progress Maximization and Idea Discovery. Unpublished Manuscript, University of Sussex, UK. Patrick, J., & Ahmed, A. (­2014). Facilitating representation change in insight problems through training. Journal of Experimental Psychology: Learning, Memory, and Cognition, 40, ­532–​­543. Reed, S. K., Ernst, G. W., & Banerji, R. (­1974). The role of analogy in transfer between similar problem states. Cognitive Psychology, 6, ­436–​­450. Ross, W. (­2021). Feeling stumped: Investigating dimensions of impasse. PsyArXiv Preprint. https://­doi. org/­10.31234/­osf.io/­4zhse. Schönauer, M., Brodt, S., Pöhlchen, D., Breßmer, A., Danek, A. H., & Gais, S. (­2018). Sleep does not promote solving classical insight problems and magic tricks. Frontiers in Human Neuroscience, 12, 72. Singley, M. K., & Anderson, J. R. (­1985). The transfer of t­ ext-​­editing skill. International Journal of M ­ an-​ ­Machine Studies, 22, ­403–​­423. Sio, U. N., & Ormerod, T. C. (­2015). Incubation and cueing effects in ­problem-​­solving: Set aside the difficult problems but focus on the easy ones. Thinking & Reasoning, 21, ­113–​­129. Smith, S. M., & Blankenship, S. E. (­1991). Incubation and the persistence of fixation in problem solving. American Journal of Psychology, 104, ­61–​­87. ­Vallée-​­Tourangeau, F., Steffensen, S. V., ­Vallée-​­Tourangeau, G.,  & Sirota, M. (­2016). Insight with hands and things. Acta Psychologica, 170, ­195–​­205. Webb, M. E., Little, D. R., & Cropper, Simon J. (­2018). Once more with feeling: Normative data for the aha experience in insight and noninsight problems. Behavior Research Methods, 50(­5), ­2035–​ ­2056. https://­doi.org/­10.3758/­­s13428- ­​­­017- ­​­­0972-​­9. Weisberg, R. W., & Alba, J. W. (­1981). An examination of the alleged role of “­fi xation” in the solution of several “­insight” problems. Journal of Experimental Psychology: General, 110, ­169–​­192. Welling, H. (­2007). Four mental operations in creative cognition: The importance of abstraction. Creativity Research Journal, 19, ­163–​­177. Wertheimer, M. (­1938). Gestalt Psychology. Source Book of Gestalt Psychology. New York: Harcourt, Brace and Co.

108

8 CREATIVE COGNITION From Ideation to Innovation Nathaniel Barr, Lucas Klein, Michael J. McNamara and Kelly Peters

Creative Cognition: From Ideation to Innovation The centrality of creativity to the story of our species is apparent across our achievements in the arts, sciences, humanities, and technology, and more fundamentally, our collective survival as primates aloft a rock hurtling through space. Included within the echelons of our accomplishments is our unique capacity to comprehend the very mind that has afforded these advances and the psychological genesis of our ideas. Understanding of creativity as a concept and the nature and workings of the mental machinery underlying creativity has progressed significantly, much like our collective creative output. Formerly beyond the scrutiny of science, then largely subsumed within the study of intelligence, the last several decades have seen creative cognition become a robust and expanding area of psychological inquiry. Human understanding of the processes by which novel and useful ideas emerge in the mind, be it the genesis of creative combinations, the nature of divergent thinking, or how insights spontaneously emerge, has benefitted from the rigorous and dedicated study of cognitive psychologists. Though these advances hold great value, some argue that the genesis of creative ideas is not the sole, or perhaps even most critical aspect of understanding the sorts of accomplishments that our species celebrates as the pinnacles of our success (­a nd are often referenced in the opening sections of papers about creativity). Rather, it has been contended that the most consequential cognitive activity involved in the transformative impact humans have had on the planet is our capacity to translate creative ideas into tangible innovation. Theodore Levitt, in a 1963 Harvard Business Review article entitled “­Creativity is Not Enough”, argues that “­ideation is relatively abundant… implementation is more scarce” and that many advocates for creativity have “­failed to distinguish between the relatively easy process of being creative in the abstract and the infinitely more difficult process of being innovationist in the concrete”. He goes on to say that “­what is often lacking is not creativity in the ­idea-​ ­creating sense but innovation in the ­action-​­producing sense, i.e., putting ideas to work” (­­pp. ­137–​­138). Interestingly, if creatively applied toward the study of creative cognition, much of Levitt’s critique remains valid. Studying idea generation in the laboratory is much easier than the messy endeavor of understanding the cognitive correlates of innovation in the real world, DOI: 10.4324/9781003009351-9

109

Nathaniel Barr et al.

where ideas are shaped into realities across space, time, and many minds. Considerable research into the “­structural, collective, and social conditions for innovation” (­see Kanter, 1988) exists, as does research into individual, group, and organizational factors influencing innovation (­e.g., Amabile, 1996), but arguably only relatively small threads are truly cognitive in orientation. Further, while many have commented on the mindsets of innovators (­D yer et al., 2011), few have approached the topic with the rigor and theoretical grounding of modern cognitive psychology in the way that creativity researchers have dissected the creative mind in the context of ideation. Despite these difficulties, it is imperative that cognitive psychologists wade into these topics. For cognitive psychologists, creativity is not enough; the cognition of innovation, too, must be subject to rigorous empirical and theoretical scrutiny. In what follows, we disentangle creativity and innovation, review the state of the science in understanding creative cognition at the level of ideation, and then turn to innovative cognition. Though less developed, we argue that a foundational base can be found, which has the potential for considerable growth. Rather than systematically reviewing what work has been done, our priority is to call upon those who have largely confined their interest in creative cognition to the realm of ideation to take aim at innovative cognition as well. We consider the barriers to cognitive research on innovation, identify opportunities for future research, and emphasize the importance of these research directions being pursued by cognitive psychologists.

Distinguishing Creativity and Innovation One particularly pernicious problem is that the terms “­creativity” and “­innovation” are often used interchangeably (­Patterson, 2002). In order that cognitive psychologists may properly orient themselves to this distinction, it is important to consider the connections and delineations between the terms. Defining “­creativity” has proven to be one of the most difficult tasks facing our field (­Sawyer, 2012). As Ausubel (­1964, ­p. 551) argued, creativity stands as “­one of the vaguest, most ambiguous, and most confused terms in psychology”. Psychologists have taken great strides toward conceptual definitional agreement and, more crucially, have found ways to operationalize that understanding, yielding considerable empirical and theoretical advance. While far from universally accepted, two basic requirements for a thought or action to qualify as “­creative” seem to have emerged in the conceptual debate, namely that to qualify as “­creative”, a thought or action: (­i) must be novel or original; and (­i i) must have some adaptive value (­Plucker et al., 2004). The measurement of a creative idea has been effectively captured in an array of tasks within the toolbox of cognitive psychologists, making it a relatively clear target for both correlational and experimental studies. “­Innovation”, on the other hand, is largely associated with the “­implementation” of creative ideas, as in the implementation of a new or significantly improved product (­good or service) or process, a new marketing method, or a new organization method in business practices, workplace organization, or external relations (­OECD, 2009). Often taken together, creativity and innovation form the process, outcomes, and products of attempts to develop and introduce new and improved ways of doing things; where the creativity stage of this process refers to idea generation, innovation refers to the subsequent stage of implementing ideas toward better procedures, practices, or products (­A nderson et al., 2014). Much as creativity researchers have articulated varying degrees of creativity (­e.g., the Four C Model; Kaufman & Beghetto, 2009), innovations are often segmented by the form and/­or type of 110

Creative Cognition: From Ideation to Innovation

creative advance that has been implemented. For example, sustaining innovations are the incremental improvements to existing products or ideas that have been implemented by the subject. In contrast, a disruptive innovation involves a high degree of imagination, a higher risk tolerance, as well as a more precarious process of implementation in which new entrants explicitly challenge incumbent firms or markets by either targeting overlooked segments of the market or by creating new markets where no market exists (­Christensen, 1997). Radical innovations, on the other hand, involve the creation of completely new forms of knowledge, followed by the implementation of this new knowledge and, possibly, its commercialization. So, while the majority of scholarly focus within cognitive psychology has been on the generation of ideas, the real mystery that needs to be solved might reside in the minds of those who are capable of that critical second act of creation, managing the implementation of ideas to produce innovation. Now that we have clarified the distinction between creativity and innovation, let us next consider what is known about both ideational and innovative cognition.

Ideational Cognition In 1950, J.P. Guilford issued an address intended both to expose a lack of research into the psychology of creativity and to motivate new interest in it. He noted that, until that point, the predominant presumption was that creativity and intelligence were largely synonymous, meaning that little attention had been paid to the distinctions. Since then, an explosion of methodologies and theories has contributed to increasingly comprehensive and nuanced definitions, an e­ ver-​­growing toolkit of assessment techniques, and a more refined understanding of the correlates of creative thinking and people, including the personality traits associated with higher creative aptitude (­see Barron, 1965, for a review of the early research on creativity). But a few decades after Guilford’s (­1950) address, the seeds of the cognitive revolution took sprout, meaning that a central approach among the increasing range of pursuits to understand ideation involved applying the lens of cognition. Today, the literature continues to grow and reach greater sophistication, yielding considerable evidence pertaining to the cognitive mechanisms that engender creativity and a more replete understanding of the neural processes that underlie them (­see Barr, 2018). Over the past decades, two primary explanations emerged for how creative ideas are generated (­Beaty et al., 2014). According to the ­controlled-​­attention theory of creativity, controlled/­executive processing selectively retrieves novel ideas and concepts from semantic memory, and the ­top-​­down control of attention helps uncover them by suppressing mundane or inappropriate ideas that tend to surface first (­Gilhooly et al., 2007). The associative theory of creativity (­e.g., Mednick, 1962) contends that differences in the structure of associative hierarchies in semantic memory facilitate or hinder the ability to combine unrelated ideas in novel configurations. Steep associative hierarchies are structured such that strongly linked ideas are readily available in semantic memory but become exhausted quickly and do not stray far from a cue. Conversely, semantic knowledge within flat associative hierarchies is structured flexibly and broadly, such that associations between disparate ideas form easily; individuals with flat associative hierarchies can access weakly linked ideas consistently over the course of a task. Creative ideation is said to be benefited by the broad structure of flat associative hierarchies because novel combinations of ideas are more likely to occur more often. Much of the evidence accrued in this arena relates to empirical studies of divergent thinking tasks. Arguably chief among these is the Alternative Uses Task (­AUT), which challenges participants to come up with as many alternative uses for ordinary objects as they 111

Nathaniel Barr et al.

can in a short period of time (­usually a few minutes). Given the cue newspaper, the response “­swat a fly” might come to mind more readily and more quickly than the response “­m ake a costume”. Responses can be assessed for quantity and quality by independent raters and correlated with scores on related tasks, personality traits, or neuroimaging data. The connection between associative abilities and divergent thinking was traditionally contrasted with the connection between executive a­ bilities—​­such as fluid intelligence (­Gf ) and retrieval (­Gr)—​­and intelligence. However, given emergent evidence, it seems that both associational processes and executive control play important roles in creative ideation. Beaty et al. (­2014) explored the differential influences of both processes. They gathered responses on a verbal fluency task (­e.g., Clark  & Mirels, 1970) and measures of executive function (­broad retrieval ability (­Gr) and Gf ) of more and less creative people (­a s measured using the AUT) and found that both metrics predicted divergent thinking abilities using structural equation modeling. Broad association is necessary to access a large search space, and ­top-​ d­ own attention inhibits inappropriate or obvious solutions in favor of more creative ones. Some have suggested that a d­ ual-​­process framework that emphasizes the interplay between traditional dichotomies inherent to cognitive theories (­e.g., Type 1 and Type 2 thinking, associative vs. executive, autonomous vs. controlled) can reconcile conflicting evidence, characterize the temporal evolution of the creative process, and integrate disparate domains of creativity research (­e.g., Barr, 2018). The view that cooperation between associative and controlled cognitive systems is necessary is supported by recent neuroscientific evidence that illustrates how creative thinking depends on connectivity between prefrontal areas linked to executive processing and cognitive control and the default mode network, which supports spontaneous associative processing (­see Beaty et al., 2016). And while divergent thinking tasks have featured prominently in the development of a cognitive understanding of ideation, there remain many other tasks in the toolkit of cognitive psychologists. One common task for assessing convergent thinking abilities is the Remote Associates Test (­R AT), the object of which is to think of a common word or phrase that relates to a collection of verbal cues (­Mednick, 1962; MacGregor  & Cunningham, 2008). For instance, the words moon, rat, and wheel all have cheese in common. Creative ­problem-​­solving tasks (­a s opposed to analytic problem solving) like the RAT are difficult or impossible to solve by simply applying a p­ re-​­defined algorithm or heuristic. In some cases, the answer arises suddenly and unexpectedly after a period of quandary, the ­so-​­called “­A ha!” moment that characterizes the experience of insight. Instead of a result of b­ rute-​­force, systematic searching through memory, insight happens when fundamental assumptions or constraints are relaxed, when the original representation of the problem is shifted in a way that allows novel connections to form, or when the problem is decomposed into perceptual chucks (­K noblich et al., 1999). While investigating insight scientifically poses certain challenges, many methods have been successful in uncovering components of it, and the field remains fertile for new approaches (­Bowden et al., 2005). First, determining whether insight has occurred is in large part a matter of interrogating the phenomenology of the p­ roblem-​­solver. Although p­ roblem-​ ­solvers typically report a consistent set of components of the e­ xperience—​­solutions are sudden and obvious, they arise after a period of impasse, and the thinker is unaware of how they arise (­the processing itself is “­nonreportable”)—​­there is no unambiguous set of criteria that defines an instance of insight. Classic “­insight problems” are designed such that they are likely to produce an impasse and then yield to sustained effort. For this reason, rebus ­puzzles—​­combinations of verbal or visual cues that lead to a common word or ­phrase—​­are often used as insight tasks because they require breaking the assumptions implicit in normal 112

Creative Cognition: From Ideation to Innovation

reading and typically take some time to solve (­MacGregor & Cunningham, 2008). But an insight problem having been solved is not evidence that insight has occurred. Nor does insight only happen for one class of problem. For this reason, operational definitions that link the insight process to neural networks in the brain would bring together various definitions and allow competing theories to be tested. Indeed, recent work taking advantage of contemporary neuroimaging [e.g., electroencephalography (­EEG) and functional Magnetic Resonance Imaging (fMRI)] has begun to elucidate the neural correlates and hemispheric asymmetry of insight (­Bowden & ­Jung-​­Beeman, 2003; Kounios & Beeman, 2009; Zhao et al., 2014). Other important work oriented toward creative ideas has focused on semantic distance, with ideas that bridge greater semantic distance seen as more creative (­Kenett, 2018). Such a quantification has featured prominently in research on analogical reasoning, much of which finds a key role for executive control processes in the generation and identification of creative, semantically distant analogies (­e.g., Barr et al., 2014; Green, 2016). Recent work has also leveraged semantic distance in the context of several new tasks. The “­Forward Flow” task developed by Gray et al. (­2019) presents participants with a single word, and they are asked to generate the next word that comes to mind given the previous one. This simple task predicts creativity in other common tasks, membership in r­eal-​­world creative groups (­performance majors, professional actors, and entrepreneurs), and can even extract estimates of celebrity creative achievement based on social media posts. Another new measure, the Divergent Association Task (­Olson et  al., 2021), differs slightly from the “­forward flow” approach. Participants are asked to name 10 words that are as different from each other as ­possible—​­the task correlates well with other creativity measures and marks another methodological approach toward the study of ideational creativity (­see also Beaty & Johnson, 2021, for another automated method of assessing creativity through semantic distance metrics). While much of cognitive psychologists’ attention has been toward the development and use of laboratory tasks to study individual ideational cognition, there have also been forays into more dynamic and complex domains like improvisation (­e.g., Beaty, 2015). During creative improvisation, unplanned, unpredictable creative productions (­musical notes, brush strokes, words in a poem) are continuously generated and evaluated in r­ eal-​­time. Such creative improvisation is similar to innovative cognition, in that the temporal window may be longer than in the more discrete ideational cognitive process but differs in that it is less oriented toward a cumulative output meant to endure and make an impact in the world. Many forms of creativity beget improvisation, from live painting to freestyle rap, but important differences make some more conducive to scientific inquiry than others. Inherent constraints on the creative domain of study, such as the abnormal conditions of laboratory experiments, limit ecological validity compared to creative processes that are allowed to occur in naturalistic environments. A prototypical example of natural but easily observable creativity is musical improvisation (­McPherson & Limb, 2013). Jazz performance, for instance, requires a high degree of acquired skill and attentional focus in pursuit of a goal that is reflective of the creative process by nature (­e.g., Lopata et  al., 2017). The music itself is conceived spontaneously and generated immediately, which makes it far easier to study than a classical composition, which may take place over the course of months or years rather than seconds. Like the aggregate findings from divergent thinking tasks and other creativity measures, the interplay between brain networks associated with associative and executive processing plays a central role in improvisational creativity (­see Beaty, 2015). The work discussed thus far, and much of the cognitive psychology research on idea generation, focuses on individual, mentalist idea generation, but it is important to consider the research on situated, distributed, and group idea generation. A considerable amount of 113

Nathaniel Barr et al.

work in the tradition of Creative Problem Solving (­CPS) is oriented toward understanding the frameworks, methods, and tools that best facilitate ideation in the context of small groups (­Treffinger & Isaksen, 2005). Not only has such work focused on the ideational strategies groups might be best served to deploy, it has illuminated the importance of individual cognitive styles in accounting for group performance (­Treffinger et al., 2008). In all, the state of understanding toward ideational cognition far surpasses what can be reasonably captured in such a small space. Across diverse tasks and measurements, cognitive psychologists have taken great strides toward apprehending a once ephemeral psychological concept, all within the span of a few decades. The rigorous experimental control and theoretical grounding in the cognitive systems that support human experience have served the field well in the pursuit of knowing how creative ideas are generated, for whom they are more likely to occur, and the conditions that support their emergence. Significant effort has also been expended on understanding the cognition underlying team creativity (­­Reiter-​ ­Palmon et al., 2012) and group creativity in the context of collaboration (­Paulus & Nijstad, 2003), in organizations (­Hargadon, 1999), and in online interactions (­Sarmiento & Stahl, 2008). Recognizing the way that many minds can come together in service of creativity, research has also built upon research on distributed cognition to explore distributed creativity, in which contributions across a group determine the creative output (­Literat & Glăveanu, 2016; Miettinen, 2006; Sawyer & DeZutter, 2009; Glăveanu, 2014). Such research on ideational cognition, which incorporates the complexity of groups in accounting for creativity, begins to capture the elements found in the noise of ­real-​­world settings in which true innovation occurs.

Innovative Cognition As the preceding section demonstrates, cognitive psychology has made great advances toward the understanding of “­idea generation” as a thinking process related to creativity and creative thought. Yet, as Levitt (­1963) reminds us, the need in applied settings is not typically creativity in the ­idea-​­creating sense but innovation in the ­action-​­producing sense. Levitt has put this distinction even more directly, describing creativity as coming up with new ideas and innovation as doing new things. It is readily apparent that doing new things requires new ideas about what should be done, but it also requires more novel, and sometimes familiar, ideas about how to realize that idea in tangible terms over time. Thus, while assuredly an essential and frequent component, generating ideas is but a piece of the puzzle toward innovation in the a­ction-​­producing sense. Unfortunately, if we are to agree with this sentiment, a disproportionate number of “­creativity studies” place their emphasis on understanding the generation of ideas (­creativity) as opposed to the implementation of novel ideas (­innovation; Axtell et al., 2000). As a result, our current understanding of the cognitive tasks, skillsets, and abilities associated with the implementation side of the creativity equation remains comparatively underdeveloped. That said, there does exist important research aimed at understanding innovative cognition that can serve as a valuable foundation for cognitive psychologists who might seek to help rectify this disparity. Arguably, much of the scholarship from outside cognitive psychology does consider cognition. Joseph A. Schumpeter, who outlined the idea of creative destruction, discussed the mind of the entrepreneur, a class of individuals whose characteristics align well with what might be called an innovator today. He described the role of “­­super-​­normal qualities of intellect and will” (­Schumpeter, 1934, p­ . 82), though he considered these intellectual qualities to be of lesser importance than the desire to do things differently: “­carrying out a new plan 114

Creative Cognition: From Ideation to Innovation

and acting according to a customary one are things as different as making a road and walking along it” (­Schumpeter, 1934, ­p.  85). Such a view aligns with Kirton’s (­1994) ­adaptation-​ ­innovation inventory, which is used to gauge whether a person’s creative style tends toward the adaptive or innovative end. Kirton sees adaptors as operating within a framework of systems and as being associated with sufficiency of originality, efficiency, and r­ule-​­group conformity, whereas innovators break away from the existing framework of systems and are associated with high interest levels in terms of originality of ideas, less concern for efficiency and ­r ule-​­group conformity. In this model, the success of innovation depends on the cultural and social context of the environment, specifically, how rigidly governed by rules an individual, society or institution tends to be. Other contemporary work supports the idea that a disposition toward novelty and an appetite for risk are important facets of innovative potential. At the individual level, research has shown how our personal need to reduce uncertainty in our lives fuels a rejection of creative ideas in favor of the practical, even when we openly espouse creativity as our desired goal (­Mueller et al., 2011). At an organizational level, research again reveals the extent of the bias against creativity. Here, the bias exists at almost every level of the organization (­K acerauskas, 2016), including its natural preference and, at times, necessity for command and control administrative hierarchies, its preference for lowering costs and efficiencies at the expense of investing in unproven ideas, and its natural tendency for hiring individuals who fit in with the homogenous culture of the firm, as well as a general tendency for stability, predictability, and order (­see Straw, 1995; Kacerauskas, 2016). Research on the factors that can kill creativity in organizations highlights the importance of challenge, freedom, resources, ­work-​­group features, supervisor encouragement, and organization support (­A mabile, 1998). In a similar vein, Ekvall’s (­1996) work on organizational climate for innovation has been instrumental in shaping our understanding of how workplace features can inhibit and/­or support the creative performances of employees. Like Amabile, Ekvall draws our attention to the importance of things like trust and openness, idea support, idea time, risk tolerance, and the key aspects of challenge and involvement in workplace tasks. Other research has shown how one’s decision to undertake the difficult choice to pursue a novel idea through to its full realization may be contingent on one’s personality or behavioral traits. As an example, overcoming the “­barrier” to creativity may depend upon creative actors maintaining an independent mindset or a salient feeling of being different from others (­K im et al., 2013). Others have highlighted the importance of motivation and openness to experience as being positively related to creativity and innovation (­Patterson & Zibarras, 2017). Patterson’s (­1999) “­Innovation Potential Indicator” was designed to identify the potential to implement innovation ideas in the workplace. This instrument uses four separate characteristics to look at behaviors that enhance or impede the development and implementation of new ideas. These characteristics include an individual’s intrinsic motivation to seek and adopt change; the extent to which an individual is comfortable challenging other people’s ways of thinking and points of view; an individual’s ­problem-​­solving style; whether they are more likely to try to evolve existing procedures or to aim for originality; whether an individual prefers an organized and structured work environment or one with variety and flexibility; and their attitude toward rules and policies. Although we have emphasized the dearth of cognitive orientation to innovation, it is also imperative to recognize that others have preceded us in articulating the importance of cognition and undertaken important research in this vein. Sund et al. (­2018) review the connections between cognition and innovation across multiple areas of inquiry (­cognition in organizations, innovation in organizations, intrapreneurship, and entrepreneurship) and 115

Nathaniel Barr et al.

invite scholars to capitalize on the “­enormous potential” for further research. Mumford et al. (­2009) identify cognition as the key to understanding creativity and innovation. These authors respond to a call by Bledow et al. (­2009) to resolve the issues surrounding the lack of coherence around creativity and innovation through the adoption of a dialectic orientation by arguing that “­attempts to understand creativity and innovation in organizations should be based on another fundamental: cognition” (­Mumford et al., 2009, ­p. 353). Reviewing evidence for this supposition, they cite work showing the success of research that approached creativity and innovation through a cognitive lens, showing for instance, that an organizational creativity training course oriented around cognitive content outperformed programs built around social, motivational, or personal work styles (­Scott et al., 2004). They further argue that cognitive approaches (­such as those described above) provide a more sophisticated means to understand ­domain-​­specific expertise and creative thinking abilities that serve as the basis for the innovation process than do other approaches, including the dialectic. Mumford and colleagues stand both as advocates and leading contributors to research on innovative cognition, and a number of other noteworthy works should interest cognitive psychologists (­e.g., Mumford  & Gustafson, 1988; Mumford, 2000; Mumford  & Hunter, 2005; Mumford et al., 2008, to name a few). Another promising area of research is the one that is focused on managerial cognition, which emphasizes the importance of studying the cognitive activity and profiles of leaders within organizations. Stubbart (­1989) notes that Schendel and Hofer’s (­1979) influential work on strategic management as a new means to view business policy and planning assumes a cognitive focus but did little to make explicit the role of cognitive science. Stubbart (­1989) took on the task of outlining the interrelations between the literature on strategic thinking and cognitive science and called for more research on managerial cognition. This paper has received a considerable number of citations, with a number of these works specifically considering innovation through the lens of managerial cognition. For instance, Barr et al. (­1992) consider the way that leaders within organizations must make changes to their mental models in response to environmental changes, underscoring the importance of cognitive maps within managers in understanding innovation and organizational renewal. Bergman et  al. (­2015) suggest that the shared cognitive maps held by management within an organization form the dominant logic, which in turn influences innovation strategy and potential. ­M iron-​ ­Spektor and Argote (­2008) place importance on paradoxical cognition within corporate leaders to balance both the production of innovations and their implementation within existing systems. Manral (­2011) argues that models of innovation that solely consider structural factors should more seriously consider the cognitive characteristics of managers, though the focus is primarily on “­cognitive attitudes” rather than aptitudes. Vecchiato (­2017) brings together research on disruptive innovation and managerial cognition, identifying beliefs about customer needs as critical for firm success in identifying and addressing new markets. Yang et al. (­2019) consider managerial cognition in the context of innovation capability, finding that managers’ perceptions of institutional pressures influence their focus on environmental strategy. Despite the value of this work, it is apparent that much of the research focuses on beliefs, perceptions, and attitudes rather than cognitive ability and dispositions, as is often under scrutiny in work on ideational cognition. Further, an inspection of these citations indicates that the work on managerial cognition primarily consists of business and management scholars leveraging ideas from cognitive psychology within business and management journals rather than cognitive psychologists fulsomely contributing to the development of a science of innovative cognition. In assessing the evidence that is currently available, significant strides have been made, yet much more work could be done toward operationalizing and understanding innovation 116

Creative Cognition: From Ideation to Innovation

in a more systematic way at the cognitive level. Scholars of diverse disciplines have taken up the task of articulating models of innovation and profiling innovators. Of the psychological work that is available, though, there are important exceptions; much more research aimed at the innovator is oriented toward ­social-​­personality variables than cognitive ones (­e.g., Loy, 1969; Zhao & Seibert, 2006; Aronson et al., 2008; Patterson & Zibarras, 2017). In ascertaining why more cognitive research has focused on ideation than innovation, it is important to consider some of the barriers to the former.

Barriers to Cognitive Approaches to Understanding Innovation Beyond the simple fact that much remains to be known about ideational cognition, to wit, there are several reasons why a focus on ideation may have prevailed over innovation in cognitive circles.

Timescales and Scope Compared to creative ideation, in which the emergence of a creative idea can be localized to a relatively tight temporal interval, the process of innovation is much more expansive. The process of taking a creative idea and realizing it in sufficiently ­far-​­reaching and tangible terms to attain the threshold definition of innovation can take many divergent paths and is often enacted over the course of months or years, sometimes even decades. This leads innovation to be more diffuse in identification than the onset of a creative insight or the production of divergent ideas, for instance, and thus less amenable to the sorts of tasks and tests most familiar to cognitive psychologists. Such a time course and the litany of cognitive processes involved over that duration necessarily complicate the challenge associated with apprehending the dynamics of innovative cognition.

Complexity and Control Many have called the human brain the most complicated and complex entity in the known universe. Consider that innovation is a feat that is rarely achieved alone and that many of the world’s most innovative organizations have employees numbering in the tens of thousands, each equipped with an unfathomably complex brain and associated cognitive system. Thus, innovation can be comprised of the interactions of many complex cognitive systems interacting under the constraints of the myriad external factors exerting influence over the creative and innovative process, including organizational structures and socioeconomic and environmental influences. As such, a m ­ ulti-​­level perspective that accounts for these key variables that span distinct arenas is required (­see Mumford & Hunter, 2005). Compared to apprehending individual creative ideation, this marks a formidable task and one that has markedly more influential factors that require accounting. For instance, R ­ eiter-​­Palmon et al. (­2012) argue that social processes can interact with cognitive processes, suggesting a specific sort of complexity in organizational environments that is often avoided in the context of research on ideation. Related to the challenges posed by complexity in terms of many moving parts and variables to consider is the unwieldy nature of environments in which innovation is most likely to occur. While in a laboratory setting, one can systematically and rigorously control variables of interest in the context of an experiment, organizational settings and other environments are far less amenable to strict controls. Further, engaging with applied partners in fields that can provide access to innovators introduces significant complexity in the logistical 117

Nathaniel Barr et al.

and bureaucratic process both generally and more specifically in the requisite b­ uy-​­in to manipulate and test factors of interest. Complicating matters even more, the observer effect, most often invoked in physics, can come into play, wherein the very process of even observing, let alone manipulating, the innovation process in the field can shape the developmental trajectory of the innovation. Working alongside other academics studying creative cognition in the lab, cognitive psychologists have ­near-​­full ­decision-​­making power and the capacity to limit many of the complex influencing factors through careful experimental d­ esign—​­this is far more difficult in the wild when organizational stakeholders have their own priorities.

Samples and Partnerships As McNemar (­1946, ­p. 333) said, “­the existing science of human behavior is largely the science of the behavior of sophomores”. While certainly an exaggeration given the advances in cognitive psychology over the decades, it remains true that psychology researchers tend to rely on college undergraduates. Undergraduates, often receiving course credit for participation, have been said to serve as “­psychology’s fruit flies” (­Rubenstein, 1982) for their abundance, accessibility, convenience, and low cost to researchers. Cognitive psychological research has been particularly heavily reliant on such students for s­ amples—​­in 1996, Kimmel stated that approximately 90% of perception and cognition studies featured student participants. The emergence and rise of online testing have surely reduced that proportion, but for many types of cognitive research, undergraduate samples remain predominant. Those seeking to study innovation need to seek more specialized samples, meaning there is an incongruence between standard participant recruitment and sample selection in cognitive psychology. Indeed, any sincere effort to understand innovative cognition will necessarily transcend traditional ­lab-​­based research. Further, the potential ­domain-​­specificity of innovation across sectors and types of expertise means that no one partnership is adequate to more broadly understand innovative cognition (­for examples of creative cognition across domains, see ­Reiter-​­Palmon & Japp, 2024, this volume; Cropley, 2024, this volume; Christensen, 2024, this volume; Shen, Ball  & Richardson, 2024, this volume; and Vartanian, 2024, this volume). Given the practical and fiscal constraints of forays beyond the campus walls, as well as the fruitful returns of laboratory work, cognitive psychologists have engaged less seriously with applied research partners in the field, a point we will turn to below when we consider future directions.

Future Directions in Innovative Cognition Research In order that cognitive psychologists may more fully apprehend creative cognition, particularly with regard to innovation, it is important to consider different methods that may help to overcome some of the barriers identified that have precluded deeper investigation to date.

Situated Experiments and In Vivo Investigation Dunbar and Blanchette (­2001) made significant strides toward understanding more fully the central role that analogies can play in creativity. They found that more distant analogies based on structural similarities were important for generating new hypotheses and ideas, whereas more superficial analogies were used in solving practical problems. While the results were illuminating, the approach is of interest here. Dunbar and Blanchette (­2001) advocated

118

Creative Cognition: From Ideation to Innovation

for the combination of in vivo (­naturalistic) and in vitro (­experimental) methods into a unified investigation. Drawing on arguments from pioneers of ecological psychology and situated cognition that suggest understanding human cognition necessitates consideration of not just mental but environmental factors (­e.g., Neisser, 1976; Gibson, 2014), these authors observed analogy use in laboratory meetings of scientists in several distinct research groups, as well as investigated the way politicians and journalists used analogies in newspapers and in political meetings and rallies. This in vivo observational research was complemented by in vitro experimental ­lab-​­based research, which is most familiar to cognitive psychologists, and found convergent evidence. Such an approach illustrates a means by which cognitive psychologists can complement traditional experimentation with more situated, naturalistic opportunities for data collection. Such an approach has not evaded the attention of those interested in creativity, innovation, and design (­e.g., Wiltschnig et al., 2011), and some have long advocated for the role of cognitive ethnography as an important research approach (­Ball & Ormerod, 2000). Further, scholars outside of the cognitive realm have made advances in “­transplanting the lab to the field” through situated experiments, an approach that arguably “­optimizes the strengths of both laboratory and field experiments in organizational research while mitigating the weaknesses of each’ (­Greenberg & Tomlinson, 2004, p­ . 703). As will be considered more fully in a subsequent section, embracing such distinct methodological tools in conjunction with collaboration across disciplines seems like a potent path to advancing understanding of innovative cognition.

Smartphones and Diary Studies In a prescient paper, Miller (­2012) sketched “­The Smartphone Psychology Manifesto”, arguing that, despite not their intended purpose, the forecasted capabilities and ubiquity of smartphones would mark a critical tool in psychological research toward conducting interactive and ecologically valid data in ­real-​­time from participants around the world. The viability of such data collection from individuals with high innovative potential appears promising toward overcoming some of the challenges that come with the broad temporal window of innovation, as described earlier. In particular, the Experience Sampling Method (­van Berkel et al., 2017) allows researchers the chance to periodically probe the experiences, thoughts, and behaviors of research participants as they engage with the world in a longitudinal manner. As an example of a methodological approach that could be fruitful in the realm of innovative cognition, consider Gable et al.’s (­2019) study that investigated the idea generation processes of writers and physicists. The investigators had professionals report the most creative ideas they had each day, tracking how h ­ igh-​­quality the idea was, whether it felt like an “­A ha!” moment, and what they were thinking about and doing when it emerged. Interestingly, a significant amount of the most creative ideas occurred spontaneously while participants were not engaging in work and thinking about something else, and these sorts of ideas were more likely to be experienced as an “­A ha!” moment and were associated with overcoming some form of block to a previously insurmountable impasse. This work sheds light on the cognitive dynamics of ideational cognition throughout the day but could easily be replicated and extended within a sample rooted in organizations and oriented toward queries that more explicitly tap the implementation aspects of innovation. Moving forward, cognitive researchers studying human cognition generally, and innovative cognition more specifically, must strive to take advantage more fully of the technological tools at their disposal.

119

Nathaniel Barr et al.

­Cross-​­Disciplinary ­Partnership-​­Based Research As much as our arguments here have oriented toward a perceived shortage of cognitive research focused on innovation, it is important to again make clear that there is no shortage of important and rigorous work from other disciplines focused on the topic (­a nd to recognize the scholars who have attacked innovation from a cognitive lens). An important task for cognitive psychologists is to forge connections and relationships with scholars outside of their usual collaborators, bringing with them their strong orientation to basic processing mechanisms and their robust psychometric toolkit for gaining insight into meaningful individual differences while being open to other approaches, such as those employed by organizational, management, business, and sociological scholars that can account for organizational, social, and environmental forces. Hennessey and Amabile (­2010) have advocated for a systems view to understanding creativity, in which distinct levels of analysis, from the neural and cognitive to the social and cultural, are taken into account. Barr (­2018) contended that those operating at the higher levels should move toward cognitive terminology and frameworks to further facilitate this sort of bridging across subfields. Going further, rather than expect scholars from outside the cognitive domain to adopt cognitive psychological perspectives, it would be even more beneficial if cognitive researchers took on the challenge of collaborating more fully with scholars focused on innovation from other perspectives, proactively bringing our expertise to the table. Not only would such collaborations yield theoretical returns through research, they could also yield considerable practical benefits in facilitating the research process. Cognitive psychology as a discipline has largely been insular in its research processes, with much of the work in the past decades amenable to being accomplished without external partners. Most collaborations are relatively clustered within common areas of expertise, and gulfs even exist between ostensibly similar subfields within cognition (­Barr et al., 2020). There are, of course, numerous notable and important counterexamples. However, the field could benefit from more engagement with external research partners, such as organizations, companies, and other teams oriented toward innovation. Organizational, business, and other sorts of scholars have a stronger tradition of engagement with such partners in their research. The future of creative cognition research should feature more in situ field experiments predicated on c­ ross-​­disciplinary collaboration with the support of external research partners than we have seen in the past.

Enhancing Innovative Cognition Research on ideational cognition, while often oriented toward understanding how creative ideas emerge, has also focused on how to enhance the creative process. Some of the more prominent efforts and an area of focus that straddles both ideational and innovative cognition relate to the CPS methodologies discussed above, which advocate for deliberate phased strategies that are based on how people organically solve problems but augmented with tools and training meant to foster original thinking (­see Treffinger & Isaksen, 2005, for a history). Other work has varied from intensive training programs to simple ­one-​­time interventions. For instance, Kienitz et al. (­2014) executed a randomized control trial in which participants completed either a ­five-​­week creativity ­capacity-​­building training program or a language ­capacity-​­building training program. Simpler interventions have focused on simple instructional manipulations that encourage creativity, assuming that creativity is a dynamic state rather than a static trait and that simply telling participants to be more creative can facilitate 120

Creative Cognition: From Ideation to Innovation

novelty. For instance, Green et al. (­2012) showed that an explicit cue to “­think creatively” improved performance on a creative analogy task. Expanding on this work, Green et  al. (­2017) coupled this cognitive intervention with a neural intervention via transcranial direct current stimulation (­tDCS) and found that tDCS enhanced the impact of the cognitive cue alone, suggesting the potential for ­non-​­invasive neural interventions as a means to promote creativity. Interventions have also been used to enhance i­ nsight—​­Subramaniam et al. (­2009), for example, showed that the induction of a positive mood facilitated insight, with neural evidence suggesting the benefits arrive through the modulation of attentional and cognitive control mechanisms. The above research represents a fraction of the work available on enhancing ideational ­cognition—​­but what about innovative cognition? There is no shortage of scholarship and practice oriented to enhancing innovation in organizations. One thread comes through the work on CPS discussed above, with the logic being that enhancing the ideational capacity of employees and leaders will translate to enhanced innovation performance, but as we have discussed, the gulf between ideation and implementation requires attention. In short, the conceptual distance between creativity and innovation lies in the realization of the inherent value latent in new ideas. This distance is not trivial. As Levitt (­1963) notes, traversing this bridge between the generation of a new idea and the realization of its latent value through implementation seems to be the hardest part of the journey. Levitt puts this bluntly when he quips: The fact that you can put a dozen inexperienced people into a room and conduct a brainstorming session that produces exciting new ideas shows how little relative importance ideas themselves actually have. Almost anybody with the intelligence of the average businessman can produce them, given a halfway decent environment and stimulus. The scarce people are those who have the k­ now-​­how, energy, daring, and staying power to implement ideas. (­­p. 138) Simply stated, originating new ideas with potential value does not guarantee their realization. And, often, it seems to be the case that the “­realization of a creative idea” seems to be the hardest part of the creativity journey. Reports of the d­ ay-­​­­to-​­day struggles of innovators seem to confirm the point that most individuals and companies are sufficiently good at generating ideas. The “­bottleneck” in the creativity/­innovation process seems to occur further down the pipeline, specifically in developing ideas into products/­services and, more acutely, diffusing proven ideas across the company (­Birkinshaw et al., 2011). Much of the work focused on interventions and strategies targeted at enhancing this end of the innovation process is focused on variables outside the cognitive. For instance, Ekvall’s (­1996) widely cited work on organizational innovation focuses on factors around culture and climate rather than the underlying cognitive facets of members within organizations more directly. If cognitive psychologists are to engage with the task of enhancing innovative cognition, they will need to both consider innovation as a process and innovators as people, much in the same way that scholars have separated the creative process and the creative person (­e.g., Rhodes, 1961). The task is to map the cognitive impacts of ­h igher-​­order organizational manipulations and interventions, to more closely understand the mechanistic interplays at work, to investigate the role of innovators and the cognitive characteristics of these individuals, and to develop cognitive and behavioral interventions that promote 121

Nathaniel Barr et al.

enhanced innovative cognition. As noted above, complex ­multi-​­level interactions are at play in innovative climates. In considering how to develop, deliver, and test such interventions, perhaps the closest allies for cognitive psychologists in this regard would be applied behavioral scientists. The application of behavioral insights (­i.e., strategies and interventions meant to induce behavior change rooted in empirical research) and the scientific method across public, private, and ­non-​­profit settings has seen a surge in recent years (­see OECD, 2017, for a summary of the proliferation of such approaches in policy settings, or Thaler, 2015, for a history). Such practitioners often hail from judgment and d­ ecision-​­making laboratories, sometimes within cognitive areas in psychology departments, and represent ideal collaborators who understand cognitive psychology and are ­well-​­versed in overcoming the challenges associated with conducting research and testing interventions within organizations. Behavioral science has been argued to constitute a framework for driving innovation itself, given its capacity for risk taking, by way of testing creative, behaviorally informed ideas in place before launching more widespread innovation (­Peters, 2020). Similar i­ntervention-​ ­based experimentation, focused on cognitive and behavioral changes, should be deployed to understand innovation. Support for such a notion can be found in Schrage’s (­2014) book, The Innovator’s Hypothesis: How Cheap Experiments Are Worth More than Good Ideas, in which he advocates a cultural and strategic shift toward ­small-​­scale experiments that can enhance the innovative performance of a firm. If adoption of a scientific mindset and the ability to execute experiments in an organizational context are critical paths to enhancing innovative cognition, important work awaits in the development and testing of cognitive interventions that enhance these attitudes and abilities. “­The mental processes underlying scientific thinking and discovery have been investigated by cognitive psychologists, educators, and creativity researchers for over half a century” (­Dunbar, 2001, ­p. 115), and this body of knowledge could be better leveraged in the pursuit of making more individuals outside the domain of science better at scientific thinking, which could mark a key to them being better at innovation.

Conclusion Although providing a full account of creative cognition, from ideation to innovation, stands as incredibly challenging, progress toward this goal can yield innumerable benefits. To do so, cognitive psychologists must venture forth from the lab, apply ambition and ingenuity, and recognize the importance of research practically connected to the human capacity to change the world (­see Barr & Pennycook, 2018). Given the advances seen thus far in our attempts to unravel the secrets of creativity and innovation, it seems likely that we will significantly advance our understanding in the decades to come. Hopefully, these efforts will allow us to further move our learning about creative cognition from ideas to action… as ideas are not enough.

References Amabile, T. (­1998, ­September–​­October). How to kill creativity. Harvard Business Review, 76(­5), ­76–​­87. Amabile, T. M., Conti, R., Coon, H., Lazenby, J., & Herron, M. (­1996). Assessing the work environment for creativity. Academy of Management Journal, 39(­5), ­1154–​­1184. Anderson, N., Potocnik, K., & Zhou, J. (­2014). Innovation and creativity in organizations: A ­state-­​ ­­of-­​­­the-​­science review, prospective communtary, and guiding framework. Journal of Management, 40(­5), ­1297–​­1333.

122

Creative Cognition: From Ideation to Innovation Aronson, Z. H., Reilly, R. R., & Lynn, G. S. (­2008). The role of leader personality in new product development success: An examination of teams developing radical and incremental innovations. International Journal of Technology Management, 44(­­1–​­2), ­5 –​­27. Ausubel, D. P. (­1964). Creativity, general creative abilities, and the creative individual. Bulletin of the Council for Research in Music Education, 3, ­i–​­v. Axtell, C., Holman, D. J., Unsworth, K. L., Wall, T. D., Waterson, P. E., & Harrington, E. (­2000). Shopfloor innovation: Facilitating the suggestion and implementation of ideas. Journal of Occupational and Organizational Psychology, 73, 2­ 65–​­285. Ball, L. J., & Ormerod, T. C. (­2000). Putting ethnography to work: The case for a cognitive ethnography of design. International Journal of ­Human-​­Computer Studies, 53(­1), ­147–​­168. Barr, N. (­2018). Intuition, reason, and creativity: An integrative d­ ual-​­process perspective. In G. Pennycook (­Ed.), The New Reflectionism in Cognitive Psychology (­­pp. ­99–​­124). Abingdon: Routledge. Barr, N., Beaty, R., & Seli, P. (­2020). Autonomy and control across cognition: Insights from creativity, memory, mind wandering, and reasoning research. In D. D. Preiss, D. Cosmelli, & J. C. Kaufman (­Eds.), Creativity and the Wandering Mind (­­pp. ­25–​­54). Cambridge, MA: Academic Press. Barr, N., Pennycook, G., Stolz, J. A.,  & Fugelsang, J. A. (­2014). Reasoned connections: A ­dual-​ ­ process perspective on creative thought. Thinking  & Reasoning, 21(­1), 6­ 1–​­75. https://­doi. org/­10.1080/­13546783.2014.895915 Barr, N., & Pennycook, G. (­2018). Why reason matters: Connecting research on human reason to the challenges of the anthropocene. In G. Pennycook (­Ed.), The New Reflectionism in Cognitive Psychology (­­pp. ­119–​­142). Abingdon: Routledge. Barr, P. S., Stimpert, J. L., & Huff, A. S. (­1992). Cognitive change, strategic action, and organisational renewal. Strategic Management Journal, Summer, 47(­92), 5­ 23–​­549. Barron, F. (­1965). The psychology of creativity. New Directions in Psychology, 2, 1­ –​­134. Beaty, R. E. (­2015). The neuroscience of musical improvisation. Neuroscience & Biobehavioral Reviews, 51, ­108–​­117. Beaty, R. E., Benedek, M., Silvia, P. J., & Schacter, D. L. (­2016). Creative cognition and brain network dynamics. Trends in Cognitive Sciences, 20(­2), 8­ 7–​­95. Beaty, R. E., & Johnson, D. R. (­2021). Automating creativity assessment with SemDis: An open platform for computing semantic distance. Behavior Research Methods, 53(­2), ­757–​­780. Beaty, R. E., Silvia, P. J., Nusbaum, E. C., Jauk, E., & Benedek, M. (­2014). The roles of associative and executive processes in creative cognition. Memory & Cognition, 42(­7 ), ­1186–​­1197. https://­doi. org/­10.3758/­­s13421-­​­­014-­​­­0428-​­8 Bergman, J. P., Jantunen, A., & Tarkiainen, A. (­2015). Managerial cognition and dominant logic in innovation management: Empirical study in media industry. International Journal of Business Innovation and Research, 9(­3), 2­ 53–​­271. Birkinshaw, J., Bouquet, C., & Barsoux, J. L. (­2011, Winter). The 5 myths of innovation. MIT Sloan Management Review. Blogpost accessible at https://­sloanreview.mit.edu/­a rticle/­­the-­​­­5 -­​­­myths-­​­­of​­i nnovation/ Bledow, R., Frese, M., Anderson, N., Erez, M., & Farr, J. (­2009). A dialectic perspective on innovation: Conflicting demands, multiple pathways, and ambidexterity. Industrial and Organizational Psychology, 2(­3), ­305–​­337. Bowden, E. M., & ­Jung-​­Beeman, M. (­2003). Aha! Insight experience correlates with solution activation in the right hemisphere. Psychonomic Bulletin & Review, 10(­3), ­730–​­737. https://­doi.org/­10.3758/­ bf03196539 Bowden, E. M., ­Jung-​­Beeman, M., Fleck, J., & Kounios, J. (­2005). New approaches to demystifying insight. Trends in Cognitive Sciences, 9(­7 ), 3­ 22–​­328. https://­doi.org/­10.1016/­j.tics.2005.05.012 Christensen, B. T. (­2024, this volume). Creative cognition and entrepreneurship. In L. J. Ball & F. V ­ allée-​­Tourangeau (­Eds.), The Routledge International Handbook of Creative Cognition. Abingdon, Oxon: Routledge. Christensen, C. (­1997). Innovator’s Dilemma. Harvard: Harvard Business School Press. Clark, P. M., & Mirels, H. L. (­1970). Fluency as a pervasive element in the measurement of creativity. Journal of Educational Measurement, 7(­2), 8­ 3–​­86. https://­doi.org/­10.1111/­j.­1745-​­3984.1970.tb00699.x Cropley, D. H. (­2024, this volume). Creative cognition in engineering and technology. In L. J. Ball & F. ­Vallée-​­Tourangeau (­Eds.), The Routledge International Handbook of Creative Cognition. Abingdon, Oxon: Routledge.

123

Nathaniel Barr et al. Dunbar, K. (­2001). What scientific thinking reveals about the nature of cognition. In K. Crowley, C. D. Schunn, & T. Okada (­Eds.), Designing for Science: Implications from Everyday, Classroom, and Professional Settings (­­pp. ­115–​­140). New York, NY: Routledge. Dunbar, K., & Blanchette, I. (­2001). The in vivo/­in vitro approach to cognition: The case of analogy. Trends in Cognitive Sciences, 5(­8), ­334–​­339. Dyer, J., Gregersen, H. B., & Christensen, C. M. (­2011). The Innovator’s DNA: Mastering the Five Skills of Disruptive Innovators. Boston, MA.: Harvard Business Press. Ekvall, G. (­1996). Organizational climate for creativity and innovation. European Journal of Work and Organizational Psychology, 5(­1), ­105–​­123. Gable, S. L., Hopper, E. A., & Schooler, J. W. (­2019). When the muses strike: Creative ideas of physicists and writers routinely occur during mind wandering. Psychological Science, 30(­3), ­396–​­404. Gibson, J. J. (­2014). The Ecological Approach to Visual Perception: Classic Edition. Hove: Psychology Press. Gilhooly, K. J., Fioratou, E., Anthony, S. H., & Wynn, V. (­2007). Divergent thinking: Strategies and executive involvement in generating novel uses for familiar objects. British Journal of Psychology, 98(­4), ­611–​­625. https://­doi.org/­10.1111/­j.­2044-​­8295.2007.tb00467.x Glăveanu, V. P. (­ 2014). Distributed Creativity: Thinking Outside the Box of the Creative Individual. Cham/­Heidelberger: Springer International Publishing. Gray, K., Anderson, S., Chen, E. E., Kelly, J. M., Christian, M. S., Patrick, J.,… & Lewis, K. (­2019). “­Forward flow”: A new measure to quantify free thought and predict creativity. American Psychologist, 74(­5), 539. Green, A. E. (­2016). Creativity, within reason: Semantic distance and dynamic state creativity in relational thinking and reasoning. Current Directions in Psychological Science, 25(­1), ­28–​­35. Green, A. E., Cohen, M. S., Kim, J. U., & Gray, J. R. (­2012). An explicit cue improves creative analogical reasoning. Intelligence, 40(­6), ­598–​­603. Green, A. E., Spiegel, K. A., Giangrande, E. J., Weinberger, A. B., Gallagher, N. M., & Turkeltaub, P. E. (­2017). Thinking cap plus thinking zap: tDCS of frontopolar cortex improves creative analogical reasoning and facilitates conscious augmentation of state creativity in verb generation. Cerebral Cortex, 27(­4), ­2628–​­2639. Greenberg, J., & Tomlinson, E. C. (­2004). Situated experiments in organizations: Transplanting the lab to the field. Journal of Management, 30(­5), ­703–​­724. Guilford, J. P. (­1950). Creativity. American Psychologist, 5, ­4 44–​­454. Hargadon, A. B. (­1999). Group cognition and creativity in organizations. Research on Managing Groups and Teams, 2(­1), ­137–​­155. Hennessey B. A. & Amabile, T. M. (­2010). Creativity. Annual Review of Psychology, 61, ­569–​­598. Kacerauskas, T. (­2016). The paradoxes of creativity management. E+M Ekonomie a Management, 19(­4), ­33–​­43. Kanter, R. M. (­1988). Three tiers for innovation research. Communication Research, 15(­5), ­509–​­523. Kaufman, J. C.,  & Beghetto, R. A. (­2009). Beyond big and little: The four c model of creativity. Review of General Psychology, 13(­1), ­1–​­12. Kenett, Y. N. (­2018). Investigating creativity from a semantic network perspective. In  Z. Kapoula, E. Volle, J. Renoult, & M. Andreatta (­Eds.), Exploring Transdisciplinarity in Art and Sciences (­­pp. ­49–​­75). Cham: Springer. Kienitz, E., Quintin, E. M., Saggar, M., Bott, N. T., Royalty, A., Hong, D. W. C.,… & Reiss, A. L. (­2014). Targeted intervention to increase creative capacity and performance: A randomized controlled pilot study. Thinking Skills and Creativity, 13, ­57–​­66. Kim, S., Vincent, L. C., & Goncalo, J. A. (­2013). Outside advantage: Can social rejection fuel creative thought? Journal of Experimental Psychology: General, 142(­3), ­605–​­611. Kimmel, A. J. (­1996). Ethical Issues in Behavioral Research: A Survey. Oxford: Blackwell Publishing. Kirton, M. (­1994). A theory of cognitive style. In M. Kirton (­Ed.), Adaptors and Innovators: Styles of Creativity and ­P roblem-​­Solving (­­pp. ­1–​­33). New York, NY: Routledge. Knoblich, G., Ohlsson, S., Haider, H., & Rhenius, D. (­1999). Constraint relaxation and chunk decomposition in insight problem solving. Journal of Experimental Psychology: Learning, Memory, and Cognition, 25(­6), ­1534–​­1555. https://­doi.org/­10.1037/­­0278-​­7393.25.6.1534 Kounios, J., & Beeman, M. (­2009). The Aha! moment. Current Directions in Psychological Science, 18(­4), ­210–​­216. https://­doi.org/­10.1111/­j.­1467-​­8721.2009.01638.x Levitt, T. (­1963). Creativity is not enough. Harvard Business Review, 41(­3), ­72–​­83.

124

Creative Cognition: From Ideation to Innovation Literat, I., & Glaveanu, V. P. (­2016). Same but different? Distributed creativity in the internet age. Creativity: ­T heories-­​­­Research-​­Applications, 3(­2), ­330–​­342. Lopata, J. A., Nowicki, E. A., & Joanisse, M. F. (­2017). Creativity as a distinct trainable mental state: An EEG study of musical improvisation. Neuropsychologia, 99, ­246–​­258. Loy Jr., J. W. (­1969). Social psychological characteristics of innovators. American Sociological Review, 34(­1), ­73–​­82. MacGregor, J. N., & Cunningham, J. B. (­2008). Rebus puzzles as insight problems. Behavior Research Methods, 40(­1), ­263–​­268. https://­doi.org/­10.3758/­brm.40.1.263 Manral, L. (­2011). Managerial cognition as bases of innovation in organisations. Management Research Review, 34(­5), ­576–​­594. McNemar, Q. (­1946). ­Opinion-​­attitude methodology. Psychological Bulletin, 43, ­289–​­374. McPherson, M., & Limb, C. J. (­2013). Difficulties in the neuroscience of creativity: Jazz improvisation and the scientific method. Annals of the New York Academy of Sciences, 1303(­1), ­80–​­83. https://­doi. org/­10.1111/­nyas.12174 Mednick, S. A. (­1962). The associative basis of the creative process. Psychological Review, 69, ­220–​­232. Miettinen, R. (­2006). The sources of novelty: A cultural and systemic view of distributed creativity. Creativity and Innovation Management, 15(­2), ­173–​­181. Miller, G. (­2012). The smartphone psychology manifesto. Perspectives on Psychological Science, 7(­3), ­221–​­237. ­M iron-​­Spektor, E., & Argote, L. (­2008). The effect of paradoxical cognition on individual and team innovation. Academy of Annual Meeting Proceedings, 1, ­1–​­6. Mueller, J., Melwani, S., & Goncalo, J. A. (­2011). The bias against creativity: Why people desire but reject creative ideas. Psychological Science, 23(­1), ­13–​­17. Mumford, M. D. (­2000). Managing creative people: Strategies and tactics for innovation. Human Resource Management Review, 10(­3), ­313–​­351. Mumford, M. D., ­Bedell-​­Avers, K. E., & Hunter, S. T. (­2008). Planning for innovation: A ­multi-​­level perspective. In M. D. Mumford, S. T. Hunter, & K. E. ­Bedell-​­Avers (­Eds.), ­Multi-​­L evel Issues in Creativity and Innovation (­­pp. ­107–​­154). Bingley: Emerald Group Publishing Limited. Mumford, M. D., & Gustafson, S. B. (­1988). Creativity syndrome: Integration, application, and innovation. Psychological Bulletin, 103(­1), 27. ­ ulti-​­level perspective on Mumford, M. D., & Hunter, S. T. (­2005). Innovation in organizations: A m creativity. In M. D. Mumford, S. T. Hunter, & K. E. ­Bedell-​­Avers (­Eds.), ­Multi-​­level Issues in Strategy and Methods (­­pp. ­9 –​­73). Bingley: Emerald Group Publishing Limited. Mumford, M. D., Hunter, S. T., & Byrne, C. L. (­2009). What is the fundamental? The role of cognition in creativity and innovation. Industrial and Organizational Psychology, 2(­3), ­353–​­356. Neisser, U. (­1976). Cognition and Reality: Principles and Implications of Cognitive Psychology. New York, NY: Freeman. OECD (­2009). Innovation in Firms: A Microeconomic Perspective. Oslo: OECD. OECD (­2017). Behavioural Insights and Public Policy: Lessons from Around the World. Paris: OECD Publishing. Olson, J. A., Nahas, J., Chmoulevitch, D., Cropper, S. J., & Webb, M. E. (­2021). Naming unrelated words predicts creativity. Proceedings of the National Academy of Sciences, 118(­25), ­1–​­6. Patterson, F. (­1999). The Innovation Potential Indicator. Manual and User’s Guide. Oxford: OPP Ltd. Patterson, F. (­2002). Great minds don’t think alike? Person level predictors of innovation at work. In C. Cooper & I. T. Robertson (­Eds.), International Review of Industrial and Organizational Psychology. New York, NY: John Wiley & Sons Ltd. Patterson, F., & Zibarras, L. D. (­2017). Selecting for creativity and innovation potential: Implications for practice in healthcare education. Advances in Health Science Education, 22, ­417–​­428. Paulus, P. B., & Nijstad, B. A. (­Eds.) (­2003). Group Creativity: Innovation through Collaboration. Oxford: Oxford University Press. Peters, K. (­Spring, 2020). Behavioral science: The answer to innovation? Rotman Management Magazine. Plucker, J. A., Beghetto, R. A., & Dow, G. T. (­2004). Why isn’t creativity more important to educational psychologists? Potentials, pitfalls, and future directions in creativity research. Educational Psychologist, 39(­2), ­83–​­96. ­Reiter-​­Palmon, R., & Japp, P. (­2024, this volume). Team cognition and team creativity. In L. J. Ball & F. ­Vallée-​­Tourangeau (­Eds.), The Routledge International Handbook of Creative Cognition. Abingdon, Oxon: Routledge.

125

Nathaniel Barr et al. ­Reiter-​­Palmon, R., Wigert, B., & de Vreede, T. (­2012). Team creativity and innovation: The effect of group composition, social processes, and cognition. In Handbook of Organizational Creativity (­­pp. ­295–​ 3­ 26). London: Academic Press. Rhodes, M. (­1961). An analysis of creativity. The Phi Delta Kappan, 42(­7 ), ­305–​­310. Rubenstein, C. (­1982). Psychology’s fruit flies. Psychology Today, 16, ­83–​­84. Sarmiento, J. W., & Stahl, G. (­2008). Group creativity in interaction: Collaborative referencing, remembering, and bridging. International Journal of ­Human–​­Computer Interaction, 24(­5), ­492–​­504. Sawyer, R. K. (­2012). Explaining Creativity: The Science of Human Innovation. 2nd Edition. Oxford: Oxford University Press. Sawyer, R. K., & DeZutter, S. (­2009). Distributed creativity: How collective creations emerge from collaboration. Psychology of Aesthetics, Creativity, and the Arts, 3(­2), 81. Schendel, D., & Hofer, C. W. (­Eds.) (­1979). Strategic Management: A New View of Business Policy and Planning. Boston: Little, Brown. Schrage, M. (­2014). The Innovator’s Hypothesis: How Cheap Experiments are Worth More than Good Ideas. Cambridge, MA: MIT Press. Schumpeter, J. A. (­1934). The Theory of Economic Development: An Inquiry into Profits, Capital, Credit, Interest, and the Business Cycle (­No. HB175 S462). London: Transaction Publishers. Scott, G., Leritz, L. E., & Mumford, M. D. (­2004). The effectiveness of creativity training: A quantitative review. Creativity Research Journal, 16(­4), ­361–​­388. Shen, W., Ball, L. J., & Richardson, B. H. (­2024, this volume). Creative cognition in advertising. In ­ allée-​­Tourangeau (­Eds.), The Routledge International Handbook of Creative Cognition. L. J. Ball & F. V Abingdon, Oxon: Routledge. Straw, B. (­1995). Why no one really wants creativity. In C. Ford & D. Gioia (­Eds.), Creative Action in Organizations: Ivory Tower Visions & Real World Voices. Thousand Oaks, CA: Sage Publications, Inc. Stubbart, C. I. (­1989). Managerial cognition: A missing link in strategic management research. Journal of Management Studies, 26(­4), ­325–​­347. Subramaniam, K., Kounios, J., Parrish, T. B., & J­ung-​­Beeman, M. (­2009). A brain mechanism for facilitation of insight by positive affect. Journal of Cognitive Neuroscience, 21(­3), 4­ 15–​­432. Sund, K. J., Galavan, R. J., & Brusoni, S. (­2018). Cognition and innovation: A framework and invitation to explore. In K. J. Sund, R. J. Galavan, & S. Brusoni (­Eds.), Cognition and Innovation. Bingley: Emerald Publishing Limited. Thaler, R. H. (­2015). Misbehaving: The Making of Behavioral Economics. London: WW Norton & Company. Treffinger, D. J., & Isaksen, S. G. (­2005). Creative problem solving: The history, development, and implications for gifted education and talent development. Gifted Child Quarterly, 49(­4), 3­ 42–​­353. Treffinger, D. J., Selby, E. C.,  & Isaksen, S. G. (­2008). Understanding individual p­ roblem-​­solving style: A key to learning and applying creative problem solving. Learning and Individual Differences, 18, ­390–​­401. Van Berkel, N., Ferreira, D., & Kostakos, V. (­2017). The experience sampling method on mobile devices. ACM Computing Surveys (­CSUR), 50(­6), ­1–​­40. Vecchiato, R. (­2017). Disruptive innovation, managerial cognition, and technology competition outcomes. Technological Forecasting and Social Change, 116, ­116–​­128. Wiltschnig, S., Onarheim, B., Christensen, B. T., Dalsgaard, P., Korsgaard, H., Ball, L. J.,… & Hebert, A. M. (­2011, October). Integrating laboratory paradigms and ethnographic field studies for advancing analyses of creative processes. In Procedings of the Second Conference on Creativity and Innovation in Design (­­pp. ­365–​­366). Yang, D., Wang, A. X., Zhou, K. Z., & Jiang, W. (­2019). Environmental strategy, institutional force, and innovation capability: A managerial cognition perspective. Journal of Business Ethics, 159(­4), 1­ 147–​­1161. Zhao, H., & Seibert, S. E. (­2006). The big five personality dimensions and entrepreneurial status: A ­meta-​­a nalytical review. Journal of Applied Psychology, 91(­2), 259. Zhao, Q., Zhou, Z., Xu, H., Fan, W.,  & Han, L. (­2014). Neural pathway in the right hemisphere underlies verbal insight problem solving. Neuroscience, 256, 3­ 34–​­341. https://­doi.org/­10.1016/­j. neuroscience.2013.10.019

126

9 INSIGHT IN THE KINENOETIC FIELD Frédéric ­Vallée-​­Tourangeau

Insight in the Kinenoetic Field The disciplinary exigencies of cognitive psychology qua science of the mind guide and constrain the nature of the research enterprise. The ontological keystone of this enterprise is the enduring notion that mental processes are the prime drivers of the construction of knowledge and the primary cause of the origin of new ideas. Coupled to this notion is the apparently inescapable tautology that these mental processes are the product of an individual’s mind. A cognitivist perspective on creativity locates it in the person, or more specifically, in the person’s head. A broad palette of instruments has been developed to measure creativity. ­Self-​­report measures range from assessing ratings of creative ideation to eliciting creative attitudes. Others, such as creative behaviour inventories, are more neutral with respect to the ontological locus of creativity. Performance tasks of divergent and convergent thinking are also employed to measure creativity. In this chapter, I focus on ­so-​­called insight problems as a platform to assess creativity. A taxonomy of such problems has been offered (­Weisberg, 1995). The common feature of these problems is that they resist, in naïve participants, the direct application of knowledge in ­long-​­term memory to transform their start state into their normative goal state. Yet these problems vary. Some problems are formulated so as to encourage a misleading interpretation or activate a prepotent response that must be abandoned or inhibited; operators might not obviously come to mind, although the problems are usually elementary riddles that can be solved with simple operators. Finally, some specify a goal state, but it is the manner with which to reach the goal state that is not obviously apparent. However, all insight problems are designed to create or compound a conceptual difficulty that must be overcome through the development of new ideas that bring the solution within a “­mental look ahead horizon” (­Ohlsson, 1984, p­ . 124, emphasis added). The current theoretical debate in the psychology of insight ­problem-​­solving aims to explain this “­mysterious leap of imagination” (­Ohlsson, 1984, ­p. 120) in one of two ways. On one side are the proponents of a special unconscious process, with semantic inferences the likely candidate (­Ohlsson, 1992). These inferences result from the propagation of activation through networks of semantic elements that evantuates in hitherto unconnected paths, linking distant and disparate elements, and result in a new interpretation or conception of the DOI: 10.4324/9781003009351-10

127

Frédéric ­Vallée-­Tourangeau

problem. This new interpretation cues new operators and a new conceptual plane within which the solution can be grasped. On the other side are researchers (­e.g., Fleck & Weisberg, 2013; Weisberg, 2018) who argue that the (­re)­a nalysis of the problem elements is evinced through conscious, deliberate, and effortful processes that promote a sounder and more productive interpretation of the problem (­the ­so-​­called ­business-­​­­a s-​­usual view). Proponents of both positions agree that the potential for discovering the solution resides, as it were, in the agent working on the problem, not unlike how stumpers1 are easily understood once a certain misleading assumption is abandoned. The bottom line is that, whether consciously or unconsciously, the reasoning agent bootstraps herself to a new plane of understanding, and the leap of imagination is the result of a mental process. She has no need to act in, on, or through the world to find her answer.

Costing the Price of a New Idea Creativity involves a cost: resources must be invested to make something new. From the special processes perspective, the price for the journey is nil since the agent is the passive vessel sailing along undercurrents of semantic activation that may take her to a promising destination. From a gradualist ­business-­​­­as-​­usual perspective, a new idea is the result of investments in cognitive capacity and ability. The more such resources are at an agent’s disposal, the more can be invested, and the more likely it is that a new idea is produced. But crucially, it is difficult to examine the exact nature of those transactions because of the covert nature of the exchange. In other words, cognitive psychology deals with unobservables. The exact manner in which old ideas are transformed into new ones cannot be directly observed. This daunting challenge is tackled through different strategies. A psychometrics strategy is adopted by some researchers: they seek to establish the positive association between measures of cognitive ­capacity—​­such as working memory and ­intelligence—​­and solution rates with both insight and analytic problems (­e.g., Chuderski & Jastrzębski, 2018). Eye tracking is another strategy employed to pin down, at least, the allocation of attentional resources to different problem elements (­e.g., Bilalić et al., 2021). Finally, verbal protocols are yet another method to identify the steps and hurdles that pave the road to a new idea (­e.g., Fleck  & Weisberg, 2013). Even so, these strategies have shortcomings, which mean that they only partially attend to the challenge. Correlational evidence cannot by itself help shed light on ­ allée-​­Tourangeau, 2020); eye tracking how a problem was solved (­­Vallée-​­Tourangeau & V data are vulnerable to many interpretations (­Orquin & Holmqvist, 2018); and it is unclear the extent to which participants’ ­meta-​­narratives reliably and validly map onto mental processes (­Ross & ­Vallée-​­Tourangeau, 2021). All scholarly and scientific enterprises should abstain from peddling tautological explanations. For example, the success of someone solving syllogistic reasoning problems may be attributable to intelligence. However, intelligence is commonly operationalized in terms of tests that involve syllogistic reasoning; hence, what’s on offer here is a circular n ­ on-​ ­explanation. Science chisels the stuff that makes up the world. The stuff is composed of substances that are sculpted through the gradual identification of attributes and actions. These substances are reified through tests and trials, and their behaviour is noted. The goal of science is to understand how a substance emerges from a set of attributes and their actions (­to adapt Latour, 2016). To explain creativity while avoiding tautology, one must therefore be careful not to draw upon attributes that have a creative pedigree of their own; there lurks an unproductive regress that begs the question and vitiates the explanation. The psychometrics strategy employed by some researchers in insight p­ roblem-​­solving appears 128

Insight in the Kinenoetic Field

to avoid this trap by seeking to establish correlations between participants’ cognitive skills and capacities with their performance on insight problems. The research programme typically proceeds by having participants work on a battery of insight problems, from which an aggregate solution rate is computed. In the same testing session (­these testing sessions can be quite long, spanning the better of eight hours, see Chuderski & Jastrzębski, 2018), participants’ cognitive capacity and skills are gauged through a battery of working memory span tests and tests of fluid intelligence. The data analysis strategy then calculates the correlation between aggregate performance on the insight problems and measures of working memory and intelligence. Typically, these correlations are positive. But the heavy cost of avoiding tautology is that this experimental procedure and analysis strategy are quite unable to explore the relationship between the microgenesis of new ideas and problem solution pathways. In other words, this research methodology does not and cannot help researchers understand the actual processes involved in the discovery of problem solutions. As such, creativity remains unexplained. The inability of the psychometrics manifesto to unveil creativity can be attributed to three key factors. First, performance on any given problem is lost to aggregate indices; it is not technically unrecoverable, but it is usually not reported. Second, the granularity of the analysis for performance on a given p­ roblem—​­the data c­ apture—​­is insufficient to reveal how participants solved any given problem. Third, the research methodology employs s­o-​­called ­second-​­order tasks, that is, problems are presented as riddles in the form of propositions rather than as ­fi rst-​­order tasks, that is, with a physical model of the problem that can be manipulated and physically transformed (­­Vallée-​­Tourangeau  & March, 2019). ­First-​­order ­problem-​­solving requires interacting with the world, which alters the cognitive terrain (­to adapt Kirsh, 2010, p­ . 443). Interactivity may scaffold thinking, resulting in better performance, but more importantly, interactivity makes the process visible. It lends itself to a careful recording of participants’ actions, the resulting changes to the physical properties of the problem, and how these changes trigger new actions and perceptions. The data are rich and invite a granular qualitative mapping of the ­trajectory—​­in space and ­time—​­of the changes to the problem from the start state to the goal state. Equally important, this granular analysis can be applied to ­problem-​­solving successes and failures symmetrically, unveiling conceptual breakthroughs as well as cul de sacs. It turns unobservable cognitive processes into observable actions and changes in the world (­observables to participants as well as to researchers). Let me wrap up these introductory remarks. Creativity does not materialize out of nothing; it involves an epistemic transformation that has a cost attached to it. Researchers should not lose sight of these costs, and taking a loan on unconscious processes or intelligence is only a temporary measure. The loans should eventually be repaid. The exact nature of the transaction is difficult to inspect because of its covert nature. This difficulty is compounded by methodological decisions that reflect a mentalist ontological commitment; the disciplinary identity and concomitant professional exigencies naturally constrain the focus of the cognitive psychologist to look at the mind and limit the exploration of creativity to a mind that is artificially carved out of the body and the world in which it is embedded. S­ econd-​ ­order ­problem-​­solving task environments can only be navigated mentally. In the absence of interactivity, no actions on the world can be recorded, no changes to a physical model of the problem can be captured, and no trajectory from the start state to the goal state can be traced objectively. A psychometrics research programme, focussing exclusively on aggregate measures of performance, does not unveil how new ideas arise. Working memory and intelligence may correlate with aggregate solution rates, but the origin of these solutions remains shrouded in a mysterious cloak. 129

Frédéric ­Vallée-­Tourangeau

The Kinenoetic Field Cognitive archaeologists deal with objects produced by ancestral hominins and formulate conjectures on the nature of these hominins’ cognitive abilities. Malafouris’s material engagement theory (­2013, 2020) introduces notions such as “­thinging” and the “­hylonoetic field”. These notions cast the association between objects and minds in transactional c­ o-​ ­constitutive terms. The term “­hylonoetic” combines matter and knowledge and was coined to capture the idea of cognition, cognitive development, and evolution as proceeding with and through engagement with materiality. Take the symmetric shape of the oldest Acheulean hand axe: it has been argued that this reflects intentionality, the enactment of a mental plan in these ancient hominins’ mind, to create an artefact of that particular shape. Stone tools are then considered: To be windows into prehistoric minds, not components of thinking itself. Assigning agency to hominins and using artifacts to document minds provided useful insights into the evolution of certain cognitive abilities, but it ignored the role that tools themselves played in early hominin cognition, and how that role influenced the evolution of cognition itself. (­Wynn et al., 2021, ­p. 101) In contrast, and from a material engagement perspective, the features of stone tools, such as handaxe symmetry, emerge out of the manufacturing process: The stone projects towards the knapper as much as the knapper projects towards the stone and together they delineate the cognitive map of what we may call an extended intentional state. The knapper first thinks through and with the stone before being able to think about the stone and hence about one’s self as a conscious reflectively aware agent. (­Malafouris, 2010a, p­ . 19) Symmetry is a ­by-​­product of engaging with the materiality of the flint and the hammer stone rather than its cause. This is a particularly fruitful perspective on innovation and toolmaking because it takes no loan on the conjectured representational or intentional abilities of these hominins. Rather, it casts intelligent innovation as a function of engaging with the world, of manipulating and transforming objects. Innovation and creativity emerge out of the concurrent transformation of the s­o-​­called subject and object. Innovation here is explained not in terms of an innovative mind and eschews the regress entirely by relying on processes that do not presume innovation but out of which it emerges. These dynamically unfolding ­object-​ ­thought mutualities are, however, themselves conjectures because their ­co-​­evolution cannot be directly observed: the objects are archaeological artefacts, the temporal scale epochal, their creators long silent. Fast forward several millennia to a cognitive psychology laboratory. The aim of the lab is not to conject but to establish how a naïve participant discovers the solution to an insight problem. The problem description is accompanied by a manipulable object, an object that can be made into a physical model of the solution. To be sure, the creative innovation is not of the same magnitude or interest as those reflected in the manufacture and manipulation of important artefacts (­e.g., such as the Mycenaean Linear B tablets; Malafouris, 2010b), and the inferences about the creative and innovative potential of ancestral hominins are a taller challenge than those of a typical university undergraduate participating in a ­problem-​­solving 130

Insight in the Kinenoetic Field

experiment. Nonetheless, a properly instrumentalized procedure can reveal the transactional ­co-​­constitutive evolution of the participant’s understanding of the solution and the physical model of the problem. In other words, ­object-​­thought mutualities can be directly observed. The movement or transformation of the object both reflects and triggers change in the participant’s understanding of the problem. New ideas from this perspective emerge from a kinenoetic field, a term that underscores the importance of object transformations in space and time (­see Ross & ­Vallée-​­Tourangeau, 2021). These transformations are not perceived passively; they are not traced through binoculars by a distant observer. The movements result from the ­agent-​­world coupling. The relief of the kinenoetic field is mapped by recording the dynamic changes in the nature and position of the object. Let me shift the narrative to a more concrete register with a classic v­ isuo-​­spatial insight problem, the triangle of coins (­top of ­Figure 9.1). In this problem, participants must discover how to transform a configuration of ten coins in the shape of a triangle that points down into one that points up. The difficulty of this problem does not lie in the nature of the operators nor in how the goal state should look like, since imagining the shape pointing up requires elementary ­v isuo-​­spatial skills and corresponds to a ­well-​­encoded and frequently encountered cultural object or symbol. The challenge lies in the following constraint in the problem definition. The goal state is realized by moving three and only three coins. Thus, the new idea that is required, the new thought that needs to be constructed, is wrought through the identification of the critical coins among the set of 10. Participants work to discover which three coins must be moved to change the shape of the triangle. The critical coins are the vertices, ­colour-​­coded in ­Figure 9.1 (­a lthough not for participants working on the problem); the solution is illustrated in the t­op-​­right panel of F ­ igure  9.1. Few participants solve this problem quickly, with a substantial proportion failing to find the answer within ten minutes. We instrumentalized the experimental procedure in the following manner (­­ Vallée-​ ­Tourangeau et al., 2021). First, we present the problem as a ­first-​­order one, that is, with a physical model of the problem where participants can move the coins and observe the results of their moves, namely the new configuration that is produced by the movements. The problem is presented as an interactive game on a computer tablet with a reset button, and participants can reset the board as often as they wish to its initial configuration, as a triangle pointing down. Second, we labelled the coins and arrayed them on a 9 × 9 grid with numbered rows and lettered columns. Third, we filmed participants as they worked on the problem, and it is this video evidence that informed a detailed granular mapping of the moves and the change in the object over time. Instrumentalizing the procedure in this manner, coupled with the video evidence, help us ­recover—​­and with ­precision—​­which coin was moved where and when, and capture the iterative relationship between the configuration of the problem and the participants’ actions. It is this iterative relationship that is responsible for producing what we are calling “­­object-​­thought mutualities”, and our videoed, instrumentalized procedure allows us to ­re-​­create, for each participant a chart of the t­emporal-​­spatial kinenoetic field. In addition, we can animate the movements an analytic strategy that clearly reveals the types of configurations that are repeatedly explored, those that are abandoned, and the trajectory to the goal state. One such animated trajectory for one of our participants can be accessed at https://­osf. io/­et4pq/: it recreates the dynamic changes to the problem configuration as the participant labours to discover the identity of the three critical c­ oins—​­the ­vertices—​­to evince a solution. The animation also illustrates the cost of discovery, the slow, ­non-​­linear, and heteroscalar evolution from a state of unknowing to a state of knowing; it captures the genesis of a new idea. In addition to this animated reconstruction of the trajectory, we created an index we call the “­m igration ratio”. The migration ratio indicates numerically the extent to which changes 131

Frédéric ­Vallée-­Tourangeau

­Figure 9.1 Top panels: start and goal states of the triangle of coins problem; vertices are c­ olour-​­coded here for illustrative purposes, but the coins were not c­olour-​­coded in this manner for participants. Middle panels illustrate the migration ratio; the number of coins along row 7 divided by the number of coins in the rows above (.29 at the start state, .67 at the goal state). Bottom panel: migration ratio (­primary y axis plotted with the black circles) and latency per move (­secondary y axis) for a participant working and solving the triangle of coins, plotted as a running average over the previous five moves Source: Adapted from Vallée-Tourangeau et al., 2021.

132

Insight in the Kinenoetic Field

in the resulting configuration approximate the target configuration (­the solution). The middle panels of ­Figure 9.1 illustrate how the migration ratio is calculated. It is the ratio of coins along row 7 that will form the base of the correct triangular configuration over those in all the rows above. At the start state, the migration ratio is 2/­7 or .29; at the goal state, it is 4/­6 or .67. In addition to creating a map of the kinenoetic field, we can create another inscription device that illustrates how coins are gradually configured to approximate the target migration ratio. The migration ratio can capture efforts to move coins down to form a base along row 7, rather than migrating coins north, a common early strategy observed with this problem (­in the above cited video animation, this is clearly observed for the first 45 moves that aim to move coins north without migrating the base of the triangle down). The bottom panel of F ­ igure 9.1 plots the migration ratio (­a long the primary y axis) and the latency per move (­a long the secondary y axis). The dynamic unfolding of these data series is also illustrated in the animated video. This inscription device provides compelling perceptual evidence of two key features of the process of discovering the critical coins with which the normative solution can be constructed: (­i) it is gradual, as reflected by the rise in the migration ratio over the last ten moves; and (­ii) it is heteroscalar; the process unfolds along different time scales, as reflected by the oscillation in move latencies. The gradual rise of the migration ratio graphically demonstrates that the participant does not know the answer before it is physically constructed in front of her. This is also confirmed by the participant making an incorrect announcement: at this point she had moved four coins to create a triangle pointing up; after receiving feedback that her solution was i­ncorrect—​­the feedback was generic, right or wrong, without specifying which coins were in the correct or incorrect ­position—​­she created the right configuration and then announced her answer. Note that participants were not forced to physically create the right configuration; that is, they could announce their solution at any time on the basis of mental projections. However, with one exception, all our participants constructed the right configuration and then announced the solution ( ­­Vallée-​­Tourangeau et al., 2021). The fact that all our successful participants laboured to construct the solution before announcing it implies an iterative transactional process between what the agent knows and the physical realization of the solution. In other words, these ­object-​­thought mutualities reveal the ­co-​­determination of epistemology and ontology. Until the solution is constructed, knowledge about the object and the nature of the object are in a relationship of unstable c­ o-​ ­dependence that finds equilibrium and sediments only as the object’s normative shape is constructed. March and Malafouris (­this volume) describe this process of becoming as “­learning into existence”. The solution is materially enacted rather than being mentally simulated first and then physically implemented. This suggests that for some participants, insight should be better thought of in terms of “­outsight” (­­Vallée-​­Tourangeau & March, 2019): and indeed, some of our participants exhibited stereotypical eureka m ­ oments—​­markers normally associated with a phenomenological response to i­nsight—​­after seeing the constructed solution but not before (­see https://­youtu.be/­ZZSC549UyTg, note at 0:00:07, the d­ ouble-​­take expressed with both hands splayed open when the participant created the solution).

Insight Reductionism and the Deceitful Inversion Insight is classically defined as a sudden gestalt, marking the swift reorganization of a mental representation (­Gilhooly & Webb, 2018). In Fleck and Weisberg’s (­2013) integrative framework developed from extensive verbal protocol data, insight, the pure ­k ind—​­that is the 133

Frédéric ­Vallée-­Tourangeau

distinct phenomenological experience that arises after a protracted moment of impasse and accompanies the sudden appearance of a new ­idea—​­is relatively rare. In their study, these pure insight moments accounted for 7% of the solutions in the verbal protocols. In Danek et al. (­2020), 15% of the correct solutions occurred suddenly, and these sudden solutions were coupled to higher Aha! ratings. The paucity of the phenomenon does not stop researchers from better understanding it. Efforts are made to triangulate it through different methods, including brain imaging as well as interventions to encourage a less controlled form of thinking (­e.g., hypofrontality, the transient or neuropsychologically chronic kind; Dietrich, 2019; de Souza et al., 2014) by weakening executive function. These efforts are sometimes prefaced with vignettes of discovery as reported by the thinkers or innovators who claim to have had new ideas of momentous importance in a dream or while engaging in unrelated pedestrian activities. As Brannigan (­1981) argues, the explanatory or the “­causal adequacy of the gestalt switch” as a model of the origin of a new idea is questionable. The gestalt switch explanation [I]n its simplest expression (…) takes the form of a recommendation that discoveries occur as the result of gestalt switches. Since the switch is the thing that is occurring, it is circular to identify it as a cause of this. (­­p. 36) Significantly, our ­fi rst-​­order approach provides compelling evidence that the gestalt switch can trip after the solution is constructed. Betting on gestalt switches, we, as others have argued, will not pay off.2 As a psychological phenomenon, such switches deserve attention (­e.g., Danek, 2018); however, elevating the phenomenological experience either as a marker of a process (­see Ross, 2021) or as a cause of creativity and discovery does not pave the way for a productive programme of research. Investing explanatory capital in gestalt switches buffers the theoretical retreat against other explanations of creativity in terms of microprocesses that operate at the ­agent-​­world interface. We turn away from the notion of insight and look instead to Latour and Woolgar (­1986, ­Chapter 4 “­The microprocessing of facts”) and their critical reflections on the “­special” nature of scientific thinking in contradistinction to any other form of thinking: [In stopping the] sociological enquiry at the level of mute individual thought (…) science will once again appear extraordinary. Our position is not unlike the opponents of vitalism in ­n ineteenth-​­century biology. No matter what progress was made by biologists to explain life in purely mechanistic and materialist terms, some aspects always remained unexplained. There were always some corners in which notions of “­soul” or the “­pure vital force” could find refuge. Similarly, the notion that there is something special about science, something peculiar or mysterious which materialist and constructivist explanations can never grasp, is pushed further and further. But this notion will remain as long as the idea lingers that there is some peculiar thinking process in the scientist’s mind. It is to complete our argument and to hamstring efforts to rescue the exotic view of science that we need tentatively to embark upon this new level of microprocessing. (­­p. 168) To kindle the faint promise of insight phenomenology as a marker and cause of creative ideation, the research programme proceeds primarily with task environments that do not 134

Insight in the Kinenoetic Field

involve participants interacting with a physical model of the problem. This methodological decision is driven by an ontological commitment to an internal locus of creativity (­namely, in the head), as well as the search for a single cause, a singular moment in space and time that explains the discontinuity from not knowing to knowing. But this discontinuity is not a salto mortale, the unbridgeable gap between routine and ­non-​­routine cognition (­Ohlsson, 2018), and it is not a mysterious leap. The distributed and protracted nature of c­ reativity—​­which in turn problematizes notions of agency and ­intentionality—​­is apparent in qualitative studies of artists, scientists, and designers (­Glăveanu et al., 2013; Ross & V ­ allée-​­Tourangeau, 2021; ­Vallée-​­Tourangeau & March, 2019; the gradual, protracted nature of scientific discovery is well illustrated in Woolgar, 1988) and calls for an entirely different explanatory model of creativity, not a reductionist one that leaps upon a single, distinct, linear cause, but a more historically and materially embedded, systemic and descriptive model of creativity. Changes in the world that are brought about through actions on objects trigger new perceptions and afford actions that pave a creative ideation trajectory. This trajectory is a route without a road: the route to the solution is not traced in advance; it is not a route to be discovered; it is a route that is enacted. The theoretical effort needed to make sense of this interactivity ought to warn us of the dangers of falling prey to a default Cartesianism that so easily seems to fit folk and subjective notions of thinking and agency (­apparent in the retrospective narrative of the trajectory of an idea; “­one day so and so had an idea”, see Latour & Woolgar, 1986, ­pp. ­169–​­171). The solution is twofold. First, we must not lose focus on the transactional nature of the interaction: agent and object are ­co-​­constituted through interaction. This encourages a shift from an implicitly dualist programme of research to one that casts ­agent-​­object as a system, a dynamic and contingent one. Second, the nature of the explanation of creative ideation should be unreservedly and unashamedly descriptive and historical without diminishing the scientific rigour of the account. As Sutton (­2010, p­ . 214) writes, “­we shouldn’t work with an overly restricted or puritanical notion of scientific explanation: ­non-​­predictive narrative explanations are common enough in the natural and social sciences of many complex systems, including branches of history, geography, geology, evolutionary biology, and meteorology”. Stengers (­1997, p­ . 171) speaks of a Darwinian narrative, not simply applied to the evolution of organisms but to a historical narrative that explores “­the diversity of causes and the diversity of ways in which the same cause can cause”.3 The detailed mapping of the kinenoetic field is a labour of granular description of actions and changes in the world. The kinenoetic field is not an irresistible cornucopia of good ideas: it maps out a trajectory, the paths that lead to solution as well as the unproductive directions along which the system evolves to failure. In other words, this granular description treats successes and failures symmetrically, revealing equally how participants’ actions reinforce unproductive configurations as well as how they can reveal more productive paths to solutions. The animated video of a participant working on the triangle of coins presented earlier illustrates that the participant at some point abandons migrating coins north and starts migrating coins south, which will eventuate in constructing a wider base along row 7, which is the key to the solution. It is tempting to identify this as a pivotal moment in the trajectory. Why did the participant choose this new strategy? (­Note how the language is imbibed in agency and mental plans: the words “­choose” and “­strategy” make the resistance to an agentic and dualist account of creative ideation difficult even if the “­why” is not clear and we have no way of knowing the exact reason for the shift in action.) Hindsight makes the question seductively tempting. The participant solved the problem, and this new pattern of coin migration invites the attribution of the success to this change in migration pattern. But this question is not helpful and injects some life into a dormant dualism. We know that the participant solved the 135

Frédéric ­Vallée-­Tourangeau

problem, and it is tempting to seek a key pivotal switch in thinking and action that sent the participant along the “­r ight” path. It’s an abductive thought reflex; the best explanation for the success of the participant is attributing foresight (­cf. Ingold, 2010). The inversion creeps in at this stage4: knowing that the participant has solved the problem, we seek to retrospectively coronate an event pivot, endowing it with causal power that it does not have. We must resist this way of analyzing creativity. A gold medallist in the 1­ 00-​­metre freestyle might have loved swimming at the age of 4, but her childhood enthusiasm and precocious talent do not explain her achievement: no single cause can be identified other than the careful description of the training and coaching unfolding in the thousands of hours of deliberate practice.5 The creative trajectory, and in the case of insight p­ roblem-​­solving, the trajectory to the normative solution, is constructed with contingent microsteps. The dual pull of looking for a cause and looking for it inside the agent’s head will forever blind researchers of creative ideation to its developmental and systemic trajectory and validate an unproductive form of reductionism. Latour (­2016) contrasts two types of reductionism. The first was alluded to earlier: the felicitous type of reductionism attempts to explain creativity from a set of actions out of which it emerges. It naturally encourages a descriptive ethnographic and historical account of the interaction with objects (“­it makes you friends of interpretable objects”; Latour, 2016, ­p. 96), mapping in granular details, actions, and object transformations in time and space. You can account for creativity in this manner by illustrating its distributed and protracted nature. The second type of reductionism prevents such accountability and curtails the detailed ethnographic mapping of the unfolding of creativity by steamrolling over “­the list of actions of which it is no longer a summary but now the source” (­­p. 96). At play here is a deceitful inversion: the end result is cast as the cause of the process that brought it to light. From Latour’s lectern, this form of reductionism is “­the bane of science” (­­p. 96).

Concluding Remarks ­ roblem-​­solving researchers who enforce and validate the limits of creativity research by P working exclusively on s­econd-​­order ­problem-​­solving ­tasks—​­tasks that can only be solved ­mentally—​­end up conjecturing exactly what they are looking for, namely that new thoughts are evinced through mental processes. Such a research strategy, based as it is upon unobservable mental shifts, is at risk of producing explanations that are analytically empty. ­First-​­order ­problem-​­solving proceeds with and through the world. Since the world is there to see and touch, things can be physically transformed and perceptually seized, the representation cost and burden are alleviated considerably. The restructuring so fondly postulated as an explanation of representational change can proceed in the world rather than (­exclusively) in the head (­­Vallée-​­Tourangeau, 2014). A ­fi rst-​­order ­problem-​­solving procedure opens the field of research by problematizing agency and instrumentality. It admits a broader class of external events that contribute to the changes in the objects, including microprocesses of a serendipitous nature (­Ross & ­Vallée-​­Tourangeau, 2021). These events do not undermine a “­psychological” account of creativity as some fear (­see Chuderski et al., 2021); they expand it and, in the process, align it more closely with the practices of artists and scientists (­Glăveanu et al., 2013). A ­fi rst-​­order task environment such as the laboratory procedure described earlier with the triangle of coins offers the opportunity to chart the transformation of an object that reifies a model; in insight ­problem-​­solving, that object is a proto model of the solution. The research procedure maps, in space and time, the transformation of that object. The mapping efforts create new inscription devices that expose these changes (­a s illustrated in 136

Insight in the Kinenoetic Field

the animated trajectory to the solution presented earlier), and these devices can be used as a communication tool with other members of the research community. The complexity of creativity is mobilized in the laboratory, and its physical traces are in turn reified through these inscription devices. The physical traces can be interrogated and analyzed, helping the researcher see the unproductive explorations as well as those that potentially pave the way to discovery. The gauntlet of the unobservables is replaced with the patient qualitative description of observables. The account that is formulated from this qualitative analysis runs the risk of being misunderstood as a reversal of agency, with problem solvers being turned into puppets at the mercy of an external environmental puppeteer, namely things in the world, but this is a false fear born out of d­ eep-​­rooted Cartesianism. It is more productive to cast agency and intentionality as extended and distributed, as emerging features of the ­agent-​­world coupling that configure a system of resources, thoughts, and affordances. Researchers employing ­fi rst-​­order task environments can bootstrap their explanation of creativity out of the tautological trap that closes on accounts formulated in terms of intelligence or individual differences measuring creativity. It is not simply that the loan on intelligence is repaid; no mortgage application is submitted in the first place. The kinenoetic field can only be mapped using an interactive ­problem-​­solving procedure; a task environment that allows and encourages an engagement with a physical model of the solution. In addition, the procedure must be instrumentalized to allow the recording of the spatial and temporal features of these transformations. Researchers must invest time to produce a granular qualitative description of the participants’ actions and the resulting changes to the physical model of the problem. These efforts map out the kinenoetic field from which a new idea emerged, unveiling the nature and cost of the transaction that purchased a new idea. It is the dynamic, ­co-​­constitutive coupling of agent and world that evinces a new idea. Author’s Note: I thank Lambros Malafouris, Paul March, Wendy Ross, and Gaëlle ­Vallée-​ ­Tourangeau for their comments on a previous version of this chapter.

Notes 1 “­A big brown cow is lying down in the middle of a country road. The streetlights are not on, the moon is not out, and the skies are heavily clouded. A truck is driving towards the cow at full speed, its headlights off. Yet the driver sees the cow from afar easily, and avoids hitting it, without even having to brake hard. How is that possible?” (­Bar Hillel et al., 2018, p­ . 112). Answer: it’s daytime. 2 In the experiments with the triangle of coins reported in V ­ allée-​­Tourangeau et al. (­2021) s­elf-​ ­report measures along four dimensions (­happiness, suddenness, relief and certainty) were recorded from participants who announced a correct solution. Ratings along the suddenness dimension were significantly lower than the other dimensions. There was little evidence that the successful participants experienced a sudden shift in comprehension. 3 Stengers (­1997, ­pp. ­171–​­172) continues: “­Darwinian authors are thus neither judges, poets, nor prophets, because the history of life as they have learned to read it does not authorize any principle of economy, that is, it does not permit the invention of the relationship of forces that would allow an object to be created and judged, or a hierarchy of questions to be established”. 4 de Vries (­2016, ­p. 34) sums up one of the main lessons of Latour and Woolgar’s (­1986) Laboratory life: Once all modalities in qualifying a statement about the world are dropped “(…) a fact has been established. It is accepted that ‘­a mammal injected with a substance with chemical structure XYZ will show an enhanced form of ABC behaviour’. At this point, an important inversion has taken place (…) history will be rewritten by inverting process and outcome. From this point on, the process will be narrated as the search for this particular outcome: this is the fact that the scientists have been looking for. It took quite some effort before it was discovered, but by now its existence has

137

Frédéric ­Vallée-­Tourangeau been revealed. From now on, the history is perceived from this vantage point: the process of scientific discovery is turned into the pursuit of a single path which inevitably led to the discovery of the structure of this substance (…)” (­emphasis added). 5 Sir Dave Brailsford explained the extraordinary success of British cycling in terms of the cumulative effects of small changes to features of the entire system: “­By experimenting in a wind tunnel, we searched for small improvements to aerodynamics. By analyzing the mechanics area in the team truck, we discovered that dust was accumulating on the door, undermining bike maintenance. So we painted the door white, in order to spot any impurities. We hired a surgeon to teach our athletes about proper h ­ and-​­washing so as to avoid illnesses during competition (­we also decided not to shake any hands during the Olympics). We were precise about food preparation. We brought our own mattresses and pillows so our athletes could sleep in the same posture every night. We searched for small improvements everywhere and found countless opportunities. Taken together, we felt they gave us a competitive advantage”. From Harrel, E. (­2015). How 1% performance improvements led to Olympic gold. Harvard Business Review, October 2015. http://­thebusinessleadership.academy/­­w p-​­content/­uploads/­2017/­03/­­How-­​­­1-­​­­Performance-­​ ­­Improvement-­​­­L ed-­​­­to- ­​­­Olympic- ​­Gold.pdf.

References ­Bar-​­Hillel, M., Noah, T.,  & Frederick, S. (­2018). Learning psychology from riddles: The case of stumpers. Judgment and Decision Making, 13(­1), ­112–​­122. Bilalić, M., Graf, M., Vaci, N., & Danek, A. H. (­2021). The temporal dynamics of insight problem ­solving – ​­restructuring might not always be sudden. Thinking & Reasoning, 27(­1), 1­ –​­37. https://­doi. org/­10.1080/­13546783.2019.1705912 Brannigan, A. (­1981). The social basis of scientific discoveries. Cambridge University Press. Chuderski, A., & Jastrzębski, J. (­2018). Much ado about Aha!: Insight problem solving is strongly related to working memory capacity and reasoning ability. Journal of Experimental Psychology: General, 147(­2), ­257–​­281. https://­doi.org/­10.1037/­xge0000378 Chuderski, A., Jastrzębski, J., & Kucwaj, H. (­2021). How physical interaction with insight problems affects solution rates, hint use, and cognitive load. British Journal of Psychology, 112(­1), ­120–​­143. https://­doi.org/­10.1111/­bjop.12442 Danek, A. H. (­2018). Magic tricks, sudden restructuring and the Aha! Experience. In F. ­Vallée-​ ­Tourangeau (­Ed.), Insight: On the origin of new ideas (­­pp. ­51–​­79). Routledge. Danek, A. H., Williams, J., & Wiley, J. (­2020). Closing the gap: Connecting sudden representation change to the subjective Aha! Experience in insightful problem solving. Psychological Research, 84, ­111–​­119. https://­doi.org/­10.1007/­­s00426- ­​­­018- ­​­­0977-​­8 de Souza, L. C., Guimaraes, H. C., Teixeira, A. L., Caramelli, P., & Levy, R. et al. (­2014). Frontal lobe neurology and the creative mind. Frontiers in Psychology, 5, 00761. https://­doi.org/­10.3389/ ­f psyg.2014.00761 de Vries, G. (­2016). Bruno Latour. Wiley. Dietrich, A. (­2019). Types of creativity. Psychonomic Bulletin  & Review, 26(­1), ­1–​­12. https://­doi.org/­ 10.3758/­­s13423-­​­­018-­​­­1517-​­7 Fleck, J. I., & Weisberg, R. W. (­2013). Insight versus analysis: Evidence for diverse methods in problem solving. Journal of Cognitive Psychology, 25(­4), ­436–​­463. https://­doi.org/­10.1080/­20445911.2013.779248 Gilhooly, K., & Webb, M. E. (­2018). Working memory and insight problem solving. In F. ­Vallée-​ ­Tourangeau (­Ed.), Insight: On the origins of new ideas (­­pp. ­105–​­119). Routledge. Glăveanu, V. P., Lubart, T., Bonnardel, N., Botella, M., & Biaisi, P.-​­M. A. et al. (­2013). Creativity as action: Findings from five creative domains. Frontiers in Psychology, 4, 176. https://­doi.org/­10.3389/ ­f psyg.2013.00176 Harrel, E. (­2015). How 1% performance improvements led to Olympic gold. Harvard Business Review, October  2015.  http://­t hebusinessleadership.academy/­­w p- ​­c ontent/­u ploads/­2 017/­0 3/­­H ow-­​­­ 1-­​ ­­Performance-­​­­Improvement-­​­­L ed-­​­­to- ­​­­Olympic- ​­Gold.pdf Ingold, T. (­2010). The textility of making. Cambridge Journal of Economics, 34, 9­ 1–​­102. https://­doi.org/ ­10.1093/­cje/­bep042 Kirsh, D. (­2010). Thinking with external representations. AI  & Society, 25, ­4 41–​­454. https://­doi. org/­10.1007/­­s00146-­​­­010-­​­­0272-​­8

138

Insight in the Kinenoetic Field Latour, B. (­2016). How better to register the agency of things (­Part 1, Semiotics). Tanner Lectures, Yale, March 2014, published in M. Matheson (­Ed.), The Tanner lectures on human values, volume 34 (­­pp. ­81–​­99). The University of Utah Press. Latour, B., & Woolgar, S. (­1986). Laboratory life. The construction of scientific facts (­second edition). Princeton University Press. Malafouris, L. (­2010a). Knapping intentions and the marks of the mental. In L. Malafouris & C. Renfrew (­Eds.), The cognitive life of things: Recasting the boundaries of the mind (­­pp. ­13–​­22). Cambridge: McDonald Institute Monographs. Malafouris’s (­2010b). Metaplasticity and the human becoming: Principles of neuroarchaeology. Journal of Anthropological Sciences, 88, 4­ 9–​­72. Malafouris, L. (­2013). How things shape the mind: A theory of material engagement. MIT Press. Malafouris, L. (­2020). Thinking as ‘­thinging’: Psychology with things. Current Directions in Psychological Science, 29, ­3 –​­8. https://­doi.org/­10.1177/­0963721419873349 Ohlsson, S. (­1984). Restructuring revisited II: An information processing theory of restructuring and insight. Scandinavian Journal of Psychology, 25, ­117–​­129. https://­doi.org/­10.1111/­j.­1467-​­9450.1984. tb01005.x Ohlsson, S. (­1992). ­Information-​­processing explanations of insight and related phenomena. In M. T. Keane & K. J. Gilhooly (­Eds.), Advances in the psychology of thinking (­Vol. 1, ­pp. ­1–​­44). ­Harvester-​­W heatsheaf. Ohlsson, S. (­2018). The dialectic between routine and creative cognition. In F. ­Vallée-​­Tourangeau (­Ed.), Insight: on the origins of new ideas (­­pp. ­8 –​­27). Abingdon: Routledge. Orquin, J. L., & Holmqvist, K. (­2018). Threats to the validity of ­eye-​­movement research in psychology. Behavior Research Methods, 50(­4), ­1645–​­1656. https://­doi.org/­10.3758/­­s13428- ­​­­017- ­​­­0998-​­z Ross, W. (­2021). On the trail of a thought: A kinenoetic analysis of problem solving. Unpublished PhD Dissertation. Kingston University. ­ allée-​­Tourangeau, F. (­2020). Microserendipity in the creative process. Journal of Creative Ross, W., & V Behavior, 55(­3), ­661–​­672. https://­doi.org/­10.1002/­jocb.478 ­ allee-​­Tourangeau, F. (­2021). Kinenoetic analysis: Unveiling the material traces of inRoss, W., & V sight. Methods in Psychology, 5, 100069. https://­doi.org/­10.1016/­j.metip.2021.100069 Stengers, I. (­1997). Power and invention. University of Minnesota Press. Sutton, J. (­2010). Exograms and interdisciplinarity: History, the extended mind, and the civilizing process. In R. Menary (­Ed.), The extended mind (­­pp. ­189–​­225). MIT Press. ­Vallée-​­Tourangeau, F. (­2014). Insight, materiality and interactivity. Pragmatics & Cognition, 22(­1), ­27–​ ­4 4. https://­doi.org/­10.1075/­pc.22.1.02val ­Vallée-​­Tourangeau, F., & March, P. L. (­2019). Insight out: Making creativity visible. Journal of Creative Behavior, 54(­4), ­824– ​­842. https://­doi.org/­10.1002/­jocb.409 ­Vallée-​­Tourangeau, F., Ross, W., Ruffatto Rech, R., & ­Vallée-​­Tourangeau, G. (­2021). Insight as discovery. Journal of Cognitive Psychology, 33(­­6 –​­7 ), ­718–​­737. https://­doi.org/­10.1080/­20445911.2020.1822367 ­Vallée-​­Tourangeau, F., & ­Vallée-​­Tourangeau, G. (­2020). Mapping systemic resources in problem solving. New Ideas in Psychology, 59, 100812. https://­doi.org/­10.1016/­j.newideapsych.2020.100812 Weisberg, R. W. (­1995). Prolegomena to theories of insight in problem solving: A taxonomy of problems. In R. J. Sternberg, & J. E. Davidson (­Eds.), The nature of insight (­­pp. ­157–​­196). MIT Press. Weisberg, R. W. (­2018). Insight, problem solving, and creativity: An integration of findings. In F. V ­ allée-​­Tourangeau (­Ed.), Insight: On the origins of new ideas (­­pp. ­191–​­215). Routledge. Woolgar, S. (­1988). Science: The very idea. Ellis Horwood Tavistock Publications. Wynn. T. Overmann, K. A., & Malafouris, L. (­2021). 4E cognition in the lower paleolithic. Adaptive Behavior, 29(­2), 9­ 9–​­106. https://­doi.org/­10.1177/­1059712320967184

139

PART II

Reflections on the Nature of Creative Cognition as Revealed through Traditional Methodologies

10 IDEA GENERATION AND ASSOCIATIVE MEMORY Richard W. Hass

Divergent thinking tasks are heavily used as a proxy for creative thinking in a variety of behavioral (­Snyder et al., 2019) and neuroscientific (­Benedek et al., 2019) studies. At their heart, divergent thinking tasks can be thought of as a ­face-​­valid procedure for measuring creative thinking. All divergent thinking tasks share the common feature that participants are to generate as many, possibly creative, responses, ideas, questions, answers, etc., to a particular prompt. Though the tasks should not be considered synonymous with creativity, nearly all theories of creativity include an idea generation phase as critical to the process (­for a historical account, see Sawyer, 2011). Though there are many different versions of divergent thinking tasks, the most popular is undoubtedly the Alternative Uses Task (­AUT), developed independently by several researchers in the 1950s. For those who are unfamiliar with such tasks, imagine that you are seated at a table and I ask you to tell me all of the different ways that you could use the common household fork that I put down in front of you on the table. This is the essence of the AUT. When used as a proxy for the idea generation component of the creative process in research, the focus has been on scoring the tasks in psychometric fashion. The various methods for scoring are explained and critiqued elsewhere (­­Reiter-​­Palmon et al., 2019), but they generally consist of separate scores for fluency (­the total number of ideas generated in some unit time) and originality. The variation in originality scoring across studies is substantial and has been the subject of controversy (­Silvia et al., 2008). Despite the issues with scoring, there is now considerable interest in cognitive studies of creative idea generation using the AUT as a proxy task. This chapter describes a variant of these studies designed not to explain how original responses are generated during the AUT and related tasks, but instead to ask whether or not processing during the AUT is related to semantic memory search and retrieval. It has been argued that the AUT provides the best paradigm for understanding how memory search supports creative thinking (­H ass, 2017a, 2017b). Though there have been rapid advances in understanding of the brain networks involved when participants complete AUTs, establishing a formal account of task processing has been slower. That is, while neuroscientific and psychometric studies have illustrated that certain brain networks and broad cognitive abilities are associated with originality scores from AUTs,

DOI: 10.4324/9781003009351-12

143

Richard W. Hass

it is still unclear how ideas are generated in the first place. The goal of this chapter is to review the connections between formal theories of associative (­semantic) memory search and empirical results of behavioral studies that use the AUT as a proxy for creative idea generation. However, theories of associative memory search are not the only formal frameworks that could support a formal theory of the AUT. Significant progress has been made on accounting for the generation of AUT responses by Olteteanu et al. (­e.g., 2016, 2019), who constructed an integrated cognitive architecture that can reproduce a number of results of empirical interest. However, a formal account of the search ­process—​­how ­ utputted—​­is still lacking. Therefore, the focus of this ideas are activated, selected, and o chapter will be on how theories of associative memory search can inform further research on divergent thinking and idea generation, more broadly defined. Indeed, it is likely that such search algorithms can be folded into other, more comprehensive theories, such as those of Olteteanu and Falomir (­2 016), and support work in neurosciences and creativity training. The chapter is organized into three parts. First, a brief overview of the details of representative theories of associative memory is given. Formal relations between the associative links within semantic memory and the process of retrieval are outlined. Second, the relevance for this work for studies of the ­AUT—​­and divergent thinking more ­broadly—​­is explained. Finally, an example of strengths and weaknesses of research using these theories is given with an extended critique of the method to aid in future work.

Theories of Associative Memory and Their Relation to the Alternative Uses Task The theoretical construct of association, inherited from philosophical studies of knowledge, has played a vital historical role in the study of human memory (­for a review see Kahana, 2012, ­Chapter 1). As a motivating example of the importance of association, imagine that you go to the market every Saturday for the same set of groceries (­e.g., milk, butter, bread, etc.). You will likely observe that, over time, those words (­a nd even the images of the products themselves) will become associated within the mental representation of either the context of a supermarket or even in other contexts in which one of the items is present (­cueing butter with bread). This is the essence of associative memory. Process models of associative memory seek to formally explain the mechanisms behind both recognition and recall tasks common in memory research. However, models that account for phenomena associated with f­ ree-​­recall from semantic categories are far more relevant to idea generation research. ­Free-​­recall from semantic categories consists of instructing a participant to simply name all of the members of that category that they know in some amount of time. Indeed, early research by Christensen et al. (­1957) explicitly compared the rate of output of participants during an AUT (­brick was the prompt) to ­free-​­recall of U.S. cities, birds, and quadruped mammals. Due to the constraints of measurement of response time in early research, the rates were not captured with the same granularity as today, leading Christensen et al. to conclude that the response rates were different for creative responses compared to other tasks. That conclusion has been challenged in recent work (­Hass, 2017a; Hass  & Beaty, 2018), showing that the rate of production of responses slows in AUT responding and in responding to various prompts in the consequences task (­Torrance, 1998). The importance of the latter results will be explained next within the larger context of research on the rate of output during the recall of category examples. 144

Idea Generation and Associative Memory

Rate of ­Free-​­Recall of Category Examples In a seminal 1944 study, Bousfield and Sedgwick asked participants to name members of several categories and tracked output over time for each to reveal a relationship between elapsed time and the speed of output. Often participants were asked to continue to name category exemplars for 20 minutes or more so that the ­long-​­term dynamics of output could be studied. There were two phenomena described by Bousfield and Sedgwick: the exponential rate of decay of the output rate and the clustering of similar responses close together in time. More rigorous process models of memory search were developed to account for these phenomena, and they all tend to describe a t­wo-​­stage process of: (­1) searching for relevant memory images, clusters, or semantic fields of responses; and (­2) retrieving all items associated with the cluster (­e.g., Raaijmakers & Shiffrin, 1981). As the process continues, the rate of successful retrieval (­i.e., naming animals) slows, and exponential functions with negatively accelerating slopes provide good fits to the cumulative number of responses plotted as a function of time (­for a review, see Wixted & Rohrer, 1994). Theoretically, the slowing of successful retrieval can be evidence of either a depletion of relevant retrieval structures or a stopping process that is sensitive to the number of times a previously retrieved item is reselected (­e.g., Gronlund & Shiffrin, 1986; Rohrer et al., 1995). The question for creativity research is whether this kind of search process governs idea generation during the AUT and similar tasks. As will be shown, it is relatively simple to fit a mathematical function, derived by Bousfield and Sedgwick (­but see Gruenewald & Lockhead, 1980) to r­ esponse-​­time data from AUT experiments and determine the success of the fit. If the function fits, it is plausible that such a ­fi nite-​­space search process operates during the AUT. Such a result provides a process model that can explain why measures of verbal fluency and fluid intelligence are often found to relate to originality scores from AUT tasks (­e.g., Beaty et al., 2014).

Clustering of Responses during F ­ ree-​­Recall As mentioned, Bousfield and Sedgewick (­1944) also observed that participants tended to output responses that were similar to one another close together in time. This phenomenon is known as clustering and has since been investigated in many kinds of recall tasks, including both ­free-​­recall from categories (­e.g., Hills et al., 2012; Rohrer et al., 1995; Troyer et al., 1997) and in ­non-​­ordered recall of learned lists (­e.g., Kahana, 1996; Raijmaakers & Shiffrin, 1981). There are several theories on why clustering is exhibited in ­free-​­recall, and though they differ in the specifics of formal aspects of the theories (­equations, parameters, etc.), they rely on the same ­t wo-​­stage sampling process as described above. Generally, clusters are thought to be reflective of a ­t wo-​­stage sampling process whereby memory images are sampled according to a global cue (­e.g., the general task context), but then i­ntra-​­image sampling also occurs according to “­local” associative links to all retrievable items in the image (­cf. Hills et al., 2012; Polyn et al., 2009; Raaijmakers & Shiffrin, 1981). The formation of associative links is ostensibly the result of learning through experience. Despite the attractiveness of these clustering theories for modeling AUT responses, assessing the degree of clustering among responses in both f­ree-​­recall and AUT is more difficult than fitting a ­response-​­time distribution. This is because most theories that account for clustering require an explicit value for association between concepts, which, in turn, is dependent upon how semantic memory organization is conceived of in the model. The nature of the organization of semantic memory and valid methods for quantifying that organization in human cognition are the subject of open debate. For example, two computational analyses 145

Richard W. Hass

of ­f ree-​­recall from taxonomic categories (­e.g., animals) showed that it is difficult to dissociate the plausibility of a memory search model from the representational structure upon which it operates (­Abbott et al., 2015; Hills et al., 2012). Hills et al. (­2012) modeled semantic memory based on a vector representation of cooccurrence data, and memory search was modeled as analogous to how animals forage for food. An animal searches its environment for patches of food (­say, berries) and then selects bits of food from the patch for some amount of time before switching back to searching for new patches. The critical component of this foraging model is that it asserts that ­patch-​­switching occurs when the rate of responding within a patch drops below an optimal threshold, which is based on the marginal value theorem (­Charnov, 1976). An optimal foraging process is one in which patch switches occur when the rate of payoff within a patch drops below the average rate of payoff between patches (­i.e., the rate at which the cost of switching patches is less than the cost of staying in a patch). In contrast, Abbot et al. (­2015) demonstrated that a network representation of semantic memory and a simple random walk algorithm without ­two-​­stage sampling could produce clustering. This called into question the need for more elaborate accounts of clustering, though the two algorithms are not directly comparable due to the differing semantic structures they operate on ( ­Jones et al., 2015). What makes this example important for modeling idea generation is the careful consideration that must be given to the interconnection between a model for semantic memory structure and a model for search that operates on such structure. A further discussion of the Hills et al. (­2012) study can illustrate the connection. The authors used the search of associate memory (­SAM; Raaijmakers & Shiffrin, 1981) model as the basis for modeling optimal foraging during ­f ree-​­recall of animals. This is a ­t wo-​­stage model wherein the probability of sampling a particular memory image is computed as a ratio of association strengths between the image and any cue with which it might be associated. Formally,

∏ S (Q ,  I ) )=   ∑ ∏ S (Q ,  I ) M



(

Ps I i|Q1,  Q2 ,   …,  QM

wj

j

j =1

N

M

k =1

j =1

i

wj

j



(­10.1)

i

In words, the probability of sampling the image, I, given the joint distribution of Q cues is the product of the retrieval strengths between the image and the cues in the cue set (­each weighted by wj), divided by the sum of the products of all retrieval strengths between cues and images. In earlier versions of SAM models that are not based on foraging theory (­e.g., Gronlund & Shiffrin, 1986; Raijmaakers & Shiffrin, 1981), there are explicit rules for updating the cue set used for each retrieval iteration based on how many times retrieval fails with respect to sampling and recovery of sampled information. Critically, among the cue set are previously output responses, so that existing semantic relations among responses would prime continued sampling of related memory images. In contrast, memory foraging theory (­H ills et al., 2012) posits that switches in the cue set correspond to points at which payoff within an image (­i.e., patch) drop below a specific rate. Regardless of the mechanism used to model search, the success of any model in which sampling probability is a ratio of association strengths is predicated upon knowledge of the full sample space (­or at least a good approximation of it). That is, to model the process of an AUT responding accurately, retrieval strengths must be known, and thus the success of any search algorithm is dependent upon the measure of association used to model semantic relations in the sample space. Without making assumptions, it is currently not clear whether 146

Idea Generation and Associative Memory

responses to AUTs form a coherent category, the size of which might then be approximated with a linguistic corpus (­a s was done by Hills et al., 2012). Existing studies on the AUT have utilized latent semantic analysis (­L SA; Beaty et al., 2014; Hass, 2017a, 2017b; Prabhakaran et al., 2014) as a model for semantic memory organization. For example, Hass (­2017b) attempted to assess whether clustering was present in AUT response sets. LSA (­Landauer & Dumais, 1997) is a method for representing word meaning using cooccurrence data from natural language corpora. Such methods are called v­ ector-​­based representations because the meaning of a word is modeled in terms of a vector of values assigned to the word mapping it to a set of latent dimensions. In such vector representations, vector cosines can be taken as measures of the similarity of the vectors in the space of word meaning. Using these “­cosine similarities”—​­derived from a LSA of the Touchstone Applied Science Associates (­TASA) corpus available at the University of Boulder’s website (­lsa.colorado.edu)—​­Hass showed that clustering was present in AUT responses but that response arrays did not tend to include as many clusters as those reported in studies of f­ ree-​­recall. However, Hass’s results did converge with other results (­e.g., Beaty & Silvia, 2012; Christensen et al., 1957; Hass, 2017a), showing that AUT responses tend to get more “­remote” as the responding progresses. There were limitations to these results, however. Importantly, there were some instances where the LSA cosines did not seem to reflect the true semantic relations between responses. As an example, using the tools at the UC Boulder website, the cosine similarity between “­keep door open” and “­keep door closed” as uses for a brick is 0.97 (­near identical similarity), whereas the cosine similarity between “­keep a door open” and “­doorstop” is 0.09, despite the ­common-​­sense interpretation of both referring to the same use of a brick as a weighty object that can keep doors in place. Despite these limitations, Hass (­2017a, 2017b) chose LSA as the basis for comparing responses because it can handle linguistic units larger than single words (­e.g., Rehder et al., 1998). However, as implied by the lack of coherence of the cosine similarities for what seem to be similar responses, there is significant disagreement about whether LSA constitutes a good model for semantic memory (­e.g., Griffiths et al., 2007). Griffiths et al. showed that a different ­vector-​­based representation, topic models, might provide a better means for modeling the meaning of text, but the analysis was constrained to single words. Steyvers and Tenenbaum (­2005) further showed that network models assembled via g­ raph-​­theoretic relations provide a more reasonable model of language development compared with vector space models. To date, none of these models of semantic memory have been used in research on semantic search.

Executive Control, Fluid Intelligence, and Strategic Search Aside from accounting for clustering and response rate, process models of associative memory usually imply or explicitly include a control process. Indeed, at the very least, there must be some mechanism that terminates search, as Hass (­2015) showed evidence that it is common for participants to terminate responding in about 90 seconds during the AUT if given the freedom to terminate without a time limit. It has been consistently shown that performance on the AUT relates to executive control processes (­for a review see Benedek et al., 2019). For example, Beaty et al. (­2014) showed that the fluency and originality of uses for objects were almost equally well predicted by measures of remote association and associative flexibility, the latter thought to be an index of executive control over lexical association. More recent analyses using network control theory suggest that frontal control networks of the brain are responsible for “­d riving” the brain into connectivity patterns that support creative thinking (­Kenett, Medaglia et al., 2018). Those results are consistent with other functional 147

Richard W. Hass

connectivity analyses that suggest that dynamic connections between the frontal cortex and the default mode network of the brain are crucial for creative thought (­Beaty et al., 2016). In a seminal experiment using verbal protocols and the AUT, Gilhooly and colleagues (­2007) found that participants employed idiosyncratic strategies to produce responses to alternative use prompts that suggested some level of cognitive control over the content of AUT responses. Strategies included repeating the name of the object, querying episodic memory for specific instances of use of the object, and mental deconstruction of the object into its constituent parts (­e.g., removing the shoelaces from shoes). The interplay of strategy and memory search has also been examined using semantic fluency tasks, again illustrating a potential relationship between the two. Unsworth et al. (­2013) showed that many different strategies emerged from the retrospective reporting of participants following semantic fluency tasks and that different recall trajectories resulted from experimental manipulation of strategies. Unsworth (­2016) further argued that working memory capacity functions to aid retrieval strategies at recall. These results are consistent with conclusions drawn by Gronlund and Shiffrin (­1986), who modeled strategic recall using a modified version of the SAM model (­Raaijmakers & Shiffrin, 1981). For example, Unsworth and colleagues (­2013) identified a “­v isualization” strategy such that participants reported visualizing walking around a zoo to identify and recall animals. Unsworth and colleagues then tested whether participants’ retrieval totals would be affected by manipulations of strategy. They found that mandating a visualization strategy led to similar retrieval rates as ­f ree-​­recall, but that constraining participants to use an alphabetic strategy (­naming an animal that begins with “­a,” then an animal that begins with “­b,” etc.) led to lower retrieval rates. So, in addition to modeling memory search based purely on association, it is likely that some executive control process needs to be explicitly invoked to fully account for phenomena associated with the AUT, and creative thinking more broadly.

An Example of a Memory Search Study with the Alternative Uses Task Given the theoretical links between memory search processes and the AUT, it is instructive to more deeply discuss how one might use behavioral data alone to investigate the nature of memory search during creative idea generation. As mentioned, Hass (­2017b) attempted to qualitatively test whether a ­foraging-​­like search process was evident in two AUT tasks: uses for a brick and uses for a bottle. The results of that study suggested that participants’ searches for ­object-​­uses did not leverage local semantic relations but instead seemed to operate on a global level, exhibiting a serial order effect: responses earlier in the task were more semantically related to the prompt. However, in that study, there was no direct comparison between AUT responses and ­free-​­recall from categories, which made firm conclusions hard to draw. Thus, a second experiment (­Hass et  al., 2019) was conducted to directly compare search during AUT responding to search during ­f ree-​­recall of examples of ­so-​­called “­­goal-​­derived” categories. This study is presented now as an example of a simple method for examining search, but also to illustrate the need for further research.

­Goal-​­Derived Categories A key aspect of this study was to propose that search during g­ oal-​­derived category exemplar generation is an apt comparison for understanding o ­ bject-​­uses generation. Alternative uses ­object-​­prompts can be viewed as a kind of ­ill-​­structured, ­goal-​­derived category. One way to define such categories is that the criteria for inclusion in the category are fuzzy or 148

Idea Generation and Associative Memory

ambiguous. For example, Walker and Kintsch (­1985) asked participants to name examples of laundry detergents. Such a prompt may cue retrieval of names of detergents (­e.g., Tide), or it may cue a more “­ad hoc” representation of the category as things that might be useful as a laundry detergent (­e.g., hand soap). That is, ­goal-​­derived categories are constructed during ­goal-​­directed activities (­Barsalou, 1982, 1983), like deciding which chair to sit on in a cafe or which personal belongings to keep from a childhood home. These categories can be distinguished from “­natural” or “­t axonomic” categories in several ways, though two are of critical importance. First, ­goal-​­derived categories are constructed when performing ­decision-​­making tasks; they are defined by personal objectives and constrained by the environment or immediate context. Second, it can be argued that ­goal-​­derived categories such as “­things to take with you in case of a fire” are not as ­well-​­established in memory as categories such as breakfast foods (­Barsalou, 1983). Omelets and pancakes are within the same category (­breakfast foods) because they are edible, made with eggs, served warm, eaten in the morning, and relatively straightforward to cook. Attributes like those, such as times of consumption and ingredients, which are the basis of category discrimination (­Rosch et al., 1976), do not ­co-​­occur as frequently in g­ oal-​­derived categories. More importantly for this analysis, Barsalou (­1982) suggested that retrieval of conceptual information for category processing involves the generation of multiple conceptual representations, each held in working memory, when the category is encountered in normal life. This reliance on multiple conceptual ­ bject-​­uses generation representations could account for the effects reviewed above, relating o to fluid intelligence and executive control. As such, these kinds of category prompts likely serve as a better comparison condition than taxonomic categories.

Overview and Method In this study, a method to identify clusters based on the ­ slope-​­ difference algorithm (­Gruenewald & Lockhead, 1980) was used along with LSA. The ­slope-​­difference algorithm identifies potential semantic clusters in terms of the difference between i­nter-​­response times (­IRT) between items, and the expected IRT given a mathematical relation between response time and output total. LSA cosine similarities were then used to examine whether responses within clusters identified in terms of response rate were indeed semantically similar. This was used across AUT prompts and g­ oal-​­derived category prompts, which all participants responded to. To create a more specific comparison between the two kinds of prompts, the prompt used for ­object-​­uses in the current study was to “­think of common and uncommon uses” of each object. That is, several studies have shown that instructing participants to “­be creative” decreases fluency (­output total) while increasing the average originality of their responses (­Forthmann et al., 2016; Nusbaum et al., 2014). Since the primary interest in the current study was the nature of the search through the category itself (­e.g., use of an object vs. ­goal-​­derived category) and not whether participants were trying to engage in creative thought, the special instruction was warranted.

Participants A total of 32 participants were recruited from undergraduate psychology courses. Participants were offered extra credit or chocolate in compensation for their time. Participants ranged in age from 18 to 25 years old, and the demographics were consistent with those of a traditional undergraduate university in the northeastern United States. All recruitment and consent procedures were approved by the university’s Institutional Review Board. 149

Richard W. Hass

Materials Participants completed the tasks using a custom MATLAB interface on an Apple iMac. Instructions and prompts appeared as text on white background above a textbox where participants entered responses. Instructions were displayed and read to participants prior to each of the three task blocks, the first of which was a practice block (­naming colors). Instructions were not visible during response generation, but the prompt remained displayed for the entire duration of each response generation interval. Demographic information was obtained using a ­pencil-­​­­and-​­paper survey after the experiment finished.

Procedure The tasks were presented in two blocks of three prompts each, with a break in between each block. Both the order of the blocks and the order of presentation of prompts within the blocks were randomized by MATLAB code. All participants responded to each prompt in each block. Each response interval was three minutes in length. The ­goal-​­derived category prompts began with “­name examples of ” and ended with one of the three prompts: things to spend lottery winnings on, things to take from your house if it were caught on fire, and things to sell at a garage sale. The ­object-​­use prompts began with “­name common and uncommon uses for a” and ended with one of the three prompt objects: a brick, a hammer, or a picture frame. The entire prompt phrase remained on the screen above the ­text-​­entry box for the entire three minutes.

Analyses and Results1 Response times were defined in terms of the time (­since presentation of the prompt) of the first keypress of each response to be consistent with studies using ­voice-​­keyed response recording. Prior to analysis, data were examined for repeated responses and malfunctions in MATLAB’s execution of the experiment. Data from three participants were excluded due to MATLAB malfunctions, reducing the final sample size to 29. Repeated responses were those that were identified as the same response given more than once by the same individual to a specific prompt. When repeats were identified, the RTs for those responses were removed from the data set. Participants gave a total of 1746 responses to the three g­ oal-​­derived prompts after the removal of 23 repeated responses. Finally, participants gave a total of 1012 responses to the ­object-​­use prompts after the removal of 11 responses.

Fluency and Cumulative Response Times One method for determining whether exponential search is evident in responding is to plot the mean number of cumulative responses, per prompt, per ­ten-​­second experimental block. This effectively estimates the rate at which participants respond and can provide clues to the process of responding. Specifically, according to the theories described above, the rate of responding should decay exponentially as participants continually ­re-​­sample ideas. ­Figure 10.1 illustrates that this tends to be true for AUT prompts, but not for g­ oal-​­derived prompts. However, it is also true that three minutes may not be enough time for responding to reach asymptote, especially for ­goal-​­derived categories. Prior studies with s­elf-​­terminating response intervals suggest that people tend to ­self-​­terminate AUT responding after about 90 seconds if not required to continue (­Hass, 2015). 150

Idea Generation and Associative Memory

­Figure 10.1  Mean cumulative response total per t­en-​­second block for all six prompts. Lines are local polynomial regression lines as approximations to exponential functions Note: GD = Goal Derived prompts, OU = Object Uses prompts.

Clustering and the ­Slope-​­Difference Algorithm Clusters were identified using a modification of the ­slope-​­d ifference algorithm (­Gruenewald & Lockhead, 1980). In its current form, the s­lope-​­difference algorithm works as follows: first, fit the following function:

(

)

R( t ) = N 1 − e − ( t τ )

(­10.2)

to the data for each participant using nonlinear least squares to obtain estimates of N, the theoretical number of responses the person would output with unlimited time, and τ, the inverse of the exponential rate parameter λ. In this parameterization, τ is the estimated mean response time for that participant. Next, for each IRT in a participant’s response array, compute the predicted IRT using the first derivative of equation 10.2:

N r (t ) =    ∗ e − (t /τ ) τ 

(­10.3)

Then, compute the observed instantaneous rate of change at each response, which is just the reciprocal of each IRT. Finally, subtract the predicted rate of change from the observed rate of change, the slope difference. If the slope difference is relatively large and positive, then 151

Richard W. Hass

­Figure 10.2 The frequency of responses categorized as w ­ ithin-​­cluster or switches by the s­lope-​ ­d ifference algorithm across both tasks

responding is faster than expected given exponential search, and the participant is likely responding in a cluster. If the difference is small or less than zero, responding is slower than predicted, and the participant is searching between clusters. Gruenewald and Lockhead (­1980) supported the validity of the algorithm such that large, positive differences between observed and predicted rates were indications that responding was faster than predicted, and thus, faster than expected responses qualify as being within clusters. For the current study, the threshold for slope differences being categorized as “­switches” was 0.10, which is the same as used by Grunewald and Lockhead. ­Figure  10.2 shows that the proportion of responses identified as within a cluster was ­significantly greater for ­goal-​­derived categories compared with object uses, χ2(­1) = 48.27, p < .001. Of the 998 o ­ bject-​­use responses, 18.8% were identified as w ­ ithin-​­cluster, while 31.3% of the 1567 g­ oal-​­derived responses were identified as w ­ ithin-​­cluster. This result is consistent with Hass (­2017b), who showed that clusters were not usually present in AUT response arrays with a ­t wo-​­minute time limit.

Pairwise Semantic Similarity The validity of the s­ lope-​­difference algorithm rests on further semantic analysis of its results. Here, the central question was whether responses in clusters corresponding to ­goal-​­derived categories were more semantically similar than those in clusters corresponding with ­object-​ ­ se generation. Pairs of sequential responses were analyzed for semantic similarity using the u tools at the UC Boulder website (­lsa.colorado.edu). The General Reading Corpus, with 300 factors, was chosen as the basis for comparisons, and the ­term-­​­­to-​­term comparator was used. Multilevel modeling was used to examine the main effects of clustering (­w ithin cluster vs. between cluster response) and prompt type (­­goal-​­derived vs. ­object-​­use) on ­L SA-​ ­derived cosine similarities and the interaction of the two fixed effects. A random intercept term was added to account for participant variation and another to account for variations across the six prompts. T ­ able 10.1 contains the results of the analysis, including 95% confidence intervals for the fixed and random effects terms. Rather, surprisingly, on average, 152

Idea Generation and Associative Memory ­Table 10.1  Results of the Multilevel Model for Pairwise Semantic Similarity between Successive Responses, with Clusters Identified by the S­ lope-​­Difference Algorithm and Semantic Similarity Defined Using LSA 95% Confidence Interval Fixed effects Intercept ­Prompt- ​­GD ­Cluster-​­between Prompt by cluster Random effects Participant Prompt Residual

­Figure 10.3 

b 0.38 –​­0.12 – ​­0.04 – ​­0.03 σ 0.04 0.03 0.20

t 14.65 –​­3.88 –​­2 .24 –​­1.49

2.5% 0.33 –​­0.18 – ​­0.08 – ​­0.07 2.5% 0.03 0.01 0.19

97.5% 0.42 – ​­0.06 – ​­0.005 0.01 97.5% 0.06 0.05 0.20

L SA cosine similarity values per prompt type for ­w ithin-​­cluster responses and switches

the pairwise similarity of responses to the g­ oal-​­derived prompts was less than the average pairwise similarity of ­object-​­uses responses. However, the results illustrate the validity of the ­slope-​­difference algorithm. In both tasks, ­w ithin-​­cluster responses were more similar than ­between-​­cluster responses. ­Figure  10.3 illustrates that there may be a small interaction between prompt type and clustering, and in ­Table 10.1, the estimate is a slightly smaller difference in the similarity of clustered and ­non-​­clustered responses for object uses compared with g­ oal-​­derived categories, though zero remains a plausible value for the interaction.

Discussion of Results and Limitations of the Method The reason to explicate the results presently was to illustrate how memory search theories can motivate analyses of creative idea generation data and also to discuss limitations of the method as it currently stands. Importantly, the ­slope-​­difference algorithm seemed to have captured 153

Richard W. Hass

actual semantic clusters in addition to purely temporal clusters (­see ­Figure 10.3). In the original paper describing the ­slope-​­difference algorithm, Gruenewald and Lockhead (­1980) provide evidence that the algorithm does, indeed, capture semantic clusters, which they term semantic fields. Though Herrmann and Pearle (­1981) criticize parts of their analysis, the algorithm is plausible in terms of a t­wo-​­stage sampling view of memory search and retrieval. In that way, the use of a cumulative response time distribution and IRTs from participants to derive clusters is theoretically related to memory search models that explicitly model how the search takes place. This modeling of the search process itself is lacking in the s­lope-​­difference algorithm, which represents a distinct limitation of the model. That is, no simulation of the AUT response process was explicitly modeled in this analysis. Again, to do so effectively requires an explicit representation of the search space, which is currently lacking. An unexpected result from the above study was that the average cosine similarity between pairs of successive responses was lower for ­goal-​­derived categories compared to the AUT prompts. One explanation for this result is that semantic relationships across clusters of ­goal-​­derived exemplars may be minimal because of the dependence of semantic similarity on context (­e.g., Barsalou, 1983). Of course, that characterization might also be said of ­object-​­uses. For example, Hass (­2017b) illustrated that ­L SA-​­derived cosine similarities may not accurately represent c­ ontext-​­dependent relationships between object uses. Indeed, the main difference between the two types of prompts is that g­ oal-​­derived prompts identify a ­ bject-​­uses context (­e.g., a garage sale), which is all items must relate to in some way, while o ­ bject-​ prompts identify an exemplar (­e.g., a brick), to which responses must relate. While o ­uses responses will likely have c­ ontext-​­dependency, it is also likely that ­context-​­dependency will be greater among g­ oal-​­derived categories such as those in this study, as the context itself is the main constraint on conceptual activation. Consider, for example, that a participant is currently naming items to take with her in case of a fire. Say a participant activates the context of “­bedroom items” as a concept and outputs two or three responses that fit this context. ­Figure 10.4 shows that responses 10 through 12 illustrate such a set of responses. Here, the participant has started the cluster with “ ­jacket,” and the cluster was preceded by “­money.” The cosine similarity according to the tool at lsa.colorado.edu between the two responses is 0.13. Contrast this with the cosine similarity between jacket and shoes, which is 0.54, and with shoes and “­extra clothing,” which is 0.47. The next transition, extra clothing to books, returns to 0.13. As can be seen, similarity drops substantially each time there is a jump to a new cluster. Since there are more such cluster switches in g­ oal-​­derived response arrays compared with AUT response arrays (­­Figure 10.2), the average pairwise similarity is lower. For comparison, ­Figure 10.5 illustrates the same participant’s brick responses. Here, the clusters are not as well defined, but the final three responses were identified by the ­slope-​ ­d ifference algorithm as being a cluster. All three responses use the word “­m ake” and indeed refer to the property of a brick as being a building material. A cursory scan through the remaining responses illustrates there are fewer explicit jumps between clusters as there are fewer responses overall. Additionally, about half of the responses refer to the use of bricks for making things, while many others correspond to the weightiness of bricks. It is important to note that the conclusions of this study are fully dependent on the validity of LSA for representing similarity between responses in both tasks. Clearly, the average similarity metric converges with the identification of clusters via the s­lope-​­difference algorithm, but ­Figures 10.4 and 10.5 also contain anomalies. First, if the word “­m ake” is removed from the three final responses, the average cosine similarity of the cluster drops from 0.48 to 0.14. It is not clear whether the use of the word “­m ake” was chosen purposefully by the participant, but one could imagine that simply stating “­sidewalk,” “­road,” and “­wall” as 154

Idea Generation and Associative Memory

­Figure 10.4 Example of one participant’s output for the category “­things to take in case of fire.” Dotted line is the best fitting exponential function relating output time to cumulative frequency (­see Eq. 10.1)

­Figure 10.5 

E xample of one participant’s output for the category “uses for a brick.” Dotted line is the best fitting exponential function relating output time to cumulative frequency (see Eq. 10.1)

responses fits the task such that they can represent the use of a brick as each of the three. As a vector representation based on word cooccurrence, when same word appears in two responses to be compared, the results are biased toward similarity. For example, the responses make a sidewalk and make a phone call result in a cosine similarity of 0.63. Obviously, one of these responses is not a use for a brick at all, and yet the word make drives the similarity comparison. Thus, a precise model for semantic organization in the AUT remains to be constructed. However, these results are indicative of a t­ wo-​­stage sampling process of search and retrieval during AUT responding. 155

Richard W. Hass

Conclusions and Future Directions The purpose of this chapter was to review the links between theories of memory search and the phenomena observable during responding to the AUT. The AUT was also discussed as a representative proxy for creative idea generation, such that while it is a simple version of idea generation, it clearly relates to a kind of thought process that might be general to creativity (­Olteteanu  & Falomir, 2016). Though much has been done to illustrate the relationships between formal theories of associative memory search and divergent thinking processes, important questions remain unanswered. Perhaps the most important is that divergent thinking tasks differ in the complexity of the categories searched and also potentially the strategies and constraints used in search. Herrmann and Pearle (­1981) rightfully stated: …It has also been known that, when memory fails, subjects treat recall as a ­problem-​ ­solving task, i.e., they often invoke a lengthy process of reasoning in order to find items that have yet to be considered. (­­p. 159) Such a rational approach is out of scope of the theories presented here. However, Olteteanu and Falomir (­2 016) outlined a formal process of reasoning for a cognitive system to produce “­a lternative” uses that was based on analogy, perception, and reasoning. Indeed, beyond pure reasoning, there are other mechanisms that might aid in divergent thinking. Most notably are ­so-​­called implicit processes, which more related to insight phenomena than to memory search (­e.g., Helie & Sun, 2010). A growing body of neuroscientific and cognitive studies of remote associates ­problem-​­solving suggests that other cognitive mechanisms beyond pure retrieval, reasoning, and executive function relate to creativity (­e.g., Salvi et al., 2015). Finally, studies using network science have also shown that the potential for remote association, as coded within individuals’ semantic networks, is crucial for creativity defined in a variety of ways (­e.g., Kenett et al., 2014; Kenett, Levy et al., 2018). The latter studies are important in that they connect semantic organization to more than just AUT performance. What remains to be determined is whether or not a true, ­SAM-​­like model of search and retrieval, with a plausible semantic representation, can recreate the phenomena described within this chapter: clustering and exponential rate of decay. It is likely that such a model would need to be endowed with specific control structures that account for switching and strategy and other representation processes that account for conceptual combination and generation. As with models of ­free-​­recall, the success of such an idea generation model is heavily dependent on an accurate representation of the conceptual space through which search operates. Thus, it is imperative that researchers focus on continuing to map out such a space, which could be done by combining research on conceptual learning and representation with research on semantic memory.

Note 1 All analyses were conducted using the R Statistical Programming Language, and all data and algorithms are available for download (­https://­osf.io/­f vne2/).

156

Idea Generation and Associative Memory

References Abbott, J. T., Austerweil, J. L., & Griffiths, T. L. (­2015). Random walks on semantic networks can resemble optimal foraging. Psychological Review, 122, ­558–​­559. Barsalou, L. W. (­1982). ­Context-​­i ndependent and c­ ontext-​­dependent ­i n-​­ formation in concepts. Memory & Cognition, 10, ­82–​­93. Barsalou, L. W. (­1983). Ad hoc categories. Memory & Cognition, 11, ­211–​­227. Beaty, R. E., Benedek, M., Silvia, P. J., & Schacter, D. L. (­2016). Creative cognition and brain network dynamics. Trends in Cognitive Sciences, 20, ­87–​­95. Beaty, R. E., & Silvia, P. J. (­2012). Why do ideas get more creative across time? An executive interpretation of the serial order effect in divergent thinking tasks. Psychology of Aesthetics, Creativity, and the Arts, 6(­4), ­309–​­319. https://­doi.org/­10.1037/­a0029171 Beaty, R. E., Silvia, P. J., Nusbaum, E. C., Jauk, E., & Benedek, M. (­2014). The roles of associative and executive processes in creative cognition. Memory & Cognition, 42(­7 ), ­1186–​­1197. https://­doi. org/­10.3758/­­s13421-­​­­014-­​­­0428-​­8 Benedek, M., Christensen, A. P., Fink, A.,  & Beaty, R. E. (­2019). Creativity assessment in neuroscience research. Psychology of Aesthetics, Creativity, and the Arts, 13(­2), ­218–​­226. https://­doi. org/­10.1037/­aca0000215 Bousfield, W. A. & Sedgewick, H. W. (­1944). An analysis of sequences of restricted associative responses. Journal of General Psychology, 30, ­149–​­165. Charnov, E. L. (­1976). Optimal foraging: The marginal value theorem. Theoretical Population Biology, 9, ­129–​­136. https://­doi.org/­10.1016/­­0 040-​­5809(­76)­­90040-​­X Christensen, P. R., Guilford, J. P., & Wilson, R. C. (­1957). Relations of creative responses to working time and instructions. Journal of Experimental Psychology, 53(­2), 8­ 2–​­88. Forthmann, B., Gerwig, A., Holling, H., Celik, P., Storme, M., & Lubart, T. (­2016). The ­be-​­creative effect in divergent thinking: The interplay of instruction and object frequency. Intelligence, 57, ­25–​­32. Gilhooly, K. J., Fioratou, E., Anthony, S. H., & Wynn, V. (­2007). Divergent thinking: Strategies and executive involvement in generating novel uses for familiar objects. British Journal of Psychology, 98, 6­ 11–​­625. https://­doi.org/­10.1348/­096317907X173421 Griffiths, T. L., Steyvers, M., & Tenenbaum, J. B. (­2007). Topics in semantic representation. Psychological Review, 114(­2), 2­ 11–​­244. https://­doi.org/­10.1037/­­0 033-​­295X.114.2.211 Gronlund, S. D., & Shiffrin, R. M. (­1986). Retrieval strategies in recall of natural categories and categorized lists. Journal of Experimental Psychology: Learning, Memory, and Cognition, 12(­4), ­550–​­561. Gruenewald, P. J., & Lockhead, G. R. (­1980). The free recall of category examples. Journal of Experimental Psychology: Human Learning and Memory, 6(­3), ­2 25–​­240. https://­doi.org/­10.1037/­­0278-​­7393. 6.3.225 Hass, R. W. (­2015). Feasibility of online divergent thinking assessment. Computers in Human Behavior, 46(­C), ­85–​­93. https://­doi.org/­10.1016/­j.chb.2014.12.056 Hass, R. W. (­2017a). Tracking the dynamics of divergent thinking via semantic distance: Analytic methods and theoretical implications. Memory & Cognition, 45(­2), ­233–​­244. Hass, R. W. (­2017b). Semantic search during divergent thinking. Cognition, 166, ­344–​­357. https://­doi. org/­10.1016/­j.cognition.2017.05.039 Hass, R. W., & Beaty, R. E. (­2018). Use or consequences: Probing the cognitive difference between two measures of divergent thinking. Frontiers in Psychology, 9, ­558–​­515. https://­doi.org/­10.3389/ ­f psyg.2018.02327 Hass, R. W., Long, J. C., & Pierce, J. (­2019). Idea generation and ­goal-​­derived categories. In A. K. Goel, C. M. Seifert, & C. Freksa (­Eds.), Proceedings of the ­Forty-​­First Annual Conference of the Cognitive Science Society (­­pp. ­408-​­414). Montreal, QB: Cognitive Science Society. Hélie, S., & Sun, R. (­2010). Incubation, insight, and creative problem solving: A unified theory and a connectionist model. Psychological Review, 117(­3), ­994–​­1024. https://­doi.org/­10.1037/­a0019532 Herrmann, D. J., & Pearle, P. M. (­1981). The proper role of clusters in mathematical models of continuous recall. Journal of Mathematical Psychology, 24(­2), ­139–​­162. https://­doi.org/­10.1016/­­0 022​­2496(­81)­­90040-​­7 Hills, T. T., Jones, M. N., & Todd, P. M. (­2012). Optimal foraging in semantic memory. Psychological Review, 119, ­1–​­10.

157

Richard W. Hass Jones, M. N., Hills, T. T.,  & Todd, P. M. (­2015). Hidden processes in structural representations: A reply to Abbott, Austerweil, and Griffiths. Psychological Review, 122(­3), ­570–​­574. https://­doi. org/­10.1037/­a0039248 Kahana, M. J. (­1996). Associative retrieval processes in free recall. Memory & Cognition, 24(­1), ­103–​­109. Kahana, M. J. (­2012). Foundations of human memory. Oxford: Oxford University Press. Kenett, Y. N., Anaki, D., & Faust, M. (­2014). Investigating the structure of semantic networks in low and high creative persons. Frontiers in Human Neuroscience, 8, 407. Kenett, Y. N., Levy, O., Kenett, D. Y., Stanley, H. E., Faust, M., & Havlin, S. (­2018). Flexibility of thought in high creative individuals represented by percolation analysis. Proceedings of the National Academy of Sciences, 115(­5), ­867– ​­872. Kenett, Y. N., Medaglia, J. D., Beaty, R. E., Chen, Q., Betzel, R. F., ­Thompson-​­Schill, S. L., & Qiu, J. (­2018). Driving the brain towards creativity and intelligence: A network control theory analysis. Neuropsychologia, 118, ­79–​­90. Landauer, T. K., & Dumais, S. T. (­1997). A solution to Plato’s problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological Review, 104(­2), ­211–​­240. Nusbaum, E. C., Silvia, P. J., & Beaty, R. E. (­2014). Ready, set, create: What instructing people to “­Be Creative” reveals about the meaning and mechanisms of divergent thinking. Psychology of Aesthetics, Creativity, and the Arts, 8, ­423–​­432. https://­doi.org/­10.1037/­a0036549 Olteţeanu, A. M.,  & Falomir, Z. (­2016). Object replacement and object composition in a creative cognitive system. Towards a computational solver of the Alternative Uses Test. Cognitive Systems Research, 39, ­15–​­32. Olteteanu, A. M., Schöttner, M., & Bahety, A. (­2019). Towards a ­multi-​­level exploration of human and computational r­ e-​­representation in unified cognitive frameworks. Frontiers in Psychology, 10, 940. Polyn, S. M., Norman, K. A., & Kahana, M. J. (­2009). A context maintenance and retrieval model of organizational processes in free recall. Psychological Review, 116(­1), ­129–​­156. Prabhakaran, R., Green, A. E., & Gray, J. R. (­2014). Thin slices of creativity: Using ­single-​­word utterances to assess creative cognition. Behavior Research Methods, 46(­3), ­641–​­659. Raaijmakers, J. G., & Shiffrin, R. M. (­1981). Search of associative memory. Psychological Review, 88(­2), ­93–​­134. https://­doi.org/­10.1037/­­0 033-​­295x.88.2.93 Rehder, B., Schreiner, M. E., Wolfe, M. B. W., Laham, D., Landauer, T. K., & Kintsch, W. (­1998). Using latent semantic analysis to assess knowledge: Some technical considerations. Discourse Processes, 25(­­2 –​­3), ­337–​­354. ­Reiter-​­Palmon, R., Forthmann, B., & Barbot, B. (­2019). Scoring divergent thinking tests: A review and systematic framework. Psychology of Aesthetics, Creativity, and the Arts, 13(­2), ­144–​­152. Rohrer, D., Wixted, J. T., Salmon, D. P., & Butters, N. (­1995). Retrieval from semantic memory and its implications for Alzheimer’s disease. Journal of Experimental Psychology: Learning, Memory, and Cognition, 21(­5), ­1127–​­1139. Rosch, E., Mervis, C. B., Gray, W. D., Johnson, D. M., & ­Boyes-​­Braem, P. (­1976). Basic objects in natural categories. Cognitive Psychology, 8(­3), ­382–​­439. Salvi, C., Bricolo, E., Franconeri, S. L., Kounios, J., & Beeman, M. (­2015). Sudden insight is associated with shutting out visual inputs. Psychonomic Bulletin & Review, 22, ­1814–​­1819. Sawyer, R. K. (­2011). Explaining creativity: The science of human innovation. Oxford: Oxford University Press. Silvia, P. J., Winterstein, B. P., Willse, J. T., Barona, C. M., Cram, J. T., Hess, K. I., Martinez, J. L., & Richard, C. A. (­2008). Assessing creativity with divergent thinking tasks: Exploring the reliability and validity of new subjective scoring methods. Psychology of Aesthetics, Creativity, and the Arts, 2(­2), ­68–​­85. https://­doi.org/­10.1037/­­1931-​­3896.2.2.68 Snyder, H. T., Hammond, J. A., Grohman, M. G., & ­K atz-​­Buonincontro, J. (­2019). Creativity measurement in undergraduate students from 1­ 984–​­2013: A systematic review. Psychology of Aesthetics, Creativity, and the Arts, 13(­2), ­133–​­143. Steyvers, M., & Tenenbaum, J. B. (­2005). The large-scale structure of semantic networks: Statistical analyses and a model of semantic growth. Cognitive Science, 29(­1), ­41–​­78. Torrance, E. P. (­1998). Torrance tests of creative thinking. Princeton, NJ: Personnel Press. Troyer, A. K., Moscovitch, M., & Winocur, G. (­1997). Clustering and switching as two components of verbal fluency: evidence from younger and older healthy adults. Neuropsychology, 11(­1), ­138–​­146. Unsworth, N. (­2016). Working memory capacity and recall from l­ong-​­term memory: Examining the influences of encoding strategies, study time allocation, search efficiency, and monitoring abilities.

158

Idea Generation and Associative Memory Journal of Experimental Psychology: Learning, Memory, and Cognition, 42(­1), ­50–​­61. https://­doi.org/­ 10.1037/­x lm0000148 Unsworth, N., Brewer, G. A.,  & Spillers, G. J. (­2013). Strategic search from l­ong-​­term memory: An examination of semantic and autobiographical recall. Memory, 22(­6), ­687–​­699. https://­doi. org/­10.1080/­09658211.2013.812736 Walker, W. H., & Kintsch, W. (­1985). Automatic and strategic aspects of knowledge retrieval. Cognitive Science, 9(­2), ­261–​­283. https://­doi.org/­10.1016/­­S0364-​­0213(­85)­­80016-​­1 Wixted, J. T., & Rohrer, D. (­1994). Analyzing the dynamics of free recall: An integrative review of the empirical literature. Psychonomic Bulletin & Review, 1(­1), ­89–​­106. https://­doi.org/­10.3758/­bf03200763

159

11 CREATIVELY SEARCHING THROUGH SEMANTIC MEMORY STRUCTURE A Short Integrative Review Yoed N. Kenett

Introduction Among his profoundly impactful work, William James also proposed a theory on the minds of highly creative and genius individuals, which he termed people with great minds (­Simonton, 2018). In his work titled “­Great men, great thoughts, and the environment,” James writes the following ( ­James, 1880; ­p. 456): Instead of thoughts of concrete things patiently following one another in a beaten track of habitual suggestion, we have the most abrupt ­cross-​­cuts and transitions from one idea to another, the most rarefied abstractions and discriminations, the most unheard of combination of elements, the subtlest associations of analogy; in a word, we seem suddenly introduced into a seething cauldron of ideas, where everything is fizzling and bobbling about a state of bewildering activity, where partnerships can be joined or loosened in an instant, treadmill routine is unknown, and the unexpected seems only law. In this passage, James describes the thought process of great minds not as a linear, sequential process but rather as a chaotic, associative process that jumps from one idea to the next. Despite recent advancements in creativity research (­Abraham  & Bubic, 2015; Benedek  & Fink, 2019; Runco  & Jaeger, 2012), we still know very little about such creative search processes through one’s mind. Recently developed computational tools in linguistics and network science are paving the way to directly study such creative search processes as they operate over semantic memory. This is the focus of the current chapter. The aim of this chapter is to highlight the role of semantic memory in individual differences in creative search processes. Semantic memory is the cognitive system of human memory that stores concepts and facts (­e.g., the Earth is round; Jones et al., 2015; Kumar, 2021). The role of semantic memory in creativity is intuitively embedded in a theory of creativity through the notion that the farther one moves from a concept in a semantic memory space, the more novel or creative the new concept will be (­Kenett, 2018b; Mednick, 1962; Volle, 

160

DOI: 10.4324/9781003009351-13

Creatively Searching Through Semantic Memory Structure

2018). For example, a key feature of divergent thinking (­DT) tests, which are widely used in creativity research to assess an individual’s ability to generate original ideas in response to ­open-​­ended cues, involves moving away from conventional to more distant, weakly related responses (­Acar & Runco, 2019; Runco & Acar, 2012). First, this chapter will review work applying computational tools to study how differences in semantic memory structure relate to individual differences in creative thinking. Next, the way in which semantic memory structure facilitates search processes that relate to creative thinking will be discussed. Finally, the effect of expertise on semantic memory structure and search processes will be examined in relation to creativity. Overall, this chapter aims to advance the theory that creativity involves search processes that are constrained by the underlying semantic memory structure.

Theory Creativity theories have long emphasized the role of semantic memory in creative thinking (­Abraham, 2014; Abraham & Bubic, 2015; Kenett, 2018b; Kenett & Faust, 2019; Mednick, 1962; Sowden et al., 2014). The main theory that relates creative thinking to semantic memory structure is the associative theory of creativity (­Mednick, 1962). This theory proposes that creativity involves the connection of weakly related or remote concepts into novel and applicable concepts. The farther apart the concepts are, the more creative the new combination will be. Importantly, this theory highlights how the structure of a person’s memory constrains their search processes operating over it, and how such an interaction relates to individual differences in creative thinking. According to this theory (­Mednick, 1962), creative individuals are characterized by ‘‘­flat” (­more and broader associations to a given concept) instead of “­steep” associational hierarchies (­few, common associations to a given concept; but see Benedek & Neubauer, 2013). On this view, creative individuals may have more associative links between concepts in their semantic memory and can connect associative relations faster than less creative individuals, thereby facilitating more efficient search processes (­Kenett, 2018b; Kenett & Austerweil, 2016; Volle, 2018). Thus, when attempting to think creatively, a less creative individual is likely to become “­stuck” on these dominant, common, associations, whereas a more creative individual can overcome them and establish more distant associations via spreading activation (­Gray et al., 2019; Kenett & Austerweil, 2016). Within this framework, semantically “­close” concepts are considered less likely to be creative, whereas semantically “­d istant” concepts are often considered creative (­Heinen & Johnson, 2018; Kenett, 2018a, 2019). Critically, search processes are theorized to operate on a semantic memory system that is organized as a network structure. A classic cognitive model of semantic memory structure was proposed by Collins and Loftus (­1975). According to their model, concepts in memory are structured as a network, according to a semantic similarity principle: the more semantic properties two concepts share, they argue, the more links there are between them (­Collins & Loftus, 1975). Furthermore, the authors argue for a spreading activation model where, once a concept in the semantic network is activated, activation spreads from it to all of its directly connected neighbors, activation that quickly decays over time and space (­Collins & Loftus, 1975). Therefore, this classic theory proposes that semantic memory structure constrains the process of spreading activation operating over it, resonating with the associative theory of creativity (­see also Gentner, 1981; Klimesch, 1987; Kroll & Klimesch, 1992 for similar models). While the role of connecting more distant concepts in memory in creativity is intuitive, it is difficult to examine empirically due to the difficulty of measuring semantic memory 161

Yoed N. Kenett

structure and semantic distance ( ­Jones et al., 2015; Kumar, 2021). However, an increasing number of studies have applied computational methods to derive quantitative measures of semantic distance in relation to creativity (­Beaty & Johnson, 2021; Dumas et al., 2020; Green, 2016; Hass, 2017b; Heinen & Johnson, 2018; Kenett, 2018a, 2019).

­Corpus-​­Based Models of Semantic Memory Recent years have seen an exciting explosion of studies of conceptual representation that take advantage of the widespread availability both of massive natural language corpora and increased computational power ( ­Jones et al., 2015; Kumar, 2021). These distributional semantics approaches yield ­large-​­scale, ­d ata-​­d riven, ­h igh-​­d imensional semantic spaces of concepts that characterize natural language and explain many aspects of behavior, such as semantic similarity, semantic priming, and creativity (­Beaty & Johnson, 2021; Günther et al., 2019; Kumar, 2021; Mandera et al., 2015, 2017). A popular computational method to represent semantic distance in creativity research is through Latent Semantic Analysis (­L SA; Landauer & Dumais, 1997; Landauer et al., 1998). LSA quantifies the semantic similarity between words in a given ­h igh-​­dimensional semantic space by determining the probability of a given word ­co-​­occurring in a specific context (­e.g., a paragraph of a document). The semantic distance between a pair of words is determined by subtracting the LSA similarity score from 1.0 (­Prabhakaran et al., 2014). In the past few years, several studies have used ­L SA-​­based, alongside additional computational linguistic models, measures of semantic distance to study creativity (­Beaty & Johnson, 2021; Bourgin et  al., 2014; Dumas  & Dunbar, 2014; Dumas et  al., 2020; Forster  & Dunbar, 2009). For example, Prabhakaran et  al. (­2014) examined LSA-based semantic distances of responses in a verb generation task, where participants were required to produce verbs in response to a series of nouns, either a verb in a cued “­be creative” condition or a verb in an u ­ n-​­cued “­typical” condition. The authors found that the ­L SA-​­based semantic distance of the verbs to the nouns was higher in the cued “­be creative” condition compared to the n ­ on-​­cued “­t ypical” condition. Furthermore, the semantic distances between the verbs and nouns generated in the cued “­be creative” condition were also related to measures of fluid intelligence, DT, and creative achievement. Heinen and Johnson (­2018) showed that ­L SA-​­based measures of semantic distance relate to measures of novelty and appropriateness, common subjective measures of creative output (­Runco & Acar, 2012). Furthermore, the authors show that such LSA scores were sensitive to instruction manipulation and changed when participants were required to generate creative responses. Such creative responses had an average intermediate ­L SA-​­based semantic distance score, compared to a low average score for common responses or a high average score for random responses (­Heinen  & Johnson, 2018). Finally, Beaty and Johnson (­2021) have s­hown—​­across multiple s­tudies—​­that semantic distance based on ­corpus-​­based models such as LSA strongly corresponds with subjective ratings of the originality of responses in DT tasks. ­ SA-​­based measures of semantic distance to examine different Other studies have used L aspects of the cognitive processes involved in creativity. Beaty et al. (­2014) used LSA to assess participants’ associative abilities by averaging the semantic distance values of their responses generated during verbal fluency tasks to specific cue words (­see also Beaty et al., 2021; Gray et al., 2019; Olson et al., 2021). In addition, the authors also assessed participants’ executive ­processes—​­processes that are critical for control, inhibition, and monitoring of behavior (­Diamond, 2013)—​­such as fluid intelligence, crystalized intelligence, and broad retrieval abilities. These measures of associative ability and executive processes were used to examine 162

Creatively Searching Through Semantic Memory Structure

the contribution of both semantic memory structure and executive function processes to creative ability. Via structured equation modeling, the authors found joint effects of associative and executive abilities on the fluency and subjectively rated creativity scores of DT responses: both ­L SA-​­based semantic distance scores and individual differences in participants executive processes abilities predicted the creative quality of responses (­Beaty et al., 2014). Finally, Hass (­2017a, 2017b) applied ­L SA-​­based measures of semantic distance to “­t rack” the dynamics of generating DT responses in a continuous DT task. This research directly linked ­L SA-​­based semantic distance scores to the ­well-​­known ­serial-​­order effect in creative thinking (­Beaty & Silvia, 2012). As a participant generates more ideas, their ideas become more creative and, as this work s­hows—​­more distant. Furthermore, using linear growth model analysis, Hass (­2017b) finds further evidence for the role of fluid intelligence in this process (­Beaty & Silvia, 2012; Beaty et al., 2019). Thus, these studies demonstrate the significance of computing ­text-​­corpora based ­models—​­such as ­L SA—​­measures of semantic distance to examine creative output. However, objections have been raised regarding the validity of this approach as a measure of semantic distance and in predicting semantic priming effects (­Hutchison et al., 2008; Kenett et al., 2017; Kumar et al., 2020; Recchia & Jones, 2009; Simmons & Estes, 2006; Vankrunkelsven et al., 2018). For example, research has indicated that the performance of LSA models strongly depends on the choice and scope of the text corpus used, which can become the determining factor in how well the model captures human performance (­Recchia  & Jones, 2009). Furthermore, while more advanced computational models of semantic spaces are being developed (­Kumar, 2021; Mandera et al., 2017), the validity of estimating semantic distance based on analysis of textual corpora is yet to be determined (­De Deyne, Verheyen, et al., 2016). Finally, Forthmann et al. (­2018) have shown how elaboration in DT responses (­number of words in DT response) can bias their ­L SA-​­based measures of semantic distance. The authors recommend that caution is exercised in the interpretation of ­L SA-​­based measures of semantic distance in DT responses and offer ways to correct for such potential biases (­Forthmann et al., 2018). Thus, while useful, application of L ­ SA-​­based measures of semantic distance in creativity research can only provide information on the creative output. Furthermore, researchers should be aware of potential methodological pitfalls. These studies show how computational methods for representing semantic memory that are based on analyzing behavioral data collected from individuals or groups outperform text ­corpus-​­based models. Thus, models that are based on behavioral data might provide a stronger, more valid approach to studying semantic memory (­Kumar, 2021).

­Network-​­Based Models of Semantic Memory Recent studies have used computational network science methodologies to represent semantic memory and analyze its properties as a semantic network. Network science is based on mathematical graph theory, providing quantitative methods to investigate complex systems as networks (­Siew et al., 2019). A network is comprised of nodes that represent the basic unit of the system (­e.g., concepts in semantic memory) and links, or edges, that signify the relations between them (­e.g., semantic similarity). A growing body of research uses network science tools at the cognitive level to investigate the structure of language and memory ­ orge-​­Holthoefer & Arenas, 2010; Hills & Kenett, 2022; Karuza (­Baronchelli et al., 2013; B et al., 2016; Siew et al., 2019). For example, cognitive network science has enabled the direct examination of the theory that highly creative individuals have a more flexible semantic memory (­Kenett & Faust, 2019), identified mechanisms of language development through 163

Yoed N. Kenett

preferential attachment (­H ills et al., 2009; Steyvers & Tenenbaum, 2005), shed novel light on statistical learning (­K aruza, 2022), shown how specific semantic memory network parameters influence memory retrieval (­Vitevitch et al., 2014; 2012), and provided new insight on second language in bilinguals (­Borodkin et al., 2016). A growing number of studies have applied network science methodologies to study creativity, focusing on the role of semantic memory structure in the creative process (­Kenett, 2018b; Kenett & Faust, 2019). Kenett and colleagues conducted a network analysis of free association data to compare the semantic memory network structure of ­low-​­and ­h igh-​ ­creative individuals, defined by performance on a battery of creative thinking tasks. The authors found that the semantic memory network of the h ­ igh-​­creative group was characterized by higher connectivity and lower overall distances between concepts (­Kenett et al., 2014), likely permitting more efficient spreading activation processes to unfold (­Kenett et al., 2018). Importantly, these network characteristics have been shown to facilitate broader search processes (­Kenett & Austerweil, 2016) that allow them to reach weaker and more uncommon concepts (­Stella & Kenett, 2019). These findings have since been replicated and extended in other ­g roup-​­based analyses (Kenett, Beaty et al., 2016), as well as at the individual level (­Benedek et al., 2017; He et al., 2021; ­O vando-​­Tellez et al., 2022). A similar semantic memory structure was also reported in a recent study of openness to ­experience—​ ­a personality trait linked to creative behavior and cognition (­Christensen et  al., 2018)—​ ­providing further support for the role of semantic memory in creativity (­Kenett & Faust, 2019; Volle, 2018). Taken together, these studies elucidate the role of semantic memory in creative thinking.

Knowledge Systems and Creative Thinking So far, we have discussed semantic memory as a unitary system, one which stores facts and knowledge. Yet, semantic memory may be composed of different knowledge systems that are differently recruited in creative thinking. Mumford et al. (­Hunter et al., 2008; Mumford et al., 2002) have argued that three different types of knowledge systems take part in the creative process: schematic, associative, and case knowledge. However, other types of knowledge systems exist, such as visuospatial knowledge. These additional knowledge systems may play separate roles in behavior, depending on task demands (­see He et al., 2021 for such a discussion). Schematic knowledge relates to generalized knowledge abstracted from experience (­Ghosh & Gilboa, 2014). Such conceptual knowledge is theorized to be organized in a way that potentially reflects causal relations between concepts in memory (­Hummel & Holyoak, 1997). Furthermore, such organized, schematic knowledge has been related to expertise (­Adams & Ericsson, 2000). It is considered to facilitate analogical transfer, which is an important process in creative thinking (­Green, 2016; Mumford et al., 2002). For example, Baughman and Mumford (­1995) examined the use of analogical transfer in conceptual combination, where one is required to combine concepts together into compounds (­Wisniewski, 1997). Conceptual combinations are considered a core process of creative thinking (­Finke et al., 1992; Mednick, 1962) and are considered to be based on either the mapping of relations between concepts or on feature search and integration strategies (­Estes, 2003). Baughman and Mumford (­1995) found that the feature search and mapping strategies applied when combining concepts influenced the quality and originality of the resulting new c­ oncepts—​­a finding that suggests that p­ rinciple-​­based concepts may provide a framework for conceptual combination efforts. 164

Creatively Searching Through Semantic Memory Structure

Associative knowledge relates to knowledge that is accumulated based on repeated ­stimulus–​ e­ vent pairings and/­or exposure to regularities in events (­Nelson et al., 2000). As discussed above, associative knowledge is theorized to be organized as a network structure that facilitates search processes operating over it (­Collins & Loftus, 1975; Mednick, 1962; Ward et al., 2004) and plays a critical role in creative thinking (­Benedek et al., 2012). This theory is empirically supported by studies showing how highly creative individuals are able to generate more associative responses that are also rated as more remote than are less creative individuals (­Gruszka & Neçka, 2002; Mednick et al., 1964; Rossman & Fink, 2010). ­Case-​­Based Knowledge relates to knowledge that is founded on experience in a particular situation (­Scott et  al., 2005). While related to schematic and associative knowledge, such knowledge is considered to be an abstracted summary of information that is consciously constructed and accessible, such as goals and performance setting. Mumford et al. (­Hunter et al., 2008; Mumford et al., 2002) argue that these three knowledge systems all contribute to facilitate creative thought but may be recruited for different types of tasks or stages of the creative process. For example, the associative system may be recruited more for idea generation to facilitate DT (­Gruszka & Neçka, 2002; Kenett & Faust, 2019). In addition, Mumford et al. (­2002) acknowledge the importance of executive processes operating over these knowledge systems in the creative process and call for new theories that integrate the roles and dynamic interactions of both knowledge systems and processes in creative thinking. The current computational approaches to studying the role of semantic memory in creativity do not differentiate across these knowledge systems. Thus, constructing such ­corpus-​­based models or semantic networks based on different types of these proposed knowledge systems will allow for more sensitive empirical investigations of the role of different types of knowledge in creative thinking.

Summary The work described above highlights the role that the structure of semantic memory plays in creative thinking, and how current computational methods are being applied to study its role in the context of creative thinking. Semantic memory structure is considered the cognitive scaffolding for creative thinking (­Abraham & Bubic, 2015). Importantly, semantic memory structure constrains the search processes that operate over it during creative thinking, which we will define here as creative search processes. We now turn to how we can use computational methods to study creative search and how it is constrained by semantic memory structure.

Creative Search The associative theory of creativity proposes that individual differences in semantic memory structure facilitate search processes that can reach deeper in memory for weakly connected concepts (­Mednick, 1962). Such search processes might be based on the spreading activation mechanism (­Collins  & Loftus, 1975). According to the spreading activation model, once a concept is activated in memory, information “­spreads” from it to all directly connected concepts and so forth (­Collins & Loftus, 1975; Klimesch, 1987; Kroll & Klimesch, 1992). This information quickly decays over time and space (­Balota & Lorch, 1986; D ­ en-​­Heyer & ­ ind-​­wandering or sponBriand, 1986). These search processes can be unguided, such as in m taneous thought, or guided, such as in g­ oal-​­directed p­ roblem-​­solving (­Volle, 2018). The development of computational models and tools, in conjunction with an increasing interest 165

Yoed N. Kenett

in cognitive search, is leading to a growing body of work that conducts either simulations or empirical studies in relation to creative thinking.

Simulating Spontaneous Creative Search Processes with Random Walks Computationally, spreading activation can be implemented as a random walk search process over a network (­De Deyne, Kenett et al., 2016; Siew, 2019; Siew et al., 2019). Starting at a particular node, a random walk selects an outbound edge with a probability proportional to the edge’s weight and moves across it. As this process progresses, it explores more nodes in the network. A random walk is determined by the number of steps it can take, and the structure of the network that it traverses over determines its performance. Siew (­2019) developed a random ­walk-​­like algorithm that better captures Collins and Loftus’ (­1975) spreading activation mechanism. This application, SpreadR, simulates an activation process that cascades, or spreads, randomly through a network as decaying process over time. While SpreadR has not been directly applied in creativity research, recent research has explored how random walk models can capture memory retrieval (­Abbott et  al., 2015; Capitán et al., 2012; Griffiths et al., 2007) and performance in creative tasks that require cognitive search, such as the remote association task (­R AT; Bourgin et al., 2014; Gupta et al., 2012; Smith et al., 2013; Smith & Vul, 2015). In the RAT (­Mednick, 1962), participants are presented with a triplet of seemingly unrelated cue words (­e.g., Cottage, Swiss, Cake) and are required to find a fourth single target word that is separately related to each of the cue words (­Cheese; Bowden & ­Jung-​­Beeman, 2003). Smith et al. (­2013) view the RAT as a multiple constraint problem, in which each cue word indicates a different attribute of the target word. Solving such a multiple constrained problem requires a ­t wo-​­stage process: a divergent search process for an alternative possible solution is conducted, and then this candidate solution is tested against all of the constraints of the problem to rate the acceptability of the solution (­Smith et al., 2013). The authors found that participants solve RAT stimuli first by selecting a set of possible answers constrained by a single cue word at a time. Furthermore, the authors show how prior candidates’ answers directly affect the following guesses, suggesting an associatively connected directed search, which is in agreement with the spreading activation model (­Collins & Loftus, 1975). These studies investigated how well a random walk over semantic memory captures general performance on cognitive search and creative tasks. However, they have not examined whether differences in creative ability can be understood in terms of the same random walk process on different semantic memory structure. Kenett and Austerweil (­2016) directly examined this issue by simulating and comparing random walk models on the semantic networks of ­low-​­and ­h igh-​­creative individuals (­Kenett & Austerweil, 2016). The authors hypothesized that the structure of the semantic network of h ­ igh-​­creative individuals enables them to use search processes that reach further and to weaker connected concepts in their memory than ­low-​­creative individuals. To test this hypothesis, the authors conducted random walk simulations on the semantic networks of the ­low-​­and ­h igh-​­creative individuals, collected by Kenett et al. (­2014). To conduct a random walk simulation over the two networks, the authors first computed a transition probability matrix for each network: the probability of a walk to transition from a specific node to any other nodes connected to it. Next, they choose a similar starting node on both networks to initiate the walk. Finally, the authors initiated walks with a varying number of steps (­­10–​­200 steps) through 10,000 simulations.

166

Creatively Searching Through Semantic Memory Structure

The authors computed two “­creative” measures based on the random walk simulations: the number of unique visited nodes by the walk as a measure of the breadth of the search and the similarity between the initial and final visited nodes as a measure of the distance between connected concepts. To control for confounding effects of the number of unique visited nodes on the similarity measure, the authors also computed the similarity score between the initial node and a unique visited node after a fixed, truncated number of unique visited nodes (­e.g., similarity score between the initial node and the fifth unique visited node; see Kenett  & Austerweil, 2016). For each number of steps in the walks examined, the authors computed each of the two measures and examined the difference between the averaged measures of the two groups. A positive/­negative difference score indicated that the ­h igh-​­creative/­­low-​­creative group had a bigger value than the ­low-​­creative/­­h igh-​­creative group. In line with the associative theory of creativity, the authors found that a random walk over the ­h igh-​­creative semantic network visits more unique nodes and that the similarity strength between initial and final nodes visited by this walk is weaker than the performance of the random walk over the ­low-​­creative semantic network. Thus, individual differences in thought processes between ­low-​­and ­h igh-​­creative individuals can be produced by an uncontrolled search process on differing semantic networks, providing support for the associative theory of creativity (­Mednick, 1962). The findings of Kenett and Austerweil (­2016) demonstrate how search processes in l­ow-​­ and ­h igh-​­creative individuals can be simulated and examined via random walk simulations and how the walk over the semantic network of ­h igh-​­creative individuals reaches farther and weaker concepts in their semantic network. The study of Kenett and Austerweil (­2016) adds to the previous applications of random walk models to examine creative tasks such as the RAT (­Bourgin et al., 2014; Smith et al., 2013). These efforts demonstrate the strength of such models in examining cognitive search processes related to individual differences in creativity.

Empirical Research on Creative Search The application of computational models to empirically study cognition is increasingly being used to study creative search processes (­Beaty & Johnson, 2021; Beaty et al., 2021; Gray et al., 2019; Hass, 2017a, 2017b; Olson et al., 2021). These approaches use t­ ext-​­based corpora measures of semantic distance, such as LSA, to compute and measure the distance between sequential responses that develop over time and relate to creative thinking. For example, Gray et al. (­2019) measure the sequential semantic distances between pairs of associates in a continuous, chained, free association task, operationalizing James’s notion of a stream of thought. The authors show across several samples how this ­measure—​­which they term forward fl ­ ow—​­correlates with individual differences in creative thinking. Others are using such computational models to empirically investigate the increase in semantic distance during a continuous DT task and how that relates to response originality (­Beaty & Johnson, 2021). In pioneering work that converged creativity research with d­ ecision-​­making research, Harms et al. (­2018) examined the role of information search in problem construction, an early stage in the creative process. The authors found that the effect of active problem construction in creative thinking was directly related to ­information-​­search behaviors, above and beyond additional cognitive factors that were controlled for (­Harms et al., 2018). This finding illustrates that creativity is an i­nformation-​­intensive process that is contingent on problem construction. The authors interpret their findings as indicating that broad construction of problems, a critical early stage of the creative process, allows for identifying more

167

Yoed N. Kenett

varied information alternatives to explore, which leads to h ­ igher-​­quality responses (­Harms et al., 2018). This study highlights the importance of operationalizing creative search behavior to advance empirical research on such behavior.

Search Variability and Effort in Creative Search All of the work described above emphasizes at least one of two critical aspects of creative search through memory. The first is the structure of one’s memory, and how heightened connectivity within this system facilitates creative search. The other is the nature of the search process itself, which allows it to propagate further and deeper in one’s memory. These two aspects were formally operationalized by Acar and van den Ende (­2016), who proposed two cognitive processes that are critical for effective creative search: Search variation and search effort. Search variation refers to the variation in elements of knowledge used when creatively combining or recombining ideas into novel and appropriate ideas (­Acar & van den Ende, 2016). This process is inspired by the work of Schilling and colleagues in innovation and technological development (­K neeland et  al., 2020; Schilling, 2005; Schilling  & Green, 2011), who have proposed that such a recombination process leads to restructuring effects in one’s semantic memory network structure (­Schilling, 2005). Schilling and Green (­2011) argue that the process of knowledge creation is related to a dynamical process over semantic networks, and that atypical links in such a network have a large impact on one’s ability to relate different ideas together, leading to breakthroughs during idea generation, similar to the associative theory of creativity (­Mednick, 1962). The authors found that when scientists cite more work from different scientific domains in their scientific articles, these articles are more likely to produce ­h igh-​­impact articles. Thus, the authors argue that their findings support the importance of search variation, or scope, for a scientific article’s impact (­Schilling & Green, 2011). Kneeland et al. (­2020) examined how innovators search through a database of outlier patents (­d iscoveries that seem very “­d istant” from known and understood technologies) alongside interviews with the inventors of such outlier patents. The authors argue that such outlier patents are due to search processes over knowledge networks, search processes that lead to discoveries that are distant from existing inventions. The authors interpret their results as demonstrating the role of long search paths that are latent in the technological development trajectory and that, alongside the use of scientific reasoning and an active process involving recombination of ideas, this leads to outlier patents. Such an interpretation strongly resonates with the associative theory of creativity and further demonstrates how one’s knowledge structure (­here technology patents) constrains search processes over it, and how a search process that can run deeper over such a structure leads to novel inventions. Search effort refers to the amount of attention devoted to creating a novel solution and is related to the intensity aspect of attention (­Acar & van den Ende, 2016). Search effort is important because it increases one’s c­ognitive-​­processing capacity to notice connections between different elements and make sense of these connections in such a way that they can be recombined to generate a novel solution to a given problem (­Acar & van den Ende, 2016). Recent cognitive neuroscience work has focused on the role of internal attention (­Benedek & Fink, 2019), as well as cognitive control (­Chrysikou, 2018, 2019), in generating creative ideas; however, much less is known on how these mechanisms mediate creative search processes over semantic memory. In this regard, Volle (­2018) proposed that creative idea generation is facilitated by executive processes, such as attention and cognitive control, which mediate search processes that operate over semantic memory structure (­Volle, 2018). 168

Creatively Searching Through Semantic Memory Structure

Furthermore, two recent variations in effortful behaviors are slowly being related to creative thinking: Curiosity and Exploration versus Exploitation behaviors. Curiosity is considered to be a motivational driver that involves the pursuit of new knowledge and experiences (­Gross et al., 2020). However, the relation of the construct of curiosity to creativity is still poorly understood (­Hagtvedt et al., 2019; Schutte & Malouff, 2020), as is its relation to searching over semantic memory structure (­but see ­Lydon-​­Staley et al., 2021). Exploration versus exploitation behaviors refer to how an individual externally (­e.g., spatially) or internally (­e.g., memory) searches, or forages, and exploits resources in various domains (­H ills et al., 2013; Hills et al., 2015; Todd & Hills, 2020). Recently, Hart et al. (­2018, 2017) developed a creative foraging game. In this game, participants search a space of geometric shapes made up of ten connected squares, moving one square at a time. Participants are asked to collect “­interesting and beautiful shapes” (­Hart et al., 2017). Importantly, Hart et  al. found that participants’ performance in this game was significantly correlated with their performance in the alternative uses task, which predicts DT, linking s­ patial-​­based creative foraging with creative thinking (­Hart et al., 2017). More recently, Kenett et al. (­2021) examined the relation between creative foraging, semantic memory structure, and creative thinking by estimating the semantic networks of groups that varied on their performance in the creative foraging game. We found that people who are high on exploration and switching between modes of exploration and exploitation in this task have a highly connected semantic memory network, possibly facilitating such heightened exploration. Furthermore, we found that this group of people had the highest creativity scores in an alternative uses task (­Kenett et al., 2021). Taken together, these two operationalized aspects of creative search allow us to advance the research of cognitive search in general, and specifically creative search. Developing new research designs that can study each of these aspects, as well as their interaction, would propel this research direction forward.

The Cognitive Control Account to Creative Search As we described above, Volle proposed that creative search is either unguided, such as in ­m ind-​­wandering, or guided, such as in ­problem-​­solving (­Volle, 2018). Such guided search processes are likely related to t­op-​­down executive processes, such as attention, intelligence, working memory, and cognitive control (­Benedek et  al., 2014; Benedek  & Fink, 2019; Chrysikou, 2019; Silvia, 2015). For example, studies using computational methods such as those described above have found contributions of both semantic memory structure and different facets of intelligence in relation to creativity (­Beaty et al., 2014; Benedek et al., 2017). These studies further highlight how such research designs, converging computational and empirical methods, can elucidate the complexity of creative search.

Summary Till now, the benefits of a richer, flexible, semantic memory structure have been argued to facilitate memory search ­that—​­according to the associative theory of ­creativity—​­leads to creative ideas via connecting concepts that are disconnected or farther apart in semantic memory (­Kenett & Faust, 2019; Mednick, 1962). Thus, an immediate conclusion is that the richer the semantic memory structure is, the more creative one is. Yet the relation between memory and creative thinking is more complex (­Ditta & Storm, 2018). In a series of studies, Beaty et al. (­2023) demonstrate how the richness of a cue’s semantic neighborhood affects 169

Yoed N. Kenett

the quantity and quality of creative ideas generated in an alternative use task. The authors find that cues with rich semantic neighborhoods (­m any associations related to them) lead participants to generate a higher number of responses than for cues that have sparse semantic neighborhoods. However, these responses were judged to be of lower originality than those generated to cues with sparse semantic neighborhoods (­Beaty et al., 2023). A salient example of the complex relations between memory (­or knowledge) and creativity is the effect of domain expertise on creative thinking.

How Semantic Memory Structure Constrains Creative ­Search—​­A Case of Expertise Previous research has emphasized the importance of d­ omain-​­knowledge expertise to come up with a novel, h ­ igh-​­impact idea (­Schilling, 2005). For example, it has been found that individuals often require at least a decade of intense study in a particular domain of knowledge prior to making a significant contribution in that domain (­Simon & Chase, 1973). Since a rich semantic memory system is critical for creative thinking (­Abraham, 2014; Abraham & Bubic, 2015), an intuitive conclusion would be that expertise facilitates creative thinking and creative search processes. However, research has demonstrated that the relation between rich semantic memory structure and creativity is more complex (­Beaty et al., 2019; Wiley, 1998)

Expertise and Rich Semantic Memory Structure Expertise has been argued to benefit d­ omain-​­specific creative thinking by developing a rich body of knowledge in that domain that allows experts to generate novel and useful ideas (­Abraham & Bubic, 2015; Mumford et al., 2002; Schilling, 2005). Experts are considered to be people who have substantial experience in a given domain, and they will have a richer, more elaborate, and better organized knowledge structure than novices (­Mumford et  al., 2002). Thus, expertise, and specifically a richly organized body of d­ omain-​­specific knowledge, has been previously linked to creative achievement across different domains (­Mumford et al., 2002). The link between knowledge, expertise, and creativity is based on the principle of knowledge accumulation over time. As associations between concepts in a body of d­ omain-​­specific knowledge develop over time, the more accurate the pattern of associations becomes, and the more efficient the expert becomes in searching for creative d­ omain-​­specific solutions (­Dosi, 1988; Schilling, 2005). A salient example of such an effect has been shown in expert chess players (­Bilalić et al., 2009; Sala & Gobet, 2017; Simon & Chase, 1973). In a recent ­meta-​­analysis, Sala and Gobet (­2017) show that an ­expertise-​­knowledge effect is not limited to chess, but to many different types of domains.

Expertise and Its Effect on Creative Search However, expertise related to a rich body of ­domain-​­specific knowledge can also inhibit creative thinking by leading to fixation or inflexibility (­Acar & van den Ende, 2016; Dane, 2010; Schilling, 2005). Functional fixedness has been related to ­problem-​­solving impasse, where participants are fixated on the most salient functions of an object, leading them to “­get stuck” (­Chrysikou et al., 2016; Smith & Blankenship, 1991). It is also related to the ­well-​­known einstellung phenomenon (­Luchins, 1947), where participants who previously 170

Creatively Searching Through Semantic Memory Structure

solved a problem in a particular way will form a p­ roblem-​­solving set that constrains them from developing further novel, creative solutions (­Wiley, 1998). Such limitations may occur because one’s knowledge structure impacts the initial representation of the problem or task, thus anchoring the types of responses that the individual can think of (­Acar & van den Ende, 2016). One salient example of this effect is research showing how misleading clues affect performance in the RAT (­Sio & Rudowicz, 2007; Wiley, 1998). Wiley (­1998) examined the effect of expertise on performance in the RAT. In a series of experiments, baseball novices and experts attempted to solve ­baseball-​­related RAT trials where the target solutions were not related to baseball. Wiley found that baseball experts solved fewer ­baseball-​­related problems than baseball novices (­Wiley, 1998). Wiley argued that expertise leads experts to search unnecessarily confined search spaces in their semantic memory while attempting to solve these ­domain-​­related RAT trials, while novices apply less constrained search processes and are thus more likely to find the appropriate solution. Thus, a rich knowledge base can be detrimental if it leads to fixation on unhelpful information.

Search Engines A final point of discussion in this chapter will examine the overlap of creative search in semantic memory with individuals’ search performance in online search engines such as Google (­w ww.google.com). Previous sections have advanced and discussed the theory that creativity involves creative search processes that are constrained by one’s semantic memory structure. This is akin to how search processes in internet search engines operate. So, one may ask: How similar are these two search processes (­mental and internet)? Can we learn anything from one process for the other? And finally, how do expertise levels relate to querying internet search engines? A fundamental study in this topic was conducted by Griffiths et al. (­2007), where they examined how the task of retrieving animal names in a semantic fluency task can be predicted by Google’s search algorithm, PageRank (­Griffiths et al., 2007). The authors show that the PageRank ranking of a word, computed from a semantic network, outperformed standard linguistic features (­e.g., frequency) in predicting the retrieval of that word by participants in various fluency tasks (­Griffiths et al., 2007). The authors conclude that the analogy between information retrieval and human memory search is worth pursuing. Importantly, insights from human creative search through memory facilitate exploratory search engine queries. While search engines have become extremely efficient at retrieving simple facts and targeted information, supporting exploratory search queries is still an open information retrieval challenge (­Mao et al., 2018). In addition to providing relevant results according to the search query, the search engine needs to provide additional context to the more exploratory results it provides (­Hassan Awadallah et  al., 2014). To do that, a better characterization of human search and its underlying factors is crucial. One such underlying factor is an individual’s domain expertise: domain experts are more successful in web search than novices; they exhibit search strategies that maximize website selection and search sequences, and their expertise has been shown to affect their query behavior and search effectiveness (­Mao et al., 2018). Mao et al. (­2018) empirically investigated how domain expertise affects individuals’ exploratory search. The aim of this study was both to elucidate why only some search queries are successful and to advance the development of new search models that capitalize on the user’s domain knowledge. The authors found that: (­1) domain experts are more effective in conducting ­in-​­domain search tasks; (­2) the results 171

Yoed N. Kenett

clicked by domain experts are not necessarily more useful; and (­3) in exploratory search, starting from the task description as an initial search query and then broadening the search via terms from landing pages facilitates exploratory search across domains. Finally, the authors found some variance in search behavior across ­domain-​­specific expertise (­Mao et al., 2018). Thus, combining research on information retrieval systems, such as search engines, with cognitive research on memory search in creative and n ­ on-​­creative settings is needed to further cross fertilize both of these fields.

Summary Overall, the findings described in this section demonstrate how expertise both benefits and hinders creative search. These findings illustrate how semantic memory structure both facilitates and constrains search processes and how it may impede the generation of novel ideas. Converging such ideas being developed in disconnected fi ­ elds—​­cognitive psychology, information retrieval, and ­m anagement—​­alongside the rapid development of computational tools that allow empirical study of such phenomena, calls for a more nuanced, complex theory on the relation between memory and creativity. We end this section with two final studies that illustrate this idea: First, Faust and Kenett (­2 014), based on their work with clinical and typical populations, proposed a cognitive continuum of semantic memory structures. At one extreme of this continuum are rigid networks, which are minimally connected (­i.e., each node in such a network is only connected to one other node). Activation decays as it spreads through more links, resulting in repetitive retrieval, because only a few nodes close to the source will be activated enough to be retrieved. Rigid networks are characteristic of the semantic memory structure of individuals with autism spectrum disorders (­Kenett, Gold et al., 2016). At the other extreme are disorganized, chaotic networks, which are highly connected. This results in disorganized retrieval because many nodes are activated from likely a single source. Chaotic networks are characteristic of the semantic memory structure of individuals with schizophrenia. According to my theory, semantic retrieval in typical individuals is achieved via a balance between rigid and chaotic network structure. This balance allows both conventional (­ordered) and flexible (­chaotic) language processing (­Faust & Kenett, 2014). Second, Kenett et al. (­2 015) proposed a cognitive theory for stimuli processing based on the interaction between two systems: an expert system that is specialized in processing conventional, typical, stimuli and a ­non-​­expert, holistic, system that is more efficient in processing unconventional, atypical stimuli. This theory is inspired by the brain laterality of linguistic processing, where it has been shown that the left hemisphere is specialized in processing conventional, dominant meanings of words, whereas the right hemisphere is more efficient in processing unconventional, creative meanings of words (­Kenett et al., 2015). We demonstrated this effect in a different m ­ odality—​­face ­processing—​­where brain laterality is reversed and the right hemisphere is specialized in face processing. Using conventional and unconventional faces, the authors show that the left hemisphere was more efficient in processing unconventional faces, which was related to participants’ higher creativity. Taken together, these findings highlight the importance of constantly maintaining a balance between these two f­orces—​­expertise and fl ­ exibility—​­for optimal performance of the cognitive system. Semantic memory provides the scaffolding for both of these “­forces,” thus its optimal structure retains enough domain generality to not induce d­ omain-​­specific 172

Creatively Searching Through Semantic Memory Structure

anchoring effects. A few studies have directly linked the structure of semantic memory to flexibility using computational tools and how typical aging leads to increasing semantic memory structuring but also decreasing flexibility (­Cosgrove et al., 2021; Kenett et al., 2018).

­Summary – Searching ​­ Creatively in Memory This chapter focused on creative search processes that operate over semantic memory, highlighting how creativity involves creative search processes that are constrained by the structure of one’s semantic memory. This was done by highlighting how recently developed computational methods aid us in directly measuring semantic memory structure, as well as in simulating and empirically measuring search processes, in relation to individual differences in creative thinking. Similar to the associative theory of creativity, such recent research on creative search is advancing by converging disconnected fields such as cognitive psychology, computational linguistics, information retrieval, and innovation research. In creativity theory, the importance of a search process that “­moves farther away” from the original stimulus is intuitively embedded (­Kenett, 2018a). Yet, much is still unknown about creative search and how it relates to general cognitive search processes over memory (­H ills et al., 2012; Todd & Hills, 2020). Future research on creative search must move beyond the issue of how semantic memory structure constrains the creative search process to questions such as: how does the search process reach a stopping criterion (­Ackerman, 2014)? How can spontaneous versus guided creative search processes be accounted for (­Volle, 2018)? How are creative evaluation processes interacting with the creative search process (­K leinmintz et  al., 2019)? And how are ­domain-​­specific versus ­domain-​­general creative search processes similar or different? We are now in an exciting era where powerful computational models and tools are being developed and converged with empirical research to study cognition like never before. Applying such an approach to study creative thinking has already begun to unpack its complexity. Since the creative process is an active process, we must push forward in elucidating its nature and how it is constrained by additional factors, such as semantic memory.

References Abbott, J. T., Austerweil, J. L., & Griffiths, T. L. (­2015). Random walks on semantic networks can resemble optimal foraging. Psychological Review, 122(­3), ­558–​­569. https://­doi.org/­10.1037/­a0038693 2014). Creative thinking as orchestrated by semantic processing versus cogniAbraham, A. (­ doi. tive control brain networks [Perspective]. Frontiers in Human Neuroscience, 8, 95. https://­ org/­10.3389/­f nhum.2014.00095 Abraham, A., & Bubic, A. (­2015). Semantic memory as the root of imagination. Frontiers in Psychology, 6, 325. https://­doi.org/­10.3389/­f psyg.2015.00325 Acar, O. A.,  & van den Ende, J. (­2016). Knowledge distance, c­ ognitive-​­search processes, and creativity: The making of winning solutions in science contests. Psychological Science, 27(­5), ­692–​­699. Acar, S.,  & Runco, M. A. (­2019). Divergent thinking: New methods, recent research, and extended theory. Psychology of Aesthetics, Creativity and the Arts, 13(­2), ­153–​­158. https://­doi. org/­10.1037/­aca0000231 Ackerman, R. (­2014). The diminishing criterion model for metacognitive regulation of time investment. Journal of Experimental Psychology: General, 143(­3), ­1349–​­1368. Adams, R. J., & Ericsson, A. E. (­2000). Introduction to cognitive processes of expert pilots. Journal of Human Performance in Extreme Environments, 5(­1), ­4 4–​­62. Balota, D. A.,  & Lorch, R. F. (­1986). Depth of automatic spreading activation: Mediated priming effects in pronunciation but not in lexical decision. Journal of Experimental Psychology: Learning, Memory and Cognition, 12(­3), 3­ 36–​­345.

173

Yoed N. Kenett Baronchelli, A., ­Ferrer-­​­­i-​­Cancho, R., ­Pastor-​­Satorras, R., Chater, N., & Christiansen, M. H. (­2013). Networks in cognitive science. Trends in Cognitive Sciences, 17(­7 ), ­348–​­360. https://­doi.org/­ 10.1016/­j.tics.2013.04.010 Baughman, W. A., & Mumford, M. D. (­1995). ­Process-​­analytic models of creative capacities: Operations influencing the c­ ombination-­​­­a nd-​­reorganization process. Creativity Research Journal, 8(­1), ­37–​­62. Beaty, R. E., & Johnson, D. R. (­2021). Automating creativity assessment with semdis: An open platform for computing semantic distance. Behavior Research Methods, 53(­2), ­757–​­780. https://­doi. org/­10.3758/­­s13428-­​­­020-­​­­01453-​­w Beaty, R. E., Kenett, Y. N., Hass, R. W., & Schacter, D. L. (­2023). Semantic memory and creativity: The costs and benefits of semantic memory structure in generating original ideas. Thinking & Reasoning, 29(2), 305–339. https://doi.org/10.1080/13546783.2022.2076742 Beaty, R. E., & Silvia, P. J. (­2012). Why do ideas get more creative over time? An executive interpretation of the serial order effect in divergent thinking tasks. Psychology of Aesthetics, Creativity and the Arts, 6(­4), ­309–​­319. https://­psycnet.apa.org/­doi/­10.1037/­a0029171 Beaty, R. E., Silvia, P. J., Nusbaum, E. C., Jauk, E.,  & Benedek, M. (­2014). The roles of associative and executive processes in creative cognition. Memory & Cognition, 42(­7 ), ­1–​­12. https://­doi. org/­10.3758/­­s13421-­​­­014-­​­­0428-​­8 Beaty, R. E., Zeitlen, D. C., Baker, B. S., & Kenett, Y. N. (­2021). Forward flow and creative thought: Assessing associative cognition and its role in divergent thinking. Thinking Skills and Creativity, 41, 100859. https://­doi.org/­10.1016/­j.tsc.2021.100859 Benedek, M., & Fink, A. (­2019). Toward a neurocognitive framework of creative cognition: The role of memory, attention, and cognitive control. Current Opinion in Behavioral Sciences, 27, ­116–​­122. https://­doi.org/­10.1016/­j.cobeha.2018.11.002 Benedek, M., Jauk, E., Sommer, M., Arendasy, M., & Neubauer, A. C. (­2014). Intelligence, creativity, and cognitive control: The common and differential involvement of executive functions in intelligence and creativity. Intelligence, 46, 7­ 3–​­83. https://­doi.org/­10.1016/­j.intell.2014.05.007 Benedek, M., Kenett, Y. N., Umdasch, K., Anaki, D., Faust, M., & Neubauer, A. C. (­2017). How semantic memory structure and intelligence contribute to creative thought: A network science approach. Thinking & Reasoning, 23(­2), ­158–​­183. https://­doi.org/­10.1080/­13546783.2016.1278034 Benedek, M., Könen, T., & Neubauer, A. C. (­2012). Associative abilities underlying creativity. Psychology of Aesthetics, Creativity and the Arts, 6(­3), ­273–​­281. Benedek, M., & Neubauer, A. C. (­2013). Revisiting Mednick’s model on c­ reativity-​­related differences in associative hierarchies. Evidence for a common path to uncommon thought. The Journal of Creative Behavior, 47(­4), ­273–​­289. https://­doi.org/­10.1002/­jocb.35 Bilalić, M., McLeod, P., & Gobet, F. (­2009). Specialization effect and its influence on memory and problem solving in expert chess players. Cognitive Science, 33(­6), ­1117–​­1143. ­Borge-​­Holthoefer, J.,  & Arenas, A. (­2010). Semantic networks: Structure and dynamics. Entropy, 12(­5), ­1264–​­1302. https://­doi.org/­10.3390/­e12051264 Borodkin, K., Kenett, Y. N., Faust, M., & Mashal, N. (­2016). When pumpkin is closer to onion than to squash: The structure of the second language lexicon. Cognition, 156, 6­ 0–​­70. https://­doi.org/­ 10.1016/­j.cognition.2016.07.014 Bourgin, D. D., Abbott, J. T., Griffiths, T. L., Smith, K. A., & Vul, E. (­2014). Empirical evidence for markov chain monte carlo in memory search. Proceedings of the 36th Annual Conference of the Cognitive Science Society, Boston, MA. Bowden, E. M.,  & J­ung-​­Beeman, M. (­2003). One hundred f­orty-​­four compound remote associate problems: Short ­i nsight-​­l ike problems with ­one-​­word solutions. Behavioral Research, Methods, Instruments, and Computers, 35, ­634– ​­639. Capitán, J. A., ­Borge-​­Holthoefer, J., Gómez, S., ­Martinez-​­Romo, J., Araujo, L., Cuesta, J. A.,  & Arenas, A. (­2012). ­L ocal-​­based semantic navigation on a networked representation of information. PLoS ONE, 7(­8), e43694. https://­doi.org/­10.1371/­journal.pone.0043694 Christensen, A. P., Kenett, Y. N., Cotter, K. N., Beaty, R. E., & Silvia, P. J. (­2018). Remotely close associations: Openness to experience and semantic memory structure. European Journal of Personality, 32(­4), ­480– ​­492. https://­doi.org/­doi:10.1002/­per.2157 Chrysikou, E. G. (­2018). The costs and benefits of cognitive control for creativity. In R. E. Jung & O. Vartanian (­Eds.), The Cambridge Handbook of the Neuroscience of Creativity (­­pp. ­299–​­317). Cambridge University Press.

174

Creatively Searching Through Semantic Memory Structure Chrysikou, E. G. (­2019). Creativity in and out of (­cognitive) control. Current Opinion in Behavioral Sciences, 27, ­94–​­99. https://­doi.org/­10.1016/­j.cobeha.2018.09.014 Chrysikou, E. G., Motyka, K., Nigro, C., Yang, S.-​­I., & ­Thompson-​­Schill, S. L. (­2016). Functional fixedness in creative thinking tasks depends on stimulus modality. Psychology of Aesthetics, Creativity and the Arts, 10(­4), ­425–​­435. https://­doi.org/­10.1037/­aca0000050 Collins, A. M., & Loftus, E. F. (­1975). A ­spreading-​­activation theory of semantic processing. Psychological Review, 82, ­407–​­428. Cosgrove, A. L., Kenett, Y. N., Beaty, R. E., & Diaz, M. T. (­2021). Quantifying flexibility in thought: The resiliency of semantic networks differs across the lifespan. Cognition, 211, 104631. https://­doi. org/­10.1016/­j.cognition.2021.104631 Dane, E. (­2010). Reconsidering the t­ rade-​­off between expertise and flexibility: A cognitive entrenchment perspective. Academy of Management Review, 35(­4), ­579–​­603. De Deyne, S., Kenett, Y. N., Anaki, D., Faust, M., & Navarro, D. J. (­2016). ­Large-​­scale network representations of semantics in the mental lexicon. In M. N. Jones (­Ed.), Big Data in Cognitive Science: From Methods to Insights (­­pp. ­174–​­202). Psychology Press: Taylor & Francis. De Deyne, S., Verheyen, S., & Storms, G. (­2016). Structure and organization of the mental lexicon: A network approach derived from syntactic dependency relations and word associations. In A. Mehler, P. Blanchard, B. Job & S. Banish (­Eds.), Towards a Theoretical Framework for Analyzing Complex Linguistic Networks (­­pp. ­47–​­79). Springer. ­Den-​­Heyer, K., & Briand, K. (­1986). Priming single digit numbers: Automatic spreading activation dissipates as a function of semantic distance. American Journal of Psychology, 99(­5), ­315–​­340. Diamond, A. (­2013). Executive functions. Annual Review of Psychology, 64, 1­ 35–​­168. 2018). A consideration of the seven sins of memory in the Ditta, A. S.,  & Storm, B. C. (­ context of creative cognition. Creativity Research Journal, 30(­4), ­402–​­417. https://­doi. org/­10.1080/­10400419.2018.1530914 Dosi, G. (­1988). Sources, procedures, and microeconomic effects of innovation. Journal of Economic Literature, 26(­3), ­1120–​­1171. Dumas, D., & Dunbar, K. N. (­2014). Understanding fluency and originality: A latent variable perspective. Thinking Skills and Creativity, 14, ­56–​­67. https://­doi.org/­10.1016/­j.tsc.2014.09.003 Dumas, D., Organisciak, P., & Doherty, M. (­2020). Measuring originality with human raters and ­text-​ ­m ining models: A psychometric comparison of methods. Psychology of Aesthetics, Creativity and the Arts. https://­doi.org/­10.1037/­aca0000319 Estes, Z. (­2003). Attributive and relational processes in nominal combination. Journal of Memory and Language, 48(­2), ­304–​­319. https://­doi.org/­10.1016/­­S0749-​­596X(­02)­­0 0507-​­7 Faust, M., & Kenett, Y. N. (­2014). Rigidity, chaos and integration: Hemispheric interaction and individual differences in metaphor comprehension [Hypothesis & Theory]. Frontiers in Human Neuroscience, 8(­511), ­1–​­10. https://­doi.org/­10.3389/­f nhum.2014.00511 Finke, R. A., Ward, T. B., & Smith, S. M. (­1992). Creative Cognition: Theory, Research, and Applications. MIT Press. Forster, E. A., & Dunbar, K. N. (­2009). Creativity evaluation through latent semantic analysis. Proceedings of the Annual Conference of the Cognitive Science Society (­­pp. ­602–​­607).UC Merced, CA: Cognitive Science Society. Forthmann, B., Oyebade, O., Ojo, A., Günther, F., & Holling, H. (­2018). Application of latent semantic analysis to divergent thinking is biased by elaboration. The Journal of Creative Behavior, 53(­4), ­559–​­575. https://­doi.org/­10.1002/­jocb.240 Gentner, D. (­1981). Verb semantic structures in memory for sentences: Evidence for componential representation. Cognitive Psychology, 13(­1), ­56–​­83. Ghosh, V. E., & Gilboa, A. (­2014). What is a memory schema? A historical perspective on current neuroscience literature. Neuropsychologia, 53(­0), ­104–​­114. https://­doi.org/­10.1016/­j.neuropsychologia. 2013.11.010 Gray, K., Anderson, S., Chen, E. E., Kelly, J. M., Christian, M. S., Patrick, J., Huang, L., Kenett, Y. N.,  & Lewis, K. (­2019). “­Forward flow”: A new measure to quantify free thought and predict creativity. American Psychologist, 74(­5), ­539–​­554. https://­psycnet.apa.org/­doi/­10.1037/­a mp000 0391 Green, A. E. (­2016). Creativity, within reason: Semantic distance and dynamic state creativity in relational thinking and reasoning. Current Directions in Psychological Science, 25(­1), ­28–​­35. https://­doi. org/­10.1177/­0963721415618485

175

Yoed N. Kenett Griffiths, T. L., Steyvers, M., & Firl, A. (­2007). Google and the mind: Predicting fluency with PageRank. Psychological Science, 18(­12), ­1069–​­1076. https://­doi.org/­10.1111/­j.­1467-​­9280.2007.02027.x Gross, M. E., Zedelius, C. M., & Schooler, J. W. (­2020). Cultivating an understanding of curiosity as a seed for creativity. Current Opinion in Behavioral Sciences, 35, ­77–​­82. Gruszka, A., & Neçka, E. (­2002). Priming and acceptance of close and remote associations by creative and less creative people. Creativity Research Journal, 14(­2), ­193–​­205. ­ ector-​­space models of semantic representation from a Günther, F., Rinaldi, L., & Marelli, M. (­2019). V cognitive perspective: A discussion of common misconceptions. Perspectives on Psychological Science, 14(­6), ­1006–​­1033. Gupta, N., Jang, Y., Mednick, S. C., & Huber, D. E. (­2012). The road not taken: Creative solutions ­ igh-​­frequency responses. Psychological Science, 23(­3), 2­ 88–​­294. https://­doi. require avoidance of h org/­10.1177/­0956797611429710 Hagtvedt, L. P., Dossinger, K., Harrison, S. H., & Huang, L. (­2019). Curiosity made the cat more creative: Specific curiosity as a driver of creativity. Organizational Behavior and Human Decision Processes, 150, 1­ –​­13. Harms, M., ­Reiter-​­Palmon, R., & Derrick, D. C. (­2018). The role of information search in creative problem solving. Psychology of Aesthetics, Creativity and the Arts, 14(­3), ­367–​­380. Hart, Y., Goldberg, H., S­ triem-​­Amit, E., Mayo, A. E., Noy, L., & Alon, U. (­2018). Creative exploration as a s­ cale-​­invariant search on a meaning landscape. Nature Communications, 9(­1), 5411. https://­ doi.org/­10.1038/­­s41467-­​­­018-­​­­07715-​­8 Hart, Y., Mayo, A. E., Mayo, R., Rozenkrantz, L., Tendler, A., Alon, U., & Noy, L. (­2017). Creative foraging: An experimental paradigm for studying exploration and discovery. PLoS One, 12(­8), e0182133. https://­doi.org/­10.1371/­journal.pone.0182133 Hass, R. W. (­2017a). Semantic search during divergent thinking. Cognition, 166, 3­ 44–​­357. https://­doi. org/­10.1016/­j.cognition.2017.05.039 Hass, R. W. (­2017b). Tracking the dynamics of divergent thinking via semantic distance: Analytic methods and theoretical implications. Memory  & Cognition, 45(­2), 2­ 33–​­244. https://­doi.org/­ 10.3758/­­s13421-­​­­016-­​­­0 659-​­y Hassan Awadallah, A., White, R. W., Pantel, P., Dumais, S. T., & Wang, Y.-​­M. (­2014). Supporting Complex Search Tasks. Proceedings of the 23rd ACM international conference on conference on information and knowledge management. He, L., Kenett, Y. N., Zhuang, K., Liu, C., Zeng, R., Yan, T., Huo, T., & Qiu, J. (­2021). The relation between semantic memory structure, associative abilities, and verbal and figural creativity. Thinking & Reasoning, 27(­2), ­268–​­293. https://­doi.org/­10.1080/­13546783.2020.1819415 Heinen, D. J. P., & Johnson, D. R. (­2018). Semantic distance: An automated measure of creativity that is novel and appropriate. Psychology of Aesthetics, Creativity and the Arts, 12(­2), 1­ 44–​­156. https://­doi. org/­10.1037/­aca0000125 Hills, T. T., Jones, M. N., & Todd, P. M. (­2012). Optimal foraging in semantic memory. Psychological Review, 119(­2), 4­ 31–​­4 40. https://­doi.org/­10.1037/­a0027373 Hills, T. T., Kalff, C., & Wiener, J. M. (­2013). Adaptive lévy processes and ­a rea-​­restricted search in human foraging. PLoS One, 8(­4), e60488. https://­doi.org/­10.1371/­journal.pone.0060488 Hills, T. T., & Kenett, Y. N. (­2022). Is the mind a network? Maps, vehicles, and skyhooks in cognitive network science. Topics in Cognitive Science, 14(­1), ­189–​­208. https://­doi.org/­10.1111/­tops.12570 Hills, T. T., Maouene, M., Maouene, J., Sheya, A., & Smith, L. (­2009). Longitudinal analysis of early semantic networks: preferential attachment or preferential acquisition? Psychological Science, 20(­6), ­729–​­739. https://­doi.org/­10.1111/­j.­1467-​­9280.2009.02365.x Hills, T. T., Todd, P. M., Lazer, D., Redish, A. D., Couzin, I. D., & Group, C. S. R. (­2015). Exploration versus exploitation in space, mind, and society. Trends in Cognitive Sciences, 19(­1), 4­ 6–​­54. Hummel, J. E., & Holyoak, K. J. (­1997). Distributed representations of structure: A theory of analogical access and mapping. Psychological Review, 104(­3), ­427– ​­466. Hunter, S. T., ­Bedell-​­Avers, K. E., Hunsicker, C. M., Mumford, M. D., & Ligon, G. S. (­2008). Applying multiple knowledge structures in creative thought: Effects on idea generation and p­ roblem-​­solving [Article]. Creativity Research Journal, 20(­2), ­137–​­154. https://­doi.org/­10.1080/­10400410802088 779 Hutchison, K. A., Balota, D. A., Cortese, M. J., & Watson, J. M. (­2008). Predicting semantic priming at the item level. The Quarterly Journal of Experimental Psychology, 61(­7 ), 1­ 036–​­1066. https://­doi. org/­10.1080/­17470210701438111

176

Creatively Searching Through Semantic Memory Structure James, W. (­1880). Great men, great thoughts, and the environment. Atlantic Monthly, 46, ­4 41–​­459. Jones, M. N., Willits, J.,  & Dennis, S. (­2015). Models of semantic memory. In J. Busemeyer  & J. Townsend (­Eds.), Oxford Handbook of Mathematical and Computational Psychology (­­pp.  ­232–​­254). Oxford University Press. Karuza, E. A. (­2022). The value of statistical learning to cognitive network science. Topics in Cognitive Science, 14(­1), 78–92. https://­doi.org/­10.1111/­tops.12558 Karuza, E. A., ­Thompson-​­Schill, S. L., & Bassett, D. S. (­2016). Local patterns to global architectures: Influences of network topology on human learning. Trends in Cognitive Sciences, 20(­8), ­629–​­640. https://­doi.org/­10.1016/­j.tics.2016.06.003 Kenett, Y. N. (­2018a). Going the extra creative mile: The role of semantic distance in c­ reativity –​ ­theory, research, and measurement. In R. E. Jung & O. Vartanian (­Eds.), The Cambridge Handbook of the Neuroscience of Creativity (­­pp. ­233–​­248). Cambridge University Press. Kenett, Y. N. (­2018b). Investigating creativity from a semantic network perspective. In Z. Kapoula, E. Volle, J. Renoult & M. Andreatta (­Eds.), Exploring Transdisciplinarity in Art and Sciences (­­pp. ­49–​­75). Springer International Publishing. https://­doi.org/­10.1007/­­978-­​­­3 -­​­­319-­​­­76054-​­4 _3 Kenett, Y. N. (­2019). What can quantitative measures of semantic distance tell us about creativity? Current Opinion in Behavioral Sciences, 27, ­11–​­16. https://­doi.org/­10.1016/­j.cobeha.2018.08.010 Kenett, Y. N., Anaki, D.,  & Faust, M. (­2014). Investigating the structure of semantic networks in low and high creative persons. Frontiers in Human Neuroscience, 8(­407), ­1–​­16. https://­doi. org/­10.3389/­f nhum.2014.00407 Kenett, Y. N., Anaki, D., & Faust, M. (­2015). Processing of unconventional stimuli requires the recruitment of the ­non-​­specialized hemisphere. Frontiers in Human Neuroscience, 9, 32. https://­doi. org/­10.3389/­f nhum.2015.00032 Kenett, Y. N., & Austerweil, J. L. (­2016). Examining search processes in low and high creative individuals with random walks. In A. Papafragou, D. Grodner, D. Mirman, & J. C. Trueswell (Eds.), Proceedings of the 38th Annual Meeting of the Cognitive Science Society (pp. 313–318). Philadelphia, PA: Cognitive Science Society. Kenett, Y. N., Baker, B. S., Hills, T. T., Hart, Y., & Beaty, R. E. (­2021). Creative foraging: Examining relations between foraging styles, semantic memory structure, and creative thinking. Proceedings of the 43rd Annual Meeting of the Cognitive Science Society (­­p. 3198). UC Merced, CA: Cognitive Science Society Kenett, Y. N., Beaty, R. E., Silvia, P. J., Anaki, D.,  & Faust, M. (­2016). Structure and flexibility: Investigating the relation between the structure of the mental lexicon, fluid intelligence, and creative achievement. Psychology of Aesthetics, Creativity and the Arts, 10(­4), ­377–​­388. https://­doi. org/­10.1037/­aca0000056 Kenett, Y. N., & Faust, M. (­2019). A semantic network cartography of the creative mind. Trends in Cognitive Sciences, 23(­4), ­271–​­274. https://­doi.org/­10.1016/­j.tics.2019.01.007 Kenett, Y. N., Gold, R., & Faust, M. (­2016). The ­hyper-​­modular associative mind: A computational analysis of associative responses of persons with Asperger Syndrome. Language and Speech, 59(­3), ­297–​­317. https://­doi.org/­0.1177/­0 023830915589397 Kenett, Y. N., Levi, E., Anaki, D.,  & Faust, M. (­2017). The semantic distance task: Quantifying semantic distance with semantic network path length. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43(­9), ­1470–​­1489. https://­doi.org/­10.1037/­x lm0000391 Kenett, Y. N., Levy, O., Kenett, D. Y., Stanley, H. E., Faust, M., & Havlin, S. (­2018). Flexibility of thought in high creative individuals represented by percolation analysis. Proceedings of the National Academy of Sciences, 115(­5), ­867–​­872. https://­doi.org/­10.1073/­pnas.1717362115 Kleinmintz, O. M., Ivancovsky, T., & ­Shamay-​­Tsoory, S. G. (­2019). The twofold model of creativity: The neural underpinnings of the generation and evaluation of creative ideas. Current Opinion in Behavioral Sciences, 27, ­131–​­138. https://­doi.org/­10.1016/­j.cobeha.2018.11.004 Klimesch, W. (­1987). A connectivity model for semantic processing. Psychological Research, 49(­1), ­53–​ ­61. https://­doi.org/­10.1007/ ­BF00309203 Kneeland, M. K., Schilling, M. A., & Aharonson, B. S. (­2020). Exploring uncharted territory: Knowledge search processes in the origination of outlier innovation. Organization Science, 31(­3), ­535–​­557. Kroll, N. E. A., & Klimesch, W. (­1992). Semantic memory: Complexity or connectivity? Memory & Cognition, 20(­2), ­192–​­210. Kumar, A. A. (­2021). Semantic memory: A review of methods, models, and current challenges. Psychonomic Bulletin & Review, 28(­1), ­40– ​­80. https://­doi.org/­10.3758/­­s13423- ­​­­020- ­​­­01792-​­x

177

Yoed N. Kenett Kumar, A. A., Balota, D. A., & Steyvers, M. (­2020). Distant connectivity and m ­ ultiple-​­step priming in ­large-​­scale semantic networks. Journal of Experimental Psychology: Learning, Memory, and Cognition, 46(­12), ­2261–​­2276. https://­doi.org/­10.1037/­x lm0000793 Landauer, T. K., & Dumais, S. T. (­1997). A solution to Plato’s problem: The latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychological Review, 104(­2), ­211–​­240. Landauer, T. K., Foltz, P. W., & Laham, D. (­1998). An introduction to latent semantic analysis. Discourse Processes, 25(­­2 –​­3), ­259–​­284. https://­doi.org/­10.1080/­01638539809545028 Luchins, A. (­1947). Proposed methods of studying degrees of rigidity in behavior. Journal of Personality, 15(­3), ­242–​­246. ­Lydon-​­Staley, D. M., Zhou, D., Blevins, A. S., Zurn, P.,  & Bassett, D. S. (­2021). Hunters, busybodies and the knowledge network building associated with deprivation curiosity. Nature Human Behaviour, 5(­3), ­327–​­336. https://­doi.org/­10.1038/­­s41562-­​­­020-­​­­0 0985-​­7 Mandera, P., Keuleers, E., & Brysbaert, M. (­2015). How useful are ­corpus-​­based methods for extrapolating psycholinguistic variables? The Quarterly Journal of Experimental Psychology, 8, ­1628–​­1642. https://­doi.org/­10.1080/­17470218.2014.988735 Mandera, P., Keuleers, E., & Brysbaert, M. (­2017). Explaining human performance in psycholinguistic tasks with models of semantic similarity based on prediction and counting: A review and empirical validation. Journal of Memory and Language, 92, ­57–​­78. https://­doi.org/­10.1016/­j.jml.2016.04.001 Mao, J., Liu, Y., Kando, N., Zhang, M., & Ma, S. (­2018). How does domain expertise affect users’ search interaction and outcome in exploratory search? ACM Transactions on Information Systems ( ­TOIS), 36(­4), ­1–​­30. Mednick, M., T., Mednick, S., A., & Jung, C., C. (­1964). Continual association as a function of level of creativity and type of verbal stimulus. Journal of Abnormal and Social Psychology, 69(­5), ­511–​­515. Mednick, S., A. (­1962). The associative basis of the creative process. Psychological Review, 69(­3), ­220–​ ­232. http://­w ww.ncbi.nlm.nih.gov/­pubmed/­14472013 Mumford, M. D., Blair, C. S., & Marcy, R. T. (­2002). Alternative knowledge structures in creative thought. In J. C. Kaufman & J. Baer (­Eds.), Creativity and Reason in Cognitive Development (­­pp. ­117–​ 1­ 36). Cambridge University Press. Nelson, D. L., McEvoy, C. L., & Dennis, S. (­2000). What is free association and what does it measure? Memory & Cognition, 28(­6), ­887–​­899. https://­doi.org/­10.3758/­BF03209337 Olson, J. A., Nahas, J., Chmoulevitch, D., Cropper, S. J.,  & Webb, M. E. (­2021). Naming unrelated words predicts creativity. Proceedings of the National Academy of Sciences, 118(­25), e2022340118. https://­doi.org/­10.1073/­pnas.2022340118 ­O vando-​­Tellez, M., Kenett, Y. N., Benedek, M., Bernard, M., Belo, J., Beranger, B., Bieth, T., & Volle, E. (­2022). Brain c­ onnectivity-​­based prediction of r­ eal-​­life creativity is mediated by semantic memory structure. Science Advances, 8(5), eabl4294, 1–16. https://doi.org/10.1126/sciadv.abl4924 Prabhakaran, R., Green, A. E.,  & Gray, J. R. (­2014). Thin slices of creativity: Using s­ingle-​­word utterances to assess creative cognition. Behavior Research Methods, 46(­3), ­641–​­659. https://­doi. org/­10.3758/­­s13428-­​­­013-­​­­0401-​­7 Recchia, G., & Jones, M. N. (­2009). More data trumps smarter algorithms: Comparing pointwise mutual information with latent semantic analysis. Behavior Research Methods, 41(­3), ­647–​­656. https://­ doi.org/­10.3758/­BRM.41.3.647 Rossman, E., & Fink, A. (­2010). Do creative people use shorter association pathways? Personality and Individual Differences, 49, ­891– ​­895. Runco, M. A., & Acar, S. (­2012). Divergent thinking as an indicator of creative potential. Creativity Research Journal, 24(­1), ­66–​­75. https://­doi.org/­10.1080/­10400419.2012.652929 Runco, M. A., & Jaeger, G. J. (­2012). The standard definition of creativity. Creativity Research Journal, 24(­1), ­92–​­96. https://­doi.org/­10.1080/­10400419.2012.650092 Sala, G., & Gobet, F. (­2017). Experts’ memory superiority for d­ omain-​­specific random material generalizes across fields of expertise: A ­meta-​­analysis. Memory & Cognition, 45(­2), ­183–​­193. Schilling, M. A. (­2005). A “­­small-​­world” network model of cognitive insight. Creativity Research Journal, 17(­­2 –​­3), ­131–​­154. https://­doi.org/­10.1080/­10400419.2005.9651475 Schilling, M. A., & Green, E. (­2011). Recombinant search and breakthrough idea generation: An analysis of high impact papers in the social sciences. Research Policy, 40(­10), ­1321–​­1331. Schutte, N. S., & Malouff, J. M. (­2020). A meta-analysis of the relationship between curiosity and creativity. The Journal of Creative Behavior, 54(­4), ­940–​­947.

178

Creatively Searching Through Semantic Memory Structure Scott, G. M., Lonergan, D. C.,  & Mumford, M. D. (­2005). Conceptual combination: Alternative knowledge structures, alternative heuristics. Creativity Research Journal, 17(­1), ­79–​­98. Siew, C. S. Q. (­2019). spreadr: An R package to simulate spreading activation in a network. Behavior Research Methods, 51(­2), ­910–​­929. Siew, C. S. Q., Wulff, D. U., Beckage, N. M., & Kenett, Y. N. (­2019). Cognitive network science: A review of research on cognition through the lens of network representations, processes, and dynamics. Complexity, 24, Article 2108423. https://­doi.org/­10.1155/­2019/­2108423 Silvia, P. J. (­2015). Intelligence and creativity are pretty similar after all. Educational Psychology Review, 27(­4), ­1–​­8. https://­doi.org/­10.1007/­­s10648-­​­­015-­​­­9299-​­1 Simmons, S., & Estes, Z. (­2006). Using Latent Semantic Analysis to Estimate Similarity. Proceedings of the 28th Cognitive Science Society, Austin, TX. Simon, H. A., & Chase, W. G. (­1973). Skill in chess. American Scientist, 61(­4), ­394–​­403. Simonton, D. K. (­2018). Creative genius as causal agent in history: William James’s 1880 theory revisited and revitalized. Review of General Psychology, 22(­4), ­406– ​­421. https://­doi.org/­10.1037/­g pr0000165 Sio, U. N., & Rudowicz, E. (­2007). The role of an incubation period in creative problem solving. Creativity Research Journal, 19(­­2 –​­3), ­307–​­318. Smith, K. A., Huber, D. E., & Vul, E. (­2013). ­Multiply-​­constrained semantic search in the Remote Associates Test. Cognition, 128(­1), ­64–​­75. https://­doi.org/­10.1016/­j.cognition.2013.03.001 Smith, K. A., & Vul, E. (­2015). The Role of Sequential Dependence in Creative Semantic Search. Topics in Cognitive Science, 7(­3), ­543–​­546. https://­doi.org/­10.1111/­tops.12152 Smith, S. M., & Blankenship, S. E. (­1991). Incubation and the persistence of fixation in problem solving. The American Journal of Psychology, 104(­1), ­61–​­87. Sowden, P. T., Pringle, A., & Gabora, L. (­2014). The shifting sands of creative thinking: Connections to ­dual-​­process theory. Thinking  & Reasoning, 21(­1), ­40– ​­60. https://­doi.org/­10.1080/­13546783. 2014.885464 Stella, M., & Kenett, Y. N. (­2019). Viability in multiplex lexical networks and machine learning characterizes human creativity. Big Data and Cognitive Computing, 3(­3), 45. https://­doi.org/­10.3390/ ­bdcc3030045 Steyvers, M., & Tenenbaum, J. B. (­2005). The large scale structure of semantic networks: Statistical analysis and a model of semantic growth. Cognitive Science, 29(­1), ­41–​­78. https://­doi.org/­10.1207/­ s15516709cog2901_3 Todd, P. M., & Hills, T. T. (­2020). Foraging in mind. Current Directions in Psychological Science, 29(­3), ­309–​­315. https://­doi.org/­10.1177/­0963721420915861 Vankrunkelsven, H., Verheyen, S., Storms, G., & De Deyne, S. (­2018). Predicting lexical norms: A comparison between a word association model and ­text-​­based word ­co-​­occurrence models. Journal of Cognition, 1(­1), 45. https://­doi.org/­http://­doi.org/­10.5334/­joc.50 Vitevitch, M. S., Chan, K. Y., & Roodenrys, S. (­2012). Complex network structure influences processing in ­long-​­term and ­short-​­term memory. Journal of Memory and Language, 67(­1), ­30–​­4 4. https://­ doi.org/­10.1016/­j.jml.2012.02.008 Vitevitch, M. S., Chan, K. Y., & Goldstein, R. (­2014). Insights into failed lexical retrieval from network science. Cognitive Psychology, 68, ­1–​­32. https://­doi.org/­10.1016/­j.cogpsych.2013.10.002 Volle, E. (­2018). Associative and controlled cognition in divergent thinking: Theoretical, experimental, neuroimaging evidence, and new directions. In R. E. Jung & O. Vartanian (­Eds.), The Cambridge Handbook of the Neuroscience of Creativity (­­pp. ­333–​­362). Cambridge University Press. Ward, T. B., Patterson, M. J., & Sifonis, C. M. (­2004). The role of specificity and abstraction in creative idea generation. Creativity Research Journal, 16(­1), ­1–​­9. https://­doi.org/­10.1207/­s15326934crj1601_1 Wiley, J. (­1998). Expertise as mental set: The effects of domain knowledge in creative problem solving. Memory & Cognition, 26(­4), ­716–​­730. https://­doi.org/­10.3758/ ­BF03211392 Wisniewski, E. J. (­1997). When concepts combine. Psychonomic Bulletin & Review, 4(­2), ­167–​­183. https://­ doi.org/­10.3758/­BF03209392

179

12 MENTAL IMAGERY AND CREATIVE COGNITION David G. Pearson

The experience of mental imagery has a long history of association with creative cognition. This includes aspects of everyday p­ roblem-​­solving (­Finke, 1994; Kaufmann, 1988), the conceptual stage of architectural design (­Purcell & Gero, 1998; Reed, 1993), and the visualisation and development of scientific models (­M iller, 1984). The term “­mental imagery” describes the ability of the brain to simulate or r­e-​­create aspects of perceptual experience (­Pearson, 2007). From the time of the philosopher Aristotle onwards, mental imagery has been linked to perception and the associated phenomenology of perceptual experience persisting in the absence of external sensory input (­Roth, 2007). There can be considerable individual differences in the phenomenological experience of imagery, including its frequency of occurrence, vividness, and level of detail (­A ndrade et al., 2014; Marks, 1973; Reisberg et  al., 2003). In the case of aphantasia, individuals can even report experiencing no conscious mental imagery at all (­Zeman et  al., 2020). However, despite such variation in the subjective experience of mental imagery, for many, it appears to be a normative representational tool across a wide range of different cognitive processes (­Dawes et  al., 2020). This includes autobiographical memory (­Greenberg  & Knowlton, 2014), ­decision-​­making for future actions (­­Blouin-​­Hudon & Pychyl, 2015), and making perceptual judgements in the absence of external referents. For example, when asked whether an elephant has a long or short tail, many people respond by visualising the appearance of an elephant from memory (­Farah et al., 1988). While the most commonly reported forms of imagery are in the form of visual and auditory experiences (­Betts, 1909; Tiggemann & Kemps, 2005), mental images can be generated in all sensory modalities (­A ndrade et al., 2014). This includes haptic imagery ( ­Juttner  & Rentschler, 2002), gustatory imagery (­Tiggemann  & Kemps, 2005), and olfactory imagery (­Stevenson & Case, 2005). A distinction can also be drawn between “­general” and “­specific” mental images (­De Beni et al., 2007; Helstrup et al., 1997). General images are more prototypical than specific images and afford a greater variety in terms of what can be subjectively experienced. A functional Magnetic Resonance Imagery (­f MRI) study conducted by Gardini et al. (­2005) suggests that the generation of both types of imagery involves different neural pathways, with general images typically generated more quickly than specific images. Visual imagery is believed to rely on pictorial rather than propositional forms of mental representation (­Kosslyn, 1994; Pearson & Kosslyn, 2015). Eye movement patterns during mental imagery have been 

180

DOI: 10.4324/9781003009351-14

Mental Imagery and Creative Cognition

shown to be similar to those made during perception of pictures depicting the same content, suggesting that oculomotor mechanisms play a role during the experience of mental imagery (­Brandt & Stark, 1997; Gurtner et al., 2021)

Mental Imagery and Anecdotal Accounts of Creative Discovery One of the most famous examples of a mental image apparently supporting creative discovery is Fredrich August von Kekule’s insight into the molecular structure of benzene (­Gardner, 1993; Miller, 2000; Weisberg, 1986). Kekule made the revolutionary proposal in 1865 that benzene comprises a ring of six carbon atoms. He gave the following account of how this scientific discovery was made: I turned my chair to the fire and dozed. Again the atoms were gambolling before my eyes… My mental eye rendered more acute by repeated visions of this kind, could now distinguish larger structures, of manifold conformation; long rows, sometimes more closely fitted together; all twining and twisting in snakelike motion. But look! What was that? One of the snakes had seized hold of its own tail, and the form whirled mockingly before my eyes. As if by a flash of lightening I awoke. (­Reproduced in Weisberg 1986, ­p. 32) Kekule’s account of his scientific discovery has been widely interpreted in the creative thinking literature as demonstrating that the mental image of a snake biting its own tail directly led to the insight that benzene’s carbon atoms were organised in a circular structure. However, some authors have disputed the strength of this interpretation. Wotiz and Rudofsky (­1954; discussed in Miller, 2000) claimed that Kekule may have concocted the story as a romanticised anecdote of scientific discovery when writing the preface to a lecture he delivered in Berlin in 1890, considering there is no record of the same account in the 30 years prior to this. Weisberg (­1986) also points out that even if Kekule’s account is genuine, he may have used the term “­snakelike” figuratively rather than descriptively, with some aspects of the story misrepresented during the process of translating the original German into English. The debate around the nature of what led to Kekule’s discovery illustrates the potential danger of taking such anecdotal descriptions of imagery supporting creative discovery too much at face value. However, the frequency with which imagery is mentioned in connection with personal accounts of creative thinking is striking. This includes accounts of architectural design (­Goldschmidt, 1991; Purcell  & Gero, 1998), sculpture (­Samuels  & Samuels, 1975), painting (­Finke, 1986), and writing (­Nin, 1969). An example of auditory imagery supporting creativity is mentioned in the first biography of Mozart written by Franz Niemetschek in collaboration with Mozart’s wife Constanze: Mozart wrote everything with a facility and rapidity, which perhaps at first sight could appear as carelessness or haste; and while writing he never came to the klavier. His imagination presented the whole work, when it came to him, clearly and vividly…. In the quiet repose of the night, when no obstacle hindered his soul, the power of his imagination became incandescent with the most animated activity, and unfolded all the wealth of tone which nature had placed in his spirit… Only the person who heard Mozart at such times knows the depth and the whole range of his musical genius: free and independent of all concern his spirit could soar in daring flight to the highest regions of art. (­Franz Xavier Niemetschek, 1798, ­pp. ­54–​­55) 181

David G. Pearson

Some of the most compelling accounts originate from scientific fields such as physics, chemistry, and engineering. Albert Einstein was a vocal proponent of the use of mental imagery, or “­v isualisation”, as a technique for reasoning about theoretical problems in physics. He combined visual thinking with “­thought experiments” (­from the German gedankenexperiment), in which he reflected on the consequences of simulated actions performed in imagined scenarios. For example, in 1895, at the age of 15, Einstein conceived of a classic thought experiment in which he imagined what an observer would perceive if they were able to match the velocity of a wave of light (­Gardner, 1993). In another thought experiment reported in 1907, Einstein imagined the possible consequences of an observer jumping from the roof of a house while simultaneously dropping a stone. I­ magery-​­based thought experiments such as these led Einstein to establish a “­principle of equivalence” that later became a key element of his general theory of relativity (­M iller, 2000). When asked to provide an account of the cognitive processes underlying his scientific reasoning, he offered the following description: The words of the language as they are written or spoken, do not seem to play any role in my mechanism of thought. The psychical entities which seem to serve as elements in thought are certain signs and more or less clear images which can be ‘­voluntarily’ reproduced and combined… From a psychological viewpoint this combinatory play seems to be the essential feature in productive thought. (­Reproduced in Gardner, 1993, ­p. 105) Stephen Hawking was also a strong advocate for the power of visualisation to support scientific reasoning and creative discovery. In his collection Black Holes and Baby Universes and Other Essays, he writes: I was sure that nearly everyone was interested in how the universe operates, but most people cannot follow mathematical ­equations – ​­I don’t care much for equations myself. This is  …. because I don’t have an intuitive feeling for equations. Instead, I think in pictorial terms, and my aim in the book was to describe these mental images in words, with the help of familiar analogies and a few diagrams. (­Hawking, 1993, p­ . 35) Other examples of prominent scientists using mental imagery to support creative thinking include Richard Feynman’s use of visualisation to represent mathematical arguments (­M iller, 1984), Nikola Tesla using imagery to mentally assemble and manipulate designs for complex electric motors (­M iller, 2000), and Michael Faraday mentally visualising lines of electromagnetic force (­Gooding, 1991).

Imagery and Creative Mental Synthesis A common thread running through the anecdotal reports of mental imagery described above is that creative insight appears linked to the active manipulation and transformation of mental images. Currie and Ravenscroft (­2002) refer to mental imagery playing a role in what they term “­recreative” imagination, which is the capacity to imagine something from a different viewpoint or perspective. Albert Einstein attributed creative insight to the process of “­combinatory play” (­Gardner, 1993), while the architect Bryan Lawson describes reasoning during creative design as involving “­a highly organised mental process capable of manipulating many kinds of information, blending them all into a coherent set of ideas, and finally 182

Mental Imagery and Creative Cognition

generating some realisation of those ideas (­Lawson, 1980, ­p. 6). This process of establishing new constructs through the combination of existing ideas has been described by the term mental synthesis (­Finke & Slayton, 1988; Thompson & Klatzky, 1978). Mental synthesis has been experimentally studied using a variety of different methodologies and techniques, including an influential programme of research on creative cognition conducted by Ronald Finke and colleagues (­Finke, 1989, 1994; Finke et al., 1992). In one series of experiments, Finke et al. created a guided synthesis task in which participants were required to mentally manipulate simple a­ lpha-​­numeric and geometric symbols in response to a sequence of verbal instructions provided by the experimenter (­Finke et al., 1989). If the sequence of instructions was followed correctly, then the resulting image was designed to resemble a familiar object or scene. At the end of each trial, participants were asked to draw their image on a piece of paper to establish whether they had followed the verbal instructions correctly or not. However, this drawing was only allowed after participants had attempted to identify their mental image. Finke et al. were interested in establishing whether participants would be able to recognise emergent patterns in their mental images without reliance on any form of external picture or sketch. Their results showed that participants were able to successfully reinterpret their mental images as resembling familiar objects or scenes on 70% of trials. These successful reinterpretations were accomplished in the absence of any perceptual support from pictures or drawings. A further 21% of trials were classified as “­partial transformations” as an error had occurred in following one or more of the verbal instructions. Out of these partial transformation trials, only 13% were correctly identified as resembling a recognisable figure. The remaining 19% participants made multiple errors in following the verbal instructions correctly and produced a final mental image that was substantially different from the intended state. For these trials, the number of correct identifications fell to zero. The implication of these findings was that participants were able to reinterpret their images as resembling familiar objects or scenes without any additional support from any external source. Finke concluded that mental imagery on its own was sufficient to provide the basis for creative discoveries in the absence of any external pictures or drawings (­Finke, 1994; Finke et al., 1989, 1992). A potential criticism of this interpretation is that participants might be able to guess the identity of the final pattern from the initial geometric and ­a lpha-​­numeric symbols rather than the final synthesised mental image (­for example, they guess a “­J” resembles the handle of an umbrella). However, this criticism appears unlikely, considering Finke et al. found participants’ ability to correctly identify the final pattern dropped sharply if they failed to correctly follow the verbal manipulation instructions. Finke et al. also ran a control study in which participants were asked to try and guess the identity of the final pattern after each verbal instruction was given. They found that none of the participants were able to correctly guess the synthesised pattern after just the first instruction (­Finke et al., 1989), and this rose to only 4% of correct identifications after the second. Another form of the experimental mental synthesis paradigm is the creative synthesis task devised by Finke and Slayton (­1988). In this procedure, participants were first presented with a set of 15 a­ lpha-​­numeric and geometric symbols, which were learnt until they could be accurately imaged in response to verbal descriptions alone. Following this, for each experimental trial, participants were verbally presented with three symbols randomly selected from the stimuli set of 15 (­i.e., “­square”, “­circle”, and capital “­V”). They were instructed to form visual images of the named symbols and were given up to two minutes to mentally combine the symbols into a novel recognisable pattern. The procedure required all named symbols to be incorporated in the final pattern, and it was not permissible to distort the original shape 183

David G. Pearson

(­e.g., stretch the square into a rectangle), but otherwise participants were free to change the size and orientation of the symbols and combine them in any way they wished. Once a participant indicated that they had been successful in creating a recognisable pattern, they were first asked to provide a verbal description before drawing the synthesised pattern on paper. This aspect of the procedure was to ensure that any ­imagery-​­based discoveries were not supported by the act of drawing. Using this paradigm, Finke and Slayton found that participants were able to produce a recognisable pattern on 40% of the creative synthesis trials. A related study conducted by Pearson et al. (­1999) used a more complex version of the creative synthesis task, in which five symbols were presented instead of three and all participants received the same sets of symbols. Despite these changes, participants were still able to produce diverse and creative synthesised patterns using the same sets of symbols. The specific cognitive processes that underlie the performance of mental synthesis remain a topic of debate. R ­ oskos-​­Ewoldsen (­1993) suggests the structural complexity of parts being combined during a synthesis task can significantly constrain the likelihood of creative discoveries being reached using mental imagery alone. Using a more complex version of the Finke et al. (­1989) guided synthesis task, Irving et al. (­2011) also found that the number of mental transformations could be a constraining factor, as participants performance decreased as the number of required image manipulations increased. Behrmann et al. (­1994) reported a patient suffering from severe visual agnosia could still successfully complete the guided synthesis task using mental imagery on its own, suggesting a dissociation between mental imagery and perceptual processes. Riquelme (­2002) found that participants who performed well on a creative image combination task were also significantly better at reinterpreting perceptually ambiguous figures, such as the ­duck-​­rabbit figure, than participants with poorer image combination performance. The ease of reinterpreting ambiguous figures has been linked to both ­self-​­rated and objective measures of creative thinking (­Doherty  & Mair, 2012; Wiseman et al., 2011). Furthermore, Sagone et al. (­2020) have reported in a sample of ­6 –­​­­12-­​­­year-​­old children that there is a significant positive association between their mental synthesis performance and measures of elaboration, flexibility, and originality derived from tests of creative thinking.

Internal and External Representations during Creative Cognition Although in the studies conducted by Finke and others, participants were required to try and make creative discoveries using mental imagery alone, in everyday life, mental imagery often interacts with external representations such as drawings, sketches, and models. The teaching of design often encourages students to use unstructured sketching as a means to further creativity and innovation (­Goel, 1995; Herbert, 1988), and reports in the literature indicate the pervasive use of sketching during creative design practice (­Do et  al., 2000; Goldschmidt, 1991). The active manipulation and transformation of mental imagery can be considered a form of working memory procedure (­Pearson, 2001; Pearson et al., 2001). Working memory refers to the temporary memory systems involved during cognitive tasks such as reasoning, learning, and comprehension (­Baddeley, 1990; Logie et al., 2020). An influential model of working ­ ulti-​­component approach first proposed by Baddeley and Hitch (­1974). memory is the m As originally stated, the model comprised three separate, l­imited-​­capacity components: the phonological loop, the visuospatial sketchpad, and the central executive. The phonological loop and sketchpad are ­modality-​­specific “­slave systems” that temporarily retain verbal and visuospatial material, respectively. Both are controlled by the central executive component, 184

Mental Imagery and Creative Cognition

an amodal a­ ttention-​­based system that coordinates the activity of the slave systems and is involved in strategy selection and the planning of complex cognitive tasks. The visuospatial component of working memory has been shown to be involved during memory for sequences of movements (­e.g., ballet moves; ­Rossi-​­Arnaud et al., 2004), the generation of visual mental images (­Baddeley & Andrade, 2000), and memory for the positions of objects (­Cattaneo et al., 2006). Baddeley (­2000) proposed a fourth component, the episodic buffer, that functions to bind together information from working memory and l­ong-​­term memory into unitary multimodal episodic representations (­Baddeley, 2000; Hitch et al., 2020). Despite widespread use of external representations during creative p­roblem-​­ solving, contemporary models of working memory seldom address the issue of how external representations interact with internal mental representations such as imagery during creative cognition. Studies that do address this interaction have tended to focus on the relationship between the formation of mental images and external auditory verbal descriptions (­e.g., Borst et al., 2012; Brandimonte & Gerbino, 1996; Denis & Cocude, 1992; Denis et al., 2002) rather than external visuospatial representations. As mental images are subject to rapid decay, active maintenance of an image is required for any inspection or transformation processes to be performed (­Kosslyn, 1994). Such maintenance processes in mental imagery are dependent on general attentional resources that can become rapidly depleted (­Borst et al., 2012; Logie, 1995; 2011; Logie & van der Meulen, 2009; Pearson, 2007; Pearson et al., 2001). Performance on a guided mental synthesis task has been shown to place a high load on ­attention-​ ­based central executive resources in working memory (­Pearson et al., 1996). Barquero and Logie (­1999) claim that successful creative insight during synthesis tasks can be constrained both by quantitative factors (­overall cognitive load) and qualitative factors (­semantic properties of the symbols being combined), and that both of these restrictions can be mitigated through the use of external representations. Bilda and Gero (­2007) suggest that restrictions in working memory capacity may impact on the conceptual stage of the design process, and that the use of external representations such as sketches can allow designers to “­­off-​­load” information and reduce cognitive demands placed on visuospatial working memory. This argument relates to the concept of “­computational ­off-​­loading” proposed by Scaife and Rogers (­1996), in which external representations help reduce the overall cognitive load experienced during completion of a task. External representations may also serve to increase participants’ ability to change the reference frame in which a synthesised mental pattern is interpreted. Daniel Reisberg (­1996; Reisberg & Logie, 1993) argues that mental imagery, unlike externally perceived pictures, cannot be inherently ambiguous. While external visual representations such as drawings and sketches can be interpreted relatively flexibly by the viewer, internally generated mental images are interpreted within a specific frame of reference specifying elements such as figure/­ ground organisation and orientation. Together, these elements determine how the mental image should be “­correctly” interpreted. For example, the ambiguous d­ uck-​­rabbit figure first devised by Jastrow (­1899) can be interpreted as representing either a duck or a rabbit, with both interpretations corresponding to opposing frames of reference. Interpreting the figure as a rabbit involves a reference frame in which the orientation of the figure runs from left to right, with the right side representing the front of the rabbit and the left side the back. In contrast, the alternative interpretation of the figure as a duck involves a different reference frame in which the perceived orientation of the figure is reversed. Clark (­2003, 2008) proposes that the externalisation of mental imagery through drawing and sketching allows participants to manipulate and transform imagery in ways that cannot easily be accomplished using internal mental representations alone. For example, Wijntjes 185

David G. Pearson

et al. (­2008) found that participants were significantly better at identifying a sketch drawn from memory of the haptic outline of an object rather than when they were perceiving the outline using touch alone. A further demonstration can be found in a study conducted by Chambers and Reisberg (­1985) using Jastrow’s ambiguous ­duck-​­rabbit figure. They found that none of their participants were able to successfully reinterpret the figure using a mental image generated from a brief visual presentation of the ­duck-​­rabbit picture. Those who initially perceived the figure as representing a rabbit were unable to reinterpret their mental image as representing anything else, and similarly for those who initially interpreted the figure as representing a duck. Strikingly, however, all participants were able to reinterpret their image almost immediately once they were allowed to sketch it on a piece of paper. Chambers and Reisberg argued that while discoveries and manipulations compatible with an existing reference frame can be accomplished fairly easily when based on an internal mental image alone, discoveries that are incompatible are much more difficult unless supported by an external representation such as sketching. ­Vallée-​­Tourangeau and March (­2019) have described these types of experiences as outsights or fi ­ rst-​­order ­problem-​­solving, in which solutions are derived through direct manipulation of objects in the external world. In contrast, ­second-​­order ­problem-​­solving relies solely on mental resources. Using a guided mental synthesis procedure developed by Irving et al. (­2011), V ­ allée-​­Tourangeau et al. (­2021) found over 100 instances of participants being able to identify synthesised patterns only after being allowed to sketch them, but not when relying solely on their mental imagery of the pattern. While the act of sketching may be considered dynamic, the external representations created from sketches and drawings remain static in nature, even when unstructured and spontaneous. In contrast, creative mental imagery can be conceived as being predominantly a dynamic process (­Pearson & Logie, 2000; Pearson et al., 1999, 2013). While sketching may be able to supplement the use of mental imagery during creative ­problem-​­solving, it cannot fully externalise all aspects of internal mental transformations. Kavakli and Gero (­2001) suggest that differences in the efficiency of using mental imagery can account for differences in cognitive activity between expert and novice designers, even when both use sketching during the design process. Mast and Kosslyn (­2002) have also found that participants’ performance on dynamic mental rotation tasks is significantly associated with their ability to successfully reinterpret perceptually ambiguous figures such as the d­ uck-​­rabbit figure. The inability of sketching to fully externalise dynamic mental imagery processes may account for why some studies of creative ­problem-​­solving have failed to detect any benefit of sketching in comparison to conditions where tasks are performed entirely mentally. Using a mental synthesis experimental paradigm (­Finke, 1994), Kokotovich and Purcell (­2000) found no significant differences between designers and n ­ on-​­designers when comparing drawing and ­non-​­drawing conditions. This replicated a similar finding reported previously by Anderson and Helstrup (­1993). Using protocol analysis of expert architects, Bilda et al. (­2006) also found no significant differences between drawing and ­no-​­drawing conditions during the conceptual phase of design tasks. A study conducted by Verstijnen et  al. (­1998) also failed to find a significant benefit of drawing on the performance of a mental ­synthesis-​­based task. Why is it that drawing seems to be ineffective in supporting mental synthesis performance but can be effective in increasing rates of successful reinterpretation for perceptually ambiguous figures? Verstijnen et al. account for this difference by drawing a distinction between the combining of different visual elements, such as occurs during mental synthesis, and the restructuring of a mental image, which they argue can be enhanced by sketching. They define combining as involving 186

Mental Imagery and Creative Cognition

known components being joined together in the absence of any change to the initial structural interpretation of the components, while restructuring involves the decomposition of components into incidental parts not previously known to exist within a configuration. It is specifically the ability to change the structural interpretation of visual patterns that Verstijnen et al. claim can be enhanced by externalisation, while the combining of components without such restructuring can occur relatively independently from any effect of externalisation such as sketching. Zhang (­1997) has postulated a “­­representational-​­determinism” in ­problem-​­solving, in which the type of representations available to participants determine the mental processes that are activated and, consequently, the type of discoveries and solutions most likely to be produced. Therefore, the form of externalisation available to participants during a creative synthesis task may be more important than simply its presence or absence. A study conducted by Pearson and Logie (­2015) tested this hypothesis by asking participants to complete trials of a guided synthesis task under different conditions of external support. The synthesis trials were performed either using mental imagery alone, drawing manipulations in the air with a finger, sketching manipulations on a piece of paper, or dynamically performing the manipulations ­on-​­screen using a graphics package. The results showed that the number of synthesised patterns correctly interpreted as recognisable objects was significantly higher in the dynamic support condition in comparison to all other conditions. This is consistent with static support such as sketching only externalising the results of mental imagery transformation, while dynamic support completely externalises all aspects of a transformation. For example, in the “­u mbrella” trial created by manipulating a “­J” and a “­D”, if participants in the sketching condition are instructed to “­d raw the capital letter ‘­D’ rotated 90 degrees to the left”, first they must imagine what the product of that manipulation looks like before they can draw the results of the manipulation. In contrast, when the synthesis trials are completed using a Personal Computer graphics package, all aspects of the trial, including the dynamic transformation, are externally represented. Kavakli and Gero (­2001) characterise sketching as a form of mental imagery processing and argue that individual differences between novice and expert designers may reflect differential cognitive activity in mental imagery processes. Successful insight in a guided synthesis task requires participants to detect an emergent configuration in the final pattern that is unrelated to the interpretation of its original constituent parts. Studies of sketching during creative design reveal an interplay between transformations in mental imagery and the use of external sketches (­Bilda et al., 2006; Fish & Scrivener, 1990; Goldschmidt, 1991). A consequence of such interplay may be that the interpretation of sketches that derive from mental imagery transformations remains influenced by frames of reference and semantic properties associated with mental imagery. For example, in the “­u mbrella” guided synthesis example, participants need to shift from interpreting the pattern as a rotated “­D” on top of a “­J” and instead detect a resemblance to the perceptual appearance of an umbrella. The findings of Pearson and Logie (­2015) suggest that this type of reinterpretation may be harder using a sketch that results from an imagined transformation in comparison to an externally performed transformation, even when the resulting patterns from both transformations represent the same thing. Dynamically perceived motion can also change the frame of reference in which perceptually ambiguous figure are perceived. For example, the ­duck-​­rabbit figure is more likely to be interpreted as a duck if it is perceived as moving from right to left, and as a rabbit if perceived as moving from left to right (­Bernstein & Cooper, 1997). Thomas (­2013) proposes that spatial working memory enables crucial “­­cross-​­talk” between performed actions such as eye movements and 187

David G. Pearson

higher order cognition that prime the correct solution in classic insight problems such as the radiation problem (­Duncker, 1945). In the context of mental synthesis performance, dynamic external support may serve to reduce or remove the need for spatial working memory to enable such “­­cross-​­talk”, allowing external representations to directly influence insight ­problem-​­solving processes.

“­Illusions of Imagery”: Cognitive Constraints on Mental Simulation Anecdotal descriptions of mental imagery experienced during creative cognition imply that its value may lie in the ability to mentally simulate the possible consequences of imagined actions (­Pearson, 2007). Shepard (­1978) proposed that imagining objects and how they are affected by transformations in space allows us to explore the consequences of actions without having to carry out the operations and transformations in physical reality. However, are there cognitive constraints on the accuracy of such mental simulations? Extensive research in domains such as perception (­e.g., Carbon, 2014) and ­decision-​­making (­e.g., Tversky, 1986) has demonstrated how heuristics employed in cognitive systems can give rise to the occurrence of systematic errors, but the operation of such processes within mental imagery is less well documented (­Ekroll, 2019; Haugland et al., 2021). One candidate for such a systematic illusion occurring in the domain of mental imagery is the cosmological model proposed by Tycho Brahe in the late 16th century (­Blair, 1990). In the Tychonic system, the planets moved around the Sun in a manner identical to the model proposed earlier by Copernicus in 1543, but at the same time, the Sun (­accompanied by the movements of the other planets) moved around the Earth. Tycho’s model was attractive to many scholars in the early 17th century as it seemed to preserve elements of Copernicus’s theory while avoiding the heresy that the Earth was not the centre of the universe. However, in the Tychonic system, there was an apparent intersection between the orbit of the Sun and Mars. This collision between the orbits was therefore inconsistent with the traditional view at the time that the Sun and planets were embedded within a system of solid nested spheres. However, Margolis (­1998a) was able to demonstrate that this perceived intersection between the orbits of the Sun and Mars is in fact illusionary. This was done by separating out the classic Tychonic sphere diagram into two components: one that contains everything that shares the Sun’s annual motion and nothing else, and the other containing only the Earth and the Moon. Margolis found that when one component of the diagram was physically rotated around the other, the illusion of an apparent intersection between two orbits was instantly dispelled. Strikingly, Margolis found that even modern experts still strongly perceived the illusionary collision until they could physically manipulate the Tychonic diagram themselves (­Margolis, 1998b). How could such a systematic error in understanding the Tychonic system persist for over 400 years? The Tycho illusion highlights a discrepancy between trying to reach a conclusion based on an externally derived percept and trying to reach the same conclusion based on internal mental imagery simulating the same information (­Pearson, 1998). The work of Margolis demonstrates that perception of a dynamic model of Tycho’s system can immediately remove the illusory collision, but that perception of a static representation of the same model is insufficient to support an equivalent insight. In terms of the Tychonic system, the mental imagery generated to represent the model is interpreted within a frame of reference based on an incorrect understanding of how the different elements dynamically move relative to each other. For modern scholars, the illusion is further strengthened by the prior knowledge that the illusory collision “­exists”, namely, that it has previously been deduced by others. 188

Mental Imagery and Creative Cognition

A topic related to this is the study of a class of magic tricks loosely defined as “­topological tricks” because they involve flexible materials (­Gardner, 2014). Ekroll (­2019) argues that such tricks represent “­illusions of imagery” and derive from cognitive limitations in our ability to mentally simulate or imagine deformations of flexible objects. Crucially, the mental representations of deformable objects created by a magician’s audience have fewer degrees of freedom than the external objects they are meant to represent. Ekroll’s working definition of a topological trick is “­a trick that is based, at least in part, on limitations, principles and heuristics in the mental processing of topological properties, such as connectedness and the possible transformations of bendable objects.” (­Ekroll, 2019, ­p. 19). The experience of such magic tricks derives from the inability of audience members to successfully imagine the secret method(­s) that underpin the trick’s effect (­Leddington, 2016). One example of an “­illusion of imagery” involves the Mobius band (­Gardner, 1977). A Mobius band is produced by joining two ends of a paper strip into a loop after twisting one of the ends by 180 degrees. If the band is then cut around the middle, it produces a single large unbroken loop instead of two smaller loops. Most people find this result highly c­ ounter-​ ­intuitive, and the principle forms the basis of a magical routine known as The Afghan Bands (­Gardner, 1977; Wilson, 1988). A research study conducted by Haugland et al. (­2021) found that 80% of a sample of undergraduate students erroneously predicted that cutting a Mobius band along the middle would produce two separate pieces instead of a single unbroken loop. Haugland et al. argue that this misleading intuition reflects the cognitive limitations of mental imagery when attempting to accurately simulate the consequences of cutting a Mobius band in real life.

Perceptual Specificity in Mental Imagery An additional constraint on mental imagery during creative cognition is the nature of the perceptual experiences from which the images derive. Constructive accounts of memory characterise the recollection of previously experienced events as involving a synthesis of features processed within different cortical regions (­e.g., Schacter et al., 1998; Slotnick, 2004). The computational theory of mental imagery proposed by Kosslyn adopts a similar approach, in which visual mental images are constructed from a variety of different pieces of information held in l­ong-​­term memory and are therefore not akin to “­mental photographs” (­Kosslyn et  al., 2006). In Kosslyn’s computational model, the transformation and manipulation of mental images involves a modification of mapping functions between pattern activation subsystems in ­long-​­term memory and a visual buffer structure in which consciously experienced mental images are represented (­Kosslyn, 1994; Pearson & Kosslyn, 2015). The occurrence of “­perceptual specificity” effects in memory reflects the process by which recollection of an event partially reactivates the same brain regions involved during the original experience (­Buckner & Wheeler, 2001; Danker & Anderson, 2010). Recognition memory for colour and ­black-­​­­and-​­white photographs of natural scenes is significantly impaired when the presentation at recall is incongruent with how the photograph was first presented (­i.e., colour at presentation vs. black and white at recall, or vice versa; Wichmann et al., 2002). Recognition memory for scenes is also impaired by incongruent changes in viewing mode between presentation and recall (­Ray & Reingold, 2003; Reingold, 2002). Findings such as these suggest that visual memory representations may incorporate surface perceptual information present during their initial encoding. If mental images are generated using the same visual memory representations, it can be expected that similar perceptual specificity effects might occur. Hitch et al. (­1995) examined 189

David G. Pearson

participants’ ability to mentally combine visual images of pairs of line drawings that were presented using either ­white-­​­­on-​­black or ­black-­​­­on-​­white contrast. They found that performance on the imagery task was significantly impaired following immediate presentation of the pairs of drawings if the contrast was incongruent (­for example, combining ­white-­​­­on-​ ­black images and ­black-­​­­on-​­white drawings). The same impairment persisted following a retention period, but only if participants were prevented from verbally labelling the figures during initial presentation. A ­follow-​­up study conducted by Walker et al. (­1997) found that the imagery impairment was specific to incongruence between the colour of the foreground figures rather than the background, implying that the underlying memory representations were ­object-​­based rather than purely pictorial in nature. A series of experiments carried out by Pearson and Hollings (­2013) exploited the fact that people’s knowledge of the appearance of individuals from the early 20th century (­e.g., Albert Einstein, Winston Churchill) derives predominantly from exposure to ­black-­​­­and-​ ­white media images. The experiments manipulated whether participants were instructed to explicitly imagine using colour or not for examples of people who were famous in the ­pre-​­and ­post-​­colour media eras (­i.e., “­imagine Albert Einstein wearing a green jacket” vs. “­imagine Albert Einstein wearing a jacket”). The results showed that colour manipulation instructions only influenced mental imagery for ­black-­​­­and-​­white era individuals, with no comparable effect on imagery for colour era individuals. These findings are consistent with the hypothesis that mental imagery can be constrained by surface features of underlying representations in memory, even when imagining something we have never directly perceived.

Individual Differences and Imagery Training in Creative Cognition Considering the functional role mental imagery appears to play during creative cognition, it follows that individual differences in imagery ability might be associated with overall creative thinking and ­problem-​­solving performance. A ­meta-​­analytic review of the imagery and creativity literature conducted by LeBoutillier and Marks (­2003) found that ­self-​ ­reported imagery was consistently associated with participants’ performance on a range of creative divergent thinking tasks. A study by Morrison and Wallace (­2001) found significant correlations between ratings of the vividness of mental imagery, the visual art subscale of the Creative Behaviour Inventory (­Hocevar, 1976), and performance on mental synthesis and imagery transformation tasks. In a similar study, Palmiero et al. (­2015) investigated the relationship between ­self-​­ratings for vividness of imagery, experimental measures of mental image generation, inspection, and transformation processes, and participants’ performance on the creative invention task (­Finke, 1994). This task requires that participants mentally synthesise sets of ­three-​­dimensional objects into novel “­inventions” within a specified category (­e.g., furniture, transportation). Palmiero et al. found that measures of image transformation ability significantly predicted originality scores for the creative inventions produced by participants, while vividness of imagery ratings predicted practicality scores. In addition to individual differences in imagery, there is also evidence that training in mental imagery can enhance creative performance. For example, mental imagery has been linked to creative improvisation in professional dancers (­Fink et  al., 2009), and imagery is frequently employed as a pedagogical technique in the teaching of contemporary dance (­Overby & Dunn, 2011). Franklin (­2013) argues that mental imagery can help support cognitive changes in the mental representation of movement, which in turn can lead to physical changes in the musculature and the execution of real movements. Dance teachers can encourage the use of different modalities of imagery (­v isual, auditory, and kinaesthetic) and 190

Mental Imagery and Creative Cognition

whether images are experienced from a fi ­ rst-​­or t­hird-​­person perspective (­Overby et  al., 1998). Metaphorical imagery can also be employed to help enhance new skill learning (­Overby  & Dunn, 2011). A longitudinal study conducted by May et  al. (­2020) with undergraduate dance students evaluated the impact of mental imagery training on creative performance in choreography exercises. They found that after four months, participants who completed the imagery workshops scored higher on teacher evaluations of creativity in choreographic assessments than controls. Imagery training was also linked to greater improvement on measures of flexible thinking (­particularly ideational and spontaneous fluency and originality) in comparison to students who did not complete the workshops. Mental imagery training has also been linked to enhancing creativity in the composition of music. Auditory imagery can be used to imagine novel music and form predictions as to how musical compositions would sound to an audience (­Gordon, 1979; Schneider & Godoy, 2001). Composition is also supported through the mental visualisation of the notation of music scores (­Mountain, 2001). Hubbard (­2013) argues that the imagery utilised by composers is multisensory in nature, incorporating both auditory and visual information. For example, Rosenberg and Trusheim (­1989) describe an orchestral player who reported visualising sounds in association with colours. Imagery can be used by expert composers to help mentally translate visualised scenes or objects into auditory equivalents (­Bailes, 2009). Wong and Lim (­2017) examined the impact of mental imagery training on compositional ­ ve-​­to e­ ight-­​­­year-​­old children. A panel of expert judges rated comcreativity in a sample of fi positions produced by children using mental imagery instructions to be significantly higher on creativity than compositions produced in a ­non-​­imagery control group. Formal training in artistic skills may also positively influence mental imagery processes. ­ rst-​­and ­final-​­year fine arts ­Perez-​­Fabello and Campos (­2007) compared the performance of fi students on measures of vividness of imagery and visual elaboration, tests of spatial representation and transformation, and visual memory. They found that ­final-​­year students who had undertaken a longer period of artistic training performed significantly better than ­first-​­year students on all measures. A similar study conducted by Pearson et al. (­2001) compared performance of the Finke and Slayton mental synthesis task by fi ­ nal-​­year architecture students to a control group without formal training in drawing or visual thinking. Although there was no difference between groups in terms of the number of synthesised patterns produced, the architecture students created patterns with a significantly higher level of transformational complexity (­a measure of the number of mental transformations necessary to produce a pattern, Anderson & Helstrup, 1993). Drake et al. (­2019) compared mental imagery processes in a sample of visual arts and ­non-​­arts students and found that the art and design students were significantly better than the controls on measures of mental imagery vividness and abstraction. If mental imagery training confers beneficial effects, what exactly is it that is developed or enhanced? One possibility is that the training improves the processing efficiency or capacity of the cognitive systems responsible for the generation, inspection, and transformation of mental images. The idea that training programmes can function as a tool for improving cognitive abilities is central to popular concepts such as “­brain training” (­Simons et  al., 2016), but evidence suggests that the generalisation of cognitive training to other skills is highly limited (­­Melby-​­Lervåg & Hulme, 2013). Rather than increasing the capacity limits of cognitive systems (­Dahlin et al., 2009), training may instead function to encourage the development of cognitive strategies that make more efficient use of existing working memory resources (­Holmes et al., 2009). Rather than any fundamental change to cognitive functioning, another explanation is that mental imagery training changes an individual’s metacognitive understanding of their own 191

David G. Pearson

imagery processes. Metacognition can be defined as awareness of one’s own cognitive processes and how they may be controlled (­Flavell, 1979). Rademaker and Pearson (­2012) conducted a study where participants received training in imagining coloured patterns over a ­five-​­day period. While training did not improve the overall strength of the participants’ images, it did significantly improve a measure of their metacognition (­the extent to which their estimates of image vividness predicted performance on a perceptual bias task). May et al. (­2020) also identified changes in participants’ metacognitive understanding as a driving factor behind their observed improvements in creative dance choreography and flexible thinking following mental imagery training.

Conclusions Overall, there is a considerable body of evidence to support an important role for mental imagery during creative cognition. However, the specific functionality of imagery in the context of creative thinking remains an area of uncertainty. I have argued that a key benefit of mental imagery may be that it allows us to mentally simulate aspects of how we consciously perceive the external world. Although these simulations are not a completely veridical representation of reality, they allow people such as artists and designers to anticipate the potential sensory impact a work might have on an audience prior to expending the effort to create it (­for example, what a picture might look like or what a musical composition might sound like). Imagery also allows simulations to be created of events that would be impossible to experience in real life, such as the scientific visualisations described by Albert Einstein and Stephen Hawking. However, while our ability to create and manipulate mental imagery may at first appear boundless, research studies have identified significant areas of constraint. These include the capacity limitations of working memory, the perceptual reference frames in which images are interpreted, the nature of the representations in memory from which images are created, and restrictions in cognitive abilities to mentally represent complex forms and dynamically animate them. Many of these constraints can be overcome by supplementing internal mental imagery with external representations such as drawing or sketching. Despite the overlap between imagery and perceptual processes in the brain revealed by cognitive neuroscience, imagining what something looks like and actually seeing it in the real world are not the same thing. Finally, the research suggesting that training in mental imagery can benefit aspects of creative cognition is promising, but it remains uncertain specifically what is being improved. The training could operate at a metacognitive level, in terms of how well people can evaluate the phenomenal qualities of their mental images, or at a more specific level, in terms of the processing efficiency and capacity of the cognitive systems responsible for the generation, inspection, and transformation of mental images. Recent research suggests that the role of attentional and oculomotor processes during the experience of mental imagery may be particularly valuable avenues to explore in the future.

References Anderson, R. E., & Helstrup, T. (­1993). Visual discovery on mind and on paper. Memory and Cognition, 21, ­283–​­293. Andrade, J., May, J., Deeprose, C., Baugh, S. J., & Ganis, G. (­2014). Assessing vividness of mental imagery: The plymouth sensory imagery questionnaire. British Journal of Psychology, 105(­4), ­547–​­563.

192

Mental Imagery and Creative Cognition Baddeley, A. D. (­1990). The development of the concept of working memory: Implications and contributions of neuropsychology. In G. Vallar & T. Shallice (­Eds.), Neuropsychological Impairments of ­Short-​­Term Memory (­­pp. ­54–​­73). Cambridge: Cambridge University Press. Baddeley, A. (­2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4(­11), ­417–​­423. Baddeley, A. D.,  & Andrade, J. (­2000). Working memory and the vividness of imagery. Journal of Experimental Psychology: General, 129(­1), 126. Baddeley, A. D., & Hitch, G. (­1974). Working memory. Psychology of Learning and Motivation, 8(­34), ­47–​­89. Bailes, F. (­2009). Translating the musical image: Case studies of expert musicians. In A. Chan & A. Noble (­Eds.), Sounds in Translation: Intersections of Music, Technology and Society (­­pp. ­41–​­59). Canberra: ANU Press. Barquero, B., & Logie, R. H. (­1999). Imagery constraints on quantitative and qualitative aspects of mental synthesis. European Journal of Cognitive Psychology, 11(­3), ­315–​­333. Behrmann, M., Moscovitch, M., & Winocur, G. (­1994). Intact visual imagery and impaired visual perception in a patient with visual agnosia. Journal of Experimental Psychology: Human Perception and Performance, 20(­5), 1068. Bernstein, L. J.,  & Cooper, L. A. (­1997). Direction of motion influences perceptual identification of ambiguous figures. Journal of Experimental Psychology: Human Perception and Performance, 23(­3), ­721–​­737. Betts, G. H. (­1909). The Distribution and Functions of Mental Imagery. New York: Teachers College, Columbia University. Bilda, Z., Gero, J. S., & Purcell, T. (­2006). To sketch or not to sketch? That is the question? Design Studies, 27(­5), ­587– ​­613 Bilda, Z.,  & Gero, J. S. (­2007). The impact of working memory limitations on the design process during conceptualization. Design Studies, 28(­4), ­343–​­367. Blair, A. (­1990). Tycho Brahe’s critique of Copernicus and the Copernican system. Journal of the History of Ideas, 51(­3), ­355–​­377. ­Blouin-​­Hudon, E. M. C., & Pychyl, T. A. (­2015). Experiencing the temporally extended self: Initial support for the role of affective states, vivid mental imagery, and future ­self-​­continuity in the prediction of academic procrastination. Personality and Individual Differences, 86, ­50–​­56. Borst, G., Niven, E., & Logie, R. H. (­2012). Visual mental image generation does not overlap with visual ­short-​­term memory: A ­dual-​­t ask interference study. Memory & Cognition, 40(­3), ­360–​­372. Brandimonte, M. A., & Gerbino, W. (­1996). When imagery fails: Effects of verbal recoding on visual memory accessibility. In C. Cornoldi, R. H. Logie, M. A. Brandimonte, G. Kaufmann, & D. Reisberg (­Eds.), Stretching the Imagination: Representations and Transformations in Mental Imagery (­­pp. ­31–​­76). Oxford: Oxford University Press. Brandt, S. A., & Stark, L. W. (­1997). Spontaneous eye movements during visual imagery reflect the content of the visual scene. Journal of Cognitive Neuroscience, 9(­1), 2­ 7–​­38. Buckner, R. L., & Wheeler, M. E. (­2001). The cognitive neuroscience of remembering. Nature Reviews Neuroscience, 2, ­624–​­634. Carbon, C. C. (­2014). Understanding human perception by ­human-​­made illusions. Frontiers in Human Neuroscience, 8(­566), 1­ –​­6. Cattaneo, Z., Postma, A., & Vecchi, T. (­2006). Gender differences in memory for object and word locations. Quarterly Journal of Experimental Psychology, 59(­5), ­904–​­919. Chambers, D., & Reisberg, D. (­1985). Can mental images be ambiguous? Journal of Experimental Psychology: Human Perception and Performance, 11, ­317–​­328. Clark, A. (­2003). Natural Born Cyborgs: Minds, Technologies, and the Future of Human Intelligence. Oxford: Oxford University Press. Clark, A. (­2008). Supersizing the Mind: Embodiment, Action, and Cognitive Extension. New York: Oxford University Press Currie, G., & Ravenscroft, I. (­2002). Recreative Minds: Imagination in Philosophy and Psychology. Oxford: Oxford University Press. Dahlin, E., Bäckman, L., Neely, A. S., & Nyberg, L. (­2009). Training of the executive component of working memory: subcortical areas mediate transfer effects. Restorative Neurology and Neuroscience, 27(­5), ­405– ​­419. Danker, J. F., & Anderson, J. R. (­2010). The ghosts of brain states past: Remembering reactivates the brain regions engaged during encoding. Psychological Bulletin, 136, ­87–​­102.

193

David G. Pearson Dawes, A. J., Keogh, R., Andrillon, T., & Pearson, J. (­2020). A cognitive profile of ­multi-​­sensory imagery, memory and dreaming in aphantasia. Scientific Reports, 10(­1), ­1–​­10. De Beni, R., Pazzaglia, F., & Gardini, S. (­2007). The generation and maintenance of visual mental images: Evidence from image type and aging. Brain and Cognition, 63(­3), ­271–​­278. Denis, M., Beschin, N., Logie, R. H. & Della Sala, S. (­2002). Visual perception and verbal descriptions as sources for generating mental representations: Evidence from representational neglect. Cognitive Neuropsychology, 19(­2), ­97–​­112. Denis, M., & Cocude, M. (­1992). Structural properties of visual images constructed from poorly or ­well-​­structured verbal descriptions. Memory and Cognition, 20, ­497–​­506. Do, E.­Y-​­L., Gross, M. D., Neiman, B., & Zimring, C. (­2000). Intentions in and relations among design drawings. Design Studies, 21(­5), ­483–​­503 Doherty, M. J., & Mair, S. (­2012). Creativity, ambiguous figures, and academic preference. Perception, 41(­10), ­1262–​­1266. Drake, J. E., Simmons, S., Rouser, S., Poloes, I., & Winner, E. (­2019). Artists excel on image activation but not image manipulation tasks. Empirical Studies of the Arts, 39(­1), ­3 –​­16. ­ roblem-​­Solving (­Psychological Monographs, No. 270). Washington D.C.: AmerDuncker, K. (­1945). On P ican Psychological Association. Ekroll, V. (­2019). Illusions of imagery and magical experiences. ­I-​­Perception, 10(­4), ­1–​­34. Farah, M. J., Hammond, K. M., Levine, D. N., & Calvanio, R. (­1988). Visual and spatial mental imagery: Dissociable systems of representation. Cognitive Psychology, 20(­4), ­439–​­462. Fink, A., Graif, B., & Neubauer, A. C. (­2009). Brain correlates underlying creative thinking: EEG alpha activity in professional vs. novice dancers. NeuroImage, 46(­3), ­854–​­862. Finke, R. A. (­1986). Mental imagery and the visual system. Scientific American, 254, ­88–​­95. Finke, R. (­1989). Principles of Mental Imagery. Cambridge, MA: MIT Press. Finke, R. (­1994). Creative Imagery: Discoveries and Inventions in Visualization. Hillsdale, NJ: Lawrence Erlbaum. Finke, R., Pinker, S. & Farah, M. J. (­1989). Reinterpreting visual patterns in mental imagery. Cognitive Science, 13, ­51–​­78. Finke, R. A., & Slayton, K. (­1988). Explorations of creative visual synthesis in mental imagery. Memory & Cognition, 16(­3), ­252–​­257. Finke, R., Ward, T. B., & Smith, S. M. (­1992). Creative Cognition: Theory, Research, and Applications. Cambridge, MA: MIT Press. Fish, J., & Scrivener, S. (­1990). Amplifying the mind’s eye: Sketching and visual cognition. Leonardo, 23(­1), ­117–​­126. Flavell, J. H. (­1979). Metacognition and cognitive monitoring: A new area of ­cognitive–​­developmental inquiry. American Psychologist, 34(­10), 906. Franklin, E. N. (­2013). Dance Imagery for Technique and Performance. Champaign, IL: Human Kinetics. Gardini, S., De Beni, R., Cornoldi, C., Bromiley, A., & Venneri, A. (­2005). Different neuronal pathways support the generation of general and specific mental images. Neuroimage, 27(­3), ­544–​­552. Gardner, H. (­1993). Creating Minds. New York: Basic Books. Galton. Gardner, M. (­1977). Mathematical games. Scientific American, 236(­4), ­129–​­139 Gardner, M. (­2014). Mathematics, Magic and Mystery. New York, NY: Dover Publications. Goel, V. (­1995). Sketches of Thought. Cambridge, MA: MIT Press. Goldschmidt, G. (­1991). The dialectics of sketching. Creativity Research Journal, 4(­2), ­123–​­143. ­ ands-​­on scientist. Physics Education, 26, ­307–​­312. Gooding, D. (­1991). Faraday was a h Gordon, E. E. (­1979). Primary Measures of Music Audiation. Chicago, IL: G.I.A. Publications. Greenberg, D. L., & Knowlton, B. J. (­2014). The role of visual imagery in autobiographical memory. Memory & Cognition, 42(­6), ­922–​­934. Gurtner, L. M., Hartmann, M., & Mast, F. W. (­2021). Eye movements during visual imagery and perception show spatial correspondence but have unique temporal signatures. Cognition, 210, 104597. Haugland, M. D., Pearson, D. G., & Ekroll, V. (­2021). Möbius band surprise: A systematic illusion in imagery. ­I-​­Perception, 12(­2), 20416695211004972. Hawking, S. W. (­1993). Black Holes and Baby Universes and Other Essays. London: Bantam Press. Helstrup, T., Cornoldi, C., & De Beni, R. (­1997). Mental images: Specific or general, personal or impersonal? Scandinavian Journal of Psychology, 38(­3), ­189–​­197. Herbert, D. M. (­1988). Study drawings in architectural design: Their properties as a graphic medium. Journal of Architectural Education, 41(­2), ­26–​­38.

194

Mental Imagery and Creative Cognition Hitch, G. J., Allen, R. J., & Baddeley, A. D. (­2020). Attention and binding in visual working memory: Two forms of attention and two kinds of buffer storage. Attention, Perception, & Psychophysics, 82(­1), ­280–​­293. Hitch, G. J., Brandimonte, M. A., & Walker, P. (­1995). Two types of representation in visual memory: Evidence from the effects of stimulus contrast on image combination. Memory & Cognition, 23(­2), ­147–​­154. Hocevar, D. (­1976). Dimensionality of creativity. Psychological Reports, 39(­3), ­869–​­870. Holmes, J., Gathercole, S. E., & Dunning, D. L. (­2009). Adaptive training leads to sustained enhancement of poor working memory in children. Developmental Science, 12(­4), ­F9–​­F15. Hubbard, T. L. (­2013). Auditory imagery contains more than audition. In S. Lacey,  & R. Lawson (­Eds.), Multisensory Imagery (­­pp. ­221–​­247). New York: Springer. Irving, L., Barry, R., LeBoutillier, N., & Westley, D. (­2011). The image control and recognition task: A ­performance-​­based measure of imagery control. Journal of Mental Imagery, 35(­3), 67. Jastrow, J. (­1899). The mind’s eye. Popular Science Monthly, 54, ­299–​­312. ­ ulti-​­modal object learning. Behavioral and Brain Jüttner, M.,  & Rentschler, I. (­2002). Imagery in m Sciences, 25(­2), 1­ 97–​­198. Kaufmann, G. (­1988). Mental imagery and problem solving. In M. Denis, J. Engelkamp & J. T. E. Richardson (­Eds.), Cognitive and Neuropsychological Approaches to Mental Imagery ( ­­pp. ­231–​­240). Dordrecht: Martinus Nijhoff Publishers. Kavakli, M.,  & Gero, J. S. (­2001). Sketching as mental imagery processing. Design Studies, 22(­4), ­347–​­364. Kokotovich, V., & Purcell, T. (­2000). Mental synthesis and creativity in design: An experimental examination. Design Studies, 21(­5), ­437– ​­4 49. Kosslyn, S. M. (­1994). Image and Brain: The Resolution of the Imagery Debate. Cambridge, MA: MIT Press. Kosslyn, S. M., Thompson, W. L., & Ganis, G. (­2006). The Case for Mental Imagery. New York: Oxford University Press, Inc. Lawson, B. (­1980). How Designers Think. London: Routledge. LeBoutillier, N., & Marks, D. F. (­2003). Mental imagery and creativity: A meta-analytic review study. British Journal of Psychology, 94(­1), ­29–​­4 4. Leddington, J. (­2016). The experience of magic. The Journal of Aesthetics and Art Criticism, 74(­3), ­253–​­264. Logie, R. H. (­1995). ­Visuo-​­Spatial Working Memory. Hove, UK: Lawrence Erlbaum Associates. Logie, R. H. (­2011). The functional organisation and the capacity limits of working memory. Current Directions in Psychological Science, 20(­4), ­240–​­245 Logie, R., Camos, V., & Cowan, N. (­Eds.) (­2020). Working Memory: The State of the Science. Oxford: Oxford University Press. Logie, R. H., & van der Meulen, M. (­2009). Fragmenting and integrating ­v isuo-​­spatial working memory. In J. R. Brockmole (­Ed.), The Visual World in Memory (­­pp. ­1–​­32). Hove, UK: Psychology Press. Margolis, H. (­1998a). Tycho’s illusion and human cognition. Nature, 392, 857. Margolis, H. (­1998b). Tycho’s illusion: How it lasted 400 years, and what that implies about human cognition. Psycoloquy, 9, 32. Marks, D. F. (­1973). Visual imagery differences in the recall of pictures. British Journal of Psychology, 64, 4­ 07–​­412. Mast, F. W., & Kosslyn, S. M. (­2002). Visual mental images can be ambiguous: Insights from individual differences in spatial transformation abilities. Cognition, 86(­1), ­57–​­70. May, J., Redding, E., Whatley, S., Łucznik, K., Clements, L., Weber, R.,  …  & Reed, S. (­2020). Enhancing creativity by training metacognitive skills in mental imagery. Thinking Skills and Creativity, 38, 100739. ­ eta-​­analytic re­Melby-​­Lervåg, M., & Hulme, C. (­2013). Is working memory training effective? A m view. Developmental Psychology, 49(­2), 270. Miller, A. I. (­1984). Imagery in Scientific Thought: Creating 20th Century Physics. Boston, MA: Birkhauser. Miller, A. I. (­2000). Insights of Genius: Imagery and Creativity in Science and Art. Cambridge, MA: MIT Press. Morrison, R. G., & Wallace, B. (­2001). Imagery vividness, creativity and the visual arts. Journal of Mental Imagery, 25(­­3 –​­4), 1­ 35–​­152. Mountain, R. (­2001). Composers and imagery: Myths and realities. In R. I. Godoy, & H. Jorgensen (­Eds.), Musical Imagery (­­pp. ­271–​­288). Abingdon: Routledge.

195

David G. Pearson Niemetschek, F. X. (­1798). Lebensbeschreibung des k. k. Kapellmeisters Wolfgang Amadeus Mozart. Hamburg: Verlag. Nin, A. (­1969). The Diary of Anais Nin ­1931–​­1934. New York: Harvest Books. Pearson, Overby, L. Y., & Dunn, J. (­2011). The history and research of dance imagery: Implications for teachers. The IADMS Bulletin for Teachers, 3(­2), ­9 –​­11. Overby, L. Y., Hall, C., & Haslam, I. (­1998). A comparison of imagery used by dance teachers, figure skating coaches, and soccer coaches. Imagination, Cognition and Personality, 17(­4), ­323–​­337. Palmiero, M., Nori, R., Aloisi, V., Ferrara, M., & Piccardi, L. (­2015). ­Domain-​­specificity of creativity: A study on the relationship between visual creativity and visual mental imagery. Frontiers in Psychology, 6, 1870. Pearson, D. G. (­1998). Imagery need not be blind to fail: Commentary on Margolis on cognitive illusion. Psycoloquy, 9(­34), ­1–​­2 . Pearson, D.G. (­2001). Imagery and the v­ isuo-​­spatial sketchpad. In J. Andrade (­Ed.), Working Memory in Perspective (­­pp. ­53–​­79) Hove: The Psychology Press. Pearson, D. G. (­2007). Mental imagery and creative thought. Proceedings of the British Academy, 147, ­187–​­212. Pearson, D. G., Alexander, A., & Webster, R. (­2001). Working memory and expertise differences in design. In J. Gero, B. Tversky & T. Purcell (­Eds.), Visual and Spatial Reasoning in Design II. Sydney: Key Centre of Design Computing and Cognition. Pearson, D. G., De Beni, R., & Cornoldi, C. (­2001). The generation and transformation of ­v isuo-​ s­ patial mental images. In M. Denis, R. H. Logie, C. Cornoldi, M. de Vega & J. Engelkamp (­Eds.), Imagery, Language and ­Visuo-​­Spatial Thinking (­­pp. ­1–​­23). Hove: Psychology Press Pearson, D. G., Deeprose, C., ­Wallace-​­Hadrill, S., Burnett Heyes, S., & Holmes, E. A. (­2013). Assessing mental imagery in clinical psychology: A review of imagery measures and a guiding framework. Clinical Psychology Review, 33(­1), ­1–​­23. Pearson, D. G., & Hollings, J. (­2013). Einstein’s jacket: Evidence for ­long-​­term perceptual specificity in mental imagery. Consciousness and Cognition, 22(­1), ­148–​­154. Pearson, D. G., Logie, R. H., & Green, C. (­1996). Mental manipulation, visual working memory, and executive processes. Psychologische Beitrage, 38(­3\4), ­324–​­342. Pearson, D. G., Logie, R. H. & Gilhooly, K. (­1999). Verbal representations and spa-tial manipulation during mental synthesis. European Journal of Cognitive Psychology, 11(­3), ­295–​­314. Pearson, D. G.,  & Logie, R. H. (­2000). Working memory and mental synthesis. In S. O’Nuallan (­Ed.), Spatial Cognition: Foundations and Applications (­­pp.  ­347–​­359). Amsterdam: John Benjamins Publishing Company. Pearson, D. G. & Logie, R. H. (­2015). A sketch is not enough: Dynamic external support increases insight on a guided synthesis task. Thinking and Reasoning, 21(­1), ­97–​­112. Pearson, J., & Kosslyn, S. M. (­2015). The heterogeneity of mental representation: Ending the imagery debate. Proceedings of the National Academy of Sciences, 112(­33), ­10089–​­10092. ­Pérez-​­Fabello, M. J., & Campos, A. (­2007). Influence of training in artistic skills on mental imaging capacity. Creativity Research Journal, 19(­­2 –​­3), ­227–​­232. Purcell, A. T., & Gero, J. S. (­1998). Drawings and the design process. Design Studies, 19(­4), ­389–​­430. Rademaker, R. L., & Pearson, J. (­2012). Training visual imagery: Improvements of metacognition, but not imagery strength. Frontiers in Psychology, 3, 224. Reed, S. K. (­1993). Imagery and discovery. In B. ­Roskos-​­Ewoldsen, M. J. ­Intons-​­Peterson  & R. Anderson (­Eds.), Imagery, Creativity and Discovery: A Cognitive Perspective (­­pp. ­287–​­312) Amsterdam: ­North-​­Holland. ­ ong-​­term perceptual specificity effects in recognition memRay, R. A., & Reingold, E. M. (­2003). L ory: The transformed pictures paradigm. Canadian Journal of Experimental Psychology, 57(­2), ­131–​­137. Reingold, R. A. (­2002). On the perceptual specificity of memory representations. Memory, 10(­5/­6), ­365–​­379. Reisberg, D. (­1996). The nonambiguity of mental images. In C. Cornoldi, R. H. Logie, M. A. Brandimonte, G. Kaufmann & D. Reisberg (­Eds.), Stretching the Imagination: Representation and Transformation in Mental Imagery (­­pp. ­119–​­172) New York: Oxford University Press. Reisberg, D., & Logie, R. H. (­1993). The ins and outs of visual working memory: Overcoming the limits on learning from imagery. In M. ­Intons-​­Peterson, B. Roskos­Ewoldsen  & R. Anderson (­Eds.), Imagery, Creativity, and Discovery: A Cognitive Approach (­­pp. ­39–​­76) Amsterdam: Elsevier.

196

Mental Imagery and Creative Cognition Reisberg, D., Pearson, D. G., & Kosslyn, S. M. (­2003). Intuitions and introspections about imagery: The role of imagery experience in shaping an investigator’s theoretical views. Applied Cognitive Psychology, 17, ­147–​­160. Riquelme, H. (­2002). Can people creative in imagery interpret ambiguous figures faster than people less creative in imagery? The Journal of Creative Behavior, 36(­2), ­105–​­116. Rosenberg, H. S., & Trusheim, W. (­1989). Creative transformations: How visual artists, musicians, and dancers use mental imagery in their work. In J. E. Shorr, P. Robin, J. A. Connella,  & M. Wolpin (­Eds.), Imagery (­­pp. ­55–​­75). Boston, MA: Springer. ­Roskos-​­Ewoldsen, B. (­1993). Discovering emergent properties of images. Advances in Psychology, 98, ­187–​­221. ­Rossi-​­Arnaud, C., Cortese, A., & Castari, V. (­2004). Memory span for movement con­figurations: The effects of concurrent verbal, motor and visual interference. Current Psychology of Cognition, 22(­3), ­335–​­349. Roth, I. (­2007). Imaginative Minds. Oxford: Oxford University Press. Sagone, E., De Caroli, M. E., Falanga, R., & Indiana, M. L. (­2020). Mental synthesis and creative thinking in typically developed Italian. New Trends and Issues Proceedings on Humanities and Social Sciences, 7(­1), ­220–​­228. Samuels, M., & Samuels, N. (­1975). Seeing with the Mind’s Eye. New York: Random House. Scaife, M., & Rogers, Y. (­1996). External cognition: How do graphical representations work? International Journal of ­Human-​­Computer Studies, 45(­2), ­185–​­213. Schacter, D. L., Norman, K. A., & Koutstaal, W. (­1998). The cognitive neuroscience of constructive memory. Annual Review of Psychology, 49, 2­ 89–​­318. Schneider, A.,  & Godøy, R. I. (­2001). Perspectives and challenges of musical imagery. In R. I. Godoy, & H. Jorgensen (­Eds.), Musical Imagery (­­pp. ­5 –​­26). Abingdon: Routledge. Shepard, R. N. (­1978). The mental image. American Psychologist, 33(­2), 125. Simons, D. J., Boot, W. R., Charness, N., Gathercole, S. E., Chabris, C. F., Hambrick, D. Z.,  & S­ tine-​­Morrow, E. A. (­2016). Do “­­brain-​­t raining” programs work?. Psychological Science in the Public Interest, 17(­3), ­103–​­186. Slotnick, S. D. (­2004). Visual memory and visual perception recruit common neural substrates. Behavioral and Cognitive Neuroscience Reviews, 3(­4), ­207–​­221. Stevenson, R. J., & Case, T. I. (­2005). Olfactory imagery: A review. Psychonomic Bulletin & Review, 12(­2), ­244–​­264. Thomas, L. E. (­2013). Spatial working memory is necessary for actions to guide thought. Journal of Experimental Psychology: Learning, Memory, and Cognition, 39(­6), ­1974–​­1981. Thompson, A. L., & Klatzky, R. L. (­1978). Studies of visual synthesis: Integration of fragments into forms. Journal of Experimental Psychology: Human Perception and Performance, 4(­2), 244. Tiggemann, M., & Kemps, E. (­2005). The phenomenology of food cravings: The role of mental imagery. Appetite, 45(­3), ­305–​­313. ­ llmann-​­Margalit (­Ed.), The Tversky, A. (­1986). Cognitive illusions in judgment and choice. In E. U Kaleidoscope of Science (­­pp. ­75–​­87). Dordrecht: Reidel Publishing Co. Vallée-Tourangeau, F., & March, P. L. (­2019). Insight out: Making creativity visible. The Journal of Creative Behavior, 54(­4), ­824–​­842. ­Vallée-​­Tourangeau, F., Ross, W., Ruffatto Rech, R., & ­Vallée-​­Tourangeau, G. (­2021). Insight as discovery. Journal of Cognitive Psychology, 33(­­6 –​­7 ), ­718–​­737. Verstijnen, I. M., van Leeuwen, C., Goldschimdt, G., Haeml, R., & Hennessey, J. M. (­1998). Creative discovery in imagery and perception: Combining is relatively easy, restructuring takes a sketch. Acta Psychologica, 99, 1­ 77–​­200. Walker, P., Hitch, G. J., Dewhurst, S. A., Whiteley, H. E., & Brandimonte, M. A. (­1997). The representation of ­non-​­structural information in visual memory: Evidence from image combination. Memory & Cognition, 25(­4), ­484–​­491. Weisberg, R. W. (­1986). Creativity: Genius and Other Myths. New York: W.H. Freeman and Company. Wichmann, F. A., Sharpe, L. T.,  & Gegenfurtner, K. R. (­2002). The contributions of colour to recognition memory for natural scenes. Journal of Experimental Psychology: Learning, Memory, and Cognition, 28(­3), 5­ 09–​­520 Wijntjes, M. W.A., van Lienen, T., Verstijnen, I. M., & Kappers, A. M. L. (­2008). Look what I have felt: Unidentified haptic line drawings are identified after sketching. Acta Psychologica, 128(­2), ­255–​­263

197

David G. Pearson Wilson, M. (­1988). Mark Wilson’s Complete Course in Magic. Philadelphia: Running Press. Wiseman, R., Watt, C., Gilhooly, K., & Georgiou, G. (­2011). Creativity and ease of ambiguous figural reversal. British Journal of Psychology, 102(­3), ­615–​­622. Wong, S. S. H., & Lim, S. W. H. (­2017). Mental imagery boosts music compositional creativity. PLoS One, 12(­3), e0174009. Wotiz, J. H., & Rudofsky, S. (­1954). Kekule’s dream: Fact or fiction? Chemistry in Britain, 20, ­720–​­723. Zeman, A., Milton, F., Della Sala, S., Dewar, M., Frayling, T., Gaddum, J.,… & Winlove, C. (­2020). ­Phantasia–​­the psychological significance of lifelong visual imagery vividness extremes. Cortex, 130, ­426–​­4 40. Zhang, J. (­1997). The nature of external representations in problem solving. Cognitive Science, 21(­2), ­179–​­217.

198

13 INCUBATION Ken Gilhooly

Introduction The basic idea of incubation in ­problem-​­solving is that a previously apparently intractable problem can be solved simply by putting it aside for a while, doing other things and awaiting inspiration. This method for solving difficult problems seems almost magical and even too good to be true! However, as we will see, research has supported the reality of positive incubation effects in ­problem-​­solving, although exactly how incubation works is still not settled. In everyday life, a person stuck on a problem is often advised to “­Leave it!” for a while, or even to “­Sleep on it!”. These are essentially recommendations to try waking or sleeping incubation, and I will discuss both forms of incubation here. First, it will be useful to look briefly at what we mean by “­problems”.

Problems and ­Problem-​­solving Problems come in many forms, but all involve finding a way to reach a goal from a starting state, where there is no already known method for directly solving them. Some problems can be tackled by fairly straightforwardly searching through a space of possible actions (­heuristic search, e.g., Thomas, 1974) or by splitting the problem up into ­sub-​­problems (­­means-​­ends analysis, e.g., Anzai & Simon, 1979) and tackling those ­sub-​­problems one by one on the basis of an initial representation of the problem. Problems amenable to straightforward search solutions may be called “­routine” problems. ­Non-​­routine problems require a change of representation or ­re-​­structuring before solution, and such problems are generally known as insight problems. An example: How could you walk across a wide deep lake without getting wet without any special equipment? Representing the water in the lake as in its liquid form needs to be changed to represent the lake as frozen solid. In tackling an insight problem, there is often a period of impasse in which all obvious actions within the original representation have been unsuccessfully tried. Creative problems require solutions that are novel to the solver and overlap considerably with insight problems in that ­re-​­structuring is often needed for novel solutions to be reached. Generally, creative problems have a wide range of possible solutions that must be divergently produced (­whether by search or through insight or by some mixture of search and insight) DOI: 10.4324/9781003009351-15

199

Ken Gilhooly

and then selected among. A creative product is novel to the solver and is intended to meet some goal (­Weisberg, 2018). The notion of incubation has been particularly invoked in the context of insight and creative problems, where impasses (­leading to the problem being set aside) are more likely, rather than in more routine problems, amenable to straightforward searching or problem reduction.

Incubation Turning now in more detail to the notion of ­incubation – ​­this idea appeared early in approaches to understanding creative and insight ­problem-​­solving and was initially based on personal reports by eminent scientists, artists, inventors and writers. Some striking regularities emerged from these accounts, including frequent references to setting problems aside for lengthy periods (­i.e., taking incubation opportunities) with subsequent sudden insightful solutions.

Personal Accounts William James (­1880) gave an early, if undetailed and generalised, description of an incubation experience as follows: “­W hen walking along the street, thinking of the blue sky or the fine spring weather...I may suddenly catch an intuition of the solution of a long unsolved problem, which at that moment was very far from my thoughts”. Henri Poincaré (­1910), a leading French mathematician in the 19th and early 20th century, provided influential accounts of a series of interlinked creative episodes. One key account is given below (­Poincaré, 1910). For 15 days I strove to prove that there could not be any functions like those I have since called the Fuchsian Functions...Every day I seated myself at my work table ~ stayed an hour or two, tried a great number of combinations and reached no results. One evening, contrary to my custom, I drank black coffee and could not sleep. Ideas rose in crowds; I felt them collide until pairs interlocked, so to speak, making stable combination. By the next morning I had established the existence of a class of Fuchsian Functions, those which come from the hypergeometric series; I had only to write out the results, which took but a few hours. This episode begins with extended conscious attempts followed by a p­ re-​­sleep, “­hypnagogic” state, between full consciousness and sleep, which induced very free searching through possible mathematical combinations, one of which eventually led to a solution. Poincaré went on to surmise that in this episode of thinking, while in a hypnagogic state, he actually became consciously aware of what is normally unconscious when a problem is set aside in incubation. Thus, this episode was taken by Poincaré as making reportable the normally hidden nature of unconscious processing as a q­ uasi-​­random, constant combining and r­ e-​­combining of ideas, until a useful new combination is reached.

Wallas’s Model In developing his often cited stage model, Wallas (­1926) built on a number of sources, but particularly on Poincaré (­1910). His model of creative thinking proposed the following five stages: Preparation, Incubation, Intimation, Illumination and Verification. It seems that 200

Incubation

Wallas was the first to use the term “­Incubation” in the context of setting a problem aside. I assume he was guided by an analogy with birds incubating eggs to allow the development of chicks from embryos or with diseases incubating undetected in patients before becoming manifest. These biological uses of the term “­incubation” suggest automatically unfolding processes that in themselves lead to complex, complete outcomes (­such as a live chick or a ­full-​­blown disease state). Thus, the very term “­incubation” itself may suggest a particular kind of explanation of any effects of setting a problem aside, in terms of an underlying automatic solution process that, if left undisturbed, will deliver a solution. Here, I will mean by “­incubation”, simply a period in which a problem is not consciously worked on, without assuming any particular processes taking place unconsciously during that period. In Wallas’s (­1926) presentation of his model, the two conscious stages of Preparation and Verification are dealt with quite briefly. Both are essential, and both involve the application of learned rules and knowledge to systematically explore possibilities and evaluate proposed results. These stages involve conscious, deliberate work that uses “­System 2” processes that are slow, reflective and analytic and require load working memory (­Evans, 2008; Kahneman, 2011). Straightforward, routine, w ­ ell-​­defined problems that could be tackled by basic search processes would involve just these two stages. In the Preparation stage, the solver would develop a representation of the problem goal and a starting state and explore possible solution actions based on relevant prior knowledge; in the Verification stage, promising solution paths would be checked and either accepted or rejected, leading to further rounds of Preparation and Verification. However, if a problem is not straightforward, the Incubation stage will likely come into play. Regarding the Incubation stage, Wallas points out that this stage has a negative ­aspect – ​­in that the solver does not consciously think about the p­ roblem – ​­and a positive aspect in that relevant mental events take place below the level of consciousness and these events may help solve the problem. If absolutely no changes occurred in mental contents during the Incubation period, it would be quite ­ineffective – ​­so it is safe to assume that something does happen during this ­stage – at ​­ least when the stage is followed by Inspiration. In terms of avoiding conscious thought about the target p­ roblem – ​­this can happen in two broad w by distraction, ­ ays – firstly, ​­ that is by conscious work on other problems, and secondly, by relaxation from all conscious mental work. Wallas suggests the first approach (­d istraction) has the benefit that a number of problems could be started in rapid succession and incubated simultaneously; in this way, the solver could benefit from a form of unconscious ­multi-​­tasking rather than by attempting to progress just one problem at a time. However, with difficult creative problems, Wallas (­­p. 87) recommends the complete rest or relaxation type of incubation, so as not to interfere with “­the free working of the unconscious”. There is an underlying idea that there is a limited total amount of mental resource (­energy or activation) that can be used, and that if conscious processes consume most of the available resources then there will be insufficient resources for the unconscious processes to use and function well. On the other extreme, too much routine drudge work could have very negative effects by reducing the chances of fruitful unconscious work. The benefits of mental relaxation, Wallas assumed, would be enhanced further by physical exercise. Note that mental relaxation does not mean the absence of any conscious mental activity, which is surely impossible in normal waking states, but rather the presence of free floating, low mental loading, unfocused, mind wandering or daydreaming. Wallas was particularly critical of the habit of continual passive reading, which apparently was very prevalent among educated people in his days. During what might be fruitful incubation periods, the habit of continually reading, rather than allowing mind wandering, 201

Ken Gilhooly

he felt, could well interfere with effective incubation. The modern equivalent would be constantly checking emails and social media on mobile devices, a very common habit today that Wallas would no doubt have deplored as likely to impair the effectiveness of incubation! Regarding the Illumination stage, Wallas acknowledged that it was impossible to directly control the occurrence of this stage, which, if all went well, would end the Incubation stage with an inspiration. Indirectly, some control could be possible in that adequate Preparation and relaxing Incubation activities should help bring about Illumination, but this could not be guaranteed to do so. He suggested an additional possibility for exercising some control over Illumination, which was to be aware of Intimation – ​­the feeling that a solution was imminent. If this feeling were recognised, then distracting activities could be set aside to maximise the chances of the intimated solution getting enough activation to cross the threshold into consciousness. He proposed that the Intimation experience was a subjective correlate of an increasingly active “­a ssociation chain”, which was about to gain ­above-​­threshold consciousness. Poincaré in his account did not identify a stage such as Intimation intervening between Incubation and Illumination, and Wallas suggested that Intimation is very fleeting and often forgotten in the excitement of Illumination. However, other early writers in the field such as Varendonck (­1921), Wundt (­1893) and Dewey (­1910) had noted I­ ntimation-​­like phenomena before Wallas, as had William James in 1890 when he wrote: “­We all of us have this permanent consciousness of whither our thought is going. It is a feeling like any other, a feeling of what thoughts are next to arise, before they have arisen” (­Principles, Vol. 1, p­ p. ­255–​­56).

Incubation: Questions Over Validity of Personal Accounts What evidence is there for the reality of Incubation effects? Personal accounts by acknowledged creative thinkers in the arts and sciences have often been taken as evidence for the existence of this phenomenon (­Csikszentmihalyi, 1996; Ghiselin, 1952). However, the long gaps in time between the supposed incubation events and their reporting must raise questions about the reliability and validity of memory for such fleeting mental events. A ­small-​ ­scale laboratory study by Russo et al. (­1989) found that retrospective ­think-​­aloud protocols given a matter of minutes after p­ roblem-​­solving episodes showed extensive forgetting and fabrication when compared with concurrent ­think-​­aloud protocols. Russo et al. concluded that retrospective protocols often fail to give a veridical reflection of the concurrent process. They say “…these findings are fully in accord with the dim view of retrospective protocols expressed in the literature”. Similar doubts about retrospective reports have been expressed by Ericsson and Simon (­1984) and Nisbett and Wilson (­1977). Other personal accounts, frequently cited elsewhere and also given long after the purported events, by Coleridge, Mozart and Kekulé, that seemed to involve incubation and inspiration, have subsequently been shown to be simply fake in the case of Mozart and highly dubious in the cases of Coleridge and Kekulé (­Weisberg, 2006, ­pp. ­73–​­78; Wotiz & Rudofsky, 1984; see also Miller, 2000, p­ p. ­340–​­341). In view of the question marks over personal reports of incubation, we now turn to look at laboratory studies aimed at establishing the reality or otherwise of incubation effects.

Laboratory Studies of Incubation Since the pioneering theoretical analysis by Wallas (­1926) discussed above, many experimental laboratory studies on incubation effects have been carried out. These studies included 202

Incubation

both (­a) insight problems, in which new ways of representing or structuring the task must be found in order to reach a solution, and (­b) creative or divergent problems, in which there are no single correct solutions but many novel and useful ideas need to be generated to reach good solutions. The most commonly used divergent task in laboratory studies is the Alternative Uses Task (­sometimes abbreviated to Uses Task or AUT), in which participants are asked to come up with as many uses different from the normal use as they can, for one or more familiar objects, such as a brick, a pencil, a tyre or a shoe, in a limited time period (­Gilhooly et al., 2007; Guilford, 1971; Guilford et al., 1978). Until recently, laboratory studies of incubation followed the Delayed Incubation paradigm, in which participants in the incubation condition work on the target problem for an ­experimenter-​­determined time ( ­preparation time) before being given an interpolated activity, working on an unrelated, distractor task, for a fixed time (­incubation period), and then resume the target problem for a fixed ­post-​­incubation work period. The performance of the Incubation group can then be compared to that of the Control group, which works without interruption on the target task for a time equal to the sum of the preparation time and the p­ ost-​­incubation working time of the incubation group. Thus, the total conscious working time is equated between Incubation and Control conditions. There are variations in which the interpolated tasks may themselves be tested for incubation effects (­a n interleaved design), but most studies have used a straightforward design without interleaving. A possible problem with the standard design, in terms of ecological validity, is that the lengths of time given to the various stages, Preparation, Incubation and P ­ ost-​­Incubation, are set by the experimenter rather than by the solver. This is convenient from the point of view of research design and analysis so that conscious work times are readily equated between Incubation and Control groups, but it may mean that some participants are forced to break when they are still making progress and would rather carry on, and some are prevented from taking the break when they are ready (­e.g., experiencing an impasse). Another departure from ­real-​­life problem situations is that the times involved for the stages are limited to what is feasible in the laboratory. Typically, the stages in laboratory studies are for periods of minutes or occasionally hours. Extension of stages over days, weeks, months or even years, as reported in r­ eal-​­world incubation accounts, is not feasible in laboratory studies. Also, if the periods for working and for incubation add up to more than around 16 hours, the special case of sleep incubation arises – ​­and we will discuss this in a later section. As well as the typical Delayed Incubation paradigm, a relatively new paradigm (­the Immediate Incubation paradigm) gives participants an interpolated task for a fixed period immediately after instructions on the target problem but before any conscious work time has been allowed on the target problem. The immediate incubation period is then followed by continuous work on the target problem for a fixed p­ ost-​­incubation time (­Dijksterhuis & Meurs, 2006). In the next section, we will review key results from early laboratory studies using the delayed incubation paradigm and then examine the more recent immediate incubation results.

Incubation Results: Narrative and M ­ eta-​­Analytic Reviews Following on from early naturalistic studies by Patrick (­1935, 1937, 1938) of poets, painters and science students, there is now a substantial body of evidence from controlled laboratory studies establishing the basic phenomenon of incubation. To make sense of the volume of results, a substantial review was carried out by Dodds et al. (­2012) in the form of a narrative review of laboratory studies carried out between 1938 and 1991. Dodds et al. identified 39 203

Ken Gilhooly

experiments on incubation, of which 26 (­75%) reported significant benefits, ten reported failures to find effects and three found incubation only for certain groups of participants. The interpolated activity method was the most common, being used in 23 of 39 experiments with 16 successful demonstrations of incubation (­67%). The remaining 16 experiments used the interleaved method, with 13 successful studies (­c. 80%). Dodds et al. concluded that a number of variables appeared to affect incubation effectiveness. Thus, longer periods of preparation were generally better than shorter periods. Incubation periods of about 30 minutes seemed best overall, but effects were still found at longer periods of 3.5 and 24 hours. Incidentally presented clues appeared to be helpful during incubation as against during other periods. No clear effects emerged from the studies reviewed of problem type or of the nature of the activities during the incubation period, but few studies had systematically manipulated problem type or incubation activities, and so no firm conclusions were possible for these variables. There was marked variation in the experimental parameters among the studies reviewed by Dodds et al. (­2012), which makes it difficult to draw strong ­cross-​­experimental conclusions from their narrative review. A quantitative ­meta-​­analysis drawing on a larger body of studies that overcame these problems to a large extent was reported by Sio and Ormerod (­2009). Sio and Ormerod collected publications through electronic searches of the major databases in psychology and educational research, including sources of unpublished data such as dissertations and conference papers. This procedure yielded 37 relevant publications with a total of 117 studies for analysis. Approximately 73% of the studies showed an effect in the direction predicted by the hypothesis that incubation effects are real. An examination of possible publication bias, that is, a bias towards positive results as against null or negative results in the published literature, using the funnel plot method, indicated no bias in the study sample. Assessing the effect sizes for each study and combining the results statistically, they found a significant positive effect of Delayed Incubation, where the overall average effect size was in the l­ow-​­medium band (­mean Cohen’s d = .29, 95% CI [.21, .39]) over a range of linguistic and v­ isuo-​­spatial insight and divergent tasks. This size of effect equates to nearly 1/­3 of a standard deviation, or about 2% of variance, similar to a Pearson r correlation coefficient of 0.15. The effects of incubation were significant for the three types of tasks (­verbal, spatial insight and divergent) considered separately but were greatest for the divergent tasks and somewhat less for the linguistic and spatial insight tasks, which were not different from each other. Other possible moderators in addition to task type were examined, and it was found that longer preparation periods yielded larger incubation effects (­a s had also been concluded by Dodds et al., 2012). Highly cognitively demanding tasks in the incubation period produced lower incubation benefits. Linguistic insight problems benefitted more from low cognitively demanding incubation activities than from rest during incubation. Overall, from both the narrative and m ­ eta-​­analytic reviews, the basic existence of Delayed Incubation effects, which had been doubted by some researchers (­e.g., Olton  & Johnson, 1976), can now be regarded as well established, particularly in the case of creative (­d ivergent) problems but also, if to a lesser degree, in visual and linguistic insight tasks. The question now becomes not whether incubation effects exist but how such effects come about. In the next sections, we will outline three broad theoretical approaches seeking to explain incubation effects. These approaches are labelled “­Intermittent Conscious Work”, “­Beneficial Forgetting/­Fresh Look” and “­Unconscious Work”; these will be explained in detail in the next section. 204

Incubation

Incubation: Explanations The main explanations for incubation effects can be summarised as follows: 1 Intermittent conscious work: This suggests that although incubation is intended to be a period without conscious work on the target task, participants may nevertheless carry out intermittent conscious work (­Seifert et al., 1995; Weisberg, 2006, ­pp. ­443–​­445). This ­ on-​­mysterious explanation of incubation effects as being due would give a simple and n to additional conscious work being carried out by incubation groups. However, checks for intermittent work on the target task during incubation periods have not supported this explanation (­Gilhooly et al., 2012). 2 Beneficial Forgetting/­Fresh Look: This view (­e.g., Segal, 2004; Simon, 1966) proposes that there is an important role for automatic reductions in the strength of misleading ideas during incubation. The idea is that misleading strategies, mistaken assumptions and mental sets weaken through forgetting during the incubation period, and thus a fresh start is facilitated when the problem is resumed after the incubation break. 3 Unconscious processing: This proposal is that incubation effects result from active but unconscious processing of the problem during the incubation period (­Campbell, 1960; Poincaré, 1910; Simonton, 1995). Supportive results were found by Dijksterhuis and Meurs (­2006) and Gilhooly et al. (­2012) in studies that found benefits for creative production in a divergent task of an immediate incubation period relative to controls. In the immediate incubation paradigm, the task is presented and then immediately set aside, and so possible explanations in terms of beneficial forgetting followed by a fresh start can be ruled out as there is no period of initial work in which misleading sets could be developed. Dijksterhuis and Meurs applied their unconscious thought theory to this paradigm, where unconscious thought has a large capacity, tends to be ­bottom-​­up, is poor at rule following and tends to divergent rather than convergent thought. Dijksterhuis and Meurs’s unconscious thought or processing can be seen as a form of “­System 1” thinking, which is characterised as automatic, fast, effortless and as not loading working memory (­Evans, 2008; Kahneman, 2011).

Unconscious Processing Models The ­O pportunistic-​­A ssimilation model proposes a particular form of unconscious processing as an explanation of incubation, in which external chance events suggest solutions to incubated problems. The model has received support from laboratory studies. For example, Seifert et al. (­1995) reported two relevant studies on opportunistic assimilation, both of which had similar designs. Participants first tackled general knowledge problems such as “­W hat is a nautical instrument used to measure the position of a ship?” (­­Answer – ​­a sextant). Around 33% of questions could not be answered on the first presentation. In a second stage, participants carried out lexical decision tasks in which some of the words were answers to the questions in the first stage and to later questions presented in a third stage. The third stage was one day later and involved further general knowledge questions, some of which were repeats from the first stage and some were new. As mentioned, some of the answers had been given to some of the old and new questions in the second stage. Thus, in the last stage, the independent factors of whether a question was old or new and whether the answer had been shown in stage 2 or not were completely crossed in a factorial design. A significant interaction was found, such that exposure to answers in stage 2 benefitted 205

Ken Gilhooly

old questions repeated in stage 3, but exposure to answers in stage 2 did not benefit new questions in stage 3. Thus, the external inputs relevant to outstanding unsolved problems aided incubation, as Opportunistic Assimilation would predict, when the relevant “­f ailure indices” had been set. This basic result was strongly replicated in a procedural variation in which the stage 1 and stage 2 trials were interleaved rather than carried out as separate phases. The persistence of activation of “­failure indices” or unachieved goals is also apparent in the phenomenon known as the Zeigarnik effect. Zeigarnik (­1927) had participants undertake a large number of tasks, including arithmetic problems, puzzles, construction tasks and so on. In half the tasks, participants were interrupted before they had completed the tasks and were moved on to other tasks. When, at the end of the task sequence, participants were asked to free recall all the tasks they had worked on, recall of the unfinished interrupted tasks was considerably higher than that for the completed tasks. For example, in one study, 26 out of 32 participants (­80%) showed this trend (­p =.008). This result is again consistent with the ­Opportunistic-​­Assimilation approach that posits lasting activation for unmet goals. Overall, in the ­Opportunistic-​­Assimilation model, Seifert et al. (­1995) proposed a key role for persisting subliminal activation to maintain unconscious t­ask-​­related processing on the target task during incubation. In particular, they point to persistent subliminal activation of the unmet goal during incubation periods following unsuccessful conscious work. During incubation, Seifert et al. suggest that incidental environmental cues related to the solution might activate the goal above threshold, causing the solver to realise that an environmental cue indicates a solution. The general idea of Opportunistic Assimilation has also been adopted in later models. For example, by Simonton (­2003) in his Blind Variation and Selective Retention approach to creativity and, in a slightly different form in Gilhooly’s (­2019) Goal + Associative Network Interaction (­GANI) model of incubation. The GANI model (­Gilhooly, 2016, 2019) proposes a mechanism for “­automatic inspiration” brought about by positive feedback loops of mutual activation between subliminally still active goals and possible solution representations. The model is most applicable to relatively s­mall-​­scale, ­k nowledge-​­rich creativity problems, such as AUTs, or ­small-​­scale insight problems involving ­re-​­structuring, such as Remote Associates Tasks and similar problems. It is consistent with facilitation of solving by external accidental inputs during incubation leading to sudden inspirations or by facilitation on returning to the task following incubation.

Sleep Incubation A common suggestion for handling impasses in ­problem-​­solving is to put the problem aside by “­sleeping on it”, which allows sleep incubation, rather than by staying awake and working on something else, which allows waking incubation. In the following sections, I will discuss sleep as an incubation opportunity. From a practical as well as a theoretical point of view, it is important to consider sleep effects since, in real life, a sleep period will intervene for any problem that is not solved within a single waking day, and so such problems will be open to possible effects of sleep incubation. ­ reco-​­Roman world of classical antiquity (­roughly It is interesting to note that, in the G from the 8th century BC to the 6th century AD), “­incubation”, which derives from Latin “­in” (­on) and “­cubare” (­to lie), was a ritual practice in which ill people slept in Incubation Temples dedicated to the God of Medicine, Asclepius. After sleep, the patients’ dreams were 206

Incubation

interpreted by ­priest-​­healers to guide the solution of their medical problems (­Remberg, 2017). Clearly, the idea of sleep incubation has a long pedigree! The everyday word “­sleep” covers a mixture of states or stages extended over several hours. How sleep in its various stages might affect ­problem-​­solving, and how cognitive processes during sleep are related to cognitive processes that occur during waking incubation and during waking conscious ­problem-​­solving, are currently contentious issues. To review the research on sleep and ­problem-​­solving, we will first outline the main stages of sleep, go on to present some evidence from personal accounts of sleep and ­problem-​ ­solving, then discuss data from surveys and experiments and finally, consider how cognitive processes during sleep might relate to processes during normal waking incubation and normal conscious ­problem-​­solving. First, let us briefly outline the main features of sleep.

Sleep and Its Stages It is useful to distinguish transitional states of consciousness between full waking alertness and sleep. Going from full alertness to sleep and back again are not ­a ll-­​­­or-​­none events like an electric light suddenly going off and back on again. There is an initial hypnagogic stage in between full alertness and sleep, and at the other end of the sleep period, a hypnopompic stage between sleep and full alertness. Since there are often brief awakenings during a night of sleep, these stages may occur briefly more than once in a prolonged sleep period (­see also Lockley & Foster, 2012, for more details on sleep). Broadly, sleep takes two main forms, namely, rapid eye movement (­R EM) and ­non-​ R ­ EM sleep. If people are woken from REM sleep, they will nearly always report that they were dreaming. REM sleep involves bursts of rapid horizontal and vertical eye movements, loss of muscle tone and “­desynchronised” electroencephalogram (­E EG) activity, comprising ­h igh-​­frequency, s­mall-​­amplitude beta waves. The EEG pattern in REM sleep is strangely similar to that observed during wakefulness (­hence REM sleep being labelled “­paradoxical” sleep, see Peigneux et al., 2010, ­p. 167). REM sleep is most common in the second half of the normal sleep period and predominates at the end of sleep. Approximately 75% of sleep is n ­ on-​­REM sleep. Dreams are estimated to occur in about 40% of the ­non-​­REM sleep periods, but dreams in this stage tend to be shorter, less vivid, simpler and less bizarre than the dreams in the REM periods. The EEG patterns in the ­non-​ R ­ EM sleep periods are quite distinct from the REM or waking patterns and feature large slow waves; hence, these periods are sometimes known as Slow Wave Sleep (­SWS). Beneficial effects of sleep periods on learning and memory have been shown for a range of tasks, for example, perceptual learning and learning to play video games (­Hobson, 2005). Stickgold and Walker (­2007) suggested that sleep is involved in converting s­hort-​­term memories into l­ong-​­term memories as well as in the subsequent strengthening of new, l­ong-​ ­term memories. Sleep seems to involve consolidation of some connections and pruning of other connections to leave only the b­ est-​­established (­strongest) connections for l­onger-​­term use. Learning new materials and hence having new connections made while awake seems to be associated with ­re-​­activation, during sleep, of the neural circuits that were initially formed when awake. Although it is established that sleep does have important beneficial effects on learning (­Hobson, 2005; Stickgold & Walker, 2007), are there also important sleep effects on ­problem-​ ­solving and creativity? We will now consider the research evidence, starting, as we did with waking incubation, with personal accounts. 207

Ken Gilhooly

Personal Accounts of Sleep Incubation Many famous anecdotal accounts implicating sleep in creative thinking have been given over the years and are very often cited in the introductory sections of papers reporting more objective studies. A frequently cited account is that of the eminent physiologist, Otto Loewi, who reported (­in 1953 and 1960) that, in 1921, during a dream, he devised an experiment to test whether communication between neurons was chemical. In the experiment, two surgically removed, but still beating, frog hearts were placed in separate glass vessels containing saline solutions. He stimulated one heart via its vagus nerve, causing it to slow down, and then added some of the saline solution from the stimulated heart’s glass vessel into the vessel containing the other heart. The second heart slowed down in response, showing a chemical component of ­inter-​­neuron transmission. He repeated the experiment by stimulating the accelerator nerve of the first heart. When the saline liquid was transferred to the second heart, it accelerated. These studies established chemical communication between neurons. He described (­1960; see also 1953) the origin of the experiment as follows: The night before Easter Sunday of that year I awoke, turned on the light, and jotted down a few notes on a tiny slip of thin paper. Then I feel asleep again. It occurred to me at six o’ clock in the morning that during the night I had written down something most important, but I was unable to decipher the scrawl. The next night, at three o’ clock, the idea returned. It was the design of an experiment to determine whether or not the hypothesis of chemical transmission that I had uttered seventeen years ago [1903] was correct. I got up immediately, went to the laboratory, and performed a simple experiment on a frog heart according to the nocturnal design. (­From Johnson, 2010, p­ p. ­100–​­101; see also Koestler, 1964, ­p. 205) Striking though this and many such anecdotes are, most accounts were first given long after the alleged events were supposed to have occurred, and so questions arise about their validity. Indeed, some commonly cited accounts involving sleep, ­semi-​­sleep states and dreams or ­d ream-​­like states have been thoroughly discredited (­e.g., Kekule’s story of a dream or possibly a daydream leading to the cyclic structure of benzene or Samuel Taylor Coleridge’s supposed dreaming up 200 or more lines of “­Kubla Khan”, see Weisberg, 2006, ­pp. ­73–​­78). Let us now go beyond intriguing but questionable anecdotes and see if “­sleeping on it” actually works in more systematic surveys and controlled experimental settings.

Empirical Studies of Sleep Effects on P ­ roblem-​­solving Surveys Survey studies have asked people to report on whether they have experienced the effects of sleep on ­problem-​­solving and creativity. Barrett (­1993, 2001) found that both students and creative professionals s­elf-​­reported dreaming about problems and finding solutions in ­d reams – especially when problem were ​­ highly visual or the solutions involved unusual uses or relationships. ­Root-​­Bernstein et al. (­1995) carried out a survey study of 38 scientists, evenly split between physical and biological fields, including four Nobel Prize winners. Among the questions, they were asked when ideas occurred: 13% reported ideas occurring during dreams, 208

Incubation

18% when falling asleep and 24% on awakening. Curiously, reporting that ideas occurred during dreams was significantly associated with a scientist’s publication impact (­i.e., how often their papers are cited by other authors.). Ovington et al. (­2015) undertook a ­large-​­scale survey of a representative sample of Australian participants (­N = 1114) on factors associated with self reports of insight in p­ roblem-​ ­solving. Of the 80% who reported having had insight experiences during p­ roblem-​­solving, some 16% reported that their insights were usually linked to some particular place or time. Of those, around 75% reported insights were tied to sleep, falling asleep (­hypnagogic stage) or waking (­hypnopompic stage). This study supports the view that s­ leep-​­related insights are fairly common, but unfortunately, no finer analysis is given of any linkage to specific stages of sleep. The survey results support a facilitating effect of sleeping incubation for p­ roblem-​­solving, but controlled laboratory studies offer more convincing tests of the matter and will now be discussed.

Experimental Studies Wagner et al. (­2004) reported that sleep facilitated the discovery of a hidden rule in number sequences (­using the Number Reduction Task or NRT). Participants were given ­eight-​­digit strings composed only of the digits 1, 4 and 9 and were to produce responses to each digit in the stimulus strings, from left to right, according to two complex rules. Correctness and speed of last responses were measured. For the last three responses in each ­eight-​­digit input, there was a simple but not obvious rule, which was not given in advance, that the last three response digits always mirrored the second, third and fourth responses. Because participants did not know at the start that there was an easy way to carry out the task, they all began using the complex rules. After initial familiarisation training, participants had eight hours of sleep, eight hours of nighttime wakefulness or eight hours of daytime wakefulness before resuming the task. Following sleep, nearly 60% of participants discovered the simple rule, but only 23% of the ­non-​­sleep participants discovered the simple rule. The participants who had slept and discovered the simple rule did so after an average of 135 trials, while the ­non-​­sleep participants who discovered the simple rule required 192 trials on average. As Stickgold and Walker (­2004) noted, there are some striking features of these results. First, the sleep participants did not wake up with the solution, as personal anecdotes and surveys have sometimes reported (­e.g., Poincaré, 1910); rather, they came to the solution faster and more often than did the ­non-​­sleep participants when they resumed the task. On the basis of subjective tiredness ratings, the sleep group and the daytime awake group were equally alert and yet differed markedly in solution rates. Also, supplementary groups that had sleep or wake periods before tackling the NRT without breaks performed equally to each other and significantly less well than the sleep incubation group. These checks indicate that the sleep group’s superior performance was not simply due to being better rested. Second, the participants did not have an explicit goal of finding a simple solution to the NRT, so the situation does not exactly mirror the typical “­sleep on it” incubation example, where the person has reached an impasse while trying to achieve a ­pre-​­specified goal. The task situation was closer to one of ­problem-​­finding ­and -​­solving as against simply ­problem-​­solving where the problem has been p­ re-​­set by the person or by an experimenter. A ­follow-​­up study (­Verleger et al., 2013, Experiment 3) indicated that the occurrence of SWS sleep was associated with ­post-​­sleep insight. Related to the involvement of SWS sleep, 209

Ken Gilhooly

Debarnot et al. (­2017) found no difference in insight occurrence following sleep (­18%) and following a control wake period (­23%), among older participants aged c. 60 years. The authors suggest that the lack of an effect of sleep on subsequent insight rates may have been due to the marked reduction in SWS generally found in older people (­Cajochen et al., 2006). Sio et al. (­2013) found that previously unsolved difficult Remote Associate Test (­R AT) items were better solved after a ­12-​­hour gap that included sleep of c. 7.5 hours than after a ­12-​­hour period of waking incubation opportunity. By presenting ­task-​­related odour cues during sleep, Ritter et al. (­2012) attempted to influence processing during sleep itself towards a creative task that had been set aside before sleep. The method was based on previous studies that had found that covertly reactivating new memories during sleep, by means of conditioned odours, improves later memory test performance (­Rasch et al., 2007). Ritter et al.’s creativity task was to generate ideas for how to motivate people to take part in volunteer work. Participants were instructed in the task before sleeping and were told that they would be asked to report solutions after sleeping. (­The procedure resembles that of Dijksterhuis’s Immediate Incubation” p­ aradigm – ​­although there is less control over the intervening activities between presentation and test than in the normal awake form of immediate incubation and so, some intermittent conscious work seems very likely in the sleep paradigm between task instruction and sleep.) When being informed about the task, two “­odour” groups were exposed to an o ­ range-​­vanilla odour. A third control group received ­no-​­odour exposure. As the participants slept, one group was exposed to a repeat of the ­orange-​­vanilla odour, one group received a control odour (­fresh tonic) and a third group received no odour. After sleeping with the “­conditioned odour” associated with the creativity task, participants’ responses were rated as more creative than those of the control odour group and those of the ­no-​­odour group. Overall, Ritter et al. concluded that task ­re-​­activation during sleep can trigger c­ reativity-​­related processes, and this method could potentially be applied in real life. Studies of sleep effects on creativity and insight have often used the RAT (­or its modern descendant, the Compound Remote Associates); however, it can be argued that the RAT primarily involves the generation of associates rather than any r­e-​­structuring and may not be a good representative of insight tasks. To meet this objection, Schönauer et  al. (­2018) have examined the effects of sleep on classic insight problems and on coming to understand how magic tricks are done. The study involved three ­well-​­known insight tasks, namely, Matchstick Arithmetic (­e.g., Öllinger et al., 2008), the N ­ ine-​­Dot task (­Scheerer, 1963) and ­ ight-​­Coin problem (­Ormerod et al., 2002). Additionally, participants were presented the E with a short video clip of ten magic tricks and instructed to try to explain how they were done (­Danek et  al., 2014). After initial presentations (­30 mins), one group had a t­hree-​ h ­ our sleeping incubation period and a second group a t­ hree-​­hour waking incubation period. There were no differences between the sleep and waking incubation groups in a 3­ 0-​­minute ­re-​­testing period. The continuous work control groups performed at a very similar level to the sleeping and awake incubation groups, and no overall incubation effects were obtained in this study. The authors suggest that effects of incubation and of s­ leep-​­based incubation in particular may be strongest for tasks highly dependent on the generation of unusual associations (­d ivergent tasks) as against tasks highly dependent on ­re-​­structuring. Using a ­re-​­activation procedure, Sanders et al. (­2019) tested whether manipulating information processing during SWS sleep impacted insight problem incubation and solving. It was predicted that reactivating previously unsolved problems during sleep could help people solve them. In the evening, participants were presented with insight puzzles (­m atchstick, rebus, verbal and spatial), each arbitrarily associated with a different sound. While participants 210

Incubation

slept overnight, half of the sounds associated with the puzzles they had not solved were presented during SWS. The next morning, participants solved 31.7% of cued puzzles, compared with 20.5% of uncued puzzles (­a 55% improvement). Moreover, ­cued-​­puzzle solving correlated with ­cued-​­puzzle memory. Overall, the results indicated that cuing puzzle information during SWS sleep can facilitate insight solving. In a f­ollow-​­up study, Sanders and Beeman (­2021) sought to examine the role of forgetting fixations in sleep incubation effects using the ­re-​­activation paradigm. Overall, they found a similar pattern of benefits from ­re-​ a­ ctivation during sleep incubation as in Sanders et al. (­2019; although not reaching statistical significance). ­Re-​­activation cueing during sleep aided problems with more fixating information more than problems with less fixating information. This pattern indicated that simple forgetting of fixating information was not a likely explanation for sleep incubation effects; rather, it suggested that reactivating fixation information enabled unconscious processing to occur that overcame fixating information. Putting together the results from a number of experimental sleep studies, involving a total of 560 participants (­Beijamini et al., 2014; Cai et al., 2009; Debarnot et al., 2017; Landmann et al., 2016; Monaghan et al., 2015; Ritter et al., 2012; Sanders et al., 2019; Schönauer et al., 2018; Sio et al., 2012; Verleger et al., 2013; Wagner et al., 2004; Walker et al., 2002), I found an average weighted effect of d = +.43 in favour of positive sleep incubation effects and concluded that, overall, sleep does aid creativity and insight (­Gilhooly, 2019, ­Chapter 6). ­ on-​­REM sleep, memory Lewis et al. (­2018) have developed a model in which, during n “­replay” abstracts rules from learned information, while in REM sleep, memory “­replay” promotes novel associations in already known information. The interleaving of REM and ­non-​­REM through a prolonged sleep period allows formation of clarified knowledge structures (­during ­non-​­R EM) and allows ­re-​­structuring during REM, which aids creative thought. Regarding the processing of problems during REM and ­non-​­REM, involving the neocortex and the hippocampus, Lewis et al. (­2018) propose an analogy with two researchers who start working on the same problem together (­a s in the n ­ on-​­REM stage, the activities of the hippocampus and the neocortex are tightly coupled), then go away and each think about it separately (­a s in the REM stage when hippocampus and neocortex are uncoupled). When the researchers meet again, they have developed different perspectives and then work to bring them together, as the hippocampus and neocortex work together during the next ­non-​­R EM episode to r­ e-​­structure knowledge. REM sleep periods seem to involve the spreading of activation widely in multiple directions, leading to ­co-​­activation of concepts that are not normally ­co-​­activated, which is reflected in the bizarre nature of dream experiences that often combine weakly connected concepts and ­attributes – ​­e.g., a talking baby or a flying dog seem unexceptional in a dream. Such ­w ide-​­ranging activation and ­co-​­activation could generate novel solution possibilities during dreams or facilitate their occurrence upon wakening (­during hypnopompic states or later) by leaving a residue of unusual connection patterns and activations. These unusual patterns of connection strengths and activations may be quite ­short-​­lived and so be most influential in the hypnopompic state and briefly afterwards, which would fit subjective reports of useful solution ideas being associated with waking. Valuable ideas arising in this way that connect with a current goal would typically go on to be rehearsed and otherwise processed (­a s in the Verification stage of Wallas’s model) for ­longer-​­term storage and use on wakening, and so escape the fate of most dream content, which is not converted into lasting memory traces and is usually lost forever from explicit memory. Can we now answer the question asked at the start of this section? If you are at an impasse in tackling a problem, is it helpful to sleep on it? There is experimental evidence that this is 211

Ken Gilhooly

beneficial. So far, the experimental results suggest small to m ­ edium-​­sized beneficial e­ ffects –​ ­not an uncommon situation in psychology. From a practical point of view, checking on waking whether solutions to problems one has slept on now seem obvious, or at least closer to solution, may be beneficial and surely can do no harm!

Conclusions Popular wisdom has it that if making no progress on a problem, one should either “­Leave it!” for a while or “­Sleep on it!” These recommendations for either waking or sleeping incubation have been supported by experimental studies as well as by surveys and anecdotal accounts. The exact mechanisms underlying incubation effects are not yet determined but seem likely to involve unconscious spreading activation coupled with and influenced by persistent subliminal goal activation.

References Anzai, Y., & Simon, H. A. (­1979). The theory of learning by doing. Psychological Review, 86, ­124–​­140. Barrett, D. (­1993). The “­committee of sleep”: A study of dream incubation for problem solving. Dreaming, 3, ­115–​­122. Barrett, D. (­2001). Comment on Baylor: A note about dreams of scientific problem solving. Dreaming, 11, ­93–​­95. Beijamini, F., Pereira, S. I. R., Cini, F. A., & Louzada, F. M. (­2014). After being challenged by a video game problem, sleep increases the chance to solve it. PLoS One, 9(­1), e84342. Cai, D. J., Mednick, S. A., Harrison, E. M., Kanady, J. C., & Mednick, S. C. (­2009). REM, not incubation, improves creativity by priming associative networks. Proceedings of the National Academy of Sciences U.S.A., 106, ­10130–​­10134. Cajochen, C., Munch, M., Knoblauch, V., Blatter, K., ­Wirz-​­Justice, A., Carrier, J., & Monk, T. H. (­2006). ­A ge-​­related changes in the circadian and homeostatic regulation of human sleep. Chronobiology International, 23, 4­ 61–​­474. Campbell, D. T. (­1960). Blind variation and selective retentions in creative thought as in other knowledge processes. Psychological Review, 67(­6), ­380–​­400. Csikszentmihalyi, M. (­1996). Creativity: Flow and the psychology of discovery and invention. New York: HarperCollins. Danek, A. H., Fraps, T., von Muller, A., Grothe, B., & Öllinger, M. (­2014). Working wonders? Investigating insight with magic tricks. Cognition, 130, ­174–​­185. Debarnot, U., Rossi, M., Faraguna, U., Schwartz, S., & Sebastiani, L. (­2017). Sleep does not facilitate insight in older adults. Neurobiology of Learning and Memory, 140, ­106–​­113. Dewey, J. (­1910). How we think. Boston, MA: D.C. Heath & Co. Dijksterhuis, A., & Meurs, T. (­2006). Where creativity resides: The generative power of unconscious thought. Consciousness and Cognition, 15, ­135–​­146. Dodds, R. A., Ward, T. B., & Smith, S. M. (­2012). A review of the experimental literature on incubation in problem solving and creativity. In M. A. Runco (­Ed.), Creativity research handbook (­Vol. 3). Cresskill, NJ: Hampton Press. Ericsson, K. A., & Simon, H. A. (­1984). Protocol analysis. Cambridge: MIT Press. Evans, J. St. B. T. (­2008). Dual processing accounts of reasoning, judgment, and social cognition. Annual Review of Psychology, 59, ­255–​­278. Ghiselin, B. (­1952). The creative process: A symposium. New York: Mentor. Gilhooly, K. J. (­2016). Incubation and intuition in creative problem solving. Frontiers in Psychology, 7, 1076. Gilhooly, K. J. (­2019).  Incubation in Problem Solving and Creativity: Unconscious Processes. Abingdon: Routledge. Gilhooly, K. J., Fioratou, E., Anthony, S. H., & Wynn, V. (­2007). Divergent thinking: Strategies and executive involvement in generating novel uses for familiar objects. British Journal of Psychology, 98, ­611–​­625.

212

Incubation Gilhooly, K. J., Georgiou, G. J., Garrison, J., Reston, J., & Sirota, M. (­2012). Don’t wait to incubate: Immediate versus delayed incubation in divergent thinking. Memory & Cognition, 40, ­966–​­975. Guilford, J. P. (­1971). The nature of human intelligence. New York: ­McGraw-​­H ill. Guilford, J. P., Christensen, P. R., Merrifield, P. R., & Wilson, R. C. (­1978). Alternate uses: Manual of instructions and interpretations. Orange, CA: Sheridan Psychological Services. Hobson, J. A. (­2005). Dreaming: A very short introduction. Oxford: Oxford University Press. James, W. (­1880). Great men and their environment. Atlantic Monthly, 46, ­4 41–​­4 49. Johnson, S. (­2010). Where good ideas come from: The natural history of innovation. London: Allen Lane. Kahneman, D. (­2011). Thinking, fast and slow. London: Allen Lane. Koestler, A. (­1964). The act of creation. London: Picador edition (­1975). Landmann, N., Kuhn, M., Maier, ­J-​­G., Feige, B., Spiegelhalder, K., Riemann, D., & Nissen, C. (­2016). Sleep strengthens but does not reorganize memory traces in a verbal creativity task. Sleep, 39, ­705–​­713. Lewis, P. A., Knoblich, G., & Poe, G. (­2018). How memory replay in sleep boosts creative problem solving. Trends in Cognitive Sciences, 22, 4­ 91–​­503. Lockley, S. W., & Foster, R. G. (­2012). Sleep: A very short introduction. Oxford: Oxford University Press. Loewi, O. (­1953). The workshop of discoveries. Lawrence, KA: Kansas University Press. Loewi, O. (­1960). An autobiographic sketch. Perspectives on Biology and Medicine, 4, ­1–​­25. Miller, A. I. (­2000). Insights of genius: Imagery and creativity in science and art. Cambridge, MA: MIT Press. Monaghan, P., Sio, U. N., Lau, S. W., Woo, H. K., Linkenauger, S. A., & Ormerod, T. C. (­2015). Sleep promotes analogical transfer in problem solving. Cognition, 143, 2­ 5–​­30. Nisbett, R. E.,  & Wilson, T. D. (­1977). Telling more than we can know: verbal reportson mental processes. Psychological Review, 84, ­231–​­259. Öllinger, M., Jones, G., & Knoblich, G. (­2008). Investigating the effect of mental set on insight problem solving. Experimental Psychology, 55, ­269–​­282. Olton, R. M., & Johnson, D. M. (­1976). Mechanisms of incubation in creative problem solving. American Journal of Psychology, 7, 6­ 17–​­630. Ormerod, T. C., Macgregor, J. N., & Chronicle, E. P. (­2002). Dynamics and constraints in insight problem solving. Journal of Experimental Psychology: Learning, Memory & Cognition, 28, 7­ 91–​­799. Ovington, L. A., Saliba, A. J., Moran, C. C., Goldring, J., & MacDonald, J. B. (­2015). Do people really have insights in the shower? The when, where and who of the Aha moment. Journal of Creative Behavior, 52, ­21–​­34. Patrick, C. (­1935). Creative thought in poets. Archives of Psychology, 178, ­1–​­74. Patrick, C. (­1937). Creative thought in artists. Journal of Psychology, 4, ­35–​­73. Patrick, C. (­1938). Scientific thought. The Journal of Psychology, 5, ­55–​­83. Peigneux, P., Schmitz, R., & Urbain, C. (­2010). Sleep and forgetting. In S. Della Sala (­Ed.), Forgetting (­­pp. ­167–​­184). Hove: Psychology Press. Poincaré, H. (­1910). Mathematical creation. The Monist, 20, ­321–​­333. (­Translation by G. Halstead). Rasch, B., Buchel, C., Gais, S., & Born, J. (­2007). Odor cues during ­slow-​­wave sleep prompt declarative memory consolidation. Science, 315, ­1426–​­1429. ­ reco-​­Roman World. Leiden: Renberg, G. (­2017). Where Dreams May Come: Incubation Sanctuaries in the G Brill. Ritter, S. M., Strick, M., Bos, M. W., van Baaren, R. B., & Dijksterhuis, A. P. (­2012). Good morning creativity: Task reactivation during sleep enhances beneficial effect of sleep on creative performance. Journal of Sleep Research, 21, ­643–​­647. ­Root-​­Bernstein, R. S., Bernstein, M., & Garnier, H. (­1995). Correlations between avocations, scientific style, work habits, and professional impact of scientists. Creativity Research Journal, 8, ­115–​­137. Russo, J. E., Johnson, E. J., & Stephens, D. L. (­1989). The validity of verbal protocols. Memory and Cognition, 17, ­759–​­769. Sanders, K. E. G., & Beeman, M. (­2021). Sleep and incubation: Using problem r­e-​­activation during sleep to study forgetting fixation and unconscious processing during sleep incubation. Journal of Cognitive Psychology, in press. Sanders, K. E. G., Osburn, S., Paller, K. A., & Beeman, M. (­2019). Targeted memory reactivation during sleep improves ­next-​­day problem solving. Psychological Science, 30, ­1616–​­1624. Scheerer, M. (­1963). Problem solving. Scientific American, 208, 1­ 18–​­131. Schönauer, M., Brodt, S., Pöhlchen, D., Breßmer, A., Danek, A. H., & Gais, S. (­2018). Sleep does not promote solving classical insight problems and magic tricks. Frontiers in Human Neuroscience, 12, ­1–​­11.

213

Ken Gilhooly Segal, E. (­2004). Incubation in insight problem solving. Creativity Research Journal, 16(­1), ­141–​­148. Seifert, C. M., Meyer, D. E., Davidson, N., Patalano, A. L., & Yaniv, I. (­1995). Demystification of cognitive insight: Opportunistic assimilation and the ­prepared-​­mind perspective. In R. J. Sternberg & J. E. Davidson (­Eds.), The nature of insight (­­pp. ­65–​­124). Cambridge, MA: MIT Press. Simon, H. A. (­1966). Scientific discovery and the psychology of problem solving. In R. Colodny (­Ed.), Mind and cosmos (pp. 22–40). Pittsburgh, PA: University of Pittsburgh Press. Simonton, D. K. (­1995). Foresight in insight? A Darwinian answer. In R. J. Sternberg and J. E. Davidson (­Eds.), The Nature of Insight (pp. 465–494). Cambridge, MA: MIT Press. Simonton, D. K. (­2003). Scientific creativity as constrained stochastic behaviour: The integration of product, person and process perspectives. Psychological Bulletin, 129, ­475–​­494. Sio, U. N., Monaghan, P., & Ormerod, T. (­2013). Sleep on it, but only if it is difficult: Effects of sleep on problem solving. Memory and Cognition, 41, ­159–​­166. ­ eta-​­analytic Sio. U. N., & Ormerod, T. C. (­2009). Does incubation enhance problem solving? A m review. Psychological Bulletin, 135, ­94–​­120. Stickgold, R., & Walker, M. (­2004). To sleep, perchance to gain creative insight. Trends in Cognitive Sciences, 8, ­191–​­192. Stickgold, R., & Walker, M. P. (­2007). ­Sleep-​­dependent memory consolidation and reconsolidation. Sleep Medicine, 8(­4), ­331–​­343. ­ obbits-​­Orcs problem. Cognitive Psychology, Thomas, J. C. Jr. (­1974). An analysis of behavior in the H 6, ­257–​­269. Varendonck, J. (­1921). The psychology of daydreams. New York: Palgrave Macmillan. Verleger, R., Rose, M., Wagner, U., Yordana, J., & Kolev, V. (­2013). Insights into sleep’s role for insight: Studies with the number reduction task. Advances in Cognitive Psychology, 9, ­160–​­172. Wagner, U., Gais, S., Haider, H., Verleger, R., & Born, J. (­2004). Sleep inspires insight. Nature, 427 (­6972), ­352–​­355. Walker, M. P., Liston, C., Hobson, J. A.,  & Stickgold, R. (­2002). Cognitive flexibility across the ­sleep-​­wake cycle: R ­ EM-​­sleep enhancement of anagram problem solving. Cognitive Brain Research, 14, ­317–​­324. Wallas, G. (­1926). The art of thought. London: Jonathan Cape. Weisberg, R. W. (­2006). Creativity: Understanding innovation in problem solving, science, invention and the arts. New York: J. Wiley & Sons. Weisberg, R. W. (­2018). Insight, problem solving, and creativity: An integration of findings. In F. ­Vallee-​­Tourangeau (­Ed.), Insight: On the origins of new ideas (­­pp. ­191–​­215) London: Routledge. Wotiz, J. H., & Rudofsky, S. (­1984). Kekulé’s dream: Fact or fiction? Chemistry in Britain, 20, ­720–​­723. Wundt, W. (­1893). Grundzuge der physiologischen psychologie, Vol. II. Leipzig: W. Engleman. p­ . 521. Zeigarnik, B. (­1927). Das Behalten von erledigten und unerledigten Handlungen. Psychologie Forschung, 9, ­1–​­85.

214

14 OF NIGHT AND LIGHT AND THE ­HALF-​­LIGHT1 The Role of Multidimensions of Emotion and Tolerance of Uncertainty in Creative Flow Genevieve M. Cseh

Cognition and emotion (­or affect2) are inextricably linked. We know, for instance, that at the most basic level, there is a difference between how ‘­negative’ affective states and ‘­positive’ affective states impact cognition. Fredrickson’s (­1998) ‘­­broaden-­​­­a nd-​­build’ theory posits that ‘­positive’ affect broadens one’s ‘­­thought-​­action repertoires’ into more varied, global, associative, and exploratory patterns; ‘­negative’ states like fear or anger, on the other hand, narrow attentional focus and promote specific patterns of thought and behaviour characterised by ‘­fight, flight, or freeze’ responses. This makes intuitive, evolutionary s­ense – ​­when we feel under threat, we are motivated by a s­ ingle-​­track thought of survival and r­ isk-​­aversion, while when we feel safe and at ease, we have the freedom to explore and experiment. The b­ roaden-­​ ­­and-​­build theory therefore suggests that it is positive affect (­PA) that can be relied on to facilitate the unusual, associative, and exploratory thinking patterns that underlie creativity. Indeed, there is a host of research evidence that supports this notion (­e.g., Baas et al., 2008; Isen, 1987). However, other research suggests that the story is not that simple and that, in fact, ‘­negative’ affect facilitates at least certain aspects or stages of the creative process more than PA (­K aufmann & Vosberg, 1997; Martin et al., 1993; Mraz & Runco, 1994). Other research suggests that we should move beyond the simplistic, primal duality of positive versus negative and discrete, unidimensional emotions. It has been shown that emotions can be: a blend of what is traditionally considered both ‘­negative’ and ‘­positive’ ­concomitantly – ​­i.e., mixed, dialectic emotions (­Larsen  & McGraw, 2011; 2014); socially constructed and not universal, varying significantly between individuals and cultures (­Feldman Barrett, 2017); and many more dimensions and prisms through which we can view emotional states (­Baas et al., 2008; Cacioppo & Berntson, 1999; Latinjak, 2012; Russell, 1980). It is becoming increasingly clear that psychology’s understanding of the link between creativity and emotion is still rudimentary and too simplistic. The relationship between emotion and the highly absorbed, ‘­peak’ state of ‘­flow’ is similarly convoluted. Flow arises from a perfect Goldilocks balance between skill and the difficulty level of a task (­Csikszentmihalyi, 1992/­2002). Although some research suggests that a ‘­positive’ emotional state may help lay the foundation for flow to occur more than less DOI: 10.4324/9781003009351-16

215

Genevieve M. Cseh

positive states (­Cseh et al., 2015), emotion during flow is said to be almost ­non-​­existent, or consciously unnoticed; only afterwards, on ‘­waking’ from the flow state, does emotion (­or awareness of it at least) pour back in, almost exclusively reported as highly positive (­A sakawa, 2010; Rogatko, 2007). However, physiological measures show that cortisol levels (­i.e. stress hormones) rise and heart rate variability falls during flow, suggesting increased mental strain (­Keller et al., 2011). By definition (­Csikszentmihalyi, 1992/­2002), flow occurs when we push ourselves out of our ‘­comfort zone’ ­ever-­​­­so-​­slightly and actively embrace challenge, and, in fact, is most likely to occur when both skill and challenge levels are high (­Massimini et al., 1987). The outcome of flow may be highly ‘­positive’ and r­ ewarding – and ​­ this is indeed the very reason why people seek out ‘­flow’ experiences and thus pursue challenging activities (­Massimini et al., 1988) – ​­but the road leading to and through it may be emotionally rockier and more nuanced than it might seem at first blush.

‘­Positive’ Versus ‘­Negative’ Emotions At the core of this complicated relationship between emotion and both creativity and flow lies the question of the appropriateness of the continued focus on the blunt polarity of ‘­negative’ versus ‘­positive’ emotion and what those two terms actually entail. The field of ‘­positive’ psychology, as it struggles to come to terms with its own definition, scope, and purpose (­Lomas et al., 2020), may be able to shed some light on this issue and inspire a more nuanced understanding of how emotion relates to both creativity and flow. ‘­Positive’ psychology was founded as a counterweight to what was seen as a field too focussed on illness and the dark side of humanity (­Seligman  & Csikszentmihalyi, 2000). Pawelski (­2015, 2016) attempted to deconstruct what the term ‘­positive’ means in relation to ‘­positive’ psychology, arguing that the definition of ‘­positive’ is no simple matter and goes beyond cultivating pleasurable experiences. Many of the points relating to what makes a psychology ‘­positive’ have since been elaborated on and incorporated into the ethos of the advent of ‘­second wave’ positive psychology (­Ivtzan et al., 2016; Wong, 2011), which sought to acknowledge and help people to balance and cope with the inevitably dual nature of e­ xistence –​ ­the light, the dark, and those things in between and simultaneously both. Some would argue that this is making ‘­positive’ psychology obsolete and just ­re-​­calibrating the field into a more balanced ‘­psychology’ (­which was ostensibly the original aim: Gable & Haidt, 2005). This philosophical unpacking of the concept of ‘­positive’ versus ‘­negative’ also should extend to our views on emotion. Psychologists have for decades been trying in vain to settle on a consensus list of core, universal emotions (­Cowen & Keltner, 2017; Ekman, 1992), and most common conceptions of affective states or emotions organise them into this b­ lack-­​­­or-​ ­white polarity. For example, the Positive and Negative Affect Schedule (­PANAS: Watson et al., 1988) remains one of the most used measurement instruments in affect/­emotion research (­Tran, 2013), despite criticisms that it fails to capture more complex dimensions of emotion (­Gaudreau et al., 2006; Killgore, 2000). Kashdan and ­Biswas-​­Diener (­2015) and others (­see Gruber  & Moskowitz, 2014) have since questioned this polarity, suggesting that when we think of ‘­positive’ emotions, we need to also consider more than just the subjective pleasantness of the emotion for the person experiencing it, the most commonly assumed reference point for the ­positive-​­negative distinction (­Cohn & Fredrickson, 2009). The wider impact of the emotional state might also need to be considered. Positive feelings may have negative consequences, and vice versa. For example, the ­broaden-­​­­and-​­build theory (­Fredrickson, 1998) suggests that an objective advantage of positive emotions is that they expand our cognitive style to be more global, to 216

Of Night and Light and the Half-Light

see the ‘­big picture’; however, this can also be their disadvantage, meaning that when we experience ‘­positive’ emotions, we may be missing the trees (­a nd the bear waiting behind one, waiting to attack) for the forest, as it were. Positive emotions, for all their benefits, also cause lapses in vigilance, leaving us open to risk and sometimes resulting in gullibility and danger, as well as a number of other unintended negative consequences such as selfishness and more susceptibility to cognitive biases (­Forgas, 2014).

Motivating Versus Demotivating Emotions The next most commonly referred to dimension of emotion/­affect refers to ‘­arousal’ or ‘­activation’, an addition most famously attributed to the circumplex model of emotion (­Russell, 1980), though similar concepts date back to at least the 19th century, where Wundt (­1897) distinguished emotions by not only their pleasantness but also their ‘­a rousing or subduing’ and ‘­strain or relaxation’ properties. This added dimension begins to acknowledge that there may be even more nuanced forces at work in various emotional states. An emotion can have both a valence (­i.e., pleasantness or unpleasantness, referring to the desirability of experiencing a feeling or not) and an energy (­arousal) ­quotient – ​­a push/­pull to either act or not. For example, relaxed, calm states may be subjectively pleasant, but they are unlikely to push a person to act; they may be content to just be or to pursue only passive exploits. Depending on the circumstances, this can be either an advantage or a disadvantage. Contentment is often thought of as a ‘­positive’ state of utter ‘­completeness’ (­Cordaro et al., 2016), with Calhoun (­2017) arguing it is a virtue, signalling the ability to appreciate the good even in an imperfect world. But it is also a state of low to no motivation (­Gable & ­Harmon-​­Jones, 2010). If one is content, or satisfied, with the status quo, there are no incentives to improve, grow, or keep searching. In fact, there is a tendency towards ‘­mood maintenance’ (­Clark & Isen, 1982, as cited in Forgas, 2014), acting in ways that preserve and prolong comfortable mood states. Contentment could be seen as an enemy of progress or ‘­the road to mediocrity’ (­Griswold, 1996, ­p. 17), and by that logic, discontent is potentially a friend to progress, learning, creating, and growth. In short, negative, undesirable, and unpleasant states can serve a positive purpose; they alert us to when there is a need to change or improve something, motivating beneficial action and giving rise to ‘­positive’ outcomes. For example, outrage at injustice can lead to movements that improve millions of lives; guilt can make us aware when we transgress values of fairness or kindness and is therefore ­pro-​­socially positive, if subjectively negative (­K ashdan & ­Biswas-​­Diener, 2015). On the other hand, someone who is bouncing off the walls with joy may be so enthused with energy that they feel compelled to act, explore, and achieve. Both contentment and joy are subjectively pleasant feelings, but they are not equally motivating towards action in the ways that the b­ roaden-­​­­and-​­build theory would suggest of the indiscriminate umbrella term ‘­positive’ emotion (­Fredrickson, 1998). There are also some suggestions that there is an inverted ­U-​­shaped relationship between positive emotions and optimal functioning. Though the energetic side of emotions is important as the initial impetus behind creativity and other actions leading to flow, beyond a certain point, positive, activating emotions turn to mania and may become counterproductive (­Fredrickson, 2013).

The Dialectics and Multidimensions of Emotion Beyond the p­ ositive-​­negative valence and arousal axes, there are a variety of other ways in which emotions have been categorised and can be understood but which so far have garnered 217

Genevieve M. Cseh

much less attention. Further exploration of this is important because, if p­ ositively-​­and negatively valenced emotions each have different ­thought-​­behaviour outcome patterns, as Fredrickson’s (­1998) ­broaden-­​­­and-​­build theory posits, this may also be true of other dimensions of emotion. There may also be subtle interaction effects of multiple dimensions of emotion on cognition and behaviour, including on creativity and flow. For instance, going beyond the basic ­t wo-​­dimensional (­valence x arousal) circumplex model of affect (­Russell, 1980), another proposed dimension is approach/­avoidance-​­motivating emotions (­Frijda, ­1986 – ​­a rguably very similar to both valence and arousal in that there is both a subjective desirability and a motivational component involved that propels or inhibits action and engagement). Latinjak (­2012) proposed more recently that time perspective also may be a dimension along which emotional experiences can be classified; for example, p­ ast-​­oriented emotions may include (­negative) resentment (­about a past occurrence or slight) or (­positive) relief that something feared in the past did not come to pass. Alternatively, ­f uture-​­oriented emotions may include anticipatory emotions like excitement/­hope (­positive) or fear/­d read (­negative) and are likely to play a significant role in motivation and achievement as people use past experiences to predict future outcomes of their present actions (­Gilbert & Wilson, 2007). These categorisations can be less or more detailed in terms of what antecedents and consequences are attributed to the emotion groupings. Robinson (­2008) has developed a taxonomy, adapted in part from Ortony and Turner (­1990), which argues that emotions can be either positive (­P) or negative (­N ), crossed with various other dimensions related to the emotion stimulus: • • • • • •

­Object-​­related (­e.g., interest (­P) or revulsion (­N ) – ​­similar to Frijda’s (­1986) approach/­ avoidance dimension); ­Future-​­oriented (­e.g., hope (­P) or fear (­N ), similar to Latinjak’s (­2012) time perspective dimension); ­Event-​­related (­e.g., gratitude (­P) or grief (­N )); ­Self-​­related (­e.g., pride (­P) or embarrassment (­N )); Social or ­other-​­related (­e.g., generosity (­P) or cruelty (­N )); And what he termed ‘­cathected’ emotions, which are invested in others/­a n object at a deep level (­e.g., love (­P) or hate (­N )).

Ellsworth and Smith (­1988) suggested that there were various ‘­shades of joy’ and that ‘­pleasant’ emotions could be further reduced depending on different appraisals of the situation triggering the e­ motion – ​­specifically, differences in effort, agency, and certainty. A group of emotions known as ‘­epistemic emotions’ are those which specifically arise from and lead to a motivation to learn and gain knowledge and understanding. These tend to include positively valenced emotions such as curiosity, interest, and awe, but also the negatively valenced emotions of confusion, dissatisfaction, or frustration (­Chevrier et al., 2019). They are similar to ‘­achievement’ emotions, which are ‘­emotions that are tied directly to achievement activities or achievement outcomes’ (­Pekrun, 2006, ­p. 317); however, the epistemic emotions are more directly associated with cognitive and ­k nowledge-​­related pursuits and are less concerned with accuracy and outcome (­Vogl et al., 2019). It is this grouping that may be particularly relevant to discussions of creativity, where often a problem is in need of being ­solved – ​­that is, moving from a state of uncertainty about a situation to one of resolution, but where goals are often nebulous or the joy is in the journey more than the final outcome. It is proposed here that this group of emotions may be at the heart of the question of how emotion relates to creativity and flow. 218

Of Night and Light and the Half-Light

But the plot thickens, and thus far the discussion has focussed on discrete emotions felt one at a time. But what about the concept of mixed emotions? Mixed emotions are two or more emotions considered to be mapped on different parts of one or more dimensions, such as ‘­happy’ (­positive valence) and ‘­sad’ (­negative valence), experienced seemingly simultaneously (­Larsen & McGraw, 2014; Scherer, 1998). There is some debate about whether mixed emotions truly exist, whether they are just separate emotions cycling between one another in quick succession (­thus only giving the illusion of them happening simultaneously), or whether they are really being felt in chorus (­Larsen, 2017; Larsen & McGraw, 2011). There is however evidence to support the concept that mixed (­sometimes called ‘­bittersweet’ or ‘­poignant’) emotions exist in tandem (­Larsen & McGraw, 2011, 2014), further complicating our understanding of the intricate dance between emotion, creativity, and flow. Hoemann et al. (­2017), arguing for the theory of constructed emotion, posit that emotional experiences are learned and constructed as internal models based on our past appraisals of links between feelings and perceived situations, which are categorised into conceptual emotion categories. Following from this theory, they suggest that mixed emotions arise when we encounter a situation that recalls aspects of multiple past experiences at the same time, from different emotion categories. They argue that the brain’s prediction/­­prediction-­​­­error-​­correction processes are so fast that the iterative process of selecting and reselecting emotions from various emotion category banks can be subjectively experienced as multiple different emotions simultaneously, though technically they are not. Whether emotions are genuinely mixed or just cycling in r­ apid-​­fire fluctuation, the message is that emotional experience can be complex and feel very paradoxical in the moment. This is potentially also the case with epistemic emotions and their positive or negative versions. It may be possible for someone to feel both curiosity and frustration, interest and confusion at once. In fact, this may be the default, necessary affective state during both creativity and flow, and the resolution of this paradoxical emotional state may be the impetus driving creative flow experiences.

Epistemic Emotion and Tolerance of Ambiguity/­Uncertainty Muis et al. (­2018, ­p. 5) define epistemic emotions as ‘­emotions that result from i­nformation-​ ­oriented appraisals (­i.e., the cognitive component of an emotion) about the alignment or misalignment between new information and existing beliefs, existing knowledge structures, or recently processed information’. Thus, epistemic emotions arise from the confirmation or disputing of previously held knowledge, the gap or tension that arises between past and present states of knowing, and potentially the future implications of that change in knowledge. Within that search for greater understanding and clarity, there are inevitably periods or areas of ambiguity and uncertainty. The exact flavour of epistemic emotions that arises during these periods (­i.e., their valence, arousal level, relation to time and self, etc.) may rely on how one reacts to and perceives u ­ ncertainty – as ​­ a thrilling adventure to be embraced and enjoyed, or as a frightening, lost meandering through a murky shadowland. Or, perhaps, it is a little of both. Williams and Aaker (­2002) found that those who had low duality tolerance (­a concept very similar to ambiguity intolerance and that they also refer to as emotional dissonance) showed less positive responses to adverts appealing to mixed emotions, demon​­ terms of strating a personality dimension to how people react to ambiguity and p­ aradox – in the situation, cognitive appraisals, or emotions. Anderson et  al. (­2019) note that most research on uncertainty or ambiguity and the ‘­tolerance’ of these states assumes a mostly negative affective r­ esponse – ​­that even the word 219

Genevieve M. Cseh

‘­tolerance’ in the terminology implies this underlying ­a ssumption – ​­that is, it is something one must ’tolerate’ or endure as an ordeal. Evolutionarily, it makes intuitive sense that humans are predisposed to want accurate, clear information that reassures them that they are not in danger, and if they are, where that danger is, and how they can deal with it. Anderson et al. suggest that an important mediator between uncertainty and affect might be the mental simulations that people conjure to help them predict and plan to resolve potential outcomes of uncertain situations. Whether a person imagines a pessimistic or optimistic future will influence their emotional response to the uncertain situation they are facing. There is a fundamental negativity bias in mental prediction in unclear situations, as people tend to want to plan for the worst in order to forestall and cope with any potential dangers (­A nderson et al., 2019; Baumeister et al., 2001); this universal tendency towards pessimistic prediction is necessary for survival, leading to negative affect (­NA) and thereby making us generally ­r isk-​­averse. Anderson et al. (­2019) note that, because of this, research to date has assumed that uncertainty is always faced with negative ­emotion – ​­specifically anxiety. There is much research that supports this (­see Carleton, 2016, for a review), and we know from cognitive dissonance research that people will readily change the way they think and behave to avoid having to face contradiction or uncertainty (­Festinger, 1959). However, Anderson et al. argue that this assumption means that research to date has largely neglected to explore and better understand the possible positive associations with uncertainty, such as the frisson of reading mystery novels or the thrill of a gamble. They note that there clearly are personality types and professions that relish and rely on uncertainty, citing scientists and explorers as ­examples – ​­in essence, those who see uncertainty as a desirable and exciting challenge that attracts rather than repels. In relation to flow and optimal performance, particularly in the domain of sports, it has been proposed that there are Individual Zones of Optimal Functioning (­Hanin, 1997, 2000): different people have different tolerance levels for the anxieties associated with challenge and performance, under which they perform at their best. Tolerance of ambiguity (­TOA), also often referred to as uncertainty tolerance and a number of other similar terms (­A nderson et al., 2019), whether used erroneously or not (­Grenier et al., 2005), could be seen in a similar light in relation to creativity; i.e. Individual Zones of Ambiguity (­or Uncertainty) Tolerance, one could say. TOA is a concept that has long been discussed and debated in relation to creativity and creative personalities (­Furnham & Marks, 2013; Merrotsy, 2013; Runco, 2014). By its nature, creativity delves into the unknown. It requires courage because e­ ver-​­looming is the prospect of failure. Although sometimes the ‘­problem’ to be solved in a creative endeavour may be quite concrete, oftentimes the goals are much more ineffable and the ­goal-​­posts are always moving or just out of reach. Arguably, this may be particularly true in artistic or scientific pursuits that deal with conceptual and ­non-​­verbal abstractions, and where predicting how a completely novel output will be received based on previous experience is not possible (­Simonton, 2000). The creative problem to be solved is ‘­d iscovered in the interaction with the elements that constitute it’ (­Getzels & Csikszentmihalyi, 1976, ­p. 247), in an iterative process of wrestling with uncertainty. It is the ability to be ‘­tolerant’ of this lack of clarity and closure, to persevere through it, be able to ignore or accept it, or even perhaps to enjoy the ‘­chase’ in some way that may be a crucial trait in creative individuals, a concept that is backed by a number of studies (­Guilford, 1970; Runco, 2014; Zenasni et al., 2008), though not all (­Merrotsy, 2013). However, the subjective, emotional experience of this ‘­TOA’ is not often discussed in any depth within these studies. Presumably, what is really meant by the phrase is the ability to 220

Of Night and Light and the Half-Light

either cope with the negative emotions that are assumed to accompany creative ambiguity or to not feel them as acutely in the first place as some others might. We know there are affective rewards to creativity (­including flow) that spur creative action, despite its uncertainties and frustrations, which may compensate for the struggles with the unknown. However, is the uncertainty in fact part of the appeal? Csikszentmihalyi (­1992/­2002) considers unambiguous feedback an important prerequisite of the flow experience; however, when it comes to creative endeavours, where ambiguity and uncertainty are inherent, perhaps clear feedback is less important to the experience of flow within this domain; or, it may be important, but only in terms of which emotions accompany uncertainty. It is important to note that ‘­a mbiguity’ and ‘­uncertainty’ often are used ­interchangeably –​ ­indeed, this is something I have been somewhat guilty of myself so far. They are not, however, completely the same, although uncertainty may include the uncertainty of ambiguity (­Grenier et al., 2005). Both might figure quite importantly in creativity and flow, but in different ways and through different emotional dimensions. Grenier et al. posit that ambiguity refers to uncertainty that is rooted in a paradox or a dialectic. The uncertainty in something ambiguous is not necessarily about what might happen in the unknowable ­f uture – ​­e.g., fear of ­failure – ​­so much as it is about having to reconcile or just make peace with potentially contradictory concepts at once. There may be subtly different abilities and characteristics that are required to cope with and thrive emotionally within each type of uncertainty. For example, while uncertainty of the yawning chasm of the unknown may cause ­future-​­oriented, ­self-​­related anxieties (­fear of failure, fear of rejection), fear of ambiguity may be a more puzzling cognitive conundrum with a less s­elf-​­focussed and a more ­present-​­focussed time orientation. Ambiguity might be approached with a feeling of unease, confusion, or tension as comfortable concepts of reality or rational ­cause-­​­­and-​­effect could potentially be thrown in the air, but it is not usually concerned with predicting the future (­Grenier et al., 2005). Therefore, it may not be as impacted by a person’s tendency towards optimism or pessimism (­­future-​­focussed time emotions) or ego concerns (­­self-​­related emotions). Ambiguous situations might be catalysts for more epistemic emotions instead of trying to balance what one thinks one knows about two polar opposites at once, and reconciling them as one, or developing a new perspective. Different people might have different Individual Zones of Tolerance to each of these types of uncertainty anxiety in different situations. Herbert (­2022) has found that creative people are not necessarily any more tolerant of uncertainties that are existential, ­self-​­related threats such as fears of rejection or loss of livelihood, but they are often more open and accepting of the ambiguities and uncertainties of grappling with paradoxical and unclear ideas during the creative process. Perhaps contradictions in previous research findings about whether creative people are more tolerant of ambiguity or uncertainty than the average person come down to a misunderstanding of the subtle differences between types of uncertainty that might be encountered during a creative/­flow experience and the accompanying emotion categories that they trigger.

Creativity and Emotion There is a rather romanticised view of the creative process in our culture; a belief still held by many to some degree is that it is a flash of magical inspiration bestowed by divine osmosis only unto a select few geniuses with minimal effort on the part of the creator (­K im, 2019). However, the cold reality is often quite different: creativity is steeped in effort, ambiguity, and risk (­Sawyer, 2006). It is described simultaneously as an exhilarating, l­ife-​­affirming, 221

Genevieve M. Cseh

positive adventure on the one hand and, on the other, a process marked by deep pain, difficulty, frustration, confusion, insecurity, and fear of failure and ridicule (­Ghiselin, 1952). Much research and popular culture have also focussed on the comorbidity of creativity and mental illness, suggesting there is a possible price to creative genius and its eureka highs, at least for some ( ­Jamison, 2011). The optimal emotional profile that is most conducive to creativity is still unclear. As noted before, the b­ roaden-­​­­and-​­build theory (­Fredrickson, 1998) suggests that pleasant emotions encourage creativity the most. Classically, cognitive experimental research by Isen and her colleagues shows this pattern throughout a number of experiments: participants in ­lab-​ ­based PA induction conditions tended to be more inclusive (­a nd novel) in categorisation tasks (­Isen & Daubman, 1984), showed improved creative p­ roblem-​­solving (­Isen et al., 1987) and made more unusual word associations (­Isen et al., 1985). Fredrickson and Branigan (­2005) showed PA condition participants making more ‘­g lobal’ rather than ‘­local’ selections on a global/­local visual processing task. Furthermore, a ­meta-​­analysis of the links between creativity and ‘­mood’ (­Baas et  al., 2008) shows that, on the whole, positive emotional states do seem to be related to more creative outcomes; however, it is worth noting that their review showed that positive mood seemed to increase creativity compared to neutral moods, not compared to negative moods, which showed no significant difference. Additionally, they found only partial support for the ‘­hedonic tone’ (­valence) hypothesis underlying the b­ roaden-­​­­and-​­build theory (­Fredrickson, 1998), finding that activating positive emotions (­e.g., happiness) increased creativity (­particularly originality and fluency), but deactivating positive emotions (­e.g., calm) did not, highlighting the interactive effect of different emotional dimensions. Other research has shown negative emotions benefitting creativity. NA has been shown to be crucial to p­ roblem-​­finding in the first place (­M raz  & Runco, 1994), perseverance (­Martin et al., 1993), and reframing (­K aufmann & Vosberg, 1997). Additionally, as noted earlier, positive emotion has downsides (­e.g., missing the trees for the forest) and negative emotion has upsides (­e.g., drive, vigilance to detail: Kashdan & ­Biswas-​­Diener, 2015). Leung et al. (­2014) suggest that negative emotion may be beneficial to creativity when it is consistent with personality traits such as neuroticism. There is also evidence that creativity is a product of mixed or ambivalent emotions. Fong (­2006) found that inducing emotional ambivalence led to higher performance on the Remote Associates Task (­Mednick, 1962) in a business school context, proposing that the tension from two conflicting, p­ olar-​­opposite states experienced at once signals that one is in an unusual or complicated situation, therefore triggering more unusual thinking patterns and associations, which are necessary to make creative links. Clearly, there may be many different variables that contribute to the ­emotion-​­creativity link we have yet to fully understand. The generally agreed definition of creativity among most creativity researchers (­Mumford, 2003; Runco & Jaeger, 2012; Sawyer, 2006) suggests that it requires two very different and seemingly contradictory cognitive skill sets: Type 1 (­d ivergent) thinking is the ability to generate unusual new ideas in a free and uninhibited manner, quite effortlessly; Type 2 (­convergent thinking) is the ability to whittle down these generated ideas critically and consciously, through analysis and evaluation, and hone a ‘­correct’ idea until it is realistic, feasible, and serves some kind of valuable end purpose (­A llen & Thomas, 2011; Kahneman, 2011). This ­dual-​­systems cognition works in constant fluctuation and along different but interconnected neural pathways (­A llen  & Thomas, 2011), and different creative processes and typologies may make use of either more or less deliberate or spontaneous emotional 222

Of Night and Light and the Half-Light

or cognitive networks (­Dietrich, 2004b). Novelty without refinement would be aimless, useless, or bizarre, while usefulness without novelty is merely rote manufacture at best or forgery at worst. The two polarities are both vitally important to the success of the whole. The correctly balanced ­tension  – ​­between open, global thought that stems from positive emotions and narrow, ­detail-­​­­and-­​­­problem-​­oriented thinking spurred by negative ­emotion –​ ­forms the holistic yin and yang of creativity (­Fang, 2012). Therefore, it is not one particular type of emotion that seems to lead to creativity; it is a complicated and seemingly paradoxical mixture. However, Baumann and Kuhl (­2005) and Isen (­1987) have argued that PA seems to cause more flexibility between these two types of cognition, but further work is needed to understand how specific ­sub-​­dimensions of positively and even negatively valenced affect may also facilitate or hinder this flexibility.

Flow and Emotion Similarly, duality is inherent in the flow experience as well. A study by Cseh et al. (­2015) showed that baseline PA, before taking part in a figural creative combination task, was associated with higher flow measures during the task. Although correlational, this may suggest that ­pre-​­activity affect might, in some way, be linked to the likelihood or intensity of flow experiences, perhaps like PA is suggested to facilitate creativity according to the b­ roaden-­​ ­­and-​­build theory (­Fredrickson, 1998). On the other hand, Cseh et al. also found that while flow was associated with an increase in PA over the course of the creativity task, there was no such relationship to change in NA, suggesting a potentially more complicated relationship to NA rather than simply a lowering or absence of it. Baumann (­2012) suggests that ‘­flow can be described as a smooth transition of intention into action through positive affect…it takes positive affect (­e.g., anticipation of success) to overcome the inhibition of action and recouple intention memory with its output system: intuitive behaviour control’ (­­pp. ­176–​­177). Baumann and Scheffer (­2010) propose the ‘­a ffective change hypothesis’ of flow, which suggests that affect fluctuates during ‘­achievement’ flow states (­i.e. flow that arises in the pursuit of an active goal to achieve, which creative activities would usually fall under; although, again, creative goals are not always clear and this uncertainty is often considered desirable; Cseh, 2014). Baumann and Scheffer’s hypothesis suggests that PA is reduced in early stages of the flow activity, when engaged in ‘­seeing’ or assessing difficulty levels of the task, and then elevated when acting towards ‘­m astering’ difficulty, and that flow is marked by fluctuating dynamic change in PA in this manner during the course of an activity. There are reminiscences here of the ­dual-​­systems thinking processes of creativity, once again highlighting the dualistic, dynamic, and paradoxical experience of the light and the dark, the positive and negative, difficulty and ease, in both flow and creativity. However, they only refer to fluctuations in PA rather than the interplay between PA and NA. It is not clear whether a decrease in PA assumes an increase in NA, and other dimensions besides valence are not examined. Physiologically, flow is characterised by a number of physical markers that are alternatively associated with either negative or positive emotional states, such as an increase in the stress hormone cortisol, released when struggling (­a nd coping) with a challenge (­Keller et al., 2011; Peifer, 2012). Neural patterns show signs of both effort and ease as well: for example, Katahira et al. (­2018) found in an electroencephalography (­E EG) study that flow was linked to increased theta activity in the frontal regions of the brain, due to mental effort and focus on a mental arithmetic task in which difficulty level was manipulated, and moderate alpha activity in frontal and central regions, indicating that the effort was being managed and not 223

Genevieve M. Cseh

too taxing. Ulrich et al. (­2013) showed that there was a reduction in negative affective arousal during flow, with concomitant deactivation of the amygdala. The hypofrontality theory (­Dietrich, 2004a) suggests that flow is characterised by an ­efficiency-​­promoting, temporary inhibition of activity in the prefrontal cortex, resulting in loss of ­self-​­consciousness and time distortion (­i.e., the proposed u ­ ncertainty-​­related emotion dimensions), and yet focus (­a lso an executive function housed in this neural region) is heightened, and presumably flow is not unduly disrupted despite the evaluative, analytic Type 2 thinking processes (­cognitively taxing activities) interspersed throughout creative activities (­Cseh, 2016; Oliverio, 2008). Ullén et al. (­2013) note that flow states in expert pianists require simultaneous activation and adequate regulation of both the parasympathetic (­relaxation and soothing) and sympathetic (­fight/­fl ight and activity readiness) systems. They suggest that flow is affectively a period of m ­ oderate-​­high positively valenced and aroused affect; however, the dual activation of both activity and relaxation systems suggests that affect is either more complex than simply positive or highly aroused, and may instead be a mixture or a state of constant affective flux. It is not just a question of whether positive or negative emotions facilitate flow, but which dimensions of each kind are at play when flow occurs and how mixed, dialectical emotions are handled to produce the optimal conditions that make the elusive flow state possible. A number of components are said to underlie the flow experience; perhaps the most central is that it requires an optimal balance between the perceived difficulty level of the task at hand and the perceived skill level of the person carrying out the task to match that difficulty (­Csikszentmihalyi, 1992/­2002). Within the list of flow components are hints of the emotional duality inherent in the experience. There is, for instance, intense attentional focus/‘­tunnel vision’ during flow, as well as its characteristic attentional absorption, which is often called being ‘­in the zone’. However, attentional focus is generally associated mainly with negative emotions and their propensity to help us zero in on danger, yet flow is considered a largely positive state that is believed to expand cognitive flexibility and therefore is often assumed to promote greater creative performance, though this is debatable (­Cseh et al., 2015; Landhäußer  & Keller, 2012). Cseh et  al., for instance, found in a ­lab-​­based figural creative construction task that flow during the task coincided with higher s­ elf-​­evaluations of creative performance but not with more objective measures or subjective creativity evaluations by others. The ­flow-­​­­self-​­confidence link is likely tied to another flow ­characteristic – a​­ ‘­paradoxical’ feeling of ‘­effortless control’ – the ​­ phenomenology of coping well with a high challenge ­endeavour – feeling capable and ultimately expecting success (­Csikszentmihalyi, ​­ 1992/­2002). There is also reference in the key components of flow to a need for clear goals and unambiguous feedback. But what of creative pursuits, where the goals are not always clear and the feedback is quite ambiguous (­or uncertain) indeed (­Cseh, 2016; Simonton, 2000)? Csikszentmihalyi (­1992/­2002) argues that in this case, creators internalise learned standards of their field and give themselves feedback. However, another explanation might be that a personality tendency towards tolerance (­or even enjoyment) of ambiguity and/­or uncertainty may be required within the person that helps to either negate the need for such clarity or that allows a creator to persevere longer without comforting certainty, until they can reach some point of satisfaction. During creative tasks, creators may be juggling both ambiguity and uncertainty. In working with complex abstractions, pushing boundaries, and making unusual associations, creators must be accepting of or even actively play with ambiguity, dialectics, and paradox to create something new and exciting. How the work will ultimately be received by others (­Csikszentmihalyi’s ­a ll-​­important ‘­field’ or ‘­domain’ (­1996) or Rhodes’ ‘­press’ (­1961)), who 224

Of Night and Light and the Half-Light

must be persuaded of the merit of the creative output, may however be uncertain as well and requires predictions of the future. Predictions, however, are based on past experience, and this is a fundamental difficulty with ­creativity – ​­past experiences cannot be relied on in predicting reactions to entirely novel outputs (­Simonton, 2000). During flow, this f­uture-​­ and ­self-​­oriented thinking and feeling is and must be suspended. The epistemic emotions accompanying consideration of ambiguous concepts during the creative process, however, may not be suspended and indeed may be crucially motivating. Thus, if the anxieties related to ambiguity or other uncertainty are kept within one’s Individual Zone of Tolerance, uncertainty can be helpful to creativity and flow in a way that unregulated ­f uture-​­and ­self-​ r­ elated uncertainty may not be, even if these epistemic emotions are not entirely pleasant (­for example, a mixture of dissatisfaction and curiosity). Muis et al. (­2018) noted that emotions arise before, during, and after actions or events but that these time periods may cross and intersect via memory of the past and prediction of the future: ‘­Activity emotions occur during engagement in an activity, whereas outcome emotions include prospective outcome emotions, related to possible future successes or failures, and retrospective outcome emotions, linked to previous successes and failures’ (­­p. 5). In the context of achievement (­or specifically creative) flow, activity and perhaps especially outcome emotions during the actual flow state are dampened, with emotion often described as absent or at least not consciously registered, perhaps due to the loss of most s­elf-​­related assessment during flow. Csikszentmihalyi notes: When we are in flow, we are not happy, because to experience happiness we must focus on our inner states, and that would take away attention from the task at hand… Only after the task is completed do we have the leisure to look back on what has happened, and then we are flooded with gratitude for the excellence of that ­experience – ​­then, in retrospect, we are happy. (­1997, ­p. 32) However, retrospective/­predictive outcome emotions, on some subconscious level at least, must still be important to the conditions that lead to and sustain creative flow, as they are a part of the assessment of skill levels and thus confidence in likelihood of success. Outcome emotions are crucial to ­decision-​­making about ­r isk-​­taking and engagement in creative action, based on memory of previous experiences of creative success or failure (­Icekson et al., 2014). Again, in creative endeavours, it can be difficult to rely on memories of previous successes to calculate future success probability, given that, by definition, creativity requires doing what has not been done before (­Simonton, 2000). This means that creativity is fraught with ‘­probability ambiguity’ (­a specific typology of uncertainty referring to the inability to form robust predictive risk assessments; Anderson et al., 2019). Again, whether this is true ambiguity or a different type of uncertainty is debatable, since this would appear to be more ­f uture-​­ and ­self-​­related. Activity emotions can be and often inevitably are ambiguous in a way that can be conducive to flow if they are epistemic, rooted in the ’here and now’ (­Grenier et al., 2005), and focussed externally on play with and resolution of ideas rather than on existential threats to the self. Greater care should be taken in future research to conceptually divide ambiguity from the broader concept of uncertainty when exploring the mixture of emotions that are the best precursors and drivers of creative flow. Additionally, exploring dimensions of emotion beyond positive and negative affect might build a more subtle view of which emotions are dampened versus experienced and helpful or unhelpful during flow. 225

Genevieve M. Cseh

Conclusions and Recommendations The aim of this review was to highlight the inherent emotional complexity within both the creative process and the flow experience. Much of that emotional complexity can be traced back to the inevitably dualistic, contradictory, paradoxical, ambiguous, risky, and uncertain aspects of both, which suggests that it is ambiguity and/­or uncertainty and how people respond to these ­emotionally – ​­and how this can be regulated for optimal functioning and ­wellbeing – ​­that deserves much more scrutiny going forward. However, it has been argued here that ambiguity in relation to creative ideas and the creative process itself has potentially different emotional consequences compared to uncertainty, which is more s­elf-​­and ­future-​ ­focussed and therefore more existentially threatening. The inherently ambiguous and uncertain nature of creativity suggests that creative individuals need to either be accepting of that inevitable ambiguity/­uncertainty, thrive on it, or find some way to resolve the ‘­problem’ by finding or constructing some form of optimistic certainty for themselves. Though the creative process carries ­in-​­built probability ambiguity (­A nderson et al., 2019), a person may trust in their own technical expertise or memories of having faced and conquered similar times of uncertainty before to give them the confidence that they will come to a solution to their creative problem in the end (­Cseh, 2014). This suggests that ­self-​­related emotions (­Robinson, 2008), at least, need to be positive to persevere with creative endeavours and thus to experience creative flow, with low personal uncertainty (­high confidence/­clear knowledge of one’s own skill levels and limitations; Anderson et al., 2019). This stands in contrast to the concept of ‘­creative mortification’, which refers to the reluctance to engage in further creative activity after negative feedback on previous work (­Beghetto, 2014). Given that it would be nearly impossible for any creator to have never experienced previous failure or negative feedback, the question then is: what distinguishes those who persevere despite these knockbacks? Beghetto found that those who were most resistant to creative mortification showed certain characteristics about their s­elf-​­concept, ­ indset – ​­i.e., they showed more of a mastery particularly their belief systems in terms of m or growth mindset orientation, believing they could improve through effort, rather than a fixed mindset that believes talent is inherent and immutable. Mindset training (­Dweck, 2006/­2017) to increase ‘­creative ­self-​­efficacy’ (­Farmer  & Tierney, 2017) might therefore be a good first step in teaching people to be more resistant to fears of failure to encourage creativity and flow. It is possible that the lack of ­self-​­consciousness during flow may help reduce the discomfort of any s­elf-​­related uncertainties and existential anxieties stemming therefrom that inhibit ­creativity – ​­the flipside of that, however, is that it may also make people less apt to notice dangers or pitfalls that uncertainty sometimes signals (­Schüler, 2012). It is also a ­chicken-​ ­egg situation; paying less attention to the self is likely to make achieving flow easier in the first place. Therefore, interventions that help to foster less ­self-​­consciousness or encourage a ​­ to ­self-​­transcendent ­focus – ​­such as mindfulness practices (­Vago & Silbersweig, 2012) – and increase ­self-​­compassion (­Neff, 2011) are likely to be protective in facilitating creativity and flow by reducing existential uncertainty. Uncertainty and ambiguity are not the same (­Grenier et al., 2005), perhaps particularly in relation to ­t ime-​­related, ­self-​­related, and epistemic emotions. Ambiguity during the creative process in terms of a creative idea’s content may trigger both positively and negatively valenced, ­object-​­focussed (­rather than ­self-​­focussed), and ­present-​­focussed (­rather than ­f uture-​ ­focussed) epistemic emotions, e.g., confusion and curiosity. Even the negative epistemic emotions in relation to ambiguous ideas are powerful motivators and work as an itch in need 226

Of Night and Light and the Half-Light

of a scratch, while the positive emotions sustain and may help promote necessary cognitive flexibility between global and local attention and Type 1 and Type 2 thought processes. What may be most important is not which particular emotions are best for creativity or the flow experience, but rather the emotion regulation strategies that are employed to balance the dialectics of different and mixed emotion dimensions inherently at play in the creative process and in flow, to ‘­sit with’ and accept the discomfort of both ambiguity and uncertainty, to view them through an optimistic, hopeful lens as an opportunity for adventure (­or at the very least as a completely normal, routine part of the process), and to nurture and sustain epistemic emotions over achievement or f­uture-​­and ­self-​­related emotions. Indeed, a number of studies show links between emotion regulation and creativity (­Fancourt et al., 2019; Ivcevic & Brackett, 2015; Yeh & Li, 2008). Ivcevic and Brackett (­2015) showed an interaction between the openness to experience personality variable (­often related to TOA: Runco, 2014), emotion regulation, and creativity, suggesting that: …emotion regulation ability appears to help individuals with high openness to transform their preference for new ideas and intellectual or artistic interests into creative behavior by enabling them to manage and influence emotions experienced in the course of the creative process. (­Ivcevic & Brackett, 2015, p­ . 484) Achieving creative flow may then come through the process of ‘­seeking optimal confusion’ (­A rguel et al., 2018), finding the perfect balance of ambiguity and clarity, difficulty and ease, dissatisfaction and satisfaction, and having the emotion regulation and acceptance skills to tolerate any discomfort and find any enjoyment that may accompany that tension. TOA and emotion regulation have garnered some interest as personality variables potentially underlying creativity, but a negativity bias means we know less about how ambiguity may be relished and desired as part of both creativity and flow than how it is often avoided. The emotional experience can be simultaneously difficult and enjoyable, and how those conflicting emotions are experienced and handled deserves more study going forward. If we as a species are typically averse to risk and uncertainty, we have a hurdle to overcome in terms of promoting creativity and flow. Future research should continue to explore the individual differences, environmental factors, and process factors that can help people to overcome this barrier to their creative potential and the bliss of flow in order to find ways to help people ­ alf-​­light of creative to better endure the dark but also find and delight in the light and the h exploration.

Notes 1 This is a reference to a quotation from the poem ‘­He Wishes for the Cloths of Heaven’ by William Butler Yeats (­1899, line 4, as cited in Ricks, 1987, ­p. 568). 2 Note: Acknowledging the complexity and the debate around definitions of affect versus emotion (­Feldman Barrett, 2017; Ketai, 1975), for ease, these terms will be used here somewhat interchangeably as they both refer to discreet, ­short-​­term, subjectively experienced feelings, which it is proposed may impact on cognition similarly.

References Allen, A. P., & Thomas, K. E. (­2011). A dual process account of creative thinking. Creativity Research Journal, 23(­2), ­109–​­118. https://­doi.org/­10.1080/­10400419.2011.571183

227

Genevieve M. Cseh Anderson, E. C., Carleton, R. N., Diefenbach, M., & Han, P. K. J. (­2019). The relationship between uncertainty and affect. Frontiers in Psychology, 10, ­1–​­17. https://­doi.org/­10.3389/­f psyg.2019.02504 Arguel, A., Lockyer, L., Kennedy, G., Lodge, J. M., & Pachman, M. (­2018). Seeking optimal confusion: A review on epistemic emotion management in interactive digital learning environments. Interactive Learning Environments, 27(­2), ­1–​­11. https://­doi.org/­10.1080/­10494820.2018.1457544 Asakawa, K. (­2010). Flow experience, culture, and ­well-​­being: How do autotelic Japanese college students feel, behave, and think in their daily lives? Journal of Happiness Studies, 11, ­205–​­223. https://­ doi.org/­10.1007/­­s10902-­​­­0 08-­​­­9132-​­3 Baas, M., De Dreu, C. K. W., & Nijstad, B. A. (­2008). A ­meta-​­analysis of 25 years of ­mood-​­creativity research: Hedonic tone, activation, or regulatory focus? Psychological Bulletin, 134(­6), ­779–​­806. https://­doi.org/­10.1037/­a0012815 Baumann, N. (­2012). Autotelic personality. In S. Engeser (­Ed.), Advances in flow research (­­pp. ­165–​­186), Cham: Springer. https://­doi.org/­10.1007/­­978-­​­­1- ­​­­4614-­​­­2359-​­1_9 Baumann, N., & Kuhl, J. (­2005). Positive affect and flexibility: Overcoming the precedence of global over local processing of visual information. Motivation and Emotion, 29(­2), ­123–​­134. https://­doi. org/­10.1007/­­s11031-­​­­0 05-­​­­7957-​­1 Baumann, N., & Scheffer, D. (­2010). Seeing and mastering difficulty: The role of affective change in achievement flow. Cognition and Emotion, 24(­8), ­1304–​­1328. https://­doi.org/­10.1080/­02699930903319911 Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (­2001). Bad is stronger than good. Review of General Psychology, 5, ­323–​­370. https://­w ww.doi.org/­10.1037/­­1089-​­2680.5.4.323 Beghetto, R. A. (­2014). Creative mortification: An initial exploration. Psychology of Aesthetics, Creativity, and the Arts, 8(­3), ­266–​­276. https://­doi.org/­10.1037/­a0036618 Cacioppo, J. T., & Berntson, G. G. (­1999). The affect system: Architecture and operating characteristics. Current Directions in Psychological Science, 8, ­133–​­137. Calhoun, C. (­2017). On being content with imperfection. Ethics, 127, 3­ 27–​­352. Carleton, R. N. (­2016). Fear of the unknown: One fear to rule them all? Journal of Anxiety Disorders, 41, ­5 –​­21. https://­w ww.doi.org/­10.1016/­j.janxdis.2016.03.011 Chevrier, M., Muis, K. R., Trevors, G. J., Pekrun, R., & Sinatra, G. M. (­2019). Exploring the antecedents and consequences of epistemic emotions. Learning and Instruction, 63, 101209. https://­doi. org/­10.1016/­j.learninstruc.2019.05.006 Cohn, M. A., & Fredrickson, B. L. (­2009). Positive emotions. In S. J. Lopez & C. R. Snyder (­Eds.), The Oxford handbook of positive psychology, 2nd ed. (­­pp. ­777–​­796) New York, NY: The Guilford Press. https://­doi.org/­10.1093/­oxfordhb/­9780195187243.013.0003 Cordaro, D. T., Brackett, M., Glass, L.,  & Anderson, C. L. (­2016). Contentment: Perceived completeness across cultures and traditions. Review of General Psychology, 20(­3), 2­ 21–​­235. https://­doi. org/­10.1037/­g pr0000082 Cowen, A. S., & Keltner, D. (­2017). ­Self-​­report captures 27 distinct categories of emotion bridged by continuous gradients. Proceedings of the National Academy of Sciences of the United States of America, 114(­38), ­E7900–​­E7909. https://­doi.org/­10.1073/­pnas.1702247114n Cseh, G. M. (­2014). Flow in visual creativity. Unpublished Doctoral Dissertation. University of Aberdeen. Cseh, G. M. (­2016). Flow in creativity: A review of potential theoretical conflict. In L. Harmat, F. Ørsted Andersen, F. Ullén, J. Wright & G. Sadlo (­Eds.), Flow experience: Empirical research and applications (­­pp. ­79–​­94). Springer International. Cseh, G. M., Phillips, L. H., & Pearson, D. G. (­2015). Flow, affect, and visual creativity. Cognition and Emotion, 29(­2), ­281–​­291. https://­doi.org/­10.1080/­02699931.2014.913553 Csikszentmihalyi, M. (­ 1996). Creativity: Flow and the psychology of discovery and invention. Harper Perennial. Csikszentmihalyi, M. (­1997). Finding flow: The psychology of engagement with everyday life. BasicBooks HarperCollins. Csikszentmihalyi, M. (­2002). Flow: The classic work on how to achieve happiness (­Rev. ed.). London, England: Rider. (­Original published 1992) Dietrich, A. (­2004a). Neurocognitive mechanisms underlying the experience of flow. Consciousness and Cognition, 13(­4), ­746–​­761. https://­doi.org/­10.1016/­j.concog.2004.07.002 Dietrich, A. (­2004b). The cognitive neuroscience of creativity. Psychonomic Bulletin & Review, 11(­6), ­1011–​­1026. Dweck, C. S. (­2017). Mindset: Changing the way you think to fulfil your potential (­updated ed.). Robinson. (­Original published 2006)

228

Of Night and Light and the Half-Light Ekman, P. (­1992). An argument for basic emotions. Cognition and Emotion, 6(­3/­4), ­169–​­200. https://­doi. org/­10.1080/­02699939208411068 Ellsworth, P. C., & Smith, C. A. (­1988). Shades of joy: Patterns of appraisal differentiating pleasant emotions. Cognition and Emotion, 2(­4), 3­ 01–​­331. https://­doi.org/­10.1080/­02699938808412702 Fancourt, D., Garnett, C., Spiro, N., West, R., & Müllensiefen, D. (­2019). How do artistic creative activities regulate our emotions? Validation of the emotion regulation strategies for artistic creative activities scale (­­ERS-​­ACA). PLoS One, 14(­2), ­1–​­22. https://­doi.org/­10.1371/­journal.pone.0211362 Fang, T. (­2012). Yin yang: A new perspective on culture. Management and Organization Review, 8(­1), ­25–​­50. https://­doi.org/­10.1111/­j.­1740-​­8784.2011.00221.x Farmer, S. M., & Tierney, P. (­2017). Considering creative ­self-​­efficacy: Its current state and ideas for future inquiry. In M. Karwowski & J. C. Kaufman (­Eds.), The creative self (­­pp. ­23–​­47). New York, NY: Academic Press. Feldman Barrett, L. (­2017). How emotions are made: The secret life of the brain. Pan Macmillan. Festinger, L., & Carlsmith, J. M. (­1959). Cognitive consequences of forced compliance. The Journal of Abnormal and Social Psychology, 58(­2), ­203–​­210. https://­doi.org/­10.1037/­h0041593 Fong, C. T. (­2006). The effects of emotional ambivalence on creativity. Academy of Management Journal, 49(­5), ­1016–​­1030. https://­doi.org/­10.5465/­a mj.2006.22798182 Forgas, J. P. (­2014). On the downside of feeling good: Evidence for the motivational, cognitive and behavioural disadvantages of positive affect. In J. Gruber & J. T. Moskowitz (­Eds.), Positive emotion: Integrating the light sides and the dark sides (­­pp. ­301–​­322). Oxford: Oxford University Press. Fredrickson, B. (­1998). What good are positive emotions? Review of General Psychology, 2(­3), ­300–​­319. Fredrickson, B. L. (­2013). Updated thinking on positivity ratios. American Psychologist, 68(­9), ­814–​­822. https://­doi.org/­10.1037/­a0033584 Fredrickson, B. L., & Branigan, C. (­2005). Positive emotions broaden the scope of attention and ­thought-​ ­action repertoires. Cognition and Emotion, 19(­3), ­313–​­332. https://­doi.org/­10.1080/­02699930441000238 Frijda, N. (­1986). The emotions. Cambridge University Press. Furnham, A., & Marks, J. (­2013). Tolerance of ambiguity: A review of the recent literature. Psychology, 04(­09), ­717–​­728. https://­doi.org/­10.4236/­psych.2013.49102 Gable, S. L., & Haidt, J. (­2005). What (­a nd why) is positive psychology? Review of General Psychology, 9(­2), ­103–​­110. https://­doi.org/­10.1037/­­1089-​­2680.9.2.103 ­ armon-​­Jones, E. (­2010). The motivational dimensional model of affect: Implications for Gable, P., & H breadth of attention, memory, and cognitive categorisation. Cognition and Emotion, 24(­2), 3­ 22–​­337. https://­doi.org/­10.1080/­02699930903378305 Gaudreau, P., Sanchez, X.,  & Blondin, J.-​­ P. (­ 2006). Positive and negative affective states in a ­performance-​­related setting: Testing the factorial structure of the Panas across two samples of ­French-​­Canadian participants. European Journal of Psychological Assessment, 22, ­240–​­249. Getzels, J. W., & Csikszentmihalyi, M. (­1976). The creative vision: A longitudinal study of problem finding in art. Wiley. Ghiselin, B. (­Ed.) (­1952). The creative process: A symposium. University of California Berkeley. Gilbert, D. T., & Wilson, T. D. (­2007). Prospection: Experiencing the future. Science, 317(­5843), ­1351–​ ­1354. https://­doi.org/­10.1126/­science.1144161 Grenier, S., Barrette, A.-​­M., & Ladouceur, R. (­2005). Intolerance of uncertainty and intolerance of ambiguity: Similarities and differences. Personality and Individual Differences, 39, 5­ 93–​­600. https://­ doi.org/­10.1016/­j.paid.2005.02.014 Griswold, C. (­1996). Happiness, tranquillity, and philosophy. Critical Review, 10, 1­ –​­32. https://­doi.org/­ 10.1080/­0 8913819608443407 Gruber, J., & Moskowitz, J. T. (­Eds.) (­2014). Positive emotion: Integrating the light sides and the dark sides. Oxford. Guilford, J. P. (­1970). Traits of creativity. In P. E. Vernon (­Ed.), Creativity: Selected readings (­­pp. ­167–​­188). Penguin. Hanin, Y. L. (­1997). Emotions and athletic performance: Individual zones of optimal functioning model. European Yearbook of Sport Psychology, 1, ­29–​­72. Hanin, Y. L. (­2000). Emotions in sport. Human Kinetics. Herbert, D. (­2022). How do organisations facilitate creativity through embracing polarity, paradox and tension? Unpublished Doctoral Dissertation. Buckinghamshire New University. Hoemann, K., Gendron, M., & Feldman Barrett, L. (­2017). Mixed emotions in the predictive brain. Current Opinion in Behavioral Sciences, 15, 5­ 1–​­57. https://­doi.org/­10.1016/­j.cobeha.2017.05.013

229

Genevieve M. Cseh Icekson, T., Roskes, M., & Moran, S. (­2014). Effects of optimism on creativity under approach and avoidance motivation. Frontiers in Human Neuroscience, 8(­A rticle 105). https://­doi.org/­10.3389/­f nhum. 2014.00105 Isen, A. M. (­1987). Positive affect, cognitive processes, and social behavior. Advances in Experimental Social Psychology, 20, ­203–​­253. Isen, A. M., & Daubman, K. A. (­1984). The influence of affect on categorization. Journal of Personality and Social Psychology, 47, ­1206–​­1217. Isen, A. M., Daubman, K. A.,  & Nowicki, G. P. (­1987). Positive affect facilitates creative problem solving. Journal of Personality and Social Psychology, 52, ­1122–​­1131. Isen, A. M., Johnson, M. M. S., Mertz, E., & Robinson, G. E (­1985). The influence of positive affect on the unusualness of word associations. Journal of Personality and Social Psychology, 48, ­1413–​­1426. Ivcevic, Z., & Brackett, M. A. (­2015). Predicting creativity: Interactive effects of openness to experience and emotion regulation ability. Psychology of Aesthetics, Creativity, and the Arts, 9(­4), ­480–​­487. https://­doi.org/­10.1037/­a0039826 Ivtzan, I., Lomas, T., Hefferon, K., & Worth, P. (­Eds.) (­2016). Second wave positive psychology: Embracing the dark side of life. Routledge. Jamison, K. R. (­2011). Great wits and madness: More near allied? The British Journal of Psychiatry: The Journal of Mental Science, 199, ­351–​­352. https://­doi.org/­10.1192/­bjp.bp.111.100586 Kahneman, D. (­2011). Thinking, fast and slow. Farrar, Straus and Giroux. Kashdan, T., & ­Biswas-​­Diener, R. (­2015). The power of negative emotion. Oneworld. Katahira, K., Yamazaki, Y., Yamaoka, C., Ozaki, H. Nakagawa, S., & Nagata, N. (­2018). EEG correlates of the flow state: A combination of increased frontal theta and moderate frontocentral alpha rhythm in the mental arithmetic task. Frontiers in Psychology, 9, ­1–​­11. Retrieved from https://­w ww. frontiersin.org/­a rticles/­10.3389/­f psyg.2018.00300/­f ull Kaufmann, G.,  & Vosberg, S. K. (­1997). ‘­Paradoxical’ mood effects on creative ­problem-​­solving. Cognition and Emotion, 11(­2), ­151–​­170. https://­doi.org/­10.1080/­026999397379971 Keller, J., Bless, H., Blomann, F., & Kleinböhl, D. (­2011). Physiological aspects of flow experiences: ­Skills-­​­­demand-​­compatibility effecs on heart rate variability and salivary cortisol. Journal of Experimental Social Psychology, 47, ­849–​­852. Ketai, R. (­1975). Affect, mood, emotion and feeling: Semantic considerations. American Journal of Psychiatry, 132(­11), ­1215–​­1217. https://­doi.org/­10.1176/­ajp.132.11.1215 Killgore, W. D. S. (­2000). Evidence for a third factor on the positive and negative affect schedule in a college student sample. Perceptual and Motor Skills, 90,­147–​­152. https://­doi.org/­10.2466/­PMS.90.1.­ 147-​­152 Kim, K. H. (­2019). Demystifying creativity: What creativity isn’t and is? Roeper Review, 41, ­119–​­128. https://­doi.org/­10.1080/­02783193.2019.1585397 Landhäußer, A.,  & Keller, J. (­2012). Flow and its affective, cognitive, and p­ erformance-​­related consequences. In S. Engeser (­Ed.), Advances in flow research (­­pp.  ­65–​­86). Springer. https://­doi. org/­10.1007/­­978-­​­­1-­​­­4614-­​­­2359-​­1_4 Larsen, J. T. (­2017). Holes in the case for mixed emotions. Emotion Review, 9(­2), ­118–​­123. https://­doi. org/­10.1177/­1754073916639662 Larsen, J. T., & McGraw, A. P. (­2011). Further evidence for mixed emotions. Journal of Personality and Social Psychology, 100(­6), ­1095–​­1110. https://­w ww.doi.org/­10.1037/­a0021846 Larsen, J. T., & McGraw, A. P. (­2014). The case for mixed emotions. Social and Personality Psychology Compass, 8(­6), ­263–​­274. https://­doi.org/­10.1111/­spc3.12108 Latinjak, A. T. (­2012). The underlying structure of emotions: A ­tri-​­dimensional model of core affect and emotion concepts for sports. Revista Iberoamericana de Psicologia del Ejercicio y el Deporte, 7(­1), ­71–​­87. Leung, A. K. Y., Liou, S., Qiu, L., Kwan, L. Y. Y., Chiu, C. Y., & Yong, J. C. (­2014). The role of instrumental emotion regulation in the e­ motions-​­creativity link: How worries render individuals with high neuroticism more creative. Emotion, 14(­5), ­846– ​­856. https://­doi.org/­10.1037/­a0036965 Lomas. T., Waters, L., Williams, P., Oades, L. G., Kern, M. L. (­2020). Third wave positive psychology: Broadening towards complexity. The Journal of Positive Psychology, 16(­5), ­660–​­674. https://­doi. org/­10.1080/­17439760.2020.1805501 Martin, L. L., Ward, D. W., Achee, J. W., & Wyer, R. S., Jr. (­1993). Mood as input: People have to interpret the motivational implications of their moods. Journal of Personality and Social Psychology, 64(­3), ­317–​­326. https://­doi.org/­10.1037//­­0 022-​­3514.64.3.317

230

Of Night and Light and the Half-Light Massimini, F., Csikszentmihályi, M., & Carli, M. (­1987). The monitoring of optimal experience: A tool for psychiatric rehabilitation. Journal of Nervous and Mental Disease, 175(­9), 5­ 45–​­549. https://­ doi.org/­10.1097/­­0 0005053-­​­­198709000-​­0 0006 Massimini, F., Csíkszentmihályi, M., & Delle Fave, A. (­1988). Flow and biocultural evolution. In M. Csíkszentmihályi & I. S. Csíkszentmihályi (­Eds.), Optimal experience: Psychological studies of flow in consciousness (­­pp. ­60–​­81). Cambridge University Press. Mednick, S. A. (­1962). The associative basis of the creative process. Psychological Review, 69, ­220–​­232. Merrotsy, P. (­2013). Tolerance of ambiguity: A trait of the creative personality? Creativity Research Journal, 25(­2), ­232–​­237. https://­doi.org/­10.1080/­10400419.2013.783762 Mraz, W., & Runco, M. A. (­1994). Suicide ideation and creative problem solving. Suicide and ­Life-​ ­T hreatening Behavior, 24, ­38–​­47. Muis, K. R., Chevrier, M., & Singh, C. A. (­2018). The role of epistemic emotions in personal epistemology and ­self-​­regulated learning. Educational Psychologist, 53(­3), ­165–​­184. https://­doi.org/­10.1080/­0 0461520. 2017.1421465 Mumford, M. D. (­2003). Where have we been, where are we going? Taking stock in creativity research. Creativity Research Journal, 15(­­2 –​­3), ­107–​­120. https://­doi.org/­10.1080/­10400419.2003.9651403 ­ elf-​­compassion: The proven power of being kind to yourself. William Morrow. Neff, K. (­2011). S Oliverio, A. (­2008). Brain and creativity. Progress of Theoretical Physics Supplement, 173, ­66–​­78. https://­ doi.org/­10.1143/­P TPS.173.66 Ortony, A., & Turner, T. J. (­1990). What’s basic about basic emotions? Psychological Review, 97, ­315–​­331. Pawelski, J. O. (­2015). Defining the “­positive” in positive psychology: Part I. A descriptive analysis. The Journal of Positive Psychology, 11(­4), ­339–​­356. https://­doi.org/­10.1080/­17439760.2015.1137627 Pawelski, J. O. (­2016). Defining the ‘­positive’ in positive psychology: Part II. A normative analysis. The Journal of Positive Psychology, 11(­4), ­357–​­365. https://­doi.org/­10.1080/­17439760.2015.1137628 Peifer, C. (­2012). Psychophsiological correlates of ­flow-​­experience. In S. Engeser (­Ed.), Advances in flow research (­­pp. ­139–​­164). Springer. https://­doi.org/­10.1007/­­978-­​­­1-­​­­4614-­​­­2359-​­1_8 Pekrun, R. (­2006). The ­control-​­value theory of achievement emotions: Assumptions, corollaries, and implications for educational research and practice. Educational Psychology Review, 18(­4), ­315–​­341. https://­doi.org/­10.1007/­­s10648-­​­­0 06-­​­­9 029-​­9 Rhodes, M. (­1961). An analysis of creativity. The Phi Delta Kappan, 42(­7 ), ­305–​­310. http://­w ww.jstor. org/­stable/­20342603 Ricks, C. (­Ed.) (­1987). The new Oxford book of Victorian verse. Oxford University Press. Robinson, D. L. (­2008). Brain function, emotional experience and personality. Netherlands Journal of Psychology, 64(­4), ­152–​­168. https://­doi.org/­10.1007/­bf03076418 Rogatko, T. P. (­2007). The influence of flow on positive affect in college students. Journal of Happiness Studies, 10(­A rticle 33), 1­ 33–​­148. https://­doi.org/­10.1007/­­s10902-­​­­0 07-­​­­9069-​­y Runco, M. A. (­2014). Personality and motivation: Tolerance of ambiguity. In M. A. Runco (­Ed.), Creativity: Theories and themes: Research, development, and practice (­2nd ed., ­pp.  ­265–​­302). Elsevier Academic Press. Runco, M. A., & Jaeger, G. J. (­2012). The standard definition of creativity. Creativity Research Journal, 24(­1), ­92–​­96. https://­doi.org/­10.1080/­10400419.2012.650092 Russell, J. A. (­1980). A circumplex model of affect. Journal of Personality and Social Psychology, 39, ­1161–​­1178. Sawyer, R. K. (­2006). Explaining creativity: The science of human innovation. Oxford University Press. Scherer, K. R. (­1998). Analyzing emotion blends. In Proceedings of the 10th Conference ofthe International Society for Research on Emotions (­­pp. ­142–​­148). Wurzburg, Germany: ISRE Publications. Schüler, J. (­2012).The dark side of the moon. In S. Engeser (­Ed.), Advances in flow research (­­pp. ­123–​­137). Springer. https://­doi.org/­10.1007/­­978-­​­­1- ­​­­4614-­​­­2359-​­1_7 Seligman, M. E. P., & Csikszentmihalyi, M. (­2000). Positive psychology: An introduction. American Psychologist, 55(­1), ­5 –​­14. https://­doi.org/­10.1037/­­0 003-​­066X.55.1.5 Simonton, D. K. (­2000). Creative development as acquired expertise: Theoretical issues and an empirical test. Developmental Review, 20(­2), ­283–​­318. https://­w ww.doi.org/­10.1006/­d rev.1999.0504 Tran, V. (­2013). Positive affect negative affect scale (­PANAS). In M. D. Gellman & J. R. Turner (­Eds.), Encyclopedia of behavioral medicine. Springer. https://­doi.org/­10.1007/­­978-­​­­1- ­​­­4 419-­​­­1005-​­9_978 Ullén, F., de Manzano, Ö., Theorell, T.,  & Harmat, L. (­2013). The physiology of effortless attention: Correlates of state flow and flow proneness. Effortless Attention, (­April), ­205–​­218. https://­doi. org/­10.7551/­m itpress/­9780262013840.003.0011

231

Genevieve M. Cseh Ulrich, M., Keller, J., Hoenig, K., Waller, C., & Grön, G. (­2013). Neural correlates of experimentally induced flow experiences. NeuroImage, 86, ­194–​­202. https://­doi.org/­10.1016/­j.neuroimage. 2013.08.019 Vago, D. R.,  & Silbersweig, D. A. (­2012, 25 October). S­ elf-​­awareness, ­self-​­regulation, and s­elf-​ ­ t ranscendence (­­ S-​­ ART): A framework for understanding the neurobiological mechanisms of mindfulness. Frontiers in Human Neuroscience, 6(­A rticle 296), 1­ –​­30. https://­doi.org/­10.3389/­f nhum. 2012.00296 Vogl, E., Pekrun, R., Murayama, K., Loderer, K., & Schubert, S. (­2019). Surprise, curiosity, and confusion promote knowledge exploration: Evidence for robust effects of epistemic emotions. Frontiers in Psychology, 10, ­1–​­16. https://­doi.org/­10.3389/­f psyg.2019.02474 Watson, D., Clark, L. A.,  & Tellegen, A. (­1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54(­6), ­1063–​­1070. https://­doi.org/­10.1037/­­0 022-​­3514.54.6.1063 Williams, P.,  & Aaker, J. L. (­2002). Can mixed emotions peacefully coexist? Journal of Consumer Research, 28(­4), ­636–​­649. https://­doi.org/­10.2139/­ssrn.945468 Wong, P. T. P. (­2011). Positive psychology 2.0: Towards a balanced interactive model of the good life. Canadian Psychology, 52(­2), ­69– ​­81. https://­doi.org/­10.1037/­a0022511 Wundt, W. (­1897). Outline of psychology (­C. H. Judd, Trans.). Williams and Norgate; Wilhelm Engelmann. https://­doi.org/­10.1037/­­12908- ​­0 00 Yeh, Y.-C.,  & Li, M.-L. (­2008). Age, emotion regulation strategies, temperament, creative drama, and preschoolers’ creativity. The Journal of Creative Behavior, 42, ­131–​­149. https://­doi.org/­10.1002/­ j.­2162- ​­6057.2008.tb01291.x Zenasni, F., Besançon, M., & Lubart, T. (­2008). Creativity and tolerance of ambiguity: An empirical study. The Journal of Creative Behavior, 42(­1), ­61–​­73. https://­doi.org/­10.1002/­j.­2162-​­6057.2008. tb01080.x

232

15 ­PROBLEM-​­SOLVING, COLLABORATION, AND CREATIVITY Jennifer Wiley and Olga Goldenberg

­Introduction Creative cognition is required when we need novel approaches and n ­ on-​­obvious solutions to problems. Creative cognition requires a delicate balance between the old and the new. New thoughts are inextricably linked to our existing knowledge (­cf. Malafouris, 2020). After all, we must use our ­long-​­term memory to interpret the situation and generate possible solutions. But, importantly, we need to move beyond routine responses and depart from the norm to achieve the n ­ on-​­obvious. Examples of past solutions or existing approaches could provide a starting point for innovations, but at the same time, they could also fixate us on inappropriate, n ­ on-​­viable, or s­ub-​­optimal solutions. We could use the opportunity to mine our past experiences for analogous cases that might apply to the present situation, or we could be stymied by mental sets and an inability to escape the solutions that are most strongly activated as we first approach a problem. In many situations, relying on past experience helps us to work productively toward a solution. But in situations where ­non-​­obvious associations are required, we often become stuck in our initial attempts to solve a problem and are unable to restructure our representations in order to perceive the most elegant or insightful solutions. Creative cognition also requires a delicate balance between divergent and convergent processes. All creative p­ roblem-​­solving involves some element of convergence as solutions are evaluated against the goals for the task. Yet, when n ­ on-​­obvious solutions are required, the distinct challenge that most clearly separates creative p­ roblem-​­solving from more analytic forms of ­problem-​­solving is the divergent thinking that allows us to access and select remote or unlikely responses. Although some creative ­problem-​­solving tasks, such as the Remote Associates Task (­R AT), are often referred to as “­convergent”, this misses the whole point of the essential creative process that these tasks were designed to c­ apture – ​­the ability to find remote, novel, or unusual responses. In Mednick’s (­1962) terms, the key to the creative process was to be able to access candidate solutions that are unlikely to come to mind. Thus, the RAT was designed to assess the ability to access uncommon associations. If one considers the relative frequency of different meanings or senses of words, then one can imagine words that have a “­steep” hierarchy where the most frequent meaning is highly accessible and infrequent meanings are less accessible. Mednick explored whether some individuals might have better access to the normatively less frequent uses of words because they had “­flatter” DOI: 10.4324/9781003009351-17

233

Jennifer Wiley and Olga Goldenberg

hierarchies, which increases the likelihood of the uncommon associates coming to mind. In subsequent research, the task has been used more generally to explore the conditions under which we can find atypical or unusual responses and to understand the conditions that prevent us from doing so (­Smith & Blankenship, 1991), while recognizing that RAT problems may sometimes be solved using more analytic approaches (­Bowden & ­Jung-​­Beeman, 2003). Other tasks, like the Alternate Uses Task, more transparently emphasize the idea generation phase and the process of sampling ideas from memory. Studies using idea generation paradigms, including those using the Torrance Tests of Creative Thinking (­TTCT; Torrance, 2008) or similar Unusual Uses Tasks (­Guilford, 1956; Wallach & Kogan, 1965), tend to focus on an individual’s ability to generate creative ideas by asking participants to generate novel, unusual, or alternate uses for everyday objects, such as a brick or a paperclip. Because the main measures that are considered are often just the number of responses that are generated (­rather than a metric that attempts to assess the quality of the ideas that are generated), and because of recent developments in the ability to use automated methods to score these types of results, divergent thinking and idea generation tasks have been one area of research that has seen an explosion of interest (­Beaty & Silvia, 2012). These tasks also generally align with the brainstorming tasks that have received the most attention in collaborative p­ roblem-​ ­solving research.

Obstacles to Creative ­Problem-​­Solving (­and How Collaboration May Help) Much of today’s research on creative p­ roblem-​­solving has its roots in early Gestalt work on human thinking. Wertheimer’s (­1945) work explicates early interests in “­productive thinking” or creative thinking rather than “­reproductive thinking”. In Wertheimer’s terms, in “­reproductive thinking”, solutions or solution methods are retrieved directly from memory, based on routine experience, and applied blindly to problems as responses associated with particular stimuli. In contrast, in “­productive thinking”, solutions are generated only after engaging in attempts at understanding the essence of a problem, considering conceptual interrelations, and engaging in a kind of reorganization. Productive thinking involves going from a situation of bewilderment or confusion to a new state of clarity, in which everything makes sense and fits together. It is an ­insight-​­based reasoning process that involves a shift in representation or interpretation. The Gestaltists, including Wertheimer, Duncker, Katona, Luchins, and Maier, explored how prior experience or practice with a particular approach could lead to mental set and how prior knowledge of objects and their usual uses could lead to “­functional fixedness” on those uses. Functional fixedness refers to the difficulty of thinking of new functions for an object other than one shaped by our everyday knowledge or experience with the object, as was demonstrated in classic studies showing the inability to see a box as useful as something other than a container or pliers as a weight rather than a gripping device (­Adamson, 1952; Duncker, 1945; Maier, 1931). Further, they found that the way problems were presented could impact the likelihood of finding a ­non-​­obvious solution. In all these cases, individuals were often fixated by their initial attempts and would hit an impasse. To reach a solution, what was needed was a restructuring in the perception or representation of the problem, or an insight. These themes continue in more contemporary research on insight and creative ­problem-​ s­olving in individuals. The most central premise is that the solver needs to avoid fixation or mental set imposed by recently primed information, by their own initial responses, and by routine solutions from their prior knowledge (­Smith, 1995, 2003; Wiley, 1998). Some benefits have been seen from incubation or taking a break from actively attempting to solve 234

Problem-Solving, Collaboration, and Creativity

a problem (­Sio & Ormerod, 2009). One candidate explanation for incubation effects focuses on the benefits that come from spreading activation in memory, often below the threshold of awareness (­Gilhooly, 2016; Smith & Kounios, 1996). Other explanations focus on how this delay can allow for the decay or forgetting of initially activated misleading information and responses (­Koppel & Storm, 2014; Smith & Blankenship, 1989). Similarly, benefits have been seen from a context change between solution attempts or from engaging in task switching (­George & Wiley, 2019; Lu et al., 2017; Smith et al., 2017). Further, being in a more diffuse attentional state can also allow for remote or n ­ on-​­obvious solutions to emerge ( ­Jarosz et al., 2012; Wieth & Zacks, 2011; Wiley & Jarosz, 2012). In each case, the proposed mechanism involves decreased activation or accessibility of initial or typical responses, allowing for a relative increase in the accessibility of weak associates. Another way that weak associates may become activated is by incidental hints or cues coming from the environment (­Maier, 1931). And, although being in a state of impasse seems like a negative event, it can make people more open to these hints as failure indices are created that allow for opportunistic assimilation of hints or cues (­Gick & McGarry, 1992; Seifert et al., 1995). Just as when hints given as part of a secondary task can be helpful after individual solvers have reached an impasse (­Moss et al., 2011), individuals may be able to benefit from hints provided by others when working in a group (­Hoffman et al., 1962; Litchfield & Ball, 2011; Maier, 1967). One key to the intuitive suggestion that groups should be more effective, flexible, and innovative at ­problem-​­solving than individuals is the assumption that each group member brings to the task a slightly different set of ­task-​­relevant knowledge and skills. Through discussion, the knowledge and skills of each member can become available for all, giving each member a larger pool of ideas to draw from. Especially if members possess different backgrounds, group ­problem-​­solving will give people a greater opportunity for novel associations, strategies, and operations. Exposure to diverse viewpoints may increase both the quantity and quality of idea generation in a group context as members with different ­task-​ ­relevant knowledge can stimulate each other’s thinking by priming of ideas that otherwise have low accessibility (­Brown et al., 1998). Many have proposed that there should be an advantage to working in diverse groups because it allows for cognitive stimulation and possible synergy from multiple perspectives, and this suggestion has received a great deal of attention in research on creative ­problem-​­solving in groups.

Brainstorming in Groups A popular technique that has been frequently used in group creativity research is brainstorming (­Diehl  & Stroebe, 1987; Goldenberg  & Wiley, 2011; Paulus  & Nijstad, 2003). First encapsulated by Osborn (­1953) in his influential book Applied Imagination: Principles and Procedures of Creative Problem-Solving, brainstorming is a formal approach to idea generation in which participants are given rules designed to help them capitalize on the cognitive stimulation that is presumed to occur naturally when people work collectively to generate solutions to a problem. Osborn asserted that the key to a successful brainstorming exercise is quantity, quantity, and more quantity. He based his approach on the assumption that ­better-​­quality ideas are produced later rather than earlier in the brainstorming session. His reasoning was that we first need to rid our minds of common, ordinary ideas before we can uncover or think up more original, creative solutions to a problem. As such, generating a greater number of ideas increases the likelihood that g­ ood-​­quality ideas will be embedded among them. These intuitions have been later corroborated by empirical evidence of significant positive correlations between the total number of ideas and the number of highly 235

Jennifer Wiley and Olga Goldenberg

original and practical ideas, although more contemporary approaches explain this result as more of an artifact of probability in sampling or an ­equal-​­odds rule, such that larger samples of ideas will necessarily contain more creative responses (­Diehl & Stroebe, 1987, 1991; Simonton, 1997). Although Osborn originally offered more than 20 different recommendations for how groups ought to be directed to maximize their ideational productivity, research has typically employed just the following four (­Goldenberg & Wiley, 2011; Paulus & Brown, 2007): criticism is ruled out (­w ithhold judgment), f­ree-​­wheeling is welcomed (­the wilder the better, it is easier to tame down than to think up), quantity is desired (­the more ideas the better, the higher the likelihood of winners), and combination and improvement are sought (­suggest how ideas of others can be turned into better ideas, or how two or more ideas can be joined into still another idea). Osborn claimed that if these rules are followed, the average person can come up with twice as many ideas when working with a group than when working alone. This bold claim has prompted decades of empirical research that has attempted to assess the actual benefits of collaboration. An initial study by Taylor et al. (­1958) both provided key empirical insights as well as methods for assessing the effectiveness of group interactions (­see also Lorge  & Solomon, 1955). In this study, subjects were asked to brainstorm either individually or in ­four-​­person groups on a series of three problems for a period of 12 minutes each. The three problems were: 1 Each year, a great many American tourists go to visit Europe. But now, suppose that our country wished to get many more European tourists to come to visit America during their vacations. What steps can you suggest that would get more European tourists to come to this country? 2 We don’t think this is very likely to happen, but imagine for a moment what would happen if everyone born after 1960 had an extra thumb on each hand… What practical benefits or difficulties will arise when people start having this extra thumb? 3 Because of the rapidly increasing birth rate that began in the 1940s, it is now clear that by 1970, public school enrollment will be very much greater than it is today. In fact, it has been estimated that if the s­tudent-​­teacher ratio were to be maintained at what it is today, 50% of all individuals graduating from college would have to be induced to enter teaching. What different steps might be taken to ensure that schools will continue to provide instruction at least equal in effectiveness to that now provided? To allow for a statistical comparison between results from individuals and group sessions, nominal groups were formed from participants who had brainstormed individually. The logic behind creating nominal groups is to equate the units of analysis based on the number of individuals, and therefore what differs between the units is whether the individuals worked alone or were able to interact during idea generation. For each unit of four individuals, the set of different (­­non-​­redundant) ideas offered across individuals within the group (­or nominal group) was used as the basis for the comparison. Any benefits in the real groups compared to the nominal groups in the quantity or quality of responses could then be attributed to interaction or collaboration processes. Although the interacting groups generated more ideas than any one individual, the more important finding was that the interacting groups generated half the number of different ideas as the nominal groups. A second analysis considered just the subset of ideas that were generated by only one group (­or nominal group). These unique ideas represented the most 236

Problem-Solving, Collaboration, and Creativity

remote, novel, or rare responses, and they were 50% more likely in the nominal groups than in the real groups. Both of these two outcome measures are quantitative counts of the number of ideas that were generated. At least in principle, a group could come up with fewer ideas total, but if many of them are highly creative, then that could also be evidence of the benefit of brainstorming on idea generation. However, there was no support for this either. The ideas generated by the nominal groups were still more likely to be those as rated higher in quality on three different dimensions (­feasibility/­plausibility, effectiveness/­significance, and generality). Contrary to Osborn’s claim, Taylor et al. found no benefit from interactive brainstorming (­in fact, the opposite). These findings provided an initial demonstration of the ineffectiveness of small group brainstorming relative to individual creative ideation, which has since been replicated many times. Interacting groups typically came up with fewer ideas than an equal number of independently working individuals (­nominal groups; Larson, 2010; Mullen et al., 1991). Many of these empirical studies of group brainstorming focused just on the total number of generated ideas, accepting Osborn’s assertion that quantity translates into quality. This ubiquitous superiority of nominal groups in terms of the total number of generated ideas relative to their interacting group counterparts has been termed “­productivity loss” or “­process loss” (­Steiner, 1972). Despite theoretical accounts that suggest that group brainstorming should lead to cognitive stimulation (­Brown et al., 1998; Nijstad & Stroebe, 2006; Paulus & Brown, 2003), synergy (­Larson, 2010), or “­process gain” (­Steiner, 1972), typically few empirical studies are able to demonstrate group advantages, and most studies find evidence for process loss. Researchers have of course been fascinated by the puzzle posed by these findings and searching for the conditions that might be responsible for these outcomes. Historically, a number of explanations have been offered for this unexpected result, including performance matching, where individuals produce only as much as others and do not attempt to maximize their performance (­Brown & Paulus, 1996; Camacho & Paulus, 1995; Paulus & Dzindolet, 1993), or decreases in effort due to a loss of motivation, social loafing, and ­f ree-​­riding effects, where individuals might be less motivated and contribute less effort, which may be especially the case when they do not perceive accountability (­Kerr & Bruun, 1983; Sheppard, 1993; Williams et al., 2003). Another set of proposals includes production blocking and the cognitive interference that results from others’ ideas (­Diehl & Stroebe, 1991; Nijstad & Stroebe, 2006). Further, other work suggests that process loss may result from the preference for discussing only information that is shared by a majority of group members. A long tradition of research on the hidden profile paradigm (­where each member is intentionally provided with unique information that is not shared by other members) shows that groups are usually unsuccessful in capitalizing on the diverse and unique cognitive resources potentially available to them (­Sohrab et al., 2015). It is well documented that groups will disproportionately favor the sampling and discussion of shared information compared with the unshared, unique information that can be critical for making an optimal decision or selecting an optimal solution approach (­Gigone & Hastie, 1993; Stasser & Titus, 1985). Members who communicate shared information receive more positive evaluations from other members for doing so; they feel better about their own task knowledge when another member’s view reinforces their own; and they evaluate each other as more competent, knowledgeable, and credible (­Wittenbaum et al., 1999, 2004). This has been referred to as the mutual enhancement effect, in which the discussion of shared information is positively reinforced. While discussing only shared information may ease the interaction by helping members relate to one another, one can see how this effect can also contribute to process loss by decreasing the likelihood that 237

Jennifer Wiley and Olga Goldenberg

unique perspectives are contributed. Finally, some newer suggestions consider the role that the nature of the problem might play, the goals that are instilled by the set of rules that are given, some possible benefits that might come from incubation periods or interleaving individual and group phases, and from dissent and diversity among the members of the group.

Production Blocking, Brainwriting, and Electronic Brainstorming During collaborative discussions, each individual needs to juggle multiple things simultaneously: they need to keep any idea they would like to contribute in memory while paying attention to the current speaker, keeping track of the ideas that have been said already, and monitoring the discussion for an appropriate time to add their two cents. Diehl and Stroebe (­1987, 1991) demonstrated that performing these cognitive tasks simultaneously can lead to forgetting the ideas that the person wanted to contribute and also make it difficult to generate additional new ideas. Group members in nominal groups do not have to worry about waiting for a turn to speak or interruptions that can cause interference and distraction because they work individually (­H insz et al., 1997). These constraints imposed by the typical ­face-­​­­to-​­face collaboration process seem likely to be responsible for “­process loss”. In a landmark study, Paulus and Yang (­2000) departed from typical ­face-­​­­to-​­face discussion methods by using a “­brainwriting” methodology in which interacting groups were asked to record their ideas on small notecards, writing up to four ideas per card, and exchanged their ideas by passing these notecards among themselves in a ­round-​­robin fashion, reading one another’s ideas before writing a new idea of their own. This procedure reduces the opportunity for production blocking in interacting groups by allowing members to contribute ideas simultaneously without waiting for a speaking turn. Paulus and Yang (­2000) found that, when using this technique, ­four-​­person interacting groups generated 40% more ideas during a ­15-​­minute period than did nominal groups. Similar results have been reported with dyads by Coskun (­2005). These groundbreaking findings suggested that new methods based on brainwriting, such as rotational techniques that use a combination of words and sketches to represent ideas (­Linsey & Becker, 2010) and utilizing sticky notes in design ideation (­Ball et al., 2021), might be useful for exploiting the creative capabilities of interacting groups. Research studies of electronic brainstorming (­E BS) began to appear in the early 1990s (­Connolly et al., 1990; Dennis & Valacich, 1993, 1994; Valacich et al., 1992). In a typical EBS session, every group member utilizes what is now familiar as chatting or texting software. Participants can type their ideas at any time while still being able to read the ideas of others on the screen. Like brainwriting, EBS offered the promise of eliminating production blocking while still allowing for cognitive stimulation effects. In a m ­ eta-​­analysis of EBS studies, Dennis and Williams (­2005) found that EBS could be used to circumvent productivity losses. They also found effects of group size. EBS was superior to f­ ace-­​­­to-​­face group brainstorming, but only for groups with more than three members. It was also superior to nominal group brainstorming, but only for groups with more than nine members. They suggested that in a small ­face-­​­­to-​­face group, the benefits associated with reduced production blocking in EBS are outweighed by the fact that typing ideas is slower than speaking. (­One wonders if this issue is perhaps less acute in today’s ­text-​­based communication culture.) Meanwhile, they suggested that in very large EBS groups, the benefits of a large pool of ideas outweighed the detrimental effects of process loss. DeRosa et al. (­2007) conducted another ­meta-​­analytic review of EBS studies that also considered performance measures of idea quality in addition to quantity. For quality, they found that EBS groups outperformed ­face-­​­­to-​­face groups in both the number of ­h igh-​­quality responses and average idea quality. However, nominal groups performed as well as EBS groups. More recent EBS, brainwriting, and brainsketching studies 238

Problem-Solving, Collaboration, and Creativity

continue to investigate group size effects further with new asynchronous and hybrid formats (­Baruah & Paulus, 2019; Paulus et al., 2013). In general, although it is rare for “­process gain” to be observed, EBS formats are superior to traditional formats in terms of “­process loss”.

The Nature of the Problems Used in Brainstorming Tasks One relatively unexplored factor that may relate to process loss is the role of the nature of the problems given as the brainstorming task. As shown in ­Table 15.1, three types of prompts have been used in research on brainstorming: (­1) improvements to an object, place, or process, such as ways to improve a university (­Barki & Pinsonneault, 2001; Baruah & Paulus, 2008; DeDreu et al., 2008); (­2) consequences of hypothetical scenarios, such as advantages and disadvantages associated with having an extra thumb (­Camacho & Paulus, 1995; Dugosh et  al., 2000); and (­3) alternate uses for a common household object, such as a paperclip (­Paulus & Yang, 2000). Studies that assess divergent thinking in individuals, including those using the TTCT, tend to employ a variety or a combination of all three question types. In contrast, group brainstorming studies typically rely either on improvements or consequences of hypothetical scenarios questions. One exception is the widely cited study by Paulus and Yang (­2000) that was able to demonstrate process gain. One key to this result may have been their “­brainwriting” methodology. However, it is possible that another factor contributing to the high productivity of interacting groups was the use of an alternate uses question. ­Table 15.1  A Sample List of Brainstorming Tasks/­Topics/­Prompts used in Group Idea Generation Research, Organized by Task Type Problem/­P rompt

Sample Reference/­Publication

Alternate uses

Other uses for a paper clip Other uses for cars, SUVs, and/­or vans Unusual uses for a coat hanger, an A4 sheet of paper, a paper clip

Paulus and Yang (­2000) Goldenberg and Wiley (­2019) Breslin (­2019)

Improvements

Ways to increase tourism in your country/­city

Dennis and Valacich (­1993) Taylor et al. (­1958) Connolly et al. (­1990) Nemeth et al. (­2004) Goldenberg et al. (­2013) Kohn and Smith (­2011) Larey and Paulus (­1999) Lichtfield et al. (­2011) Paulus et al. (­2013) Goldenberg and Wiley (­2019) Diehl and Stroebe (­1989)

How to solve the parking problem/­t raffic in your city Ways to improve your university/­department/­ college experience

Ways to improve cars, SUVS, and/­or vans Ways to improve entertainment programs Hypothetical scenarios

Consequences of having an extra thumb

Consequences of requiring students to own a portable computer

239

Camacho and Paulus (­1995) Dugosh and Paulus (­2005) Korde and Paulus (­2017) Paulus and Dzindolet (­1993) Taylor et al. (­1958) Dennis and Valacich (­1994)

Jennifer Wiley and Olga Goldenberg

Alternate uses questions may be particularly conducive to allowing groups to capitalize on the creative potential of the different knowledge, mindsets, and perspectives that each member brings to a group. When asked to generate uses for an object, individuals will be likely to experience functional fixedness or mental blocks from thinking about traditional uses. If an individual working alone is likely to become “­stuck” and unable to come up with a new idea on a task, then this is precisely when group interaction may lead to facilitated ­problem-​­solving. Also, because the objective of the alternate uses task is to come up with various novel uses or ideas, each response requires the generation of new ideas, presumably through the combination of remote and previously unrelated concepts (­Mednick, 1962). Thus, many solutions to alternate uses problems are unlikely to be ­pre-​­formed in an individual’s l­ong-​­term memory prior to the brainstorming session. Exposure to others’ ideas could provide external stimulation that may help individuals generate new ideas and overcome fixation imposed by alternate uses prompts, which may help to balance out, and possibly even offset, other causes of process loss. In contrast to alternate uses prompts, questions that prompt participants to consider ways to improve a place, an object, or an organization (­i mprovements prompts), such as one’s university, may differ in both idea ­pre-​­formation and the likelihood of difficulties resulting from functional fixedness. Individuals may have already thought about potential improvements to their university or to common objects. As a result, for improvement questions, many candidate solutions may be available in ­long-​­term memory (­i.e., they are high in idea ­pre-​ ­formation) and need to be simply accessed and retrieved rather than generated. With these prompts, it may be less necessary to generate solutions using novel combinations of concepts and features. Further, solvers can base their improvements on their current knowledge about objects, meaning they do not need to restructure their understanding or overcome fixation due to prior knowledge to generate solutions for improvement prompts. Both high idea p­ re-​ ­formation and low likelihood of functional fixedness on the improvement brainstorming questions should lead to relatively less difficulty in generating solutions for individual brainstormers. As a result, there is less need for group input and fewer opportunities for cognitive stimulation from other people’s responses. Further, it is even possible that exposure to others’ ideas will be distracting because it can disrupt the natural flow of a group member’s ideas (­Nijstad et al., 2002) and can result in fixation on the ideas of others (­Smith, 2003). This could decrease the number and variety of ideas that groups generate (­Kohn & Smith, 2011). Thus, there are several reasons why improvement prompts may result in less of a benefit for groups over individuals than the alternate uses prompts. In a recent line of research, Goldenberg and Wiley (­2019) explored whether alternate uses and improvement problems were more susceptible to process loss. These experiments used an EBS format (­a Google chat interface) instead of f­ace-­​­­to-​­face brainstorming to minimize production blocking and create conditions similar to those that have been found to produce the best group results in the literature. In the first study, students brainstormed alone for 20 minutes about novel uses for cars, sports utility vehicles, and/­or vans other than for transportation (­a lternative uses prompt) or about potential improvements to these vehicles (­i mprovement prompts). Participants in the alternative uses condition reported more fixation (­I felt stuck; my earlier ideas got in the way of generating later ideas), while participants in the improvements condition reported having thought more about the question prior to the study. In the second study, participants brainstormed in response to one of the same two prompts, but did so either individually or in interactive groups of three. There was no idea sharing in the individual condition, and they just entered their own responses using the chat interface. In the interacting group condition, participants were able to read each other’s 240

Problem-Solving, Collaboration, and Creativity

responses in the chat. This experiment allowed for the traditional i­ nteracting-​­nominal group comparison. Nominal groups were created by combining all of the ideas generated by successive sets of three participants who had brainstormed individually on the same prompt (­Taylor et al., 1958). The most common index of brainstorming performance, idea quantity, was computed by counting the total number of ­non-​­redundant responses each person generated. Two measures of quality were also computed: the number of highly original ideas and the number of semantic categories. To assess the originality of ideas, first, the frequency of each solution in the sample was computed (­Friedman et al., 2003; Kohn & Smith, 2011; Taylor et al., 1958). Ideas suggested by only a small percentage of the sample (­less than 5%) were considered more original (­this is similar to the rate for the unique ideas generated by only 1 out of the 24 groups in Taylor et al.) and were used to compute the total number of highly original ideas for each group. The variety and flexibility of ideas were assessed using the number of t­ ask-​­relevant semantic categories sampled by each participant (­De Dreu et al., 2008; Goldenberg et al., 2013; Nijstad et al., 2002). The results showed that the interacting groups brainstorming on improvements to vehicles were less productive in terms of both idea quantity and quality than the nominal groups, but this gap was not present when groups brainstormed about alternate uses for vehicles. These results are consistent with the ­well-​­documented process loss that has been observed in studies using improvement prompts for brainstorming. However, on the alternate uses prompts, interacting and nominal groups performed equally well in terms of both idea quantity and quality. Although no benefit from interacting groups was observed, the closing of the performance gap between interacting and nominal groups with the alternate uses questions is a valuable result and still provides an exception to the more typical findings of process loss. This result also suggests that one possible reason for so consistently observing process loss in the group brainstorming literature is o ­ ver-​­reliance on improvement prompts (­Baruah & Paulus, 2008; Diehl & Stroebe, 1987; Kohn & Smith, 2011; Nijstad et al., 2002), and that one of the reasons contributing to the finding of group synergy in the study by Paulus and Yang (­2000) could be because, unlike most other research on the topic, they used an alternate uses prompt (­other uses for a paper clip).

The Effect of Brainstorming Goals on Performance Another issue that has been largely overlooked in previous research concerns the possibility that groups and individuals may pursue different goals when generating their ideas and that the rules they are given may or may not be interpreted the same way in interacting and nominal groups. Osborn suggested several rules for group brainstorming that are generally presented together as a set, following the assumption that each in its own way can help to improve the ideational performance of groups and that their benefits are cumulative. Recently, however, research has begun to critically examine these assumptions, and studies have tested the impact of the rules when presented separately. The quantity rule, particularly as it impacts idea quality, has received perhaps the greatest attention (­Litchfield, 2008, 2009; Litchfield et al., 2011; Paulus et al., 2011; Runco et al., 2005). Some attention has been paid as well to the ­no-​­criticism rule, which will be discussed more later (­Nemeth & Ormiston, 2007; Nemeth et  al., 2004). The effects of the ­free-​­wheeling and ­build-​­on rules, on the other hand, seem especially important to consider to the extent that they may be working at cross purposes (­Goldenberg et al., 2013; Kohn et al., 2011). They imply rather incompatible goals and may invoke opposing cognitive processes. On the one hand, the ­build-​­on rule suggests that the goal of ideation is to generate ideas that extend those that have already 241

Jennifer Wiley and Olga Goldenberg

been proposed. This could encourage fixation, perseverance, or conformity with already suggested ideas or examples ( ­Jansson & Smith, 1991; Kohn & Smith, 2011; Smith, 2003). In brainstorming research, a set of responses spanning fewer semantic categories can be seen as a sign of greater fixation (­Brown et al., 1998; Kohn & Smith, 2011; Larey & Paulus, 1999; Ziegler et al., 2000). The ­f ree-​­wheeling rule, on the other hand, is an explicit call for novelty and distinctiveness. While the b­ uild-​­on rule prompts convergent, incremental thinking (­Guilford, 1959; Guilford & Hoepfner, 1971), which could lead to the use of a narrower range of semantic categories, ­free-​­wheeling demands a more divergent orientation that should result in ideas from a wider range of semantic categories. When presented together as part of the standard set of four instructions given at the start of a brainstorming session, these two rules may establish incompatible goals. If interacting groups are more inclined to pursue the ­build-​­on goal while individuals are more apt to respond to the f­ ree-​­wheeling goal, then this could be another reason for reduced creativity in interacting groups. An added complication is that fixation effects resulting from exposure to the ideas of others may make it harder to follow the ­free-​­wheeling rule than the ­build-​­on rule. It may simply be easier to pursue the goal of building on and modifying already suggested ideas than to strive to generate entirely new ideas (­Ward, 1994). To examine the role of different sets of brainstorming rules, Goldenberg et al. (­2013) had ­four-​­person groups and individuals spend 15 minutes generating ideas for improving their university. One third of participants saw the standard set of four rules. The remaining two thirds saw sets that eliminated either the b­ uild-​­on or f­ree-​­wheeling rule. The performance of the interacting groups was compared to that of f­our-​­person nominal groups who worked alone in the same three instructional conditions. When ­build-​­on and ­free-​­wheeling rules were presented separately, interacting groups and nominal groups responded similarly. Both generated more diverse ­ideas – ​­ideas drawn from a larger number of semantic c­ ategories – ​­in the f­ree-​­wheeling condition. Further, both generated more practical ideas in the b­ uild-​ ­on condition. By contrast, when the two rules were presented simultaneously, interacting groups performed as they did in the ­build-​­on condition, and nominal groups performed as they did in the f­ ree-​­wheeling condition. This study is not the first to suggest that interacting groups draw their ideas from smaller numbers of semantic categories than nominal groups. It is, however, the first to demonstrate that interacting groups are capable of generating ideas from a range of categories when the two conflicting brainstorming rules are presented separately.

Benefits of Incubation and Interleaving Group Discussion with Individual Brainstorming Inspired by Wallas’ (­1926) model of discovery, researchers have explored the importance of incubation or taking time off from a creative ­problem-​­solving task (­Baruah & Paulus, 2019; Dodds et  al., 2003; Sio  & Ormerod, 2009). A number of studies by Steven Smith (­1995, 2003) have shown that taking brief breaks during a creative ­problem-​­solving task can help individuals overcome fixation effects, or mental blocks that result from initial attempts or exposure to examples. When people brainstorm alone, their ­per-​­minute idea generation rates gradually decline as the session progresses, and the same trend is seen in group brainstorming (­Kohn & Smith, 2011). Osborn himself suggested that periods of incubation could promote solution insights in group settings. Kohn and Smith (­2011) empirically tested whether taking a break can help alleviate fixation in an EBS session. To induce fixation, confederates 242

Problem-Solving, Collaboration, and Creativity

presented participants with typical ideas at fixed time intervals. After brainstorming for ten minutes, half of the participants were interrupted and asked to work on an unrelated task for five minutes, after which brainstorming resumed for another ten minutes. In contrast, participants in the control conditions generated ideas without interruptions. However, participants in the interrupted condition came up with 86% more ideas and explored 57% more ­t ask-​­relevant categories than participants who did not take a break. Similarly, a recent study found that engaging in unrelated activities during a break led groups to generate more original solutions in a second attempt at an Unusual Uses Task than simply resting during a break (­Breslin, 2019). Neither of these experiments testing for the effects of incubation or taking a break on group creativity included conditions where individuals participated on their own, which precluded a comparison to nominal groups. However, a study by Christine Smith and colleagues (­Smith et al., 2010) has shown that incubation can also aid group p­ roblem-​­solving above the level of nominal groups. Incubation periods led to increased group performance on rebus problems not only when a priming manipulation provided helpful cues, but also when fixation was induced by misleading cues. Other recent studies provide some support for the suggestion originally proposed by Osborn that interleaving individual and group attempts may be beneficial (­Korde & Paulus, 2017). Because idea generation rates gradually decline as the session progresses, Dennis et al. (­2005) suggested that individuals might benefit from external stimulation provided by others’ ideas, specifically when it becomes difficult to keep generating new solutions. Thus, there may be a benefit to working alone first and then with others. Alternatively, Dugosh et  al. (­2000) found that priming individuals with ideas can increase productivity and the quality of responses. They found that exposing participants to a wider variety of stimulus ideas in the initial phase led to improved variety in their responses (­similar results were also obtained by Nijstad et al., 2002). This suggests that if a range of ideas are proposed during an initial group phase, this might benefit a later individual phase. Nemeth et al. (­2004) also report that the quality of an initial group interaction led to better individual productivity following the session. However, in Kohn and Smith (­2011), exposure to the ideas of others reduced the novelty of ideas (­a s measured by typicality) as well as increased conformity to other participants’ ideas. These results are consistent with those of other researchers, who also noted the convergent tendency of groups in interactive contexts to generate ideas from fewer semantic categories compared to nominal groups (­Larey & Paulus, 1999; Ziegler et al., 2000). Exposure to the ideas of others does not always produce a positive, stimulating effect, but it can sometimes narrow the scope of the idea generation process. This parallels the ­sometimes-​­fixating, ­sometimes-​­facilitating effect of seeing examples of creative design in individuals (­George et al., 2019; 2021; George & Wiley, 2020; Jansson & Smith, 1991; Smith et al., 1993). It is also worth exploring whether it might be helpful to stagger the introduction of the rules in phases so that there is just one goal (­or one subset of compatible goals) to pursue at a time. For example, the f­ree-​­wheeling rule could be emphasized during an initial, divergent brainstorming phase, and the b­ uild-​­on rule emphasized in a separate, convergent brainstorming phase (­de Vreede et al., 2010; Kohn et al., 2011).

Dissent and Diversity Although Osborn advised that judgment should be withheld, some studies are now suggesting that one way to bring group performance to the same, or even higher, level as nominal groups is by actually harnessing the potential of dissent or evaluative comments (­Connolly 243

Jennifer Wiley and Olga Goldenberg

et al., 1990; Jetten & Hornsey, 2014; Wiley & Jensen, 2006; Yong et al., 2014). For example, Nemeth and colleagues (­2004) compared the effectiveness of brainstorming with the usual withhold criticism rule versus the opposite rule: encourage debate and criticism. Although there was no difference among conditions in productivity during the group phase, Nemeth et al. (­2004) found an overall benefit in total productivity (­combined performance across the group phase and a subsequent individual phase) from the debate condition. Similarly, Smith (­2008) explored the benefits of dissent on creative idea generation in a dyad paradigm. When dyads were informed that they represented minority and majority positions, they generated ideas from a broader range of semantic categories than agreeing dyads. Another factor that is closely intertwined with dissent or conflict is that of diversity. In order for there to be a difference of opinion, group members need to have different perspectives. However, the potential impact of group diversity on performance is not c­ lear-​­cut. From one perspective, homogeneous groups may communicate more effectively and form shared representations of the problems that they are solving more easily than diverse groups. If group members with different backgrounds represent and process information differently, this may contribute to inefficiencies in group performance (­H insz et al., 1997; Stout et al., 1999; van Knippenberg & Schippers, 2007). Further, diversity among group members may incur more negative impressions of group members or of group performance and less satisfaction in the group process, while communicating shared knowledge is more rewarding (­Curseu et al., 2007; Wittenbaum et al., 1999). In contrast, an alternative approach proposes that diversity along cognitive dimensions is beneficial for group performance. In an investigation of several molecular biology laboratories, Dunbar (­1997) found that, when scientists in a laboratory are from diverse backgrounds, they are better able to generate alternative hypotheses in the face of unexpected findings, which can, in turn, lead to scientific breakthroughs. Diversity in backgrounds means that each individual has a disparate set of prior experiences that they can mine to offer analogies that may spark insightful solutions (­Dunbar, 1997; George & Wiley, 2018). A group with heterogeneous knowledge should be able to approach p­ roblem-​­solving more flexibly because others can provide the new perspectives that can help free each individual member from fixation or mental set. There is a growing number of empirical studies that have been able to demonstrate that groups high in cognitive diversity may experience advantages in creative ­problem-​­solving. In Stroebe and Diehl (­1994), ­four-​­member groups were asked to generate as many ideas for ways to protect the environment as they could. Cognitively diverse brainstorming groups (­w ith different types of specialized knowledge) generated more ideas for protecting the environment than did the more homogenous groups. Ohtsubo (­2005) manipulated diversity in knowledge by either giving all members of a triad the same knowledge of all of the clues needed to solve a puzzle (­the homogenous condition) or by giving each member a different subset of the clues (­the diverse condition). Cognitively diverse triads solved the puzzles better than the cognitively homogenous triads. Using a very different creative p­ roblem-​­solving task (­a variant of Mednick’s RAT), Wiley and Jolly (­2003) found that cognitively diverse dyads were more effective than more homogenous dyads at finding remote associates. Generally, expertise allows for access to a large amount of ­domain-​­related information as well as fast and easy retrieval of typical solutions in familiar ­problem-​­solving contexts. However, when an atypical or creative solution is required, ­h igh-​­knowledge participants can actually be slower and less likely to reach a solution than novices (­Wiley, 1998). Similar findings have been found with chess experts who are unable to find optimal moves for novel positions because they are blocked by other alternatives 244

Problem-Solving, Collaboration, and Creativity

(­Bilalić et al., 2008). In these cases, expertise can cause fixation or a mental set. Warnings not to use domain knowledge do not help, and giving ­problem-​­solvers an incubation period or break between p­ roblem-​­solving attempts improves ­problem-​­solving, but only for novices (­Wiley, 1998). To help them break out of their mental set, experts may need external cues in order to prime new associations, divert them from incorrect solutions, or direct them toward the correct solution. To test this hypothesis, Wiley and Jolly (­2003) asked students to solve RAT problems in one of three conditions: either in pairs where both partners had low knowledge; in pairs where both partners had high knowledge; or in mixed knowledge ­ igh-​­and one l­ow-​­knowledge member pairs. Only the mixed knowledge pairs with one h performed significantly better than nominal groups (­estimated from solution rates in Wiley, 1998). This result shows that experts may sometimes need the assistance of novices in order to be most effective, flexible, or innovative in their ­problem-​­solving. Canham et al. (­2012) manipulated cognitive diversity in another way by training pairs of students to either use the same or different approaches to solve unfamiliar problems. The results indicated that homogeneously trained dyads were more accurate at solving new problems that directly matched the training problems. However, the diversely trained dyads were better able to come up with novel and creative solutions to transfer problems that required them to apply their knowledge more flexibly. Gijlers and de Jong (­2005) also found that dyads engaging in discovery learning generated more hypotheses when they were heterogeneous in prior knowledge than when they were homogeneous. Some studies have also explored the benefits of diversity among groups with more than two members. In contrast to larger group sizes, dyads have a number of possible advantages: each student has more of an opportunity to participate, and there are fewer group members to distract them from their own thinking (­Dugosh et al., 2000; Moreland, 2010). In addition, there is less of a chance that students will “­­free-​­ride” or “­loaf ” in a group of two, and the student is more likely to feel more highly invested in the product or activity (­Stroebe & Diehl, 1994). This suggests that smaller groups may offer less possibility for loafing, ­free-​­riding, interruption, or distraction. But moving beyond dyads, while the presence of more people in a group may increase the chances of production blocking, interference, and distraction, it also increases the number of perspectives that can be offered and the potential for heterogeneity, dissent, and conflict. If diversity and conflict make the group more likely to think about multiple perspectives, then positive effects may occur. To test for effects from both diversity of knowledge and the potential for conflict (­due to the size of the group), Wiley and Jensen (­2006) used a ­problem-​­solving task where the goal was to decipher a random coding of letters to numbers. Unlike traditional cryptarithmetic problems, where all of the information necessary for solution is contained within the problem itself, the ­Letters-­​­­to-​­Numbers task requires participants to engage in hypothesis testing, think creatively, and achieve insights during solution that will allow them to discover more than one letter at a time. This task seemed particularly interesting because previous studies had found a benefit for groups versus individuals (­a nd nominal groups, Laughlin et al., 2002). The main questions were not only whether heterogeneity in relevant math skills would contribute to performance, but also which group size might be most likely to take advantage of this diversity. Wiley and Jensen found that triads had more letter solutions available per trial than dyads and singletons. Further, triads were better able to take advantage of group heterogeneity in math skills. Perhaps the most important comparison to note is between individuals with math skill, the dyads with a single skilled math member, and the triads with a single skilled math member. Here the number of skilled members is held constant at one, but only in the triad context was there an improvement in p­ roblem-​­solving performance. 245

Jennifer Wiley and Olga Goldenberg

Other research has tested whether similar benefits from collaboration among diverse triads might extend into educational contexts with creative p­ roblem-​­solving tasks called l­earning-­​ ­­by-​­invention activities ( ­Jarosz et al., 2017; Schwartz & Martin, 2004; Sinha & Kapur, 2021). For example, Wiedmann et  al. (­2012) tested for benefits from working in diverse triads on a ­problem-​­solving activity where students were asked to invent a formula for standard deviation before receiving a lesson on the canonical solution. The math background of the members was varied in the triads so that some groups were heterogeneous in their level of math skills, while others were homogeneous. Although one might have expected that groups comprised of only high math members would outperform all other groups, it was the mixed groups that generated the widest variety of solution attempts and more ­h igh-​­quality solution approaches. The discussion of a wider range of possible solution approaches helps students recognize the constraints and affordances of each. This then prepares students to appreciate the critical features of the canonical solution once it is presented to them. Groups including at least one member with high math skills and one with lower math skills gained the most from attempts to develop their own mathematical formulas or procedures. Because the main contrast in this paradigm is generally between groups that are exposed to a lesson first versus groups that engage in ­problem-​­solving first (­a nd not between groups and individuals), benefits from working in diverse groups on ­learning-­​­­by-​­invention activities have not been considered with reference to nominal group estimates. However, these instructional activities were developed precisely to help students gain a deeper understanding of important and difficult concepts that they typically fail to grasp individually (­Schwartz & Martin, 2004). These studies show that cognitive diversity can have a positive effect. Two main explanations can be offered. First, more diverse groups can draw on a broader knowledge base than more homogeneous groups. When different members have different perspectives, they can share those with each other, giving the group access to multiple representations or multiple alternative solution paths. Access to multiple representations may then allow for more flexibility in future ­problem-​­solving. Further, interaction among individuals can yield additional positive effects. During idea generation, an idea proposed by one group member can activate related knowledge in another member, which in turn can lead to a new idea being generated. Through discussion, new, more complex representations emerge that were not held by any member a priori (­Curseu et al., 2012; Schwartz, 1995; Wiedmann et al., 2012; Wiley & Jensen, 2006). In this way, a synergistic effect is created (­Baruah & Paulus, 2019; Larson, 2010; Salazar, 1995).

Idea Elaboration, Development, and Implementation Most of the past research efforts on collaborative creativity have been dedicated to exploring idea generation and selection processes (­Cropley, 2006; Paulus et al., 2019), with much less attention paid to other, equally important stages of the creative process, such as idea development and implementation (­Mumford, 2003; Osborn, 1963). As teams of engineers, scientists, or other individuals collaborate, the process invariably transforms initial ideas into new forms. Through c­ ross-​­fertilization, new ideas emerge that cannot be attributed to any one source. In relation to this point, the editors of this volume were reminded of Halban’s recollection of his work with Kowarski and Joliot on nuclear fission leading up to their 1939 Nature paper: ... we were 12 to 14 hours a day in the laboratory and spent many hours discussing frantically. It is now and was then impossible to say who contributed what. The 246

Problem-Solving, Collaboration, and Creativity

question arose from time to time of somebody saying you had a good idea yesterday, I thought about it, and soon we had a kind of conspiracy agreement, we said really we must stop attributing ideas to one of us, because it might cause one day jealousy and there is a c­ ross-​­fertilization of ideas if three people work together, and you must attribute whatever we did to the group. (­­p. 68, Weart, 1979) In recent years, scholars have begun to investigate development and implementation processes from both theoretical and practical angles, zooming in on the variables that might affect team performance. For example, Paulus et al. (­2019) proposed a set of variables that affect the divergent (­exploration, generation of remote associates, and new idea generation) and convergent (­evaluation, narrowing down, and refinement) processes in group creativity, as well as factors that link the two phases. Their model predicts that facilitating factors in convergent creativity include variables such as task structure and instructions, the cognitive diversity of team members, and affect. The number and quality of ideas in the brainstorming stage, as well as group members’ responses to these, are a few important intervening processes between the divergent and convergent phases. Coursey et al. (­2019) tested a few of these propositions empirically by asking participants to create and develop an idea for a new sport. The divergent phase consisted of asynchronous interactions using online discussion boards (­to avoid process loss), while the synchronous convergent stage focused on idea selection and refinement of the final choice into a more detailed plan. While the number and originality of ideas generated during brainstorming did not affect the originality of the final products, these metrics of performance were influenced by the number and novelty of team members’ replies to each other. This research highlights the importance of the elaboration process. McMahon et  al. (­2016) compared interacting and nominal groups on idea generation, selection, and development by tasking them with the creation of a new language learning game. Initially, no differences between groups and individuals were observed on the quantity and quality of ideas for any of these stages. However, when the researchers zoomed in on the idea development stage by providing participants with only one idea to develop into a full design, they found that the proposals of interacting groups were superior to nominal ones in marketability ratings and superior to individuals in fun ratings. Moreover, groups considered a wider range of game design categories than individuals during the development phase. These results suggest that it might be worthwhile to implement ­t ask-​­focused groups: some for idea generation, others for selection, and yet others for development and implementation. In contrast, research by Goldschmidt (­2016) exploring neurological differences between divergent and convergent thinking in creative design showed that shifts between the two types of processes are so frequent that they are almost inseparable. Overall, this new area of research on stages that go beyond idea generation appears promising toward the discovery of group synergy in creative ­problem-​­solving.

Summary If the main obstacle to creative thinking is that we become fixated in our initial attempts to solve a problem, then collaborating with others seems like an obvious suggestion to help us escape our own thoughts and reach n ­ on-​­obvious solutions. Intuitively, it seems that individuals collaborating in groups should be more effective, flexible, and innovative than individuals working alone. Yet a great deal of work has shown this is not generally the case. The 247

Jennifer Wiley and Olga Goldenberg

emergence of virtual interfaces that circumvent many of the issues that can occur in ­face-­​­­to-​ ­face contexts has been shown to help reduce “­process loss”. Some newer areas of research that are still emerging may help unlock the potential for process gain or synergy. These include considering the role that the nature of the ­problem-​­solving task might play, the goals that are instilled by the set of rules that are given, the benefits that might come from incubation periods, the advantages of interleaving different ­problem-​­solving phases, possible advantages that may arise from dissent or conflict among the members of the group, and the inclusion of more convergent stages of the creative group process. When the goal is to be able to find ­non-​ ­obvious solutions, or to solve novel problems, then cognitive diversity in ­problem-​­solving groups may be an asset, but only if members share their unique knowledge with each other. Author’s Note: This chapter was supported in part by the National Science Foundation grant number 2120658.

References Adamson, R. E. (­1952). Functional fixedness as related to problem solving: A repetition of three experiments. Journal of Experimental Psychology, 44, ­299–​­291. Ball, L. J., Christensen, B. T., & Halskov, K. (­2021). Sticky notes as a kind of design material: How sticky notes support design cognition and design collaboration. Design Studies, 76, 101034. Barki, H., & Pinsonneault, A. (­2001). Small group brainstorming and idea quality: Is electronic brainstorming the most effective approach? Small Group Research, 32, ­158–​­205. Baruah, J.,  & Paulus, P. B. (­2008). Effects of training on idea generation in groups. Small Group Research, 39, ­523–​­541. Baruah, J., & Paulus, P. B. (­2019). Collaborative creativity and innovation in education. In C. Mullen (­Ed.), Creativity Under Duress in Education? Creativity Theory and Action in Education (­­pp. ­155–​­177). Cham: Springer. Beaty, R. E., & Silvia, P. J. (­2012). Why do ideas get more creative across time? An executive interpretation of the serial order effect in divergent thinking tasks. Psychology of Aesthetics, Creativity, and the Arts, 6, ­309–​­319. Bilalić, M., McLeod, P., & Gobet, F. (­2008). Why good thoughts block better ones: The mechanism of the pernicious Einstellung (­set) effect. Cognition, 108, ­652–​­661. Bowden, E. M., & ­Jung-​­Beeman, M. (­2003). Aha! insight experience correlates with solution activation in the right hemisphere. Psychonomic Bulletin & Review, 10, 7­ 30–​­737. Breslin, D. (­2019). ­O ff-​­task social breaks and group creativity. The Journal of Creative Behavior, 53, ­496–​­507. Brown, V. R., Tumeo, M., Larey, T. S., & Paulus, P. B. (­1998). Modeling cognitive interaction during brainstorming. Small Group Research, 29, ­495–​­526. Camacho, L. M., & Paulus, P. B. (­1995). The role of social anxiousness in group brainstorming. Journal of Personality and Social Psychology, 68, ­1071–​­1080. Canham, M., Wiley, J., & Mayer, R. E. (­2012). When diversity in training improves dyadic problem solving. Applied Cognitive Psychology, 25, ­421–​­430. Connolly, T., Jessup, L. M., & Valacich, J. S. (­1990). Effects of anonymity and evaluative tone on idea generation in c­ omputer-​­mediated groups. Management Science, 36, ­689–​­703. Coskun, H. (­2005). Cognitive stimulation with convergent and divergent thinking exercises in brainwriting: Incubation, sequence priming, and group context. Small Group Research, 36, ­466–​­498. Coursey, L. E., Gertner, R. T., Williams, B. C., Kenworthy, J. B., Paulus, P. B., & Doboli, S. (­2019). Linking the divergent and convergent processes of collaborative creativity: The impact of expertise levels and elaboration processes. Frontiers in Psychology, 10, 699. Cropley, A. (­2006). In praise of convergent thinking. Creativity Research Journal, 18, ­391–​­404. Curseu, P. L., Schruijer, S. G. L., & Boros, S. (­2007). The effects of groups’ variety and disparity on groups’ cognitive complexity. Group Dynamics: Theory, Research and Practice, 11, ­187–​­206. Curseu, P. L., Schruijer, S. G. L., & Boros, S. (­2012). Socially rejected while cognitively successful? The impact of minority dissent on groups’ cognitive complexity. British Journal of Social Psychology, 51, ­570–​­582.

248

Problem-Solving, Collaboration, and Creativity De Dreu, C. K. W., Baas, M., & Nijstad, B. A. (­2008). Hedonic tone activation and activation level in ­mood-​­creativity link: Toward a dual pathway to creativity model. Journal of Personality and Social Psychology, 5, ­739–​­7456. Dennis, A. R., Pinsonneault, A., McNamara Hilmer, K., Barki, H., Gallupe, B., Huber, H., & Bellavance, F. (­2005). Patterns in electronic brainstorming. International Journal of E ­ -​­Collaboration, 1, ­38–​­57. Dennis, A. R., & Valacich, J. S. (­1993). Computer brainstorms: More heads are better than one. Journal of Applied Psychology, 78, ­531–​­537. Dennis, A. R., & Valacich, J. S. (­1994). Group, ­sub-​­group, and nominal idea generation: New rules for a new media? Journal of Management, 20, ­723–​­736. Dennis, A. R., & Williams, M. L. (­2005). A m ­ eta-​­analysis of group size effects in electronic brainstorming: More heads are better than one. International Journal of ­E -​­Collaboration, 1, ­24–​­42. DeRosa, D. M., Smith, C. L., & Hantula, D. A. (­2007). The medium matters: Mining the long prom­ eta-​­analysis of the electronic ised merit of group interaction in creative idea generation tasks in a m group brainstorming literature. Computers in Human Behavior, 23, ­1549–​­1581. de Vreede, G. J., Briggs, R. O., & ­Reiter-​­Palmon, R. (­2010). Exploring asynchronous brainstorming in large groups: A field comparison of serial and parallel subgroups. Human Factors, 52, ­189–​­202. Diehl, M., & Stroebe. W. (­1987). Productivity loss in brainstorming groups: Toward the solution of a riddle. Journal of Personality and Social Psychology, 53, ­497–​­509. Diehl, M.,  & Stroebe, W. (­1991). Productivity loss in ­idea-​­generating groups: Tracking down the blocking effect. Journal of Personality and Social Psychology, 61, ­392–​­403. Dodds, R. A., Ward, T. B., & Smith, S. M. (­2003). A review of the experimental literature on incubation in problem solving and creativity. Creativity Research Handbook, 3, ­285–​­302. Dugosh, K. L., Paulus, P. B., Roland, E. J., & Yang, H. C. (­2000). Cognitive stimulation in brainstorming. Journal of Personality and Social Psychology, 79, ­722–​­735. Dugosh, K. L., & Paulus, P. B. (­2005). Cognitive and social comparison processes in brainstorming. Journal of Experimental Social Psychology, 41, ­313–​­320. Dunbar, K. (­1997). How scientists think: ­On-​­line creativity and conceptual change in science. In T. B. Ward & S. M. Smith, (­Eds.), Creative Thought: An Investigation of Conceptual Structures and Processes (­­pp. ­461–​­493). Washington, DC: American Psychological Association. Duncker, K. (­1945). On problem solving. Psychological Monographs, 58, whole no. 270. Washington, DC: American Psychological Association. Friedman, R. S., Fishbach, A., Forster, J., & Werth, L. (­2003). Attentional priming effects on creativity. Creativity Research Journal, 15, ­277–​­286. George, T., Mielicki, M. K., & Wiley, J. (­2021). Great expectations: Misleading effects of images in the alternate uses task. Psychology of Aesthetics, Creativity, and the Arts. Advance online publication. George, T., & Wiley, J. (­2018). Breaking past the surface: Analogical transfer as creative insight. In F. ­Vallée-​­Tourangeau (­Ed.), Insight: On the Origin of New Ideas (­­pp. ­143–​­168). New York: Routledge. George, T., & Wiley, J. (­2019). Fixation, flexibility, and forgetting during alternate uses tasks. Psychology of Aesthetics, Creativity, and the Arts, 13, ­305–​­313. George, T., & Wiley, J. (­2020). Need something different? Here’s what’s been done: Effects of examples and task instructions on creative idea generation. Memory & Cognition, 48, ­226–​­243. George, T., Wiley, J., Koppel, R. H., & Storm, B. C. (­2019). Constraining or constructive? The effects of examples on idea novelty. The Journal of Creative Behavior, 53, ­396–​­403. Gick, M. L., & McGarry, S. J. (­1992). Learning from mistakes: Inducing analogous solution failures to a source problem produces later success in analogical transfer. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18, ­623–​­639. Gigone, D.,  & Hastie, R. (­1993). The common knowledge effect: Information sharing and group judgment. Journal of Personality and Social Psychology, 65, ­959–​­974. Gijlers, H., & De Jong, T. (­2005). The relation between prior knowledge and students’ collaborative discovery learning processes. Journal of Research in Science Teaching, 42, ­264–​­282. Gilhooly, K. J. (­2016). Incubation and intuition in creative problem solving. Frontiers in Psychology, 7, ­1–​­9. Goldenberg, O., Larson, J. R., Jr., & Wiley, J. (­2013). Goal instructions, response format, and idea generation in groups. Small Group Research, 44, ­227–​­256. Goldenberg, O., & Wiley, J. (­2011). Quality, conformity, and conflict: Questioning the assumptions of Osborn’s Brainstorming Technique. The Journal of Problem Solving, 3, ­96–​­118. Goldenberg, O., & Wiley, J. (­2019). Individual and group brainstorming: Does the question matter? Creativity Research Journal, 31, ­261–​­271.

249

Jennifer Wiley and Olga Goldenberg Goldschmidt, G. (­2016). Linkographic evidence for concurrent divergent and convergent thinking in creative design. Creativity Research Journal, 28, ­115–​­122. Guilford, J. P. (­1956). The structure of intellect. Psychological Bulletin, 53, 267. Guilford, J. P. (­1959). Creativity. American Psychologist, 5, ­4 44–​­454. Guilford, J. P., & Hoepfner, R. (­1971). The Analysis of Intelligence. New York: ­McGraw-​­H ill. Hinsz, V. B., Tindale, R. S., & Vollrath, D. A. (­1997). The emerging conceptualization of groups as information processors. Psychological Bulletin, 121, ­43–​­64. Hoffman, L. R., Harburg, E., & Maier, N. R. F. (­1962). Differences and disagreement as factors in creative group problem solving. Journal of Abnormal and Social Psychology, 64, ­206–​­214. Jansson, D. G., & Smith, S. M. (­1991). Design fixation. Design Studies, 12, ­3 –​­11. Jarosz, A. F., Colflesh, G. J. H., & Wiley, J. (­2012). Uncorking the muse: Alcohol intoxication facilitates creative problem solving. Consciousness & Cognition, 21, ­487–​­493. Jarosz, A., Goldenberg, O., & Wiley, J. (­2017). Learning by invention: Small group discussion activities that support learning in statistics. Discourse Processes, 54, ­285–​­302. Jetten, J., & Hornsey, M. J. (­2014). Deviance and dissent in groups. Annual Review of Psychology, 66, ­461–​­465. Kerr, N., & Bruun, S. (­1983). Dispensability of member effort and group motivation losses: Free rider effects. Journal of Personality and Social Psychology, 44, ­78–​­94. Kohn, N. W., Paulus, P. B., & Choi, Y. H. (­2011). Building on the ideas of others: An examination of the idea combination process. Journal of Experimental Social Psychology, 47, ­554–​­561. Kohn, N. W., & Smith, S. M. (­2011). Collaborative fixation: Effects of others’ ideas on brainstorming. Applied Cognitive Psychology, 25, ­359–​­371. Koppel, R. H., & Storm, B. C. (­2014). Escaping mental fixation: Incubation and inhibition in creative problem solving. Memory, 22, ­340–​­348. Korde, R., & Paulus, P. B. (­2017). Alternating individual and group idea generation: Finding the elusive synergy. Journal of Experimental Social Psychology, 70, ­177–​­190. Larey, T. S., & Paulus, P. B. (­1999). Group preference and convergent tendencies in small groups: A content analysis of group brainstorming performance. Creativity Research Journal, 12, ­175–​­184. Larson, J. R., Jr. (­2010). In Search of Synergy in Small Group Performance. New York: Psychology Press. Laughlin, P. R., Bonner, B. L., & Miner, A. G. (­2002). Groups perform better than the best individuals on ­letters-­​­­to-​­numbers problems. Organizational Behavior and Human Decision Processes, 88, ­605–​­620. Linsey, J. S. & Becker, B. (­2010). Effectiveness of brainwriting techniques: Comparing nominal groups to real teams. In T. Taura & Y. Nagai (­Eds.), Design Creativity 2010 (­­pp. ­165–​­171). London: Springer. Litchfield, R. C. (­2008). Brainstorming reconsidered: A ­goal-​­based view. Academy of Management Review, 33, ­649–​­668. Litchfield, R. C. (­2009). Brainstorming rules as assigned goals: Does brainstorming really improve idea quantity? Motivation and Emotion, 33, ­25–​­31. Litchfield, D., & Ball, L. J. (­2011). Rapid communication: Using another’s gaze as an explicit aid to insight problem solving. Quarterly Journal of Experimental Psychology, 64, ­649–​­656. Litchfield, R. C, Fan, J., & Brown, V. R. (­2011). Directing idea generation using brainstorming with specific novelty goals. Motivation and Emotion, 35, ­135–​­143. Lorge, I., & Solomon, H. (­1955). Two models of group behavior in the solution of ­Eureka-​­type problems. Psychometrika, 20, ­139–​­148. Lu, J. G., Akinola, M., & Mason, M. F. (­2017). “­Switching on” creativity: Task switching can increase creativity by reducing cognitive fixation. Organizational Behavior and Human Decision Processes, 139, ­63–​­75. Maier, N. R. F. (­1931). Reasoning in humans. II. The solution of a problem and its appearance in consciousness. Journal of Comparative Psychology, 12(­2), ­181–​­194. Maier, N. R. F. (­1967). Assets and liabilities in group problem solving: The need for an integrative function. Psychological Review, 74, ­239–​­249. Malafouris, L. (­2020). Thinking as ‘­thinging’: Psychology with things. Current Directions in Psychological Science, 29, ­3 –​­8. McMahon, K., Ruggeri, A., Kämmer, J. E., & Katsikopoulos, K. V. (­2016). Beyond idea generation: The power of groups in developing ideas. Creativity Research Journal, 28, ­247–​­257. Mednick, S. A. (­1962). The associative basis of creativity. Psychological Review, 69, ­220–​­232. Moreland, R. L. (­2010). Are dyads really groups? Small Group Research, 41, ­251–​­267.

250

Problem-Solving, Collaboration, and Creativity Moss, J., Kotovsky, K., & Cagan, J. (­2011). The effect of incidental hints when problems are suspended before, during, or after an impasse. Journal of Experimental Psychology: Learning, Memory, and Cognition, 37, 140. ­ eta-​­analytic Mullen, B., Johnson, C., & Salas, E. (­1991). Productivity loss in brainstorming groups: A m integration. Basic and Applied Social Psychology, 12, ­332–​­340. Mumford, M. D. (­2003). Where have we been, where are we going? Taking stock in creativity research. Creativity Research Journal, 15, ­107–​­120. Nemeth, C. J., & Ormiston, M. (­2007). Creative idea generation: Harmony versus stimulation. European Journal of Social Psychology, 37, ­524–​­535. Nemeth, C. J., Personnaz, B., Personnaz, M., & Goncalo, J. A. (­2004). The liberating role of conflict in group creativity: A study in two countries. European Journal of Social Psychology, 34, ­365–​­374. Nijstad, B. A., Stroebe, W., & Lodewijkx, H. F. M. (­2002). Cognitive stimulation and interference in groups: Exposure effects in an idea generation task. Journal of Experimental Social Psychology, 38, ­535–​­544. Nijstad, B. A., & Stroebe, W. (­2006). How the group affects the mind: A cognitive model of idea generation in groups. Personality and Social Psychology Review, 10, ­186–​­213. Ohtsubo, Y. (­2005). Should information be redundantly distributed among group members? Effective use of group memory in collaborative problem solving. Applied Cognitive Psychology, 19, ­1219–​­1233. ­ roblem-​­Solving. New York: Osborn, A. F. (­1953). Applied Imagination: Principles and Procedures of Creative P Scribner. ­ roblem-​­Solving (­3rd ed.). Osborn, A. F. (­1963). Applied Imagination: Principles and Procedures of Creative P New York: Scribner. Paulus, P. B., & Brown, V. R. (­2003). Enhancing ideational creativity in groups: Lessons from research on brainstorming. In P. B. Paulus & B. A. Nijstad (­Eds.), Group Creativity: Innovation Through Collaboration (­­pp. ­110–​­136). Oxford: Oxford University Press. Paulus, P. B., & Brown, V. R. (­2007). Toward more creative and innovative group idea generation: A ­cognitive-​­social motivational perspective of brainstorming. Social and Personality Compass, 1, ­248–​­265. Paulus, P. B., Coursey, L. E.,  & Kenworthy, J. B. (­2019). Divergent and convergent collaborative creativity. In I. Lebuda, & V. P. Glǎveanu (­Eds.), The Palgrave Handbook of Social Creativity Research (­­pp. ­245–​­262). Cham: Palgrave Macmillan. Paulus, P. B., & Dzindolet, M. T. (­1993). Social influence processes in group brainstorming. Journal of Personality and Social Psychology, 64, 5­ 75–​­262. Paulus, P. B., Kohn, N. W., & Arditti, L. E. (­2011). Effects of quantity and quality instructions on brainstorming. The Journal of Creative Behavior, 45, 3­ 8–​­46. Paulus, P. B., Kohn, N. W., Arditti, L. E., & Korde, R. M. (­2013). Understanding the group size effect in electronic brainstorming. Small Group Research, 44, ­332–​­352. Paulus P. B., & Nijstad, B. A. (­Eds.) (­2003). Group Creativity: Innovation through Collaboration (­­pp. ­110–​ ­136). Oxford: Oxford University Press. Paulus, P. B., & Yang, H.-​­C. (­2000). Idea generation in groups: A basis for creativity in organizations. Organizational Behavior and Human Decision Processes, 82, ­76–​­87. Runco, M. A., Illies, J. J., & Eisenman, R. (­2005). Creativity, originality, and appropriateness: What do explicit instructions tell us about their relationships? Journal of Creative Behavior, 39, ­137–​­148. Salazar, A. J. (­1995). Understanding the synergistic effects of communication in small groups: Making the most out of group member abilities. Small Group Research, 26, ­169–​­199. Schwartz, D. L. (­1995). The emergence of abstract representations in dyad problem solving. Journal of the Learning Sciences, 4, ­321–​­354. Schwartz, D. L., & Martin, T. (­2004). Inventing to prepare for future learning: The hidden efficiency of encouraging original student production in statistics instruction. Cognition and Instruction, 22, ­129–​­184. Seifert, C. M., Meyer, D. E., Davidson, N., Patalano, A. L., & Yaniv, I. (­1995). Demystification of cognitive insight: Opportunistic assimilation and the ­prepared-​­mind perspective. In R. J. Sternberg & J. E. Davidson (­Eds.), The Nature of Insight (­­pp. ­65–​­124). Cambridge, MA: MIT Press Sheppard, J. A. (­1993). Productivity loss in performance groups: A motivation analysis. Psychological Bulletin, 113, 6­ 7–​­81. Simonton, D. K. (­1997). Creative productivity: A predictive and explanatory model of career trajectories and landmarks. Psychological Review, 104, ­66–​­89.

251

Jennifer Wiley and Olga Goldenberg Sinha, T., & Kapur, M. (­2021). When problem solving followed by instruction works: Evidence for productive failure. Review of Educational Research, 91, ­761–​­798. Sio, U. N., & Ormerod, T. C. (­2009). Does incubation enhance problem solving? A ­meta-​­analytic review. Psychological Bulletin, 135, 9­ 4–​­120. Smith, R. W.,  & Kounios, J. (­1996). Sudden insight: A ­ ll-­​­­or-​­none processing revealed by ­speed–​ ­accuracy decomposition. Journal of Experimental Psychology: Learning, Memory, and Cognition, 22, ­1443–​­1462. Smith, S. M. (­1995). Getting into and out of mental ruts: A theory of fixation, incubation, and insight. In R. J. Sternberg & J. E. Davidson (­Eds.), The Nature of Insight (­­pp. ­121–​­149). Cambridge, MA: MIT Press. Smith, S. M. (­2003). The constraining effects of initial ideas. In P. B. Paulus & B. A. Nijstad (­Eds.), Group Creativity: Innovation through Collaboration (­­pp. ­15–​­31). Oxford: Oxford University Press. Smith, S. M. (­2008). Adding minority status to a source of conflict: An examination of influence processes and product quality in dyads. European Journal of Social Psychology, 38, ­75–​­83. Smith, S. M., & Blankenship, S. E. (­1989). Incubation effects. Bulletin of the Psychonomic Society, 27, ­311–​­314. Smith, S. M., & Blankenship, S. E. (­1991). Incubation and the persistence of fixation in problem solving. The American Journal of Psychology, 104, ­61–​­87. Smith, C. M., Bushouse, E., & Lord, J. (­2010). Individual and group performance on insight problems: The effects of experimentally induced fixation. Group Processes & Intergroup Relations, 13, ­91–​­99. Smith, S. M., Gerkens, D. R., & Angello, G. (­2017). Alternating incubation effects in the generation of category exemplars. The Journal of Creative Behavior, 51, ­95–​­106. Smith, S. M., Ward, T. B., & Schumacher, J. S. (­1993). Constraining effects of examples in a creative generation task. Memory & Cognition, 21, ­837–​­845. Sohrab, S. J., Waller, M. J., & Kaplan, S. (­2015). Exploring the hidden profile paradigm: A literature review and analysis. Small Group Research, 46, ­1–​­47. Stasser, G., & Titus, W. (­1985). Pooling of unshared information in group decision making: Biased information sampling during discussion. Journal of Personality and Social Psychology, 48, ­1467–​­1478. Steiner, I. D. (­1972). Group Process and Productivity. New York: Academic Press. Stout, R. J., ­Cannon-​­Bowers, J. A., Salas, E., & Milanovich, D. M. (­1999). Planning, shared mental models, and coordinated performance: An empirical link is established. Human Factors, 41, ­61–​­71. Stroebe, W., & Diehl, M. (­1994). Why groups are less effective than their members: On productivity losses in ­idea-​­generating groups. European Review of Social Psychology, 5, ­271–​­303. Taylor, D. W., Berry, P. C., & Block, C. H. (­1958). Does group participation when brainstorming facilitate or inhibit creative thinking? Administrative Science Quarterly, 3, ­23–​­47. Torrance, E. P. (­2008). Torrance Tests of Creative Thinking: N ­ orms-​­Technical Manual, Verbal Forms A and B. Bensenville, IL: Scholastic Testing Service. Valacich, J. S., Dennis, A. R., & Nunamaker, J. F., Jr. (­1992). Group size and anonymity effects on ­computer-​­mediated idea generation. Small Group Research, 23, ­49–​­73. van Knippenberg, D., & Schippers, M. C. (­2007). Work group diversity. Annual Review of Psychology, 58, ­515–​­541. Wallach, M. A., & Kogan, N. (­1965). Modes of Thinking in Young Children. New York: Holt, Rinehart, and Winston. Wallas, G. (­1926). The Art of Thought. New York: Harcourt Brace. Ward, T. B. (­1994). Structured imagination: The role of conceptual structure in exemplar generation. Cognitive Psychology, 27, ­1–​­40. Weart, S. R. (­1979). Scientists in Power. Harvard: Harvard University Press. Wertheimer, M. (­1945). Productive Thinking. New York: Harper. Wiedmann, M., Leach, R., Rummel, N., & Wiley, J. (­2012). Does group composition affect learning by invention? Instructional Science, 40, ­711–​­730. Wieth, M., & Zacks, R. (­2011). Time of day effects on problem solving: When the ­non-​­optimal is optimal. Thinking & Reasoning, 17(­4), ­387–​­401. Wiley, J. (­1998). Expertise as mental set: The effects of domain knowledge in creative problem solving. Memory & Cognition, 26, 7­ 16–​­730. Wiley, J., & Jarosz, A. F. (­2012). Working memory capacity, attentional focus, and problem solving. Current Directions in Psychological Science, 21, 2­ 58–​­262.

252

Problem-Solving, Collaboration, and Creativity Wiley, J., & Jensen, M. (­2006). When three heads are better than two. Proceedings of the Annual Conference of the Cognitive Science Society (­­pp. ­2375–​­2380). Vancouver: Cognitive Science Society. Wiley, J., & Jolly, C. (­2003). When two heads are better than one expert. Proceedings of the Annual Conference of the Cognitive Science Society (­­pp. ­1242–​­1246). UC Merced: Cognitive Science Society. Williams, K. D., Harkins, S. G., & Karau, S. J. (­2003). Social performance. In M. A. Hogg & J. Cooper (­Eds.), Sage Handbook of Social Psychology (­­pp. ­327–​­346). Thousand Oaks, CA: Sage. Wittenbaum, G. M., Hollingshead, A. B.,  & Botero, I. C. (­2004). From cooperative to motivated information sharing: Moving beyond the hidden profile paradigm. Communication Monographs, 71, ­286–​­310. Wittenbaum, G. M., Hubbell, A. P.,  & Zuckerman, C. (­1999). Mutual enhancement: Towards an understanding of the collective preference for shared information. Journal of Personality and Social Psychology, 77, ­967–​­978. Yong, J., Sauer, S. J., & Mannix, E. A. (­2014). Conflict and creativity in interdisciplinary teams. Small Group Research, 45, ­266–​­289. Ziegler, R., Diehl, M., & Zijlstra, G. (­2000). Idea production in nominal and virtual groups: Does ­computer-​­mediated communication improve group brainstorming? Group Processes and Intergroup Relations, 3, ­141–​­158.

253

16 METAPHORIC CREATIVITY AS EMBODIED PERFORMANCE IN SOCIAL INTERACTION Thomas Wiben Jensen

Introduction The relationship between metaphor and creativity is a close fit, as glove fits hand. According to Lakoff and Johnson (­1980), “­the essence of metaphor is understanding and experiencing one kind of thing in terms of another” (­­p. 5), which in varying degrees may involve a creative dimension. Indeed, many novel literary metaphors can be considered prime manifestations of human creativity. In the same vein, creative processes and thinking are often enabled by metaphorical reasoning or imagination. Metaphors may work as tools for developing new ideas and perspectives on established reality since creative processes often rely on our ability to see something in terms of something else. Within the tradition of conceptual metaphor theory (­CMT), this process (­to see something in terms of something else) is approached via the notion of embodiment in the sense that the focus is on how we understand abstract concepts and ideas via a transfer from more concrete, typically embodied, areas of experience (­Lakoff and Johnson 1980 and 1999; Kövecses 2005; Gibbs 2017). The basic argument is that metaphors are grounded in bodily experiences that structure the way we think and speak about notions and experiences that do not in themselves have a physical character. Popular examples are both primary metaphors such as AFFECTION IS WARMTH (­she cares warmly for her children) and more complex metaphors such as LIFE IS A JOURNEY (­he has come a long way since his youth). This way of looking at metaphor is closely linked to the general notion of embodiment, emphasizing that the brain and body evolved together and are intrinsically coupled. The brain is part of a larger cognitive system that includes the nervous system and sensorimotor capabilities, and this fundamental connectedness can be witnessed in metaphorical structures (­Lakoff and Johnson 1999; Gibbs 2005 and 2017; Johnson 2007). Despite this intense focus on embodied structures and processes within CMT, most studies of metaphor have not addressed creativity as a process but rather as a product, that is, by analyzing the presence of creative metaphors in various contexts (­­Hidalgo-​­Downing 2020). In this qualitative study, however, I present another way into the study of creative cognitive processes in relation to metaphor, that is, metaphoric creativity as embodied performance in social interaction. This means that the object of study is not primarily complex and creative linguistic metaphors but rather embodied metaphorical actions embedded in the dynamics of social interaction. This perspective on metaphorical action is closely connected to the 

254

DOI: 10.4324/9781003009351-18

Metaphoric Creativity in Social Interaction

­so-​­called radical embodied approach within cognitive science, which claims that cognition is fundamentally for action and as such is best studied as part of ­person–​­environment dynamics rather than described via ­rule-​­based information processing as in more classical computational approaches to cognition (­Chemero 2011; Anderson 2014; Gallagher 2017; Di Paolo et al. 2018). Likewise, from a linguistic perspective, the focus is on languaging as a ­whole-​ b­ ody ­sense-​­making activity taking place in socially and semiotically structured environments (­Maturana 1978; Bottineau 2010; Thibault 2011). Based on empirical examples analyzed from a combined e­cological-​­enacted approach to language and cognition, I argue that metaphoric creativity is not reserved to the use of complex linguistic metaphors but can also be understood as embodied creative actions that afford new perspectives and new action possibilities. These metaphorical performances can be defined as creative since they appear adaptive in the sense of being novel (­i.e., original and unexpected) and appropriate (­i.e., adaptive concerning task constraint) as defined in creativity research (­Sternberg and Lubart 1999; Runco 2004; Veale et al. 2013). Adaptiveness as a phenomenon is what “­enables us to understand novelty as conditioned by contextual factors” (­­Hildalgo-​­Downing 2020, ­p. 5). These embodied performances may include language, but the metaphorical value of the actions does not solely rely on words. Rather, from a linguistic point of view, these performances are activities “­in which wordings play a part” (­Neumann and Cowley 2013, p­ .  18), emphasizing that from a vantage point in languaging, “­rather than treating form or meaning as primary, it is recognized that we perceive bodily events as wordings” (­I bid.). Wordings do “­play a part”, but they are always to be seen as embedded in larger complexes of bodily actions, and it is this totality that is the center of attention for this study. Furthermore, the metaphorical performances bear a resemblance to Gallagher and Lindgren’s notion of enacted metaphor as something “­that we ­enact—​­that is, one that we put into action or one that we bring into existence through our action” (­Gallagher and Lindgren 2015, ­p. 392). Again, the focus is on how we “­do metaphor” in and through our embodied actions rather than on the structural complexities and semantic duality of certain verbal (­or literary) metaphors. In this way, this qualitative empirical study adds to research in creativity by showing that creative cognitive processes, such as metaphorical meaning making, are often closely intertwined with mundane activities; they are not necessarily something unique, detached from everyday doings, but rather they are part and parcel of how we do things and jointly coordinate our actions.

Structure of the Chapter In the following, a brief overview of contemporary research in metaphor is presented with a special focus on the role of embodiment. This is followed by an introduction to ­ecological-​ e­ nactive trends within cognitive science with a specific focus on the relationship between language, action, and creativity. Following this, the new developments within a dynamic approach to both metaphor and metaphoricity are addressed, followed by a short description of the body of research on metaphor and creativity. This leads to the method section and the analytical part, which consists of two i­n-​­depth analyses of metaphorical performances in social interaction from two different settings: a learning session in a kindergarten with a teacher and three children and a therapy session involving a therapist and a client. The analyses will be carried out on the basis of both transcriptions and anonymized drawings. Finally, in the concluding section, the analyses will be put into perspective in relation to the topics at hand. 255

Thomas Wiben Jensen

Theory Section Conceptual Metaphor Theory and Embodiment The rise of CMT in the 1980s and 1990s involved a shift of focus from seeing metaphor as primarily a linguistic phenomenon chiefly involving rhetorical or aesthetic dimensions typically employed by poets and writers to a focus on metaphor as first and foremost a cognitive phenomenon (­Lakoff and Johnson 1980, 1999; Johnson 1987; Lakoff 1993; Kövecses 2005). This also led to a dramatic expansion of the field of metaphor studies, now involving the mundane and much more widespread use of both conventional and novel metaphors in all areas of language use. The focus was now on the cognitive mechanisms at play in metaphor use in structuring our basic ­sense-​­making and experience. Metaphor was now studied as a cognitive process of mapping between source and target domains in the human mind. This mapping process is motivated by a transfer from ­pre-​­linguistic embodied experiences onto the ways we comprehend and talk about abstract concepts. The basic claim is that metaphorical expressions at the level of language are grounded in sensorimotor experiences. In this way, metaphor use is seen as motivated and structured by conceptual metaphors on a ­cognitive-​­experiential level in ways we are rarely aware of when speaking metaphorically (­Lakoff 1993). Thus, in using expressions like “­I don’t see your point”, “­your arguments are blurred by opaque language”, “­she is extremely bright”, and “­can you shed some light on your ideas”, we talk about and understand the abstract concepts of knowing and understanding as if they had a visual or optical quality; hence the conceptual metaphor KNOWING IS SEEING. The strong focus on embodiment in CMT, as part of s­econd-​­generation cognitive science, was an important breakthrough in relation to understanding the vital role of the body in virtually all aspects of our thinking, including language production. However, seen from a contemporary, more ecological approach, it also came with a price. The seemingly logical assumption that since cognition is profoundly embodied and our bodies visibly have physical boundaries that separate us from the surrounding environment, cognition too must, by definition, be a bounded phenomenon tied to an individual body and reserved to processes in the head (­Gibbs 2017; Jensen 2017). In this way, even though the notion of embodiment within CMT in the outset intended to resolve the ­m ind-​­body dualism, the strong focus on embodied experience unwillingly came to refine and consolidate already established distinctions between individual/­ social, bodily/­ contextual, cognition/­ communication, thought/­ language, and so forth (­Cuffari and Jensen 2014). Thus, in relation to studies of metaphor, the ability to see and create potential doubleness via words, images, and gestures has been traditionally understood and described as a process taking place inside the human mind. Today, another trend within metaphor studies is emerging. Metaphor, the human tendency to enact doubleness, to see, feel, experience, and understand one kind of thing in terms of another, can also be seen as stemming from perceived invariances in the environment and not chiefly from mental operations of ­cross-​ ­domain mappings. This way of approaching metaphor relies chiefly on recent tendencies in cognitive science as proposed by an ­ecological-​­enactive perspective.

­Ecological-​­Enactive Cognition Over the last 1­ 5–​­20 years, new ways of understanding and investigating cognition have emerged. In short, the notion of cognition has been pushed, expanded, and elaborated 256

Metaphoric Creativity in Social Interaction

(­Hutchins 1995; Gibbs 2006; Johnson 2007; Stewart et al. 2010; Clark 2011; Cowley and ­Vallée-​­Tourangeau 2013; Fuchs 2017; Gallagher 2017; Malinin 2019). Today, the term cognition does not just refer to mental or psychological ­in-​­skull processes (­even though their role and existence are not denied), nor does it just refer to neural processes in the brain (­even though they of course play a necessary, yet not constitutive, part), and likewise, it is not sufficient to say that cognition is grounded in bodily processes (­even though they are clearly part of cognition). Instead, cognition is to be understood as emerging from both internal and external processes that are distributed across the ­brain-­​­­body-​­environment. In this sense, an e­ cological-​­enactive approach to cognition focuses on the ways in which cognition arises via couplings with environments (­physical and ­socio-​­cultural), which shape, and are shaped by, the actors who inhabit them. The implication is that actors perceive their environments in terms of the affordances (­opportunities for action) they provide to them (­Gibson 1977; Gallagher 2017). In short, the notion of affordances implies that we see and understand entities in the world in terms of what we can do with them; we perceive them in and through their action potential, not in terms of what they “­a re” independent of their use. That is, the round shape of a cup is seen in and through its “­­grasp-​­ability” (­a nd “­d rinkability” of its content), the flat surface of a table in and through its “­­place-​­ability”, and the smiley face of another human being in and through its “­sociability” or “­­talk-​­ability”, depending on the circumstances of course. In this view, cognition is about grasping the affordances of the environment (­physical and/­or social) and as such more than an inner mental structure. The idea of cognition as enactive refers to the basic idea that cognition is for action. It is in and through meaningful actions that we make sense of our environment, allowing us to maintain our existence within it (­Varela et al. 1991; Stewart et al. 2010). The enactive thesis was developed by Varela et al. (­1991) around the notion of autopoiesis, a concept that describes how organisms create their own experiences by initiating actions in environments. In more contemporary terms, the core idea of enactive cognition is often described in terms of couplings that brings forth a world: To bring forth a world means to enact dimensions of meaning and significance through the living body in action and through multiple kinds of physiological, sensorimotor, and interpersonal couplings. The mind is what occurs in these enactments and not what goes on in the head. (­Di Paolo et al. 2018, ­pp. ­17–​­18) Recently, Baggs and Chemero (­2018) have argued for a synthesis between ecological and enactive approaches to cognition, combining the ecological focus on affordances and the enactive focus on constructing (­bringing forth) the environment, or umwelt. The suggestion is to distinguish between the physical world as the enduring structure of the environment irrespective of species inhabiting it and then, secondly, between habit and umwelt as designating ­species-​­specific environments and ­actor-​­specific environments, respectively (­Malinin 2019, p­ . 4). In relation to approaching metaphor as embodied performance in the present study, the focus will be on both ­human-​­specific environments (­habit) and ­actor-­​­­specific-​­environments (­u mwelt).

An E ­ cological-​­Enactive Approach to Metaphor and Metaphoricity In the last 1­ 0–​­15 years, different dynamic and ecological perspectives on metaphor have been proposed (­Cameron et al. 2009; Cameron 2012; Jensen 2017, 2018; Gibbs 2019; Müller 257

Thomas Wiben Jensen

2019; Jensen and Greve 2019; Szokolszky 2019). Overall, there is a shift from a focus on ­cross-​­domain conceptualization in individual cognition to emergent experiencing in dynamic social interactions. This implies that metaphor can now be seen as a product of a ­person–​­environment ­relation  – ​­rather than a product of an inner mental process, which emphasizes the need to study the ways in which metaphor performance is embedded in environmental structures. One way to approach this shift is via the notion of metaphoricity (­Müller 2008; Müller and Tag 2010; Jensen and Cuffari 2014; Jensen 2017). Metaphoricity refers to the way we “­do metaphor” in online processes of speaking, writing, gesturing, and so on. As such, it is bound to active s­ense-​­making in the h ­ ere-­​­­and-​­now, embedded in a ­socio-​­cultural context, and constrained by elements of the environment. Importantly, it implies a view on metaphoric meaning as a gradable phenomenon; a scalar value that can be more or less present. Thus, it entails a shift of focus from metaphor as a cognitive product (­of ­ ell-​­defined and restricted linguistic entity to a focus on the a ­cross-​­domain mapping) and a w process of creating and enacting some kind of metaphorical, or double, meaning. In this way, metaphoricity can be examined as part of adaptive and coordinated behavior among various participants in interaction. Furthermore, as mentioned in the introduction, it has recently been proposed by Gallagher and Lindgren (­2015) that certain metaphors can be viewed as enacted phenomena rather than, for instance, ­pre-​­existing metaphors we encounter in a written text. An example of enactive metaphor could be a child picking up a banana and pretending it is a phone, thereby understanding and experiencing the banana metaphorically as a phone. Being part of a playful engagement with the environment, such a metaphorical action is tightly linked to affective and intersubjective dimensions as a way of manipulating the environment and thereby inviting further playful actions. In continuation of this, the important aspect here is that, from an ­ecological-​­enactive perspective, this kind of metaphoric action is not part of an inner mental operation in which the banana is somehow represented as a phone; rather, it is literally in the doing, in the grabbing movements of the hand and following manipulation of the environment, that the metaphor is enacted. The child, “­in effect, enacts a metaphor that builds on an affordance presented by the shape of the banana and on her previous experience with phones” (­Gallagher and Lindgren, ­p. 396). In the analytical section, I will look at two different examples of creative metaphorical performances that, to some extent, bear a resemblance to the banana example, even though they both involve a more complex combination of bodily and verbal actions. In line with the e­ cological-​­enactive approach, the focus is on the ­co-​­available felt sense in the interaction and what is made possible in the conversational interaction as it unfolds ( ­Jensen and Cuffari 2014). In this way, metaphors on an ­ecological-​­enactive understanding of languaging are not just words spoken but significant behaviors that serve in the moment to mark, capture, constrain, organize, and redirect the experiential flow, thereby allowing for creative processes to emerge. This perspective is also in line with new tendencies within creativity research.

Creativity Research and Metaphor Within creativity studies, researchers often differentiate between creativity as a product or process and, likewise, between artistic, big C creativity, and small c creativity (­Runco 2004). Small c creative acts may appear in many mundane settings and discourse practices, and the analytical focus is often on recontextualization and social value of creativity in various social contexts. Within the big C, creativity is more aligned with an analytical focus on the result of creative acts or endeavors, whether it is in the context of painting, music, literature, 258

Metaphoric Creativity in Social Interaction

film, advertisements, dance, etc. (­­Hildalgo-​­Downing 2020). In this study, the focus is on the small c creativity approached primarily via the concepts of adaptiveness, flow, and emotion. As mentioned in the introduction, adaptiveness in relation to creativity is closely related to contextual factors in the sense that novelty may appear as a consequence of recontextualization. That is, the process of taking already familiar and in itself unoriginal concepts into new contexts thereby activating their novelty value. It is often the context motivating the creative performance that determines the degree of creativity. The notion of flow has been examined by Csikszentmihalyi (­1996), who concludes that creative moments often arise when participants (­artists, athletes, and scientists) are in a state of flow, which is “­a n almost automatic, effortless, yet highly focused state of consciousness” (­­p. 110). In relation to the present study, this is important since the notion of flow can enable us to see creativity as arising spontaneously on the spot as people are completely engaged in the interaction. Likewise, emotion is a key component since a core quality of emotions lies in the way in which they saturate experiences with value. In and through emotions, we experience something as ­valuable – ​­fearful, exciting, boring, scary, attractive, or perhaps new and creative (­LeDoux 1996; Damasio 1999). In this sense, emotions may disclose what a situation affords in terms of potential doings, and the specific efforts required in these doings. In relation to creativity, this is crucial since emotions often are the key factor that spark spontaneous actions also in a creative fashion (­Slaby et al. 2013; Jensen and Pedersen 2016). In relation to metaphor, this study is conducted from a vantage point, in which metaphoric creativity is contextually driven and shaped. A metaphoric performance (­verbal or ­non-​­verbal) or expression is rarely creative in itself. Also, it is the context of use (­a nd ­re-​­use) that determines the degree of creativity (­­Hildalgo-​­Downing 2019). This is an important insight since it makes metaphoric creativity more accessible and, even more importantly, it illustrates how (­metaphoric) creative processes are often part of other activities, for instance various forms of coordination and communication that provide for social action and change, as we will witness in the analyses.

Analytical Section Data and Method In this chapter, two r­eal-​­life examples from different settings are analyzed: a talk about emotions in a kindergarten in the first example and a therapy session in the second example. The examples are embedded in what one might call “­organizational ­eco-​­systems”, in which certain issues with different levels of specificity are addressed. In the first example, the case concerns the rather abstract topic of reflecting on the nature of the emotion of sadness, whereas the second example concerns a more concrete (­but still complicated) case of the patient having received a letter of concern from the municipality potentially concerning her parental authority. A consequence of dealing with organizational e­ co-​­systems is that, in both settings, one finds a set of expectations for how actions can be carried out by the participants to achieve the desired goals. However, in both examples, surprising and creative actions are accomplished during the ongoing interactions. As the participants engage in the task, they engage with the affordances in the s­pecies-​­specific and a­ ctor-​­specific environments in creative and unpredictable ways. This, in turn, highlights the intertwined nature of metaphoric performance in relation to the affordance structure of the environment. The method used in the following analyses is Multimodal Interaction Analysis (­M MIA; Norris 2004; Goodwin 2017). MMIA is devised to investigate social interaction as a 259

Thomas Wiben Jensen

­ hole-​­bodied activity embedded in a physical and social environment. At the heart of the w method lies the assumption that the verbal and bodily ­non-​­verbal dimensions of language are equally important dimensions in the act of “­doing language” (­or languaging) with other people. Thus, MMIA takes into account the full array of situated embodied actions, including gesture, gaze, facial expression, posture, and head movement in synchronization with verbal utterances. Likewise, MMIA makes it possible to investigate languaging as a w ­ hole-​ ­bodied activity encompassing affect and emotion. Thus, textual transcriptions (­both Danish and English) are combined with drawings to better show the dynamics of interaction. The drawings are conducted by a professional illustrator and are completely anonymized. The names used in the first example are all pseudonyms. This study was reported to the South Danish Regional Committee on Health Research Ethics and the Legal office of the University.1 Written informed consent was obtained beforehand from all involved parties for participation and publication of the extracts as well as the accompanying drawings.

Example 1: Performing Sadness2 Kay, Louise, and Peter all attend a Danish kindergarten for f­our-​­to ­six-­​­­year-​­olds.3 A popular learning activity in the kindergarten is the s­o-​­called emotion talks. These are led by one of the teachers, who initiates the talk by showing the children a picture of another child with a distinct emotional expression. Then they all talk about the child in the picture and the emotion expressed there. The goal of these emotion talks is to get the children to articulate their emotional experiences and help them make sense of different emotions. In this case, as shown in ­Figure 16.1, the teacher presents the children with a photo of a girl with a sad expression on her face. The following sequence takes place right after the picture has been presented to the children. Extract 1 Time code: Danish Original T: teacher, P: Peter, L: Louise, K: Kay (­A nna: the name given to the girl on the picture) 1 P: er hun da død? 2 T: om hun er død hende her? (­0.5) nej men måske er hun ked af det fordi hun kender hun nogen der 3 er død det ved vi jo ikke (.) vi kan jo kun se på Annas ansigt [vi kan jo ikke vide] hvad hun er ked af 4 K: [åh jeg har ondt i hjertet] 5 K: åh jeg har ondt i hjertet nu at jeg hører om de følelser😊 6 L: og jeg har 7 T: får du ondt i hjertet Kay? 8 K: [ ja😊] 9 L: [og jeg] (.) får ondt I maven af at høre om følelserne 10 T: ja det kan jeg godt forstå 11 K: a.. AV English translation 1 P: Is she dead then? 2 T: if she is dead this girl (­0.5) no but maybe she is sad because she knows someone who is dead we 260

Metaphoric Creativity in Social Interaction

3 don’t know that (.) we can only look at Anna’s face [we cannot know] what makes her sad 4 K: [oh I have a pain in the heart] 5 K: oh I have a pain in the heart now that I hear about these feelings😊 6 L: and I have 7 T: do you get a pain in the heart Kay? 8 K: [yes😊] 9 L: [and I] (.) and I get a pain in the stomach in hearing about these feelings 10 T: yeah I understand that 11 K: o..OUCH As stated in the theory section, the ­ecological-​­enactive approach to metaphor and languaging entails a focus on more than the words spoken, also allowing for analytical consideration of significant behaviors in the flow of interaction. In this example, we will zoom in on the behaviors of the two girls, Kay and Louise, which constrain and redirect the experiential flow in unexpected and creative ways. Still, to provide the context for these actions, we shall first look closer at the verbal interaction initiated by the teacher. In the first line, the boy Peter asks whether the girl in the picture (­A nna) might be dead. In response to that, in lines 2­ –​­3, the teacher outlines what we might call a possibility zone. The teacher explains that surely the girl in the picture is not dead, but she might know someone who is dead. In this way, it is noticeable that the children do not at first interpret the face as sad; rather, it is the teacher who facilitates this interpretation, in which she states that, as observers, we cannot know what has caused her sadness. We only have access to the expression of sadness, “­we can only look at Anna’s face”, not the reason causing sadness. Put another way, the possibility zone offers only the epistemic mode of seeing (­f rom a ­third-​­person perspective) rather than knowing (­from a first person perspective). However, this structuring of seeing versus knowing as the only liable option is challenged by the two girls, Kay and Louise. Indeed, in the following turns, they enact and populate an experience zone in the flow of interaction, thus reconfiguring the experiential affordances in the ongoing next moments of the interaction. Zooming in on Kay, we can notice that, in the beginning of the sequence (­l. ­1–​­3), she is sitting still, alternating between gazing at the picture and sometimes the teacher. But in line 4, she suddenly shifts into a more active mode. Just before she utters, “­oh I have a pain in the heart”, a conventional expression in Danish that is often semantically associated with sadness, Kay starts performing various bodily actions. Most notably, she opens her mouth wide and makes a contorted face as if she is in pain (­see ­Figure 16.1a). The teacher’s attention, however, is still on the other girl, Louise (­not visible on the drawing but sitting behind the boy). After not having received the attention of the teacher, Kay then repeats her turn in line 5 about having a pain in the heart and adds, ”now that I hear about these feelings”.

­Figure 16.1 (­­a–​­c) Teacher and children engaged in interaction in a kindergarten

261

Thomas Wiben Jensen

She thereby puts her emotional outburst into context, and notably, at the end of this turn, Kay now changes her emotional style. She smiles, gestures vividly, and gazes at the teacher, as well as the other children, with a playful look on her face, probably due to the fact that she now got the teacher’s attention (­see ­Figure 16.1b). Finally, at the end of the sequence in line 11, Kay again performs a whole body expression of ­pain—​­or perhaps rather of being ­wounded—​­by leaning forward grimacing with an open mouth and smiley eyes while uttering the sound “­OUCH” with high volume (­see ­Figure 16.1c). In doing this, Kay performs pain, or perhaps she enacts an embodied performance related to the use of the expression “­pain in the heart” by playing with the emotion rather than trying to describe it in words. A traditional metaphor (­CMT) analysis might see the utterance, “­I have a pain in the heart”, as first and foremost an instance of the conceptual metaphor THE BODY IS THE MIND or a metonymy like PHYSICAL AGITATION FOR THE EMOTION (­Kövecses 2000, ­p. 82). However, such a ­word-​­based approach cannot capture the particularities of the metaphoric performances. Kay is neither in physical pain, nor emotional pain, but, in a playful way, she performs both. In this sense, she engages in emotional learning in and through a metaphoric performance. She is engaging in emotional learning by physically acting wounded and thereby metaphorically “­doing sadness”. Thus, the embodied connection between “­wounded” and “­sad” entails a double meaning and enacts metaphoricity. Kay enacts a metaphorical meaning in the sense that she seems to understand and, to some degree, experience sadness as pain in the heart. Crucially, the actions of Kay afford a further elaboration by Louise in line 6, when she uses the related expression, “­a pain in the stomach”. Together and in competition, Kay and Louise enact a zone of language play as well as an emotional and creative engagement in which they explore and play with the experience of sadness and pain. When treated as a whole, the metaphoric performances are c­ o-​­experiential. They draw on experience and knowledge about sadness, and they also use the situational affordances for playing out ­sadness-​­related feelings in a group. As they enact the emotions of others, they understand such expressions as embodied ­self-​­expression. By playing out expressions through her body, Kay enacts and plays with the emotion of sadness of the pictured girl, which invokes an emotional change or attachment to the story. The Russian philosopher and literary critic Mikhail Bakhtin (­­1895–​­1975) has famously described how a word in a language is never entirely your own. Words exist on a longer time scale due to the ways in which they have been used before: The word in language is half someone else’s. It becomes one’s “­own” only when the speaker populates it with his own intentions and his own accent, when he appropriates the word, adapting it to his own semantic and expressive intention. (­Bakhtin 1981, ­p. 294) Related to this example, one might add that the girls are appropriating not only the words but also the behavior and expressions of sadness, thereby adding to the creative value of the example. They are not necessarily conscious of the metaphoric potential of their actions. Rather, it could be argued that they may be in a state resembling that of flow, as described by Csikszentmihalyi (­1996), in which they, in a spontaneous fashion, elaborate on sadness by trying out conventional, as well as more unconventional, images of sadness as pain. The connection between physical pain and sadness in itself is far from novel, but in and through this recontextualization, it gains a creative value since the link (­between sadness and physical pain) is not just described via a linguistic metaphor; it is physically enacted and 262

Metaphoric Creativity in Social Interaction

performed, allowing for a zone of experience in which these connections are explored in a playful manner in the context of a learning environment.

Example 2: Grabbing the Stain4 In the first example, we saw how a conventional metaphoric expression was recontextualized in a creative way by physically enacting or performing the conventional metaphoric expression rather than just saying it. This implies that there is a creative potential in shifting from words to actions, from verbally describing something via a metaphor to physically enacting the metaphor in engaging in the learning activity. In the second example, the situation is not a learning activity with children but instead a therapy session with adults. Still, the same mechanism is at play, albeit in a slightly different manner, since this example concerns not only embodied emotional performance but also the actual physical actions of grabbing and holding an artifact. The extract below is from a therapy session between a client (­a single mother of a ­three-­​ ­­year-​­old daughter) and a therapist selected from a large cognitive, ethnographic study on psychotherapy.5 Just before the session, the client received a s­ o-​­called letter of concern from the local municipality, which questions her ability to take care of her daughter due to an incident that occurred a while ago in the day care. She is very upset about this and has brought the letter with her for the therapist to read it. In the therapy session, they discuss in detail the potential consequences of the letter. In the beginning of the session, the therapist addresses this problematic topic by referring to it as a “­stain”: “­you talk about that this issue becomes like a stain on you (.) so there is something here in which everything is not perfect”. Approximately five minutes later in the conversation, the therapist and client return to the topic, this time in relation to the client’s fear of the social exposure she might face due to the letter of concern becoming public knowledge: Extract 2 Time code: 00:13.­01 – ​­0 0:14:04 Danish Original C: Client, T: Therapist. 1 2 3 4 5 6 7 8 9 10 11 12 13 14

C: ja og jeg sad også bare og tænkte prøv at se sådan en (.) T: ja C: som NAVN (.) får han nu af vide (.) at nu har vi sådan en plet der på os T: mm (.) så den her plet må jeg se din plet ha ha (.) det denne her ikke også C: ja T: der er din plet (.) [må jeg] have lov C: [ ja] T: og kigge lige sådan (.) må jeg have lov og skimme det C: ja ja] T: hvad det er som ­du-​­fordi det det her med pletten du siger (.) på en eller anden måde så hører jeg dig forsvare (.) det du fortæller det er at (.) du bliver aldrig påvirket (.) ­over-​­i forhold til din datter (.) du øh gør altid det bedste for hende hun er altid det vigtigste (.) og øh du sætter dig selv til side for hende C: og det det jeg gør 263

Thomas Wiben Jensen

15 T: ja (.) øhm (.) og øhm (.) o ­ g-​­og den fortælling den forsvarer du (.) og det her det siger du det er en 16 plet på det det passer ikke sammen med dit billede af hvilken [du er] okay (.) 17 C: [nej] 18 C: det der det mit brev (.) det mine kommentarer 19 T: øhm okay (.) fint English translation 1 C: yeah (.) and I was just thinking look at someone like NAME [­ex-​­husband] (.) will he know about it 2 now (.) that we have such a stain on us 3 T: right mm (.) so this stain (.) can I see your stain he he (.) it’s this one right? 4 C: yes 5 T: this is your stain (.) [may I?] 6 C: [yeah] 7 T: look like (.) can I be allowed to skim it? 8 C: yeah yeah 9 T: it is like ­you-​­cause this about this stain (.) in some way I hear you defend (.) what you tell me is (.) 10 that you never get affected by (.) in relation to your daughter eh (.) you always do what 11 is best for her she is always the most important (.) that you always put yourself aside for her 12 C: and that is what I do 13 T: right (.) eh (.) and eh (.) and that narrative you defend (.) and this is why you say this is a stain on 14 it (.) it doesn’t fit with the image of what kind of mom [you are] right 15 C: [no] 16 C: this here is my letter (.) it’s with my comments 17 T: eh (.) okay fine The focus of the analysis is on the different uses of the stain metaphor, both verbal and in a more physical manner. “­Stain” in this example is used in a metaphorical way since it does not refer to a physical mark or stain but rather to the experience of the client in relation to her problematic situation. Generally, people may talk about traumatic, damaging, or socially unacceptable situations or experiences in their lives in terms of stains. We tend to think about and experience the impact of problematic events as if they had a physical character capable of leaving a stain on us. Within CMT, a large number of studies have documented expressions using words like “­stain”, “­spot”, “­taint”, or “­blemish” that can be seen as part of a larger cognitive schema relying on a dichotomy between clean and dirty objects or substances (­or experiences with these). Following from this, the claim is that we understand and experience unmoral or problematic and socially unacceptable behaviors in terms of (­interaction with) dirty or filthy objects, motivating the contrasting conceptual metaphors GOOD IS CLEAN, BAD IS DIRTY, leading to expressions like “­a clean/­d irty conscious”, a stained CV”, “­a tainted reputation”, “­a dirty mind”, “­a pure heart”, and so forth (­Lakoff and Johnson 1999; Deignan 2005; Kövecses 2005; Stefanowitsch 2011; Negro 2015; Yu et al. 2016; Gibbs 2017). However, an overlooked dimension in relation to the stain metaphor is the way in which it entails a causal or indexical relation between a problematic event and its social effects. A 264

Metaphoric Creativity in Social Interaction

stain points to the action that has caused it (­like a stained shirt points to the action of spilling coffee), which, in turn, makes it a powerful expression in dealing with issues of problematic behaviors, moral accountability, and social reputation ( ­Jensen, 2022). In this example, the use of the stain metaphor points to the social consequences for the patient and its implications on her ­self-​­perception. Stain as a metaphor is well suited in this context. In one word, it makes visible a relation between the problematic incident in the life of the client (­leading to the letter of concern) and the present social exposure. Thus, on a concrete level, the client fears that her e­ x-​­husband will hear about the letter of concern and perhaps use it against her in relation to rights for visitation, and these concerns are conveyed by a ­re-​­use of the stain metaphor (­previously introduced by the therapist), as we can see in lines ­1–​­2. Then, in line 3, a series of unexpected actions starts. First, the therapist refers to the expression of stain by using the definitive article “­this stain”, thereby accentuating both an awareness of the previous use (­a s mentioned above) and an increased focus ­here-­​­­and-​­now on the expression as something specific, something to be examined and valued. Then, after a mini pause, the therapist suddenly exclaims, “­can I see your stain?”, while making an open palm gesture toward a piece of paper on the table between them, that is, the letter of concern that the patient has brought with her (­see ­Figure 16.2a). The therapist now laughs briefly, which might also indicate a heightened awareness of this rather unusual use of the expression. Then she continues by asking, “­it is this one, right?”, while leaning over the table and touching the letter (­see ­Figure 16.2b). Further, after the patient has confirmed the therapist’s question in line 6, the therapist continues her new line of inquiry by asking, “­this is your stain, may I?”, while grabbing the letter in her hand (­see ­Figure 16.2c). These physical actions by the therapist are not only unconventional but also creative in different ways. They involve a move from a verbal metaphorical description to physically enacting a metonymic dimension of the stain metaphor, thereby recontextualizing it in a novel way. While metaphor is characterized by a domain mapping (­for instance, understanding the target domain of life in terms of the source domain’s physical journeys), metonymy is typically characterized by highlighting certain elements within the same domain. Metonymic expressions typically come about via a process in which one ­well-​­understood or easily perceived aspect of something stands for the thing or concept as a whole. Examples could be “­Number 10 remains silent” or “­Moscow denies any part in the recent aggressions”. In both cases, a physical location stands for the institution placed at that location (­i.e., the British and Russian government, respectively). This physical aspect of the domain matrix of “­government” is now highlighted as an easy way of referring to these rather complex institutions. In relation to the present example, the physical actions of the therapist are in effect a way of highlighting a certain aspect of “­stain” (­as a metaphor) in a metonymic fashion. This metonymic relation comes about in two different ways. First, the pointing and grabbing of the letter explicitly identified as “­the stain” can be seen as a ­part-­​­­for-​­whole metonymy in which the letter (­identified as the stain) stands for the larger domain matrix associated with stain in

­Figure 16.2 (­­a–​­c) Therapist and patient in a therapy session

265

Thomas Wiben Jensen

this example. The use of the stain metaphor in this conversation involves different aspects, such as the incident in the day care, the following damaged social reputation of the client (­in particular in relation to her ­ex-​­husband), the fear of losing custody of her child, as well as the letter from the municipality as its most concrete element. In this light, the referential use of the stain metaphor in Extract 2 can be seen as highlighting one particular element within the whole domain matrix of stain, that is the physical letter. The letter now comprises these different aspects of the domain matrix, and as, such “­stain” now functions as a ­part-­​­­for-​ ­whole metonymy based on the previously established metaphorical meaning. The new way of both talking about, as well as physically manipulating, the stain changes the meaning of the expression in a creative and unexpected way since the letter is now identified as a physical manifestation of the complex idea of stain. This creative domain highlighting leading to a new metonymic quality is illustrated in ­Figure 16.3. Second, this physical dimension of the metonymy affords a new type of action potential for the therapist. She can use the concrete artifact as a way to not just reinforce previously introduced metaphorical expressions but also for using the metonymy as a “­counseling resource” (­Tay 2017) that leads her to identify the letter as “­your stain” and thereby take it in her hands and look at it more closely. Importantly, there is a therapeutic benefit in compressing the multifaceted and somewhat unmanageable notion of “­your stain” into a concrete and much more manageable physical artifact. The metonymic use endows the stain with the physical affordances of a letter: you can hold it in your hand and examine it. Clearly, this does not solve the problematic situation of the client, but it may, for a brief while, make it easier for her to deal with. In relation to the task constraints of therapy, this seems as a highly appropriate, albeit unconventional, action that is also characterized by a certain element of emotional ease despite the severity of the topic (­e.g., the brief laughter of the therapist). Finally, it can be seen as an example of an enacted metonymy. It is in the physical actions of pointing and grabbing the letter while referring to it as “­your stain”, that the metonymic qualities emerge. The creative dimension arises via a shift from verbal metaphoric descriptions to physical actions as part of languaging behavior. Thus, the enacted metonymy in a creative fashion brings forth a new perspective on the already established reality.

Self-perception Social reputation

Letter of concern

Social authorities

Incident in the day care Losing custody of her child

­Figure 16.3 Stain as a metonymy based on a domain highlighting

266

Metaphoric Creativity in Social Interaction

Conclusion The two examples analyzed in the analytical section can be seen as manifestations of “­­creative-­​ ­­cognition-­​­­in-­​­­the-​­w ild”, to paraphrase Hutchins’ (­1995) famous book on distributed cognition. Creativity is often loosely defined as the ability to produce something that is both novel and useful. But the examples also documented that novelty is as much in the process as in the product. Put another way, in the examples analyzed, creativity is a situated practice in which process and product cannot be distinguished in any meaningful way. The metaphorical creativity is not just a case of producing novel linguistic expressions; it can also involve an embodied performance of an enacted metaphor or metonymy, which is unpredictable and arises spontaneously via couplings with the umwelt or ­actor-​­specific environment (­but also draws on elements from ­species-​­specific environment). In both cases analyzed, the creative metaphorical process was sparked by both physical artifacts in the environment (­i.e., the image of the sad girl in Example 1 and the letter of concern in Example 2) as well as the interactional dynamics with other participants. From an e­ cological-​­enactive perspective, it makes good sense to approach creativity as enacted through p­ erson–​­environment interactions and distributed among material artifacts and other people. As such, the creative niche construction (­u mwelt) is constituted via a process of couplings between persons and artifacts and between task constraints and emotional embodied experience. This process is the creative product. Finally, the examples pointed to the importance of c­ ross-​­modality in spontaneous creative processes. As already noted in the literature on metaphor and creativity, “­the great potential of metaphoric creativity resides precisely in the capacity of metaphor to enable the crossing of boundaries between modes and the senses” (­­Hidalgo-​­Downing 2020, ­p. 9). This has primarily been examined via examples of multimodal products involving words, images, and sometimes movements and sound as well (­El Refaie 2013; Domínguez 2020; Naciscione 2020; ­Perez-​­Sobrino and Littlemore 2020). This study contributes to this line of research by showing that metaphoric creativity can also be studied as embodied performances, not chiefly relying on words. Rather, the creative potential emerges from a transfer from using verbal metaphoric expression to enacting the metaphoric/­metonymic meanings via bodily actions. This way of physically doing metaphor adds another experiential dimension to the ways in which we not only understand something in terms of something else but also experience and feel one type of thing in terms of another. Author’s Note: This work was supported by the Velux Foundation (­Grant no. 10384). In relation to Example 2, I thank all employees and clients at the outpatient clinic for anxiety and personality disorders at Brønderslev Psychiatric Hospital, Denmark, for their willingness to participate in the project.

Notes 1 The second example in this article is selected from a large cognitive, ethnographic study conducted at a Danish Psychiatric Hospital. The dataset consists of ­v ideo-​­recordings of authentic therapeutic conversations between therapists and patients diagnosed with social anxiety disorders and/­or personality disorders. 2 Different analyses of this example have previously been published in Jensen and Cuffari (­2014) and Jensen (­2018). 3 Only Kay and Peter are visible on the drawings which are made on the basis of still images from a video recording. Unfortunately, the video camera was placed in a way that that did not visually catch all of the participants. 4 A different analysis based on this example figures in Jensen (­2022).

267

Thomas Wiben Jensen 5 The dataset consists of ­v ideo-​­recordings of therapeutic conversations between therapists and clients diagnosed with social anxiety disorders and/­or personality disorders. This study was reported to the South Danish Regional Committee on Health Research Ethics and the Legal office of the University in which the project is registered.

References Anderson, M. (­2014). After Phrenology: Neural Reuse and the Interactive Brain. Cambridge: MIT Press. Baggs, E., & Chemero, A. (­2018). Radical embodiment in two directions. Synthese, 198, ­1–​­16. Doi: 10.1007/­­s11229-­​­­018-­​­­02020-​­9 Bakhtin, Mikhail M. (­1981). The Dialogic Imagination: Four Essays. Austin: University of Texas Press. Bottineau, D. (­2010). Language and enaction. In J. Stewart, O. Gappene and E. A. Di Paolo (­Eds.), Enaction Toward a New Paradigm for Cognitive Science. Cambridge, MA: MIT Press, ­267–​­306. Cameron, L. J. (­2012). Metaphor and Reconciliation: The Discourse Dynamics of Empathy in ­Post-​­Conflict Conversations. Abingdon: Taylor & Francis. Cameron, L. J., Maslen, R., Todd, Z., Maule, J., Stratton, P. & Stanley, N. (­2009). The discourse dynamics approach to metaphor and ­metaphor-​­led discourse analysis. Metaphor and Symbol, 24(­2), ­63–​­89. Chemero, Anthony. (­2011). Radical Embodied Cognitive Science. Cambridge, MA: A Bradford Book. Clark, A. (­2011). Supersizing the Mind: Embodiment, Action, and Cognitive Extension. Oxford: Oxford University Press. Cowley, S. J. & ­Vallée-​­Tourangeau, F. (­Eds.) (­2013). Cognition beyond the Brain: Computation, Interactivity and Human Artifice. Berlin: Springer. Csikszentmihalyi, M. (­1996). Creativity: Flow and the Psychology of Discovery and Invention. New York: Harper/­Collins. Cuffari, E. & Jensen, T. W. (­2014). Living bodies: c­ o-​­enacting experience. In C. Müller, A. Cienki, E. Fricke, S. Ladewig, D. McNeill and J. Bressem (­Eds.), Handbook: ­B ody–­​­­L anguage–​­Communication. Vol. 2. Berlin: De Gruyter Mouton, ­2016–​­2025. Damasio, A. (­1999). The Feeling of What Happens: Body and Emotion in the Making of Consciousness. New York: Hartcourt Brace. Deignan, A. (­2005). Metaphor and Corpus Linguistics. Amsterdam: John Benjamins Publishing Company. Di Paolo, E. A., Cuffari, E. C. & De Jaegher, H. (­2018). Linguistic Bodies. The Continuity between Life and Language. Cambridge, MA: The MIT Press. Domínguez, M. (­2020). Disentangling metaphoric communication: The origin, evolution and extinction of metaphors. In L. ­H ildalgo-​­Downing and B. Mujic Kraljevic (­Eds.), Performing Metaphoric Creativity across Modes and Contexts. Amsterdam/­Philadelphia: John Benjamins (­Figurative Thought and Language 7), ­175–​­195. El Refaie, E. (­2013). ­Cross-​­modal resonances in creative multimodal metaphors: Breaking out of conceptual prisons. Review of Cognitive Linguistics, 11(­2), ­236–​­249. DOI: 10.1075/­rcl.11.2.02elr Fuchs, T. (­2017). Ecology of the Brain: The Phenomenology and Biology of the Embodied Mind. Oxford: Oxford University Press. Gallagher, S. (­2017). Enactivist Interventions: Rethinking the Mind. Oxford: Oxford University Press. Gallagher, S. & Lindgren, R. (­2015). Enactive metaphors: Learning through ­full-​­body engagement. Educational Psychology Review, 27(­3), ­391–​­404. Gibbs, R. W. (­2005). Embodiment and Cognitive Science. Cambridge: Cambridge University Press. Gibbs, R. W. (­2017). Metaphor Wars. Cambridge: Cambridge University Press. Gibbs R. W. (­2019). Metaphor as ­dynamical–​­ecological performance. Metaphor and Symbol, 34(­1), ­33–​ ­4 4. DOI: 10.1080/­10926488.2019.1591713 Gibson, J. J. (­1977). The Ecological Approach to Visual Perception. Hilldale, NJ: Erlbaum. Goodwin. C. (­2017). ­Co-​­operative Action (­L earning in Doing: Social, Cognitive and Computational Perspectives). Cambridge: Cambridge University Press. ­H ildalgo-​­Downing, L. (­2020). Introduction: Towards an integrated framework for the analysis of metaphor and creativity in discourse. In L. ­H ildalgo-​­Downing and B. Mujic Kraljevic (­Eds.), Performing Metaphoric Creativity across Modes and Contexts. Amsterdam/­Philadelphia: John Benjamins (­Figurative Thought and Language 7), ­1–​­17. Hutchins, E. (­1995). Cognition in the Wild. Cambridge, MA: MIT Press.

268

Metaphoric Creativity in Social Interaction Jensen, T. W. (­2017). Doing metaphor: An ecological perspective on metaphoricity on discourse. In B. Hampe (­Ed.), Metaphor: Embodied Cognition and Discourse (­­pp. ­257–​­276). Cambridge: Cambridge University Press. Jensen, T. W. (­2018). The world between us: The social affordances of metaphor in ­f ace-­​­­to-​­face interaction. Rask. International Journal of Language and Communication, 46, ­45–​­76. Jensen, T. W. (­2022). “­We have such a stain on us”: The indexical affordances of the stain metaphor. Metaphor and Symbol, 37(­3), ­208–​­228. Jensen, T. W., & Cuffari, E. (­2014). Doubleness in experience: Toward a distributed enactive approach to metaphoricity. Metaphor and Symbol, 29(­4), ­278–​­297. DOI: 10.1080/­10926488.2014.948798 Jensen, T. W.  & Greve, L. (­2019). Ecological cognition and metaphor. Metaphor and Symbol, 34(­1), ­1–​­16. DOI: 10.1080/­10926488.2019.1591720 Jensen, T. W. & Pedersen, S. B. (­2016). Affect and affordances: The role of action and emotion in social interaction. Cognitive Semiotics, 9(­1), ­97–​­103. Johnson, M. (­1987). The Body in the Mind: The Bodily Basis of Meaning, Imagination, and Reason. Chicago: University of Chicago Press. Johnson, M. (­2007). The Meaning of the Body: Aesthetics of Human Understanding. Chicago, IL: University of Chicago Press. Kövecses, Z. (­2000). The scope of metaphor. In A. Barcelona (­Ed.), Metaphor and Metonymy at the Crossroads (­­pp. ­79–​­92). Berlin: Mouton de Gruyter. Kövecses, Z. (­2005). Metaphor in Culture: Universality and Variation. New York: Cambridge University Press. Lakoff, G. (­1993). The contemporary theory of metaphor. In A. Ortony (­Ed.), Metaphor and Thought (­­pp. ­202–​­251). Cambridge: Cambridge University Press. Lakoff, G. & Johnson, M. (­1980). Metaphors We Live By. Chicago: University of Chicago Press. Lakoff, G. & Johnson, M. (­1999). Philosophy in the Flesh: The Embodied Mind and Its Challenge to Western Thought. New York: Basic Books. LeDoux, J. (­1996). The Emotional Brain. New York: Simon and Schuster. Malinin, L. H. (­2019). How radical is embodied creativity? Implications of 4E approaches for creativity research and teaching. Frontiers in Psychology, 10, 2372. DOI: 10.3389/­f psyg.2019.02372 Maturana, H. R. (­1978). Biology of language: The epistemology of reality. In G. A. Miller,  & E. Lenneberg (­Eds.), Psychology and Biology of Language and Thought: Essays in Honor of Eric Lenneberg (­­pp. ­27–​­63). New York: Academic Press. Müller, C. (­2008). Metaphors Dead and Alive, Sleeping and Waking: A Dynamic View. Chicago, IL: University of Chicago Press. 2019). Metaphorizing as embodied interactivity: What gesturing and film viewMüller, C. (­ ing can tell us about an ecological view on metaphor. Metaphor and Symbol, 34(­1), ­61–​­79. DOI: 10.1080/­10926488.2019.1591723 Müller, C. & Tag, S. (­2010). The dynamics of metaphor: Foregrounding and activationg metaphoricity in conversational interaction. Cognitive Semiotics, 6(­S1), ­85–​­120. Naciscione, A. (­2020). Multimodal creativity in figurative use.  In L. ­H idalgo-​­Downing,  & B. K. Mujic (­Eds.), Performing Metaphoric Creativity across Modes and Contexts (­­pp. ­249–​­280). Amsterdam: John Benjamins Publishing Company. Negro, I. (­2015). Corruption is dirt: Metaphors for political corruption in the Spanish press. Bulletin of Hispanic Studies, 92, ­213–​­237. Neumann, M. & Cowley, S. J. (­2013). Human agency and the resources of reason. In S. J. Cowley and F. ­Vallée-​­Tourangeau (­Eds.), Cognition beyond the Brain: Computation, Interactivity and Human Artifice. Berlin: Springer, ­13–​­30. Norris, S. (­2004). Analyzing Multimodal Interaction: A Methodological Framework. New York: Taylor & Francis. ­Pérez-​­Sobrino, P. & Littelmore, J. (­2020). What makes an advert go viral? The role of figurative op­ ildalgo-​­Downing and B. Mujic Kraljevic (­Eds.), erations in the success of Internet videos. In L. H Performing Metaphoric Creativity across Modes and Contexts. Amsterdam/­Philadelphia: John Benjamins (­Figurative Thought and Language 7), ­119–​­152. Runco, M. A. (­2004). Creativity. Annual Review of Psychology, 55, ­657–​­687. Slaby, J., Paskaleva, A. & Stephan, A. (­2013). Enactive emotion and impaired agency in depression. Journal of Consciousness Studies, 20(­­7–​­8), ­33–​­55.

269

Thomas Wiben Jensen Stefanowitsch, A. (­2011). Cognitive linguistics as a cognitive science. In M. Callies, W. R. Keller and A. Lohöfer (­Eds.), ­Bi-​­directionality in the Cognitive Sciences: Avenues, Challenges, and Limitations (­­pp. ­295–​­310). Amsterdam: John Benjamins. Sternberg, R. J. & Lubart, T. I. (­1999). The concept of creativity: Prospects and paradigms. In R. J. Sternberg (­Ed.), Handbook of Creativity. Cambridge: Cambridge University Press, ­3 –​­15. Stewart, J. Gapenne, O. & Di Paolo, E. A. (­Eds.) (­2010). Enaction. Toward a New Paradigm for Cognitive Science. Cambridge: MIT Press. Szokolszky, A. (­2019). Perceiving metaphors: An approach from developmental ecological psychology. Metaphor and Symbol, 34(­1), ­17–​­32. DOI: 10.1080/­10926488.2019.1591724 Tay, D. (­2017). An analysis of metaphor hedging in psychotherapeutic talk. In M. Yamaguchi, D. Tay and B. Blount (­Eds.), Approaches to Language, Culture, and Cognition. (­­pp. ­251–​­226). New York: Palgrave Macmillan. DOI: 10.1057/­9781137274823_11 Thibault, P. J. (­2011). ­First-​­order languaging dynamics and ­second-​­order language: The distributed language view. Ecological Psychology, 23, ­210–​­245. Varela, F. J., Rosch, E. & Thompson, E. (­1991). The Embodied Mind. Cognitive Science and Human Experience. Cambridge: MIT Press. Veale, T., Feyaerts, K. & Forceville, C. (­Eds.) (­2013). Creativity and the Agile Mind. Berlin: Mouton de Gruyter. Yu, N., Wang, T. & He, Y. (­2016). Spatial subsystem of moral metaphors: A cognitive semantic study. Metaphor and Symbol, 31(­4), 1­ 95–​­211. DOI: 10.1080/­10926488.2016.1223470

270

17 ANALYZING CHANGING PATTERNS OF EXTERNAL REFERENCE USE FROM INFORMAL LAB GROUP PRESENTATIONS TO FORMAL COLLOQUIA Christian D. Schunn, Lelyn D. Saner and Susannah B. F. Paletz

Introduction Scientific discovery and innovation involve a long continuum from relatively social processes (­e.g., peer review, u ­ ser-​­testing, lab groups, and collaborations) to more solitary processes (­e.g., reading, writing, incubation, evaluating, and deciding; Crowley  & Azmitia, 2001). Cognitive science investigations have generally tended to emphasize solitary processes (­e.g., Chan et al., 2011; Chen & Klahr, 1999; Chinn & Malhotra, 2002; Dunbar, 1993; Gentner, 2002; Huang et al., 2017; Klahr & Dunbar, 1993). But there has also been important cognitive science work on dyads (­e.g., Okada & Simon, 1997), as well as lab groups and teams (­e.g., Christensen & Schunn, 2007; Dunbar, 1997, 2000; Paletz, Chan, & Schunn, 2017; Saner & Schunn, 1999). A commonly examined issue, with implications across the continuum, is the extent to which novel ideation (­e.g., insight and discovery) results directly from referring to and then building upon others’ ideas, particularly from domains far from the current area of investigation. Such external references may be very simple and mundane, such as simply citing the original source for a statement or quotation or providing a concrete example that instantiates an abstract idea. But they can also be more cognitively complex, such as describing how another theory might explain a result or drawing an analogical comparison to explore similarities and differences. Specifically, the use of analogies for cognitive processing has been investigated extensively across the cognitive science disciplines (­e.g., Boicho  & Kokinov, 2001; Gentner, 2002; Holyoak & Thagard, 1989; Vendetti et al., 2015). Because each pair of selected analogs produces a unique set of mappings, which may be unpacked and useful to different degrees under different circumstances, analogies can be very fruitful for generating new ideas to consider. But external references, as a broader construct, can play a role in science beyond just being part of analogical reasoning, for example by priming certain ideas

DOI: 10.4324/9781003009351-19

271

Christian D. Schunn et al.

(­K aplan & Simon, 1990; Schunn & Dunbar, 1996), helping explain ideas (­Dunbar, 1993), or providing support for claims. Importantly, external references can be useful across the continuum of relatively solitary to relatively social processes, such that the analytic focus can be on the individual level (­Chan et al., 2011; Christensen & Schunn, 2005; Gentner, 2002) or the group level (­Chan, Dow, & Schunn, 2015; Paletz et al., 2013). One controversial question within both analogies and external references is what effect topical similarity or associative closeness has on one’s ability to draw new ideas from a referenced ­source – ​­that is, the topic distance between what is referred to and the matter at hand (­e.g., within Kepler’s famous analogy, atoms and solar systems are considered very different scientific topics). We avoid the complex issue of whether there is or is not a deep structural similarity between the ideas, since, in many situations, too little is explicitly said about each idea to make judgments about that kind of similarity. Further, the debate in the field (­a nd the focus of our study) is about topic ­d istance – ​­do more creative ideas come from closely related source domains or from source domains that are quite different from the current discovery or innovation subject matter? Sometimes sources further away are found to be best (­Chan et al., 2011; Chan & Schunn, 2015a; Gentner, 2002), and sometimes closer sources are found to be best (­Chan et al., 2015; Dunbar 1997; Fu et al., 2013). Embedded within the continuum of relatively solitary to relatively social activities is the correlated dimension of time. In past research, when the focus was on short time scales (­e.g., quick judgments and inspirations within seconds to minutes), the analytic lens that researchers applied tended to be on individuals (­e.g., Dunbar & Schunn, 1996; Klahr & Dunbar, 1988; Mynatt, Doherty, & Tweney, 1977). However, when the focus was on long time scales (­e.g., major scientific discoveries of an era; final outputs of a ­real-​­world engineering process), the analytic lens included the social influences of collaborators, mentors, and competitors (­e.g., Gentner, Brem, Ferguson, & Wolff, 1997; Tweney, 1985). There are reasons to doubt the external validity of lab studies focusing on short time scales (­e.g., the artificial and ­k nowledge-​­lean nature of the tasks; Chinn & Malholtra, 2002). Moreover, any analysis at a short time scale may miss the ­longer-​­term and large cumulative effects of a knowledge source, given that iterative ­problem-​­solving may eventually produce a stronger outcome (­Chan & Schunn, 2015b). At the same time, any analysis at longer time scales (­e.g., Gentner et al., 1999) may rely on incomplete data since people sometimes forget sources of inspiration (­K aplan & Simon, 1990). One solution is to obtain data at multiple time points along a discovery process such as data from both early and later presentations. In addition, social contexts involve a mix of uses for external references, including cognitive functions (­e.g., argument building and ­decision-​­making) and communicative functions (­e.g., building common ground and persuasion), and it may be important to tease apart those different kinds of functions of inspirational sources. For example, did a team member refer to something just to help another team member understand their idea, or was the reference “­the source” of their idea (­Christensen & Schunn, 2007; Dunbar, 1997)? As the time scale grows, the communicative roles of references may become increasingly dominant in the discourse when, for instance, the group or team attempts to persuade outsiders about the importance or value of their idea or replace inductive analogical reasoning with additional deductive evidence that was acquired gradually to support the earlier inductive reasoning. Historical analyses of scientific discovery and innovation must therefore be careful in drawing strong inferences about the function of external references found in retrospective accounts. Indeed, this caution is a primary motivator of the current study; we seek to understand whether later presentations of a scientific line of work may misrepresent the role of distant knowledge sources in the discovery process. 272

Analyzing Changing Patterns of External Reference Use

In this chapter, we report on an exploratory investigation of the use of external sources within scientific reasoning through a contrast between external references that occur in lab group meetings (­earlier, m ­ eaning-​­making stages of a project) and those that occur in colloquia presentations (­later, ­persuasion-​­focused stages of a project). Comparisons of a­t-­​ ­­the-​­bench ­problem-​­solving (­from early work) to the contents of written documents and presentations (­from later stages) inherently involve many confounding differences. By comparison, there is some general similarity in the overall structure of these two kinds of events that makes for a cleaner contrast: a spoken format following a roughly similar structure focused on a line of research work. Of particular interest to our research questions, colloquia are expected to have significant polish and include a stronger persuasion function. In this way, we can see whether the later, more formal presentations include kinds of external references that may not be as common during the discovery process itself. In addition, lab group presentations involve a range of completeness of the research, from descriptions of work about to be conducted to a set of studies that have been conducted and fully analyzed. Thus, it is possible to verify within lab group presentations whether certain kinds of external references “­accumulate” or “­d isappear” as the work becomes more mature. We do acknowledge that the two event types also differ in other important ways. Colloquia tend to involve very successful researchers describing relatively successful lines of research, whereas ­lab-​­group presentations involve a mixture of junior and senior researchers describing lines of work that may not prove to be successful. For this reason, we focused on two lab groups at a ­research-​­intensive university that were very successful and included research that was later featured in external colloquia. We focus on one major research question: What is the relative role of external references that are close to the target domain versus external references that are from other scientific domains or everyday (­­non-​­scientific) domains within lab group presentations and colloquia? Following Dunbar’s claims about the greater functional value of near analogies in science as well as Christensen and Schunn’s (­1997) work on functions of near and far analogies in engineering design, external references to distant domains may tend to serve a communicative function, whereas external references to the same domain may tend to serve a ­problem-​ ­solving function. Thus, we expect more distant external references in colloquia (­especially in the introduction and literature review phases) and more n ­ ear-​­topic external references in lab group presentations (­especially within the ­sense-​­making of results). Further, we expected the main differences in frequency and types of external references between lab groups and colloquia to be carried by the presenters rather than appearing in the audience members’ contributions. Critically, similarity in audience members’ use of external references across the two contexts would serve as a useful sanity check that the topics of research presented in the lab groups were not inherently more or less ripe for external references than were the topics of research presented in the colloquia. On a more exploratory basis, we also considered other features of external references that might account for different proportions of l­ong-​­and ­short-​­distance references and different use goals, such as the types of things specifically referenced and the degree to which the intended meanings were unpacked.

Methods To answer these questions, we examined a series of audio recordings of the proceedings of two psychology lab groups and one colloquium series, all from within one psychology department. The psychology department was part of a major university in a large city in the 273

Christian D. Schunn et al.

US Midwest. At the time of the recordings, the department had over 100 faculty, graduate students, ­post-​­docs, and research assistants; except for the research assistants, most individuals were from other cities originally and had obtained prior degrees from other universities. Each lab group presentation involved one researcher presenting research that was about to be conducted or had just been conducted but was not yet fully analyzed. The colloquium series presenters were from around the US and had been invited as speakers because of the established excellence of their work.

Participants The sample set included recordings from 15 e­ vents  – ​­eight scientific colloquia and seven research lab group meetings (­three from one group and four from another). All the participants were aware that the sessions were being recorded, but they were not aware of the goals of this study. One of the study authors was a member of the community and attended both colloquia and lab group sessions. The colloquia were composed of the primary speaker and 3­ 0-­​­­to-​­60 audience members, a mix of faculty, ­post-​­docs, graduate students, and research assistants. Three of the eight colloquium speakers were senior faculty members at the university where the event was recorded, which helped rule out the possibility that the researchers at this university had a different style in the use of external references. The other five speakers were ­well-​­known, senior researchers invited to present from other universities. The colloquia ranged from 70 to 90 minutes in length; topics included social psychology, cognitive psychology, cognitive neuropsychology, and cognitive development. With respect to the lab groups, one was a developmental psychology group with regular attendance by two faculty, three ­post-​­docs, three graduate students, and two research assistants. The other was a cognitive psychology lab group with regular attendance by three faculty, six ­post-​­docs, five graduate students, and five research assistants. The exact number of participants varied from meeting to meeting. These lab group meetings were also approximately ­70–​­90 minutes long. The lab groups were selected because they were supervised by ­well-​­established researchers whose level of eminence within their fields was comparable to those of the colloquium speakers.

Data Preparation and Coding We fully transcribed all event session recordings, including the questions and comments made by audience members, producing a total data corpus of 182,599 words. Since the colloquia and lab groups were roughly similar in duration, the transcript lengths were also very similar across lab groups (­M=11,908 words per session) and colloquia (­M=12,405 words per session). We then segmented the transcripts by changes of speaker and highlighted full sentences within speaker turns where proper names, concrete objects, and acronyms were mentioned. External references were fundamentally references to anything that was “­external” to the research being presented: not the theory being tested, not the researchers or participants, not the stimuli or instruments or analysis tools, and not the findings of the study. External references could be to other research teams, other studies, other findings, other phenomena, or everyday events and objects. To identify external references within the discussions, the second author conducted a first pass. Direct ­self-​­references (­e.g., “­I then did…”) were not counted unless the reference was to prior published work with which the speaker was 274

Analyzing Changing Patterns of External Reference Use

directly involved (­e.g., “­In a previous experiment I found that…”). All of the authors reviewed the highlighted (­full sentence) segments and came to consensus through discussion on which would be included in the final data set. To code the external references on the dimensions of interest, the second and third authors ­double-​­coded a subset of the data to establish ­inter-​­rater reliability. Two subsets were selected ­pseudo-​­randomly to ensure that some references from every transcript were included in the calibration process. With both sets ­pre-​­selected, we coded the first subset independently, compared codes, and resolved disagreements through discussion. We then repeated the same procedure with the second subset to assess improvement in reliability and verify that agreement was at criterion. At both stages, the speaker’s entire statements were included in the test data set to provide context for making judgments. Overall, there were 424 references across all transcripts, 108 of which were repeats of unique references made earlier in the same discussions, leaving a total of 316 unique references. Approximately 25% of these unique references were d­ ouble-​­coded using this procedure to assess reliability, the criterion for which was set at a minimum Cohen’s Kappa (­κ) value of 0.50. Statements about the general topics of each presentation, whether made by those introducing colloquia speakers or those leading lab group meetings, were also marked in each transcript to be excluded from the main analysis but used as documentation for coding the relative distance of external references.

Context Features External references were coded for both the content of the reference itself and the local context (­w ithin the transcript). There were four context features. Event type. The main context variable is the event type in which the external reference was found (­i.e., lab group vs. colloquium). Speaker type. We also distinguished whether the external reference was made by the main presenter or by one of the audience members. Stage of research. Each lab group discussion was coded for the general stage of research being presented. If the main presenter was presenting the design of a new study or analyses run on data from a pilot study, the session was coded as pertaining to early stage research. If the topic was the description and results of a full study, or an overview of several studies, it was coded as ­late-​­stage research. Because colloquia were focused on late stage research, this distinction was only applied to the lab group discussions. Phase of presentation. Transcripts were chunked according to their general phase of presentation. Comments related to specifying the topic and scope, framing the key questions and motivation, and addressing any prior work related to the current effort were labeled as introduction. Methods included any statements about the specific methodologies and procedures used (­or to be used) and any ideas about their merits and disadvantages. We associated dialogue related to the outcomes of hypothesis tests and observations about how the findings might answer the research questions with the results phase, and we attributed any dialogue about the implications and contributions of the research with the discussion phase. Although the typical order of these phases in published work is as we have described them ­here – ​­introduction, methods, results, and then discussion – ​­the dynamic nature of the conversations during these r­eal-​­time events meant that segments of any category could occur at any time. For example, some presentations involved the description of several studies, such that speakers would cycle through the phases. In addition, comments and questions from the audience could pertain to any of these phases, regardless of where the speaker was in the presentation. 275

Christian D. Schunn et al.

Content Features There were three content features coded for each external reference (­a nd we acknowledge that the presentation phase is potentially about both content and context). In the examples given below, the primary referent is underlined, and enough surrounding text is included to illustrate the distinction being made. Reference distance. The first f­eature – ​­reference distance – ​­was adapted from Dunbar’s (­1996) coding scheme applied to biology labs to fit the current domain of psychology. As a first step, statements made by the primary presenters, or those introducing them, about the presentation topic were identified and marked in each protocol. These statements were used as anchors to determine the distance of the external references. Three levels of distance were initially coded: if a reference was directly related to the topic or to the main speaker’s focal domain of expertise within their general discipline, it was coded as ­within-​­domain (­e.g., “…we start with a quote from the King of the Mountain, ah (­laughs) which is, ah, Baddeley. Ah, this is his view, ah, of what is involved in doing memory span”). In contrast, if a reference was made to some other s­ ub-​­area of the main speaker’s general discipline (­i.e., their larger domain of expertise) or to some other academic discipline, the reference was coded as being ­between-​­domain (­e.g., during a cognitive psychology talk, “­Now the psychological approach sort of goes back to Freud,”). Finally, when speakers made ­ on-​­scientists alike, could experia reference to things that people in general, scientists and n ence in everyday life, it was coded as everyday (­e.g., “­I mean think of talking as just a bunch of cars driving really fast down the freeway…”). ­Within-​­domain references were most common (­proportion=0.71, n=225), followed by ­between-​­domain (­0.10, n=32), and everyday (­0.19, n=59). Because of the low frequency of ­between-​­domain references, we combined the b­ etween-​­domain and everyday reference counts into the category of distant references. In conceptual support of this category collapse, we note that analyses of analogical reasoning have often used a binary near/­far distinction. Further, prior work on historical cases of scientific discovery has treated both ­between-​ d­ omain and everyday as a form of “­d istant” analogy. For example, the snake swallowing itself as the source of Kekule’s benzene ring discovery and the Kepler analogy between the solar system and the structure of an atom are both canonical examples of distant analogies. Reference use. We also coded references for how they were used to convey information to others involved in the discussion. The reference use dimension focused on the degree to which references were elaborated upon by those who made them. In some cases, a reference was simply made or identified, and these were coded as mentioned (­e.g., “­Well, what about the ‘­­remember-​­know’ paradigm that’s gotten so much press lately?”). In other cases, a reference might be explicitly defined, evaluated, or compared with another concept being d­ iscussed –​ ­i.e., elaborated. For example: So this is... might be viewed as an alternative to a recently dominant, s­ emi-​­dominant position in the field which involves posing or positing multiple memory systems that operate with different rules, in different places in the brain and so on and so on. Reference type. Finally, we coded reference type, which was whether the reference was specifically to the work of another researcher (­e.g., “­Sternberg has done a lot of work on it…”) or to an idea, item, or object (­e.g., “­OK, another hierarchical model is the competitive cue model…” or “­Now, one additional thing that really is a promissory note is the time course information”). This variable served as a control variable but was not of primary interest. 276

Analyzing Changing Patterns of External Reference Use

Researchers are expected to make brief connections to the work of other researchers in presentations, particularly in external presentations such as colloquia.

Coding Reliability As is often found with in vivo research, the contents and use of the references were challenging to code, but all codes met acceptable levels of reliability. There was 83% agreement (­κ = 0.57) for reference distance, 87% agreement for reference use (­κ = 0.63), and 83% agreement for reference type, (­κ = 0.67). The reliability of event type, speaker type, stage of research, and phase of presentation was established by reaching consensus through discussion among the authors.

Analyses To measure the base and relative rates of referencing, we calculated the frequencies of references per thousand words of discourse and used ­repeated-​­measures Analysis of Variance (­A NOVA) to assess differences among contextual conditions. A different analytic approach was needed to examine the effects of content and their interactions with context. Because nearly all of the data in this study were categorical, we used ­log-​­linear analysis with backward elimination to identify which variables and interactions had significant effects on the overall distributions of different reference categories. To examine specific effects, we ran t­hree-​­way ­log-​­linear analyses to generate G2 values, which are functionally interpretable as χ2 values. We will report odds ratios and η2 as measures of effect size.

Results General Rates of External Referencing: More Common in Colloquia Overall, colloquia participants averaged 2.08 (­SD = 0.69) references per thousand words, and participants in lab group discussions made an average of 1.35 (­SD = 0.44) references for every thousand words F(­1,13) = 6.14, p = 0.028, η2 = 0.91. External references occurred during all phases of presentation, although they were particularly common in the introduction, and lab groups very rarely had external references in their general discussions.

Model Optimization for Categories of References To identify the simplest model that best represents the patterns of influence across potential key features, we ran a saturated model with full interactions consisting of event type, speaker, reference distance, reference use, and reference type. After systematically eliminating variables, starting with the highest order interactions, we assessed their effects on the overall model in terms of changes in likelihood ratio ­chi-​­squares. Lower ­chi-​­square values correspond to higher model fit, and after the elimination procedure was completed, the reduced model had an acceptable ­fit – ​­χ2 (­10) = 5.58, p=0.85, Cramer’s V=0.04. We examined the influence of different effects on the overall model in terms of the c­ hi-​­square change resulting from their elimination. Neither the saturated model, Δχ2 (­1) = 0, p = 0.99, nor the ­four-​­way models, Δχ2 (­5) = 3.73, p = 0.59, had significant impacts on the fit of the model when eliminated. Only main effects, Δχ2 (­5) = 311.48, p 0, it suggests that the blue participant is driving the coordination, and when DI < 0, the red participant is driving the coordination. Thus, the figures below show that while there is some variability in the DI estimates per condition, the blue and red conditions show the expected pattern where the highest densities are greater than zero or less than zero, respectively. And additionally, in the both condition, the DI densities are at or very close to zero, indicating a general tendency for participants to follow the goals of each condition. Note that this pattern holds for both novices and experts. For our analyses, we ran a mixed model with DI as the outcome variable with condition (blue leads, red leads, or both), and expertise level (­novice or expert) as dummy coded fixed effects with random intercepts for trial number nested within game (­to account for repeated observations of the same dyad within a given mirror game). Note that for Group, experts are the referent group, and for Condition, blue leading is the referent category. We also ran a model that allowed for interactions between the two fixed effects. The BIC and likelihood ratio suggest the model that included the interaction effect was a better fit than the model without the interaction (­see ­Table 35.2), although the change in marginal R 2 was approximately 3%. 632

Collaborative Creativity

­Figure 35.3 Novice and expert transfer entropy analyses for the mirror game

­Table 35.2  Comparison of Models Directionality Index for Mirror Game Data Model

Df

AIC

BIC

logLik

Deviance

χ2

χ 2df

p

Fixed effects only Fixed effects with interaction

6 8

−1248.5 −1253.5

−1229.7 –1228.5

630.28 634.78

−1260.5 −1269.5

9.00

2

0.01

­Table 35.3  Mirror Game Directionality Index Model Output Effect

Estimate

Std. Error

t

p Bootstrapped 95% CIs

Intercept Group (­novice) Condition ( ­both) Condition (red) Group (novice) × Condition (both) Group (novice) × Condition (red)

0.007 0.002 −0.007 −0.012 0.0004

0.001 0.001 0.001 0.002 0.002

5.812