Situated Cognition Research: Methodological Foundations (Studies in Brain and Mind, 23) [1st ed. 2023] 3031397436, 9783031397431

Citation preview

Studies in Brain and Mind  23

Mark-Oliver Casper Giuseppe Flavio Artese   Editors

Situated Cognition Research Methodological Foundations

Studies in Brain and Mind Volume 23

Series Editor Gualtiero Piccinini, University of Missouri - St. Louis, St. Louis, MO, USA Editorial Board Members Berit Brogaard, University of Oslo, Norway, University of Miami, Coral Gables, FL, USA Carl Craver, Washington University, St. Louis, MO, USA Edouard Machery, University of Pittsburgh, Pittsburgh, PA, USA Oron Shagrir, The Hebrew University of Jerusalem, Jerusalem, Israel Mark Sprevak, University of Edinburgh, Scotland, UK

The series Studies in Brain and Mind provides a forum for philosophers and neuroscientists to discuss theoretical, foundational, methodological, and ethical aspects of neuroscience. It covers the following areas: •  Philosophy of Mind •  Philosophy of Neuroscience •  Philosophy of Psychology •  Philosophy of Psychiatry and Psychopathology • Neurophilosophy • Neuroethics The series aims for a high level of clarity, rigor, novelty, and scientific competence. Book proposals and complete manuscripts of 200 or more pages are welcome. Original monographs will be peer reviewed. Edited volumes and conference proceedings will be considered provided that the chapters are individually refereed. This book series is indexed in SCOPUS. Initial proposals can be sent to the Editor-in-Chief, prof. Gualtiero Piccinini, at [email protected]. Proposals should include: •  A short synopsis of the work or the introduction chapter •  The proposed Table of Contents •  The CV of the lead author(s) •  If available: one sample chapter We aim to make a first decision within 1 month of submission. In case of a positive first decision the work will be provisionally contracted: the final decision about publication will depend upon the result of the anonymous peer review of the complete manuscript. We aim to have the complete work peer-reviewed within 3 months of submission. For more information, please contact the Series Editor at [email protected].

Mark-Oliver Casper  •  Giuseppe Flavio Artese Editors

Situated Cognition Research Methodological Foundations

Editors Mark-Oliver Casper Department of Philosophy University of Kassel Kassel, Germany

Giuseppe Flavio Artese Department of Philosophy University of Kassel Kassel, Germany

ISSN 1573-4536     ISSN 2468-399X (electronic) Studies in Brain and Mind ISBN 978-3-031-39743-1    ISBN 978-3-031-39744-8 (eBook) https://doi.org/10.1007/978-3-031-39744-8 © The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Paper in this product is recyclable.

Contents

1

Introduction����������������������������������������������������������������������������������������������    1 Mark-Oliver Casper and Giuseppe Flavio Artese

Part I Philosophy of Science 2

 Methodological Problem of Choice for 4E Research������������������������   17 A Mark-Oliver Casper

3

A Professional Guide to Explanation. Commentary on “A Methodological Problem of Choice for 4E Research” ��������������   45 Guilherme Sanches de Oliveira

4

Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates Pluralism��������������������������������������������������������������   51 Guilherme Sanches de Oliveira

5

Commentary on “Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates Pluralism” ��������������������������   77 Mark-Oliver Casper

6

Ecological Psychology, Enaction, and the Quest for an Embodied and Situated Cognitive Science ��������������������������������   83 Manuel Heras-Escribano

7

Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward����������������������������������������������������������  103 Pierre Steiner

Part II Ontology of the Mind 8

Extended Cognition and the Search for the Mark of Constitution – A Promising Strategy? ����������������������������������������������  129 Beate Krickel

v

vi

Contents

9

Commentary on “Extended Cognition and the Search for the Mark of Constitution – A Promising Strategy?”����������������������  147 Julian Kiverstein and Michael Kirchhoff

10 Dissolving  the Causal-Constitution Fallacy: Diachronic Constitution and the Metaphysics of Extended Cognition������������������  155 Julian Kiverstein and Michael Kirchhoff 11 Why  Diachronic Constitution Won’t Help. Commentary on “Dissolving the Causal-­Constitution Fallacy”����������������������������������  175 Beate Krickel 12 Predictive  Processing and Extended Consciousness: Why the Machinery of Consciousness Is (Probably) Still in the Head and the DEUTS Argument Won’t Let It Leak Outside����������������������������������������������������������������������������������������  181 Marco Facchin and Niccolò Negro 13 Commentary  on “Predictive Processing and Extended Consciousness” ����������������������������������������������������������������������������������������  209 Julian Kiverstein and Michael Kirchhoff 14 Plural  Methods for Plural Ontologies: A Case Study from the Life Sciences������������������������������������������������������������������������������  217 Luis H. Favela and Anthony Chemero Part III Applications 15 Chronic  Pain, Enactivism, & the Challenges of Integration����������������  241 Sabrina Coninx and Peter Stilwell 16 Affordances,  the Social Environment, and the Notion of Field: State of the Debate and Methodological Insights������������������  277 Giuseppe Flavio Artese Index������������������������������������������������������������������������������������������������������������������  307

Chapter 1

Introduction Mark-Oliver Casper and Giuseppe Flavio Artese

Abstract  In the last three decades, an interdisciplinary research program - called “the 4E approaches” - highlighted the necessity of rethinking the theoretical assumptions and experimental practices of mainstream cognitive science. However, the claim of the disruptive potential of 4E research grows old without seeing the announced and wanted effects on how cognitive phenomena are studied. This introduction to the edited volume “Situated Cognition Research: Methodological Foundations” focuses on the method(olog)ical issues of the 4Es, the loosely connected but promising methodical options available in the debate, and the foundational questions that need to be answered if we aim to develop a 4E methodology.

1.1 The Quest for a Methodological Grounding In the last three decades, an interdisciplinary research program highlighted the necessity of rethinking the theoretical assumptions and experimental practices of mainstream cognitive science (e.g., Chemero & Silberstein, 2008). While the mainstream  view presents cognition as a brain-bound computing process, other approaches claim that cognitive phenomena cannot be understood independently of a cognitive system’s body and its relation to the (social) environment. This non-­ orthodox branch is known under labels such as the “4E approaches” (Newen et al., 2018) or “situated cognition research” (Robbins & Aydede, 2009). The philosophical and scientific antecedents of these approaches reach back to philosophers such as Husserl, Dewey, Wittgenstein, Dreyfus, and Merleau-Ponty as well as to works of developmental and ecological psychologists such as Vygotski and Gibson, but also biologists such as von Uexküll and computer scientists like Winograd. They built the basis for views according to which cognition is an extended, enacted, embodied, or embedded phenomenon. The interdisciplinary debate about these M.-O. Casper (*) · G. F. Artese Institute of Philosophy, University of Kassel, Kassel, Germany e-mail: [email protected]; [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_1

1

2

M.-O. Casper and G. F. Artese

views spiked in the 1990s and is still gaining support. However, the purported disruptive potential of the 4Es for the cognitive sciences was not accessed yet, despite their influence on fields such as robotics (Pfeifer et al., 2007), neuroscience (Moseley et al., 2015), art theory (Krois, 2018), psychology (Ohl & Rolfs, 2017), and psychiatry (De Haan, 2020). The main objections held against 4E research are well-known. The most notorious ones are the coupling-constitution fallacy (Adams & Aizawa, 2001), the cognitive bloat objection (Allen-Hermanson, 2013), the mark of the cognitive question (Rowlands, 2009), and the scaling-up problem (van den Herik, 2021; Casper, 2019; Satne, 2015; Clark, 1998). Additional to these objections, the situated cognition approach suffers from different forms of fragmentation, such as methodical and methodological fragmentation. Although using different methods is standard in most disciplines and sometimes researchers even strive to compensate for weaknesses of particular methods by using several ones in an approach known as “triangulation” (Hopper & Hoque, 2006), the prevailing methodical and methodological fragmentation in 4E research is a severe challenge to the field. One aspect of this fragmentation is that – until now – no overview of possible methods has been provided, so 4E researchers are not aware of all their options for approaching cognitive phenomena. In addition, it is unclear whether specific methods exclude each other or which methods are compatible with the individual 4E positions. Consequently, a synthesis of results in the 4E context rarely takes place, and progress in situated cognition research is seriously impeded. The methodical fragmentation gives rise to further specific 4E counter-arguments and problems such as the motley crew argument (Shapiro, 2010), the problem of replicability (Wagenmakers et al., 2016), an ambivalent connection to experimental psychology and neuroscience, and the sometimes raised problem of triviality (e.g., Goldinger et al., 2016). Overall, the influence of the 4E approaches on the cognitive sciences stagnates at best. This edited volume is motivated by this unrewarding setting. The mentioned problems justify an explicit and detailed quest for a methodological grounding of situated cognition research which will contribute to a possible future solution to 4Es method(olog)ical issues. This volume intends to do so by dealing with questions that need to be answered before a 4E methodology can be developed. The next section of this introduction looks at the methodological criticism against the 4Es. The following paragraphs outline recent approaches to confront these problems and argue for the necessity of a comprehensive 4E methodology. Once all these points and their consequences are made clear, the relevance of the different issues tackled in each chapter of this book conveys. At the end of this introduction, the philosophical contributions in this edited volume are summarized and presented as preliminary considerations for answering some of the most pressing methodological issues in 4E context.

1.2 Method(ologi)cal Criticism In the following, we understand the term “methodical” as referring directly to structured proceedings of scientific practices (methods) and the term “methodological” as referring to investigations of such proceedings. Different from “methodical”, the

1 Introduction

3

term “methodological” bears on the analysis, evaluation, and philosophical contextualization of methods. If properly understood, then the motley crew argument is a crucial methodical criticism against 4E research. It was introduced by Shapiro (2010) and presents a point from where also  methodological considerations can start. Essentially, the argument questions the scientific credibility of claims made in the 4E context. The concern is that once premises of situated cognition research are accepted (e.g., that cognition unfolds beyond the boundaries of neural events), we lose our scientific-methodical access to the phenomena we are interested in as cognition researchers. Suppose it is argued that a conglomerate of neural, bodily, and environmental entities must be orchestrated in the right way to constitute cognitive phenomena. In that case, cognition researchers are confronted with a “motley crew” of entities. The issue with this motley crew is that the relevant entities and processes responsible for cognitive powers to arise are not well defined and, therefore, hardly identifiable. In this theoretical setting, the definition of variables, as well as their quantification for experimental reasons, have been proven to be problematic issues (see also Willems & Francken, 2012 for the “erosion of the concept of embodiment” due to poor definitions which lead to experimental problems; similar problems are highlighted by Körner et al., 2015). The radical claim that cognition is a process that spans the brain, body, and environment (see Kirchhoff & Kiverstein, 2019; Thompson & Varela, 2001; Clark & Chalmers, 1998) forces sympathizers of the 4E approaches to clarify how their phenomena of investigations are tracked in experimental practice. What needs to be considered in this context is that cognitive systems might sometimes rely on one crew of entities and processes to cope with a given situation while they employ a different crew to handle another situation (in this regard, see the notion of adaptivity elaborated by defenders of the enactive approach, e.g., Barandiaran & Moreno, 2008). Under this premise, a cognitive system’s constitutive structures are highly flexible (for this soft-assembly see, e.g., Anderson et al., 2012), they only work as a whole (see the concept of emergence often endorsed by dynamical and non-­ representational approaches, e.g., Thompson, 2007, pp. 37ff.), and the responses of cognitive systems vary in the face of similar stimuli (for a further elaboration of notions such as the ones of enacted dynamics or non-linear behavior see, e.g., Tschacher & Dauwalder, 2003). Hence, the critical punchline of the motley crew argument seems plausible even if we have a more detailed look at how 4E approaches characterize their phenomena. All these mentioned concepts represent challenges for empirical 4E research. The questions that 4E theorists are called to answers are, but not limited to, the following: How is it possible to identify and track the variables that constitute a cognitive system if such a system can spread in different situations over different constituents? How to keep track of its patterns of interaction while those patterns (can) alter irregularly? These questions call for method(olog)ical competencies that can force a scientific entry into cognition’s motley crew of entities and processes. Indeed, many experimental studies already aim at phenomena of situated cognition (Rolfs & Schweitzer, 2022; Favela et al., 2021; Rolfs & Ohl, 2021; Kyselo & Tschacher, 2014). Some of their results became important milestones for 4E research (Diefenbach et al., 2013; Strack et al., 1988).

4

M.-O. Casper and G. F. Artese

However, even these well-known outcomes stand under the suspicion that their replication is problematic or has failed (Wagenmakers et al., 2016). Although some philosophers of science argue that replication is overrated (Feest, 2019), the suspicion and criticism concerning (4E related) premises, empirical results, and experimental settings should be dispelled by a comprehensive methodology of situated cognition research. Despite these problems, situated cognition research aims to understand the development and constitution of cognitive systems in changing environments or in real-world situations. Suggestions to solve this problem, such as strictly limiting the options for a cognitive system to interact in laboratory settings, appear to be an issue for the 4E branch. The controlled conditions prevent or at least severely obstruct the examination of flexible-constitutive patterns produced by cognitive systems. Hence, 4E researchers end up roughly in the following dilemma. On the one hand, overly restrictive conditions prevent 4E researchers from investigating the adaptive potential and dynamics of cognitive systems. On the other hand, real-world situations (or other settings that follow the real world very closely in their design) leave 4E researchers with a motley crew of entities and processes that are scientifically hardly accessible. Similar concerns were voiced early in psychological debates. It was argued that experimental results, if achieved in strictly controlled conditions, seem to lack “external validity” (Campbell & Stanley, 1966) or “ecological validity” (Brunswik, 1956). The bone of contention is either the artificial character of laboratory, experimental settings (Orne & Holland, 1968) or the artificiality of the stimuli used in psychological experiments (Brunswik, 1956). The debate about both aspects is ongoing (Ibanez, 2022; Sonkusare et al., 2019; Kingstone et al., 2008). Each or both aspects might obstruct the generalizability of results found in experimental settings. 4E research can fall precisely into this criticism if empirical research is performed in a certain way. However, apprehensions against empirical studies under controlled conditions are sometimes also presented as misguided since generalizability is not always the aim of experimental research (Mook, 1983). Further, even experimental results generated in real-world surroundings can have limitations to their generalizability. While these counter-arguments are reasonable, they do not hold when generalizability is an explicit goal of 4E research. It is, therefore, not too pugnacious to claim that, under such conditions, the received versions of experimental psychology or neuroscience stand in an ambivalent relation to the 4E project. In addition to the problems presented, 4E research is sometimes confronted with the issue of triviality. That cognitive systems have a body and interact with their surroundings seems trivially true. Hence, the benefits of supporting these approaches are not immediately apparent for everyone – neither for philosophical thinkers nor for empirical cognition research. While some researchers therefore attempt to belittle the contribution of 4E research for cognitive science (Goldinger et al., 2016), others show that there are influential contributions worthy of being noted broadly (e.g., Farina, 2021). A first and rough reply to the triviality concern is that 4E premises imply a change of perspective that respects cognition’s biological and normative grounding. In the wake of that change, the research practices of cognitive

1 Introduction

5

science can alter significantly since such grounding steers them away from solipsistic methods towards classes of phenomena (such as emergent ones) that were rarely considered before. Whether this kind of reply is entirely persuasive is an issue of ongoing discussion. If these points of criticism are connected with the further concern that situated cognition theories do not or cannot explain their phenomena of interest since they do not employ proper and genuine types of explanation, then 4E researchers are confronted with a comprehensive method(olog)ical challenge. Methodological fragmentation, the motley crew argument, difficult experimental prospects, and missing explanatory strategies delineate the tasks ahead for this branch of the cognitive sciences.

1.3 Sketch of Current Method(olog)ical Approaches and Resources Theoretical and practical issues for 4E approaches are numerous. However, it is incorrect to claim that there is little or no attention to the mentioned problems in the situated cognition research community (Wagman & Chemero, 2014). There have been attempts to develop methodological approaches to understand which methods can be employed to (empirically) corroborate 4E claims. There have been efforts to integrate historical disciplines in the cognitive sciences  (Sutton, 2010). Tracking and analyzing cultural techniques, such as mnemo techniques (Tribble, 2005), understood as non-biological scaffoldings of cognition, is a major research direction of this proposal. Also, a focus on historical-archaeological and cultural analyses of cognitive gadgets and tools is available (Malafouris, 2021; Heyes, 2018). A more influential approach relates concepts, methods, and theories usually employed by dynamical system theories, complexity science, or synergetics in the context of cognitive science (e.g., Favela, 2020; Lamb & Chemero, 2014). This line of research aims to understand the principles that govern the development of systems, networks of systems, and their self-enacted as well as reciprocally triggered dynamics through their mathematical formalization. The role of these mathematical formalizations has often been highlighted and appreciated by philosophers and psychologists working on enactivism, ecological psychology, or a combination of the two. Such efforts can help to grasp the reciprocal action patterns between a cognitive system and the outer environment in flux. Enactivists even applied dynamical system theory to social phenomena such as dyadic and triadic relationships to understand how they are established and how their dynamics are enacted (Kyselo & Tschacher, 2014). Similarly, dynamical systems theory has also been used to analyze the unfolding of neural dynamics and behavioral patterns during various social performances (Reséndiz-Benhumea et al., 2020). Interestingly, the tools of dynamical systems theory have been, at least in principle, accepted by all forms of enactivism regardless of whether it is autopoietic enactivism (Maturana & Varela, 1991), sensorimotor enactivism (O'Regan & Noë, 2001) or radical enactivism (Hutto & Myin, 2012).

6

M.-O. Casper and G. F. Artese

Besides these mathematical strategies, the 4E approach includes phenomenological methods. The two most relevant strategies are neurophenomenology (Varela, 1996) and front-loaded phenomenology (Gallagher, 2003). The former wants to train participants to suspend their beliefs and theoretical commitments regarding their subjective perception. Hence, when asked to do so in experimental settings, they produce more accurate descriptions of their phenomenal experiences. The latter incorporates phenomenological concepts and premises in experimental designs. Such concepts and premises are applied to reach theory-driven experiments whose results are supposed to support phenomenological hypotheses. The broader incorporation of these and similar phenomenologically inspired approaches in cognitive science is ongoing (e.g., Miyahara et al., 2020). However, their explanatory power concerning cognitive phenomena is contested (Casper & Haueis, 2022; Pinna & Conti, 2021). Of course, phenomenology is not only employed by 4E research (that can be anti-reductionist, anti-mechanistic, and anti-functionalistic). Phenomenological approaches are also applicable in theories and research on cognitive phenomena based on the new mechanist movement (Pokropski, 2021). A further set of strategies come from the so-called normativist approaches. The supporters of these approaches highlight social interaction patterns and their constitutive role in high-level cognitive states and processes (van den Herik, 2021; Casper, 2019;  Satne, 2015;  Menary, 2012;  Steiner, 2009;  Brandom, 2009,  1994). It is assumed that such states and processes can be held by organisms capable of aligning their behavior to normative structures of social and linguistic interactions. The salient point of this methodical approach is to give collectively enacted practices pride of place in analyses of cognitive phenomena. Hence, social interactions and their investigations become a guiding theme for cognition researchers (De Jaegher, 2018;  De Jaegher & Froese, 2009;  Barnier et  al., 2008;  De Jaegher & Di Paolo, 2007). Also, social interaction in interspecies groups has been considered (Amon & Favela, 2019). Another attempt to methodologically enrich the explanatory strategies of 4E approaches consists in highlighting the importance of a possible combination of ethology, the biological study of behavioral patterns, and 4E research. Such a combination opens up further methodical perspectives and resources for 4E investigations (Casper & Artese, 2023). E.g., ethology intertwines why- and how-questions. It pursues multi-level explanations that imply the micro-level of organism-intern mechanisms. It also includes morphological analyses of body parts (the so-called meso-level of an individual’s behavior) and the macro-level of social interactions and environmental forces. That ethological studies consider the dynamical relations between different levels makes them a particular fit with enactivism. So far, historical, dynamical, phenomenological, normativist and ethological approaches were listed. Still, an open issue is how the different strategies and methods can be combined and integrated. Are 4E researchers confronted with a methodical patchwork that cannot be further structured? Or is it feasible to orchestrate several methods whose interrelation goes beyond spontaneous conjunctions based on contingently shared research problems?

1 Introduction

7

1.4 Methodological Preliminaries and the Contributions of this Volume The points of criticism mentioned above are confronted with various sketched options for approaching cognitive phenomena methodically. However, the relations between those options, their theoretical groundings, and practical results remain a methodological challenge. This challenge can be handled by answering questions coming from the philosophy of science and the history of the cognitive sciences, metaphysics, and partly phenomenology. What questions need to be asked, and what philosophical choices need to be made to understand how the methods in the 4E context relate and when to use them? Suppose we neither care about these questions nor about providing a methodology that covers them. In that case, there is a risk that 4E approaches will be slightly modified and incorporated by non-situated approaches. Such contamination could suppress the 4E debate in the long run. This process is already ongoing, and several attempts can be taken as case examples. Piccinini (2022) suggests a situated form of neural representationalism, Nowakowski (2017) claims that the role of embodiment for cognitive phenomena can be understood through morphological computation, Villalobos and Dewhurst (2018, 2017) argue for a compatibilist view of enactivism and computationalism, Bechtel and Bich (2021) advertise an understanding of enactive concepts such as self-maintenance in mechanistic terms, and Zednik (2011) integrates dynamical explanation procedures into the mechanist approach, while others (Miłkowski et al., 2018) announce a “silent mechanist revolution” in cognitive science. Against this backdrop, methodological questions of 4E research are urging. Though such questions are pressing, it cannot be taken for granted that their solutions are contrary to integrationist or compatibilist views of 4E approaches and cognitivist or mechanistic positions. However, without a closer look, it is not evident whether such compatibilism is tenable. The first methodological question we should ask in this context is whether the number of available methods calls for an appreciation or limitation of such plurality. An answer to this question goes beyond the level of specific methods. It can be developed by a philosophy of science asking whether we should support explanatory pluralism or explanatory unification. Depending on its specific form, pluralism entitles different methods, such as different types of explanations, to work equally well for a problem of interest. A patchwork or integration of different methods is generally deemed possible without establishing a hierarchy of relevance between them. Also depending on its special definition, explanatory unification tends to support the reduction of applicable methods, to develop a grading of explanatory significance between them, and to determine which approach has what function within the explanation process – e.g., being merely descriptive instead of being ultimately explanatory. The whole issue just described is tackled in the first section of this volume labeled “philosophy of science”. The opening chapter by Casper frames the methodical plurality in the 4E debate as a problem that requires a philosophical response. More specifically, methodical

8

M.-O. Casper and G. F. Artese

plurality forces a methodological problem of choice upon 4E researchers between explanatory pluralism and explanatory unification. The author argues that 4E researchers should explicitly and carefully commit themselves to one of these positions before starting method(olog)ical inquiries. Because our understanding of what proper methods are, how they relate, and under what conditions more than one method can be simultaneously employed differs considerably after such a commitment is incurred. It is assumed that only by taking sides in this issue a methodological orientation is available which allows for investigations of 4E methods. His chapter develops a strategy for solving the problem of choice and offers preliminary guidance on starting a 4E methodology. The second contribution by Sanches De Oliveira emphasizes that explanatory plurality, as available in the mind sciences, is often understood as a philosophical challenge or a difficult issue to handle. The author argues in opposition to such views that the debate concerning methodological questions draws unnecessary philosophical attention. Instead of getting bogged down by questions about the relations, compatibility, and quality of methods used in 4E context, this contribution suggests a stance in which such questions do not play an essential role. This philosophical position’s driving idea is that explanation procedures are different explanatory styles. In particular, the theoretical relevance of assumed (in-)compatibilities between different explanation types is likely to decrease significantly if the variance of methods is thus understood. Once 4E researchers commit themselves to a pluralist or unificationist strategy, they are still faced with a whole bundle of positions that could function as starting points for empirical cognitive research. In the third contribution of this volume, Heras-Escribano asks which investigative strategies – that roughly share the incentives of 4E cognitive science – can provide solid and non-representationalist scientific access to cognitive phenomena. The main philosophical punchline of this paper is based on the fact that enactivism has been, at times, prominently depicted as a philosophy of nature and not a scientific research program for cognitive systems. While Heras-Escribano suggests that enactivism is scientifically disqualified since it does not offer a framework in which cognitive phenomena can be empirically studied, ecological psychology is instead outlined as an approach without such shortcomings. Ultimately, ecological psychology is presented as a philosophical approach to the mind and a proper scientific framework for studying cognitive phenomena. The project of situated cognition research developed in the wake of phenomenology and philosophical pragmatism. A shared idea between phenomenology, pragmatism, and situated cognition is that philosophical problems should be solved by analyzing practices. If translated into an interdisciplinary philosophy of cognition, then this convincement implies that studying normative structures of practices should receive investigative priority. Those practices imply dynamics that unfold according to their own rules; the discursive practice is an example for this (Brandom, 1994). Those rules are deemed constitutive preconditions for specific cognitive phenomena (e.g.,  “high-level” ones  such as beliefs). For some 4E researchers, normativity is hence an important – and sometimes even seen as a constitutive – aspect of cognitive states and processes. The family of approaches that point out the high relevance of

1 Introduction

9

normativity for cognitive phenomena can be labeled “normativism”. However, there are different normativist positions and it is sometimes unclear how they relate to each other and how normativist strategies find their access into experimental practices in cognitive science. In his contribution, Steiner identifies the most important normativist claims and elaborates their connection. Three major claims on normativity and cognition are presented and their affiliations to 4E research are unfolded. Based on his analysis, Steiner further confronts the issue of how to deal with folk psychology and its concepts in context of (4E) cognitive science. Despite its assumed normative nature, folk psychology occupies an intricate status also in situated cognition approaches and poses a challenge for neuroscientific theorizing and experimenting. The second section of the volume, titled “ontology of the mind”, is concerned with the inherent relationship of metaphysical and ontological problems (mostly concerning the constitution of cognitive phenomena) with establishing a 4E methodology. Authors in this section generally assume that ontological claims concerning the nature of cognitive processes affect how (cognitive) phenomena are investigated. From their very early formulation, theories such as enactivism or ecological psychology imply an ontological break from existing cognitivist theories of cognition and challenge the orthodox picture of cognitive processes, mental representations, the boundaries between the mental and the physical, etc. Different ontological statements concerning the nature of mental phenomena entitle us to focus solely on parts of an individual organism, the whole organism, or the interaction patterns between the organism and its surroundings to study cognitive phenomena. Krickel is the first author of the section. She analyzes possible criteria to identify the “the mark of constitution” for cognitive phenomena. She presents minimal criteria that, if satisfied, count as genuinely constitutive of cognitive capacities. However, the author argues that none of the available prominent accounts of constitution – property-based, diachronic, and mechanistic  – fully satisfy all criteria. It is ultimately argued in favor of a pragmatist understanding of the nature of cognition in which it is possible to maintain a pluralistic and liberal view dictated by the scope and the aims of individual scientific questions. In the second contribution of this section, Kiverstein & Kirchhoff aim to overcome the coupling-constitution fallacy and offer a “mark of the cognitive” suited for non-representational approaches to cognition. They claim that the theoretical power behind the coupling-constitution fallacy strongly depends on an implicit idea of constitution that is inherently atemporal and synchronic. In contrast, the authors argue in favor of a diachronic notion of constitution. The notion of diachronic constitution challenges the distinction between causal and constitutive factors and instead emphasizes how cognitive systems are involved in a non-linear interplay of various factors. The chapter by Facchin & Negro can be seen as a direct response to the recent monography of Kirchhoff and Kiverstein “Extended Consciousness and Predictive Processing” (2019). The two authors critically discuss the concrete applications of Kirchhoff and Kiverstein’s notion of diachronic constitution. The main idea that the two authors aim to challenge in their contribution is that not only cognition but also phenomenal experience can be articulated as a phenomenon emerging from the interplay of neural, bodily, and environmental factors. While Facchin and Negro

10

M.-O. Casper and G. F. Artese

does not dismiss the hypothesis of extended consciousness as such, they deem Kirchhoff and Kiverstein’s claim unsatisfactory  since it leads to a “phenomenal bloat”. In the following chapter by Favela & Chemero, it is argued that empirically-­ supported theories of situated phenomena demand ontological pluralism. While pragmatist philosophers such as William James supported similar claims, Favela & Chemero’s argumentation does not arise from mere philosophical considerations but from an example of empirical studies directed at bird flocks. The behavior of such flocks is presented as a situated phenomenon, and understanding these phenomena requires a mutual development of conceptual and methodological issues. In particular, these flocks exhibit ontologically plural features such as component and interaction dominance. Hence, research on situated phenomena should start from an ontological pluralism without rejecting metaphysical monism. As mentioned, some 4E researchers support the idea that a radical change in the research practices of cognitive science is a non-negotiable necessity. Such researchers are confronted with the crucial task of offering methodical resources that can guarantee a qualitative shift in the production of scientific knowledge in cognitive science. That is why this edited volume includes a third section, called “applications”. It develops philosophical avenues to connect theoretical 4E positions with consciousness research. The contribution of Coninx & Stilwell directly aims to analyze the conditions necessary for an integrative model in the study of chronic pain. According to them, a scientifically respectable theory of chronic pain needs to be ontologically clear, provide a fine-grained conceptualization of the phenomenon under investigation, be clear on the methodologies used, and ultimately, be able to make explicit the direct therapeutic applications that follow from a specific approach. In their paper, they argue in favor of an enactive and non-reductionist approach to pain that goes hand in hand with a form of “integrative pluralism”. Their contribution hence connects with the texts by Casper and Sanches de Oliveria, who discuss the pros and cons of supporting explanatory pluralism, unification, or none of both. In the last contribution of the volume, Artese investigates the methodological consequences of the different understanding of affordance perception. In particular, he addresses how affordance perception can be characterized in relation to environmental niches that are socially and culturally organized. The main problem the paper is concerned with is whether the presence of normative, socio-cultural practices and socially structured contexts can influence affordance experience. After elaborating on the different positions present in the debate, the author, inspired by the phenomenological work of Aron Gurwitsch, holds that the experience of affordances in context needs to be studied as a perceptual “field” always having three dimensions: the theme, the thematic field, and the margin. Following Artese, these dimensions can help to access how functionally and interconnected social settings shape and modulate affordance experience. Several chapters of this edited volume are stronger related to each other than others. This is the case, e.g., with the two chapters on explanatory pluralism and the contributions about (diachronic) constitution. This philosophical proximity does not include argumentative consensus. Still, where deemed profitable for furthering

1 Introduction

11

discussion and re-sharpening the theoretical overlaps between certain chapters, the authors commented on each other’s texts. Those commentaries follow Chaps. 2 (“A Methodological Problem of Choice for 4E Research”), 4 (“Explanatory Diversity and Embodied Cognitive Science”), 8 (“Extended Cognition and the Search for the Mark of Constitution”), 10 (“Dissolving the Causal-­Constitution Fallacy”), and 12 (“Predicting Processing and Extended Consciousness”).

References Adams, F., & Aizawa, K. (2001). The bounds of cognition. Philosophical Psychology, 14(1), 43–64. Allen-Hermanson, S. (2013). Superdupersizing the mind: Extended cognition and the persistence of cognitive bloat. Philosophical Studies, 164(3), 791–806. Amon, M. J., & Favela, L. H. (2019). Distributed cognition criteria: Defined, operationalized, and applied to human-dog systems. Behavioural Processes, 162, 167–176. Anderson, M. L., Richardson, M. J., & Chemero, A. (2012). Eroding the boundaries of cognition: Implications of embodiment. Topics in Cognitive Science, 4(4), 717–730. Barandiaran, X., & Moreno, A. (2008). Adaptivity: From metabolism to behavior. Adaptive Behavior, 16(5), 325–344. Barnier, A. J., Sutton, J., Harris, C. B., & Wilson, R. A. (2008). A conceptual and empirical framework for the social distribution of cognition: The case of memory. Cognitive Systems Research, 9(1–2), 33–51. Bechtel, W., & Bich, L. (2021). Grounding cognition: Heterarchical control mechanisms in biology. Philosophical Transactions of the Royal Society B, 376(1820), 20190751. Brandom, R. (1994). Making it explicit: Reasoning, representing, and discursive commitment. Harvard University Press. Brandom, R. (2009). How analytic philosophy has failed cognitive science. TAP-2009 Towards an Analytic Pragmatism, 121. Brunswik, E. (1956). Perception and the representative design of psychological experiments. University of California Press. Campbell, D.  T., & Stanley, J.  C. (1966). Experimental and quasi-experimental designs for research. Rand McNally & Company. Casper, M. O. (2019). Social enactivism: On situating high-level cognitive states and processes. De Gruyter. Casper, M. O., & Artese, G. F. (2023). A methodological response to the motley crew argument: Explaining cognitive phenomena through enactivism and ethology. In J. M. Viejo & M. Sanjuán (Eds.), Life and Mind (pp. 27–48). Springer. Casper, M.  O., & Haueis, P. (2022). Stuck in between. Phenomenology’s explanatory dilemma and its role in experimental practice. Phenomenology and the Cognitive Sciences, 22(7), 1–24. Chemero, A., & Silberstein, M. (2008). After the philosophy of mind: Replacing scholasticism with science. Philosophy of Science, 75(1), 1–27. Clark, A. (1998). Embodiment and the philosophy of mind. Royal Institute of Philosophy Supplements, 43, 35–51. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58(1), 7–19. De Haan, S. (2020). Enactive psychiatry. Cambridge University Press. De Jaegher, H. (2018). The intersubjective turn. In A. Newen, L. De Bruin, & S. Gallagher (Eds.), The Oxford handbook of 4E cognition (pp. 453–467). Oxford University Press. De Jaegher, H., & Di Paolo, E. (2007). Participatory sense-making. Phenomenology and the Cognitive Sciences, 6(4), 485–507.

12

M.-O. Casper and G. F. Artese

De Jaegher, H., & Froese, T. (2009). On the role of social interaction in individual agency. Adaptive Behavior, 17(5), 444–460. Diefenbach, C., Rieger, M., Massen, C., & Prinz, W. (2013). Action-sentence compatibility: The role of action effects and timing. Frontiers in Psychology, 4, 272. Farina, M. (2021). Embodied cognition: Dimensions, domains and applications. Adaptive Behavior, 29(1), 73–88. Favela, L. H. (2020). Cognitive science as complexity science. Cognitive Science, 11(4), e1525. Favela, L. H., Amon, M. J., Lobo, L., & Chemero, A. (2021). Empirical evidence for extended cognitive systems. Cognitive Science, 45(11), e13060. Feest, U. (2019). Why replication is overrated. Philosophy of Science, 86(5), 895–905. Gallagher, S. (2003). Phenomenology and experimental design. Toward a phenomenologically enlightened experimental science. Journal of Consciousness Studies, 10(9–10), 85–99. Goldinger, S. D., Papesh, M. H., Barnhart, A. S., Hansen, W. A., & Hout, M. C. (2016). The poverty of embodied cognition. Psychonomic Bulletin & Review, 23(4), 959–978. Heyes, C. (2018). Cognitive gadgets: The cultural evolution of thinking. Harvard University Press. Hopper, T., & Hoque, Z. (2006). Triangulation approaches to accounting research. In Z. Hoque (Ed.), Methodological issues in accounting research: Theories and methods (pp.  562–572). Spiramus Press. Hutto, D. D., & Myin, E. (2012). Radicalizing enactivism: Basic minds without content. MIT press. Ibanez, A. (2022). The mind’s golden cage and cognition in the wild. Trends in Cognitive Sciences, 26(12), 1031–1034. Kingstone, A., Smilek, D., & Eastwood, J. D. (2008). Cognitive ethology: A new approach for studying human cognition. British Journal of Psychology, 99(3), 317–340. Kirchhoff, M. D., & Kiverstein, J. (2019). Extended consciousness and predictive processing: A third-wave view. Routledge. Körner, A., Topolinski, S., & Strack, F. (2015). Routes to embodiment. Frontiers in Psychology, 6, 940. Krois, J. M. (2018). Philosophy and iconology. In M. Lauschke, J. Schiffler, & F. Engel (Eds.), Ikonische Formprozesse. Zur Philosophie des Unbestimmten in Bildern (pp. 1–27). De Gruyter. Kyselo, M., & Tschacher, W. (2014). An enactive and dynamical systems theory account of dyadic relationships. Frontiers in Psychology, 5, 452. Lamb, M., & Chemero, A. (2014). Structure and application of dynamical models in cognitive science. Proceedings of the Annual Meeting of the Cognitive Science Society, 36(36). Malafouris, L. (2021). How does thinking relate to tool making? Adaptive Behavior, 29(2), 107–121. Maturana, H. R., & Varela, F. J. (1991). Autopoiesis and cognition: The realization of the living (Vol. 42). Springer Science & Business Media. Menary, R. (2012). Cognitive practices and cognitive character. Philosophical Explorations, 15(2), 147–164. Miłkowski, M., Clowes, R., Rucińska, Z., Przegalińska, A., Zawidzki, T., Krueger, J., et al. (2018). From wide cognition to mechanisms: A silent revolution. Frontiers in Psychology, 2393. Miyahara, K., Niikawa, T., Hamada, H. T., & Nishida, S. (2020). Developing a short-term phenomenological training program: A report of methodological lessons. New Ideas in Psychology, 58, 100780. Mook, D. G. (1983). In defense of external invalidity. American Psychologist, 38(4), 379–387. Moseley, R., Kiefer, M., & Pulvermüller, F. (2015). Grounding and embodiment of concepts and meaning: A neurobiological perspective. In Y. Coello & M. H. Fischer (Eds.), Perceptual and Emotional Embodiment (pp. 101–122). Routledge. Newen, A., De Bruin, L., & Gallagher, S. (2018). The Oxford handbook of 4E cognition. Oxford University Press. Nowakowski, P. R. (2017). Bodily processing: The role of morphological computation. Entropy, 19(7), 295.

1 Introduction

13

Ohl, S., & Rolfs, M. (2017). Saccadic eye movements impose a natural bottleneck on visual short-term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 43(5), 736. O’Regan, J. K., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24(5), 939–973. Orne, M.  T., & Holland, C.  H. (1968). On the ecological validity of laboratory deceptions. International Journal of Psychiatry, 6(4), 282–293. Pfeifer, R., Lungarella, M., & Iida, F. (2007). Self-organization, embodiment, and biologically inspired robotics. Science, 318(5853), 1088–1093. Piccinini, G. (2022). Situated neural representations: Solving the problems of content. Frontiers in Neurorobotics, 16–846979. Pinna, B., & Conti, L. (2021). Illusory figures: From logic to phenomenology. Psychology of Consciousness: Theory, Research, and Practice, 8(2), 164. Pokropski, M. (2021). Mechanisms and consciousness: Integrating phenomenology with cognitive science. Routledge. Reséndiz-Benhumea, G. M., Sangati, E., & Froese, T. (2020). Levels of coupling in dyadic interaction: An analysis of neural and behavioral complexity. 2020 IEEE Symposium Series on Computational Intelligence (SSCI), 2250–2256. Robbins, P., & Aydede, M. (Eds.). (2009). The Cambridge handbook of situated cognition. Cambridge University Press. Rolfs, M., & Ohl, S. (2021). Moving fast and seeing slow? The visual consequences of vigorous movement. Behavioral and Brain Sciences, 44, e131. Rolfs, M., & Schweitzer, R. (2022). Coupling perception to action through incidental sensory consequences of motor behaviour. Nature Reviews Psychology, 1(2), 112–123. Rowlands, M. (2009). Extended cognition and the mark of the cognitive. Philosophical Psychology, 22(1), 1–19. Satne, G. (2015). The social roots of normativity. Phenomenology and the Cognitive Sciences, 14(4), 673–682. Shapiro, L. (2010). Embodied cognition. Routledge. Sonkusare, S., Breakspear, M., & Guo, C. (2019). Naturalistic stimuli in neuroscience: Critically acclaimed. Trends in Cognitive Sciences, 23(8), 699–714. Steiner, P. (2009). Mind the consequences of inferentialism and normativism: Conceptual mental episodes ain’t in the head (at all). TAP-2009 Towards an Analytic Pragmatism, 112. Strack, F., Martin, L. L., & Stepper, S. (1988). Inhibiting and facilitating conditions of the human smile: A nonobtrusive test of the facial feedback hypothesis. Journal of Personality and Social Psychology, 54(5), 768–777. Sutton, J. (2010). Exograms and interdisciplinarity. History, the extended mind, and the civilizing process. In R. Menary (Ed.), The extended mind (pp. 189–226). MIT Press. Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of mind. Harvard University Press. Thompson, E., & Varela, F. J. (2001). Radical embodiment: Neural dynamics and consciousness. Trends in Cognitive Sciences, 5(10), 418–425. Tribble, E. (2005). Distributing cognition in the globe. Shakespeare Quarterly, 56(2), 135–155. Tschacher, W., & Dauwalder, J. P. (Eds.). (2003). The dynamical systems approach to cognition: Concepts and empirical paradigms based on self-organization, embodiment, and coordination dynamics (Studies of nonlinear phenomena in life science Vol. 10). World Scientific. van den Herik, J. C. (2021). Rules as resources: An ecological-enactive perspective on linguistic normativity. Phenomenology and the Cognitive Sciences, 20(1), 93–116. Varela, F.  J. (1996). Neurophenomenology: A methodological remedy for the hard problem. Journal of Consciousness Studies, 3(4), 330–349. Villalobos, M., & Dewhurst, J. (2017). Why post-cognitivism does not (necessarily) entail anti-­ computationalism. Adaptive Behavior, 25(3), 117–128.

14

M.-O. Casper and G. F. Artese

Villalobos, M., & Dewhurst, J. (2018). Enactive autonomy in computational systems. Synthese, 195(5), 1891–1908. Wagenmakers, E.  J., Beek, T., Dijkhoff, L., Gronau, Q.  F., Acosta, A., Adams, R.  B., Jr., et  al. (2016). Registered replication report: Strack, Martin, & Stepper (1988). Perspectives on Psychological Science, 11(6), 917–928. Wagman, J. B., & Chemero, A. (2014). The end of the debate over extended cognition. In T. Solymosi & J. R. Shook (Eds.), Neuroscience, neurophilosophy and pragmatism (pp. 105–124). Palgrave Macmillan. Willems, R. M., & Francken, J. C. (2012). Embodied cognition: Taking the next step. Frontiers in Psychology, 3, 582. Zednik, C. (2011). The nature of dynamical explanation. Philosophy of Science, 78(2), 238–263.

Part I

Philosophy of Science

Chapter 2

A Methodological Problem of Choice for 4E Research Mark-Oliver Casper

Abstract  The 4E approaches have been frequently linked to the claim that they fundamentally change how we theoretically analyze and empirically investigate cognitive phenomena. Despite this aspiration, the methodology of 4E researchers received surprisingly little attention, although the envisaged change of mainstream cognitive studies presupposes elaborated methodical competencies. This chapter argues that a methodology is necessary for the 4Es. However, several preliminary considerations need to precede a methodological analysis of the branch. It is claimed that methodological analyses of 4E research first need to answer the problem of choice whether to support a version of explanatory pluralism or explanatory unification. The answer to this problem significantly influences how a methodology can be developed. The problem can only be properly met, however, if the different versions of pluralist and unificationist positions are related so that we can understand where exactly they diverge or overlap. Below, the spectrum of integration is introduced, which suggests a relation between pluralist and unificationist views. The spectrum is a provisional guide to a 4E methodology as it invites 4E researchers to place themselves within this spectrum and approach methodological and methodical questions from there. Keywords  4E cognition · Situated cognition · Methodology · Motley crew argument · Problem of choice · Spectrum of integration · Explanatory pluralism · Explanatory unification · Explanation types

2.1 Introduction The claim that cognition is an extended, enacted, embodied, and embedded phenomenon gained increasing attention over the last decades (Newen et  al., 2018; Robbins & Aydede, 2009). The main goal of those supporting this claim is to M.-O. Casper (*) Institute of Philosophy, University of Kassel, Kassel, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_2

17

18

M.-O. Casper

understand how cognitive phenomena emerge in real-world situations from the interaction of the brain, body, and environment. From this so-called 4E perspective, entities and processes that are seen as constitutive of cognitive phenomena have special characteristics. They are decentralized (there is no localizable instance of control or coordination between them), at least sometimes heterogeneous (there are biological and non-biological constituents), they are dynamically interacting, and adaptively changed if the situation allows or requires it (in terms of cognitive offloading or plastic change). Such premises multiplied the possible disciplines included in cognition research (Hohol, 2021; Sutton, 2010). This heterogeneous stage of disciplines can be deemed a benefit for the situated cognition research community since it allows to approach shared target phenomena in many methodical ways. However, there are also neglected problems that come along with this plurality of disciplines. It is far from clear how the different methods applied relate to each other and how interdisciplinary collaboration between disciplines might be realized methodically. E.g., how do researchers, who use primarily dynamical explanations, connect with the outputs of the new mechanist movement (Craver & Darden, 2013; Craver, 2007)? Under what conditions is such integration feasible? A methodology is necessary to confront these and further questions that bugged cognitive science in general (e.g., Newell, 1973) as well as 4E research since it beginnings. In the long run, such a methodology is a reality check for the ambitious aim that is frequently linked to 4E research: To fundamentally change the way how we theoretically analyze and empirically investigate cognitive phenomena (Stewart et al., 2010). This chapter asks questions the answers to which structure a methodological analysis of 4E research. In the first section, some of the pressing reasons to call for a 4E methodology are detailed. The second section presents key preliminary considerations for a 4E methodology. Especially the problem of choice between explanatory pluralism and explanatory unification is expressed. Which side a researcher chooses influences the understanding of what a 4E methodology should cover. Suppose a researcher commits to, e.g., a form of explanatory unification. In that case, entitlements can originate to seek certain methods and to avoid others – not because they are deemed inappropriate for a specific research question but because they are not treated as methods at all. A 4E methodology will hence vary according to the answer given to the problem of choice. The problem of choice is reified in Sect. 2.3, in which a specific kind of method is focused on: the different explanation types used in the debate. While the following text focuses primarily on explanation types to substantiate the problem of choice and its consequences, it is not claimed that a methodology should only cover such types. Also other aspects of cognitive research practice such as data mining, experiment design, tool development, etc. should be considered but they are beyond the scope of this contribution. Instead it is asked whether 4E researchers should use different explanation types to compensate for possible weaknesses each of them might have, or whether they should concentrate on applying a single explanation type for as many target phenomena as possible. If we support the first option and, therefore, a version of explanatory pluralism, how can we be sure the outputs of different explanation

2  A Methodological Problem of Choice for 4E Research

19

types are compatible? If we focus on one explanation type and therefore support explanatory unification, how do we know which type to choose? Section 2.4 elaborates on options to answer these questions while the “spectrum of integration” is presented. This spectrum involves different positions of explanatory pluralism and explanatory unification. It hence describes a landscape of abstract but important options for researchers who work on a methodological analysis of 4E research. These options are mainly defined by a set of conditions under which the integration of varying explanation types might be feasible. This section also gives an example of a cross-explanatory integration. In Sect. 2.5, I argue for choosing a specific place in the spectrum of integration and outline the next steps for a 4E methodology that starts from there.

2.2 In Need for a Methodology Even though the methodical shortcomings of the 4Es were mentioned early in the debate (Kiverstein & Clark, 2009) and highlighted again recently (Goldinger et al., 2016; Körner et al., 2015), a structured solution or even an attempt to react comprehensively to these shortcomings is missing. That we miss such a solution is even more surprising since a solid methodical repertoire might be critical for influencing cognitive science’s mainstream and inducing the ambitious “radical changes” 4E researchers frequently announced. This section therefore presents the most pressing reasons why a methodology is needed for the 4E branch. (i) The specific driver of a 4E methodology is the so-called “motley crew argument” (Shapiro, 2010). This argument claims that, if the core assumptions of 4E approaches are true – for instance the hallmark thesis that constituents of cognitive phenomena are decentralized, heterogeneous, and dynamically interacting –, then cognition researchers need to include so many different entities and processes when studying cognitive phenomena (empirically) that no scientifically interesting regularities can be found. One reason for this is that they might not have the methods for this research. Also, to identify and quantify the relevant variables that constitute cognitive phenomena is almost impossible due to the quick and flexible adaption of cognitive systems to their surroundings. These problems might disqualify situated cognition research as a scientific research project. The 4Es simply lack the tools for the identification and analysis of scientifically relevant regularities. The actual question raised by the motley crew argument, hence, is whether the available heterogeneous repertoire of methods can support 4E claims. Researchers like Gallagher (2017a, p. 21) might have given the motley crew argument momentum by suggesting that enactivism, a subdivision of the 4Es, should be seen as a philosophy of nature and not as a scientific research program. The mentioned argument might have been strengthened further by enactivists such as Di Paolo who explicitly avoid solid methodical commitments.

20

M.-O. Casper

Their main argument is that such commitments would come along with unnecessary restrictions of the enactive theory. Other enactivist philosophers concentrated on criticizing the representationalist posits of orthodox cognitive science. While their critique gained international reach and attention (Hutto & Myin, 2014), a positive description of how a non-representationalist cognitive science will most likely look like is hardly provided. As a consequence, the motley crew argument is still a pressing issue. A feasible non-representationalist approach to understand cognitive systems includes the examination of their soft-assembly and interaction patterns in real-world situations. However, such an approach needs strong interdisciplinary, methodologically transparent cooperation to deal with the epistemological challenges coming along with the 4E approach. (ii) The panoply of situated cognition research includes at least four different theses (4Es) that are linked to varying historical sources of philosophy, such as phenomenology (Gallagher & Zahavi, 2020) and pragmatism (Casper, 2019; Menary, 2016). These theses have various theoretical premises, such as representationalism & non-representationalism, and diverse interdisciplinary connections. In this complex setting, 4E researchers are usually unaware of all their methodical options. A comprehensive overview of the applied methods, such as experiment design or explanation types, is not provided, yet. This methodical opacity impedes research progress because the best-suited methods for investigating 4E claims might not be known by those who (empirically) examine them. Based on this opacity, possible interdisciplinary connections might be undetected and the integration of research efforts is hence obstructed. A 4E methodology can remove this non-transparency to offer a so far missed methodical mapping across the approaches in situated cognition. (iii) The concerns following cognitive science since its beginnings are connected with its interdisciplinary approach, ad hoc theorizing, search for effects without proper theoretical grounding, and missing systematic accumulation of knowledge (Miłkowski & Hohol, 2021; Newell, 1973). The main issue in this context can be summed up as the fragmentation worry. What seemed to apply to cognitive science in general applies even more precisely to 4E research. Not only increased the 4E branch the number of possible disciplines that can be involved in cognition research. It has also been concerned with oppositions between the synchronic and diachronic constitution of cognitive phenomena (Kirchhoff & Kiverstein, 2020), and the antagonism between representational and non-representational theories of cognition (Hutto et  al., 2014). Possible solutions to these disputes are typically depicted as either/or choices for one side (with rare exceptions such as Bich & Bechtel, 2021; Carls-Diamante, 2019). Consequently, these disputes deflect from the connecting aspects of 4E approaches and force a decision for (or against) one of the available sides. By structuring discussions in this way, fragmentations between the 4E positions are consolidated, leading to duplication of effort and lack of (mutual) understanding, a premature consensus within 4E positions, and a lack of productive disputes. These issues can be called the disciplinary, theoretical, and method(olog)ical fragmentation of situated cognition research.

2  A Methodological Problem of Choice for 4E Research

21

Why exactly should fragmentation lead to a fragmentation worry? Because 4E research might remain a minority view if it does not get rid of or at least minimizes the fragmentation in terms of methods (e.g., various applicable explanation types with unknown or problematic relations among each other) and/or in terms of theoretical commitments (e.g., representationalism vs. enactivism). It might stay a minority view since the fragmented perspectives encouraged by it either do not lead to significantly altered research practices that differ from those already developed in the cognitivist framework of cognitive science – or 4E research does not lead to a comprehensive, alternative view of cognition. It instead might just embody a possible, theoretical add-on that can be considered optionally in the context of cognitivist research. A 4E methodology that examines how theories and methods mutually require, constrain, and exclude each other in 4E research limits these problematic parts of fragmentation. (iv) The motley crew argument, the methodical opacity, and the fragmentation worry lead to what I call the speculation worry. Philosophers of mind are less restricted in theorizing about cognitive phenomena compared to, e.g., cognitive neuroscientists. This can be a benefit as usual methodical limitations of other disciplines do not impede the conceptual engineering (of philosophers) in cognition research. Still, philosophers reach premises and posits which are inferentially coherent with central assumptions in cognition research. However, such posits might be of slim value for actual research practice. Take for example the claim that action-oriented representations are distributed over several levels of a cognitive system (Clark, 1997, pp. 168 f.). Although the reasoning about mental representations and their action-oriented characteristics made representationalism more flexible  – regarding, e.g., the bottleneck problem (Clark, 2015a) – it also came along with theses about how such representations are constituted that are scientifically not or hardly suitable. In other words, if 4E research does not minimally restrict itself by committing to methodical options and protocols, its speculative side might outweigh its potential for progressive change in cognitive science. Assuming the points mentioned above are valid, it is still not settled how methodological investigations should be structured and what issues need to be focused on. Hence, some preconditions of such a methodological analysis should be determined first. The consideration of preconditions leads to the problem of choice.

2.3 The Problem of Choice A first easy step towards a 4E methodology is to confirm that a plurality of methods exists. Methods can be understood as structured proceedings applied in scientific practice for various purposes (developing concepts, issuing theories and models, making observations, gathering data, constructing experimental settings, offering

22

M.-O. Casper

explanations, providing predictions) (Kaplan, 2017, pp.  18ff.). This plurality of methods is also known as explanatory plurality (Mitchell, 2002) which refers to the simple fact that science in general (and situated cognition research in particular) employs a number of methods. This trivial assessment is followed by the daunting task how to philosophically react to this fact with a claim that sheds light on what this plurality is good for. There are roughly two modes of response to this question. In the first mode, the plurality is seen as a problem since the more methods are applied, the more demanding their coordination and integration of their results will be. In the second mode, the plurality is understood as a benefit that helps to stabilize observed phenomena and gain as much information about them as possible. Many positions between these extremes are possible (Muszynski & Malaterre, 2021; Miłkowski et al., 2019). The first mode of response can be subsumed under “explanatory unification” while the second mode is part of “explanatory pluralism”. These two options determine the problem of choice. Depending on the choice, far-reaching consequences result for a methodological analysis of the 4E branch. The following succinctly recapitulates the two positions (including certain subversions of it). These two positions will be depicted below as the ends of a spectrum we need to place ourselves in if we aim at methodological analyses of the situated cognition debate.

2.3.1 Explanatory Unification Newell’s (1990, 1973) prominent pursuit of unification in cognitive science is an attempt to circumvent what he assessed deterioration or even possible failure of cognition research. As mentioned above, a major problem for cognition researchers is their ad hoc reasoning and fishing for effects in experimental settings. It is fishing for effects if research is not guided by proper theoretical groundings (Milkowski & Hohol, 2021). No theory is proposed from which the effects are being predicted. Instead they are noticed “on the spot” in experimental settings. Recently however, exploratory experimenting is a frequently accepted form of conducting experiments that is basically understood as an empirical investigation of a target phenomenon without a proper theoretical framework or even without explicit hypotheses (Haueis, 2014; Feest, 2012). Still, exploratory experiments are only applicable on a large-­ scale and can function as a principal research practice if neither the fragmentation worry is shared nor a comprehensive theory of cognition is valued. Also an entire unification of cognitive science needs to be dismissed or at least seen as dispensable (Barker, 2019; Taylor & Vickers, 2017) if exploratory experiments gain methodical weight. On the contrary, mechanistic explanation procedures – that take pride of place in mainstream cognitive science (Miłkowski et al., 2018; Craver & Tabery, 2015) – imply a moderate form of reductionism, representationalism intermittently, and the effort of an integrative coordination between various kinds of evidence in order to arrive at a “mosaic unity” (Craver, 2007). Mechanistic explanations hence seem to

2  A Methodological Problem of Choice for 4E Research

23

be at least partly at odds with theory free research and exploratory experiments. This kind of explanatory unification hence revolves around coordination of evidence and reduction of phenomena to a material structure whose activity is constituting it. It is possible to rephrase this perspective and claim that unification is either about building “a theory of entities (or a single entity) that is responsible for all observed phenomena [of a certain kind], or study general principles that govern them” (Miłkowski & Hohol, 2021). This general description, however, does not do justice to the different versions of unification available. In the following, three forms of unification are presented to highlight the rarely acknowledged flexibility of this position. (i) Unification by reduction is the first version that was already touched upon and that can take two forms. In one of these forms, what is being unified are different theories. Unifying theories means to unify (parts of) science and that theoretical unification is first and foremost achieved by reduction (Grantham, 2004, p. 135). A crucial aspect of reduction is known as the “condition of derivability” (Nagel, 1961, p. 354). The general idea is that reduction is successful if crucial theoretical aspects like laws or general principles of a specific science (the secondary science to be reduced) are inferential consequences or premises of another science (the primary science to which reduction is made). If this condition can be satisfied, then the secondary science is derivative from the primary science and the former can be reduced. Reduction is hence based on a logical relation between theories. For a review of literature about this kind of reduction see Schaffner (1993) and Sarkar (1998). In another form, unification by reduction is referring to a specific way of explaining target phenomena. The significant difference necessary to keep apart the theory-based unification by reduction and the explanation-based unification by reduction is a difference between theories and explanations as stated by Morrison (2000) and Halonen & Hintikka (1999). They claim that explanations often provide additional information about a phenomenon that is not simply derivable from theoretical positions or theories. The theories might be unified and grant a certain pattern of explanatory inferences (an explanation). But the concrete explanation of a phenomenon is not already implicit in a theory. In these lights, let us focus on mechanistic explanations. Mechanistic explanations are sometimes conceived of as reductive explanations (Bechtel & Abrahamsen, 2008; Craver, 2007; Machamer et al., 2000). They are reductive explanations because “phenomena manifested at one level of organization are explained in terms of components, parts and operations at lower level(s) of organization” (Zednik, 2011, 261; others may label mechanistic explanations differently, see Krickel, 2020, 2018). All kinds of phenomena (including cognitive ones) are supposed to be explained in a unifying manner by identifying, describing, and analyzing the mechanisms responsible for them. If an explanation type like the mechanistic explanation is playing the role of a great unifier of how phenomena get explained, then all other attempts to explain phenomena are evaluated in terms of their integration with, similarity to, and useful-

24

M.-O. Casper

ness for mechanistic explanation (Miłkowski et  al., 2019, 2018; Kaplan & Bechtel, 2011). (ii) Unification can also be understood as interfield unification which is understood as a relation between fields. A field is supposed to be an area of science with distinct features: it is usually smaller than a discipline, it exhibits a core problem or a significant target phenomenon, it contains domain-specific facts, includes explanatory goals as well as expectations how to solve the core problem, and it implies conceptual resources like concepts, laws, and theories (Darden & Maull, 1977, p. 44). We can take situated cognition research as an example of a field: it focuses on a core problem (“how is cognition – or a specific cognitive state – constituted?”), it contains facts about, e.g., the embodied grounding of cognitive states and processes, and involves expectations of how the core problem should be solved (e.g., cognitive states and processes are constituted by an interplay of decentralized entities and processes). Interfield unification does not refer to logical relations between theories but to the empirical connections between entities and processes that are analyzed by researchers from different fields. By bringing different fields in contact, a unification is envisaged that is first and foremost led by using insights from a specific field in another one. This unification is successful if it helps to understand better on which grounds the research premises are based that can be found in the research agenda of a specific field. The success of unification also requires a refinement of what is important in a field (e.g. modified focus of research, modified framing of research problems, improvement of explanation types) and the enlargement of the contextual knowledge concerning the phenomena a field is interested in (Grantham, 2004, p. 137). (iii) The last version is representational unification (Miłkowski & Nowakowski, 2021). Representational unification is about the unity of texts that explain a phenomenon. The explanation does not need to work just with language but can also include pictorial elements (like graphs and diagrams). Since representational unification sets a strong focus on how a phenomenon is explained, specific types of explanation need to be focused on for representational unification to work. Supporters of representational unification often select the mechanistic explanation type (Miłkowski, 2016a). Representational unification is presupposing four dimensions. The more of these dimensions are covered by the text(s) that are supposed to explain a phenomenon mechanistically the more unified they are: (a) Simplicity and non-redundancy: If texts involve superfluous complexity or unnecessarily repeat parts from other texts or sections of themselves, they are deemed too complex and redundant. (b) ­Generality and scope: the more phenomena can be explained by referring to a body of texts the more general, wider and unified the explanatory scope of an explanatory text will be. (c) Non-monstrosity: The text should not be a random collection of claims and inferences. To avoid unstructured eclecticism means to avoid monstrosity. (d) Systematization: using prestructured ways of phenomena description and aiming at completeness of explanations regarding address-

2  A Methodological Problem of Choice for 4E Research

25

able phenomena (Miłkowski & Nowakowski, 2021). Representational unity is about the coherent, elaborated, and principled application of a specific type of explanation. It is not discussed by proponents of representational unification if only approaches of mechanistic explanations can be unified in this representational sense or if other explanation types are existent and capable of being unified similarly. Both claims, that other explanations types might or might not be subject of representational unification, will need careful argumentation and elaboration. Prima facie, there do not seem to be strong arguments against the idea that also texts of dynamical, normative or ethological explanations can each be unified in the representational sense posed by Miłkowski & Nowakowski. These three presented versions of explanatory unification determine one side of the problem of choice. There might be more positions possible on this side of the spectrum. Still, to understand how the problem of choice develops, “only” the two sides of the problem need to be plausible regardless of how many positions such a spectrum might encompass. The other side of the problem is posed by an explanatory pluralism that also includes at least three subversions.

2.3.2 Explanatory Pluralism As a first and general approximation, “explanatory pluralism” means that different methods are available and applied in a particular research branch even if the methods might imply opposing premises about how a phenomenon under study is constituted. Although different types of explanation differ with regard to central theoretical and methodical commitments, they can still be part of the same research progamme (Shapiro, 2007; Lakatos, 1968) or research traditions (Laudan, 1989). On the one hand, explanatory pluralism can be conceptualized as a competitive pluralism. Such view describes the competitive situation different methods (theories, models, types of explanation) are exposed to and in which a certain set of epistemic procedures will supersede or marginalize all the other ones. If competitive pluralism describes the relation between methods correctly, then, eventually, one option or set of options out of all methods available becomes dominant. The slogan of such an approach is “Pluralism: The way to scientific unity” (see Mitchell, 2002). On the other hand, there is an alternative to competitive pluralism that can be labeled integrative pluralism. The idea is that, between different methods, division of labor is pursued. They might identify, describe, and explain distinct levels of a complex system or situation while their results might be put together to illuminate a concrete particular. Bringing together results is – in practice – more or less easily achievable because the connections made between the results do not need to live up to strict standards of integration. Everything that helps associate important facts, make unforeseen inferences or gain impetus in further research is appreciated. Problems emerge where not only the results but also different theories or types of explanation

26

M.-O. Casper

are supposed to be connected. Such connection “at the theoretical level is unlikely to be very robust” in integrative pluralism (Mitchell, 2002, p. 67) and it hence deals differently with the plurality of methods compared to competitive pluralism. Its aim is not to decrease the number of methods in order to reach a theoretical and practical unification of science. The available variety of methods is applied to gain ever more information about a phenomenon and to get of it the most comprehensive picture possible. In the following, we focus on integrative pluralism and its subversions because to understand the problem of choice we need to present pluralistic positions that are decidedly different from unificationist approaches (while competitive pluralism would be too close to unificationist positions to make the difference between pluralism and unification graspable). The following subversions of explanatory pluralism can be stated. (i) Pluralism by Co-Existence. Dale et al. (2009) propose that the necessity for more than one “theoretical paradigm for explanation” (p. 741) derives from the variety of organizational levels of a system. Such variety is (often) necessary to bring cognitive phenomena about. For example, it is frequently claimed that cognitive phenomena are complexly entangled in organism-environment-­ interactions (Gallagher, 2017a). If these interactions are supposed to be understood, different levels (lower-scale, meso-scale, macro-scale) should be considered that facilitate the interaction. Different levels or cluster of levels can be approached with different explanation types. Ethological and dynamical explanations are good examples here. Ethological explanations frequently require endogenous and exogenous aspects to cohere for a behavioral profile to be triggered. They involve morphological structures and mechanisms, the phylogeny, the ontogeny, and the survival value of behavior to understand why a behavioral profile is set up as it is (Bateson & Laland, 2013; Tinbergen, 1963). Dynamical explanations, i.e., mathematical descriptions of a (cognitive) system that can be used to predict how the system will behave in the near or long-­ term future (Nielsen, 2006), aim to find dynamical laws. Dynamical laws are the force of (regular) changes in observed variables. Both types do not share their level focus. However, their results can be used to illuminate a specific phenomenon. A behavioral profile can be seen as the phylogenetic outcome of an ongoing evolutionary process of a species while dynamical explanations cover the behavioral flexibility of a species mathematically. This means results that are brought together do not need to be results of various attempts to answer the same question. Furthermore, the types of explanation used to reach such results do not need to be integrated (for details see “conditions of integration” below in Sect. 2.4). They can simply co-exist. ­Pluralism by co-existence supports the idea that there are genuine alternatives of explanation types and that they do not necessarily need to interact with one another in order to be endorsed individually or in order to use their results to illuminate a concrete particular. The explanatory value of this kind of pluralism is to be found in the amount of information it can easily produce. To actually

2  A Methodological Problem of Choice for 4E Research

27

explain a piece of the world with this information is simply left to those who put the information together. (ii) Patchwork pluralism is based on a non-redundant use of multiple explanation types in order to account comprehensively for a (cognitive) phenomenon of interest. These are types of explanation that either allow representationalist vocabulary or avoid it altogether. The explanatory pluralism followed here is pragmatic in the sense that researchers should employ any type of explanation available, independent of whether it implies representationalist posits or not. Such posits are accepted as long as they shed light on a phenomenon (or its parts). Less philosophical ideology and less building of theoretical camps are as important as the free use of available explanation types to reach a specific explanatory aim with as many methods (such as explanation types) as possible. We can call this approach a patchwork pluralism since the different explanation types are not just co-existing and yielding insights that can be listed and then used to illuminate a concrete particular. Patchwork pluralism means that different types of explanation are employed at the same time and with regard to the same explanatory aim. Such a position supports the idea that there are genuine alternatives of explanation types and that they need to be applied concomitantly if the research question requires it. A good example of this subversion of integrative pluralism is the idea of “hybrid explanation” (Gervais, 2021). It allows the usage of different “explanatory inferences” or “explanation schemes” to account for a certain phenomenon – like explaining the phenomenon of motor control in an octopus that fetches or reaches for objects with one of its arms (Carls-Diamante, 2019). In the latter example, a mixture of types of explanation is used and presented as the best option to explain point-to-point arm movements of an octopus. Hence, it is an example of a patchwork pluralism that allows different explanation types to be applied simultaneously to account for a (cognitive) phenomenon. (iii) Cluster pluralism accepts the existence of different levels of systems without calling for a reduction of one level to another one. It also recognizes interlevel relations (between, e.g., the micro- and meso-scale of a system) and is hence an alternative to reductionism and eliminativism (de Jong, 2001). Cluster pluralism allows different types of explanations to be applied to illuminate a concrete particular under the condition that those types need to enhance each other. This kind of pluralism hence is not only about integrating results from different explanatory sources but changing premise or theoretical commitments that come along with one of the used tools of investigation. For example, in the case of the concurrent use of the dynamical and mechanistic  type of explanation one might wonder whether the non-decomposability commitment usually endorsed by those who apply the dynamical explanation type can still be incurred since mechanistic explanations require the decomposability commitment (Zednik, 2011). Whether the bracketing of the non-decomposability commitment or other modifications in other explanatory tools is an enhancement or not is in need for discussion.

28

M.-O. Casper

The position of cluster pluralism supports the idea that there are genuine alternatives of explanation types. However, a condition for cluster pluralism is that they both need to theoretically benefit from each other. If they do not generate more knowledge concerning each other’s premises, they will be part of other pluralisms. Pluralism by co-existence allows for association of results from different explanation types without any required interaction between these types. If they are directed at the same question but not develop, e.g., each other’s premises, then they are part of patchwork pluralism. If the research results are the consequences of applying several types of explanations and if the development of these explanation types is a sequel as well, then we have a case of cluster pluralism. How the mutual improvement of explanation types might be possible is elaborated on below in Sect. 2.4. It should become increasingly clear that the problem of choice consists in committing oneself to a subversion of explanatory unification or explanatory pluralism. Whoever supports the explanation-based unification by reduction will allow less types of explanations to be applied in 4E research compared to researchers who work in the context of patchwork pluralism. As a consequence, a methodology of the situated cognition branch will vary significantly when based on unification by reduction or patchwork pluralism. Whether the problem of choice really exists for 4E researchers apart from the above determined characteristics of unification and pluralism depends on how many explanatory methods, such as explanation types, are employed in the debate. The next section substantiates the claim that the problem of choice is an actual problem for 4E research.

2.4 Explanation Types in 4E Research The 4E debate is methodologically complex. The differences between “radical embodied cognitive science” (Chemero, 2013) and “predictive processing” (also known as “predictive coding”, Clark, 2015b, c) exemplify the difficult theoretical setting. While radical embodied cognitive scientists follow a non-representational research strategy and apply primarily dynamical explanations, i.e., mathematical descriptions of how a (cognitive) system will behave in the near or long-term future (Nielsen, 2006), they also generally support a pluralist account that allows different and incompatible methods to be employed. However, we find a possible contrast to this position already in a specific version of predictive processing. This theory claims that the brain generates predictive mental models about how the perceived scenery will unfold. There are researchers arguing that predictive processing involves representationalist posits and features mechanistic explanations (e.g., Hohwy, 2020; Gładziejewski, 2019). Employing mechanistic explanations is the attempt to explain a phenomenon decidedly different from dynamical explanations by referring to a material structure whose behavior constitutes the phenomenon under study (Craver, 2006). “Predictive processing” presents a unifying account for perception, cognition, and action that limits explanatory resources more strictly than the pluralist outlook of radical embodied science. However, non-­representational

2  A Methodological Problem of Choice for 4E Research

29

Fig. 2.1  Overview of available explanation types in the 4E debate

understandings of predictive processing are available (Kirchhoff & Robertson, 2018; Orlandi, 2018) as well and, if supported, subvert the just sketched opposition. Still, as long as the prominent representational interpretation of predictive processing is supported, 4E research involves theoretical and methodical oppositions. Furthermore, in addition to the just mentioned types of explanation, the dynamical and the mechanistic one, the situated cognition debate involves at least five more types. This figure gives a rough overview (Fig. 2.1). Similar incoherencies between 4E approaches, as just exemplified by the differences between radical embodied cognitive science and predictive processing, are numerous in the debate. Especially the (in-)compatibilities of mechanistic (Craver & Darden, 2013), dynamical (Lamb & Chemero, 2014), normative (Casper, 2019; Satne, 2015), teleological/etiological (Garson, 2011; Millikan, 1984), ethological (Jamieson & Bekoff, 1992), phenomenologically-inspired (Bruineberg & Rietveld, 2014; Gallagher, 2003, 2017b), and functional (Weiskopf, 2011; Wheeler, 2010) explanations are not fully determined, yet. Whether these (in-)compatibilities are of interest for a 4E methodology, again, depends on how the problem of choice is answered and what is understood as a proper explanation of something. Supporters of the explanation-based unification by reduction might claim that the mismatches between explanation types do not matter since there is only one serious candidate to explain cognitive phenomena. Supporters of cluster pluralism might say that how many types of explanation a 4E methodology should consider can only be estimated after a detailed examination is conducted that shows which of these types can get integrated. Before committing oneself to one or the other position, there is need for clarity regarding what counts as an explanation in the first place. While broader consensuses are very rare in philosophical debates, we find a strong consent regarding the distinctive feature of what makes an explanation explanatory. It is widely accepted that every explanation needs to exhibit explanatory asymmetry (Hausmann, 1993). This means that if A explains B, B cannot explain A. Consider the well-known flagpole example. Assuming that the pole is standing on an even surface, that light is emitted by a steady source in a specific angle from the distance, and that light travels in straight lines, then it is possible to calculate the pole’s height, its shadow length or the light’s angle of elevation, if two of these three variables are known (Bromberger, 1966). It is usually said that (i) the pole’s height and the light’s angle explain the length of the flagpole’s shadow but (ii)

30

M.-O. Casper

it is not the case that the same angle and the shadow’s length explain why the flagpole is as high as it is. It is usually agreed that the dependencies which we find in (i) are taken to be explanatory, whereas the dependencies in (ii) are non-explanatory. The reason often provided for this difference is that the dependency in (i) is asymmetrical because it tracks causal relationships whereas the dependencies in (ii) do not: the height is a cause of the shadow’s length, but the shadow’s length does not cause the height of the flagpole. This causal account of explanatory asymmetry is not the only one since also inferentialist approaches are being developed (Khalifa et al., 2021). How exactly explanatory asymmetry is constituted is secondary for the main issue of this article. Relevant is that it can be tested whether the seven above suggested explanation types really offer explanations. If these types include explanatory asymmetry, it is likely that they offer proper explanations. There are already articulated doubts that, e.g., phenomenologists can provide a sui generis explanation type (as claimed by  Albertazzi, 2019). Some claim that phenomenological explanations do not adhere to explanatory asymmetry, others say that they do but by the cost of merging with other explanation types and therefore phenomenological explanations are not sui generis (see Casper & Haueis, 2022 for a discussion of this issue). Hence, the availability of explanation types in the 4E debate might decrease (and therefore support unificationist approaches) if several of the employed types turn out not to match the explanatory asymmetry condition. Whether this is the case goes beyond the scope of this article. Discussions about this issue are, however, already ongoing. Assuming that such a scrutiny will still leave more than one explanation type available in the 4E context, we need to ask how those types should or can connect. This question is handled by introducing the spectrum of integration.

2.5 A Spectrum of Integration Two issues are now relevant when dealing with the problem of choice. We have seen that in addition to the mere plurality of explanatory tools in situated cognition research we are also confronted with a number of stances we can take towards such plurality (pluralism or unification strategies). The mere listing of these stances and their (sub-)versions, however, leaves us with the task of how to organize them. Is there a specific theoretical dimension along which we can arrange them, understand better their relations and differences? Once answered, the latter points structure a methodological study of 4E research. Moreover, the number of different explanation types presented above corroborates not only that the problem of choice is not an abstract issue. The different ways of explaining prompt, again, the specific question what this plurality is good for and how to handle it. Both issues, the organization of stances and the handling of methodical plurality, can be approached by pointing out a dimension along which explanatory pluralism and explanatory unification can be organized. Once they are organized, we understand how they relate to each other and how the choice of a stance directly shapes the role of methods in a possible methodology. This dimension is the intensity of

2  A Methodological Problem of Choice for 4E Research

31

integration that those stances demand from applied explanatory tools. As a rule of thumb, we can say: The less integration is required, the more genuine alternatives of explanatory tools we can find in a specific research branch. The more integration is required the less alternatives will be available. If we work with this dimension of integration, then it is possible to build a spectrum of integration in which we can place the (sub-)versions of explanatory pluralism and unification. The following hence has a twofold function. First, it pre-structures a possible 4E methodology and is therefore aiming at solving the motley crew argument by enabling methodological studies for the 4Es. Second, since there is no working account of what the integration of different explanation types exactly looks like (Taylor, 2021; Miłkowski, 2016a), the spectrum below is a step towards a scheme of viable, cross-explanatory integration and its conditions. The spectrum begins with the status “no integration” and ends up with “full integration” and can be pictured as follows (Fig. 2.2). The spectrum highly depends on the idea of “integration”. “Integration” is a largely debated term that can be applied to various kind of methods. In this chapter, integration of explanation types is understood as their extension of explanatory range and depth (cf. Taylor, 2021). The explanatory range can be determined by the number of addressable phenomena an explanation type can cover. Explanatory depth is defined by the details of a phenomenon (including the relation of different levels that might be necessary to constitute it) that an explanation type can consider. The more phenomena and details can be referred to, the more extended the range of explanation types will be. How strong such range and depth can be enlarged depends on how many conditions an explanation type needs to satisfy for such extension. An important premise at this point is that an explanation type most notably extends its

Fig. 2.2  Idealized depiction. Different positions of pluralism and unification are arranged along the dimension “intensity of integration”

32

M.-O. Casper

range and depth through integration with other explanation types. At this point, the problem of choice filters through again. Pluralist positions tend to accept the extension of explanatory range and depth with fewer conditions compared to unificationist positions that tend to increase the number. The number of conditions required for integration determines a position in the spectrum of integration. Hence, the problem of choice could be answered by discussing how many conditions are relevant when integrating different explanation types. The conditions of integration are distributed over several approaches of the pluralist and unificationist literature. The following is open to amendments while at the same time it is assumed that possible amendments do not essentially change the organization of pluralist and unificationist positions. The following five conditions can be found in the debate and be used to determine the intensity of integration between explanatory tools. While presenting the conditions it is ensured that they are already linked to the tools focused on in this chapter – explanation types: (i) Mutual increase of information. A step towards integration can be made by gathering the outputs of different explanation types. Those results are used to gain more knowledge about a target phenomenon without presupposing that the explanation types producing the results are mutually compatible or explicitly related. The intensity of integration, if only this condition is considered, is low since the explanatory tools are not necessarily linked and gathered information does not necessarily cohere. The most relevant point here is that details of a particular phenomenon are available and that these details might get paired up in context of a specific question the answer to which is not possible without them (Grantham, 2004). (ii) Shared heuristic and confirmational dependence. This means that data generated in the context of one explanation type may be used to confirm hypotheses produced in a different context (Grantham, 2004). This condition increases the intensity of integration since without an actual purpose (confirmation), the increase of information will simply be seen as ineffective. What is only optional in condition (i) – namely the usage of information opposed to the mere availability of information from different sources – is a requirement in condition (ii). The condition “shared heuristic and confirmational dependence” cannot be satisfied in the context of pluralism by co-existence, since explanation types and their results are not supposed to significantly interact with one another. However, it is satisfied within, e.g., patchwork pluralism. (iii) Inferential enhancement (Kitcher, 1984). Inferential enhancement can be but not necessarily is a bidirectional relationship between explanation types. Explanation type A may be able to elaborate on issues that are taken for granted or that are neglected by type B but which are relevant (e.g., type B focuses on entities and their role for a phenomenon but does not elaborate on how those entities are produced). Type A can further identify issues that were neither addressed nor known by B but which fall in B’s domain (problems with premises, new phenomena to be explained). While type A sheds light on neglected

2  A Methodological Problem of Choice for 4E Research

33

issues of type B, it explicates possible avenues of development for B and thereby inferentially enhance it. This explication can be a side effect of the work done with type A explanations and it is not mandatorily their determined focus. In some instances, the development of B due to results of A, can take effect on A as well. This reciprocal influence is possible although types of explanation do not necessarily share historical contexts, reference relations (different phenomena might be addressed), or explanatory goals. However, this mutual modification is not always the case (Hardcastle, 1992). (iv) Multi-level perspective. Beyond the idea that results from different explanation types can be used to solve a scientific problem, this condition demands that the information acquired from different explanatory sources is put together in such a way that an explanation of a target phenomenon spans various levels. These levels can cover the macro-level (such as behavioral patterns of an organism), the meso-level (when focusing on, e.g., functions of organic tissues such as cluster of neural cells), and the micro-level (e.g., workings of single neurons). This condition is partly a combination of condition (ii) and (iii) – but it is different from them since it requires not just information to be shared/ applied. The stronger postulation is made that all of this needs to be done to reach a coherent, multi-level overview of the relevant aspects and processes that constitute a target phenomenon. It is frequently assumed that such multi-­ level consideration is at its core mechanistic (Craver, 2019; Povich & Craver, 2017) and many follow that claim (Piccinini, 2020;  Miłkowski, 2016b). However, ethological attempts exist to explain behavioral profiles of organisms or groups of individuals by covering the function, the ontogeny, the phylogeny, and the mechanisms of behavior (Bateson & Laland, 2013; Bekoff, 1999; Tinbergen, 1963). Hence, the idea to fuse several constitution-relevant levels of a target phenomenon does not necessarily presuppose a mechanisms-only attitude. Ethology is a good example for how different explanation types (teleological, functional, mechanistic explanations) might get used to form explanations. Ethological, multi-level explanations might therefore also be framed as “hybrid explanation” (Carls-Diamante, 2019). (v) Conceptual refinement (based on Kitcher, 1984). Conceptual refinement is analogous to inferential enhancement. It does not concentrate on premises but on concepts used in the context of an explanation type. The definition of concepts and the entitlement to act that comes along with them (Brandom, 1994, Chap. 3) can be changed based on the outputs of empirical research that employs specific explanation types. These results, reached in the context of a certain explanation type, can highlight the need for inferential reorganization between concepts in the context of another explanation type. Also similar to inferential enhancement, conceptual refinements may be subject to reciprocity. A classic example of conceptual refinement is the modification of the Mendelian concept “mutation” through research in molecular genetics. The latter version of genetics, which first and foremost focus on biological microlevel mechanisms, led to a detailed analysis of the physical procedure of muta-

34

M.-O. Casper

tions and to elaborated classificatory systems of mutations that refined the original Mendelian concept, that was not in the same way explicitly concerned with micro-level mechanisms. This elaboration, however, allowed for new experimental settings to study biological inheritance and cell change. If such conceptual refinement between explanation types can be observed, a condition of integration between explanation types is satisfied. (vi) Shared commitments. Every type of explanation involves core commitments which are incurred by anyone who applies it. Such commitments are, so far, not studied for every explanation type used in situated cognition research. The mechanistic explanation is largely analyzed in this regard (e.g., Craver & Darden, 2013). Researchers who explain phenomena mechanistically are bound to heuristic guidelines such as localizability and decomposability. The potential integration between explanation types depends on the commitments that come along with them and whether they stand in opposition to each other. If it is the case that a type of explanation implies a non-decomposability commitment  – this means that a system of interest might not be decomposable without losing track of the phenomenon due to which researchers focus on that system (as in case of “dynamical explanations”, see e.g., Silberstein & Chemero, 2013) –, then it is obvious that an integration of such an explanation type with mechanistic explanation needs careful and detailed exploration. However, it is not necessary for explanation types to share commitments in the context of e.g. patchwork pluralism. Since for pluralists of that sort the results achieved by using different explanation types can be brought together to illuminate a concrete particular without further evaluation of the commitments involved by them. Commitment sharing is a presupposition for cluster pluralism or unity by reduction. Only if no opposition between core commitments is given, then reciprocal modification becomes possible (as in the case of “computational post-cognitivism” that is supposed to be possible due to the reworked representational commitment of computational approaches (Villalobos & Dewhurst, 2017)). The more of the mentioned condition of integration are shared and satisfied by types of explanation, the higher the intensity of their mutual integration will be. The more intense the integration is, the more intense the unification will be. Unification is therefore presented as an achievement or (interim) result of an integration process that first and foremost focuses on different explanation types.1

 It is important to note that Miłkowski (2016a) distinguishes between integration and unification in a different manner. While, for him, integration is the process of combining multiple explanations in a coherent manner, unification is the process of developing general, simple, elegant explanations. His account of integration is therefore different from the one in this article that suggests that the intensity of integration is a measurement for unification. 1

2  A Methodological Problem of Choice for 4E Research

35

2.5.1 Relation of Conditions and Stances The above mentioned conditions need to be related with the different versions of explanatory pluralism and unification. This helps to understand how strong the intensity of integration between explanatory tools, in this case explanation types, needs to be from perspective of those versions – in order to conceive those tools as proper ones. Below, the allocation of the conditions to the stances partly follows explicit statements of the researchers supporting the stances. However, not always can such statements be found. The distribution is straightforward. Each condition corresponds to a position in the spectrum of integration while the conditions get accumulated when wandering down the spectrum from “no integration” to “full integration”. Hence, pluralism by co-existence holds condition (i); mutual increase of information. Patchwork pluralism implies conditions (i) and (ii); the latter is shared confirmational dependence. Cluster pluralism exhibits conditions (i), (ii), and (iii); inferential enhancement. Go on like this until unification by reduction is reached that implies conditions (i), (ii), (iii), (iv – multi-level perspective), (v – conceptual refinement), and (vi – shared commitments). Unsurprisingly, the position requiring the strongest integration exhibits the highest number of conditions while the positions accepting lower intensity of integration accumulate less conditions.

2.5.2 Example of Cross-Explanatory Integration While cross-explanatory integration, i.e. the integration of different explanation types, might be a theoretically possible process, thorough examples of such integrations are hard to find. This section briefly presents an instance of a promising path of cross-explanatory integration. It is an integration example of dynamical and mechanistic explanations. The dynamical explanation type is understood as something that tracks the development of structures in complex systems (self-organizing networks with non-linear dynamics, showing collective behavior and pattern formation) (Silberstein & Chemero, 2013). The focus of attention for this type lies on the organizational network principles of those systems. How networks behave more or less uninfluenced by the detailed workings of their material small-scale features is a central point of interest; such networks are hence not decomposed to understand them. Oppositely, mechanistic explanations concentrate on individual mechanisms. Their aim is to locate and decompose material structures whose behavior constitutes a phenomenon of interest. Several attempts to integrate mathematical tools and concepts of dynamical explanations with the mechanist approach exist (Zednik, 2011; Bechtel, 1998; Clark, 1997). Especially Zednik highlights that there are no a priori reasons to assume that dynamical models and analyses are excluded from mechanistic ways of

36

M.-O. Casper

explaining. One of his points in this regard is that mechanistic explanation procedures imply identifications & descriptions of a mechanism’s parts, their organization and operations, and that a number of viable identification & descriptive schemes exist to achieve such description. One of these schemes is to be found in the area of dynamical explanations. It is claimed that the HKB model, a dynamical explanation of bimanual coordination (which is “a robust behavioral phenomenon in which the rhythmic oscillatory motion of two opposing index fingers spontaneously becomes coordinated in a way that depend on the frequency of the oscillator” (Zednik, 2011, p. 242)), includes structural decomposition to identify the working parts that realize bimanual coordination and a description of their spatiotemporal organization. Since decomposition and localization are heuristic core principles of the mechanist approach, procedures of dynamical explanations (viz structural decomposition and description of spatiotemporal organization) clearly overlap with the mechanist approach and are hence claimed to be mechanistic in their nature. As a further consequence, dynamical network analysis can be, under certain conditions, be integrated in the mechanistic framework. This idea, that the description and analysis of dynamical networks can work together with other explanatory strategies, is recently substantiated in a series of articles by Bechtel (especially Bechtel, 2020,  2019). Bechtel examines how researchers in system biology determine the components in cells that contribute to their own workings and self-maintenance as a whole. Such a successful determination is very difficult since cell life is dependent on proteins and protein-protein interactions. While proteins are significant parts of biological cells whose structure is partly determined by them, the specific role of proteins for triggering, maintaining or stopping cell mechanisms is rarely clear. Especially when proteins form temporarily semi-stable complexes, their influence on short- and long-term procedures in the cell is hard to underestimate although not completely understood. Even if no complexes are formed, a short contact between proteins can lead to an altered functioning of them, leaving the cell workings modified. This interaction between proteins and their influence on the cell is said to be hardly possible to investigate by detailed examinations of individual mechanisms. As a consequence, system biologists developed a strategy to track protein-­ interaction that is not entirely led by mechanistic premises. They track, on the basis of large datasets, how often, how long and how strong entities (such as proteins) interact with one another and develop a network representation of it. The resulting network models of these interactions are known as “node-edge representations of large networks”. They use the value of certain variables (such as epistatic interaction between genes) to quantify and relate interaction with each other. The important aspect of these network analyses is that they do not focus on the affiliation of the interacting entities to specific mechanisms. Outcome of such a research strategy are network models that demonstrate the multitude of interactions within a network independent from their material realization. This “relational data from which networks are constructed often only provides information about which nodes interact in the selected manner with each other and not additional information about the

2  A Methodological Problem of Choice for 4E Research

37

entities represented such as where they appear in the cell or what biological processes they are involved in” (Bechtel, 2020, p. 4). The heuristic principles of localization and decomposition are at least strongly diminished in the context of such network research if not declined. Instead, the heuristic of these analyses is the “guilt-by association” premise that relates entities within a network based on their interaction. Still, network research connects to mechanistic approaches since it considers the intensity of interactions between concrete, individual entities. “Clusters of interaction” are presented within large network representations (see Bechtel, 2019). Those clusters can also function as the epistemic basis for further hypotheses about involved mechanisms behind the interactions tracked. New parts of mechanisms, broader mechanisms or whole new mechanisms can be detected based on network research that does not work with mechanistic heuristics. Network research as presented by Bechtel is an example of cross-explanatory integration since, in the context of the spectrum of integration, the connection of the dynamical explanation (understood as the attempt to track the development of self-­ organizing networks) and the mechanistic explanation type satisfies three out of six conditions. The connection of both types in the manner described includes the increase of information about the phenomenon of interest (condition (i): mutual increase of information). It involves that information generated in the context of one type can be used in context of the other type (condition (ii): shared heuristic and confirmational dependence). In addition, explanations of one type might elaborate on issues that are taken for granted or that are neglected by explanations of the other type (condition (iii): inferential enhancement). The connection of mechanistic and dynamical explanations does not hold features such as having a multi-level perspective (condition iv) since it does not reach a coherent, multi-level involving overview of the relevant aspects that bring about a target phenomenon. It also misses the refinement of concepts (condition v) because network research does not explicitly criticize (or directly modifies) mechanistic terms applied in investigations of material structures. As shown above, node-edge network representations and therefore parts of dynamical explanations require other heuristic principles compared to mechanistic approaches. Hence, they do not share the same basic theoretical commitments (condition vi). The connection of network analyses and the mechanistic explanation type satisfies the first three out of six conditions of integration. Researchers who support cluster pluralism, as presented above, can see a proper cross-explanatory integration fulfilled in the connection of network research in system biology and the mechanistic approach. A similar perspective is valid for supporters of pluralism by co-existence or patchwork pluralism, but not for unificationist strategies since condition (iv)-(vi) are not satisfied. This circumstance is a nice illustration of the effect an answer to the problem of choice can have. Based on where researchers locate themselves in the spectrum, network research and mechanistic positions can coalesce. Since so much depends on the stance one commits oneself to in the spectrum, the next section addresses concisely whether a specific stance should be selected from where a possible 4E methodology can start.

38

M.-O. Casper

2.6 Where to Be in the Spectrum of Integration Is a specific position in the spectrum of integration to be favored? A thorough discussion of this issue requires more space than is left in this article. However, I succinctly present possible reasons that cluster pluralism is the most reasonable spot to take. A first but not entirely satisfying reason for that choice is that explanatory pluralism seems to be the default attitude for researchers in cognitive science, at least when roughly following the founding ideas of cognitive science as an interdisciplinary project (Miłkowski & Hohol, 2021). Further reasons to support cluster pluralism connect with the conditions in play sketched above. Stances such as pluralism by co-existence and patchwork pluralism that relate to condition (i), mutual increase of information, and (ii), shared heuristic and confirmational dependence, are methodologically too liberal. It is a positive aspect that they allow high flexibility concerning what counts as an explanation or an integration of different explanation types. Many resources that might get applied in situated cognition research follow this way of methodological categorization. However, at the same time, some might argue that these conditions alone facilitate monstrosity (Miłkowski, 2016a) and, in the light of (i) & (ii), explanation procedures become hardly distinguishable from epistemic luck. Hence, although they allow for interesting positions that are embodied by temporarily stable explanations for specific research problems, one should be skeptical about pluralism by co-existence and patchwork pluralism. In addition, condition (iii), inferential enhancement, and (vi), shared commitments, produce a philosophical tension if applied together. Succinctly put, inferential enhancement is problematic to achieve if all fundamental commitments are shared between explanation types. An essential aspect of inferential enhancement is that the results of an explanation type can have ramifications on premises or issues that are taken for granted in another explanation type. To achieve this critical but constructive attitude towards each other is hardly possible for different explanation types if they already share fundamental commitments. Accepted premises and issues are then actually not re-evaluated but incurred and applied. Therefore, inferential enhancement is a complicated issue for the stance introduced above as Unity by Reduction, which accepts conditions (i) – (vi). A similar issue appears for condition (iv), multi-level perspective, and therefore for the stance Interfield Unity. Combined with inferential enhancement, the condition of multi-level perspective presupposes inferential connections between explanation types that are too strong compared to the condition (iv). Inferential enhancement does not aim at coherent overviews of levels assumingly responsible for bringing about a phenomenon. Interfield Unity, a unificationist stance that presupposes everything from (i) to (iv), appears to be likewise problematic regarding the conditions of integration it involves. Although Representational Unification (a stance that involves integration condition (i) – (v)) is easily confronted with the same hitch as just described, the implied conceptual refinement (v) is, in principle, a

2  A Methodological Problem of Choice for 4E Research

39

possible sequel of inferential enhancement. Altered premises might influence the significance of concepts, as the example of the term “mutation” in the Mendelian context clarifies above. Researchers who want to defend unificationist strategies are, as a result, well advised to discuss the relation between inferential enhancement and multilevel perspective or shared commitments. As things appear in the context of the above sections, cluster pluralism should be the favored position to assume in the spectrum of integration as it is neither too liberal in accepting explanations of cognitive phenomena nor does it exhibit philosophical tensions between the conditions that define it. If such reasoning for cluster pluralism, that needs to be extended and substantiated in the future debate, is endorsed, then a methodology of 4E research should consequently check whether attempts to integrate assumingly different explanation types respect integration conditions (i) – (iii). It shows how explanations relate or could relate in future research to support claims of situated cognition research.

2.7 Conclusion While 4E research accepts a large variety of theories, hypotheses, and explanation procedures, it is notoriously hard to navigate methodologically. The presented spectrum of integration forces us to commit ourselves to methodological presuppositions that structure how we manage the complex methodical setting of situated cognition research. Such presuppositions are the conditions we expect of cross-­ explanatory integrations. Such integration is determined by six conditions of integration (mutual increase of information, shared heuristic and confirmational dependence, inferential enhancement, multi-level perspective, conceptual refinement, shared commitments) which, depending on how many of these conditions one accepts, define different positions in the spectrum of integration. These positions are pluralism by co-existence, patchwork pluralism, cluster pluralism, interfield unity, representational unity, and unity by reduction. This chapter hence displays the methodological intricacies required to be handled once the aim is to methodologically and methodically structure 4E research. However, the chapter focused solely on explanatory issues such as the relations of different explanation types. How an answer to the problem of choice influences the evaluation of other methodical tools such as data mining protocols, experimental design, etc., still needs to be discussed. The chapter further argued that cluster pluralism is the most reasonable position in the spectrum of integration since it is neither too liberal when it comes to cross-­ explanatory integration nor does it imply theoretical tensions between the conditions of integration claimed necessary for such integration. In total, these considerations aim to be important steps towards a much needed methodology of situated cognition research.

40

M.-O. Casper

References Albertazzi, L. (2019). Experimental phenomenology. What it is and what it is not. Synthese, 198, 2191–2212. Barker, M. J. (2019). Eliminative pluralism and integrative alternatives: The case of species. The British Journal for the Philosophy of Science, 70(3), 657–681. Bateson, P., & Laland, K. N. (2013). Tinbergen’s four questions: An appreciation and an update. Trends in Ecology and Evolution, 28, 712–718. Bechtel, W. (1998). Representations and cognitive explanations: Assessing the dynamicist challenge in cognitive science. Cognitive Science, 22(3), 295–317. Bechtel, W. (2019). Analysing network models to make discoveries about biological mechanisms. The British Journal for the Philosophy of Science, 70, 459–484. Bechtel, W. (2020). Hierarchy and levels: Analysing networks to study mechanisms in molecular biology. Philosophical Transactions of the Royal Society B, 375(1796), 20190320. Bechtel, W., & Abrahamsen, A. (2008). From reduction back to higher levels. Proceedings of the Annual Meeting of the Cognitive Science Society, 30(30). Retrieved (24.07.2023) from https:// escholarship.org/uc/item/8zn5q0j8 Bekoff, M. (1999). Cognitive ethology. In W. Bechtel & G. Graham (Eds.), A companion to cognitive science. Blackwell Publisher. Bich, L., & Bechtel, W. (2021). Mechanism, autonomy and biological explanation. Biology and Philosophy, 36, 53. Brandom, R. (1994). Making it explicit: Reasoning, representing, and discursive commitment. Harvard University Press. Bromberger, S. (1966). Why-questions. In R. Colodny (Ed.), Mind and cosmos: Essays in contemporary science and philosophy (pp. 86–111). University of Pittsburgh Press. Bruineberg, J., & Rietveld, E. (2014). Self-organization, free energy minimization, and optimal grip on a field of affordances. Frontiers in Human Neuroscience, 8, 599. Carls-Diamante, S. (2019). Make up your mind: Octopus cognition and hybrid explanations. Synthese, 199(Suppl 1), 143–158. Casper, M. O. (2019). Social enactivism. On situating high-level cognitive states and processes. De Gruyter. Casper, M.  O., & Haueis, P. (2022). Stuck in between. Phenomenology’s explanatory dilemma and its role in experimental practice. Phenomenology and the Cognitive Sciences, 22(7), 1–24. Chemero, A. (2013). Radical embodied cognitive science. Review of General Psychology, 17(2), 145–150. Clark, A. (1997). Being there: Putting brain, body, and world back together again. MIT Press. Clark, A. (2015a). Embodied prediction. Open mind. Frankfurt am Main. Clark, A. (2015b). Surfing uncertainty: Prediction, action, and the embodied mind. Oxford University Press. Clark, A. (2015c). Predicting peace: The end of the representation wars. A reply to Michael Madary. In T. Metzinger & J. M. Windt (Eds.), Open mind: 7(R). MIND Group. https://doi. org/10.15502/9783958570979 Craver, C. F. (2006). When mechanistic models explain. Synthese, 153(3), 355–376. Craver, C.  F. (2007). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. Clarendon Press. Craver, C. F. (2019). Levels of mechanisms: A field guide to the hierarchical structure of the world. In S. Robins, J. Symons, & P. Calvo (Eds.), The Routledge companion to philosophy of psychology (pp. 427–439). Routledge. Craver, C. F., & Darden, L. (2013). In search of mechanisms: Discoveries across the life sciences. University of Chicago Press. Craver, C., & Tabery, J. (2015). Mechanisms in science. Retrieved (01.04.2023) from https://plato. stanford.edu/entries/science-­mechanisms/

2  A Methodological Problem of Choice for 4E Research

41

Dale, R., Dietrich, E., & Chemero, A. (2009). Explanatory pluralism in cognitive science. Cognitive Science, 33(5), 739–742. Darden, L., & Maull, N. (1977). Interfield theories. Philosophy of Science, 44(1), 43–64. https:// doi.org/10.1086/288723 de Jong, H. L. (2001). Introduction: A symposium on explanatory pluralism. Theory & Psychology, 11(6), 731–735. Feest, U. (2012). Exploratory experiments, concept formation, and theory construction in psychology. Scientific concepts and investigative practice, 3, 167–189. Gallagher, S. (2003). Phenomenology and experimental design. Toward a phenomenologically enlightened experimental science. Journal of Consciousness Studies, 10(9–10), 85–99. Gallagher, S. (2017a). Enactivist interventions. Oxford University Press. Gallagher, S. (2017b). The past, present, and future of time-consciousness  – From Husserl to Varela and beyond. Constructivist Foundations, 13(1), 91–116. Gallagher, S., & Zahavi, D. (2020). The phenomenological mind. Routledge. Garson, J. (2011). Selected effects and causal role functions in the brain: The case for an etiological approach to neuroscience. Biology and Philosophy, 26, 547–565. Gervais, R. (2021). The multiplicity of explanation in cognitive science. Foundations of Science, 26(4), 1089–1104. Gładziejewski, P. (2019). Mechanistic unity of the predictive mind. Theory & Psychology, 29(5), 657–675. Goldinger, S. D., Papesh, M. H., Barnhart, A. S., Hansen, W. A., & Hout, M. C. (2016). The poverty of embodied cognition. Psychonomic Bulletin & Review, 23, 959–978. Grantham, T. A. (2004). Conceptualizing the (dis)unity of science. Philosophy of Science, 71(2), 133–155. Halonen, I., & Hintikka, J. (1999). Unification: It’s magnificent but is it explanation? Synthese, 120(1), 27–47. Hardcastle, V. (1992). Reduction, explanatory extension, and the mind/brain sciences. Philosophy of Science, 59, 408–428. Haueis, P. (2014). Meeting the brain on its own terms. Frontiers in Human Neuroscience, 8, 815. Hausmann, D.  M. (1993). Linking causal and explanatory asymmetry. Philosophy of Science, 60(3), 435–451. Hohol, M. (2021). Cognitive science: An interdisciplinary approach to mind and cognition. In B. Brożek, M. Jakubiec, & P. Urbańczyk (Eds.), Perspectives on interdisciplinarity (pp. 33–55). Copernicus Center Press. Hohwy, J. (2020). New directions in predictive processing. Mind & Language, 35(2), 209–223. Hutto, D.  D., & Myin, E. (2014). Neural representations not needed-no more pleas, please. Phenomenology and the Cognitive Sciences, 13(2), 241–256. Hutto, D.  D., Kirchhoff, M.  D., & Myin, E. (2014). Extensive enactivism: Why keep it all in? Frontiers in Human Neuroscience, 8, 706. Jamieson, D., & Bekoff, M. (1992). On aims and methods of cognitive ethology. PSA: proceedings of the biennial meeting of the Philosophy of Science Association, 1992(2), 110–124. Kaplan, A. (1962/2017). The conduct of inquiry. Methodology for behavioral science. Routledge. Kaplan, D. M., & Bechtel, W. (2011). Dynamical models: An alternative or complement to mechanistic explanations? Topics in Cognitive Science, 3(2), 438–444. Khalifa, K., Millson, J., & Risjord, M. (2021). Inference, explanation, and asymmetry. Synthese, 198, 929–953. Kirchhoff, M. D., & Kiverstein, J. (2020). Attuning to the world: The diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11, 1966. Kirchhoff, M.  D., & Robertson, I. (2018). Enactivism and predictive processing: A non-­ representational view. Philosophical Explorations, 21(2), 264–281. Kitcher, P. (1984). 1953 and all that: A tale of two sciences. Philosophical Review, 93, 335–373. Kiverstein, J., & Clark, A. (2009). Introduction: Mind embodied, embedded, enacted: One church or many? Topoi, 28, 1–7.

42

M.-O. Casper

Körner, A., Topolinski, S., & Strack, F. (2015). Routes to embodiment. Frontiers in Psychology, 6, 940. Krickel, B. (2018). The mechanical world. Studies in brain and mind. Krickel, B. (2020). Extended cognition, the new mechanists’ mutual manipulability criterion, and the challenge of trivial extendedness. Mind & Language, 35(4), 539–561. Lakatos, I. (1968). Criticism and the methodology of scientific research programmes. Proceedings of the Aristotelian Society, 69, 149–186. Lamb, M., & Chemero, A. (2014). Structure and application of dynamical models in cognitive science. Proceedings of the Annual Meeting of the Cognitive Science Society, 36, 809–814. Laudan, L. (1989). From theories to research traditions. In B. A. Brody & R. E. Grandy (Eds.), Readings in the Philosophy of Science (pp. 368–379). Prentice Hall. Machamer, P., Darden, L., & Craver, C.  F. (2000). Thinking about mechanisms. Philosophy of Science, 67(1), 1–25. Menary, R. (2016). Pragmatism and the pragmatic turn in cognitive science. In A.  K. Engel, K.  J. Friston, & D.  Kragic (Eds.), The pragmatic turn. Towards action-oriented viewers in cognitive science (pp. 215–234). MIT Press. Miłkowski, M. (2016a). A mechanistic account of computational explanation in cognitive science and computational neuroscience. Computing and Philosophy: Selected Papers from IACAP, 2014, 191–205. Miłkowski, M. (2016b). Unification strategies in cognitive science. Studies in Logic, Grammar and Rhetoric, 48(1), 13–33. Miłkowski, M., & Hohol, M. (2021). Explanations in cognitive science: Unification versus pluralism. Synthese, 199, 1–17. Miłkowski, M., & Nowakowski, P. (2021). Representational unification in cognitive science: Is embodied cognition a unifying perspective? Synthese, 199(1), 67–88. Miłkowski, M., Clowes, R., Rucińska, Z., Przegalińska, A., Zawidzki, T., Krueger, J., et al. (2018). From wide cognition to mechanisms: A silent revolution. Frontiers in Psychology, 9, 2393. Miłkowski, M., Hohol, M., & Nowakowski, P. (2019). Mechanisms in psychology: The road towards unity? Theory & Psychology, 29(5), 567–578. Millikan, R. (1984). Language, thought and other biological categories. MIT Press. Mitchell, S. (2002). Integrative pluralism. Biology and Philosophy, 17, 55–70. Morrison, M. (2000). Unifying scientific theories: Physical concepts and mathematical structures. Cambridge University Press. Muszynski, E., & Malaterre, C. (2021). A roadmap to explanatory pluralism: Introduction to the topical collection ‘The Biology of Behaviour’. Synthese, 199(1), 1777–1789. Nagel, E. (1961). The structure of science. Problems in the logic of explanation. Harcourt, Brace & World, Inc. Newell, A. (1973). You can’t play 20 questions with nature and win: Projective comments on the papers of this symposium. In W. G. Chase (Ed.), Visual information processing (pp. 283–308). Academic. Newell, A. (1990). Unified theories of cognition. Harvard University Press. Newen, A., De Bruin, L., & Gallagher, S. (2018). The Oxford handbook of 4E cognition. Oxford University Press. Nielsen, K. S. (2006). Discussions dynamical explanation in cognitive science. Journal for General Philosophy of Science, 37, 139–163. Orlandi, N. (2018). Predictive perceptual systems. Synthese, 195(6), 2367–2386. Piccinini, G. (2020). Neurocognitive mechanisms: Explaining biological cognition. Oxford University Press. Povich, M., & Craver, C. F. (2017). Mechanistic levels, reduction, and emergence. In S. Glennan & P.  M. Illari (Eds.), The Routledge handbook of mechanisms and mechanical philosophy (pp. 185–197). Routledge. Robbins, P., & Aydede, M. (Eds.). (2009). The Cambridge handbook of situated cognition. Cambridge University Press.

2  A Methodological Problem of Choice for 4E Research

43

Sarkar, S. (1998). Genetics and reductionism. Cambridge University Press. Satne, G. (2015). The social roots of normativity. Phenomenology and the Cognitive Science, 14, 673–682. Schaffner, K. (1993). Discovery and explanation in biology and medicine. University of Chicago Press. Shapiro, L. (2007). The embodied cognition research programme. Philosophy Compass, 2(2), 338–346. Shapiro, L. (2010). Embodied cognition. Routledge. Silberstein, M., & Chemero, A. (2013). Constraints on localization and decomposition as explanatory strategies in the biological sciences. Philosophy of Science, 80(5), 958–970. Stewart, J., Gapenne, O., & Di Paolo, E. A. (Eds.). (2010). Enaction. Toward a new paradigm for cognitive science. MIT 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. Taylor, S. D. (2021). Two kinds of explanatory integration in cognitive science. Synthese, 198(5), 4573–4601. Taylor, H., & Vickers, P. (2017). Conceptual fragmentation and the rise of eliminativism. European Journal for Philosophy of Science, 7(1), 17–40. Tinbergen, N. (1963). On aims and methods of ethology. Zeitschrift für Tierpsychologie, 20, 410–433. Villalobos, M., & Dewhurst, J. (2017). Why post-cognitivism does not (necessarily) entail anti-­ computationalism. Adaptive Behavior, 25, 117–128. Weiskopf, D.  A. (2011). Model and mechanisms in psychological explanation. Synthese, 183, 313–338. Wheeler, M. (2010). In defense of extended functionalism. In R. Menary (Ed.), The extended mind (pp. 245–270). MIT Press. Zednik, C. (2011). The nature of dynamical explanation. Philosophy of Science, 78(2), 238–263.

Chapter 3

A Professional Guide to Explanation. Commentary on “A Methodological Problem of Choice for 4E Research” Guilherme Sanches de Oliveira

On the face of it, the perspective that Casper puts forward in his paper (Chap. 2) and the one I offer in mine (Chap. 4) are completely at odds with each other. He sees the current diversity of explanations, theories and methods in embodied cognitive science as problematic and as calling for some kind of integration, whereas I defend an extreme pluralist stance and don’t see the diversity as a problem. Not only that, but while he argues for coordination between the disparate approaches, I propose that expecting such coordination is in many cases unrealistic and moreover that, even when it’s possible, it might be counterproductive. The straightforward thing to do now would be for me to double down and use this space to try to show why I’m right and Casper is wrong. But I don’t think he’s wrong. He and I do disagree, but I don’t think it’s that sort of disagreement where only one side can be right and the other must be wrong. In particular, I don’t think that focusing on the different conclusions he and I draw in our respective chapters is a fruitful way to understand the nature of our disagreement. My goal in this commentary will be to explain why. I don’t know if Casper will agree with my assessment or not, but I won’t be surprised if he doesn’t—in fact, as I will suggest, this would make perfect sense given what I propose in my chapter. Still, my hope is that, by clarifying how (from my perspective) our different views relate to each other, I can help some readers better appreciate both his proposal and mine. The starting point for Casper is the observation that “4E research” is characterized by great diversity when it comes to theoretical commitments, methods, hypotheses and explanation types, a diversity that makes the field “notoriously hard to navigate” (Chap. 2, Sect. 2.7). Casper concedes that this heterogeneity has its benefits, as it can lead to more varied insights on the same phenomena. Ultimately, however, he sees the situation as problematic. Above and beyond the potential benefits, “there are also neglected problems that come along with this plurality of disciplines” G. Sanches de Oliveira (*) Institute of Psychology and Ergonomics, Technical University of Berlin, Berlin, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_3

45

46

G. Sanches de Oliveira

(Chap. 2, Sect. 2.1), including, among others, the worry that, without integration, 4E research will remain a scattered minority in cognitive science: “fragmentations between the 4E positions are consolidated, leading to duplication of effort and lack of (mutual) understanding, a premature consensus within 4E positions, and a lack of productive disputes” (Chap. 2, Sect. 2.2). In Casper’s view, 4E needs a methodology: it needs a system (importantly, a single one) that guides research and shows how different theories, procedures, hypotheses etc. come together in productive ways. His chapter takes a first step toward this unitary methodology by identifying the “problem of choice” between different philosophical positions along the “spectrum of integration”—this is the set of philosophical positions ranging from those favoring unification by reduction, on one extreme, to pluralist positions favoring completely independent co-existence without integration, on the other extreme. Having thus described the options on the philosophical menu, Casper invites 4E researchers to make their choice. He strongly recommends a middle-­ground position he calls “cluster pluralism,” a position that “allows different types of explanations to be applied to illuminate a concrete particular under the condition that those types need to enhance each other” (Chap. 2, Sect. 2.3.2, emphasis added)—this position, he claims, is the one “to be favored” and the one that “should be selected” (Chap. 2, Sect. 2.5.2) because it’s “the most reasonable position” (Chap. 2, Sect. 2.7). In contrast, the starting point for my chapter is the observation that the starting point for philosophers writing on the diversity of explanations at play in science (including embodied cognitive science) is typically the idea that explanatory diversity is a problem that needs to be addressed. Some philosophers on one extreme believe that the problem should be solved by reduction of competing explanations to a single one (effectively doing away with the diversity), others somewhere in the middle hold that the way to go is to integrate or at least coordinate different explanations (thereby showing the diversity to be superficial or at least only partial), while others falling toward a pluralist extreme usually work out different reasons for continued heterogeneity (in practice addressing the problem by giving some justification for why the diversity should be seen as beneficial). Still, despite disagreeing in these ways, what these different philosophical positions agree on is that explanatory diversity is problematic and needs to be confronted somehow. In my chapter I sketch a different approach to explanatory diversity, viewing it not as a problem to be tackled, but a phenomenon to be understood. Here is what I mean by that. People trained in different research traditions, coming from different conceptual backgrounds, working with different tools and methods, sometimes doing different types of research (theoretical or empirical, quantitative or qualitative, and so on), often disagree with each other on what they see as the best way to explain some of the same phenomena—this heterogeneity is a fact. But rather than adopting the usual normative philosophical stance on this fact (i.e., arguing about what should be done about it), my intervention is to attempt to naturalize it, that is, to show how the fact of diversity or heterogeneity makes sense given the sorts of things that we humans are. I do this by arguing that, given a broadly embodied, situated perspective on cognition as a feature of entire brain-body-environment systems, if we “reflexively” (technical term) apply this perspective towards understanding the

3  A Professional Guide to Explanation. Commentary on “A Methodological Problem…

47

explanation-seeking practices that humans engage in when doing science, then it is only natural that people in different “situations” (another technical term) will disagree in the way that we see 4E researchers disagreeing. I conclude by proposing that this reflexive turn motivates a kind of pluralism—but I will come back to this point later. What I’ve said so far is enough to highlight what I see as the most interesting difference between Casper’s view and mine. Above I described Casper’s “spectrum of integration” as a philosophical menu. Casper himself doesn’t use this culinary analogy, but I think it can be helpful here. The idea of a menu suggests the need to weigh the pros and cons of different options in order to decide in favor of one or another (this is the “problem of choice,” as Casper calls it). So, a superficial reading could suggest that Casper and I disagree because we argue in favor of competing options on the menu: he recommends an integrative/coordinative option, while I favor a more extreme kind of pluralist option. But that’s not how I see it. Underlying the perspective I put forward in my chapter is the intuition that this disagreement between options takes too narrow a view on the phenomenon in question. The usual normative philosophical focus is on the variety of options on the menu and the criteria for deciding what the best option is: should we maximize flavor, maximize nutritional value, maximize long-term satiation, a compromise between these options, or something else entirely? But I don’t think there can be a right answer to this question. The naturalized, reflexive perspective I advocate suggests that there’s no single best option in absolute terms, because different options might be reasonable for different people in different contexts. Not only that, but it motivates going further and thinking not just about “food” (or “explanations”) as objects to be consumed but rather thinking in terms of “eating” (or “explaining”) as a complex set of activities that includes the objects but goes far beyond them. Besides what you eat, there is the how: knife and fork, chopsticks, bare hands? sitting at a table, sitting on the floor around a shared pot, standing up? alone or with others? quickly while multitasking or slowly, intentionally savoring each bite? and so on. My worry is that disagreements about the criteria for picking “the best” option on the menu neglect the fact that different people choose different things for different reasons, and that sometimes we don’t all have to be eating the same things for the same reasons to have a decent time together. This is why the specific type of pluralism I recommend in my chapter is not a pluralism about explanations in general, but a pluralism about explanatory styles. That is, starting from the assumption of cognition as inherently embodied and situated (or situationed?), I think it makes sense to see scientists as embodied, situated cognitive agents for whom different styles of explaining may be more or less attractive given many different factors, crucially including the different conceptions that they may have developed over time of what needs explaining and how. And, unlike the image of a menu suggests, I don’t think there’s much choosing going on here: these different embodied and situated developmental factors inform our thinking to the point of shaping the range of options that are “live” and those that are “dead” for us, as James (1896) put it in a different discussion. This view obviously resonates with Kuhnian ideas about paradigms as well as more recent “perspectivisms,” as I

48

G. Sanches de Oliveira

recognize in my chapter—my point, however, is to suggest that we in the 4E field have independent reasons to think this way, because this view follows from how we understand people (including scientists) as embodied, situated cognitive agents. Does this mean that philosophers like Casper are wrong to engage in normative theorizing about criteria that make some options better than other ones? I really don’t think so. And my reason for thinking this is in line with what I’m saying about understanding scientists as embodied, situated cognitive agents. As I’ve alluded to above, in my chapter I articulate a shift away from thinking about “explanations” as objects that could in principle be compared and analyzed on their own, independently of their context and use, and instead I favor thinking about “explanatory styles,” that is, styles of situated explanation-seeking practices. Another image that I think can be helpful here is that of scratching an itch. If you’ve ever tried to scratch someone else’s itch or have them scratch yours, you know that it’s a perilous endeavor. Try as they might, the “scratcher” simply cannot know if they are succeeding unless the “scratchee” tells them so. What counts as a good scratch depends on the itch, and the person with the itch has privileged access to the itch and to what makes a satisfactory scratch: one person’s itch is not another’s. Broadly speaking, I think this is a good way to approach “explanation.”1 Explanations (as countable nouns) can take many different shapes and forms, both in terms of their material embodiment (whether they are verbal, pictorial, mathematical, etc) and in terms of their logic or functional structure, that is, how they work and what they do (whether they are deductive-nomological, statistical, mechanistic, historical, etc). But, in all of this heterogeneity, explanations are just the products of embodied, situated cognitive practices that aim to address very particular needs and answer particular questions. To talk about the objects (analyzing, comparing, and evaluating them against each other) without reference to the practices (and practitioners) that occasioned them is like categorizing and judging ways of scratching without taking into account differences between different itches. So in contrast with thinking about how different explanations relate to each other (which one is better, which can be integrated or not, etc), I think it’s important to ask how successful different explanation-seeking practices are on their own terms, that is, in light of what they were meant to accomplish, or what (and whose) itch they were meant to scratch. Only then is it possible to ask of the different products of these practices (i.e., the different “explanations” as sentences, drawings, equations or other objects) whether they can work together: first we need to know whether they are even trying to answer the same types of questions, that is, we need to know whether they are different attempts to scratch the same itch, or whether they target different itches altogether.

 This idea is not new, and dates back at least to the classical pragmatists: Peirce talks about “The irritation of doubt” as causing “a struggle to attain a state of belief” (1966, p. 99), while James describes “The transition from a state of puzzle and perplexity to rational comprehension” as being “full of lively relief and pleasure” (1879, p.  317); Dewey makes similar remarks (see, e.g., Luntley, 2016). 1

3  A Professional Guide to Explanation. Commentary on “A Methodological Problem…

49

My pluralism of explanatory styles is not only a pluralism about scratchings, but also a pluralism about itches: I’m in principle open to the possibility that different projects within embodied cognitive science (and definitely outside of it) have different interests and goals based on which they can best be evaluated as more or less successful—I don’t think there’s such a thing as purpose- and situation-neutral “explanatoryness”. My chapter’s call for consideration of different embodied cognitive situations can thus be seen as a call to entertain the possibility that people can have very different itches (rather than there being a single absolute, universal itch), and that, from the outside, we can’t reasonably tell someone that they are failing to scratch their own itch. This is why, to play with what Casper calls his “provisional guide to a 4E methodology” (Chap. 2, Abstract), I propose that we take a professional stance on explanation. My use of “professional” here is meant in the etymological sense of the term, having to do with a fundamental commitment you publicly affirm (i.e., “profess”) to some vocation (i.e., a calling). Historically, this first had the religious connotation of professing your faith or allegiance to a set of beliefs and/or institution, only later coming to mean a commitment to a trade or occupation (i.e., different “professions” in the modern sense). In my view, explanation is a fundamentally professional business: it has to do with deep-seated and wide-ranging embodied, situated commitments (conceptual, theoretical, practical, etc.) that inform the ideas that people have about what needs explaining and how to do it—and different people have different commitments, ideas, and so on. In contrast with normative philosophical attempts to define ‘explanation in general’ (that is, in absolute terms that apply universally across the board), the idea here is that what counts as “explanatory” depends on the commitments that people profess, the commitments that characterize their practice—and these can vary considerably between different people in different inquiry situations. Now here’s the key point. What I’ve just said about understanding explanation in terms of different attempts to scratch potentially different itches is what I had in mind in putting forward, in my chapter, my view about the differences between different approaches within embodied cognitive science. But now I see that the same point applies to the disagreement between Casper and me. Just as I think researchers in different corners of the 4E field might be attempting to scratch different itches, so it is with our views: the view that Casper holds (along with many other philosophers) of explanatory diversity as a problem to be solved (and in a certain way) corresponds to one itch, and the view that I sketch, of explanatory diversity as a phenomenon to be understood, corresponds to a completely different itch. The normative itch that’s dominant in philosophy involves theorizing about what should be the case independently of whether people agree with you. This normative mode of theorizing is what leads some philosophers to say things like nonlinear dynamical models only describe but don’t explain: this is because these philosophers have a normative ideal of what should count as an explanation (i.e., what makes an explanation truly explanatory) and they think that the way some scientists try to explain certain phenomena falls short from that ideal.

50

G. Sanches de Oliveira

Given this dominant normative itch, I think Casper’s chapter is spot on: his “provisional guide to a 4E methodology” gives compelling philosophical reasons for why greater coordination between disparate 4E projects would be beneficial, as well as compelling philosophical reasons in favor of his preferred version of integration/ coordination. My own naturalized, reflexive itch makes me wonder about whether— given the potentially very different itches that 4E researchers can have—philosophical argument will be sufficient to motivate different researchers to change. I’m not convinced that the “fragmentation worry,” for instance, can convince particular researchers (e.g., ecological psychologists, or phenomenologists) to incorporate theories and methods from other traditions (especially if they didn’t already see these as continuous with what they are doing), all in the name of furthering the 4E community as a whole. The same goes for Casper’s requirement that “different types of explanations (…) need to enhance each other” (Chap. 2, Sect. 2.3.2): I see the point broadly speaking and in the abstract, but, in the real world, I’m skeptical that it would convince particular researchers in particular inquiry situations to abandon theories and methods they were trained in just because these don’t enhance the explanations of others (e.g., ecological psychologists abandoning key tenets in the Gibsonian tradition or phenomenologists adopting quantitative, nonlinear dynamical methods, say), all just to satisfy supposed requirements for an integrated, coordinated 4E methodology. This skepticism makes me less enthusiastic about the whole normative endeavor: does my saying that “scientists should do XYZ” still make a sound if nobody listens? Still, none of this entails that the normative itch is not real, nor that it is wrong to scratch it. If you feel the itch, Casper’s provisional guide is an excellent starting point for figuring out how to scratch it. It’s just that this is not the only itch that a philosopher of embodied cognitive science can have, as I hope to have made clear. Of course, I don’t know if Casper will agree with me. The view I’m putting forward here will probably fail to scratch his itch—but maybe this is as it should be.

References James, W. (1879). The sentiment of rationality. Mind, 4, 317–346. James, W. (1896). The will to believe: And other essays in popular philosophy. Longmans. Luntley, M. (2016). What’s the problem with Dewey? European Journal of Pragmatism and American Philosophy, 8(VIII-1). Peirce, C. S. (1966). The Fixation of Belief. In Philip P. Wiener (Ed.), Selected Writing (Values in a Universe of Chance) (pp. 91–112). Dover Publications.

Chapter 4

Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates Pluralism Guilherme Sanches de Oliveira

Abstract  Explanatory diversity is a salient feature of the sciences of the mind, where different projects focus on neural, psychological, cognitive, social or other explanations. The same happens within embodied cognitive science, where ecological, enactive, dynamical, phenomenological and other approaches differ from each other in their explanations of the embodied mind. As traditionally conceived, explanatory diversity is philosophically problematic, fueling debates about whether the different explanations are competing, compatible, or tangential. In contrast, this paper takes the perspective of embodied cognitive science as its starting point and accordingly approaches explanatory diversity not as a problem to be solved, but as a phenomenon to be understood. Recent work has explored how the view of cognition as embodied motivates reflexively viewing science as a situated embodied cognitive practice. Here I argue that this reflexive turn motivates adopting a pluralistic stance when it comes to questions about theoretical and methodological disagreements. In particular, it motivates moving away from thinking in terms of explanations as disembodied entities that compete with one another, and instead thinking in terms of different explanatory styles as embodied practices of explaining, many of which might be legitimate and warranted independently of whether and how the explanations themselves relate to one another.

4.1 Introduction Scientists use a wide variety of approaches to explaining the world, and in many cases this diversity of explanations seems perfectly straightforward. It’s not surprising that researchers in different disciplines come up with different sorts of explanations when they are investigating entirely unrelated phenomena. Along these lines, there’s nothing particularly mysterious about the many differences between explanations in, say, G. Sanches de Oliveira (*) Institute of Psychology and Ergonomics, Technical University of Berlin, Berlin, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_4

51

52

G. Sanches de Oliveira

inorganic chemistry, entomology, cultural anthropology, and astronomy. It also seems intuitive that, over time, the dominant approach to explaining a given phenomenon changes and gets superseded by a novel, better approach, such as in the shift from geocentric to heliocentric models, or from explanations in terms of phlogiston to oxygen. What is particularly philosophically interesting is when different explanations targeting the same phenomena run in parallel and even flourish at the same time. This is precisely what we see in the sciences of the mind today. Some explanations of human behavior make exclusive reference to events happening in the brain, such as localized patterns of neural activity; other explanations focus on psychological features such as personality traits, or they focus on cognitive states and processes such as representational structures and computational algorithms, and they often do so without paying attention to neural goings-on; other approaches favor explanation in terms of supra-individual factors, such as the ones found in parts of social and environmental psychology, setting aside the individual neural or computational aspects of mind; and other approaches still seem to fall in neither of these groups. Explanatory diversity is also very clearly present within embodied cognitive science, where researchers coming from different traditions (e.g., dynamical systems theory, ecological psychology, enactivism, phenomenology, and neuroscience) often disagree in how they investigate the embodied mind. The existence of many different ways of explaining mind and behavior has traditionally fueled philosophical debate about how the different explanations relate to one another, and in particular about whether different explanations can be integrated or whether some explanations are more fundamental and can ultimately replace the others. From this common philosophical perspective, explanatory diversity is a problem to be solved. Some philosophers try to solve the problem by advocating for the reduction of one type of explanations to the other, a move in the direction of eliminating diversity or at least reducing the quantity of legitimate explanation types in a given domain of inquiry. Other philosophers try to show, instead, that the different types of explanations are compatible and complement each other, or that they are tangential and can coexist even without integration: in positions like these, the goal is to legitimize the continued existence of a plurality of explanations. But whatever the position in each case, the starting point seems to be the same: philosophers typically see explanatory diversity as problematic on its own, something that calls for resolution or justification, something that we need to address somehow. In contrast with this usual philosophical framing, and taking instead the perspective of embodied cognitive science as a starting point, this paper approaches the question of explanatory diversity not as a problem to be solved, but as a phenomenon to be understood. In embodied cognitive science we use a variety of theoretical and methodological resources to understand all sorts of behavioral phenomena in terms of cognition as embodied, situated, enactive, ecological and so on. In line with this, it makes sense to see scientists themselves as embodied cognitive agents and, accordingly, to understand all aspects of scientific practice as aspects of situated embodied cognitive activity. This idea has been described in the recent literature as amounting to a reflexive shift—that is, a shift where we continue using the same familiar theoretical and methodological resources we already use to

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

53

understand behavioral phenomena in the cases of “ordinary people,” but we reflexively apply those resources to develop an understanding of scientists and scientific practices, including ourselves and our research practices. The goal of this paper is to spell out the implications of this reflexive embodied shift for how we think about explanatory diversity. As I will argue, reflexivity motivates reframing theoretical and methodological disagreements: in contrast with the usual focus on explanations as disembodied entities that compete with one another, we instead move toward thinking of explanatory diversity in terms of alternative explanatory styles as embodied practices of explaining, of making sense of the world. This reflexive reframing, in turn, motivates thinking that multiple explanatory styles might be perfectly legitimate and warranted even if they are not compatible or integratable. The paper is organized into two parts, with two sections each. I begin in Sect. 4.2 with a broad characterization of the traditional positions in the philosophical debate surrounding explanatory diversity in science, which I illustrate with examples relating to research in embodied cognitive science. Given the paper’s overarching goal of arguing that reflexivity motivates pluralism, it is important to consider in greater depth the different types of pluralist perspectives that philosophers have developed. This is what I do in Sect. 4.3, where I focus on pluralism of explanatory styles as a compelling variety of pluralism. Having up to that point focused on perspectives on explanatory diversity coming primarily from debates in philosophy of science, I then move to the second part of the paper where we shift to a perspective informed by embodied cognitive science. There the proposal is that, independently of the philosophical reasons in its favor, pluralism of explanatory styles is particularly attractive for researchers in embodied cognitive science who adopt a reflexive stance. Section 4.4 begins to work toward this conclusion by considering reflexivity itself, and identifying efforts within embodied cognitive science to develop reflexive accounts of scientific practice. Section 4.5 then concludes by exploring how, having taken this reflexive shift, there is good reason to see pluralism of explanatory styles not only as an ally but as a natural consequence of our views. I close by discussing the philosophical and scientific implications of the paper’s proposal. Besides reorienting, for embodied cognitive science researchers, the philosophical debates about the relation between different explanatory approaches, these steps also open up new avenues for theoretical and empirical work that more self-consciously applies the tools from embodied cognitive science to understand scientific practice itself.

4.2 The Philosophical Problem of Explanatory Diversity The classical philosophical framing for the question of explanatory diversity was in terms of a debate concerning the unity or disunity of science. The common assumption was that the world comprises discrete levels of organization that account for the differences between distinct scientific disciplines: “Reality, in the modern conception, appears as a tremendous hierarchical order of organised entities, leading, in a superposition of many levels, from physical and chemical to biological and

54

G. Sanches de Oliveira

sociological systems” (von Bertalanffy, 1950, p. 164; cited in Oppenheim & Putnam, 1958). This basic picture of the world as a nested hierarchy of parts that decompose into smaller parts that decompose into even smaller parts motivated two sorts of responses. Reductionists saw the unity of science as an ideal to be accomplished through the reduction of the vocabulary and explanatory principles used at the higher levels to those used in the lower and more fundamental ones; in contrast, anti-reductionists believed instead in the “autonomy” of explanation at higher levels in the so-called “special sciences” (see, e.g., Oppenheim & Putnam, 1958; Putnam, 1967; Fodor, 1974, 1997). More recent philosophical debate about explanatory diversity in the sciences of the mind has moved away from discussions about the prospect of unification of all of science, yet talk of “levels” has remained central for reductionists and anti-­ reductionists alike. On one side of the contemporary debate, for example, John Bickle’s “ruthless reductionism” sees the explanations provided in cognitive psychology and even cognitive neuroscience as useful only when it comes to “phenomena for which we don’t yet have experimentally verified cellular and molecular mechanisms” (Bickle, 2005, p. 237); ultimately these “higher level” explanations are steps “in the search for lower level neuronal mechanisms” (Bickle, 2003 p. 130) as the goal is always to go “‘further down’ to cellular, synaptic, and ultimately molecular biological mechanisms” (p.  4). In contrast, on the other side, anti-­ reductionists typically also appeal to differences in levels but they do so for the opposite goal, namely to justify the current existence, and permanence into the future, of explanatory diversity. Within this camp, some anti-reductionists advocate for integration, holding that good explanations need to combine contributions from multiple disciplinary perspectives and their respective ontological levels. Along these lines, for example, Paul Thagard claims: I expect that computational models will increasingly aid our understanding of the relations between different levels of mechanisms—for example, helping to relate the social to the cognitive, the cognitive to the neural, and the neural to the molecular. (Thagard, 2005, pp. 218–219)

Other anti-reductionists instead highlight the independence of the different approaches relative to one another and their individual value as legitimately explanatory: it is plausible to think that there are as many levels of description available as there are levels of organization. (...) We may describe humans in terms of their physical make-up, their chemical constitution, their physiological structure, their gross anatomy, their cognitive capacities, their social role, and much more. All these give a distinctive perspective on human psychology. (Richardson, 2009, p. 474; emphasis added)

Anti-reductionists of different orientations thus coincide in rejecting the idea that neuroscience (or, more specifically, cellular and molecular neuroscience, in the case of ruthless reductionists like Bickle) is “the” science of the mind, the one that all explanations should ultimately boil down to. But the question remains, even for anti-reductionists, of just how much independence we take the different explanatory approaches to enjoy: this is a question about the prospect of integration between

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

55

different explanations (rather than of reduction of one to the other), which concerns when and why combining different explanations and explanatory approaches might be possible, justified, useful, and desirable.

4.2.1 Explanatory Diversity in Embodied Cognitive Science It’s in the context of these philosophical ideas and positions that debate about explanation in embodied cognitive science—the focus of this paper—has unfolded in recent years. Broadly construed, embodied cognitive science eschews brain-centric reductionism and instead views cognition as embodied, embedded, situated, extended, enactive, ecological, and so on.1 This has occasioned debates of two types. On the one hand, there has been controversy about how the explanations provided by embodied cognitive science sit with approaches such as the ones already mentioned which favor explanation in neural, psychological, computational, social or other terms. One example of this is the debate about the dynamical systems approach that many advocates of embodied cognition use. Here the disagreement centers on questions about the compatibility between dynamical explanation and representational/computational explanation, as well as on questions about the explanatory or descriptive nature of the dynamical approach vis-a-vis mechanistic approaches (see, e.g., Van Gelder, 1995, 1998; Bechtel, 1998; Chemero, 2000; Shapiro, 2010; Kaplan & Bechtel, 2011; Stepp et al., 2011; Zednik, 2011; Piccinini & Craver, 2011; Gervais, 2015; Favela, 2020a, b). On the other hand, besides debates about how embodied cognition relates to other explanatory approaches in the sciences of the mind, there is also internal disagreement, as researchers working within embodied cognitive science have debated amongst themselves about how our different projects and views relate to one another. One example of this is the current debate concerning the compatibility between ecological psychology and enactivism. This debate is often framed around tensions supposed to exist between the direct realist approach to perception developed in the Gibsonian tradition (Gibson, 1966, 1979; see also, e.g., Michaels & Carello, 1981; Heft, 2001; Turvey, 2018) and the constructivism of the Maturana and Varela tradition (Maturana & Varela, 1980; Varela et al., 1991; see also, e.g., Di Paolo et al., 2010, 2017; Gallagher, 2017). While some take these and other differences to make ecological psychology and enactivism irreconcilable, other researchers have suggested that enactivist explanation at the level of individual agents and agency can complement, and be integrated with, explanations from ecological psychology that focus on organism-environment relations at the species level (see, e.g., Fultot et al., 2016; Di Paolo, 2016; Heras-­Escribano, 2016, 2019; Baggs & Chemero, 2021; Ryan & Gallagher, 2020; Heft, 2020).

 The handbooks edited by Newen et  al. (2018), Shapiro (2014), and Calvo and Gomila (2008) offer a comprehensive picture of the diversity of research in embodied cognitive science. 1

56

G. Sanches de Oliveira

For a last example, some of the controversy surrounding work in (and inspired by) the phenomenological tradition can be seen as straddling the internal and external types of debates in embodied cognitive science. This is the case of the disagreement about neurophenomenology, for instance. Here the source of contention is the apparent tension in using the third-person methods of neuroscience (seen by some as necessarily reductive and outside the scope of embodied cognitive science) in conjunction with the study of subjective experience from an anti-reductionist stand point (see, e.g., Varela, 1996; De Preester, 2002; Lutz & Thompson, 2003; Bayne, 2004; Overgaard, 2004; Zahavi, 2010; Pollard, 2014; Gallagher, 2012, 2017, 2018). A related but clearly distinct line of research draws insights from the phenomenological tradition to make sense of phenomena studied empirically using the tools of dynamical systems theory rather than those of neuroscience. This approach avoids the tension between subjective experience and scientific measurement attending neurophenomenology, yet it gives rise to some of its own questions. Here, an example of an internal debate (i.e., fully within embodied cognitive science) concerns how best to apply and interpret the empirical notion of interaction-dominant dynamics (Holden et al., 2009), whether in terms of a Heideggerian view of tool use or in light of Merleau-Ponty’s views on sensorimotor unity and embodied interaction (see, e.g., Dotov et al., 2010, 2017; Sanches de Oliveira et al., 2019); naturally, this line of research is additionally subject to external criticisms such as the ones mentioned previously concerning the status of dynamical approaches more generally, i.e., whether they are explanatory or ‘merely descriptive’. As these examples reveal, debates surrounding the different approaches at play in embodied cognitive science research have by and large been shaped by the traditional philosophical perspective according to which the existence of explanatory diversity in a given domain is a problem to be addressed, either by doing away with that diversity (e.g., by showing that some explanations are more legitimate than others) or somehow by justifying the continuation of that diversity (e.g., by identifying compatibilities and complementarities between the different explanations). The goal of this paper, once again, is to offer a different perspective on explanatory diversity, one that’s reflexively informed by embodied cognitive science (rather than only applied to it) and that approaches explanatory diversity as a phenomenon to be understood. The outcome of this reflexive approach on scientific practice from an embodied perspective, I will propose, is motivation to embrace a pluralistic take on disagreement and explanatory diversity. But to be able to elaborate on this idea, we first need to consider pluralism in more detail.

4.3 Explanation and Pluralism(s) Pluralism is a kind of anti-reductionism. But what does this mean? There are different types of reductionism, and different views one may hold as a reductionist. Brigandt and Love (2017), for instance, distinguish between ontological, methodological and epistemic reductionism, which they identify as logically independent

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

57

but, in practice, often tacitly held in combination with one another. These differences aside, a common point of agreement for reductionists seems to be their evaluation of explanatory diversity as, at best, only temporarily useful: in this view, lower-level descriptions are more basic, more fundamental, and ultimately more explanatory than descriptions in terms of processes at higher levels, such that having access to a diversity of explanations and explanatory approaches is beneficial only insofar as it enables developing descriptions that are reductive and, for that reason, better positioned to get at what (supposedly) truly explains the phenomena. Pluralists reject this overall picture. As Steel (2004) puts it, the difference between the views is that “while reductionism asserts that there is a privileged, fundamental level of description, pluralism maintains that there are multiple autonomous perspectives from which the same phenomena can be studied” (p. 66). To be clear on the contrast, then, reductionists may favor reduction when it comes to the relations between different scientific disciplines and their methods, and they may, additionally or instead, favor reduction when it comes to aspects of phenomena studied in the same field or domain; pluralists, in contrast, are anti-reductionists in both senses. But agreement on the negative aspect (i.e., rejecting reductionism) doesn’t entail agreement on the positive details, and the recent literature in fact suggests that there are multiple ways of adopting a pluralist stance with regard to explanatory diversity—there is a plurality of pluralisms, if you will. This is precisely what van Bouwel (2014) proposes. Following Mitchell (2009), van Bouwel offers a taxonomy that distinguishes between five types of pluralism that vary in the degree of autonomy each assigns to different explanatory approaches and, relatedly, in how much emphasis each places on the prospect of integration between explanations from different research projects and disciplinary perspectives. Commenting on these distinctions, Gijsbers (2016) argues that the differences between the five types of pluralism boil down to a contrast between views that endorse integrationism and views that endorse isolationism. Integrative pluralists, on the one hand, uphold the ideal of combining explanations “into one unified explanation that is superior to each of the individual ones”; the radical view he calls isolationist pluralism, on the other hand, is “the view that [different] explanations all add to our understanding” even if they “cannot be combined into a single integrated account” (Gijsbers, 2016, pp. 1–2).2 Following Gijsbers (2016), then, while integrationists and isolationists agree that explanations at different levels enjoy some degree of autonomy relative to one another, integrationists hold that we ultimately gain explanatory power by combining the different levels of description, a view that isolationists reject in favor of continued separation between explanations. In the sciences of the mind, this can be seen as translating into a disagreement about the role of explanations at the neural  Given that the views differ primarily on how they evaluate the prospect of integration (in particular whether they see integrated explanations as necessarily superior), some have argued that integrationism should be seen as its own category, as a middle ground between reductionism and pluralism rather than as a type of pluralism (see, e.g., Brigandt, 2010). 2

58

G. Sanches de Oliveira

level. To be sure, both integrationists and isolationists would agree in thinking that explanation at the neural level is insufficient—this is because both reject brain-­ centric reductionism. Yet, the two views might part ways when it comes to the question of whether inclusion of the neural level is even explanatorily necessary. For integrationists the best explanations will likely incorporate explanations of phenomena at the neural level, which need to somehow be combined with explanations at the higher levels. Isolationists, on the other hand, hold that many different explanations may be perfectly legitimate even if it’s not possible to integrate them, and, accordingly, they may think that an explanation can be perfectly legitimate and successful even if it leaves the brain entirely out of the picture.

4.3.1 Problems with Integrationist and Isolationist Pluralisms Although explanatory integrationism and explanatory isolationism have their advantages, there are also significant philosophical drawbacks to each of them. Beginning with integrationism, a difficulty is that the ideal of non-reductive integrated explanations seems to be not only unrealistic but potentially even counterproductive. This is what Potochnik and Sanches de Oliveira (2020) suggest. While granting that combining insights from different explanations and disciplines can be incredibly helpful in many cases, they argue that philosophers should not be overly optimistic about explanatory integration—and not simply because integration is difficult to accomplish in some cases, but crucially because in many cases it could even be detrimental. As they propose, scientists typically aim not for completeness of explanation, but only for elucidating the causal patterns that are most directly relevant for answering the questions that occasioned their research in the first place. And pursuing integrated explanations can be unfruitful when, as they put it, “additional causal information occludes the causal pattern of immediate interest with details that may be causally important but are incidental to the immediate research aims” (Potochnik & Sanches de Oliveira, 2020, p. 1315). Scientists develop explanations to gain understanding of phenomena. Sometimes additional information can provide better understanding. But we can’t assume that integrating different kinds of explanations will always necessarily get us there, because sometimes too much information can hinder rather than support understanding. Integration can be a good thing, but integrationism—the normative view on integration as always good, the ultimate ideal—isn’t. Isolationism, on the other hand, seems problematic in at least two ways. The first is that, if the view is framed as a normative perspective that mandates the continued separation between research projects and disciplines, then isolationism is just plain unrealistic and out of touch with real-world science. Although disciplines remain independent and autonomous, there is an undeniable growing tendency toward collaborative initiatives bringing together not only different perspectives within a given field but even across research programs and disciplines. This means that, understood as a normative philosophical view about science (i.e., an -ism), isolationism

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

59

flies in the face of a wide range of good research being done today—and this suggests that it’s not the norm that scientists actually use to guide their own work. But, second, even if isolationism is more negatively framed simply as a rejection of integrationism, the reason for rejecting it matters. As seen earlier, the autonomy of the different explanations we find in science (including the sciences of the mind) is, for many philosophers, due to the independence between the levels of description and associated levels of organization that these explanations are concerned with. Pluralists of an isolationist bent typically hold that “there is a plurality of legitimate and autonomous levels of description and explanation of a given phenomena [sic]” (Steel, 2004, p. 55; see also, e.g., Richardson, 2009). But appeal to levels seems like a problematic basis for justifying pluralism. This is not only because there are so many different conceptions of levels at play in science (see, e.g., Craver, 2007; Eronen, 2013; Potochnik, 2017) but also because, for any given conception of levels, it is doubtful that it can in fact capture the differences between the phenomena focal to different explanations such that it can account for those explanatory differences. As Potochnik puts it, “no account of general, hierarchical levels of organization succeeds, nor do the metaphysical, explanatory, epistemic, or causal significances variously ascribed to levels of organization” (Potochnik, 2017, p. 176). Even if relationships of compositionality or of realization and implementation are common and unproblematic, it doesn’t follow that all of reality falls into “discrete, stratified levels” (ibid.), nor does it follow that the different types of explanation neatly correspond to each of the supposed levels (if there were any).

4.3.2 Pluralism of Explanatory Styles For the reasons just considered, a particularly attractive alternative to both integrationist and isolationist versions of explanatory pluralism is the pluralism of explanatory styles put forward by Potochnik and Sanches de Oliveira (2020). They summarize their account as follows: On our view of pluralism of explanatory styles, different approaches offer independent explanations, even of the same phenomena, in terms of the distinct causal patterns they focus on. The value of these explanations partly derives from their identification of broad patterns in causal influence; this distinguishes our emphasis on causal patterns from an emphasis on mere causal information or causal mechanism. This places the researchers who seek explanation center stage: Which kind of potential cause is of immediate interest to them is crucial in determining which kind of explanatory style is called for. (Potochnik & Sanches de Oliveira, 2020, p. 1316)

Two points are worth emphasizing. The first is that, unlike pluralisms about explanation such as integrationism and isolationism, the pluralism in question here is one about explanatory styles. More traditional pluralisms focus on the differences between competing explanations, and can thus seem to treat explanations as if they were self-contained entities that can be compared without consideration of their context and the questions they were designed to answer. In contrast, the pluralism in

60

G. Sanches de Oliveira

question here focuses on the diversity of styles of explaining at play in the different sciences and even within a given field. Different explanatory styles may favor different causal patterns. For this reason, even though different explanations of the same phenomena might differ, say, in the temporal and spatial scales they focus on, the justification for this diversity doesn’t depend on any notions of levels, but may instead be due to a diversity of causal patterns that are focal for the different researchers and research groups. This makes a big difference. Pluralists about explanation typically reject reduction to a fundamental level and, accordingly, ask which other levels might also be explanatory. Pluralism of explanatory styles proposes something radically different: “the question is not which levels are explanatory, but which explanatory style is called for given the potential causal pattern of interest” (Potochnik & Sanches de Oliveira, 2020, pp. 1317–1318). The second point has to do with how pluralism about explanatory styles differs specifically from the integrationist type of pluralism about explanations. As seen above, one challenge for integrationism is the fact that, for scientists, sometimes less is more: simpler explanations can be, and often are, more useful and illuminating than others that are more complete and more true but also more complex. Holding a pluralist perspective about explanatory styles makes it possible to circumvent this kind of challenge. Again, in this view explanatory diversity is understood as arising because scientists sometimes adopt different styles of explanation that favor distinct causal patterns even for explaining the same phenomena. This difference in the causal patterns of interest is at once a difference that concerns the phenomena in question (and the causal patterns they exhibit) and a difference concerning the scientists themselves (and their interests). This means that, put differently, pluralism of explanatory styles is an interpretation of explanatory diversity that explicitly takes into account both what’s being explained (the phenomena) and who’s doing the explaining (the scientists). This difference in focus shows why we can’t evaluate the prospect of integration of different explanations only in terms of the explanations themselves. Trying to combine causal information from different explanations can be counterproductive for some scientists if the information added occludes the information most directly relevant to the causal patterns that are of interest to those scientists; after all, the causal patterns are what the scientists are trying to understand, and they are what will lead scientists to see some candidate explanations as being more or less illuminating and explanatory. By emphasizing how different explanations sometimes focus on different causal patterns even when they target the same phenomena, pluralism of explanatory styles damps down the philosophical expectation that it is always beneficial to (try to) integrate explanations.

4.4 Reflexivity, Mind, and Embodiment The previous two sections focused on different views relating to explanatory diversity as discussed primarily in, and from the perspective of, philosophy of science. One way to proceed now would be to consider how those philosophical views can

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

61

be applied to disagreements in embodied cognitive science such as the ones reviewed in Sect. 4.2.1. Along these lines, we could draw from a particular philosophical perspective on explanatory diversity in order to adjudicate between two competing explanations in embodied cognitive science, or perhaps to argue instead that the explanations are compatible and integratable such that the diversity of explanations is merely superficial. This is the usual approach in philosophical discussions about embodied cognitive science (e.g., in philosophy of cognitive science), and I don’t think there is anything terribly wrong with it. But my goal here is to offer a different, less common perspective: rather than applying philosophical views about scientific explanation to make sense of the theories and methods of embodied cognitive science, the proposal here is that we work in the opposite direction, starting with views from embodied cognitive science and exploring how those views can inform our thinking about scientific explanation and the diversity we see in science. And the crucial first step here is to see how embodied cognitive science can inform our view of science, scientific practice and scientists—that is, how embodied cognitive science is, or can be, reflexive.

4.4.1 Reflexivity? There’s a very broad sense in which it’s reasonable to see embodied cognitive science as unavoidably reflexive, because reflexivity lies at the foundation of any and all scientific approaches to mind and behavior. This is a conclusion we might reach, for instance, following Kurt Danziger’s historical account of the rise of modern psychological categories: Psychological discourse presupposes that individuals have learned to relate to themselves in some sense as objects, that they have learned to introspect their own mental states, or self consciously monitor their own actions, or describe their own character traits in a common language. These are all examples of psychological self-objectification. (Danziger, 1997, p. 23)

In this view, engaging in scientific theorizing about the mind is in and of itself already a reflexive activity. This is clear by comparison with the use of scientific methods to study various phenomena in, say, chemistry, physics, or geology. Even if these phenomena are all related to human life in some way, the investigation in these cases does not typically involve “self-consciousness” or “self-objectification,” as Danziger put it, because the primary target of inquiry is not humans. In contrast, the sciences of the mind necessarily consist in a form of scientific self-reflection: in neuroscience, psychology and cognitive science—including embodied cognitive science—we employ the same broadly naturalistic, scientific explanatory lens used in other sciences, but we turn that lens to understanding human beings. But there is a different, more restricted sense of “reflexivity” used in the social sciences that is particularly relevant for us. In this narrower sense, reflexivity in the social sciences concerns not simply theorizing about humans in general, but rather, applying the theories and methods from the social sciences specifically to

62

G. Sanches de Oliveira

understand science, including social science itself. This perspective is nicely illustrated by David Bloor’s approach in the sociology of scientific knowledge: in this sense, in line with the way sociologists understand human behavior more generally, a sociological perspective on science is reflexive when it sees sociological “theories and methods as emanating from society, that is, as the product of collective influences and resources and as peculiar to the culture and its present circumstances” (Bloor, 1976/1991, p. 44).3 Understood in this narrower sense, as a specific and substantive theoretical and methodological approach (rather than in the broad sense seen in Danziger’s quote above), reflexivity is not something we can take for granted. No doubt, it makes sense to expect the sciences of the mind to have interesting things to say about scientific practices and scientific research. But it’s perfectly conceivable, and in fact it’s common, for research on human behavior to go on without taking this extra step: more often than not, researchers study many different human behaviors without explicitly addressing the specific behaviors that some humans engage in when they participate in scientific practices of theorizing, experimenting, explaining, and so on. So even if the sciences of the mind are always reflexive in the broad sense, research on mind, behavior and cognition does not by necessity include the narrower and substantive reflexivity characteristic of theoretical and methodological work that takes scientific practice itself as an object of investigation.

4.4.2 Reflexivity in the Enactive Tradition Within embodied cognitive science, there is growing interest in reflexivity, but substantive reflexive research is still rare. Some of the earliest examples of reflexive thinking appear in the work of the founders of enactivism, Humberto Maturana and Francisco Varela. In his essay Biology of Cognition,4 for example, Maturana explicitly sets out to give an account of cognition that also applies to what he calls the “observer,” that is, the cognitive agent who cognizes even about cognition. As he puts it: “The observer is a human being, that is, a living system, and whatever applies to living systems applies also to him” (1980, p. 8); and “The observer is a living system and an understanding of cognition as a biological phenomenon must account for the observer and his role in it” (p. 9). For Maturana this means that the same biological principles of self-organization, autonomy and adaptation that explain the simplest forms of life also explain how our functional organization “gives rise to such phenomena as conceptual thinking, language, and self-­ consciousness” (p. 6).

 For other discussions of “reflexivity” in the social sciences, see, e.g., Ashmore (1989) and Lynch (2000). 4  Originally written in 1970, Biology of Cognition was published as the first part of Maturana & Varela’s, 1980 book Autopoiesis and Cognition: The Realization of the Living. Citations refer to this edition. 3

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

63

Having developed an approach for making sense of cognition as a biological function of living systems, Maturana and Varela extend the same approach to speak of themselves as observers, a specific type of living system. But it’s interesting to see that their comments stay at a general level, applying to meta-cognitive, self-­ referential thinking of the type that all of us engage in when we contemplate our own thoughts, behavior and experience. And this is different from offering a way of making sense of the particular types of observation, understanding, explanation and experience that only some observers engage in when they cognize about cognition as scientists. Understanding meta-cognitive “self-observation” can be a helpful starting point, but it doesn’t automatically translate into a substantive account of the peculiar “self-observing” practices we engage in when we do work that contributes to the shared enterprise of the sciences of the mind. This early work is thus reflexive more in the broad sense (e.g., the one we find in Danziger) than in the narrower, specific one we are interested in (e.g., in analogy to the one used by Bloor and others in the social sciences). Only later—especially with (and following) Maturana & Varela’s, 1987 book The Tree of Knowledge: The Biological Roots of Human Understanding—are there clear steps in the direction of an enactive understanding of scientific practice as scientific, that is, one that sheds light on how doing science differs from other forms of thinking, solving problems and so on. There we find, for instance, a suggestive discussion about the nature of explanation. Maturana and Varela point out that their goal is to provide an account of cognition in terms of “bringing forth a world,” and so they ask how they can know that they have succeeded. In trying to answer their own question, they start by considering what it means to offer a successful explanation: “An explanation is always a proposition that reformulates or recreates the observations of a phenomenon in a system of concepts acceptable to a group of people who share a criterion of validation” (1987, p. 28). As they recognize, this conception of explanation is so broad that it applies equally to magical/superstitious thinking and to scientific explanations. The difference that makes an explanation scientific rather than magical, in their view, has to do with the “criterion of validation” at play in the system of explanations that scientists share. Their account comprises four different steps or conditions that can, but need not be, chronologically ordered in this particular sequence: (a) Describing the phenomenon (or phenomena) to be explained in a way acceptable to a body of observers (b) Proposing a conceptual system capable of generating the phenomenon to be explained in a way acceptable to a body of observers (explanatory hypothesis) (c) Obtaining from (b) other phenomena not explicitly considered in that proposition, as also describing its conditions for observation by a body of observers (d) Observing these other phenomena obtained from (b) (Maturana & Varela, 1987, p. 28). With this account on hand they can then answer the question of how to judge the scientific adequacy of their own (enactive) account of cognition as the activity of living beings bringing forth a world: they will know that they have succeeded in giving a satisfactory scientific explanation

64

G. Sanches de Oliveira when we have set forth a conceptual system that can generate the cognitive phenomenon as a result of the action of a living being, and when we have shown that this process can produce living beings like ourselves, able to generate descriptions and reflect on them as a result of their fulfillment as living beings operating effectively in their fields of existence. (Maturana & Varela, 1987, p. 30)

Based on this picture and other insights discussed in the book, in the afterword Varela expresses particularly clearly the reflexivity of their project when he describes its logical conclusion: the result is that of “tracing autonomy from the realm of the biological all the way up to the human—including the activity of scientists like ourselves” (1987, pp. 253–254). More recent contributions to the enactivist literature have acknowledged the role of reflexivity in enactivist thought and developed different aspects of the insights by Maturana and Varela just reviewed (see, e.g., Stewart, 2010; Bottineau, 2010; Di Paolo et al., 2017). Some of these can be seen as pursuing step d in Maturana and Varela’s formulation cited above, identifying additional phenomena (including those part of scientific practice) that follow from the enactive starting point. Other noteworthy substantive examples that focus on particular dimensions of scientific practice include work taking the enactive explanatory system to elucidate the nature of model-based research as an embodied practice (Rolla & Novaes, 2022) as well as work on scientific observation, understood as “socio-materially augmented” cognition and as “irreducibly self, other-, and world-involving” (Froese, 2022).

4.4.3 Situated Reflexivity In parallel to work in the enactive tradition, another example of a substantive reflexive approach offered in the recent literature in embodied cognitive science is the perspective of “situated reflexivity” (Sanches de Oliveira, 2023). This perspective draws from Dewey’s notion of “situation,” first, to put forward a transactional understanding of the embodiment of cognition and, based on this, second, to explore a way to reflexively make sense of the embodied cognitive activities at play in scientific practice. In embodied cognitive science it’s common to see cognition as a feature of brain body-environment systems. In this view, to understand cognition as embodied is to understand it as in principle not reducible to what goes on in the brain nor even in the body as such: cognition is always an interplay of neural and bodily resources situated in an environment. But what does it mean for cognition, as an embodied phenomenon, to be “situated”? Dewey argued that we aren’t “in” the environment in the same sense that you can have a coin “in” your pocket. We aren’t simply surrounded by space. Rather, we perceive, act and think in situations, and what we do, what we are and what we become is defined through transaction with objects and other people in the environment. Noting that “situation” and “interaction” are thus intertwined, Dewey claims: “An experience is always what it is because of a transaction taking place between an

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

65

individual and what, at the time, constitutes his environment” (Dewey, 1938a/1997, p. 43). Importantly, the situation is not just where cognition happens: to think this way is to assume (erroneously) that cognition is fully on the side of the organism and that the situation is fully on the side of the environment. Instead, Dewey explains, the situation arises from a transaction between subjective (or organismic) and objective (or environmental) conditions: “Any normal experience is an interplay of these two sets of conditions. Taken together, or in their interaction, they form what we call a situation” (Dewey, 1938a/1997, p. 42). So the situation is relational. But so is cognition itself, which cannot be understood purely in individualistic terms because it is inextricable from the (transactional) situation: “control of individual actions is effected by the whole situation in which individuals are involved, in which they share and of which they are co-operative or interacting parts” (Dewey, 1938a/1997, p. 53). In sum, then, the situation arises through transaction between subjective and objective conditions, and, as situated, so does cognition: “a qualitative and qualifying situation is present as the background and the control of every experience” (Dewey, 1938b/2008, p. 76). On the first level, then, this Deweyan perspective helps clarify what it means to think of cognition in terms of brain-body-environment systems: it means to think of cognition as constituted by a “situation” transactionally made up by organismic activity in, and variously constrained (i.e., enabled and restricted) by, the environment. This is illustrated in Fig. 4.1a. The point of situated reflexivity, then, is to take this way of thinking to the next level, and, from this embodied view of cognition in general, to extend the scope of consideration so as to account for the embodied cognitive activities of scientists, including researchers in embodied cognitive science. This is illustrated in Fig. 4.1b. From the perspective of situated reflexivity, if embodied cognition is always fundamentally “situated” in the sense of arising from a transaction between subjective and objective conditions, then it follows that we can’t straightforwardly extrapolate from theoretical or experimental observations of embodied cognition in the case of “ordinary folk” to draw definite conclusions about the activities and practices people engage in as scientists. But this is not because science is unique in having a special status compared to non-scientific activities. Rather, it’s because in some sense, every situation is unique: although there may be general principles at play in cognitive phenomena everywhere, the situated nature of cognition as an embodied phenomenon means that there’s no such thing as “situation-neutral cognition.” It follows that we can’t properly understanding problem solving, for instance, without taking into account how the situation shapes both what constitutes a problem, what counts as a solution, and what the process of solving the problem looks like, that is, how it unfolds in that kind of situation (on this, see also, e.g., Kirsh, 2009). The perspective of situated reflexivity thus opens up avenues for research focusing precisely on the situations that make up scientific practice in different disciplinary contexts, different research programs, and at the interface of science and society—after all, situations can differ not only between different scientists and scientific groups, but also between scientists and non-scientists in the general public, policy makers and other stakeholders. This means that reflexive thinking is not

66

G. Sanches de Oliveira

Fig. 4.1 (a) Illustration of the common understanding of cognition in terms of brain-body environment systems. Dewey’s transactional notion of “situation” helps elucidate how, as an embodied phenomenon, cognition is “situated.” (b) Taking seriously the situated embodiment of all of cognition motivates, reflexively, considering the cognitive situations at play in scientific practice, including our own practice as researchers in embodied cognitive science. See text for details. (Images from Sanches de Oliveira, 2023; used with permission)

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

67

the end point (such as when we conclude that the way we think about embodied cognition for people in general must also apply to scientists like us): rather, in situated reflexivity it becomes the starting point, the motivation for substantive theoretical and empirical work applying ideas and methods from embodied cognitive science to understand science, scientific practice, and scientists. Finally, something that situated reflexivity has in its favor is that, although specific and substantive, it is still broad enough to be attractive for researchers coming from different theoretical backgrounds and of different orientations and goals— researchers in potentially very different inquiry “situations.” The examples of reflexive thinking reviewed in Sect. 4.4.2 were all grounded exclusively in the enactive tradition. In contrast, situated reflexivity is in principle a “nonpartisan” embodied perspective. Perhaps the notion of situation at play here could be further fleshed out following the principles of ecological psychology, for instance, in terms of the ambient energy array that organisms with different developmental trajectories encounter through their education of attention. But this Gibsonian ecological framing is not mandatory: alternative enactive interpretations of the Deweyan notion of situation have been explored (see, e.g., Gallagher, 2017), and other applications following potentially different frameworks within embodied cognitive science seem possible as well. Research in embodied cognitive science inspired by situated reflexivity can, therefore, be taken in a direction that prioritizes explanation in terms of affordances, direct perception and ecological information, for instance, or alternatively in terms of sensorimotor contingencies, autopoiesis and sense-making. Either way, the perspective of situated reflexivity provides a specific entry point and angle from which to apply our conceptual and experimental tools toward investigating the situation-specific embodied cognitive practices of scientists, including ourselves.

4.5 From a Situated Reflexive Stance on Explanatory Diversity to Pluralism of Explanatory Styles As stated at the beginning, the goal of this paper is to sketch and motivate a new perspective on explanatory diversity in science inspired by ideas from embodied cognitive science. The existence of multiple explanations in a given scientific domain is traditionally seen as philosophically problematic, that is, something to be addressed and solved somehow. In contrast, this paper pursues an alternative approach, seeing explanatory diversity in science as a phenomenon to be understood, and in particular a phenomenon to be examined reflexively, through the lens of embodied cognitive science. The proposal is that taking the step toward situated reflexive research in embodied cognitive science (as reviewed in the previous section) provides a way to reconceptualize or reframe theoretical and methodological disagreements in science. The goal of this concluding section is to explore precisely what this reframing is, and how, in turn, it motivates a pluralist take on explanatory diversity.

68

G. Sanches de Oliveira

The clarified conception of the situation, in Deweyan terms as the interplay of objective and subjective conditions, sheds light on important ways in which researchers working in different projects, fields and disciplines can differ from one another. As embodied agents, individual scientists harness brain-body-world resources in particular ways that are specific to the situations they find themselves “in.” It thus follows that their inquiry practices and the products of those practices (e.g., particular descriptions, accounts, models, theories, laws) need to be evaluated against the background of the objective and subjective conditions out of which they arise. Some of these conditions have to do with whether the phenomena of interest do in fact exhibit the focal causal patterns that scientists in a given research project investigate—this is straightforwardly an objective condition. But other conditions have to do with the factors that make some phenomena (rather than others) be of special interest, as well as factors that make some causal patterns in those phenomena (rather than other patterns in the same phenomena) become focal. And these are neither purely objective nor purely subjective factors, but an interaction of both. That is, although individuals can have an idiosyncratic sense of curiosity that informs what sparks their interest, even this seemingly subjective aspect is not static and detached, but it changes over time through experience and interaction with other people, including through scientific training in particular research traditions. The suggestion, then, is that when theoretical and methodological differences arise in a given field or between disciplines, it becomes crucial to consider the extent to which those differences might be due to a situational incompatibility, that is, due to the fact that the parties disagreeing are operating in effectively distinct inquiry situations. Rather than assuming that the different approaches are competitors in a zero-sum epistemic game where only one can be right (more explanatory, more accurate and true, and so on) in absolute terms, each might be presenting a reasonable solution to the problems that arise in their respective inquiry situations. This view is, of course, in line with well-known social and historical analyses in philosophy of science such as Thomas Kuhn’s (1970) and many after him.5 The point, however, is to see that these aren’t simply unrelated philosophical insights that run parallel to the questions we are concerned with in embodied cognitive science. Instead, our own scientific commitments motivate understanding the difference between inquiry situations in terms of differences in the interplay of the subjective and objective conditions that shape how scientists as embodied cognitive agents determine certain research questions and attempt to answer them. Thinking carefully about our scientific commitments, then, should drive us toward thinking about explanatory diversity in pluralist terms, and, in particular, viewing the diversity of explanatory styles as potentially stemming from a diversity of embodied cognitive situations.  This includes, for instance, at least some of the different philosophical perspectives that have been labeled “perspectivism” (see discussion in, e.g., Giere, 2010, Teller, 2018, 2020, Massimi, 2022), although the fact that perspectivism is often framed in the context of debate about realism and antirealism is important to bear in mind: that debate, while related in interesting ways, is still distinct from and even tangential to the debate surrounding explanation and explanatory diversity that is our focus here. 5

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

69

In Sect. 4.3 we saw that there are good reasons for adopting a pluralist stance on explanatory diversity, and in particular, reasons for favoring pluralism of explanatory styles rather than other types of “explanatory pluralism.” But the reasons explored at that point in the paper all stemmed from debate within philosophy of science. Given this paper’s focus and orientation, what matters now is to see how pluralism of explanatory styles is well positioned to capture key consequences of situated reflexivity for embodied cognitive science researchers, and is therefore especially appealing for those of us working in embodied cognitive science.6 The consequences of adopting pluralism of explanatory styles for those of us inspired by embodied, situated reflexivity are twofold. The first is that this perspective helps reorient debates about competing explanatory approaches in the sciences of the mind. Traditional philosophical approaches, even pluralistic ones, tend to focus on particular explanations (i.e., the products of practices of explaining, such as particular propositions, physical or mathematical models, depictions and visualizations, simulations and so on) as abstract, disembodied, context-free entities and, from this starting point, they attempt to determine the compatibility or integratability of those entities. In contrast, thinking in terms of explanatory styles from an embodied, situated perspective motivates focusing on the practices and practitioners at work in the pursuit and generation of explanations, and to do so through consideration of the situations that occasion the different research projects in the first place. Different researchers are often in rather different inquiry situations, and the relative situation-specificity of their explanations makes it so that multiple cross-cutting explanatory styles may be legitimate and warranted even when it comes to investigations of the same phenomena. In more practical terms, this perspective enables researchers in embodied cognitive science to make sense of their critics, of the challenges they level against embodied cognition, and of the competing explanations they advance instead. Rather than brushing off the differences as due to a focus on different levels (of phenomena or of analysis) or, perhaps worse, brushing off competing explanations as plain wrong and misguided, the disagreement is reframed as possibly arising from a difference in inquiry situations for which different explanatory styles, even if radically incompatible, might be legitimate. Along these lines, for example, the continued success of some brain-centric reductionist approaches suggests that they must have at least some purchase on the way the world actually works and that they must also be addressing at least somewhat satisfactorily questions that, within that situation, call for answers. The converse is also true, of course: many explanatory approaches that focus solely on regularities at the level of organism-environment relations but make no reference to neural states and processes surely provide some empirical traction and some support for the scientists’ understanding of important aspects of the relevant phenomena. This doesn’t mean that brain-centric reductionism is the right way to go for  Naturally, this is not to say that cognitive scientists who don’t subscribe to a transactional embodied and situated view can’t or shouldn’t adopt pluralism of explanatory styles. Maybe they can and should, maybe not. The claim is simply that this pluralist stance is particularly attractive to those of us who are already working within embodied cognitive science given our specific commitments to thinking of cognition in embodied, situated terms. 6

70

G. Sanches de Oliveira

everyone nor that explanation at the ecological scale (i.e., focusing on organism-­ environment relations) is the right way to go for everyone. But neither does it mean that an integration of the two is what is needed. This is because ‘what is needed’ is necessarily anchored to someone in some situation. Given the reflexive embodied view of cognition in situations, the pluralist conclusion is that not everyone has the same needs, and not all scientific inquiry situations have the same types of problems with the same types of solutions. Consider recent work in Gibsonian, ecological approaches to neuroscience. Here, researchers develop an understanding of the neural aspects of behavior, yet they do so without giving up their primary focus on organism-environment relations, that is, without adopting brain-centric reductionism (see, e.g., Favela, 2014; Kiverstein & Miller, 2015; de Wit et  al., 2016; van Dijk & Myin, 2019; Raja & Anderson, 2019; Bruineberg & Rietveld, 2019). Developments like these illustrate how researchers might expand the scope of their explanations without shifting explanatory styles. Although they include investigation of certain types and aspects of neural activity, it doesn’t seem right to describe these research projects as integrating ecological and reductive neuroscientific insights: the inquiry situation for each camp (ecological and reductionist neuroscience) remains generally unchanged despite the local partial convergence, and accordingly so do their explanatory needs also remain largely unchanged. A similar move applies to disagreements within embodied cognitive science itself, such as in the cases described earlier in the paper, in the clash between the theories of ecological psychology and of enactivism, or between phenomenologically-­ inspired approaches that adopt neuroscientific methods, dynamical systems methods, or neither. Rather than simply comparing the resulting explanations and evaluating them (as if) in a context-free and situation-neutral manner, researchers would do well to consider whether, and to what extent, the seemingly competing theories and methods belong in similar inquiry situations: they may not, and may not be strictly competing after all. Settling this question would help determine, in cases of different explanations that target the very same phenomena, if they focus on different causal patterns in those phenomena: if so, then the different explanations might each be warranted as conducive to greater understanding (i.e., for someone, some community, some discipline etc.) even if integration is not viable or helpful. In fact, adopting a pluralist stance means that settling the question of how competing explanations and explanatory approaches relate to one another might, in some cases, have little to do with the prospect of integration. Succeeding in making sense of the differences doesn’t mean determining which competitor is right and which one wrong, nor that a combination of both is the preferable outcome. Adopting a pluralist stance with regard to explanatory styles and cognitive situations could just mean learning to carry on and co-exist peacefully and respectfully. This brings us to the second point. In addition to reorienting these and other debates (which, from a philosophical standpoint, are often framed in an adversarial manner, as competitors), this proposal has the additional consequence of directing attention to new or understudied issues in embodied cognitive science. As we have seen, adopting the perspective of situated reflexivity amounts to applying to

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

71

ourselves the same explanatory principles we employ when we understand “ordinary” cognition in terms of harnessing brain-body-world resources to solve problems in particular situations, i.e., in particular configurations of objective and subjective conditions. Taking a naturalized perspective on science as a human activity calls for further elaboration in ways appropriate for an embodied approach in the sciences of the mind. As a practice that is in continuity with other human practices, science can be studied theoretically and empirically just as much, and just as well, as any other aspect of human behavior. Depending on one’s theoretical inclinations, a possible path for this kind of project would be to apply in this domain insights related to wide computationalist and extended versions of embodied cognitive science such as the ones defended by Wilson (1994, 2004) and Clark (1997, 2008). A body of work already exists that aims to shed light on the distributed nature of the many cognitive systems at play in science (see, e.g., Giere, 2002, 2007; Giere & Moffatt, 2003; Hutchins, 2010, 2014; Nersessian, 2005, 2019). The path here, then, could be one of further elaboration, expanding on this body of work, or one of reacting to it and proceeding in novel directions. In contrast, for those of us working in anti-representational, ecological and enactive versions of embodied cognitive science, the path is less obvious but the payoff perhaps proportionally more significant. Scientific practices involve abstract reasoning, description and conceptual problem-solving, all of which can be seen as instances of so-called “representation-hungry” cognition (Clark & Toribio, 1994). Some interesting starting points exist, for example, in research on the embodied aspects of STEM education and science learning (see, e.g., Hutto et al., 2015; Abrahamson & Sánchez-García, 2016). Building upon initial findings such as these and following through with developing reflexive approaches within embodied cognitive science might therefore even help address the challenge of the “scaling up” of radical embodied approaches to account for so-called “high” or “offline” cognitive processes. Many proposals have been offered in recent years for addressing the scaling up problem in general (see, e.g., Haselager et  al., 2003; van Dijk & Withagen, 2016; Gallagher, 2017; Di Paolo et al., 2017; Casper, 2019; Bruineberg et al., 2019; Zahnoun, 2021; Sanches de Oliveira et al., 2021), but there is still very little work exploring exactly how these insights can translate into a specifically anti-­representational reflexive embodied understanding of science (see Sanches de Oliveira, 2022, 2023). Taking the situatedness of cognition seriously motivates viewing science—in line with the pragmatist roots of embodied cognition (Crippen & Schulkin, 2020)— as in continuity with the rest of life, and in fact as one among many behaviors humans engage in to address the demands of life. It also motivates understanding the cognitive aspects of science as “situated” in complex configurations of subjective and objective conditions: this means—in line with the phenomenological roots of embodied cognition (Käufer & Chemero, 2021)—understanding scientists as being “in the world” in a strong relational sense, radically unlike the weak internalist sense in which we might speak of a coin “in” a pocket or a ghost “in” the machine. As a result, taking the situatedness of cognition seriously motivates reconsidering scientific disagreements, reevaluating (and even empirically investigating) their

72

G. Sanches de Oliveira

roots, and adopting a pluralist attitude in the face of the diversity of explanatory styles we find in science. In this light, we begin taking into account the potentially radically different inquiry situations from which our critics might be raising challenges to our views, which motivates approaching those challenges with more grace and empathy. Of course, we would appreciate the same courtesy in return. Acknowledgements  I’m thankful to Louie Favela, Tony Chemero and Angela Potochnik for inspiring discussions of earlier versions of ideas that appear in this chapter, as well as to participants in the Methodology of Situated Cognition Research online workshop in 2021 for their helpful comments and questions.

References Abrahamson, D., & Sánchez-García, R. (2016). Learning is moving in new ways: The ecological dynamics of mathematics education. Journal of the Learning Sciences, 25(2), 203–239. Ashmore, M. (1989). The reflexive thesis: Wrighting sociology of scientific knowledge. University of Chicago Press. Baggs, E., & Chemero, A. (2021). Radical embodiment in two directions. Synthese, 198(Suppl 9), 2175–2190. Bayne, T. (2004). Closing the gap? Some questions for neurophenomenology. Phenomenology and the Cognitive Sciences, 3(4), 349–364. Bechtel, W. (1998). Representations and cognitive explanations: Assessing the dynamicist’s challenge in cognitive science. Cognitive Science, 22(3), 295–317. Bickle, J. (2003). Philosophy and neuroscience: A ruthlessly reductive account. Springer. Bickle, J. (2005). Precis of philosophy and neuroscience: A ruthlessly reductive account. Phenomenology and the Cognitive Sciences, 4(3), 231–238. Bloor, D. (1976/1991). Knowledge and social imagery (2nd ed.). University of Chicago Press. Bottineau, D. (2010). Language and enaction. In J. Stewart, O. Gapenne, & E. A. Di Paolo (Eds.), Enaction: Toward a new paradigm for cognitive science (pp. 267–306). MIT Press. Brigandt, I. (2010). Beyond reduction and pluralism: Toward an epistemology of explanatory integration in biology. Erkenntnis, 73(3), 295–311. Brigandt, I., & Love, A. (2017). Reductionism in Biology. In E. N. Zalta (Ed.), The stanford encyclopedia of philosophy. Metaphysics Research Lab, Stanford University, spring 2017 edition. Bruineberg, J., & Rietveld, E. (2019). What’s inside your head once you’ve figured out what your head’s inside of. Ecological Psychology, 31(3), 198–217. Bruineberg, J., Chemero, A., & Rietveld, E. (2019). General ecological information supports engagement with affordances for ‘higher’ cognition. Synthese, 196(12), 5231–5251. Calvo, P., & Gomila, T. (2008). Handbook of cognitive science: An embodied approach. Elsevier. Casper, M. O. (2019). Social enactivism. On situating high-level cognitive states and processes. De Gruyter. Chemero, A. (2000). Anti-representationalism and the dynamical stance. Philosophy of Science, 67(4), 625–647. Clark, A. (1997). Being there: Putting brain, body, and world together again. MIT Press. Clark, A. (2008). Supersizing the mind: Embodiment, action, and cognitive extension. Oxford University Press. Clark, A., & Toribio, J. (1994). Doing without representing? Synthese, 101(3), 401–431. Craver, C.  F. (2007). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. Oxford University Press.

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

73

Crippen, M., & Schulkin, J. (2020). Mind ecologies: Body, brain, and world. Columbia University Press. Danziger, K. (1997). Naming the mind: How psychology found its language. Sage. De Preester, H. (2002). Naturalizing Husserlian phenomenology: An introduction. Psychoanalytische Perspectieven, 20(4), 633–647. de Wit, M., van der Kamp, J., & Withagen, R. (2016). Gibsonian neuroscience. Theory and Psychology, 26, 413–415. Dewey, J. (1938a). Experience and education. Simon & Schuster. Dewey, J. (1938b). The Later Works of John Dewey, 1925–1953: 1938, Logic  – The theory of inquiry (Collected Works of John Dewey Series). Southern Illinois University Press. Di Paolo, E. A. (2016). Across the uncanny valley: The ecological, the enactive, and the strangely familiar. Constructivist Foundations, 11(2), 327–329. Di Paolo, E., Rohde, M., & De Jaegher, H. (2010). Horizons for the enactive mind: Values, social interaction, and play. In J. Stewart, O. Gapenne, & E. Di Paolo, E. A. (Eds.), Enaction: Toward a new paradigm for cognitive science (pp. 33–88). MIT Press. Di Paolo, E., Buhrmann, T., & Barandiaran, X. (2017). Sensorimotor life: An enactive proposal. Oxford University Press. Dotov, D. G., Nie, L., & Chemero, A. (2010). A demonstration of the transition from ready-to-hand to unready-to-hand. PLoS One, 5(3), e9433. Dotov, D., Nie, L., Wojcik, K., Jinks, A., Yu, X., & Chemero, A. (2017). Cognitive and movement measures reflect the transition to presence-at-hand. New Ideas in Psychology, 45, 1–10. Eronen, M. I. (2013). No levels, no problems: Downward causation in neuroscience. Philosophy of Science, 80(5), 1042–1052. Favela, L. H. (2014). Radical embodied cognitive neuroscience: addressing “grand challenges” of the mind sciences. Frontiers in Human Neuroscience, 8, 796. Favela, L. H. (2020a). The dynamical renaissance in neuroscience. Synthese, 199(1–2), 2103–2127. Favela, L. H. (2020b). Dynamical systems theory in cognitive science and neuroscience. Philosophy Compass, 15(8), e12695. Fodor, J.  A. (1974). Special sciences (or: The disunity of science as a working hypothesis). Synthese, 28, 97–115. Fodor, J. (1997). Special sciences: Still autonomous after all these years. Philosophical Perspectives, 11, 149–163. Froese, T. (2022). Scientific observation is socio-materially augmented perception: Toward a participatory realism. Philosophies, 7(2), 37. Fultot, M., Nie, L., & Carello, C. (2016). Perception-action mutuality obviates mental construction. Constructivist Foundations, 11(2), 298–307. Gallagher, S. (2012). On the possibility of naturalizing phenomenology. Oxford handbook of contemporary phenomenology, 4, 70–93. Gallagher, S. (2017). Enactivist interventions: Rethinking the mind. Oxford University Press. Gallagher, S. (2018). Rethinking nature: Phenomenology and a non-reductionist cognitive science. Australasian Philosophical Review, 2(2), 125–137. Gervais, R. (2015). Mechanistic and non-mechanistic varieties of dynamical models in cognitive science: Explanatory power, understanding, and the ‘mere description’ worry. Synthese, 192(1), 43–66. Gibson, J. J. (1966). The senses considered as perceptual systems. Houghton Mifflin. Gibson, J. J. (1979). The ecological approach to visual perception. Houghton Mifflin. Giere, R. (2002). Scientific cognition as distributed cognition. In P.  Carruthers, S.  P. Stich, & M. Siegal (Eds.), The cognitive basis of science (pp. 285–299). Cambridge University Press. Giere, R.  N. (2007). Distributed cognition without distributed knowing. Social Epistemology, 21(3), 313–320. Giere, R. N. (2010). Scientific perspectivism. University of Chicago press. Giere, R.  N., & Moffatt, B. (2003). Distributed cognition: Where the cognitive and the social merge. Social Studies of Science, 33(2), 301–310.

74

G. Sanches de Oliveira

Gijsbers, V. (2016). Explanatory pluralism and the (dis)unity of science: The argument from incompatible counterfactual consequences. Frontiers in Psychiatry, 7, 32. Haselager, W. F., Bongers, R. M., & Van Rooij, I. (2003). Cognitive science, representations and dynamical systems theory. In W.  Tschacher & J.  P. Dauwalder (Eds.), The dynamical systems approach to cognition: Concepts and empirical paradigms based on self-organization, embodiment, and coordination dynamics (pp. 229–241). World Scientific. Heft, H. (2001). Ecological psychology in context: James Gibson, Roger Barker, and the legacy of William James’s radical empiricism. Psychology Press. Heft, H. (2020). Ecological psychology and enaction theory: Divergent groundings. Frontiers in Psychology, 11, 991. Heras-Escribano, M. (2016). Embracing the environment: Ecological answers for enactive problems. Constructivist Foundations, 11(2), 309–312. Heras-Escribano, M. (2019). Pragmatism, enactivism, and ecological psychology: towards a unified approach to post-cognitivism. Synthese, 198(Suppl 1), 337–363. Holden, J. G., Van Orden, G. C., & Turvey, M. T. (2009). Dispersion of response times reveals cognitive dynamics. Psychological Review, 116(2), 318. Hutchins, E. (2010). Cognitive ecology. Topics in Cognitive Science, 2(4), 705–715. Hutchins, E. (2014). The cultural ecosystem of human cognition. Philosophical Psychology, 27(1), 34–49. Hutto, D. D., Kirchhoff, M. D., & Abrahamson, D. (2015). The enactive roots of stem: Rethinking educational design in mathematics. Educational Psychology Review, 27(3), 371–389. Kaplan, D. M., & Bechtel, W. (2011). Dynamical models: An alternative or complement to mechanistic explanations? Topics in Cognitive Science, 3(2), 438–444. Käufer, S., & Chemero, A. (2021). Phenomenology: An introduction (2nd ed.). John Wiley & Sons. Kirsh, D. (2009). Problem solving and situated cognition. In P. Robbins & M. Aydede (Eds.), The cambridge handbook of situated cognition (pp. 264–306). Cambridge University Press. Kiverstein, J., & Miller, M. (2015). The embodied brain: Towards a radical embodied cognitive neuroscience. Frontiers in Human Neuroscience, 9, 237. Kuhn, T.  S. (1970). In 2nd Edn (Ed.), The structure of scientific revolutions, volume 2 of International Encyclopedia of Unified Science. The University of Chicago Press. Lutz, A., & Thompson, E. (2003). Neurophenomenology integrating subjective experience and brain dynamics in the neuroscience of consciousness. Journal of Consciousness Studies, 10(9–10), 31–52. Lynch, M. (2000). Against reflexivity as an academic virtue and source of privileged knowledge. Theory, Culture & Society, 17(3), 26–54. Massimi, M. (2022). Perspectival realism. Oxford University Press. Maturana, H. R., & Varela, F. J. (1980). Autopoiesis: The organization of the living. Autopoiesis and cognition: The realization of the living, 42, 59–138. Maturana, H. R., & Varela, F. J. (1987). The tree of knowledge: The biological roots of human understanding. New Science Library/Shambhala Publications. Michaels, C. F., & Carello, C. (1981). Direct perception. Prentice Hall. Mitchell, S.  D. (2009). Unsimple truths: Science, complexity, and policy. University of Chicago Press. Nersessian, N. J. (2005). Interpreting scientific and engineering practices: Integrating the cognitive, social, and cultural dimensions. In M. Gorman, R. Tweney, D. Gooding, & A. Kincannon (Eds.), New directions in scientific and technological thinking (pp. 17–56). Erlbaum. Nersessian, N. J. (2019). Creating cognitive-cultural scaffolding in interdisciplinary research laboratories. In A. C. Love & W. C. Wimsatt (Eds.), Beyond the Meme: Development and Structure in Cultural Evolution (pp. 64–94). The University of Minnesota Press. Newen, A., De Bruin, L., & Gallagher, S. (2018). The Oxford handbook of 4E cognition. Oxford University Press. Oppenheim, P., & Putnam, H. (1958). The unity of science as a working hypothesis. Minnesota Studies in the Philosophy of Science, 2, 3–36.

4  Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates…

75

Overgaard, M. (2004). On the naturalising of phenomenology. Phenomenology and the Cognitive Sciences, 3(4), 365–379. Piccinini, G., & Craver, C. (2011). Integrating psychology and neuroscience: Functional analyses as mechanism sketches. Synthese, 183(3), 283–311. Pollard, C. (2014). Merleau-ponty and embodied cognitive science. Discipline Filosofiche, 24(2), 67–90. Potochnik, A. (2017). Idealization and the aims of science. The University Chicago Press. Potochnik, A., & Sanches de Oliveira, G. (2020). Patterns in cognitive phenomena and pluralism of explanatory styles. Topics in Cognitive Science, 12(4), 1306–1320. Putnam, H. (1967). Psychological predicates. Art, mind, and religion, 1, 37–48. Raja, V., & Anderson, M.  L. (2019). Radical embodied cognitive neuroscience. Ecological Psychology, 31(3), 166–181. Richardson, R. C. (2009). Multiple realization and methodological pluralism. Synthese, 167(3), 473–492. Rolla, G., & Novaes, F. (2022). Ecological-enactive scientific cognition: modeling and material engagement. Phenomenology and the Cognitive Sciences, 21(3), 625–643. Ryan, K. J., & Gallagher, S. (2020). Between ecological psychology and enactivism: Is there resonance? Frontiers in Psychology, 11, 1147. Sanches de Oliveira, G. (2022). From something old to something new: Functionalist lessons for the cognitive science of scientific creativity. Frontiers in Psychology, 12, 750086. Sanches de Oliveira, G. (2023). The strong program in embodied cognitive science. Phenomenology and the Cognitive Sciences, 22, 841–865. https://doi.org/10.1007/s11097-­022-­09806-­w Sanches de Oliveira, G., Riehm, C., & Annand, C. (2019). Bee-ing in the world: Phenomenology, cognitive science, and interactivity in a novel insect tracking task. In A. Goel, C. Seifert, & C. Freksa (Eds.), Proceedings of the 41st Annual Conference of the Cognitive Science Society (pp. 1008–1013). Cognitive Science Society. Sanches de Oliveira, G., Raja, V., & Chemero, A. (2021). Radical embodied cognitive science and “real cognition”. Synthese, 198(1), 115–136. Shapiro, L. (2010). Embodied cognition. Routledge. Shapiro, L. (2014). The Routledge handbook of embodied cognition. Routledge. Steel, D. (2004). Can a reductionist be a pluralist? Biology and Philosophy, 19(1), 55–73. Stepp, N., Chemero, A., & Turvey, M.  T. (2011). Philosophy for the rest of cognitive science. Topics in Cognitive Science, 3(2), 425–437. Stewart, J. (2010). Foundational issues in enaction as a paradigm for cognitive science: From the origin of life to consciousness and writing. In J. Stewart, O. Gapenne, & E. Di Paolo (Eds.), Enaction: Toward a new paradigm for cognitive science (pp. 1–31). MIT Press. Teller, P. (2018). Referential and perspectival realism. Spontaneous Generations: A Journal for the History and Philosophy of Science, 9(1), 151–164. Teller, P. (2020). What is perspectivism, and does it count as realism? In M. Massimi & C. D. McCoy (Eds.), Understanding perspectivism (pp. 49–64). Routledge. Thagard, P. (2005). Mind: Introduction to cognitive science. MIT press. Turvey, M. T. (2018). Lectures on Perception: An Ecological Perspective. Routledge. Van Bouwel, J. (2014). Pluralists about pluralism? different versions of explanatory pluralism in psychiatry. In M. C. Galavotti, D. Dieks, W. J. Gonzalez, S. Hartmann, T. Uebel, & M. Weber (Eds.), New directions in the philosophy of science (pp. 105–119). Springer. van Dijk, L., & Myin, E. (2019). Ecological neuroscience: From reduction to proliferation of our resources. Ecological Psychology, 31(3), 254–268. van Dijk, L., & Withagen, R. (2016). Temporalizing agency: Moving beyond on-and offline cognition. Theory & Psychology, 26(1), 5–26. Van Gelder, T. (1995). What might cognition be, if not computation? The Journal of Philosophy, 92(7), 345–381. Van Gelder, T. (1998). The dynamical hypothesis in cognitive science. Behavioral and Brain Sciences, 21(5), 615–628.

76

G. Sanches de Oliveira

Varela, F.  J. (1996). Neurophenomenology: A methodological remedy for the hard problem. Journal of Consciousness Studies, 3(4), 330–349. Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind: cognitive science and human experience. MIT Press. von Bertalanffy, L. (1950). An outline of general system theory. The British Journal for the Philosophy of Science, 1(2), 134–165. Wilson, R. A. (1994). Wide computationalism. Mind, 103(411), 351–372. Wilson, R. A. (2004). Boundaries of the mind: The individual in the fragile sciences. Cambridge University Press. Zahavi, D. (2010). Naturalized phenomenology. In S.  Gallagher & D.  Schmicking (Eds.), Handbook of phenomenology and cognitive science (pp. 2–19). Springer. Zahnoun, F. (2021). On representation hungry cognition (and why we should stop feeding it). Synthese, 198(Suppl 1), 267–284. Zednik, C. (2011). The nature of dynamical explanation. Philosophy of Science, 78(2), 238–263.

Chapter 5

Commentary on “Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates Pluralism” Mark-Oliver Casper

Chapter 2 and Chap. 4, Casper’s “A Methodological Problem of Choice for 4E Research” and Sanches de Oliveira’s “Explanatory Diversity and Embodied Cognitive Science: Reflexivity Motivates Pluralism”, seem to address the same issue: the plurality of methods in the 4E debate. On the one hand, my contribution considers this plurality a severe challenge for 4E researchers and their future studies  – since it implies the question of whether and how different methodical approaches and their results can connect on an interdisciplinary level. Sanches de Oliveira, on the other hand, intends to offer a description of this plurality under which such problems disappear. In this comment I point out four problematic aspects of his account. The “methodological problem of choice” requires 4E researchers to choose either a form of explanatory pluralism or explanatory unification for developing a methodology of the 4Es. The former rather accepts and justifies various methodical attempts – without establishing a hierarchy of relevance between them. The latter tends towards rejecting the equivalence of methods and seeks hierarchies of relevance. Depending on which side researchers join, the methodological assessment of 4E research will vary. Finally, those who do not react to the problem of choice seem to be of no help in developing a strongly needed methodology of the 4Es. Only such a choice establishes benchmarks for evaluating methods in the 4E branch and hence options to develop a methodology thereof. Sanches de Oliveira’s attempt to (dis)solve problems with explanatory plurality is twofold and based on the idea that we need to modify our perspective on it. (i) Instead of understanding explanations as if they were purely mental and non-­ material processes, we may modify our perspective and instead talk about embodied and situated practices of explaining. His line of reasoning unfolds in the following M.-O. Casper (*) Institute of Philosophy, University of Kassel, Kassel, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_5

77

78

M.-O. Casper

way: If at least some cognitive processes (such as explaining) are practices continuously realized in different research situations, then the relevance of problems frequently connected to explanatory plurality grows smaller or even dissolves because practices do not have clear-cut boundaries. This thought applies to explanatory practices as well. If no essential differences between explanatory practices can be identified, we hardly encounter a problem of choice since there are no clear options we can choose from. There are just different explanatory styles, different ways of realizing a practice according to a situation. No further distinctions are needed. In personal communication, Sanches de Oliveira highlighted (ii) that the talk of explanatory styles is also closer to real-life explanation procedures in the sciences. The problem of choice might be a problem in specific philosophical contexts. However, this problem is not necessarily a problem for researchers studying and explaining cognitive phenomena. Put differently, explanatory plurality can be seen as problematic depending on one’s philosophical stance. For him, the problems with such plurality are hence enacted by a specific philosophical style of thinking. Any theory of explanation that comes from the “explanatory style approach”, he claims, is philosophically humble, since it does not think that philosophy needs to structure other disciplines’ research, closer to actual research practices, and not focused on problems but on the dissolution of problems. Sanches de Oliveira’s approach has a solid philosophical character since it tries to change the whole frame of how we engage with explanatory plurality. It uses premises of 4E theories to analyze the cognitive procedures of studying cognition and the general constitution of cognition simultaneously. Concurrently, he tries to establish a position from where explanatory plurality is not seen as a knotty issue that requires work (which might bog down the studies on phenomena we are interested in as philosophers of mind – since we are forced to develop a 4E methodology first instead). Still, this comment raises four significant points that should be considered when explanatory diversity/pluralism is investigated:

5.1 The Problem of Bypassing the Question I think that establishing the talk of “explanatory styles” does not circumvent the problems often seen as accompanying explanatory plurality, especially the problem of choice. The change of terms from, e.g., “explanation” to “embodied, situated practices of explaining realized in specific styles” only modifies the number of phenomena connected to explanation procedures, but not the problem that comes with the plurality of explanatory strategies. A good example of this might be the following: There are explanation procedures that are strongly number-based (mathematical explanations (e.g., Lange, 2013; Baker, 2009)). However, there are also explanation procedures that are image-based or visual (Tufte, 1997). Let us now go on and call these different ways of explaining “explanatory styles”. We still can ask, “Do these explanatory styles work together?” and if they match and work together, then we ask further, “How is it possible that they work together?”. Another more

5  Commentary on “Explanatory Diversity and Embodied Cognitive Science…

79

general example to illuminate the point being made here is this: Sport includes different categories. There are people playing basketball and people who are fencing. Both practices are practices of sport or different styles of doing sport (if this way of putting it is permissible). Nevertheless, fencing and basketball are hardly compatible, and it is not a coincidence that we never see a fencer and a basketball player on the same court, joining the same team and trying to win a game together. We can draw from this that if we assume that different explanatory styles are not essentially different but different “only” because they refer to individual situations in which they are carried out, then we still do not know how those styles interrelate, whether they are a good fit for the questions we ask, and whether we need only one explanatory style or more to explain something. The issue is that the talk of explanatory styles still allows rephrasing the problem of choice. This is why the “explanatory style approach” bypasses the actual question regarding problems of explanatory plurality. It simply offers a possible reformulation of a problematic situation that remains unchanged under new names. A minor aspect connected to this problem is that if identifying different patterns of explaining is problematic, we struggle to differentiate between proper explanatory practices. If this struggle is plausible, then this problem might be reiterated on a more general level: How do we distinguish the explanatory practice from other forms of acting, such as discursive practice (that is not always in the business of explaining phenomena)? Is there a cluster of conditions implied by explanatory practices only they can satisfy and which simultaneously are satisfied by all different styles of explaining? If we cannot ascertain these conditions, we risk losing the phenomenon of explanation altogether.

5.2 The Problem of a 4E Precondition While Sanches de Oliveira is eager to reset the conditions under which something can be seen as a conflict of explanatory strategies, he bases his claims on 4E premises. The latter is an intelligible move. Suppose we want to establish a cognition theory or an approach to study cognition. In that case, we should be able to apply the theory to cognitive processes such as scientific research and explanations. This so-called “reflexive turn” is the attempt to make our ordinary and scientific thinking scientifically transparent for us. However, in Sanches de Oliveira’s view, the reflexive turn only makes sense if premises of the 4E framework are accepted. This is a more or less explicit commitment he requires us to incur so the talk of “explanatory styles” can reach its goal. If we do not accept those premises, we are not necessarily directed toward the specific material processes and social interactions he clusters as “an explanation of something in a specific situation”. Suppose, e.g., we accept a strong representationalism in cognition research. Then, we are not coerced to examine how explanations are carried out in the world because an explanation (a cognitive process) is a specific version of how information from the transcranial environment is being taken up and processed internally by a cognitive system.

80

M.-O. Casper

Whether Sanches de Oliveira’s claims are acceptable depends on a researcher’s tendency towards enactivism or similar positions. This aspect makes the talk of explanatory styles hard to embrace because it is inferentially connected to a bundle of assumptions regarding the decentralized constitution of cognitive phenomena. This bundle is usually not an integral part of a philosophy of science. There is one more aspect connected to the 4E precondition. As I pointed out in “A Methodological Problem of Choice” in this volume, a 4E methodology is a consequence of the motley crew argument (Shapiro, 2011). The argument essentially says that if someone thinks that the constituents of cognition are at least sometimes also located in the surroundings of our body (constituents are hence sometimes non-­ neural and non-bodily), then cognition researchers need to include many different entities when studying cognitive phenomena. It then becomes doubtful that scientific regularities of cognitive constitution can be identified and traced. The talk of “explanatory styles” is similarly constricted. What methods do we have to elaborate on practices of explaining, and how do those methods relate to each other? The talk of explanatory styles seems to invite the motley crew argument through the back door again. At the same time, supporters of the reflexive turn do not appear interested in countering the motley crew argument and hence allow a critical point to undermine the reflexive turn’s effort.

5.3 The Problem of a Small Vicious Circle Assume that the reflexive turn is a philosophical necessity to fully develop a theory of scientific research and explanations (hence, it is preconditioned that claims in philosophy of science and assumptions of philosophy of mind interrelate). Consequently, a vicious circle shows. Explanatory practices involve researchers with specific research interests. This implies that there are certain beliefs about what and how it should be investigated and explained; individual motivations might also be part of the picture. The issue arising at this point is that research interests, at least the involved beliefs implied by the interests, are relevant cognitive phenomena for supporters of the reflexive turn. It follows that cognitive phenomena (such as researcher’s beliefs) are supposed to explain other cognitive phenomena (e.g. explanations). This means that the investigated phenomenon is presupposed (petitio principii). There might be a way to relativize this vicious circle: The assumption that cognitive phenomena are situated practices implies that cognitive phenomena can be constituted in myriad ways. If a cognitive phenomenon “Ω” in a certain situation is constituted in a manner that significantly differs from the constitution of another cognitive phenomenon “Φ”, then it might be possible that those disparate constitutive modalities nevertheless allow “Ω” to be a relevant part in explanations of “Φ” – since the entities and processes which bring either about clearly vary. Putting them in relations to build a more comprehensive theory of cognitive capacities does not seem to move in circles. The above-sketched petitio principii might be fended off in

5  Commentary on “Explanatory Diversity and Embodied Cognitive Science…

81

this or similar ways – but until Sanches de Oliveira does not detail this kind of philosophical defense for the “explanatory style approach”, his approach seems to be confronted with a small vicious circle.

5.4 The Incorporation Problem As somebody who is  skeptical about 4E’s methodological status, I doubt that Sanches de Oliveira’s ambitious approach supports the development of a 4E methodology. The consequence of this neglect is not that 4E research is further carried out but just without a proper methodology. It is more likely that if 4E researchers hesitate to develop methodical procedures and a methodological frame that steers situated cognition research into more structured investigations, then the 4E branch is modified and incorporated into other approaches of cognitive science that are already methodically and methodologically organized. The talk of explanatory styles is neither designed to help answer this challenging situation nor does it help to clarify the problems and consequences of explanatory plurality. However, this “incorporation problem” is only a problem if the just sketched potential (negative) prospects for situated cognition research are realistic. Only then the 4E branch urgently needs a methodology. If you think otherwise and work in the 4E branch, then Sanches de Oliveira’s philosophical position could be an elegant and persuasive commitment. Still, I claim that if no clear statements come soon about how to study, experiment with, and explain phenomena of situated cognition, then we have seen the peak of 4E research already. That is why we should answer the problem of choice and, depending on our answer, elaborate methodically and methodologically on 4E research.

References Baker, A. (2009). Mathematical explanation in science. The British Journal for the Philosophy of Science, 60(3), 611–633. Lange, M. (2013). What makes a scientific explanation distinctively mathematical? The British Journal for the Philosophy of Science, 64(3), 485–511. Shapiro, L. (2011). Embodied cognition. Routledge. Tufte, E.  R. (1997). Visual explanations: Images and quantities, evidence and narrative. Graphics Press.

Chapter 6

Ecological Psychology, Enaction, and the Quest for an Embodied and Situated Cognitive Science Manuel Heras-Escribano

6.1 Towards an Embodied and Situated Cognitive Science This chapter evaluates which approach within 4E cognition is in a better position to offer a research program and a scientific framework for the embodied and situated cognitive sciences that could outcompete cognitivism once and for all. The main thesis of this work is that ecological psychology is in a better position to develop that scientific framework. This is because enactivism can be defined as a philosophy of nature rather than a scientific framework with its own research program (following Gallagher, 2017). This means that enaction did not develop its own scientific methods, models, and metrics for human in vivo experimentation, which are key aspects in order to offer a whole new scientifically feasible foundation of a new science of the mind from a purely embodied and situated perspective. Although I agree that a philosophy of nature is a necessary aspect for any general theory of cognition, it is not enough: we also need a research program and a scientific framework in order to establish a fully-fledged science of the mind. Ecological psychology, as I show in this work, includes a philosophy of nature and a research program, which makes it suitable to settle the foundations for a fully embodied and situated cognitive science. In conclusion, I think that everyone working on embodiment and situated cognition should follow the ecological lead. In this section one I introduce both ecological psychology and enactivism from a historical perspective, strengthening their similarities and differences. Section 6.2 analyzes Gallagher’s (2017) main reasons for defining enactivism as a philosophy of nature rather than a research program, which are based on Godfrey-Smith’s distinction between these two ideas. Section 6.3 shows how ecological psychology includes both a philosophy of nature and a scientific framework with its own M. Heras-Escribano (*) Department of Philosophy 1, University of Granada, Granada, Spain e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_6

83

84

M. Heras-Escribano

research program, which leads me to the conclusion that it is in a better position to style the main scientific framework for an embodied and situated cognitive science.

6.1.1 Where Do We Come From? The two main theories among all embodied and situated approaches that aim to offer a fully embodied, situated, and non-representational cognitive science that could outcompete cognitivism are ecological psychology and enactivism. Here I will sketch, both in a systematic and historical way, the main ideas and proposals of both theories. 6.1.1.1 Ecological Psychology Ecological psychology or the ecological approach was born between the 1960s and the 1970s in the United States, and it was proposed by James J. Gibson and Eleanor J. Gibson, two experimental psychologists. James Gibson was trained in the traditions of Jamesian pragmatism and behavioral experimental psychology: the former was a well-established although not a mainstream approach to the mind, whereas the latter was an incipient scientific field of study in the United States (Heft, 2001; Reed, 1988). James Gibson studied psychology (both the graduate and undergraduate studies) at Princeton University, where he worked under the supervision of Herbert Langfeld (who in turn studied under the supervision of Franz Brentano, the thinker who settled the antecedents of both phenomenology and the analytic philosophy of mind). However, his major influence at Princeton was Edwin Holt, a radical empiricist and behavioral psychologist -in fact, Holt developed a motor theory of consciousness based on James’ radical empiricism. Holt’s views on consciousness were more Jamesian than behaviorist, as he included desires, goals, purposes, and many other aspects that were neglected by orthodox behaviorism (Holt et al., 1912). In this sense, the behaviorism that influenced Gibson through Holt was not the stimulus-response view of Watsonian, orthodox behaviorism, but a radical empiricism strongly oriented towards experimentation at a behavioral scale (Costall, 2004; Raja, 2019). This is because pragmatism and behaviorism were reactions against structuralism, the view pioneered by Wilhelm Wundt and Edward Titchener by which consciousness was a structuration of different elements (sensations, affections, images, etc.) under certain relations, and that we can access through introspection (Reed, 1988). Both pragmatism and behaviorism were inspired by Darwinism and led a reaction against structuralism in psychology, which provoked an initial alliance between them, but their ontological and epistemic commitments were certainly different: whereas behaviorism considered organisms as bundles of mechanistic responses governed by the stimulus-response scheme, pragmatism strongly emphasized agency and personal autonomy guided by organic coordinations with the environment. James Gibson considered himself a radical empiricist at

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

85

the time, and that view guided his work and also the principles of ecological psychology1 (Gibson, 1982). Ecological psychology is based on six key principles: (1) the main unit of analysis is neither the brain nor the organism, but the organism-environment system; (2) the scale for the analysis of cognition is the ecological scale; (3) behavior is self-­ organized through the coupling with fluxes and forces of the environment; (4) perception is continuous with action, they are two sides of the same process; (5) perception is not based on sensations, but on ecological information; (6) perception is basically the perception of affordances or opportunities for acting (M. J. Richardson et al., 2008). Principle (1) comes from the Darwinian origins of pragmatism and it has been key for the ecological approach, as this idea of the organism-environment system advanced the ambitious definition of situated cognition proposed around the 1990s (Heras-Escribano, 2019). This unit of analysis is a system in the ecological approach because cognition is explained systematically as the set of mutual interactions between organisms and their environments. This leads us to the second principle: the organism-environment system established a unit of analysis and an explanatory level of description. It is pointless to explain cognition with a physical or biological language when it should be properly explained in its own level of description. In this sense, ecological psychology offers a non-reductive approach to perception and action. Principle (3) is committed to the idea that organisms, as they form a system with their environments, they explore their surroundings and establish relations of mutual affection between the environment and them. This leads to the emergence of behavioral patterns of exploration and perception, which gives rise to Principle (4), the idea that cognition is a continuous and spatiotemporally extended process, not a series of discrete states, hence perception and action are two sides of the same process, two different moments in the same flux of interaction with the environment. These interactions, according to Principle (5), are not sensation-­based: organisms are sensitive to ecological information, which is the pattern of energy arrays that is structured so as to reveal the affordances of the environment. Take, for example, vision: when light rays from a source bounce from different surfaces in a given space (for example, a room), they produce a structure. The mere encounter with that structure reveals the shape of the room, showing the organism what it can or can’t do. The organism directly perceives the affordances of the environment, as stated in the sixth principle, because the structure of the light specifies  For reasons of space, it is impossible to explain in detail all the approaches and ideas that influenced the Gibsons in the birth of ecological psychology. For a more detailed view on the issue, please see Harry Heft’s (2001) masterpiece on the origins of ecological psychology entitled Ecological psychology in context and the paper that I co-authored with Lorena Lobo and David Travieso entitled “The history and philosophy of ecological psychology” (Lobo et  al., 2018). Vicente Raja’s (2019) paper emphasizes the development of James Gibson’s intellectual history as championing the idea that psychology needed to develop its own laws at its own scale. Reed’s (1988) biography of James Gibson and also Eleanor Gibson’s (2001) autobiography are the best biographies to read about the birth and development of ecological psychology. You can find in Spanish another recent summary of the philosophical antecedents of ecological psychology in Heras-Escribano and Lobo (2022). 1

86

M. Heras-Escribano

the shape of the environment. These affordances of the environment are the possibilities for acting available for particular organisms. They are aspects of the environment related to particular bodies and their actions. In the same room with the same illumination, a human will find graspable things, but an animal without opposable thumbs will not find anything graspable. What we directly and primarily perceive are the available affordances. James and Eleanor Gibson developed these ideas together, but they divided tasks around 1966: she focused on perceptual learning, while he focused on perceptual theory. The main goal for them was to establish a cognitive psychology starting from perception and action and increasingly scaling up to other capacities (Reed, 1991). Both were obsessed with finding a lawful or scientific explanation of perception and action, one that was based on causal relations (Raja, 2019). James Gibson wrote two major works between the 1960s and 1970s: one in which he tried to reformulate the senses from mere passive receptors to active perceptual systems that included postures, orientations, and the whole body and its movement, which anticipated the idea of embodiment (Gibson, 1966); and another one which was the first attempt to offer an ecological approach to vision (Gibson, 1979). Eleanor Gibson wrote many books, sometimes with different co-authors, on topics such as perceptual learning and the psychology of writing (Gibson, 1969, 1991). She received the highest scientific award in the United States, the National Medal of Science (being one of the few women to achieve it). After James Gibson’s death, different colleagues and students of his developed the main lines of ecological psychology; in particular, it is worth emphasizing the vast amount of empirical and theoretical work at the Center for the Ecological Study of Perception and Action at the University of Connecticut (CESPA). This center was devoted to operationalize and empirically support the main ideas of the ecological approach (Jacobs et al., 2012; Michaels & Carello, 1981; Shaw & Bransford, 1977; Turvey et  al., 1981, 1982; Turvey, 1996; van Orden et al., 2005). Some other students, colleagues and followers around the world expanded the limits of ecological psychology through the years outside CESPA (Fajen, 2007; Favela et al., 2018; Heft, 1989, 2003; Lee, 2009; Lee & Reddish, 1981; Richardson & Chemero, 2014; Schmidt & Richardson, 2008; Favela et al., 2021; de Wit et al., 2016; Jacobs & Michaels, 2002; Lobo et al., 2019; Lobo et al., 2014; Lobo & Travieso, 2012; Michaels et al., 2001; Paz et al., 2019; Withagen & Michaels, 2004). The influence of ecological psychology has gained momentum since then, especially with the rise of the 4e approaches to the mind (Chemero, 2009; Heras-Escribano, 2019; Rietveld & Kiverstein, 2014; Segundo-­ Ortin et al., 2019; van Dijk et al., 2015; Lobo, 2019; Travieso et al., 2020; Di Paolo et al., 2021). 6.1.1.2 Enaction Enactivism, enaction, or the enactive approach was born in the 1990s as an alternative to cognitivist-oriented cognitive science (Di Paolo & Thompson, 2014; Stewart et al., 2010). The foundational work of the approach was the book The embodied

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

87

mind (Varela et al., 1991), authored by the biologist Francisco Varela, the philosopher Evan Thompson, and the psychologist Eleanor Rosch. The main aim of the authors was to settle the bedrock of an innovative approach for explaining human experience through a combination of phenomenology, autopoiesis, sensorimotor coordinations, and Buddhism. Starting from the idea that living systems are already cognitive systems, this approach takes ideas from biophysics, autopoiesis, and organizational biology for providing a philosophical worldview based on a common explanatory pattern for every form of life, showing that life consists of self-­organized and self-assembled networked systems that produce an organism or agent, which is constantly struggling against environmental perturbations for keeping its own stability ongoing, for which it self-regulates compensating the effects of the environment (Maturana & Varela, 1980; Maturana, 1975; Thompson, 2007; Varela, 1979). This idea of self-organization and self-regulation, according to enactivists, can be applied from cells to society and, just like unicellular and multicellular organisms, sociality can be explained under the same enactive principles. Organisms enact their own meaningful environments, and that meaning emerges when organisms interact with their environments and value those interactions as positive or negative with regard to their contribution to self-organization or self-maintenance (Barandiaran, 2016; Barandiaran et al., 2009). This is called embodied “sense-making”, as it is the activity by which the agent gives meaning to different aspects of the environment thanks to its embodied exploration and interaction (Thompson & Stapleton, 2009; De Jaegher & Di Paolo, 2007). The way in which organisms interact with their environments is via sensorimotor contingencies,2 the idea that every sensation is linked to a particular action, such that the mastery of these sensorimotor contingencies occurring at different timescales leads to skillful action. This skillful performance is at the very basis of sense-making. This very idea scales to the social level: “In the social realm, the interactive coordination of embodied sense-making activities with others lets us participate in each other’s sense-making (social understanding = participatory sense-making)” (De Jaegher, 2013). Thus, the enactive approach offers a general view of cognition from the most simple forms of life to human sociality as governed by the same patterns and principles, showing that where there is life, there is mind. The sequel to The embodied mind came more than a decade later, in 2007, and it was written by Evan Thompson, one of the three original authors. The book, Mind in life, explored some of the ideas that were merely presented in the first book, and developed further conclusions for a wide range of topics, from sensorimotor subjectivity to mental causation. Ten years later, Di Paolo, Barandarian, and

 The issue of sensorimotor contingencies is problematic for ecological psychology, as it claims that perception is not based on sensations, but on ecological information. According to ecological psychology, enactivism’s emphasis on sensations is clearly a mistake, as sensations should be conceived as a vestige from cognitivism. In fact, enactivism neither rejects nor reformulates the idea of sensation, it simply takes it as it is understood within the cognitivist approach, and connects sensations with movement via sensorimotor schemes. A more ambitious view from an embodied and situated perspective should include a critical analysis of sensations. For more on the problems of sensations from an ecological approach, see Heft (2020). 2

88

M. Heras-Escribano

Buhrmann wrote Sensorimotor life, a work that aimed to be a continuation of the two previous works and that strengthen the enactive approach by providing new evidence thanks to computational models and the reinterpretation of debates in philosophy of biology and psychology (Di Paolo et al., 2017). This approach is sometimes referred to in the literature as autopoietic enactivism (Degenaard & O’Regan, 2017). Although in the original enactive approach there has been a tight connection between life and mind, different variations of enactivism does not emphasize this point. The first variation, called sensorimotor theory, only focuses on these already-­ mentioned sensorimotor contingencies and leaves aside the biological basis of agents: “Autopoietic processes are then only relevant for perceptual experience in so far as they impact on the perceptual capacities and their behavioral expressions” (Degenaard & O′ Regan, 2017). The philosopher Alva Noë has applied the evidence gathered by sensorimotor theory to epistemic and perceptual problems, such as the problem of perceptual presence (Noë, 2004, 2012; O’Regan & Noë, 2001). Another branch is radical enactivism, proposed by Dan Hutto and Erik Myin, and whose project consists in sanitizing the 4e approaches to the mind, leaving aside any trace of computationalism or representationalism in perception and action. From a more analytic style, these authors have offered genuine arguments that support the main claims of enaction, creating new ideas and theses such as the Hard Problem of Content (Hutto & Myin, 2012; Hutto & Satne, 2015). As we can see, there are at least three different ways of understanding enaction, although I will focus here on what has been defined as autopoietic enactivism.

6.1.2 Where Are We Now? Ecological psychology and enaction have been widely developed over the last decades. In the case of ecological psychology, a great corpus of human in vivo experimental evidence has been gathered since the 1970s until nowadays. This is of vital importance, not only because we are talking about human in vivo empirical evidence: the data collection has been achieved not just through the models and procedures of a cognitivist-oriented cognitive science, but developing its own ecological models, metrics, and procedures. This allowed ecological psychology to develop fields of study of its own, as it happens with the case of dynamic touch, the scientific study of the sense of touch as based on action. This aspect is decisive and, as we will see, it will play a critical role in the rest of the chapter. At the same time, many ecologically-oriented philosophers applied the principles of ecological psychology to conceptual debates and they have shown the potential of this scientific discipline for the philosophy of mind, epistemology, and the life sciences. In sum, the combination of a vast experimental data gathering, a revolutionary scientific methodology, and an innovative and challenging theoretical framework show that ecological psychology is a real contender of cognitivism in the cognitive sciences.

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

89

In the case of enactivism, its three different branches allowed for the division of scientific labor and the development of both empirical and conceptual issues. The enactive approach focused mainly on combining different computational models with reinterpretations of different experiments in psychology in order to offer a general new conception of the mind that spreads through different forms of life, from bacteria to homo sapiens, showing that all different kinds of cognitive abilities, from a bacterium’s movement towards a sucrose gradient to language and social cognition are nothing but different species of the same genus. A great work in this respect is, for example, that of Di Paolo et al. (2017) and Di Paolo et al. (2018). Sensorimotor enactivism has been focused on the collection of evidence from computational models and in vivo evidence, and it was applied both to epistemology and the philosophy of perception thanks to the work of Alva Noë, and also to the philosophy of mind thanks to Kevin O’Regan’s book that summarizes his main achievements and its application to theoretical problems (O’Regan, 2011). The third branch, radical enactivism, has been focused on proposing a theoretical alternative to that of autopoietic enactivism for attacking cognitivism: instead of simply denying the need for mental representations, radical enactivists dealt with the specific arguments from cognitivist-oriented analytic philosophers to show that their arguments for naturalizing content and postulating representations is unfounded. As we can see, and despite the tensions among these three branches, enactivism aims to offer a general view of nature and mind understood as dynamic-based and agent-based that confronts the view of cognitivism, which takes representations and computational processes as the very basis of cognition. The embodied and situated cognitive sciences were born at the same time as enactivism; this is, in the 1990s. Sometimes authors use two labels interchangeably to refer to this constellation of theories: ‘embodied and situated cognition’, and ‘4E cognition’. The former underlines the importance of embodiment and situatedness as key concepts, whereas the latter includes these two theses (embodiment and embeddedness) but also includes enaction and the extended mind (hence the four ‘E’s) as two main approaches that incorporate the theses (Ward & Stapleton, 2012). Another label is ‘post-cognitivism’, but the very term is equivocal: whereas embodiment and situatedness as such were proposed in the 1990s, we have seen how theories such as ecological psychology were contemporaries of cognitivism. As such, if we stick to ‘4E cognition’, there are more approaches beyond the extended mind and enaction within the 4E approach: interactivism, phenomenological cognitive science, etc. As we can see, 4E cognition or the embodied and situated approach in general is not a monolithic perspective, and it is well known that the interests, methodologies, and commitments are not the same for all approaches within the theory. The thin line that sustains the minimum agreement of all these approaches was defined by a loose conception of embodiment and situatedness that in the case of ecological psychology and enaction was more similar than in the case of the extended mind. Whereas in the extended mind the idea of embodiment and situatedness was some kind of ‘extended functionalism’ by which the world and the body were part of mental states inasmuch as they could fulfill the same functional role of other mental/neural elements (implementational materiality, as Wheeler (2010)

90

M. Heras-Escribano

defined it), for ecological psychology and enactivism the role of the body and the world was not merely implementational: the body and the world played a crucial role understood from a developmental, diachronic perspective, where the history of interactions was the main object of study. The body and the world were not just a static, neutral, indifferent substratum where multiple functions could be implemented as states, but the vehicle through which a continuous, spatiotemporally extended and interactive process was taking place and that contributes substantially to that process. Wheeler refers to this view of embodiment and situatedness as ‘vital materiality’ (Wheeler, 2010). Although there is a similarity in ecological psychology and enaction regarding this way of understanding embodiment and situatedness, it is simply fair to acknowledge that the original source for this way of understanding embodiment originally comes from ecological psychology, mainly because it is temporarily previous: these ideas of organic coordination, of mutuality between organism and environment from a diachronic perspective was the first principle that guides ecological psychology, a principle that was directly inspired by Darwinism and pragmatism. In this sense, ecological psychology advanced the vital materiality view on embodiment and situatedness decades before embodiment and situatedness were born, and decades before 4e cognition was even conceived. From this perspective, enaction and the extended mind can be understood as an attempt of different traditions, like phenomenology and analytic philosophy, to reach the radical and groundbreaking principles that ecological psychology established decades before.

6.1.3 What Do We Need? The answer is plainly simple, yet highly complex: we need a new science of the mind, and that new science should be built over embodied and situated principles. This implies that the whole scientific framework should be operationally designed from that perspective, which means that the models, metrics, and the very design of the experiments should be directly inspired by those theses. This is not quantitative change, but a qualitative one: it is not that we should use the same models, metrics, devices, and strategies that are already present within the mainstream (cognitivist) cognitive science and reformulate the results of the experiments. We need a brand new methodology that is built from scratch within the principles of embodiment and situatedness. Someone could ask why the methods, metrics, models, and devices that are already present in the scientific community are not enough, or why a mere reinterpretation of data is not sufficient for shaping this new science. The answer is that all these devices, methods, and metrics were created from a cognitivist mindset, from a view of what is the mind that somehow determines the goals in the mind of the scientist and the design of the devices, models, and metrics for achieving that

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

91

goal.3 Thus, if there is a science that aims to be the leading framework for understanding and investigating the mind, that science cannot simply be dependent on the whole general framework, methods, and techniques for studying the mind that were built from a fully incompatible view. Such a strategy is simply doomed to failure, as those resources are clearly insufficient for achieving what we are looking for. On the contrary, creating a new science with methods, metrics, devices and models that incorporate in their very design and conception the principles of embodiment and situatedness is key for defining the mind in those terms. Now the pertinent question is the following: which one of the 4e approaches is the most suitable for achieving this goal? Here I will defend that ecological psychology is already shaping this new science, whereas enactivism is stuck in a process of reinterpretation of data plus using cognitivist-designed experimental setups and models that are not fully suitable for a science with a fully embodied and situated approach. They are doing cognitivist science with an embodied and situated flavor, and this is far from establishing an embodied-embedded cognitive science. This idea can be traced back in the literature, but it has also been acknowledged explicitly in the work of Gallagher, when he claimed that, for different reasons, “enactivism may be better thought as a philosophy of nature than a scientific research agenda” (Gallagher, 2017, p.  22). Section 6.2 analyzes and strengthens Gallagher’s rationale, and Sect. 6.3 focuses on how ecological psychology is not only a philosophy of nature, but also a full-blown research program. In particular, it is the only suitable research program for establishing a fully embodied and situated cognitive science.

6.2 Enaction: Philosophy of Nature or Research Program? In his influential work Darwinian populations and natural selection, the philosopher Peter Godfrey-Smith distinguishes between a philosophy of nature and a scientific research program. Such a distinction is mostly implicit in every discipline (as every scientific research program in every discipline includes a philosophy of nature), and the tasks associated with each one of them are very different. As such, according to this distinction, a philosophy of nature “can use its own categories and concepts (…) developed for the task of describing the world as accurately as possible when a range of scientific descriptions are to be taken into account” (Godfrey-­ Smith, 2001, p. 284). In this sense, a philosophy of nature includes a conceptual repertoire to offer a particular view of nature that competes with other views in the task of offering the most accurate description of our cosmos. In this line, and relating this idea to the embodied and situated cognitive sciences, both enactivism and ecological psychology are naturalist and non-reductive approaches to mind (and nature). They are non-reductive because they do not understand cognition as brain-­ centered, and they emphasize the centrality of agency as a key level of explanation

 Thanks to Lorena Lobo for making me aware of this important point.

3

92

M. Heras-Escribano

in the science of the mind. There are differences, though: while the history of enactivism started from an organism-centered perspective and gradually accepted the importance of the environment in the shaping of agency over the years,4 ecological psychology (due to its pragmatist, Darwinian-based origins) started from the centrality of organisms as active explorers of their environments, so the main unit of analysis for understanding cognition was not just the organism, but the organism-­ environment system: the systematized history of interactions between them. Enaction, as such, has provided innovative categories and concepts for describing the world according to its view and principles. Concepts already introduced such as ‘sense-making’, ‘adaptivity’, ‘precariousness’, or even ‘enaction’ itself offer a view of mind in nature that provides its own Weltanschauung, a way of understanding not only the functioning of the mind, but the role of nature in general. If all organisms are self-assembled, open systems that need to re-organize given the perturbations and effects of the environment in them, then we have a view of life and a view of nature: life is precarious and nature is constantly affecting organisms, which have to re-assemble to compensate the perturbations. It is a view that puts the agent in the center of the explanation of life and cognition, a view that vindicates the organismal point of view that is neglected by the gene-centered view in evolutionary biology, and that rejects the mechanistic approach of the explanation of cognition in favor of a more dynamic view of cognitive processes.5

 For example, Varela remained ambiguous with respect to the environment. In the same work, he first wrote that “environmental elements intervening between the efector and the sensory surfaces of the organism [are] irrelevant, because the nervous system can be defined as a network of neuronal interactions in terms of the interactions of its component neurons, regardless of intervening elements. (Varela, 1979, p. 242, emphasis added) and, later on, that “[t]he domain of the possible states that the nervous system can adopt as a closed system is at any moment a function of this history of interactions, and implies it.” (Varela, 1979, p. 245, emphasis added). This ambiguity remained until Di Paolo’s introduction of the idea of adaptivity (2005), or the way in which agents internally regulate themselves to compensate for environmental perturbations. However, I consider that this understanding of the environment is completely reductive to the mere effects of it in the organism: interventions for Varela (1979, p. 242) or perturbations for Di Paolo (2005, p. 437). The effect of these interventions or perturbations are understood as internal structural organismal changes produced by stimuli/perturbations (Di Paolo, 2005, p. 437–8), as it happens in cognitivism (at least in the sense that they share both the inner dimension of the organism as an explanatory level and the reduction of the environment to mere effects or perturbations in the organism). In this sense, the enactive view reduces and subsumes the environment into changes within the organism, and does not conceive it as part of an organism-environment system. On the contrary, the organism-environment system is the main unit of analysis in the ecological approach (which gives an equal leading role to the organism and the environment), which makes ecological psychology a fully situated approach to cognition. 5  Although it is true that in the early works of autopoiesis the words ‘mechanistic’ and ‘naturalistic’ were almost interchangeable (especially in the work of Maturana), it is fair to say that the evolution of the enactive approach lead to a differentiation between a mechanistic explanation and a dynamic explanation, and that the latter was used by enactivists to explain how neural processes work. However, not all enactive-friendly approaches reject the use of mechanistic views (see the work of Dewhurst, Villalobos, or Korbak, for example). 4

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

93

This leads us to Gallagher’s idea6 that enaction is not a scientific research program, but a philosophy of nature. As he claimed, “enactivism may be better thought of as a philosophy of nature than a scientific research agenda” (Gallagher, 2017, p. 22). This is an extremely radical statement, as it has the risk of stripping away enactivism from any scientific potential. While this could be understood as something harmful for this approach, Gallagher himself claims that this does not mean that enactivism has nothing to do at a scientific level. As he wrote: “[E]ven if enactivism were to be considered a philosophy of nature, it wouldn’t be right to conclude that it cannot offer concrete hypotheses or raise novel scientific questions” (Gallagher, 2017, p. 24). It follows from the words of Gallagher that this is what enaction has been doing all this time since it was born: take concepts, ideas, a proposal from traditions such as phenomenology or sensorimotor theory, and apply them to the current scientific study of the mind so as to illuminate the aspects of cognition that have been traditionally overlooked or even neglected. This strategy has already been vindicated by Gallagher for phenomenology, which he named ‘front-loading phenomenology’, and that consists in using “the insights developed in phenomenological analyses (modeled on Husserlian description, or the more empirically oriented phenomenological analyses found, for example, in Merleau-­ Ponty or in neurophenomenology) to inform the way experiments are set up” (Gallagher, 2003, p. 92). This enlightens current experimental science and allows for the obtaining of new data that couldn’t be acquired otherwise. This is highly promising and exciting, and implies that in the long-run the methods of cognitivist cognitive science and cognitivist experimental psychology could be affected by this new perspective. As we can see, the same strategy of front-loading phenomenology has been applied to enactivism in the fields of computational modeling, sensorimotor theory, and experimental psychology: experimentalists get inspiration from enactivism so as to inform the ways in which they can perform their perceptual experiments, and also computer scientists test enactive ideas via computational models of unicellular life. For example, in Sensorimotor life, the authors apply different computational models to test enactivist ideas such as networked sensorimotor schemes in fields like skillful coping or perceptual learning. Similar computational models have been used to analyze enactive features of cognition in cellular automata, as it happens with the idea of normativity (Barandiaran & Egbert, 2014). As Evan Thompson shows in Mind in life, there have been authors that aimed to front-­ load enactivism by making computational models of autopoiesis. This strategy of front-loading enaction in science is complemented with another claim of Gallagher: “[f]or a philosophy of nature to take scientific data seriously does not require that it take any particular scientific interpretation as necessary truth” (Gallagher, 2017, p. 126). With this claim, he proposes to reinterpret scientific data from a particular philosophy of nature (in this case, enaction). This other strategy is also what enaction has been doing all this time, and the work of Gallagher in his books How the

 Although here I focus on Gallagher (2017), the idea of understanding enaction as a philosophy of nature is becoming popular among enactivists (Meyer & Brancazio, 2022). 6

94

M. Heras-Escribano

body shapes the mind and The phenomenological mind (with Dan Zahavi) is a clear example of that when he has reinterpreted well-known experimental evidence so as to support enactivist theses in the fields of agency, free will, perceptual learning, or even discussing philosophical issues (like the Molyneux problem, for example). At the same time, we can find clear examples of data reinterpretation in The embodied mind and its sequel, Mind in life (for example, when Thompson combines different ideas from Stephen Jay Gould or Lynn Margulys with empirical data from evolutionary biology to support his autopoietic views). I think no one could claim, contrary to Gallagher (2017), that enaction does not offer an ambitious philosophy of nature that also includes two challenging tasks: first, the purpose of reinterpreting empirical data under a new, richer light; second, that it makes use of current scientific methods in their aim of front-loading enactive ideas in scientific practice. In this sense, enactivism is a thrilling perspective for anyone who works in the field of the embodied and situated cognitive sciences. Nevertheless, I think this project is not sufficient to achieve the most ambitious goal of this field: to offer a full-blown research program in cognitive science that aims to outcompete cognitivism. For this, the tasks of front-loading ideas in the current scientific frameworks and reinterpreting scientific evidence gathered by cognitivism are simply not enough. To become a competitor of cognitivism, we should not just use the current cognitivist-inspired scientific methodologies; we should also offer a new scientific framework with its own models, metrics, and practices that are directly built upon embodied and situated principles. In the same vein, we should not only reinterpret data, but also offer new data using the resources of this scientific framework, with the aim of producing scientific results already focused on the aspects that are valuable for an embodied and situated cognitive science. I think this road has already been established by ecological psychology.

6.3 Ecological Psychology: A Philosophical and Scientific Framework In this section I want to show that ecological psychology is both a philosophy of nature and a research program. It is both things because it offers a new ontology of nature of its own, one that starts from the organism as an explorer that directly perceives affordances thanks to its movement, and that forms a system with its environment. Also, it is a scientific framework and a research program because it develops its own scientific models, metrics, and methodologies with the aim of offering a scientific, naturalist explanation of experience (perception and action) from a firstperson perspective.

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

95

6.3.1 Ecological Psychology as a Philosophy of Nature As we have seen previously, ecological psychology is based on six principles that sustain the approach. If we pay attention, we reach the conclusion that some of them do not just focus on a scientific analysis of perception and action: they also hide important ideas with respect to how nature is understood in ecological terms. First, the idea that the main unit of analysis is the organism-environment system means that nature itself is presented as the product or the mutual affection and interaction between organisms and their environments. This is key, because the organism is an explorer of its surroundings, an agent. An organism is basically animate, and psychology starts with animacy (Gibson, 1979/2015, p. 3). Organisms establish mutualistic relations with their environments, which means that they are both dependent on each other7 (Gibson, 1979/2015, p. 4). This mutual dependency or mutuality of organism and environment is not implied by the physical sciences, but by the cognitive or psychological sciences. This is because the magnitudes, units, or metrics of physics are not useful for explaining the cognitive or psychological dimension of the organism and how it is shaped from the interaction with its environment (Gibson, 1979/2015, pp. 4–7). The environment of the organism is the surroundings understood from the perspective of what is meaningful or valuable for the organism, but how can we explain this in scientific terms? The naturalization of meaning or value in perception comes with the explanation of how we perceive affordances thanks to ecological information, something that was depicted in Sect. 6.1. For this, ecological psychology developed an ecological ontology (Käufer & Chemero, 2021, pp. 156–164) that includes descriptions of the environment at an experiential level, dividing it into medium (the space in which perception and action take place), substances (the materials that make up environmental objects), and surfaces (the interfaces between medium and substances). It is thanks to that ontology that we can explain the direct perception of affordances (because thanks to this we can explain how energy arrays form a structure through reverberation in the surfaces of the environment). This ecological ontology is not only restricted to these elements, but it includes the idea of organisms as agents or explorers, the environment as the meaningful surroundings of the organism, and a new ontology of perception in which we can highlight affordances. We can also derive important ethical and political implications from this view of nature (Sanches de Oliveira, 2018). This should be sufficient to show that ecological psychology includes a philosophy of nature. Besides this, several philosophers have included aspects of ecological psychology in their own approaches in order to advance in philosophical discussions; take, for example, the influence of affordances in the work of

 The mutualistic approach of the ecological approach and the understanding of experience and cognition as the systematization of the history of interactions between organisms and their environments (hence the term ‘organism-environment system’) makes it suitable to be related with current developments in evolutionary biology, such as niche construction theory (Heras-Escribano, 2020), (Heras-Escribano & de Pinedo-García, 2018). 7

96

M. Heras-Escribano

phenomenologists like Dreyfus or enactivists like Gallagher. Needless to say, philosophers like Chemero or Reed, within the ecological tradition, applied ecological ideas and concepts to conceptual problems in the philosophy of mind and the philosophy of biology. This ecological philosophy of nature is maybe the aspect that has been less exploited from the ecological account. But now we are going to focus on ecological psychology as a scientific framework that is genuinely based on embodied and situated principles.

6.3.2 Ecological Psychology as a Research Program and Scientific Framework I have said that the main challenge of 4e approaches is to offer a genuine scientific framework and research program that is directly built upon embodied and situated principles. The main challenge is, then, to create scientific methodologies, models, and metrics that are conceived from a full-blown embodied and situated perspective. This is important because the scientific program that is built upon these principles is conceived to illuminate processes and aspects that have been largely overlooked or even neglected by previous cognitivist programs. This is essential, as current methodologies use models and metrics that have been conceived from a cognitivist standpoint. The need for offering purely embodied and situated methodology is key. And, as I see it, ecological psychology is the only approach within the 4e perspective to cognition that can offer what we need. In the rest of the section, I will illustrate why we should follow the ecological lead for establishing a full-blown embodied and situated cognitive science that could outcompete cognitivism once and for all thanks to the example of William Warren’s analysis of the affordance of climbability. Warren (1984) analyzed the affordance of climbability from a purely ecological (this is, embodied and situated) approach. Instead of studying the brain or the nervous system of the participants, he explained why we perceive the climbability of steps by taking the organism-environment system as a unit, establishing a model at this level of analysis, and using new agent-related, ecological metrics. It is one of the first examples of a purely embodied and situated cognitive science with human participants. Warren wondered what can make participants estimate that a step is climbable or not. First, he observed how participants raised their legs to step on different obstacles or steps. He noticed that (as it was expected) tall participants stepped on higher obstacles than short participants. Then he changed the focus from absolute metrics to a first-person perspective, and he studied every participant’s estimation about the climbability of a step by analyzing the relation between the participant’s leg length and the height of the step. As we can see, this is an agent-­ related, mutualistic, embodied perspective for doing it. Following this, Warren proposed an elegant biomechanical model that captures this idea of the organism-environment reciprocity for perceiving climbability. The results were

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

97

appealing: according to Warren’s model, a human being (with two fully functional legs), independently of being tall or short, will perceive as climbable a step that is no more than 88% of the length of her leg. These agent-related metrics work for each person, tall or short. This had huge consequences from an ecological (embodied, situated) perspective: there is no need to postulate mental operations or gymnastics for explaining how we can estimate whether a step is climbable or not. These estimations are not based on abstract operations: it is a kind of implicit knowledge that we have of our body as related to the affordances of the environment. As Warren himself wrote: The affordance boundary between climbable and unclimbable stairways (…) can thus be seen to emerge from the dynamics of the animal-environment system. Analyzing the biomechanical basis for critical riser height (…) would provide the makings of a lawful account of perceptual categories without an appeal to mediating constructs such as mental categories or concepts (Warren, 1984, p. 691, emphasis added).

This is how Warren proposed the use of agent-related metrics for describing the affordance of climbability from an ecological approach. Warren’s experiment was somehow the starting point for some studies that make the concept of affordance truly operative in experimental terms. As we can see, Warren offered a full-blown embodied and situated model for explaining the affordance of climbability. The model included mathematical operations, variables, and everything that is needed in a model. But it also included something groundbreaking: new agent-based, embodied, situated metrics. These ecological metrics were not based on absolute measures (inches, centimeters, meters, etc.), but on agent-related measures: a step is climbable if it is no more than 88% of the length of my own leg. This 88% works for every length in the world, no matter how tall or short the human being. It is a purely embodied and situated metric that lawfully explains how we perceive the climability of a step from a first-person perspective. In sum, Warren’s innovative model for the affordance of climbability is an example of a scientific, lawful explanation of the perception of affordance from a first-person perspective and the illustration that a fully embodied and situated cognitive science is possible. Needless to say, this is not the only model offered in purely embodied and situated terms within the ecological approach. For example, Warren himself developed different models for several affordances, such as those of pass-through-ability in apertures (Warren & Whang, 1987), and the literature of ecological psychology is filled with many others (Craig & Watson, 2011; Fajen, 2008; Fajen & Devaney, 2006; Fajen et al., 2009; Fitzpatrick et al., 1994; Jacobs & Michaels, 2007; Michaels et al., 2001, etc.). This shows how the ecological scientific framework offers a science that is directly based upon embodied and situated terms and methodological resources that offer a highly ambitious research program: explaining perception and action scientifically and from a first-person perspective at the same time. Something else should be said regarding this last point. Since the scientific revolution, philosophers and scientists aimed to naturalize consciousness. It has always been puzzling for them how can we offer a detailed description of brain and nervous

98

M. Heras-Escribano

processes from a third-person perspective and, at the same time, how this explanation cannot exhaust a description of a first-person perspective experience of what it is like to be me, or you, or a bat. This conflict is at the very basis of the hard problem of consciousness and experts have written extensively about it. Some of them have proposed the idea of developing a first-person science of consciousness, but no serious attempt has been sketched since then. This led many others to claim, like Dennett, that a “[f]irst-person science of consciousness is a discipline with no methods, no data, no results, no future, no promise. It will remain a fantasy” (Dennett, 2001). After what we have seen in this section, I dare to say that ecological psychology and its own embodied and situated framework has enough resources, methods, metrics, data, and results for offering a first-person science of consciousness from an embodied and situated perspective. I think this is a great opportunity for ecologically-­oriented philosophers and ecological psychologists to start working on this promising field of study to show that a first-person naturalization of experience and consciousness is possible.

6.4 Conclusion: Following the Ecological Lead I started this chapter showing that, in order to outcompete cognitivist once and for all, it is necessary to develop not only a philosophy of nature that front-loads embodiment and situatedness and reinterprets empirical data, but also a scientific framework completely based on embodied and situated principles, with its own models, metrics, resources, etc. As I see it, situated and embodied conception of nature and cognition demands a new science and a new research program that makes justice to the goals and commitments of this new way of understanding the mind and nature, and also to highlight key aspects of cognition that have been largely overlooked by cognitivist research programs. In this sense, I agree with Gallagher (2017) in the idea that enaction is better conceived as a philosophy of nature rather than a research program or a scientific framework. But this is not enough to establish a complete embodied and situated cognitive science. Ecological psychology, as I showed in the last section, offers a philosophy of nature as well, but it also offers a research program and a scientific framework with fully-fledged embodied and situated credentials. For this reason, we should follow the ecological lead in order to establish a serious embodied and situated scientific framework in the cognitive sciences. Funding  The work for this research has been generously funded by the following sources: Juan de la Cierva-Incorporación Postdoctoral Fellowship (Ministerio de Ciencia e Innovación, Spain), Proyecto de Consolidación Investigadora 2022 “Toward an Ecological Approach to the Natural Origins of Content: From Direct Perception to Social Norms (ECOCONTENT)” (Ministerio de Ciencia e Innovación, Spain), and the research project “De la experiencia a los conceptos: Una reformulación del problema de Molyneux a través de la sustitución sensorial ecológica (ECOCONCEPT)” funded by the Ayudas a Proyectos de Investigación Científica 2022 Program of the BBVA Foundation (Spain).

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

99

References Barandiaran, X. (2016). Autonomy and enactivism: Towards a theory of sensorimotor autonomous agency. Topoi, 36, 409–430. Barandiaran, X. E., & Egbert, M. D. (2014). Norm-establishing and norm-following in autonomous agency. Artificial Life, 20(1), 5–28. Barandiaran, X. E., Di Paolo, E., & Rohde, M. (2009). Defining agency: Individuality, normativity, asymmetry, and spatio-temporality in action. Adaptive Behavior, 17(5), 367–386. Chemero, A. (2009). Radical embodied cognitive science. MIT press. Costall, A. (2004). From Darwin to Watson (and cognitivism) and back again: The principle of animal- Environment mutuality. Behavior and Philosophy, 32(1), 179–195. Craig, C., & Watson, G. (2011). An affordance based approach to decision making in sport: Discussing a novel methodological frame work. Revista de Psicologia Del Deporte, 20(2), 689–708. De Jaegher, H. (2013). Embodiment and sense-making in autism. Frontiers in Integrative Neuroscience, 7, 15. De Jaegher, H., & Di Paolo, E. (2007). Participatory sense-making. Phenomenology and the Cognitive Sciences, 6(4), 485–507. De Paz, C., Travieso, D., Ibáñez-Gijón, J., Bravo, M., Lobo, L., & Jacobs, D. M. (2019). Sensory substitution: The affordance of passability, body-scaled perception, and exploratory movements. PLoS One, 14(3), e0213342. de Wit, M. M., van der Kamp, J., & Withagen, R. (2016). After phrenology: Neural reuse and the interactive brain. Theory & Psychology, 26(3), 413–415. Degenaar, J., & O’Regan, J.  K. (2017). Sensorimotor theory and Enactivism. Topoi, 36(3), 393–407. Dennett, D. (2001). The fantasy of frst-person science. Jean Nicod Lectures. Di Paolo, E.  A. (2005). Autopoiesis, adaptivity, teleology, agency. Phenomenology and the Cognitive Sciences, 4(4), 429–452. Di Paolo, E., & Thompson, E. (2014). The enactive approach. In L. Shapiro (Ed.), The Routledge handbook of embodied cognition (pp. 68–78). Routledge. Di Paolo, E. A., Buhrmann, T., & Barandiaran, X. (2017). Sensorimotor life. Oxford University Press. Di Paolo, E. A., Cuffari, E. C., & De Jaegher, H. (2018). Linguistic bodies: The continuity between life and language. MIT Press. Di Paolo, E. A., Heras-Escribano, M., Chemero, A., & McGann, M. (Eds.). (2021). Enaction and ecological psychology: Convergences and complementarities. Frontiers Media SA. https://doi. org/10.3389/978-­2-­88966-­431-­3 Fajen, B. R. (2007). Affordance-based control of visually guided action. Ecological Psychology, 19(4), 383–410. Fajen, B.  R. (2008). Perceptual learning and the visual control of braking. Perception & Psychophysics, 70(6), 1117–1129. Fajen, B.  R., & Devaney, M.  C. (2006). Learning to control collisions: The role of perceptual attunement and action boundaries. Journal of Experimental Psychology: Human Perception and Performance, 32(2), 300–313. Fajen, B. R., Riley, M. A., & Turvey, M. T. (2009). Information, affordances, and the control of action in sport. International Journal of Sport Psychology, 40(1), 79–107. Favela, L. H., Riley, M. A., Shockley, K., & Chemero, A. (2018). Perceptually equivalent judgments made visually and via aaptic sensory-substitution devices. Ecological Psychology, 30(4), 326–345. Favela, L. H., Amon, M. J., Lobo, L., & Chemero, A. (2021). Empirical evidence for extended cognitive systems. Cognitive Science, 45(11), e13060. Fitzpatrick, P., Carello, C., Schmidt, R. C., & Corey, D. (1994). Haptic and visual perception of an affordance for upright posture. Ecological Psychology, 6(4), 265–287.

100

M. Heras-Escribano

Gallagher, S. (2003). Phenomenology and experimental design toward a phenomenologically enlightened experimental science. Journal of Consciousness Studies, 10(9–10), 85–99. Gallagher, S. (2017). Enactivist interventions. Oxford University Press. Gibson, J. J. (1966). The senses considered as perceptual systems. Houghton-Mifflin. Gibson, E. J. (1969). Principles of perceptual learning and development (The century psychology series). Prentice-Hall. Gibson, J. J. (1979). The ecological approach to visual perception. Psychology Press. Gibson, J. J. (1982). Autobiography. In E. Reed & R. Jones (Eds.), Reasons for realism (pp. 16–89). Routledge. Gibson, E. J. (1991). An odyssey in learning and perception. Learning, development, and conceptual change. MIT Press. Gibson, E. J. (2001). Perceiving the affordances. A portrait of two psychologists. Psychology Press. Godfrey-Smith, P. (2001). Three kinds of adaptationism. Adaptationism and optimality, 122. Heft, H. (1989). Affordances and the body: An intentional analysis of Gibson’s ecological approach to visual perception. Journal for the Theory of Social Behaviour, 19(1), 1–30. Heft, H. (2001). Ecological psychology in context: James Gibson, Roger Barker, and the legacy of William James’s radical empiricism. Erlbaum Press. Heft, H. (2003). Affordances, dynamic experience, and the challenge of reification. Ecological Psychology, 15(2), 149–180. Heft, H. (2020). Ecological psychology and enaction theory: Divergent groundings. Frontiers in Psychology, 11, 991. Heras-Escribano, M. (2019). The philosophy of affordances. Palgrave Macmillan. Heras-Escribano, M. (2020). The evolutionary role of affordances: Ecological psychology, niche construction, and natural selection. Biology and Philosophy, 35(2), 1–27. Heras-Escribano, M. (2021). Pragmatism, enactivism, and ecological psychology: Towards a unified approach to post-cognitivism. Synthese, 198(Suppl 1), 337–363. Heras-Escribano, M., & de Pinedo-García, M. (2018). Affordances and landscapes: Overcoming the nature-culture dichotomy through niche construction theory. Frontiers in Psychology, 8, 1–15. Heras-Escribano, M., & Lobo, L. (2022). Los orígenes teóricos de las affordances y la psicología ecológica: una introducción histórica. In M.  Heras-Escribano, L.  Lobo, & J.  Vega (Eds.), Affordances y ciencia cognitiva. Introducción, teoría y aplicaciones. Tecnos. Holt, E. B., Marvin, W. T., Montague, W. P., Perry, R. B., Pitkin, W. B., & Spaulding, E. G. (1912). The new realism: Cooperative studies in philosophy. The Macmillan Company. Hutto, D. D., & Myin, E. (2012). Radicalizing enactivism. MIT Press. Hutto, D. D., & Satne, G. (2015). The natural origins of content. Philosophia, 43(3), 521–536. Jacobs, D.  M., & Michaels, C.  F. (2002). On the apparent paradox of learning and realism. Ecological Psychology, 14(3), 127–139. Jacobs, D. M., & Michaels, C. F. (2007). Direct learning. Ecological Psychology, 19(4), 321–349. Jacobs, D.  M., Silva, P.  L., & Calvo, J. (2009). An empirical illustration and formalization of the theory of direct learning: The muscle-based perception of kinetic properties. Ecological Psychology, 21(3), 245–289. Jacobs, D.  M., Vaz, C., & Michaels, C. (2012). The learning of visually-guided action: An information-­space analysis of pole balancing. Journal of Experimental Psychology: Human Perception and Performance, 28, 1215–1227. Käufer, S., & Chemero, A. (2021). Phenomenology: An introduction. John Wiley & Sons. Lee, D. N. (2009). General tau theory: Evolution to date. Perception, 38(6), 837–850. Lee, D., & Reddish, P. (1981). Plummeting gannets: A paradigm of ecological optics. Nature, 293(5830), 293–294. Lobo, L. (2019). Current alternatives on perceptual learning: Introduction to special issue on post-­ cognitivist approaches to perceptual learning. Adaptive Behavior, 27(6), 355–362. Lobo, L., & Travieso, D. (2012). El patrón de exploración modula la percepción de longitudes a través del tacto dinámico. Psicothema, 24(1), 55–61.

6  Ecological Psychology, Enaction, and the Quest for an Embodied and Situated…

101

Lobo, L., Travieso, D., Barrientos, A., & Jacobs, D.  M. (2014). Stepping on obstacles with a sensory substitution device on the lower leg: Practice without vision is more beneficial than practice with vision. PLoS One, 9(6), e98801. Lobo, L., Heras-Escribano, M., & Travieso, D. (2018). The history and philosophy of ecological psychology. Frontiers in Psychology, 9, 2228. Lobo, L., Nordbeck, P.  C., Raja, V., Chemero, A., Riley, M.  A., Jacobs, D.  M., & Travieso, D. (2019). Route selection and obstacle avoidance with a short-range haptic sensory substitution device✰. International Journal of Human-Computer Studies, 132, 25–33. Maturana, H.  R. (1975). The organization of the living: A theory of the living organization. International Journal of Man-Machine Studies, 7(3), 313–332. Maturana, H., & Varela, F. (1980). Autopoiesis and cognition: The realization of the living. Springer. Meyer, R., & Brancazio, N. (2022). Putting down the revolt: Enactivism as a philosophy of nature. Frontiers in Psychology, 13, 948733. Michaels, C. F., & Carello, C. (1981). Direct perception. Prentice-Hall. Michaels, C. F., Zeinstra, E. B., & Oudejans, R. R. D. (2001). Information and action in punching a falling ball. The Quarterly Journal of Experimental Psychology: Section A, 54(1), 69–93. Noë, A. (2004). Action in perception. MIT Press. Noë, A. (2012). Varieties of presence. Harvard University Press. O’Regan, J. K. (2011). Why red doesn’t sound like a bell. Understanding the feel of consciousness. Oxford University Press. O’Regan, J., & Noë, A. (2001). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24(5), 939–973. Raja, V. (2019). J. J. Gibson’s most radical idea: The development of a new law-based psychology. Theory & Psychology, 29(6), 789–806. Reed, E. S. (1988). James J. Gibson and the psychology of perception. Yale University Press. Reed, E.  S. (1991). James Gibson’s ecological approach to cognition. In A.  Still & A.  Costall (Eds.), Against cognitivism: Alternative foundations for cognitive psychology (pp. 171–198). Harvester Wheatsheaf. Richardson, M., & Chemero, A. (2014). Complex dynamical systems and embodiment. In L. Shapiro (Ed.), The Routledge handbook of embodied cognition (pp. 39–50). Routledge. Richardson, M. J., Shockley, K., Fajen, B. R., Riley, M. A., & Turvey, M. T. (2008). Ecological psychology: Six principles for an embodied—Embedded approach to behavior. In P.  Calvo & T. Gomila (Eds.), Handbook of cognitive science: An embodied approach (pp. 161–187). Elsevier. Rietveld, E., & Kiverstein, J. (2014). A rich landscape of affordances. Ecological Psychology, 26(4), 325–352. Sanches de Oliveira, G. (2018). Ecological psychology and the environmentalist promise of affordances. In Proceedings of the 40th annual conference of the cognitive science society (pp. 1061–1021). Schmidt, R. C., & Richardson, M. J. (2008). Dynamics of interpersonal coordination. In A. Fuchs & V.  K. Jirsa (Eds.), Coordination: Neural, behavioral and social dynamics (pp.  281–308). Springer. Segundo-Ortin, M., Heras-Escribano, M., & Raja, V. (2019). Ecological psychology is radical enough: A reply to radical enactivists. Philosophical Psychology, 32(7), 1001–1023. Shaw, R., & Bransford, J. (1977). Perceiving, acting, and knowing: Toward an ecological psychology. Erlbaum Press. Stewart, J.  R., Gapenne, O., & Di Paolo, E.  A. (2010). Enaction: Toward a new paradigm for cognitive science. MIT Press. Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of mind. Harvard University Press. Thompson, E., & Stapleton, M. (2009). Making sense of sense-making: Reflections on enactive and extended mind theories. Topoi, 28(1), 23–30.

102

M. Heras-Escribano

Travieso, D., Lobo, L., De Paz, C., Langelaar, T. E., Ibáñez-Gijón, J., & Jacobs, D. M. (2020). Dynamic touch as common ground for enactivism and ecological psychology. Frontiers in Psychology, 11, 1257. Turvey, M. T. (1996). Dynamic touch. American Psychologist, 51(11), 1134–1152. Turvey, M. T., Shaw, R. E., Reed, E. S., & Mace, W. M. (1981). Ecological laws of perceiving and acting: In reply to Fodor and Pylyshyn (1981). Cognition, 9(3), 237–304. Turvey, M., Fitch, H., & Tuller, B. (1982). The Bernstein perspective: I. The problems of degrees of freedom and context-conditioned variability. In J. S. Kelso (Ed.), Human motor behavior: An introduction (pp. 293–252). Erlbaum Press. van Dijk, L., Withagen, R., & Bongers, R. M. (2015). Information without content: A Gibsonian reply to enactivists’ worries. Cognition, 134, 210–214. van Orden, G. C., Holden, J. G., & Turvey, M. T. (2005). Human cognition and 1/f scaling. Journal of Experimental Psychology: General, 134(1), 117–123. Varela, F. J. (1979). Principles of biological autonomy. Kluwer Academic. Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind. MIT Press. Ward, D., & Stapleton, M. (2012). Es are good: Cognition as enacted, embodied, embedded, affective and extended. Advances in Consciousness Research, 86, 89–104. Warren, W.  H. (1984). Perceiving affordances: Visual guidance of stair climbing. Journal of Experimental Psychology: Human Perception and Performance, 10(5), 683–703. Warren, W., & Whang, S. (1987). Visual guidance of walking through apertures: Body-scaled information for affordances. Journal of Experimental Psychology: Human Perception and Performance, 13(3), 371–383. Wheeler, M. (2010). Minds, things, and materiality. In L.  Malafouris & C.  Renfew (Eds.), The cognitive life of things. Recasting the boundaries of the mind, McDonald Institute for Archaeological Research. Withagen, R., & Michaels, C. F. (2004). Transfer of calibration in length perception by dynamic touch. Perception & Psychophysics, 66, 1282–1292.

Chapter 7

Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward Pierre Steiner

Abstract   In this chapter, I pursue two aims. Firstly, I propose an original survey and analysis of the way proponents of 4E cognition have until now defined the relations between normativity and cognitive science. A first distinction is made between making normativity an explanandum of 4E cognitive science, and turning normativity into a property or part of the explanantia of 4E cognitive science. Inside of the latter option, one must distinguish between methodological, ontological and semantic claims on the value of normativity for studying and defining cognitive phenomena. The second aim of the paper consists in developing the further claim that normativity is an essential property of daily intentional concepts and of scientific concepts. I show how 4E cognition might further develop this latter claim in the context of recent debates about cognitive ontologies in neuroscience.

7.1 Introduction “Normativity” refers here to one important property of the activities of human agents. When they act and think, they are not only subject to natural laws; their behaviour is governed and motivated by norms. Take our ways of organizing the environment, our ways of conducting ourselves in spatial orientation to one another, our ways of using voice and body, our ways of dressing or our ways of conveying our thoughts and feelings to one another: we do not only seek to achieve some ends in virtue of some means; we exhibit a “sensitivity to the issue of rightness” (Taylor, 2016, p. 7). Human creatures act and think on the basis of what they take to be good, right, rational, appropriate or optimal to do. To put it differently, they do not act only on the basis of what they are disposed or inclined to do: those dispositions and inclinations encompass what they – human agents – are supposed or are expected to P. Steiner (*) Technology and Human Sciences Department, The University of Technology of Compiègne, Compiègne, France e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_7

103

104

P. Steiner

do, i.e. which behaviours are deemed appropriate or inappropriate within a given community. Their deeds may be assessed with reference to criteria of correctness; and they can themselves blame, criticize or correct other fellows for having failed to respect those criteria. Normativity is thus manifested when we distinguish, in our judgements and action, better from worse, correct from incorrect, optimal from suboptimal or appropriate from inappropriate. The norms I will discuss here are norms having their origins and existence in social and cultural practices. Moral norms, epistemic norms, norms of rationality (Kiesewetter, 2017), biological norms (FitzPatrick, 2000) and perceptual norms of correctness (Doyon & Breyer, 2015), for instance, are not similar to those social norms. Even though I shall restrict myself here to social norms in human communities, it should be noted that social norms are not exclusively human: nonhuman animal communities can also manifest social normativity (Andrews, 2020; Fitzpatrick, 2020). Many norms are implicit: following a norm does not require being able to linguistically articulate or reflect it. A rule is an explicit norm. Unlike laws of nature, norms do not have brute physical force. Acting from a norm is also distinct from acting from mere coercion: following a norm is being able to break it (Mulligan, 1999). Norms have a binding force (Korsgaard, 1996): one follows a norm because one feels or knows one should act this way. With this clarification made on what normativity is, we can now articulate the basic question at the source of this paper: Wherein lies the originality and scope (if any) of the 4E research program in cognitive science concerning normativity ?

By the “4E research program in cognitive science” (Newen et al., 2018), I mean the theoretical models and empirical investigations which underline the constitutively embodied, affective, situated and action-oriented dimensions of mind and cognition, up to the point that cognitive processes are not solely restricted to what takes place in the central nervous system but rather encompass bodily, motor and environmental processes. “4E” (as encompassing “embodied, embedded, enactive, extended”) is often a convenient short-hand for “4 + n E”: one can add “ecological”, “emotional”, “enculturated” etc. for underlining other relevant dimensions of cognition. 4E cognitive science also downplays and sometimes even rejects the claim that cognitive processes necessarily involve the manufacture or retrieval of mental representations, defined as intracranial and naturally contentful physical structures. Some approaches are locally non-representationalist (some forms of cognition do not involve mental representations, or only involve peripheral and emergent mental representations); others endorse global non-representationalism (no cognitive state or process will be usefully understood as made of mental representations). 4E cognition harbours works devoted to the study of normativity: what are the mechanisms and cognitive architectures involved in the perception, understanding, transgression, enforcement, negotiation and teaching of commitments, standards, rights, customs or obligations? As a matter of fact, there are embodied-enactive-­ embedded-extended-ecological accounts of normativity and normative phenomena.

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

105

For example, Rietveld (2008) investigates the normative aspects of unreflective skillful actions, by insisting on their situated and embodied dimensions. Schmidt and Rakoczy (2018, forthcoming) study how human creatures acquire and exercise normative attitudes towards themselves, other fellows, other animals, but also objects. Their investigations focus on properties of interactions between individuals. Segundo-Ortin (2022) explains the possibility  how socio-cultural norms can influence our perceptions of affordances, not appealing to the mental representations of norms, but to the education of intention and attention. Van den Herik (2021) proposes an explanation of meta-linguistic normativity not appealing to the primary application of learning rules, but to the regulation of linguistic activities by various mechanisms, including linguistic rules as resources. All these works share common tenets. The norms and routines that structure our interactions are out there in the world. Our ways of behaving have their normative significance built into them in virtue of being embodied and situated. Explaining how we apprehend norms does not require complex mental representations or an internalization of norms. Matthew Ratcliffe expresses that point clearly, and its consequences for the question of knowing how we can understand and interpret each other: Norms are integral to the relationships that comprise situations (…). The systematic structure we require in order to interpret people is out there in the shared world. So the burden need not be carried by a complicated understanding of the relationships between mental states. (Ratcliffe, 2007, pp. 97–98)

In all those cases, normativity is here the explanandum of 4E cognition. But we cannot restrict the originality and the scope of 4E cognition on normativity to normativity as an explanandum. Another, complementary approach, is to see how normativity affects the methodology of 4E cognition: which difference(s) – if any – do normativity and normative phenomena make to the way 4E cognition approaches cognition and cognitive phenomena? Normativity becomes here a property of various explanantia. The remainder of this paper will propose an answer to this question, by proposing several original distinctions allowing us to apprehend the 4E cognition literature more clearly (Sects. 7.2 and 7.3). In Sects. 7.4, 7.5, and 7.6, I will not focus on what 4E cognition does and says, but on what it could and should better do and say concerning one type of normativity, the one associated with daily and scientific concepts. The context of the discussion will be recent discussions in neuroscience about cognitive ontologies.

7.2 Three Claims on Normativity and Cognition Let us start with a methodological claim: the scientific study of cognition involves normative commitments. Cognitive scientists, in their daily work, follow norms: objectivity, accurateness, clarity are examples of epistemic norms, involved in any scientific enterprise. Those norms are ideally shared by the members of the

106

P. Steiner

community, along with legal and ethical regulations. Since at least the work of Robert Merton (1942), we know how much the scientific activity is not norm-free. One did not have to wait for 4E cognition for understanding or providing reasons for this claim. What 4E cognition can bring here is a fresh look on its importance. For instance, Sanches de Oliveira (2022) argues that in order to be coherent, what he calls “strong embodied cognitive science” should also be applied to cognitive science itself, making it more reflexive on its own embodied and situated dimensions. Without a 4E cognitive science of cognitive science, 4E cognitive science would be drastically incomplete. Those new dimensions to be studied might include scientific norms, and how they are concretely learnt, exercised, transmitted, and enforced by cognitive scientists in their daily practices.  With 4E cognition, one would add another object (namely: scientific normativity) to the explanatory agenda of cognitive science. It is not an object of study amongst others; it is a constitutive dimension of scientific rationality. 4E cognition here converges with the classical studies by Michael Polanyi and Thomas Kuhn on the tacit and practical dimensions of science. This methodological claim on the importance of studying the normative dimension of cognitive science itself is interesting, but it is not helpful if we want to understand what many proponents of 4E cognition mean when they are saying that cognition (and not only scientific cognition) has a normative nature. Norms might be constitutive of the study and description of cognitive systems, but is it the case that those systems cannot be cognitive independently of social norms? Some 4E accounts answer positively to this question, concerning at least some cognitive phenomena. In order to better understand what is at stake here, consider a metaphysical claim and a semantic claim. These two claims make normativity a basic property of cognition. What makes it the case that certain cognitive phenomena obtain is that, at least in part, certain normative facts obtain. Metaphysical claim: some cognitive facts are partially constituted by normative facts (this is a metaphysical claim, for it is about the essence of some cognitive states). Semantic claim: the meaning or content of cognitive states is partially constituted by normative facts. These two claims can be combined. One may also infer the metaphysical claim from the semantic claim; it is also possible to justify the semantic claim from the metaphysical claim. These claims are not new in the philosophical literature. They existed well before 4E cognition. What I want to show now is how some proponents of 4E cognition endorse, justify and develop them in an original way. Endorsing the metaphysical claim, some proponents of 4E cognition argue that social institutions can be vehicles of cognitive processes. From the semantic claim, some proponents of 4E cognition propose a theory of mental content which turns representationalism into an unjustified position. 4E cognition may also articulate these two claims together: the repudiation of representationalism (from the semantic claim) does not lead to an impoverishment

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

107

of explanatory resources, for social institutions are now – in virtue of the metaphysical claim – parts of the enabling mechanisms of cognition. Let us now examine these two strategies in turn. By “examination” I mean here a description, not an assessment of the truth value.

7.3 Normativity as Constitutive of Cognition: Ontological and Semantic Ascents Quarks, earthquakes, or galaxies behave or proceed following laws of nature. Physiological and biological processes as digestion, homeostasis or reproduction depend on mechanisms which can malfunction. There are norms in nature (Davies, 2001), justifying the careful use of teleology in the description and explanation of physiological and biological processes. What about cognition? The neural mechanisms underlying cognition may clearly dysfunction. Their purposes were not fixed by a designer or by a community; they emerged and were shaped by natural selection. But is the normativity of cognition exhausted by the biological normativity of neural processes? According to 4E cognitive science, some essential dimensions of cognition are constituted by normative facts, “normative facts” denoting here social and cultural facts. What does it mean? Some objects, events and practices are constituted by social and cultural norms, in the following sense: without these norms, those objects and practices would not exist. Take chess: moving pieces on a board in a universe in which no chess rules existed would not count as playing chess. Take a regicide: the event of killing someone counts as a regicide only if the killed person is a king. Being a king does not supervene on the intrinsic properties of a person, but on social and legal properties. Take a wedding: in order for two persons to be married to each other, they need to pass by some institutional rites which are framed by legal rules. When Richard Menary writes about the “normative nature of cognition” (Menary, 2007, p. 135), he takes the case of what he calls cognitive practices: solving problems, making inferences, planning, working out answers to questions,… Those activities rely on the manipulation of external representations (books, pictures, graphs, written inscriptions on a surface,…) and they are guided by cognitive norms: purposive norms (which define the aim of the activity), corrective norms (which enable to correct the activity), manipulative norms (for manipulating inscriptions), interpretative norms (interpreting inscriptions of a representational system), etc. (2007, pp. 137–138). The meaning and the purpose of these activities are not intelligible unless one takes these norms into account. Only organisms that behave appropriately by following those norms can be said to reason, solve problems or make inferences. The naturalism of 4E cognition is here a liberal naturalism (De Caro & Macarthur, 2022), including normative facts in the nature which makes cognition possible.

108

P. Steiner

Whereas Menary shows that a context of norms is a necessary condition for the development and exercise of some cognitive tasks, Gallagher and Crisafi (2009) argue that normative institutions may be the vehicles of cognitive activities such as legal judging and legal reasoning. It is not only the case that normative practices and institutions enable the emergence and exercise of cognitive processes; they may be constituents of those processes: Certain social practices and institutions (…) are not only institutions with which we accomplish certain cognitive processes, but also are such that without them such cognitive processes would no longer exist. They are at least enabling conditions, and on the most liberal reading, constitutive of those processes. (2009, p. 46)

Consider how a judge is producing a judgment for solving a legal problem: the process involves many components, not restricted to hand-held conveniences as books and documents: testimonies, expertises, debates,… The legal system is a part of the vehicles from which cognition is produced. Take also a museum. In virtue of the arranged works and information it displays, of its architecture and organization, the museum can mould my understanding of past events. It is not a system in which my cognition is located; it is a system which is a component of my cognitive processes. This liberal reading of what makes a part of the world a part of cognitive processes is obviously questionable, and indeed questioned. It raises the well-known objection of “cognitive bloat” (Allen-Hermanson, 2013): once they become too extended, cognitive systems turn into motley crews, threatening the unity of cognition and the possibility of a unified science of it. Let us now turn to the reading and use 4E cognition proposes of the semantic claim according to which the meaning or content of cognitive states is partially constituted by normative facts. Following Mark Greenberg, normative theories of content hold that “what makes it the case that a symbol has a particular content is at least in part that certain normative facts obtain” (Greenberg, 2005, p. 306). Inside of the genus of normative theories, distinctions ought to be made between theories that make normativity a matter of social and linguistic practices, and theories which start from biological normativity, as teleological theories of content (teleosemantics; cf. Dretske, Millikan, Papineau, Neander,…). Even though they will acknowledge that some sophisticated forms of mental content require the existence of linguistic and social norms, the latter theories will have no problem with the recognition of neural states and events as contentful and representational (in virtue of teleological and functional norms). This is not the case for non-representational theories of 4E cognition, which will hold that any form of content constitutively depends on social and linguistic normativity. Contentful intentional states are normatively constituted within linguistic and social practices (Cash, 2009; Hutto & Satne, 2015). As John Haugeland wrote: The fundamental pattern becomes a culture or way of life, with all its institutions, artifacts, and mores. The idea is that contentful tokens, like ritual objects, customary performances, and tools, occupy determinate niches within the social fabric—and these niches “define” them as what they are. Only in virtue of such culturally instituted roles can tokens have contents at all. (Haugeland, 1998, p. 156)

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

109

Here too, the claim is not new. Wilfrid Sellars and Donald Davidson are important forerunners. Tutelary figure of the “Pittsburgh School” (Maher, 2012) to which the works of John Haugeland, John McDowell and Robert Brandom are related, Wilfrid Sellars famously held that some cognitive attitudes and their contents were normatively constituted: In characterizing an episode or state as that of knowing, we are not giving an empirical description of that episode or state; we are placing it in the logical space of reasons, of justifying and being able to justify what one says (1956, § 36).

Being (recognized as) a knower, and more broadly producing conceptual mental episodes (beliefs, judgements, intentions,…) requires being a member of linguistic and inferential practices from which the content of what is known, believed, or judged is defined. In order to describe the structure of those normative, content-­ conferring practices, more precise theories will invoke “the social practices of giving and asking for reasons” or “deontic scorekeeping” (attributing and undertaking normative statuses such as commitments and entitlements, see Brandom, 1994). According to Donald Davidson, when we attribute a propositional attitude to someone, we do not describe a fact which is independent of a “constitutive ideal of rationality” (Davidson, 1980, p. 222). The basic principles of logic define the structure of a person’s attitudes and actions, and therewith what it is to have a thought or perform an action. There are logical and rational relations amongst beliefs, desires and intentions, and these relations make those states the intentional states they are: The meaning of a sentence, the content of a belief or desire, is not an item that can be attached to it in isolation from its fellows. We cannot intelligibly attribute the thought that a piece of ice is melting to someone who does not have many true beliefs about the nature of ice, its physical properties connected with water, cold, solidity, and so forth. The one attribution rests on the supposition of many more—endlessly more. (Davidson, 2004, p. 183)

For Davidson, it is only in a triangular relation between the agent, the interpreter and the world that the possibility of thought occurs: Someone cannot have a belief unless he understands the possibility of being mistaken, and this requires grasping the contrast between truth and error – true belief and false belief. But this contrast, I have argued, can emerge only in the context of interpretation, which alone forces us to the idea of an objective, public truth. (Davidson, 1984, p. 170)

One possible implication of this view is that conceptual mental events do not have material vehicles: they are first proper to persons situated in social practices (Steiner, 2014). But 4E cognition does not need to take that road. Indeed, another and more widespread originality of 4E cognition lies in the use it makes of the semantic claim on the normative nature of meaning for rejecting representationalism as the thesis that there exist naturally contentful intracranial states. Dan Hutto and Erik Myin’s radical enactivism (Hutto & Myin, 2013, 2017) is here a case in point. Hutto and Myin do not deny the contingent existence of contentful or representational states or attitudes, such as sentences, pictures, or judgments. Some, but emphatically not all, forms of cognition are content-involving: they represent the

110

P. Steiner

world in ways that might not obtain. They can be true or false. This content-­involving cognition derives its content from public language and cultural symbol systems, supported by normative practices. What Hutto and Myin’s radical enactivism puts under question is the existence of mental representations as naturally contentful states. There is no natural content: that is, there is no content which would be a product of natural relations of covariation or indication. Hutto and Myin’s basic strategy against natural content is the following: they claim that respectable naturalistic theories cannot accommodate naturally contentful cognitive states (also called “mental representations”), so that these states should be theoretically eliminated. Since their first book, Hutto and Myin argue that neither informational theories nor teleosemantics are able to provide a satisfactory non-intentional explanation of the emergence of semantic properties in nature (be they a matter of truth conditions, satisfaction conditions, accuracy conditions; be they conceptual or non-conceptual; be they propositional or non-propositional): either these theories beg the question by already coming with intentional notions, or they merely deliver covariation and indication, which are not sufficient for giving semantic or representational content. Following the semantic claim, only social and cultural norms can generate content. These failures to naturalize content (what radical enactivism also calls “The hard problem of content”) entail that representationalism has no foundations in the naturalistic ontology proponents of representationalism generally assume (but see Piccinini, 2022 for a recent dissenting view). Unable to be integrated in the naturalistic ontology it claims to be a part of, the representationalist program would be “plagued with toxic debt, financed by loans it cannot pay back” (Hutto & Myin, 2013, p. 160). Since mental content has no place in a naturalistic ontology, there are good reasons to think it does not exist as an entity conveyed or produced by natural processes, including subpersonal and intracranial ones. According to the authors, when it comes to contentful cognition (thoughts, imaginings or reasonings), the contents are not natural or subpersonal contents: they are derived from the integration of cognitive agents in socio-cultural normative practices. In addition, non-­ representational means and models are available and plausible for explaining basic cognitive phenomena, but also more complex phenomena. Those means include institutions and other normative practices, as we have seen above (see Casper, 2018 for this use of normativity as a way to solve the “scaling up” problem: how can we get from non-representational basic cognition to more sophisticated forms of cognition?). Normativity has thus here ontological and semantical forces when we want to define and explain cognition in its basic and sophisticated forms. This goes well beyond the fact normativity is an explanandum and a property of the scientific study of cognition (as we have seen in Sects. 7.1 and 7.2). For the moment, I believe this is the most complete characterization one may offer of the originality and scope of the way 4E cognition gives importance to normativity in the study of cognition. I would like now to develop a further step which is somehow latent in some works of 4E cognition. It concerns the relations between neuroscientific facts and concepts, normativity, and the concepts of folk psychology.

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

111

7.4 Folk Psychology and the Normativity of Daily Intentional Concepts The place I will start from is the normative nature of folk psychology according to some theories of 4E cognition. Folk psychology is classically defined as the practice of predicting and explaining behaviour by the attribution of propositional attitudes, including intentional verbs and concepts (belief, thought, intention, desire,…). According to the first version of functionalism in philosophy of mind, mental states were identified with functional states, whose functions or roles were specified by folk psychology. For David Lewis, the language of folk psychology could be treated as “a term-introducing scientific theory” (Lewis, 1972, p. 256). The defects of this version of functionalism were numerous: It is doubtful that there really is a suitable body of folk psychological platitudes about the mind; if there were such a body, it is doubtful that it would constitute a theory in the relevant sense; if it did constitute such a theory, it is doubtful that it would be correct or precise enough to pick out people’s (or animals’) actual mental states. (Piccinini, 2020, p. 97)

Is folk psychology a defective yet proto-scientific theory of mind and behaviour? Are its concepts theoretical concepts that stand to be refined in mature sciences of mind and behaviour? Is the primary task of folk psychology the explanation and prediction of behaviour in virtue of the attribution of propositional attitudes? Some 4E accounts of folk psychology answer negatively to all these questions (Ratcliffe, 2007; Hutto, 2008; Gallagher, 2020; Zawidzki, 2013; McGeer, 2007).1 “Folk psychology” rather refers to the various ways ordinary people come to understand and navigate the social world around them (Andrews et al., 2020). These ways are not necessarily propositional, and they are not primarily theoretical. There is no equivalence between the attribution of propositional attitudes and folk psychology: whereas some authors argue for the legitimate use of propositional attitudes in scientific psychology and even in neuroscience (Hochstein, 2017), 4E proponents underline that folk psychology does not necessarily involve the attribution of propositional attitudes. Folk psychology is neither a (proto)scientific theory nor a method (as simulation theories of folk psychology hold), because it is not primarily a descriptive enterprise, aiming at predicting and explaining behaviour from sideways on: it has a normative dimension (Dewhurst & Burr, 2022). This normative dimension is not exhausted by the normative presuppositions accompanying the use of intentional concepts. Folk psychology is first “a practice of showing how people’s performances live up to certain norms and thereby become, in that special way, intelligible” (McGeer, 2007, p. 141). Explaining and predicting behaviour in terms of beliefs and desires requires making sense of that behaviour by putting it in a network of normative practices  – not only norms of rationality, but also cultural and social norms. There are sorts of beliefs and desires that rational creatures are expected to form under some circumstances, and there are sorts of actions that rational creatures are

 The commonality between the approaches is here more important than the differences.

1

112

P. Steiner

expected to pursue in light of their beliefs and desires. According to this view, our attributions of intentional states contribute to regulate the behaviours of people and our interactions with them (we make sense of what they do by considering what they ought to do); those attributions rely on shared normative structures such as customs, standards and stereotypes. Prediction and explanation work because they capture those normative patterns of behaviour, for instance under the form of narratives. According to this view, explaining an action is like telling a story, making the action intelligible to us. And those justificatory explanations of behaviour exert a regulative influence on future instances of behaviour (Zawidzki, 2013). Consider the following example: you are explaining the fact someone is leaving a meeting room because she believed that someone in the audience has insulted her and because her desire not to be publicly insulted is stronger than her desire to stay in the room. This explanation is intelligible provided one understands the balance of desires at stake here. This understanding resorts to an attunement with normative facts (being insulted is worse than leaving a meeting). In view of that origin and background, explanations and predictions in folk psychology are not proto-­scientific attitudes. Interpreting behaviour by using intentional concepts is placing that behaviour in a network of norms and commitments. If it is the description of a fact, this fact is a relational and normative fact. Those norms and commitments are intertwined with the norms and commitments from which persons are defined, being ontologically different (neither superior nor inferior) from organisms and entities. The locus classicus here is Wilfrid Sellars’s “Philosophy and the scientific image”: To think of a featherless biped as a person is to think of it as a being with which one is bound up in a network of rights and duties. From this point of view, the irreducibility of the personal is the irreducibility of the ‘ought’ to the ‘is’. But even more basic than this (though ultimately, as we shall see, the two points coincide), is the fact that to think of a featherless biped as a person is to construe its behaviour in terms of actual or potential membership in an embracing group each member of which thinks of itself as a member of the group. (Sellars, 1962, p. 77)

Following those observations, here is a fourth claim on the normativity of cognition: Normative facts are built in our daily concepts of intentional states.

Of course, it is possible to hold that every concept, and not only intentional concepts, involves normative facts: this is notably the case of scientific concepts, if one understands them from an inferentialist perspective (Brandom, 1994). That being said, this broader acknowledgment of the normative nature of any concept is compatible with the recognition that the normative facts associated with daily intentional concepts are different from the operational and instrumental norms which define the meaning or reference of scientific concepts. But different in which sense?2 In order to be studied and identified, the structural and causal properties of neural mechanisms need to be qualified with concepts, but the role and the identity of those mechanisms are what they are independently of those concepts. Moreover, these  I will not deal here with the issue of deciding whether those differences entail (or not) theoretical, ontological or logical irreducibility. 2

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

113

concepts are not applied by considering whether those mechanisms are committed to norms of rationality or to inferential norms. By contrast, beliefs and desires are attributed against a background conception of what the agent ought to believe, desire or do. At some point, too many irrational beliefs threaten the intelligibility of the fact the agent really entertains beliefs, and goes against their identification in terms of content. Intentional explanations are informative if we assume the agent acts in accordance with the norms defining the content and the inferential connections between these intentional states. Unlike observational and theoretical concepts used in science, intentional concepts are not used for describing and labelling a reality (events, properties, mechanisms, states,…) which is independent of them and of our norms. The application of the concepts of “belief”, “desire” and “intention” is inseparable from a normative evaluation of what an agent ought to do and say, turning his sayings and doings into bearers of beliefs, desires and intentions. Moreover, this normative evaluation is not based on the consideration of a fixed set of law-governed conditions (as in “if this powder is potassium chloride, it should produce a purple flame when it is heated”). Davidson spoke as follows about the “normative character of mental concepts”: Beliefs, desires, intentions, and intentional actions must (…) be identified by their semantic contents in reason-explanations. The semantic contents of attitudes and beliefs determine their relations to one another and to the world in ways that meet at least rough standards of consistency and correctness. Unless such standards are met to an adequate degree, nothing can count as being a belief, a pro-attitude, or an intention. But these standards are norms, our norms, there being no others. (2004, p. 114)

I would like now to assess the scope of this claim on the normativity of daily intentional concepts when we consider the differences between folk psychology concepts and scientific concepts, including neuroscientific concepts.

7.5 Beyond Neuroscience as a Conveyor of Confirmations for 4E Cognition According to the mainstream view of 4E cognition, cognition is enacted in the couplings between living, embodied organisms and their environments: sensorimotor coupling, technological coupling, socio-cultural coupling, and linguistic coupling. In this mashup, the brain is not an organ of representation and of central control, but of mediation and integration (Di Paolo & De Jaegher, 2012; Fuchs, 2017). It is a part of an enactive system (Gallagher et al., 2013). For the cognitive activities of cephalic creatures, one will hold that there are neural correlates of these activities, without turning these neural facts into total realizers of those activities. Cognitive systems may be seen as extended and distributed across intracranial and environmental factors, with the brain being decentered (Gallagher, 2018): it is an unprivileged part of the mechanisms enabling cognitive agency. As Lambros Malafouris writes:

114

P. Steiner The centrality of the brain does not lie in its ability to constitute mentality by internalising and representing the world; rather, it lies in the ability of the brain to connect, to attend, to respond, to attune and relate to the world using its extraordinary plasticity and sensitivity. (Malafouris, 2019, p. 6)

It is therefore not surprising that when they take notice of what is happening in theoretical neuroscience, proponents of 4E cognition are seduced by dynamical models eschewing the classical functional decomposition/localization methodology, and by “beyond the brain”3 approaches in which neuroscientists claim that “mind is not traceable to any one structure” in so far as “bodily engagement and practiced patterns that characterize psychic life distribute across brain regions and infuse action as it occurs in the world” (Crippen & Schulkin, 2020, p. 15), that the brain is first a device delicatly geared to and for embodied action (Engel et  al., 2016), that the brain is a “complex causal mediator of the relationship between body and environment” (Anderson, 2014, p. xxii), or that “human cognition is not human brain function but human brain-body-environment function” (Anderson, 2014, p. 79; see also Northoff, 2018). Moreover, many neuroimaging data make it increasingly difficult to maintain that our classical cognitive categories, inherited and refined from folk psychology, are neatly and stably realized in the brain, even when the brain is defined and studied from its relations with the non-neural body and world. The observed functional differentiations in the brain are not necessary the results of functional specializations.4 There is a functional diversity of individual regions of the brain, and a frequent functional overlap between the constituents of different brain networks. On this issue, the mapping between cognitive functions (identified by concepts that figure in contemporary models of attention, object perception, reading,…) and neural structures becomes notoriously complex. Different functions cross-correlate with different structures. There are one-to-many mappings (a single cognitive function activates different structures) and many-to-one mappings (the same structure is implicated in many functions). For example, the so-called fusiform face area also responds to cars and birds. Broca’s area is more frequently activated by nonlanguage tasks (movement preparation, movement imitation, action recognition, music perception,…) than by language-related ones. The mapping is often many-to-many: cognitive processes require or activate different neural regions, and any given brain region supports multiple cognitive processes. This problem of multiple mapping between brain and psychological categories can be seen as a confirmation of the claim that the brain is decidedly not the sole unit of cognition. In  The expression has been proposed by Crippen and Schulkin (2020, p. 202).  As Price and Friston clearly argue,

3 4

it might be easy to ascribe a brain area a given function, but the validity of the functional label depends on the validity and scope of the cognitive model applied. This, in turn, depends on whether the model includes cognitive processes that are both necessary for the task and invoked incidentally. Furthermore, even if the cognitive model is fully specified, inferring that an area is involved in a function of interest is not the same as inferring that the area is specific to that function. To infer functional specificity requires a demonstration that an area is activated only by tasks that engage its function and no others (2005, p. 265)

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

115

order to identify the functional role of a brain region, one needs to consider the environment (bodily, worldly) in and with which this area is coordinated, and not only the anatomical regions the area coactivates with. The same neural structures can subserve different psychological functions in different situations. The mapping is not between cognitive functions and the brain alone, but between cognitive functions and brain-body-world system. Instead of intrinsic and abstract functions, we should instead opt in favour of context-sensitive (Klein, 2012) attributions of functions, the context including here other activated neural structures and networks, but also non-neural variables such as the environment in which the subject performs a task, or the etiology of the functions (Khalidi, 2017). This is another point of possible convergence with the methodology of 4E cognition.5 Nevertheless, there is another, more radical diagnosis that can be made on the basis of the complexity of the mapping between cognitive functions and neural structures. Instead of interpreting discoveries on the diverse tasks activating Broca’s area as confirmations of, for example, the idea that perception, action and language are dynamically intertwined (Crippen & Schulkin, 2020, pp. 61–62, 116, 182–183, 194), one can wonder whether the categories of folk psychology and of scientific psychology are still appropriate for describing and labelling what the brain does at different levels, in terms of functions and structures. If neural categories do not map onto psychological categories, it may be because the latter ones are ill-defined. More generally, it might be the case that “cognitive neuroscience is awash in a sea of conflicting terms and concepts” (Poldrack & Yarkoni, 2016, p. 588). One of the problems of cognitive neuroscience might be that the taxonomy of mental functions which is still widely used has been inherited from cognitive psychology models at odds with the functional organization of the brain – and just consider how psychology has long revendicated its autonomy from neuroscience!6 There is no reason for continuing to suppose that the ontology of the brain squares with the ontology of cognitive psychology at least if the latter descends from folk psychology, “our stone age psychological ontology” (Bunzl et al., 2010, p. 54). As Michael Anderson7 puts it, it would be something of a miracle if the functional categories developed for an autonomous science of the mind were simply the same as those that turn out to do the most explanatory work in a science of the brain (Anderson, 2014, pp. 131–132).

Classically, one starts from a folk or psychological specification of a cognitive operations, and ask “what are the neural correlates of this operation?”. For example, it is now well known that the left ventral occipito-temporal cortex is reliably associated with visual word recognition in adult normal readers (Cohen & Dehaene, 2004;

 See de Wit and Matheson (2022).  As we have seen above, other neuroscientists argue that the problem is not the taxonomy by itself, but the continuation of the method of describing neural structures in terms of intrinsic functions. 7  Unlike Poldrack and Price & Friston, Anderson does not maintain the project of arriving to oneto-one mappings between cognitive functions and neural structures. But he shares the project of revising our cognitive ontology. 5 6

116

P. Steiner

Dehaene & Cohen, 2011). However, should we infer from this association that this area is to be identified as the visual word form area (McCandliss et  al., 2003; Kronbichler et  al., 2004)? Should we label the explanans  – the observed neural structure – with the same term as the explanandum? The “visual word form area”, is ill-named: it responds not only to the viewing of words, but also to other visual objects (Price & Friston, 2005). The area does not have the specialization that the label implies. Saying that one area is involved in a function is not the same as identifying the role the area plays in that function. Ideally, one should redescribe tasks in terms of the functions of the areas involved: function should predict structure, and structure predicts function (Price & Friston, 2005). One of the new questions should be: “what is the function associated with the activation of this brain region?”. Here, labels matter, and our old labels, inherited from folk psychology, might prove to be misleading. The nomenclature of neuroscience  – the set of standardized terms researchers use in a systematic way in order to promote mutual understanding – and mind-based concepts, we have just seen, are often hard to square. For the same function, there might be a discrepancy between two labels: the mind-based label, specified on the basis of behavioural performances; and the anatomical label, specified on the basis of the regions that respond to the task. The “mind-based” (Buzsáki et  al., 2022) labels and concepts, are derived from, and nourish, a picture of the mind as consisting in a set of mental faculties or powers. The neuroscientist György Buzsáki expresses the problem as follows: Most of our terms to describe cognitive behavior have been inherited from philosophy rather than emerging from objective investigation of the brain itself. Although the ultimate goal of neuroscience is to understand how the brain generates behavior (i.e., how neuronal activity causes it), current mainstream behavioral/cognitive neuroscience appears to work in the opposite direction. Typically, we choose a subjectively defined term (e.g., volition, imagination, emotion) as the independent variable and look for dependent variables in the brain. This strategy is based on the assumption that the independent variable represents a real, objectively existing entity. More rarely, we begin with a distinct brain pattern (e.g., g frequency oscillation) and search for external correlates from our existing vocabulary. I will argue that most of our behavior-related terms emerged before and independent of neuroscience, and there is little guarantee that these terms correspond to circumscribed brain mechanisms. (2020, p. 2)

There is a huge a gap between neuroimaging data and our ways to name and label the structural, functional and anatomical properties of the brain if we rely on folk concepts (as when we speak about the primary visual cortex, or the dorsal attention network). What are the legitimate uses (if any) that one can make of those folk concepts for labelling what the brain does? The danger here is a problematic circularity: the same concepts – “attention”, “vision”, imagination” – could figure in both the explanans and the explanandum. Conversely, what is the degree of re-examination and update one should bring to these folk concepts, in light of neuroscientific findings? Some categories have already been eliminated from the scientific vocabulary of neuroscience (“concupiscence”, “avarice”, “parental love”, the Freudian “principle of pleasure”,…), other categories once thought to be unified have been splitted (“memory”), new categories have been coined (“response inhibition”)…. Consider the following example

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

117

given by Anderson (2015): the brain patterns observed during humour tasks are more similar to the patterns observed during semantic tasks than patterns observed during emotional tasks. Is the concept of “humour” appropriate here for naming the explanandum? Or should we infer that what we mean by “humour” is not an emotional experience? If we answer positively, we take for granted that what folk psychology means by “humour” aims at denoting a specific set of processes in the brain. According to Buzsáki, neuroscience cannot be an applied science: a technological practice consisting in measuring and observing phenomena previously defined and studied by philosophy and cognitive psychology, in continuity with the mind-­ based definitions of folk psychology. Neuroscientific observations and data must become landmarks for the proper or appropriate use of scientific concepts in neuroscience: ideally, those concepts should carve the brain at its joints. According to Poldrack, a putative cognitive process is real if an only if only one specific neural structure or network is activated in association with it (Poldrack, 2010, p. 754). We here reach issues pertaining to cognitive ontology: what is exactly the set of cognitive functions and structures investigated by the neurosciences? What are the categories that best describe the mind’s organization and the functions of the brain? Price and Friston introduced the term “cognitive ontology” in 2005. “Cognitive ontology” refers to the set of cognitive functions one appeals to when conducting neuroimaging studies. For the authors, an “ontology” is “a systematic definition of structure-function relations whereby structure predicts function and functions predict structures” (2005, p. 263). For Poldrack, a cognitive ontology is “a theory about the structure of the mind that specifies the component operations that comprise mental functions” (Poldrack, 2010, pp. 753–754). Poldrack’s Cognitive Atlas is an online database of experimental results, providing an inventory of known structures and relations. Following Janssen et al. (2017), besides being a nomenclature (a set of standardized terms used by researchers), a cognitive ontology can also be a domain (a set of entities to which a theory refers), and a set of metaphysical categories (events, processes, tropes, mechanicisms,…). Those three senses are interdependent. For instance, agreeing on a nomenclature ideally requires a consensus on the referents of the terms, and on the general metaphysical categories which are the most appropriate for characterizing the domain under study. How should we label the neural structures and areas? How should we name the functions which are correlated with the areas? In both cases, do we need to appeal to the concepts of folk psychology, philosophy and scientific psychology? Or to terms similar to those used by folk psychology? Are the processes that neuroscientists investigate realizers of the intuitive mental categories proper to folk psychology? Until now, proponents of 4E cognition have not paid much attention to those issues and debates in neuroscience.8 They should have done so. Crippen and Schulkin are correct in their diagnosis that “too many in the 4E camp pass over detailed discussions of the nervous system” (Crippen & Schulkin, 2020, p. 196).

 One exception being Hutto et al. (2017).

8

118

P. Steiner

They also often pass over a methodological reflection on the value of the concepts from which and by which we discuss about cognition, the brain and their relations. The plea here is to invite 4E researchers to go beyond the (true and important, if not trivial) claim that the brain is not the sole unit of cognition, and to widen their interests beyond the (debated) non-computational and non-representational nature of brain processes: it is to invite them to be (more) explicit on the status of the relations between folk concepts, scientific concepts, and neuroscientific concepts. One cannot but be amazed by the relative scarcity, in 4E cognition, of reflections concerning the origins and status of the concepts and preconceptions every scientist or philosopher starts from when cognition is under study. Everything happens as if a form of naive introspectionism were still the founding norm of research: before disagreeing on their localization and nature, and eventually appealing to sophisticated introspection or phenomenology as a way to study cognition, everybody is using terms such as “perception”, “understanding”, “reading”, “thinking” or “reasoning” by assuming we all know (by experience, by acquaintance) what these terms roughly mean or refer to (Buzsáki, 2020). One rarely considers that it might be methodologically valuable, in a preliminary fashion, to pay attention to our modes of ascription of psychological predicates, and to their differences with the other scientific predicates we use for defining and explaining the complex interplays between brain, body and the world. This issue does not only concern the nature of cognition and its relations with the brain; it more basically raises the question of the relations between folk psychology concepts and neuroscientific concepts. It is here that normativity matters, and especially the claim that normative facts are built in our daily concepts of intentional states. 4E cognition indeed harbours some potential resources for dealing with this issue of the differences between mind-based concepts and neuroscientific concepts. Let us now see how.

7.6 Neuroscience without Folk Psychology, Folk Psychology Without Literalism “Folk psychology concepts” are our daily intentional concepts, or what Francken and Slors (2014) call “commonsense cognitive concepts”: they are the concepts we use in daily life for explaining, predicting and interpreting behaviour. “Perceive”, “think”, “intend”, “believe”, “conscious”, “desire” or “reason” are examples of such concepts and verbs. On the scale of history, we have developed and used those concepts independently of the development of the brain sciences, or of the belief that the mind is in the brain (Danziger, 1997). As Buzsáki underlined in the quotation given in the last section, these concepts were not born out of scientific inquiries, and were not developed for scientific purposes. In a sense, we know what it is to see, believe or reason independently of scientific investigations: we implicitly know what are the (typical yet open-ended) normative patterns associated with the behaviours qualified by the use of those concepts. Any individual learns the correct use of

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

119

those concepts without getting acquainted with neuroscientific facts or evidence. Commonsense cognitive concepts may be used by neuroscientists themselves not only in their daily lives, but also for intraprofessional communication. They are also used to convey neuroscientific results to wider audiences. Sometimes, our daily intentional concepts and the scientific concepts used by cognitive scientists share the same words and predicates. Other scientific concepts have in common the name but not the meaning or the reference of daily concepts: the terms are the same, but not the concepts. “Memory”, “remembering” and “reading” are examples: what a layman means and associates with those terms is different from the understanding, distinctions and classification proposed by psychologists and neuroscientists. In order to study the neural basis of a cognitive function such as reading, one needs to operationalize that function in the form of a specific task. One may then measure the neural activity when the task is achieved, in order to obtain a mapping between neural structures and the cognitive function. Both functions and tasks often carry the mark of folk psychology (Francken et al., 2022). Commonsense cognitive concepts get often surreptitiously incorporated or maintained in the neuroscientist’s work. This may result in ambiguities and misunderstandings, as we have seen in the previous section. As Daniel Dennett warns: When neuroscientists help themselves to the ordinary terms that compose the lore I have dubbed “folk psychology” they need to proceed with the utmost caution, since these terms have presuppositions of use that can subvert their purposes and turn otherwise promising empirical theories and models into thinly disguised nonsense. (2007, p. 74)

Unlike Buzsáki, Poldrack or Anderson, Dennett does not dampen the use of attenuated versions of folk psychology concepts in neuroscience. According to Dennett, empirical discoveries in neuroscience show us that parts of our brain engage in processes that are strikingly like believing, deciding,… They resemble enough to our personal level behaviours so that we may name them with the same terms (“belief”, “decide”), with an attenuated sense. Brains do not have fully-fledged beliefs, desires, or memories but attenuated sorts of beliefs, memories and desires.9 Dennett here criticizes Maxwell Bennett and Peter Hacker’s criticism of the (alleged) mereological fallacy in neurosciences: the use of psychological predicates proper to persons for naming or labelling what parts of persons do. Dennett’s liberalism on the use of attenuated psychological and folk predicates in neuroscience can be made clearer and less ambiguous if one underlines the fact that folk psychology predicates and daily intentional concepts do not have the same purposes and norms of use than scientific concepts, including neuroscientific concepts.

 Bennett and Hacker reply as follows to Dennett’s suggestion:

9

Cognitive neuroscientists ask real questions  – they ask how the prefrontal cortices are involved in human thinking, why re-entrant pathways exist, what precisely are the roles of the hippocampus and neocortex in a human being’s remembering. Being told that the hippocampus sort of remembers for a short while and that the neocortex has a better sort of long-term memory provides no explanation whatsoever (2008:249).

120

P. Steiner

In the enactive accounts of folk psychology I mentioned earlier on, the application of intentional concepts to observable behaviour is mediated by norms and expectations. Those norms and expectations are different from the norms and expectations present in scientific circumstances. Indeed, when we use folk concepts, it is not for naming lawful natural relations, but for characterizing/predicting/rationalizing behaviour, inside of a normative network. When we use “see”, “believe” or “desire” for characterizing what persons do, we expect those persons to behave in specific ways. As Dennett himself underlined a long time ago, when we attribute beliefs, desires or intentions on the basis of behavioural performances, we do not speculate about the concrete causes responsible for that behaviour. Folk psychology concepts are very poor trackers of fine-grained neuroscientific states and processes, and more basically of cognitive functions and architectures. In truth, those concepts are not denotative concepts – or, at least, their function is not to designate the same events and processes as the ones which are investigated by neuroscience and scientific psychology. Thinking that these daily concepts, in order to work, should be identified with inner mechanisms, is – to quote Dennett’s nice words – “a gratuitous bit of misplaced concreteness” (Dennett, 1987, p. 51). The relations between those concepts and brain processes are not direct; they are mediated by factors including normative factors. The characterization of behaviour with mind-based concepts, we have seen, makes sense against a background of norms and normative expectations. The use of scientific concepts also involves normative commitments, but those are not the same norms. This difference between (neuro)scientific concepts and folk psychology concepts does not entail a rivalry. Folk psychology concepts and neuroscientific concepts are mutually articulated in the following way: when we use folk psychology concepts for making sense of behaviour, we rely on behavioural performances. Those behavioural performances are produced by a rich array of mechanisms and processes, including the neural mechanisms and processes studied by neuroscience. As Mason Cash writes, any neurological investigation will look at neural states causally related to the publicly observable features by virtue of which [an] intentional state is typically judged to be appropriately ascribed: the features of the person’s situation that they perceive and the actions they perform in that situation. (Cash, 2009, p. 137)

The predictive and explanatory successes of both folk psychology and neuroscientific concepts are different. In folk psychology, one succeeds in explaining and understanding the behaviour of a person by placing it in a network of commitments and norms. In neuroscience, one captures the subpersonal mechanisms at the basis of this observable behaviour. In both cases, we have explanatory powers, but folk psychology constructs are not explanatory because they are ultimately instantiated in the processes and events investigated by neuroscientists. Conversely, those processes and events explain behaviour, but not in virtue of instantiating some particular mental functions as they are defined by folk psychology (see Burnston, 2021, for a similar view). Intentional explanations have causal powers (they allow us to track, predict and explain how agents will behave) by identifying real patterns in a world including social norms. Those patterns are not reflections of the patterns taking place in the brain (Dennett, 1991).

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

121

In spite of the recent and aforementioned appeals in neuroscience to demarcate mind-based concepts from scientific concepts, one can be a literalist about folk psychology: the concepts of folk psychology are denotative concepts; they aim at denoting real processes and entities inside of individuals, the same processes as those studied by mature sciences of the mind. Folk psychology is a (proto)theory. Since neuroscientific concepts are also denotative and are different from folk psychology concepts, both family of concepts are concurrent concepts. This literalism was assumed by both Fodor’s propositional attitudes realism, and Churchland’s eliminativism (according to which folk psychology concepts designate grossly; they ought to be replaced by neuroscientific concepts). More recently, Carrie Figdor (2018) defended a new version of literalism. Literalism, in her case, is the claim that the use of psychological predicates for reporting what neurons, bacteria, or flies do should be understood literally. Saying that neurons prefer, plants decide or that bacteria communicate linguistically is not a façon de parler or a metaphorical talk. Be they used for characterizing humans or non-human animals, organisms and organs, those predicates refer to the same facts. Literalism surely goes against our common intuitions and daily uses of those predicates, but it comes with the idea that the extensions of predicate denoting natural phenomena is not fixed by common sense and intuitions, but by science. The scientific uses of psychological predicates are univocal across human and nonhuman contexts: The scientific uses of psychological predicates across a wide range of unexpected domains are literal with the same reference that they have in the human domain. (Figdor, 2018, p. 166)

An attenuated version of literalism – but not under this name – might occasionally  be endorsed by some proponents of 4E cognition: autopoietic enactivism (Thompson, 2007), and its claim that cognition and life are (more or less) equivalent phenomena, is keen to attribute cognitive activities (preferences, evaluations, spatial orientation) and “sense-making” to basic forms of life such as bacteria. Contrary to literalism, one can be a pluralist (McCaffrey & Wright, 2022; Hochstein, 2016; Sullivan, 2017). This is the option I am defending here, and it resonates with other stances in the 4E research program: the aims and purposes of folk psychology and neuroscientific vocabulary are different.10 They serve different aims. Folk psychology is a tool dealing with coarse-grained behaviour, whereas neuroscientific vocabulary aims at describing and explaining fine-grained mechanisms and the functional organization of the brain. Folk psychology concepts are not about neural organization: they are about behavioural performances and patterns caused by neural organization, but it is not this neural etiology which directly turns these performances into bearers of intentional properties and patterns. It is the entanglement of those performances in normative practices. When we attribute

 Other versions of pluralism will insist that inside of neuroscience itself, there is and should be a plurality of purposes, aims and categories - some of those categories being congruent with folk psychology categories, and involving the use of propositional attitudes (see for instance Hochstein, 2016, 2017). 10

122

P. Steiner

thoughts, intentions, beliefs or desires to someone, we make no commitments about the structures and entities causally responsible for that behaviour. And we do not only make dispositional attribution: because of their beliefs, persons are not only disposed but also expected to behave a certain way. We are placing their behaviour in a network of norms and sociocultural constraints. Folk psychology concepts do not refer to discrete neural processes or states that causally explain predicted behaviour: their fruitful applicability to human creatures rather show that our brains harbour mechanisms that causally explain the relevant behavioural patterns on which we rely for applying folk psychology concepts, in normative environments (Francken & Slors, 2014, pp. 254–255). At one place in her book Pieces of mind, Figdor expresses a view which can be interpreted as a form of local pluralism. She considers the case of beliefs, desires, and intentions, and concedes that explaining behaviour with those concepts presumes, she writes in a Davidsonian/Dennettian spirit, “some form of human rationality” (Figdor, 2018, p. 60), a “second order normative assessment of first-order cognitive states” (2018, p. 60 n.1). Attributing “some form of human rationality” is a “second order normative assessment of first-order cognitive states”, but not all second order normative assessments of first-order cognitive states consist in the attribution of some form of human rationality. The assessment of rationality consists here in placing the observed behaviour in a network of norms and duties. Figdor expresses cautiousness about the prospect that, one day, beliefs, intentions and desires will be the subject of scientifically motivated extension and revision, as other psychological predicates such as “expect” or “prefer” already are. Therefore, even in her view, some psychological predicates are predicates whose use and ascription primarily makes sense from a specific normative background. In Sect. 7.4, I have started from the observation that some works in the 4E tradition emphasize the normative nature of folk psychology and of folk psychological concepts. In this section, I have just developed the consequences of this observation when one considers the relations between those concepts (even used by scientists) and neuroscientific concepts and phenomena. There is neither continuity nor discontinuity: we are on different perspectives on behaviour, characterized by different purposes, norms and contexts of exercise.

7.7 Conclusion In this paper, I have proposed several distinctions for describing and clarifying the various ways 4E cognitive science dealt with the pervasive phenomenon of normativity. Normativity can be an explanandum of 4E approaches, but also an essential property of its explanantia. Methodological, metaphysical and semantic priorities can be given to normativity in the definition and explanation of cognitive phenomena. But 4E cognition also accommodates resources for appreciating the normative dimensions of concepts, and especially daily intentional concepts, in relation with folk psychology. It is here that new stakes appear for 4E cognition in relation with

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

123

neuroscience. Indeed, I have exhorted proponents of 4E cognition to go beyond the claim that cognition is distributed across brain, body and world, for taking position in recent debates on cognitive ontology. The methodology of cognitive science – including 4E cognitive science – should include reflections on the status of the concepts by which we understand the contribution of the brain to cognitive systems. On this topic, I have pleaded for a pluralism insisting on the normative differences between folk psychology concepts and neuroscientific concepts.

References Allen-Hermanson, S. (2013). Superdupersizing the mind: Extended cognition and the persistence of cognitive bloat. Philosophical Studies, 164, 791–806. Anderson, M. (2014). After phrenology. Neural reuse and the interactive brain. MIT Press. Anderson, M. (2015). Mining the brain for a new taxonomy of the mind. Philosophy Compass, 10(1), 68–77. Andrews, K. (2020). Naïve normativity: The social foundation of moral cognition. Journal of the American Philosophical Association, 6(1), 36–56. Andrews, K., Spaulding, S., & Westra, E. (2020). Introduction to folk psychology: Pluralistic approaches. Synthese, 199(1–2), 1685–1700. Bennett, M., & Hacker, P. M. S. (2008). History of cognitive neuroscience. Wiley-Blackwell. Brandom, R. (1994). Making it explicit: Reasoning, representing, and discursive commitment. Harvard University Press. Bunzl, M., Hanson, S. J., & Poldrack, R. (2010). An exchange about localism. In S. J. Hanson & M. Bunzl (Eds.), Foundational issues in human brain mapping (pp. 49–54). MIT Press. Burnston, D. (2021). Cognitive ontologies, task ontologies, and explanation in cognitive neuroscience. In J. Bickle, C. F. Craver, & A. S. Barwich (Eds.), The tools of neuroscience experiment: Philosophical and scientific perspectives (pp. 259–283). Routledge. Buzsáki, G. (2020). The brain-cognitive behavior problem: A retrospective. Eneuro, 7(4). Buzsáki, G., McKenzie, S., & Davachi, L. (2022). Neurophysiology of remembering. Annual Review of Psychology, 73(1), 187–215. Cash, M. (2009). Normativity is the mother of intention: Wittgenstein, normative practices and neurological representations. New Ideas in Psychology, 27(2), 133–147. Casper, M. O. (2018). Social enactivism: On situating high-level cognitive states and processes. De Gruyter. Cohen, L., & Dehaene, S. (2004). Specialization within the ventral stream: The case for the visual word form area. NeuroImage, 22(1), 466–476. Crippen, M., & Schulkin, J. (2020). Mind ecologies: Body, brain, and world. Columbia University Press. Danziger, K. (1997). Naming the mind: How psychology found its language. SAGE Publications. Davidson, D. (1980). Essays on actions and events. Clarendon Press. Davidson, D. (1984). Inquiries into truth and interpretation. Oxford University Press. Davidson, D. (2004). Problems of rationality. Oxford University Press. Davies, P. (2001). Norms of nature. MIT Press. De Caro, M., & Macarthur, D. (Eds.). (2022). The Routledge handbook of liberal naturalism. Routledge. de Wit, M. M., & Matheson, H. E. (2022). Context-sensitive computational mechanistic explanation in cognitive neuroscience. Frontiers in Psychology, 13, 903960. Dehaene, S., & Cohen, L. (2011). The unique role of the visual word form area in reading. Trends in Cognitive Sciences, 15(6), 254–262.

124

P. Steiner

Dennett, D. (1987). The intentional stance. MIT Press. Dennett, D. (1991). Real patterns. Journal of Philosophy, 88(1), 27–51. Dennett, D. (2007). Philosophy as naive anthropology: Comment on Bennett and Hacker. In M. Bennett, D. Dennett, P. M. S. Hacker, & J. Searle (Eds.), Neuroscience and philosophy: Brain, mind, and language (pp. 73–95). Columbia University Press. Dewhurst, J., & Burr, C. (2022). Normative folk psychology and decision theory. Mind and Language, 37(4), 525–542. Di Paolo, E., & De Jaegher, H. (2012). The interactive brain hypothesis. Frontiers in Human Neuroscience, 6, 163. Doyon, M., & Breyer, T. (Eds.). (2015). Normativity in perception. Palgrave Macmillan. Engel, A.  K., Friston, K.  J., & Kragic, D. (Eds.). (2016). The pragmatic turn: Toward action-­ oriented views in cognitive science. MIT Press. Figdor, C. (2018). Pieces of mind: The proper domain of psychological predicates. Oxford University Press. FitzPatrick, W. J. (2000). Teleology and the norms of nature. Routledge. Fitzpatrick, S. (2020). Chimpanzee normativity: Evidence and objections. Biology and Philosophy, 35(4), 1–28. Francken, J. C., & Slors, M. (2014). From commonsense to science, and back: The use of cognitive concepts in neuroscience. Consciousness and Cognition, 29, 248–258. Francken, J. C., Slors, M., & Craver, & C. F. (2022). Cognitive ontology and the search for neural mechanisms: Three foundational problems. Synthese, 200(5), 1–22. Fuchs, T. (2017). Ecology of the brain: The phenomenology and biology of the embodied mind. Oxford University Press. Gallagher, S. (2018). Decentering the brain: Embodied cognition and the critique of neurocentrism and narrow-minded philosophy of mind. Constructivist foundations, 14(1), 101–134. Gallagher, S. (2020). Action and interaction. Oxford University Press. Gallagher, S., & Crisafi, A. (2009). Mental institutions. Topoi, 28(1), 45–51. Gallagher, S., Hutto, D., Slaby, J., & Cole, J. (2013). The brain as part of an enactive system (commentary). Behavioral and Brain Sciences, 36(4), 421–422. https://doi.org/10.1017/ S0140525X12002105 Greenberg, M. (2005). A new map of theories of mental content: Constitutive accounts and normative theories. Philosophical Issues, 15(1), 299–320. Haugeland, J. (1998). Having thought: Essays in the metaphysics of mind. Harvard University Press. Hochstein, E. (2016). Categorizing the mental. Philosophical Quarterly, 66(265), 745–759. Hochstein, E. (2017). When does ‘folk psychology’ count as folk psychological? British Journal for the Philosophy of Science, 68(4), 1125–1147. Hutto, D. D. (2008). Folk psychological narratives: The sociocultural basis of understanding reasons. Bradford. Hutto, D. D., & Myin, E. (2013). Radicalizing enactivism: Basic minds without content. MIT Press. Hutto, D. D., & Myin, E. (2017). Evolving enactivism: Basic minds meet content. MIT Press. Hutto, D. D., & Satne, G. (2015). The natural origins of content. Philosophia, 43(3), 521–536. Hutto, D. D., Peeters, A., & Segundo-Ortin, M. (2017). Cognitive ontology in flux: The possibility of protean brains. Philosophical Explorations, 20(2), 209–223. Janssen, A., Klein, C., & Slors, M. (2017). What is a cognitive ontology, anyway? Philosophical Explorations, 20(2), 123–128. Khalidi, M.  A. (2017). Crosscutting psycho-neural taxonomies: The case of episodic memory. Philosophical Explorations, 20(2), 191–208. Kiesewetter, B. (2017). The normativity of rationality. Oxford University Press. Klein, C. (2012). Cognitive ontology and region-versus network-oriented analyses. Philosophy of Science, 79(5), 952–960. Korsgaard, C. (1996). The sources of normativity. Cambridge University Press.

7  Normativity and the Methodology of 4E Cognition: Taking Stock and Going Forward

125

Kronbichler, M., Hutzler, F., Wimmer, H., et  al. (2004). The visual word form area and the frequency with which words are encountered: Evidence from a parametric fMRI study. NeuroImage, 21(3), 946–953. Lewis, D.  K. (1972). Psychophysical and theoretical identifications. Australasian Journal of Philosophy, 50(3), 249–258. Maher, C. (2012). The Pittsburgh school of philosophy: Sellars, McDowell, Brandom. Routledge. Malafouris, L. (2019). Mind and material engagement. Phenomenology and the Cognitive Sciences, 18, 1–17. McCaffrey, J., & Wright, J. (2022). Neuroscience and cognitive ontology: A case for pluralism. In F. De Brigard & W. Sinnott-Armstrong (Eds.), Neuroscience and philosophy (pp. 427–466). MIT Press. McCandliss, B. D., Cohen, L., & Dehaene, S. (2003). The visual word form area: Expertise for reading in the fusiform gyrus. Trends in Cognitive Sciences, 7(7), 293–299. McGeer, V. (2007). The regulative dimension of folk psychology. In D. D. Hutto & M. Ratcliffe (Eds.), Folk psychology re-assessed (pp. 137–156). Kluwer/Springer Press. Menary, R. (2007). Cognitive integration: Mind and cognition unbounded. Palgrave Macmillan. Merton, R. (1942). The normative structure of science. In N. W. Storer (Ed.), The sociology of science (pp. 267–278). University of Chicago Press. Mulligan, K. (1999). Justification, rule-breaking and the mind. Proceedings of the Aristotelian Society, 99(2), 123–139. Newen, A., De Bruin, L., & Gallagher, S. (Eds.). (2018). The Oxford handbook of 4E cognition. Oxford University Press. Northoff, G. (2018). The spontaneous brain. MIT Press. Piccinini, G. (2020). Neurocognitive mechanisms: Explaining biological cognition. Oxford University Press. Piccinini, G. (2022). Situated neural representations: Solving the problems of content. Frontiers in Neurorobotics, 16, Article 846979. Poldrack, R. (2010). Mapping mental function to brain structure: How can cognitive neuroimaging succeed? Perspectives in Psychological Sciences, 5(6), 753–761. Poldrack, R., & Yarkoni, T. (2016). From brain maps to cognitive ontologies: Informatics and the search for mental structure. Annual Review Psychology, 67, 587–612. Price, C. J., & Friston, K. J. (2005). Functional ontologies for cognition: The systematic definition of structure and function. Cognitive Neuropsychology, 22(3–4), 262–275. Ratcliffe, M. (2007). Rethinking commonsense psychology: A critique of folk psychology, theory of mind and simulation. Macmillan. Rietveld, E. (2008). Situated normativity: The normative aspect of embodied cognition in unreflective action. Mind, 117(468), 973–1001. Sanches de Oliveira, G. (2022). The strong program in embodied cognitive science. Phenomenology and the Cognitive Sciences. https://doi.org/10.1007/s11097-­022-­09806-­w Schmidt, M., & Rakoczy, H. (2018). Developing an understanding of normativity. In A. Newen, L.  De Bruin, & S.  Gallagher (Eds.), The Oxford handbook of 4E cognition (pp.  685–705). Oxford University Press. Schmidt, M., & Rakoczy, H. (forthcoming). On the uniqueness of human normative attitudes. In K. Bayertz & N. Roughley (Eds.), The normative animal? On the anthropological significance of social, moral and linguistic norms. Oxford University Press. Segundo-Ortin, M. (2022). Socio-cultural norms in ecological psychology: The education of intention. Phenomenology and the Cognitive Sciences. https://doi.org/10.1007/s11097-­022-­09807-­9 Sellars, W. (1956). Empiricism and the philosophy of mind. Minnesota Studies in the Philosophy of Science, 1, 253–329. Sellars, W. (1962). Philosophy and the scientific image of man. In R. Colodny (Ed.), Science, perception, and reality (pp. 35–78). Humanities Press. Steiner, P. (2014). The delocalized mind. Judgements, vehicles, and persons. Phenomenology and the Cognitive Sciences, 13(3), 1–24.

126

P. Steiner

Sullivan, J. (2017). Coordinated pluralism as a means to facilitate integrative taxonomies of cognition. Philosophical Explorations, 20(2), 129–145. Taylor, C. (2016). The language animal: The full shape of the human linguistic capacity. Harvard University Press. Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of mind. Harvard University Press. van den Herik, J. (2021). Rules as resources: An ecological-enactive perspective on linguistic normativity. Phenomenology and the Cognitive Sciences, 20(1), 93–116. Zawidzki, T. W. (2013). Mindshaping: A new framework for understanding human social cognition. MIT Press.

Part II

Ontology of the Mind

Chapter 8

Extended Cognition and the Search for the Mark of Constitution – A Promising Strategy? Beate Krickel

Abstract  The disagreement between defenders and opponents of extended cognition is often framed in terms of constitution. The underlying principle of this discussion is what I will call the co-location principle: cognition is located where its constituents are located. The crucial question is under which conditions something is to be counted as a constituent of cognition. I will formulate three criteria of adequacy that an account of constitution must satisfy to be applicable to the dispute on extended cognition. I will evaluate different accounts of constitution as to whether they satisfy the criteria of adequacy. Three different accounts of constitution can be identified in the literature on extended cognition, which I will call property-based constitution, diachronic constitution, and mechanistic constitution. I will show that none of the accounts satisfies all three criteria. Thus, the search for the mark of constitution does not seem to be a promising strategy for defending extended cognition. I will briefly sketch a new strategy of how to deal with this situation.

8.1 Introduction Many proponents of situated cognition argue that cognition does not only happen in the brain. Rather, cognition is said to be extended. This claim is usually spelled out in terms of constitution: cognition is constituted by elements external to the brain, such as bodily factors, tools, or social interactions. For example, according to Richard Menary the disagreement between defenders and opponents of extended cognition concerns the question “whether the manipulation of external vehicles constitutes a cognitive process” (2006, p. 329). This constitution-claim is taken to imply that the external factors, such as manipulations of external vehicles, are not merely causal inputs or outputs of an otherwise isolated cognitive system. Apart B. Krickel (*) Institute of History and Philosophy of Science, Technology, and Literature, Technical University Berlin, Berlin, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_8

129

130

B. Krickel

from this negative characterization, however, there is no agreement on what constitution implies. Proponents of extended cognition are aware of the need for a consistent account of constitution. Indeed, according to Baumgartner and Wilutzky (2017), it has become common to frame the discussion about extended cognition not in terms of the search for the mark of the cognitive, as it has traditionally been done, but as the search for the mark of constitution.1 However, so far, no account of constitution has been proposed that is  generally accepted. Three accounts of constitution can be identified in the literature on extended cognition: what I call property-based constitution, diachronic constitution, and mechanistic constitution. I will argue that none of these accounts can be used to settle the debate on extended cognition – at least not how the debate is currently framed. To show this, I develop criteria of adequacy based on which the three accounts of constitution can be evaluated. I will show that none of the accounts satisfies all criteria. To still be able to fruitfully discuss the question of the location of cognition, I will conclude that we should adopt the mechanistic account and allow for pragmatic norms to partially determine the location of cognition. In other words, I will argue that we should give up on a fully realistic notion of the location of cognition. The paper proceeds as follows: in Sect. 8.2, I briefly summarize what the debate on extended cognition is about, show that the co-location principle is one of its background assumptions, and formulate three criteria of adequacy that an account of constitution in the context of the debate on extended cognition needs to satisfy. In Sects. 8.3, 8.4, and 8.5, I will present the different accounts of constitution that have implicitly or explicitly been applied in the literature on extended cognition. In Sect. 8.3, I will present the first account of constitution that has been put forward in the debate on extended cognition, which I will call “property-based constitution”. I will argue that it violates at least one criterion of adequacy. In Sect. 8.4, I will present “diachronic constitution” and argue that  it cannot account for all criteria of adequacy either. In Sect. 8.5, I will present “mechanistic constitution” – which is the most elaborate account – and argue that, in the end, it fails at least one criterion of adequacy as well. In Sect. 8.6, I will conclude that, therefore, the search for the mark of constitution does not appear to be promising for the settlement of the debate on extended cognition. I will present different options of how to deal with the situation.

8.2 Extended Cognition, Constitution & Criteria of Adequacy The debate on extended cognition is a debate about the location or the boundaries of cognition (Gallagher, 2018b; Rupert, 2010). The dispute is whether cognitive processes or cognitive states are located only inside the brain (or parts thereof), or  Note that it is unclear whether the two strategies are distinct. As will become clear below, the search for the mark of the cognitive can be regarded as one version of the search for the mark of constitution. 1

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

131

whether cognition may also occur in the body and the physical or social environment of an agent. This dispute is often framed in terms of constitution: defenders of extended cognition hold that at least sometimes cognition is at least partly constituted by elements external to the brain. Thus, there are some cognitive processes or states that extend beyond the brain, i.e., that are (at least partially) located outside the brain. Opponents of extended cognition argue that cognition is always constituted only by elements inside the brain. Thus, the brain is the locus of cognition. The arguments of both camps have a similar underlying structure, which I will call the “Argument from Constitution”: (Argument from Constitution) 1. Cognition is constituted by X. 2. All (entities that instantiate) X are inside the brain. / There is an (entity instantiating) X that is outside the brain. 3. Thus, cognition is located only in the brain. / Thus, cognition is sometimes (partially) located outside the brain. For this argument to be valid, a further assumption must be made, which is an implicit assumption of both camps. This further assumption concerns the relation between constitution and the location of cognition. I will call this implicit assumption the Co-location principle: Co-location Principle: Cognition is located where and only where its constituents are located. Only if the Co-location Principle is presupposed, the inference from 1 and 2 to 3 is valid. Two points need to be noted: first, strictly speaking defenders of extended cognition only need a weaker version of the Co-location Principle. To render their version of the Argument from Constitution valid, it would suffice to assume that the spatiotemporal region of the constituents is a sub-region of the spatiotemporal region of cognition (i.e., delete the “… and only where” form the formulation above). However, also for defenders of extended cognition, the motivation for using constitution-talk in the present context is the assumption that the constituents form the supervenience base of cognition, which implies that there can be no cognition if no constituent is present. Second, while the quest for the “location of cognition” may sound like the quest for the “spatial location of cognition” (Kirchhoff & Kiverstein, 2020, p. 11), the Co-location Principle clearly is to be understood in terms of spatial as well as temporal location (Clark, 2008, pp.  77, 114). Constitution is a dependence relation between a constituee and a set of constituents. The constituee’s existence depends on the existence of the constituents. Thus, the constituee exists only when and where its constituents exist. The Co-location Principle implies that cognition exists at the same time and place where its constituents exist. The overall question discussed in the literature on extended cognition is whether at the time of the occurrence of a cognitive process, such as Otto’s remembering the way to MOMA or a person’s playing Tetris, an external element is involved in the relevant way. Indeed, I do not see how to make sense of the idea that constitution in the present context could be understood only in terms of spatial co-location without temporal co-location.

132

B. Krickel

Based on the previous considerations, in their search for the location of cognition, defenders and opponents of extended cognition fight about what the constituents of cognition are; e.g., whether “the manipulation of external vehicles constitutes a cognitive process” (Menary, 2006, p. 329). But how to determine what the constituents of cognition are? Under which conditions does something count as a constituent of something else? Defenders as well as opponents of extended cognition alike must have answers to these questions to get their version of the Argument from Constitution running. Different notions of constitution have been explicitly defended or implicitly suggested by different authors. Before I present these accounts, I want to identify criteria of adequacy that help us evaluate whether the respective account of constitution can be plugged into the Argument from Constitution to render it valid. A first criterion can be derived from the need of the implicit premise, i.e., the Co-Location Principle: 1. Preservation of the Co-Location Principle: The account of constitution must validate the Co-location Principle. That is, whatever constitution turns out to be, for the Argument from Constitution to be valid, constitution should be such that the Co-Location Principle turns out to be true. A further criterion is: 2. Neutrality regarding Framework of Cognition: The account of constitution must in principle be acceptable by defenders as well as opponents of extended cognition. If (2) is violated, the respective account of constitution would just beg the question against the other camp. Furthermore, the real dispute between defenders and opponents of extended cognition would turn out to concern not the location of cognition, but the nature of constitution. 3. Neutrality regarding Location: The account of constitution must not presuppose assumptions about the location of cognition. An account of constitution that tells us what the constituents of cognition are by already presupposing where cognition is, would obviously be of no help in the present context. The whole idea of the search for the mark of constitution is to determine the location of cognition by identifying its constituents, not the other way around. Furthermore, if condition (3) is violated probably also condition (2) is violated. In the next sections, I will introduce three notions of constitution that have explicitly or implicitly been suggested in the debate on extended cognition.2 The  Kirchhoff (2015) argues that opponents of extended cognition, especially Adams and Aizawa, presuppose material constitution in their criticism of extended cognition. Material constitution is discussed in analytic metaphysics. It denotes the relation between an object and what it is made of, such as statues and lumps of clay (Wasserman, 2018). Kirchhoff argues that the critics’ objections against extended cognition (more specifically, the coupling-constitution fallacy) presuppose material constitution. However, this notion cannot be applied to putative cases of extended cognition because extended cognition is about cognitive processes, rather than objects. While I agree with 2

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

133

first I will call “property-based constitution”,3 the second is called “diachronic constitution”, the third is “mechanistic constitution”.

8.3 Property-Based Constitution Defenders of property-based constitution, how I will understand it here, identify one or more properties that they take to characterize cognition and that, therefore, an entity must instantiate to count as a constituent of cognition. This corresponds to the way the expressions “to constitute” and “is constitutive of” are generally used in academic contexts. For example, Wieland (2014) uses the following example to illustrate what it means for something to be constitutive of something else: reporting extraordinary thoughts arrived at by non-empirical means is constitutive of writing a philosophical paper – “for if you do not obey this rule you are just writing something else” (Wieland, 2014, p. 250). Accounts of property-based constitution identify a property (or a set of properties) that is necessary and/or sufficient for some element that instantiates the property to be a cognitive element. The academic-­ everyday use of “being constitutive of” is ambiguous with respect to whether it expresses a necessary and sufficient condition, only a necessary condition, or only a sufficient condition. Opponents of extended cognition who adopt a property-based account of constitution interpret the term at least in terms of necessary conditions. For defenders of extended cognition, it is sufficient to interpret “constitutive of” in terms of providing sufficient conditions that need not be necessary. Indeed, many authors in the debate on extended cognition seem to rely on property-­based constitution. For example, Fred Adams and Ken Aizawa (2008); (Aizawa & Adams, 2005) take underived content to be the property that an entity needs to instantiate in order to count as cognitive. They seem to take this to be a necessary as well as sufficient condition for being cognitive.4 Accordingly, they identify those spatiotemporal regions as the location of cognition that are occupied by entities that instantiate underived content—which according to them are to be found only in the brain. Clark and Chalmers (1998), in contrast, assume that cognitive states are constituted by states that have a specific functional role for which it is Kirchhoff that material constitution is not adequate to spell out constitution in the context of extended cognition for that reason, I disagree that anybody (at least not Adams and Aizawa) opposed extended cognition based on material constitution. Indeed, Adams and Aizawa’s line of reasoning is independent of any specific notion of constitution—it just makes use of the intuitive judgement that, say, the heater or air conditioner is not constitutive of the expansion of the bimetallic strip in a thermostat just because the former is causally linked to the latter (Adams & Aizawa, 2001). Furthermore, everyone seems to agree that the location of cognition concerns at least also the location of cognitive processes, which is obvious already in Adams and Aizawa (2001). 3  This is similar to what David Kaplan calls “proprietary demarcation criteria” (Kaplan, 2012, p. 548). 4  Note that it is not entirely clear whether Adams and Aizawa take underived content to be necessary and sufficient for being cognitive. For a discussion see Menary (2010, p. 606).

134

B. Krickel

true that “were it done in the head” it would count as cognition. An entity that instantiates the property of having a specific functional role is thereby cognitive. Thus, Clark and Chalmers identify spatiotemporal regions occupied by entities with that functional role as the location of cognition – and those may be located outside the brain. Further property-based accounts of constitution have been suggested. I won’t discuss them in all detail as this has been done in many other publications already (see for example Kaplan (2012)). Do property-based accounts of constitution satisfy the criteria of adequacy formulated in the previous section? Do property-based accounts of constitution justify the Co-location Principle? They do: if being X is what makes something cognitive, the cognitive will be wherever and whenever X is instantiated (compare: writing a philosophical essay happens wherever and whenever there is someone writing a text reporting extraordinary thoughts arrived at by non-empirical means).5 Furthermore, property-based constitution does not make any presuppositions about the location of cognition (as long as the relevant property is not taken to be “being located inside the brain” or the like). Still, property-based accounts of constitution are problematic. The reason is that they are likely to fail the second criterion of adequacy: every version so far suggested begs the question against the respective opponent (this has been observed by, e.g., Kaplan (2012)). In other words, the problem for relying on property-based accounts of constitution is that the real disagreement, then, is how to define cognition in the first place, or in other words, what “the mark of the cognitive” is. Some authors (such as Kaplan (2012)) have argued that there is no definition of cognition that opponents as well as defenders of extended cognition agree on, and that thus, any stipulation on what cognition is begs the question against the respective other camp. Kaplan, therefore, proposes to leave the search for the mark of the cognitive behind and concentrate on the search for the “mark of constitution” (also diagnosed by Baumgartner and Wilutzky (2017) as a common trend; see introduction). Of course, these considerations do not imply that there cannot be a variant of property-based constitution that could satisfy the second criterion of adequacy and be acceptable by both camps. However, given that the dispute is already going on for some time and no account that both agree on has been suggested, it might be more fruitful to look for a different understanding of constitution that is independent of how cognition is to be defined in the first place. This is what the other notions of constitution are taken to be.

 Again, note that there is a difference between how the Argument from Constitution is to be formulated based on property-based constitution for the defender compared to the opponent of extended cognition. While both opponents and defenders of extended cognition may identify a disjunction of different features as each sufficient for the cognitive, the opponent of extended cognition must show that none of these features is exemplified by something outside the brain. The defender of extended cognition must only show that there is one element outside the brain that exemplifies one of the features in the disjunction. 5

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

135

8.4 Diachronic Constitution Standardly, constitution is taken to be a synchronous relation, implying that the relata of constitution occur at the same time (property-based constitution as well as mechanistic constitution are in this sense standard accounts). Defenders of diachronic constitution reject this standard assumption. According to defenders of diachronic constitution, constitution is a dependence relation that holds between entities that are causally connected in a specific way. Usually, this “specific way” is spelled out in terms of “reciprocal causation” or “reciprocal coupling” (Clark, 2014; Kirchhoff, 2015; Menary, 2006). When looking at the different discussions of diachronic constitution or reciprocal causation, however, it becomes clear that there are important differences between the writings of different authors. Take Menary’s analysis of the popular Otto-example: Schematically, X is manipulations of the notebook reciprocally coupled to Y – neural processes – which together constitute Z, the process of remembering. (Menary, 2007, p. 63)

This is often interpreted as a formulation of the diachronic account (Kirchhoff, 2015, p. 323). However, in Menary’s quote the relata of the causal coupling (X and Y) and the relata of the constitution relation (X + Y and Z) are not the same. But it is only the relation between X and Y that is diachronic (as they are causally related). The relation between X + Y and Z is left unspecified. Therefore, I think Menary’s view is best understood in terms of functionalist property-based constitution: the manipulations of the notebook and the neural process in the right kind of causal organization together constitute the process of remembering because they play a certain functional role, i.e., they enable the agent to succeed in a cognitive task (Menary, 2006). A diachronic account of constitution in line with the characterization above can be found in Kirchhoff’s publications (Kirchhoff, 2015; Kirchhoff & Kiverstein, 2020) and in Leuridan and Lodewyckx (2020).6 In Kirchhoff’s work it is often not clear what the diachronicity amounts to. Sometimes he seems to hold that constitution is diachronic because the relata are extended in time, i.e., they are processes (Kirchhoff, 2015, pp. 324–325). However, this alone does not suffice to render constitution diachronic, which is apparent already from the fact that also proponents of synchronous constitution (which is the standard view) assume that the relata of constitution are processes (see fn. 2). This is crucial: the mere fact that it is processes that are related by constitution does not render constitution diachronic. Clearly, processes can occur at the same place and time, and can therefore be synchronously related. Kirchhoff does not provide any reason why processes cannot depend on each other in a synchronous way. Indeed, it seems to make perfect sense to say, for example, that the process of baking a cake is constituted by pouring the soy milk into the bowl at a certain time t – and this does not even require that the baking and the pouring occur at the same time scales. Similarly, propagations of  Note that Leuridan and Lodewyckx are not directly concerned with extended cognition.

6

136

B. Krickel

action potentials occur at the same time as openings of ion-channels, and my brain is active at the same time as I am writing this article. All that is required for synchrony is that the one process happens in a sub-region of the spatiotemporal location of the other. Thus, for diachronic constitution to be a real alternative (call it “diachronic constitution proper”) to synchronous view of constitution and to be a diachronic relation rather than a synchronous one, it needs to imply that the relata of constitution occur at different points in time. This, according to most philosophers, is a core feature of causation. Indeed, in a later paper (Kirchhoff, 2017), Kirchhoff seems to argue that constitution is a relation between relata that occur at different points in time  – simply because constitution is a causal relation. He argues that “continuous reciprocal causation is a species of constitution” (p. 866) and he goes on arguing that whenever talk of constitution is invoked in this context it is conceivable that what is referred to as constitutive dependence, in order to sharply distinguish from causal dependence, is really a special kind of causation: the kind of recurrent and reciprocal complexity one finds in brains and in extensive systems. (…) Thus, the idea that certain forms of causation are really forms of constitution revises the distinction between causation and constitution such that simple kinds of linear causation do not qualify as constitution, while nonlinear and reciprocal kinds of causation do. (Kirchhoff, 2017, p. 867)

If, however, constitution is reciprocal causation, it is not clear which of the relata is the constituent and which is the constituee. The view is confronted with the problem that constitution is standardly (by defenders and opponents of extended cognition alike) taken to be an asymmetric relation: If X constitutes Y, Y does not constitute X. If the coupling of Otto and his notebook constitutes a given cognitive process, that cognitive process does not constitute the coupling between Otto and the notebook. In a recent paper, Kirchhoff seems to have revised his account of diachronic constitution once again (Kirchhoff & Kiverstein, 2020). Kirchhoff and Kiverstein argue that constituents are indeed simply a subset of linear causes: A common strategy by which to identify constituents for specific phenomena is by determining what plays the most salient causal role(s) with regards to the constitution of some phenomenon. So the relevant distinction is not between causation and constitution, per se; rather, it is between mere causes and constituent causes (…). (Kirchhoff & Kiverstein, 2020, p. 8)

This, of course, does not tell us which causes are the constituents as Kirchhoff and Kiverstein do not explain what “most salient causes” are. But let us ignore the problems that arise from the details of the accounts. Let us ask whether diachronic constitution proper can successfully be plugged into the Argument from Constitution, i.e., whether it satisfies the criteria of adequacy formulated in Sect. 8.2. The notion of diachronic constitution has the advantage that it is independent of any explicit definition of cognition, and it does not rely on any implicit or explicit assumptions regarding the location of cognition. Thus, diachronic constitution can account for criteria of adequacy no. 2 and 3. However, it is still useless as an approach to constitution in the context of extended cognition. The reason is that it fails criterion 1, i.e., it cannot be used to verify the Co-location Principle.

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

137

The reason why diachronic constitution proper fails criterion 1 is obvious: diachronic constitution is a relation between things that occur at different times (while they may occur in the same spatial location). However, the Co-location Principle implies that co-location of a constituee and its constituents is a matter of space and time. This makes perfect sense: if Otto remembers something between t1 and t2 but he does not use his notebook between t1 and t2, the notebook will not be a constituent of this remembering process. Otto must use his notebook at least at some time interval during a remembering episode for the notebook to potentially be a part of the remembering process. However, based on diachronic constitution proper, if using the notebook is a constituent of the remembering process, necessarily, the using of the notebook does not happen during the remembering - because if they would, they could not be related by diachronic constitution proper. Consequently, if diachronic constitution is plugged into the Argument from Constitution, the argument becomes invalid. Note that my argument against diachronic constitution proper does not rest on the claim that it may not be an adequate account of constitution in general. This argument could be made based on the contrast between causation and constitution that motivated the whole dispute. As explained in the introduction, the main claim of defenders of extended cognition is usually characterized as the claim that extracranial elements are not just causal inputs to cognition but constituents of it. Opponents of extended cognition accept that extracranial elements provide (important) causal inputs to cognition but deny the constitution claim (Rupert, 2004). If constitution is characterized as a specific kind of causation (either “reciprocal” or “most salient”; see above), then the whole dispute seems to be trivialized. One may get the impression that diachronic constitution somehow misses the target by failing to account for what the dispute was originally. Kirchhoff and other defenders of diachronic accounts of constitution may turn the argument around and hold that the original motivation was blurry and, in the end, misguided: there is no substantial difference between causes and constituents of cognition. Even if I find this diagnosis interesting and partially convincing, my main argument against diachronic accounts of constitution in the context of extended cognition is that diachronic constitution cannot be plugged into the Argument from Constitution as it cannot be used to show that cognition is located in the same spacetime region as its constituents. Thus, it cannot be used to draw any inferences about the location of cognition.

8.5 Mechanistic Constitution A further notion of constitution that has been put forward in the literature on extended cognition is mechanistic constitution (Abramova & Slors, 2019; Gallagher, 2018a; Kaplan, 2012; Kirchhoff, 2017; Krickel, 2020). This account has been developed independently of the debate on extended cognition. Rather, mechanistic constitution is a notion taken from philosophy of science, i.e., from the discussion of

138

B. Krickel

the so-called new mechanistic approach to scientific explanation. In this literature, mechanistic constitution is taken to be a relation between a mechanism composed of different entities and activities in a certain organization and a phenomenon-to-be-­ explained. The most prominent account of mechanistic constitution is the so-called mutual manipulability account of constitutive relevance (Craver, 2007a, b). It consists of two conditions: X’s Φ-ing is constitutively relevant for S’s Ψ-ing (if and) only if: (i) Parthood: X’s Φ-ing is a spatiotemporal part of S’s Ψ-ing, (ii) Mutual Manipulability: one can change S’s Ψ-ing by changing X’s Φ-ing, and one can change X’s Φ-ing by changing S’s Ψ-ing. X’s Φ-ing is a behaving entity, such as an ion channel’s opening. S’s Ψ-ing is the phenomenon – a behaviour of a system, such as a neuron’s firing (Craver, 2007b; Krickel, 2018a). Parthood implies that X’s Φ-ing is located within the spatiotemporal region of S’s Ψ-ing. Mutual Manipulability is spelled out in terms of interventionism (Woodward, 2003): the changing of X’s Φ-ing by means of changing S’s Ψ-ing, and vice versa, is brought about by conducting ideal interventions (the details of ideal interventions are not crucial here). The mechanism for the phenomenon (S’s Ψ-ing) is composed of all and only those entities and activities (X’s Φ-ings) that are constitutively relevant for S’s Ψ-ing. The motivation for defining constitutive relevance in terms of mutual manipulability is that by means of bottom-up interventions, i.e., those that change the phenomenon via changing a component, one establishes that the component is indeed a difference-maker for the phenomenon. The top-down intervention, i.e., changing the component via changing the phenomenon, is crucial for excluding background conditions. Background conditions are often difference-makers for the phenomenon without being explanatorily relevant to it. For example, intervening into the heart such that the heart stops will have an influence on the performance in a word-stem completion task (Craver, 2007b, p. 157). Changing the impact that gravity has on the body (e.g., by accelerating the body), will change, for example, a person’s performance in a memorizing task. Still, the heart and gravity are not taken to be components of the mechanisms for word-stem completion or memorizing. According to the mutual manipulability account, background conditions are not components of mechanisms as they are not top-down manipulable. For example, there is no intervention into word-stem completion that will have an impact on the activities of the heart or on gravity. While top-down interventions help to distinguish between components and external background conditions such as gravity, alas, they do not help to exclude internal background conditions, such as the heart’s activity relative to word-stem completion. The statement above is simply false: there surely is an intervention on word-stem completion that changes the activity of the heart (Craver, 2007b, p. 158). For example, if the words to be completed are emotional words that are highly arousing, this could lead to an increase of the heartbeat frequency. Thus, at least some background conditions turn out to satisfy the mutual manipulability criterion.

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

139

In order to solve this problem, some authors have suggested to add a further condition to the mutual manipulability account: causal specificity (Hewitson et al., 2018). This notion is, for example, developed in Woodward (2010). Applied to mechanistic constitution, the idea is that in contrast to interventions on background conditions, interventions into mechanistic components have rather specific effects on the phenomenon and will not simply function as on/off switches. Furthermore, changes in a component influence only the phenomenon or at least not a vast number of other phenomena that are not directly connected to the phenomenon at issue. For example, in the investigation of word-stem completion factors that will influence all cognitive mechanisms are not specific to word-stem completion and are, thus, not components in the mechanism for word-stem completion. Thus, by adding causal specificity, background conditions such as the heart’s activity in word-stem completion tasks are excluded. Note that in the present context, the idea of causal specificity should be described as “difference-making specificity” – I will use this term in what follows. The reason is that mechanistic constitution is taken to be a non-causal difference-making relation (Craver, 2007b, p.  153). However, as will become clear later, this application of causal notions to mechanistic constitution is indeed problematic. X’s Φ-ing is constitutively relevant for S’s Ψ-ing if and only if: (i) Parthood: X’s Φ-ing is a spatiotemporal part of S’s Ψ-ing, (ii) Mutual Manipulability: one can change S’s Ψ-ing by changing X’s Φ-ing, and one can change X’s Φ-ing by changing S’s Ψ-ing, (iii) Difference-making Specificity: the interventions described in (ii) lead to specific changes rather than to on/off changes or changes in a vast number of phenomena. Modified in this way, does mechanistic constitution account for our criteria of adequacy? Before we can address this question, one problem that is independent of the issue of extended cognition needs to be addressed. According to many authors, the mutual manipulability account is self-contradictory (Baumgartner & Casini, 2017; Baumgartner & Gebharter, 2016; Kästner, 2017; Krickel, 2018b; Leuridan, 2012; Leuridan & Lodewyckx, 2020; Romero, 2015). Roughly, the contradiction arises because the account rests on the following three assumptions that cannot all be true at the same time: 1. Mechanistic constitution is a non-causal dependence relation. 2. Mechanistic constitution can be spelled out in terms of mutual manipulability. 3. Mutual manipulability consists in intervening into the phenomenon with respect to one of the components that changes the component (top-down intervention), and in intervening into a component with respect to the phenomenon that changes the phenomenon (bottom-up intervention). These three claims are inconsistent because if constitution is a non-causal relation (claim 1) and consists of mutual manipulability (claim 2), then it cannot be the case that what is mutually manipulable is the phenomenon as a whole and each component (claim 3) as this (according to interventionism) would imply that they

140

B. Krickel

are causally related which would contradict the first claim. At least one of the claims must be rejected to render the mutual manipulability account consistent. Different suggestions have been made for how to react to this inconsistency. One strategy is to reject claim (1) and argue that mechanistic constitution is a causal relation. As already indicated in the previous section, this strategy has been defended by Leuridan (2012) and Leuridan and Lodewyckx (2020). However, as this strategy is a version of defending diachronic causation, it suffers from the same problems when applied to the Argument from Constitution as Kirchhoff’s account of diachronic constitution. Others have rejected claim (2) by rejecting the idea that mechanistic constitution relies on mutual manipulability (Baumgartner et  al., 2020; Baumgartner & Casini, 2017; Baumgartner & Gebharter, 2016; Couch, 2011; Harbecke, 2010, 2015; Romero, 2015). This, however, denies the core of the mutual manipulability account, and thus, indeed amounts to a rejection of it. Furthermore, each alternative account comes with its own problems.7 Therefore, I will focus on the third strategy here, i.e., rejecting that mutual manipulability consists in interventions into or with respect to the phenomenon as a whole. This strategy has been adopted by Baetu (2012), Harinen (2018), Krickel (2018b), and most recently by Prychitko (2021) and by Craver et al. (2021). There is one important difference between these different versions of the third strategy:  Among accounts of mechanistic constitution that reject mutual manipulability, one can distinguish between accounts that are still within the interventionist framework and those that reject interventionism as an approach to mechanistic constitution. Among the first are fat-handedness approaches (Baumgartner et al., 2020; Baumgartner & Casini, 2017; Baumgartner & Gebharter, 2016; Romero, 2015) that spell out mechanistic constitution in terms of interventions that target variables at the top as well as the bottom level at the same time (the interventions are performed with a ‘fat hand’). Based on these fat-handed interventions it is either abductively inferred that an observed change is brought about by an underlying constitutive relationship (Baumgartner & Casini, 2017; Baumgartner & Gebharter, 2016; Romero, 2015); or substantial further on the relevant interventions are made (they also have to be “horizontally surgical”, (Baumgartner et  al., 2020)) that allow for a distinction between fat-handed interventions that target constitutively related variables and those that accidentally target a higher- and a lower-level variable at the same time. A critical discussion of the former types of accounts can be found in Krickel (2018b). The horizontal surgicality account is not applicable in the present context as the account requires that interventions lead to simultaneous changes in the phenomenon and the mechanistic component. However, as phenomena are behaviors of systems/organisms, changes to, e.g., the brain, will not instantly lead to a change in behavior as this change needs to be transmitted to, e.g., the muscles to affect behavior. Approaches to mechanistic constitution that do not rely on interventions can be found in Couch (2011) and Harbecke (2015). These approaches defend a regularity-based notion of mechanistic constitution along the lines of Mackie’s analysis of causation in terms of INUS-­ conditions Mackie (1974). According to these accounts, roughly, mechanisms are sets of components that are non-necessary but sufficient for the occurrents of a phenomenon, where each component/member of the set is a necessary member of the set that alone is insufficient for the phenomenon to arise. The problem of these accounts is that they take mechanisms to be sufficient for phenomena. Thereby, they render background conditions mechanistic components. Surely, to be able to perform word-stem completion one’s heart needs to be active in the right way. A functioning word-stem completion mechanism in a dead body will not give rise to word-stem completion behavior. Still, scientists do not take the heart’s activity to be a component in the mechanism for word-stem completion. 7

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

141

most accounts in this ballpark think of top-down interventions as interventions into the input of a mechanism with respect to a component, and of bottom-up interventions in terms of interventions into a component with respect to the mechanism’s output (Baetu, 2012; Craver et al., 2021; Harinen, 2018; Prychitko, 2021). Krickel (2018b), in contrast, presupposes a process-interpretation of mechanistic phenomena (or what I call “entity-involving occurrents” or “EIOs”) and characterizes top-­ down interventions as interventions into time-slices of the phenomenon (what I call “temporal EIO-parts” (Krickel, 2018b, p.  55)) with respect to a component, and bottom-up interventions as interventions into components with respect to a later time-slice of the phenomenon. My account has the advantage that it saves the idea that top-down and bottom-up interventions cross mechanistic levels. However, my account is in trouble when it comes to identifying the very first and the very last component of a phenomenon (Krickel, 2018b, p. 66). Regarding these components no bottom-up or top-down interventions exist. One might solve this problem by combining the two types of accounts. Here, I do not want to discuss this suggestion in detail but leave it for future research. For the purposes of this paper, I will assume that some version of the third strategy is successful in solving the contradiction depicted above. Mechanistic constitution has one clear advantages over diachronic constitution: it validates the Co-location Principle. Given that the first condition, Parthood, requires that the constituent is a spatiotemporal part of the phenomenon, it follows that the phenomenon is located where its constituents are located. However, some have argued that Parthood is the reason why mechanistic constitution is useless for the debate on extended cognition (Baumgartner & Wilutzky, 2017). The reason is that it seems to violate our third criterion of adequacy: it presupposes assumptions about the location of cognition. If for X to be a constituent of Y in the mechanistic sense, X must be a part of Y, we thereby already presuppose that X and Y are co-­ located. Thus, we already presuppose that cognition is located wherever X is located, and then determine whether it is a constituent. Its application to extended cognition, therefore, trivializes the whole endeavour as one presupposes what one needs to show, namely the co-location of cognition and extra-cranial elements (Baumgartner & Wilutzky, 2017). Whether this really is question-begging is not so clear. In (Krickel 2020) I argue that it is not. Roughly, I argue that assuming that some element is a spatiotemporal part of a cognitive phenomenon does not yet clarify the location of cognition. The criteria for the location of cognition are more demanding than mere spatiotemporal inclusion – it requires mutual manipulability and difference-making specificity as well. This apparent stretch succeeds because I think of cognitive phenomena as behaviors of agents. Qua being behaviors, these phenomena will involve all sorts of non-cognitive elements, such as leg movements in navigation behaviors, or arm movements in remembering tasks. The arm and leg movements are spatiotemporal parts of the cognitive phenomenon as they occur in the same space-time region as the behavior. However, they are not cognitive parts. I argue  that  to differentiate between mere spatiotemporal parts and cognitive parts, the mutual manipulability account is indeed insufficient. The mutual manipulability account  is confronted

142

B. Krickel

with what I call the “challenge of trivial extendedness” (Krickel, 2020, p. 549). This challenge arises because leg and arm movements indeed turn out to satisfy the mutually manipulability criterion as well as the difference-making specificity criterion. Thus, on the mutual manipulability account alone, extension of cognition would become trivial as cognitive phenomena would extend the brain qua being behaviors. However, the dispute is not whether behavior extends the brain (clearly it does) but whether cognition extends the brain. Therefore, I argue we need to add a further condition to the mutual manipulability account to render it fruitful for the dispute over extended cognition. I call this additional criterion “behavior unspecificity” (Krickel, 2020, p. 553). It implies that a spatiotemporal part of a cognitive behavior that satisfies the mutual manipulability condition and the difference-making specificity condition is a cognitive component only if it is part of the behavior qua the behavior being a manifestation of a cognitive capacity. The spatiotemporal part is not only a component of the mechanism for the behavior because, say, the experimental setup to investigate spatial navigation was constructed in the way that subjects had to move their legs. Rather, the spatiotemporal part is a component of the mechanism because the mechanism is a manifestation of a cognitive capacity that may manifest in different behaviors. Behavior unspecificity is tested by comparing different mechanisms underlying different operationalizations of the same cognitive capacity. Only those spatiotemporal parts that occur in all these operationalizations are crucial for the cognitive capacity at hand, and thus, only those will count as cognitive components. One potential objection against applying my 2020 account to extended cognition is that this seems to presuppose clarity on the mark of the cognitive. Mechanistic components, according to the account, are cognitive if and only if they are constitutive of a cognitive capacity. But what renders a capacity cognitive? It seems that my account can help us answer the extended cognition question only if some version of property-based constitution is presupposed. Thus, this approach is afflicted with the same problems that property-based accounts of constitution have. This problem, however, can be solved. My 2020 account starts from cases of cognitive capacities, such as memory and problem solving, that are accepted by cognitive scientists as paradigmatic cases of cognition. Since defenders of extended cognition only need to show that there is one actual case in which cognition is extended, I can accept that there are capacities for which it is unclear whether they count as cognitive or not. Thus, while a definition of cognition may, in the end, be necessary to distinguish cognitive from non-cognitive capacities, the dispute amongst defenders and opponents of extended cognition is independent of this clarification – as long as it can be shown that there is one paradigmatic cognitive capacity that is partially constituted by extracranial elements. My 2020 account, still, is not unproblematic. Independently of the issue of extended cognition, my account is as an account of the constitution of capacities. While the original mutual manipulability account specifies which elements are constitutive of a behavior, my account specifies the conditions under which an element is constitutive of a capacity. However, the account remains incomplete as I do not explain what I take capacities to be and how they are to be individuated. This,

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

143

however, is crucial for the account and the evaluation of the notion of behavior unspecificity. Remember that my idea was that a mechanistic component is behavior unspecific, i.e., a cognitive component, only if it occurs in all operationalizations of the relevant cognitive capacity. Thus, whether hand gestures (this is the examples I use to illustrate my account) turn out to be cognitive components in spatial problem solving depends on whether the tasks in which hand gestures occur and those in which they do not occur are manifestations of the same capacity, or not. And if they turn out to be manifestations of the same capacity, the question is whether this implies that the capacity is multiple realized, or whether this shows that hand gestures are indeed redundant as they do not occur in all manifestations of the capacity. The individuation-question crucially depends on what capacities are taken to be. On a prominent account (Cummins, 1975, 1983), capacities are functional kinds. They are individuated by their functional roles or, as Cummins phrases it, in terms of “input-output conditions” (Cummins, 1983, p. 53). As functional kinds, capacities are in principle multiple realizable. Problem solving that is done purely in the head and problem solving that relies on hand gestures may turn out to be different realizations of the same cognitive capacity. However, others have argued that cognitive capacities are to be individuated in a more fine-grained manner in terms of more specific causal features that scientists identify (Rupert, 2004). For example, problem solving that is done purely in the head will have specific causal features that cases in which hands are used do not have. For example, the former will be influenced by changes in cognitive load, whereas the latter will be influenced by whether the light is turned on or off. On this characterization, the two manifestations of problem solving turn out to be manifestations of different capacities as they have different functional roles. Interestingly, despite the unclarity of how capacities are to be individuated, the outcome may speak in favor of my 2020 account and my claim that it verifies extended cognition. On a coarse-grained individuation of the functional role of a given capacity, problem solving that is done purely in the head and problem solving that relies on hand gestures turn out to be different realizers of the same capacity. Surely, my account needs to be modified to allow for multiple realization. That is, we should not require that a component occurs in all manifestations of a capacity, but only in all manifestations of one and the same realizer of that capacity. If problem solving that relies on hand gestures indeed turns out to be a realizer of the capacity to solve problems, then extended cognition is verified at least in so far as there are extracranical cognitive components in the body. On a fine-grained individuation of the functional roles of capacities, problem solving that is done purely in the head and problem solving that relies on hand gestures turn out to be distinct capacities. As long as hand gestures indeed occur in all manifestations with the same causal profile, hand gestures turn out to be constituents of the respective capacity. However, at this point a problem re-appears. On the fine-grained individuation of the functional roles of capacities, internally realized capacities and partially externally realized capacities will turn out to be different capacities due to their different causal profiles. Surely, the internally realized capacities correspond to the paradigmatic cases of cognitive capacities that are investigated by cognitive scientists. But

144

B. Krickel

what about the partially externally realized capacities? Are they cognitive capacities given that it is exactly their cognitive status that is questioned by many philosophers and cognitive scientists? Similarly, on the coarse-grained individuation of the functional roles of capacities, one might ask whether problem solving that involves hand gestures is indeed a cognitive realization of problem solving. One could simply hold that problem solving is a capacity that can be realized by cognitive mechanisms and by non-cognitive mechanisms. Indeed, we seem to be back at the beginning: we need to know what the mark of the cognitive is to argue whether partially externally realized capacities are indeed cognitive capacities. Thus, in the end, the application of my 2020 account to settle the dispute between opponents and defenders of extended cognition may fail for the same reason as property-based constitution by presupposing a potentially question-begging notion of cognition and thereby violating criterion of adequacy no. 2.

8.6 Conclusion The debate on extended cognition is a debate about the location of cognition. According to a recent trend, extended cognition is framed in terms of the ‘search for the mark of constitution’. In this paper, I specified a list of criteria of adequacy that an account of constitution needs to satisfy to be applicable to the dispute on extended cognition: 1. The account of constitution must validate the Co-location Principle. 2. The account of constitution must be acceptable by defenders as well as opponents of extended cognition. 3. The account of constitution must not presuppose assumptions about the location of cognition. I discussed three accounts of constitution that have been used in the literature on extended cognition: property-based constitution, diachronic constitution, and mechanistic constitution. None of these accounts can be used to defend extended cognition. Property-based accounts fail regarding criterion (2). Diachronic constitution fails criterion (1). And while mechanistic constitution can be defended against the objection that it fails criterion (3) it seems to presuppose a property-based account and thus fails criterion (2). How to deal with this situation? Going back to discussing the mark of the cognitive does not seem to be promising as it has already been proven to be a dead end. Defenders and opponents of extended cognition seem to have to agree to disagree. In my view, the most promising strategy would be to take a pragmatic turn and ask why it is relevant to determine the location of cognition. This strategy would accept that there is no realistic interpretation of the location of cognition and that we may have to accept a pluralistic notion of cognition. Still, it may provide us with clear criteria for identifying the location of cognition relative to the respective context in which localizing cognition is relevant. Working out this proposal, however, must be the objective of future research.

8  Extended Cognition and the Search for the Mark of Constitution – A Promising…

145

References Abramova, E., & Slors, M. (2019). Mechanistic explanation for enactive sociality. Phenomenology and the Cognitive Sciences, 18(2), 401–424. Adams, F., & Aizawa, K. (2001). The bounds of cognition. Philosophical Psychology, 14(1), 43–64. Adams, F., & Aizawa, K. (2008). The bounds of cognition. Blackwell Publishing Ltd. Aizawa, K., & Adams, F. (2005). Defending non-derived content. Philosophical Psychology, 18(6), 661–669. Baetu, T. M. (2012). Filling in the mechanistic details: Two-variable experiments as tests for constitutive relevance. European Journal for Philosophy of Science, 2(3), 337–353. Baumgartner, M., & Casini, L. (2017). An abductive theory of constitution. Philosophy of Science, 84(2), 214–233. Baumgartner, M., & Gebharter, A. (2016). Constitutive relevance, mutual manipulability, and fat-­ handedness. The British Journal for the Philosophy of Science, 67(3), 731–756. Baumgartner, M., & Wilutzky, W. (2017). Is it possible to experimentally determine the extension of cognition? Philosophical Psychology, 30(8), 1104–1125. Baumgartner, M., Casini, L., & Krickel, B. (2020). Horizontal surgicality and mechanistic constitution. Erkenntnis, 85(2), 417–430. Clark, A. (2008). Supersizing the mind  - embodiment, action, and cognitive extension. Oxford University Press. Clark, A. (2014). Mindware  – An introduction to the philosophy of cognitive science. Oxford University Press. Clark, A., & Chalmers, D. J. (1998). The extended mind. Analysis, 58(1), 7–19. Couch, M. B. (2011). Mechanisms and constitutive relevance. Synthese, 183(3), 375–388. Craver, C. (2007a). Constitutive explanatory relevance. Journal of Philosophical Research, 32(Section II), 3–20. Craver, C. F. (2007b). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. Oxford University Press. Craver, C. F., Glennan, S., & Povich, M. (2021). Constitutive relevance & mutual manipulability revisited. Synthese, 199(3–4), 8807–8828. Cummins, R. (1975). Functional analysis. The Journal of Philosophy, 72(20), 741–765. Cummins, R. (1983). The nature of psychological explanation. MIT Press. Gallagher, S. (2018a). New mechanisms and the enactivist concept of constitution. In M. P. Guta (Ed.), Consciousness and the ontology of properties (pp. 207–220). Routledge. Gallagher, S. (2018b). The extended mind: State of the question. The Southern Journal of Philosophy, 56(4), 421–447. Harbecke, J. (2010). Mechanistic constitution in neurobiological explanations. International Studies in the Philosophy of Science, 24(3), 267–285. Harbecke, J. (2015). Regularity constitution and the location of mechanistic levels. Foundations of Science, 20(3), 323–338. Harinen, T. (2018). Mutual manipulability and causal inbetweenness. Synthese, 195(1), 35–54. Hewitson, C. L., Kaplan, D. M., & Sutton, J. (2018). Yesterday the earwig, today man, tomorrow the earwig? Comparative Cognition & Behavior Reviews, 13, 25–30. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology and Philosophy, 27(4), 545–570. Kästner, L. (2017). Philosophy of cognitive neuroscience: Causal explanations, mechanisms and empirical manipulations. De Gruyter. Kirchhoff, M.  D. (2015). Extended cognition & the causal-constitutive fallacy: In search for a diachronic and dynamical conception of constitution. Philosophy and Phenomenological Research, 90(2), 320–360. Kirchhoff, M. D. (2017). From mutual manipulation to cognitive extension: Challenges and implications. Phenomenology and the Cognitive Sciences, 16(5), 863–878.

146

B. Krickel

Kirchhoff, M. D., & Kiverstein, J. (2020). Attuning to the world: The diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11, 1966. Krickel, B. (2018a). The mechanical world (Vol. 13). Springer International Publishing. Krickel, B. (2018b). Saving the mutual manipulability account of constitutive relevance. Studies in History and Philosophy of Science Part A, 68, 58–67. Krickel, B. (2020). Extended cognition, the new mechanists’ mutual manipulability criterion, and the challenge of trivial extendedness. Mind & Language, 35(4), 539–561. Leuridan, B. (2012). Three problems for the mutual manipulability account of constitutive relevance in mechanisms. British Journal for the Philosophy of Science, 63(2), 399–427. Leuridan, B., & Lodewyckx, T. (2020). Diachronic causal constitutive relations. Synthese, 198, 9035–9065. Mackie, J. L. (1974). The cement of the universe. Clarendon Press. Menary, R. (2006). Attacking the bounds of cognition. Philosophical Psychology, 19(3), 329–344. Menary, R. (2007). Cognitive integration - mind and cognition unbounded. Palgrave Macmillan. Menary, R. (2010). The holy grail of cognitivism: A response to Adams and Aizawa. Phenomenology and the Cognitive Sciences, 9(4), 605–618. Prychitko, E. (2021). The causal situationist account of constitutive relevance. Synthese, 198(2), 1829–1843. Romero, F. (2015). Why there isn’t inter-level causation in mechanisms. Synthese, 192(11), 3731–3755. Rupert, R.  D. (2004). Challenges to the hypothesis of extended cognition. The Journal of Philosophy, 101(8), 389–428. Rupert, R.  D. (2010). Systems, functions, and intrinsic natures: On Adams and Aizawa’s “The Bounds of Cognition”. Philosophical Psychology, 23(1), 113–123. Wasserman, R. (2018). Material constitution. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Fall 2018). Metaphysics Research Lab, Stanford University. Wieland, J. W. (2014). Internalism does entail scepticism. In E. Weber, D. Wouters, & J. Meheus (Eds.), Logic, reasoning, and rationality (pp. 247–260). Springer. Woodward, J. (2003). Making things happen: A theory of causal explanation. Oxford University Press. Woodward, J. (2010). Causation in biology: Stability, specificity, and the choice of levels of explanation. Biology and Philosophy, 25(3), 287–318.

Chapter 9

Commentary on “Extended Cognition and the Search for the Mark of Constitution – A Promising Strategy?” Julian Kiverstein and Michael Kirchhoff

Abstract  The discussion of extended cognition is premised on a metaphysical distinction between causation and constitution. For example, Rowlands (2009) notes that “EM [extended mind] is a claim about the composition or constitution of (some) mental processes” (2009, p. 54). Or, as Wheeler puts it: “Bare causal dependency of mentality on external factors […] is simply not enough for genuine cognitive extension. What is needed is constitutive dependence” (2010, p.  246). In this sense, Krickel (this volume) rightly notes that the current disagreement between friends and foes of extended cognition turns on a metaphysical dispute about what is constitutive of mentality and how to locate those constituents. Krickel (this volume) argues that our account of diachronic constitution fails to provide a convincing answer to the metaphysical dispute about ‘what is constitutive’ of mentality and ‘how to locate’ the relevant constituents. The central reason is that diachronic constitution fails the so-called ‘co-location condition,’ which Krickel argues is a necessary condition that an account of constitution must meet in the discussion over the extended mind. The co-location condition asserts that ‘cognition is located where and only where its constituents are located.’ We expose an ambiguity in the co-­ location condition, and show why this condition is no threat to our account of diachronic constitution. We conclude that the co-location condition is not a necessary condition that an account of constitution in the discussion over the extended mind must adhere to.  The discussion of extended cognition is premised on a metaphysical distinction between causation and constitution. For example, Rowlands (2009) notes that “EM [extended mind] is a claim about the composition or constitution of (some) mental J. Kiverstein (*) Department of Psychiatry, Amsterdam University Medical Research, Amsterdam, The Netherlands e-mail: [email protected] M. Kirchhoff School of Liberal Arts, University of Wollongong, Wollongong, NSW, Australia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_9

147

148

J. Kiverstein and M. Kirchhoff

processes” (2009, p. 54). Or, as Wheeler puts it, “[b]are causal dependency of mentality on external factors […] is simply not enough for genuine cognitive extension. What is needed is constitutive dependence” (2010, p. 246). In this sense, Krickel (this volume) rightly notes that the current disagreement between friends and foes of extended cognition turns on a metaphysical dispute about what is constitutive of mentality and how to locate those constituents. To this end, Krickel identifies three conditions some entity (event, process, object or property) must satisfy to qualify as a bona fide constituent of cognition. They are: 1. A co-location condition: Cognition is located where and only where its constituents are located; 2. An agreeability condition: Whatever account of constitution one advances, it must in principle be acceptable by both parties in the discussion; and 3. A location-neutral condition: The account of constitution must not presuppose whether the constituents of cognition are internal (e.g., neural vehicles) or external (e.g., including, necessarily, non-neural bodily and worldly entities). The first condition may be taken to be uncontroversial on pain of standard bloat objections. However, a closer look reveals an ambiguity concerning how to read the notion of location. This can be read in a spatial sense as concerning what elements fall spatially within, or outside of, the boundaries of a cognitive system. Crucially, however (1) can also be read in a temporal sense as requiring that a relation of co-­ temporality hold between constituting relata. Relata are cotemporally located if for any point (or interval) in time the relata can be clearly and distinctly individuated as located at the same spatiotemporal regions. Our account of diachronic constitution denies that constitution requires the co-temporal location of constituting relata. We will say more about our reasons for doing so below. Thus, we reject the co-location condition, if the location of constituents is understood as entailing the co-temporal location of constituents. The co-location condition should be interpreted spatially, not temporally, if it is to avoid begging the question against diachronic views of constitution. The other two conditions should be taken as straightforwardly uncontroversial; a denial of either principle is simply to beg the question against the opponent’s view. Krickel targets three accounts of constitution: property constitution, diachronic constitution and the mechanistic view of constitution. All three accounts are argued to violate  one or more of the three principles listed above. Hence, none of them work as a proper account of constitution in the discussion over the extended cognition thesis. In the rest of this commentary, we focus on Krickel’s objections to our work in developing the diachronic account of constitution (in, e.g., Kirchhoff (2015), Kirchhoff (2017), Kirchhoff & Kiverstein (2020)). Here are the issues we shall confront: (A). There is unclarity about how to understand what is diachronic about diachronic constitution; (B). No justification is given for the claim that processes cannot depend on one another in a synchronic way;

9  Commentary on “Extended Cognition and the Search for the Mark of Constitution …

149

(C). Diachronic constitution violates the standard view that constitution is an asymmetric relation of dependence; and (D). Diachronic constitution violates the co-location condition. We start by clarifying (A) and (B). In doing so, we shall argue that the consensus that one cannot dispense with the view that constitution is asymmetric need not be accepted. So, we deny (C). Regarding (D), we shall argue that even if diachronic constitution violates the co-location principle, this is no problem for diachronic constitution for discussions over extended cognition. Hence, we shall show that the co-location condition is not a necessary condition on constitutive accounts of cognition.

9.1 What Is Diachronic Constitution? Krickel points out that what is ‘diachronic’ about diachronic constitution has been stated differently in different papers over the years. Here, then, is the official statement: Metaphysical orthodoxy has it that constitution relates entities (events, or processes) synchronically, whereas causation is a diachronic relation between things (or events, processes, etc.) over time. We argue that the metaphysical orthodoxy is mistaken. Some constitutive relations are special cases of causal relations; specifically, continuous and reciprocal causal relations. We are not the only ones making this proposal. It has been put forth by Leuridan and Lodewyckx (2020) as well as in Ingthorsson (2020). In our work we have provided different examples of how constitution can be diachronic that don’t work in exactly the same way. The first type of case is one in which X is diachronically constituted because X occurs simultaneously and continuously over time with its constituents but where neither X nor its constituents are wholly present at an instant in time. The second type of case is one in which some of the constituents of X occur at different times from the constituted. At this stage, Krickel may well respond that when we define diachronic constitution in opposition to synchronic constitution, we are helping ourselves to a problematic notion of ‘synchronic’. From our perspective, if X and Y stand in a synchronic relation to one another they do so because X and Y are instantiated atemporally. However, it is also possible to articulate ‘synchronicity’ in terms of X and Y being co-temporal as discussed above (Krickel, personal communication). We note here that many metaphysical philosophers have understood constitution in atemporal terms. Horgan, for instance, claims: “In philosophical contexts, it [constitution, supervenience …] is primarily used non-temporally to signify a metaphysical and/ or conceptual determination-relation” (1993, p.  555). Gillett stipulates that: “Compositional relations are non-causal determination relations that are synchronous” (2013, p. 9). Diachronic constitution questions this basic assumption about the non-temporality of constitutive relations.

150

J. Kiverstein and M. Kirchhoff

Causation is often distinguished from constitution on the grounds that the constituents or relata of a causal relation have a wholly distinct existence, whereas this is not the case for the constituents of a constitution relation. Constitution holds between relata that spatiotemporally and materially co-exist. However, in systems that exhibit continuous, reciprocal causation, the relata in question need not be wholly distinct given their mutual co-dependence in the here-and-now and over time. This point is also echoed in Ingthorsson (2020), arguing that an entirely unidirectional view of causation (where one object is acting, and another acted upon) finds no home in the empirical sciences. A simple non-cognitive example of this is the relation between the ensemble dynamics of convection rolls and the dynamics of the molecules making up the larger, ensemble behavior. It is because of the mutual co-determination of parts and wholes in cases of circular causality that we make use of the notion of circular causality to analyse diachronic constitution. Finally, and this addresses (C) above, it is widely accepted that constitution is a dependence relation that holds asymmetrically. The constituted, X, depends on its parts, Y; not the other way around. This assumption has been challenged. Barnes (2018) questions this assumption in making the case for what she calls ‘symmetric dependence’. One example she provides is the relation between WWII and the evacuation of Dunkirk. It seems plausible that WWII would not have been the same event without the evacuation of Dunkirk. It would have been a very different war without the evacuation. Similarly, the evacuation of Dunkirk would not have been what it was unless it were part of WWII. That is, what it is to be the evacuation of Dunkirk is to be part of WWII.  Of course, this will ultimately depend on one’s account of the individuation of events. However, at first glance, it seems a strong case can be made that the part-whole constitution relation for events is an example of symmetric dependence. We think the same is true for the convection roll example: the dynamics of the constituents get to be what they are (in terms of their configuration and degrees of freedom) in virtue of the whole they are partaking in; and vice-versa. So, if Barnes is correct, then ontological dependence relations such as constitution need not, always and necessarily, hold asymmetrically. The violation of asymmetric dependence is not a problem for our account of diachronic constitution. This provides some initial clarification of what we mean by ‘diachronic constitution’ in response to Krickel’s first criticism. It also allows us to address one of Krickel’s more specific worries: that “it seems to make perfect sense, for example, to say that the process of baking a cake at t1 is constituted by pouring the soy milk into the bowl at t1 – and this does not even require that the baking and the pouring occur at the same time scales.” From our perspective, talk of events unfolding at times t1 is an abstraction. It is a convenient way to understand what is happening in the world; yet, it is just that – an epistemic convenience. From the perspective of diachronic constitution, events or processes cannot manifest themselves at an instant in time. Nor can they necessarily be individuated at spacetime regions, intervals or phases of times except as abstractions. What we would want to say, is that two events can co-evolve; that two events can unfold simultaneously and over different time-scales. Our assessment of the example provided by Krickel is that

9  Commentary on “Extended Cognition and the Search for the Mark of Constitution …

151

it signifies a slippage from talk of simultaneity to talk of synchronicity. From the perspective of diachronic constitution, two events can unfold simultaneously without unfolding synchronically. To illustrate this idea, consider the non-­ cognitive case of convection rolls: the ensemble dynamics unfold simultaneously and continuously with the dynamics of the individual molecules, the latter unfolding over faster timescales than the former. In this sense, we agree with Krickel that two events can unfold simultaneously. However, it does not follow that the events in question co-occur synchronically. The relata of a constituting relation, in the case of biological and cognitive systems, do not do their constituting work synchronically or cotemporally. So far we have responded to (A)–(C). The crux of Krickel’s argument against the use of diachronic constitution in the debate about extended cognition relates to (D); the violation of the co-location principle, to which we now, more fully, turn our attention.

9.2 Diachronic Constitution and the Co-location Principle Krickel’s argument can be reconstructed as an apparent dilemma for diachronic constitution. On the one hand, the co-location principle states that ‘cognition is located where and only where its constituents are located.’ On the other hand, according to Krickel, for our account of constitution to be a diachronic rather than a synchronic account, ‘it needs to imply that the relata of constitution occurs at different points in time.’ According to Krickel, a proper diachronic account of constitution must violate the co-location principle if co-location is taken to mean co-temporal location. However, the violation of the co-location principle renders diachronic constitution useless in debates concerning the constitution of cognitive systems. So, it looks like we have been pushed into a dead-end. Setting aside the sense in which this argument trades on an ambiguity in the use of the term “location” as discussed earlier, we will argue that the co-location condition does not create a genuine dilemma for diachronic constitution. Consider initially that diachronic constitution is not the claim that the relata of constitution must occur at different points in time. Diachronic constitution is the view that the relata of constitution unfold over different timescales: this is the case with the convection roll example and with the case of emotion co-regulation which we discuss in our chapter contribution to this edited collection. Suppose we instead understand the relata of a diachronic constitution relation as located at different spacetime regions. Can events or entities be located at different spacetime regions and also satisfy the co-location condition? It would appear not. However, we have denied that the events or processes that are the constituents of cognitive and biological systems can be individuated as located at distinct spacetime regions other than, as a matter of epistemic convenience, as abstractions. The relation of mutual co-­ determination between parts and wholes in cases of diachronic constitution implies what we have called a symmetric relation of dependence.

152

J. Kiverstein and M. Kirchhoff

Take the much-discussed example of Otto’s notebook as an example. (Our thanks to Krickel for discussion of this example.) One can point to the notebook as occupying a spacetime region in Otto’s pocket for instance. Doesn’t this contradict what we have just claimed about entities located at discrete spacetime regions being abstractions? We would certainly agree that one can individuate the notebook as a self-­ contained object. However, the constituents of a relation of diachronic constitution are not objects but events and processes unfolding over different time scales. The notebook forms a part of Otto’s mind when it closes a looping cycle of perception and action exhibiting precisely the kind of reciprocal causality that we see in the convection roll example. The notebook can only be taken to make a well-defined contribution to the larger perception-action cycle as an abstraction. The degrees of freedom of the notebook and its exact configuration within the cognitive system Otto forms with his notebook is determined by the larger perception-action cycle of which the notebook is a part. Conversely, the perception-action cycle unfolds in the way it does in part because Otto is coupled with the notebook. The two unfolding processes evolve together simultaneously and continuously and it is only because of this relation of mutual co-determination that the notebook can be said to diachronically constitute a part of Otto’s mind. One might now object: if diachronic constitution holds between relata that unfold simultaneously and continuously, in which sense is this diachronic? The straightforward answer is that the relata are temporal, and continuously unfolding. At no point in time are the relata instantiated at a single (snapshot) moment in time or even at an interval or spacetime region. We will finish up by noting that we need not commit ourselves to the claim that all cases of diachronic constitution are cases in which causation is simultaneous and continuous. Ultimately this boils down to the phenomenon to be explained. In Kirchhoff and Kiverstein (2020) we made the observation that ultimate causes are usually treated as wholly distinct from proximate causes. For example, we can ask why birds migrate, and we can ask how they migrate. The former question can be answered by reference to the evolutionary and developmental history of birds. The latter how-question is answered by reference to muscle mass, morphology, and so on, at a specific point in time. We take diachronic constitution to imply that the choice between ultimate and proximate explanations is sometimes a false choice. We gave the example of Eva’s culture shock, arising from the misalignment of expectations that were formative of Eva’s experiences due to her growing up in Poland and those that are operative in her new home of Vancouver. The expectations that have historically come to shape her perception and action (an example of ultimate explanation) are out of step with those of her surroundings. To explain her experience of culture shock proximally we have to appeal to an ultimate cause. We suggest Eva’s experience of cultural shock is constituted by her coupling with the cultural environment and her failure of her expectations to align with her new cultural environment. Does the misalignment with her cultural environment that is constitutive of her experience satisfy the co-location condition? Insofar it requires making reference to ultimate causes – expectations she formed growing up

9  Commentary on “Extended Cognition and the Search for the Mark of Constitution …

153

in Poland  – it would seem not to satisfy the co-location condition. The ultimate cause is neither co-temporally nor spatially located where Eva is currently spatially and temporally located. As an ultimate cause, her developmental history is in the past, and concerns a time when she lived in Poland. Krickel will likely respond that the violation of the co-location conditions shows that culture shock is not an example that can be used to argue for the extended mind. Unsurprisingly, we draw a different conclusion. The co-location condition doesn’t carve the cake in the right way to settle what is or is not constitutive of the mind. The appearance that it is a necessary condition stems from the ambiguity in the term ‘location’. We agree that cognition is located only where its constituents are located if location is understood spatially and materially. However, we have been arguing that the co-location condition fails when location is understood as requiring the relation of co-temporality between constituents. It fails because the very idea of co-temporality requires that one locate constituents at discrete spacetime regions and we have been arguing that in biological and cognitive cases, neither the constituted phenomenon nor its constituents are wholly present at an instant in time.

References Barnes, E. (2018). Symmetric dependence. In R. Bliss & G. Priest (Eds.), Reality and its structure: Essays in fundamentality (pp. 50–69). Oxford University Press. Gillett, C. (2013). Constitution, and multiple constitution, in the sciences: Using the neuron to construct a starting framework. Minds & Machines, 23, 309–337. Horgan, T. (1993). From supervenience to superdupervenience: Meeting the demands of a material world. Mind, 102(408), 555–586. Huemer, M., & Kovitz, B. (2003). Causation as simultaneous and continuous. The Philosophical Quarterly, 53(213), 556–565. Ingthorsson, R. (2020). A causal theory of constitution and persistence. In M. Szlatkowski (Ed.), E.J. Lowe and Ontology (pp. 290–310). Routledge. Kirchhoff, M.  D. (2015). Extended cognition & the causal-constitutive fallacy: In search for a diachronic and dynamical conception of constitution. Philosophy and Phenomenological Research, 90(2), 320–360. Kirchhoff, M. D. (2017). From mutual manipulation to cognitive extension: Challenges and implications. Phenomenology and the Cognitive Sciences, 16(5), 863–878. https://doi.org/10.1007/ s11097-­016-­9483-­x Kirchhoff, M. D., & Kiverstein, J. (2020). Attuning to the world: The diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11. Kirchhoff, M. D., & Kiverstein, J. (2021). Diachronic constitution. Retreived from: http://philsci-­ archive.pitt.edu/19690/1/Diachronic%20Constitution.pdf. Krickel, B. (this volume). Extended cognition and the search for the mark of constitution  – A promising strategy? Leuridan, B., & Lodewyckx, T. (2020). Diachronic causal constitutive relations. Synthese, 198, 1–31. Rowlands, M. (2009). Enactivism and the extended mind. Topoi, 28, 53–62. Wheeler, M. (2010). In defense of extended functionalism. In R. Menary (Ed.), The extended mind (pp. 245–270). MIT Press.

Chapter 10

Dissolving the Causal-Constitution Fallacy: Diachronic Constitution and the Metaphysics of Extended Cognition Julian Kiverstein and Michael Kirchhoff Abstract  This chapter questions the causal-constitution fallacy raised against the extended mind. It does so by presenting our signature temporal thesis about how to understand constitutive relations in the context of the extended mind, and with respect to dynamical systems, more broadly. We call this thesis diachronic constitution. We will argue that temporalising the constitution relation is not as remarkable (nor problematic) as it might initially seem. It is (arguably) inevitable, given local interactions between microscale and macroscale states of (coupled) dynamical systems. We focus primarily on the metaphysics of the extended mind in this paper. However, we also show how our account of diachronic constitution has important implications for the metaphysics of dependence relations more generally as well as an emerging literature on inter-level explanations in the mechanistic framework applied to the discussion over extended, enactive and embodied cognition. Keywords  Extended mind · Embedded cognition · Causal-constitutive fallacy · Diachronic constitution · Metaphysical dependence · Reciprocal causation · Mutual manipulability

J. Kiverstein (*) Department of Psychiatry, Amsterdam University Medical Research, Amsterdam, The Netherlands e-mail: [email protected] M. Kirchhoff School of Liberal Arts, University of Wollongong, Wollongong, NSW, Australia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_10

155

156

J. Kiverstein and M. Kirchhoff

10.1 Introduction This chapter shows how to dissolve the causal-constitution fallacy raised in response to arguments for the extended mind (EM) (Adams & Aizawa, 2001, 2008; Block, 2005; Rupert, 2009; Aizawa, 2010).1 EM is the thesis that neural, extra-neural bodily and worldly elements jointly constitute some cognitive processes – such as semantic memory or problem solving – and mental states – such as belief or perceptual experiences (Clark & Chalmers, 1998; Hurley, 1998; Noë, 2004; Menary, 2010; Rowlands, 2009; Wheeler, 2010; Kirchhoff & Kiverstein, 2019a). Although EM is a thesis about the constitution of cognition, including action and perception, appeals to causal coupling lie at the heart of most arguments for EM. Consider the following representative quote from Van Gelder and Port (1995): In this vision, the cognitive system is not just the encapsulated brain; rather since the nervous system, body and environment are all constantly changing and simultaneously influencing each other, the true cognitive system is a single unified system embracing all three. (Van Gelder & Port, 1995, p. 373; cited by Adams & Aizawa, 2008, p. 89)

Haugeland (1998) describes how the ant in making its way across a sandy beach forms a coupled system with the beach. The ant’s movements and the surface of the beach reciprocally influence one another so as to form what Haugeland describes as “an integrated unit”, not “a pair of distinct components” (1998, p. 217). Along similar lines, Anderson et al. (2012) give examples of cognitive systems that are softly assembled on the fly, out of heterogeneous neural, bodily, and worldly elements. Like the ant and the beach these elements alter each other’s dynamics in such a way as to make it difficult to assign individual roles to each element. Instead, neural, bodily and environmental elements work together to form what Anderson and colleagues describe as “synergies” – functional groupings temporarily held together to act as a single coherent unit (2012, p. 719). The best arguments for EM are in our view premised on explanations like these of the behaviour of an agent that make appeal to the two-way, reciprocal causal coupling of agent and world (cf. Chemero, 2009; Palermos, 2014). This is also the case for so-called first-wave accounts of EM that stress the role of functional parity between internal and external states (Clark & Chalmers, 1998). The causal-constitution (CC) fallacy is the mistake of inferring from the claim that ‘X and Y are causally related’, like in the coupling examples above, the conclusion that ‘X and Y are constitutively related’. Such an inference is argued to be invalid in part because causation and constitution are taken to be distinct

 Note that this attempt to dissolve the causal-constitution fallacy is relevant for providing positive arguments for the entire field of so-called 4E (extended, enactive, embodied and ecological) philosophy of cognitive science. All these frameworks emphasise the constitutive role of the body and environment, in addition to the brain, in the formation and constitution of cognitive processes and mental states. Hence, getting clear about how to understand metaphysical talk of constitution will help researchers in this area to articulate their positions in a way that does not fall prey to the usual kinds of metaphysical worries about 4E claims in philosophy of cognitive science. 1

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

157

metaphysical relations of dependence, and in part because there are counterexamples that show that two things can be causally related while not being constitutively related – e.g., the relation between sunburn and solar radiation (Rowlands, 2009). Opponents of EM insist one needs additional arguments to justify the move from claims about causation to the conclusion that minds are extended. Adams and Aizawa (2008) for instance claim: “It simply does not follow from the fact that process X is in some way causally connected to a cognitive process that X is thereby a part of that cognitive process” (2008, p. 91). If they are right, facts about causation cannot tell us anything about facts pertaining to the constitution of a cognitive system. To attempt to argue for EM on the basis of a claim about the causal coupling between agent and world is to conflate causation and constitution. We set out to show how this classic line of argument against EM fails. Thus we aim to remove one of the longstanding obstacles to acceptance of EM in the philosophical literature. Our defence of EM rests on a treatment of constitution as inherently temporal. We call this conception of constitution: diachronic constitution (Kirchhoff, 2015, 2017; Kirchhoff & Kiverstein, 2019b, 2020). Although the notion of constitution is standardly cast in atemporal terms, this chapter argues that temporalising the constitution relation is not as remarkable (nor problematic) as it might initially seem. On the contrary, such a temporalisation of the constitution relation is arguably inevitable, given local interactions between microscale and macroscale states in (coupled) dynamical systems. The upshot of our argument will be that the notion of constitution required for making sense of the constitution of biological processes like minds is a diachronic notion. However, the debate about EM has so far relied upon a more standard understanding of constitution as a synchronic relation of dependence. We will argue this is a mistake. Once we recognise that minds, in the same sense as all other dynamical processes one finds in nature, are diachronically constituted, this calls into question the distinction between causal and constitutive relations of dependence. That is to say: there is less of a difference between constitution and causation than it might initially seem. Constitution and causation are alike precisely because constitution, as this metaphysical concept applies to dynamical systems, is a type of multiscale and reciprocal causation. We will suggest the relevant metaphysical distinction that is needed is not between causal and constitutive dependence but between linear and nonlinear, reciprocal causation. The notion of constitution needed for making sense of the metaphysics of dynamical processes is that of nonlinear and reciprocal causation. We end by considering recent discussions that propose to understand constitution in mechanistic terms. Gallagher (2018), who adopts an interpretation of Krickel’s (2017) work on mechanistic explanation – has argued that it is possible to salvage a notion of synchronic constitution by appealing to work on mechanistic explanation. We consider his account of mechanistic constitution, and point to three key problems with it: (a) it keeps intact the distinction between causation and constitution that this paper challenges; (b) it cannot – contra its own ambitions – make sense of the constitution of dynamical processes, precisely because it assumes that constitution must be synchronic; and (c) it renders the discussion over how to

158

J. Kiverstein and M. Kirchhoff

determine the boundaries of the mind empirically and scientifically intractable given an unarticulated commitment to four-dimensionalism about the nature of the existence of entities. We conclude that the notion of constitution applicable to the mind is best thought of dynamically and thus diachronically (see also Leuridan & Lodewyckx, 2020).2

10.2 Preliminary Remarks on Constitution and Causation In metaphysics, constitution is typically characterised as an atemporal, synchronic one-one relation between (typically) materially and spatially co-located objects that differ in their kind membership (Wasserman 2004). Constitution can be compared to other metaphysical dependence relations such as ‘composition’ – a synchronic relation that holds between many non-overlapping things (e.g. H2O molecules and water, Gillett, 2007; Hawley, 2006), or ‘realisation’, a synchronic dependence relation that holds between property instances (e.g., certain patterns of neuronal activity ‘realise’ pain at some time in virtue of instantiating the functional role of pain at that time, Polger, 2004). Many philosophers engaged in the debate about the extended mind have found the synchronic (atemporal) aspect of metaphysical dependence relations a compelling reason for thinking that causation is fundamentally distinct from metaphysical dependence relations. Leading proponents of EM (and indeed opponents) adhere to this stark delineation between causation and metaphysical dependence relations. One reads for example: EM [Extended Mind] is a claim about the composition or constitution of (some) mental processes. (Rowlands, 2009, p. 54); or Bare causal dependency of mentality on external factors – even when that causal dependency is of the “necessary” kind […] is simply not enough for genuine cognitive extension. What is needed is constitutive dependence […]. (Wheeler, 2010, p. 246)

At the same time Clark and Chalmers (1998) describe the orchestration of an extended cognitive system in terms of a temporally extensive, causal coupling of an agent with entities located external to the body. They say that in cases of extended cognitive processing: […] the human organism is linked with an external entity in a two-way interaction, creating a coupled system that can be seen as a cognitive system in its own right. All the components play an active causal role, and they jointly govern behaviour in the same sort of way that cognition usually does. (1998, p. 8)

 We note here that we cannot and do not aim to provide a full and detailed discussion of the recent surge of work on mechanistic constitution. We focus here on what is most important for the argument for diachronic constitution that we are developing. A full treatment of this literature will be a task for a different paper. 2

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

159

We agree with these authors that EM is best understood as a thesis about the constitution of the mind. Recall the key claim of EM is that mental states like beliefs or memories are sometimes jointly constituted by neural and bodily processes and tools and technologies like notebooks or mobile phones located in the environment. The key issue between opponents and proponents of EM can be framed as follows: under what conditions does a tool or technology that, taken on its own, has no cognitive properties, become a part of a cognitive system, for example a cognitive agent (Clark, 2010, p. 83). What is at stake is the following: what is it for an item that is located externally to the body of an agent to be incorporated as a proper part of a cognitive system. Classical (so-called “first wave”) arguments for the extended mind have claimed that so long as the external item is coupled to the agent through cycles of perception and action in such a way as to drive cognitive behaviour in a similar way to internal cognitive processes, this is sufficient for the item to count as a proper part of a cognitive system. The kind of causal coupling that Clark and Chalmers (1998) argued would be sufficient for an item to count as a part of a cognitive system required that certain conditions, sometimes referred to as “glue and trust” conditions must be satisfied. These conditions include that the item must be easily and reliably accessed (so-called “glue conditions”), and that the item be relied upon automatically (“trust conditions”) as is typically the case for processes that are internal to the individual. Adams and Aizawa (2010) have objected that the claim that “an object or process X is coupled to a cognitive agent does not entail that X is a part of the cognitive agent’s cognitive apparatus” (Adams & Aizawa, 2010, p. 85). To think that coupling with X suffices to make X a part of the agent’s cognitive system is to conflate causation and constitution. This is what has come to be referred to as the “causal-­ constitution fallacy” in the literature (cf. Block, 2005). Constitution is however not a univocal concept but bifurcates into at least two different kinds. The causation-constitution distinction is premised on what we will call “material constitution” (see e.g. Baker, 2000). Minimally, it requires that constitution has the kind of properties exhibited by the notion of material constitution. However, the notion of constitution required for making metaphysical sense of the mind is diachronic. We will argue that EM is not guilty of making a causation-­ constitution fallacy because the notion of constitution required for understanding extended minds is best understood in terms of cycles of nonlinear or reciprocal causality that unfold over time (cf. Kirchhoff, 2017).

10.3 Two Kinds of Constitution We have shown in detail elsewhere how it is possible to delineate two notions of constitution in such a way that the respective notions share none of their key properties with one another (Kirchhoff & Kiverstein, 2019b; Kirchhoff & Kiverstein, 2020; see also Kirchhoff, 2015 for the earliest treatment of diachronic constitution in the discussion over EM). Here, our aim is to briefly rehearse the key differences between material constitution and diachronic constitution.

160

J. Kiverstein and M. Kirchhoff

10.3.1 Material Constitution Take the relation between Michelangelo’s David and a particular piece of marble, Piece. It is widely agreed that a necessary condition for Piece to constitute David is that the relation of constitution involves two coincidence conditions: spatial coincidence – Piece constitutes David at t only if Piece and David have the same spatial location at t; and material coincidence – Piece constitutes David at t only if Piece and David share all the same material parts at t. In addition to these coincidence conditions, the constitution relation in metaphysical debates is said to have the following three formal properties: • Asymmetry: Piece constitutes David, not vice-versa. • Relative fundamentality: if Piece constitutes David, then David is (in some ontological) sense less fundamental than its base, Piece. • Finally, the standard notion of constitution tracks relations between objects of different kinds (as mentioned above). It thus adheres to a species of object ontology: the relata are objects, i.e., discrete spatio-temporally located things with definite properties.

10.3.2 Diachronic Constitution Contrast the case of David with an example of extended cognition. Extended cognitive systems are examples of what Anderson and colleagues call “interaction-­ dominant systems”  – systems in which the constituent elements constrain each other’s dynamics in such a way as to act as single coherent units (Anderson et al., 2012). We will take as our example of extended cognition, dyadic emotion co-­ regulation between infant and caregivers (Krueger, 2011; Varga, 2016; Jurgens & Kirchhoff, 2019).3 We have just seen how synchronic constitution works best when it is used to understand the constitution of material objects such as tables and chairs (Kirchhoff, 2015; Kirchhoff & Kiverstein, 2019a). Unlike material objects, emotions are processes and are therefore inherently diachronic. Processes take up time from the moment they begin to when they end. In contrast to material objects, whose identity conditions are wholly determined at each moment of their existence, processes are  It may be objected that emotions are non-cognitive states on many views of emotion (e.g. Prinz, 2004). Thus, to use emotion regulation as an example of extended cognition is incoherent even if it can be shown that emotions exhibit reciprocal causality. To fully respond to such a line of objection would require us to delve into the debate about how to carve off the cognitive from the non-cognitive. We will not attempt to answer this question in this paper. We do however contend that the distinction between emotion and cognition is now somewhat outdated. Cognitive neuroscience suggests a picture in which cognitive processes like attention and executive control are reciprocally influencing and influenced by emotional processes (Pessoa, 2013). These processes are tightly coupled in the brain. We would argue on the basis of this tight coupling that emotion is partially constituted by cognition and vice versa, in line with the treatment of constitution we are giving in this paper. 3

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

161

dependent on temporal unfolding to be what they are, for processes such as emotions “are creatures of time” (Noë, 2006, p. 28). As Peter Goldie says about emotion: “it is a complex pattern of activity and passivity, inner and outer, which unfolds over time, and the unfolding pattern over time is explanatorily prior to what is the case at any particular time” (2011, p. 119). Emotion regulation has been argued to be a feature of a socially extended process between two or more individuals, and not merely something constituted by a single individual (Krueger, 2011; Varga, 2016; Jurgens & Kirchhoff, 2019). The reason for this turns on a principle from statistical mechanics: generalised synchrony. Generalised synchrony refers to the synchronisation of chaotic dynamics, most often in skew-product (master-slave) systems (Friston & Frith, 2015). In the case of emotion co-regulation, it is considered in the context of reciprocally coupled (dynamical) systems. Generalised synchrony is ubiquitous in the natural world. It shows how multiple systems and their respective dynamics assemble into a coherent ensemble with an overall state of stability. As Varga puts it: “Emotion regulation is not a feature of a single individual, but a socially extended phenomenon. A core aspect that enables this interactive, co-regulatory phenomenon to arise is synchrony: interactions are in synchrony if the gestures, vocalizations expressions, the rhythms of the interactions – coincide” (2016, p. 2474). Varga helpfully highlights another fundamental feature of emotion regulation: temporality in generalised synchrony. If the synchrony between infant and caregiver breaks down, or if it is deliberately broken by the caretaker putting on a neutral face (as in the famous Still Face experiment), infants become distressed. Hence, the generalised synchrony between infant and caretaker can be lesioned, breaking the shared ‘dynamical narrative’ (Friston & Frith, 2015, p.  12). Finally, the idea of generalised synchrony implies the slaving principle. The interaction between the individuals unfolds over slower timescales than does the microscale elements (e.g. the gestures, expressions and patterns of neural activity). This implies that the dynamics at the macroscale of the co-regulating system made up of infant and caregiver can be understood as impacting on the degrees of freedom of the dynamics at the microscale in each of the individuals.4

 It could be argued that emotion regulation is not a case of extended cognition as we have defined it above. Even if it is characterised by reciprocal causality and thus best understood in terms of diachronic constitution, the processes that are involved in emotion regulation are just neural and bodily. The contribution of the environment has not yet been established. This shortcoming in our argument is easily corrected by pointing to examples discussed in the literature on extended emotions that illustrate how artefacts can play a role in emotion regulation  (see e.g. Colombetti & Roberts, 2014; Colombetti & Krueger, 2015). Consider for instance Colombetti and Roberts’s (2014) example of Renier, a person of melancholic temperament. Renier keeps a diary with him containing inspirational quotations that help him to feel less miserable and dejected. Colombetti and Roberts suppose that after a while Renier may through the use of the notebook even come to develop a positive temperament. Renier’s use of the notebook (just like in the famous case of Otto) we suggest counts as an example of reciprocal causality. The cycles of perception and action that couple Renier to his notebook stand in a relation of reciprocal causality. They do so in such a way as to regulate his mood. His use of the notebook to regulate his mood is thus an example of mood regulation in which the notebook gets to play a constitutive role. 4

162

J. Kiverstein and M. Kirchhoff

There is an immediate complication with trying to shoehorn the concept of material constitution (as defined above) to make it fit interaction-dominant systems, such as the socially extended cognitive system that forms between infant and caregiver. One cannot begin by asking how the parts, X, constitutes the whole, Y, at an atemporal instant t. The main reason is that both the parts and the whole are processes. Crucially, what there is of a process at any given time is not sufficient to determine that a particular macroscale process is unfolding. Just think here of a Mexican wave. One cannot screen-off the immediately preceding dynamics; the temporal unfolding of a Mexican wave is essentially part of what it is for something to be such a phenomenon. Similarly, emotion co-regulation is inherently temporal, or diachronic. The synchronous patterns of interaction that form between infant and caregiver unfold over time can never be wholly present at a synchronic instant t. Second, the relation of co-regulation itself takes time, which implies that the relata themselves occur at different points in time. Specifically, the exchange of gestures and facial expressions unfolds over slower and therefore over longer timescales than any of the patterns of neural activity. Moreover, the interaction between infant and caregiver is continuous, in the precise sense that temporality cannot be broken down into discrete quanta. So were one to claim that the system in question is in a particular ‘state’, X, at a particular point in time, this would be an approximation – it would “boil down to saying that the average of the system’s states during that period of time was X” (Spivey, 2007, p. 30). This point has both metaphysical and methodological implications. Processes are temporally spread out in nature, which implies that a process cannot be wholly present at any given point in time. To insist that processes can be described as being present at some synchronic point in time is possible only by taking an average of the states the process goes through over a period of time. To see this more clearly, consider what Ladyman and Ross (2007) say about the relation between water and H20. Under the synchronic model of constitution, if H2O composes or constitutes water at some point in time t, or at each temporal stage over t1… tn, then both of the hydrogen molecules and the oxygen molecule must be entirely present at that point in time or at each moment over t1… tn. According to Ladyman & Ross, this synchronic view is however mistaken. Water is constituted “by oxygen and hydrogen in various polymeric forms, such as (H2O)2, (H2O)3), and so on, that are constantly forming, dissipating, and reforming over short time periods in such a way as to give rise to the familiar properties of the macroscopic kind water” (2007, p. 21; italics added). Because water is constituted in a complex dynamical system it “makes no sense to imagine it having its familiar properties synchronically” (Ross & Ladyman, 2010, p. 160; italics added). To claim that a dynamical process has the properties it has at some point in time is an idealisation that should not be mistaken for how the properties of dynamical processes manifest themselves over time. In other words, from the fact that a system can be described synchronically, it does not follow that it has any of its special characteristics synchronically. The behaviour of interaction-dominant systems is best captured within a broadly speaking process-based ontology. Process-based ontologies are proposed as an alternative to substance ontologies that start from self-subsisting entities and their

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution… Table 10.1  Two concepts of constitution across two examples

Synchronic constitution Michelangelo’s David  Synchronicity  Asymmetry  Object/entity-based  Non-causal

163

Diachronic constitution Emotion co-regulation  Diachronicity  Symmetry (co-constitution)  Process-based  Reciprocal causation

properties. Processes differ ontologically from entities and their activities that can be located discretely in space-time. Instead of focusing on the material and spatial composition of the whole by its parts as is the case with synchronic constitution relations, we are proposing to individuate dynamical processes in terms of their behaviour over time in particular contexts. Diachronic constitution shifts focus away from material elements and their co-location (without denying that this is a part of constitution) to a perspective of dynamical unfolding over time of the behaviour of whole complex non-linear systems. The differences are summarised in Table 10.1, and they suggest the need to distinguish between two (mutually exclusive) notions of the relation of constitution.

10.4 Dissolving the Causal-Constitution Fallacy We have been arguing that synchronic constitution is not the notion of constitution that is needed for the metaphysics of the extended mind, precisely because extended cognitive systems are dynamical and processual in nature. Yet it is the synchronic notion of constitution that both proponents and opponents of the extended mind have relied upon. Thus consider the so-called first-wave arguments for the extended mind. They have been focussed on the conditions under which an element locatable at a specific space-time point that has no cognitive properties of its own, such as a notebook, could count as a proper part of a cognitive system. Clark argues that under certain conditions coupling can suffice to make an element located external to an individual a proper part of that individual’s cognitive system. Opponents of EM deny this in part on the grounds that to think otherwise is to conflate causation and constitution. This debate has played out however based on what Wilson has described as a “compositional” understanding of constitution (Wilson, 2009). Such a compositional understanding of constitution is synchronic: it is the relations of spatial and material coincidence of the components of a system at an instant in time t that determines what does or does not count as a proper part of a system. We have suggested that what is distinctive about putative cases of EM is that they are characterised by reciprocal causality. If the claim that cognitive systems exhibit reciprocal causality can be defended, it will follow that the notion of constitution required to understand such systems is diachronic, not synchronic. The classical objection to EM, that it is premised on a confusion of causation and constitution, fails. There is no such fallacy and no such confusion since the notion of constitution needed to understand EM is one that is best analysed in terms of reciprocal

164

J. Kiverstein and M. Kirchhoff

causality. The notion of constitution needed to understand extended cognitive systems is, we submit, just a type of causation: reciprocal causation. This observation challenges a number of common assumptions across philosophy of mind, philosophy of science and metaphysics. First, causal and constitutive dependence are commonly distinguished on the grounds that cause and effect are distinct relata, whereas this is not the case (so the assumption goes) for constitutive dependence. Krickel (echoing Craver, 2007) for instance writes: “one platitude about causation is that its relata have to be wholly distinct. Thus, it follows that causal relevance and constitutive relevance are mutually exclusive relations” (Krickel, 2019, p. 5).5 We are making what may be seen as a radical move by many analytic metaphysicians of rejecting this platitude. Causal and constitutive dependence are not necessarily premised upon distinct relata, for both kinds of dependence relations back explanations by appeal to notions such as ‘because’ and ‘in virtue of’ structures. What they share in common is that both types of talk can answer questions about the generation of structures like the patterns of generalised synchrony that temporarily form between infant and caregiver. What can make a causal explanation also count as a constitutive explanation is, we have argued, the relations of reciprocal two-way dependence that hold between micro- and macroscale processes. Moreover, in cases of reciprocal causality cause and effect are not wholly distinct relata but enter into circular, and symmetrical relations with each other as we just described. Indeed, once one acknowledges that reciprocal causation in dynamical systems, one must also acknowledge that cause and effect are not always wholly distinct. On the contrary, such an assumption likely only holds in cases of linear causality. It does not hold for processes in which microscale and macroscale dynamics stand in reciprocal and circular causal relations to one another. Second, causation is often treated as a strictly intralevel relation (e.g., between parts and their activities), while constitution is cast as an entirely interlevel relation. In introducing the notion of diachronic constitution, Kirchhoff (2015) makes the same delineation between interlevel (diachronic) constitution and intralevel causation. Here we make an important amendment to Kirchhoff (2015): diachronic constitution implies that causation can be interlevel because it can operate over multiple spatial and temporal scales. Reciprocal causation is a micro- macro-level relation (see also Leuridan & Lodewyckx, 2020 for discussion).6

 It should be noted this is a common assumption in the literature on the metaphysics of causation. Thus it is common to read examples like event x – the boy kicking the football caused event y – the glass in the window shattering (e.g. Salmon, 1998). We are arguing this feature of causation is absent in cases of reciprocal causality and thus cannot be used to distinguish causation from constitution. 6  There are some interesting points of overlap between Leuridan and Lodewyckx (2020) and our arguments here for diachronic constitution (building on Kirchhoff (2015) and Kirchhoff & Kiverstein (2019a, b)). Of special interest is the push towards eliminating the standard view that causation and constitution are independent relations in virtue of causation being intralevel and constitution an interlevel relation of dependence. In this paper we do not discuss this overlap with Leuridan and Lodewyckx (2020). This will be an important task for another occasion. 5

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

165

In sum, it is not at all obvious why one should accept that causal and constitutive relevance are mutually exclusive relations when it comes to understanding the metaphysics of dynamical systems. One might of course ask the following: could authors working on EM not use a process-based and synchronic notion of constitution? The short answer is: ‘no’. One cannot both claim that cognition is processual and that the constitutive basis of cognition is synchronic. The reason is that processes cannot be what they are at a synchronic point in time. Note, to persist, in the sense of a process, is to persist as unfolding over time. If the very nature of a process is to unfold over time then how can its existence be determined at an atemporal instant? In the case of David and Piece this dilemma does not arise, simply because the relation of constitution holds between objects and is articulated at a very high degree of abstraction. Objects such as Piece and David, however, have their identity determined at every moment at which they exist. Hence, unlike processes, objects such as David and Piece do not depend for their existence on dynamical unfolding over time. Objects are therefore said to endure. By contrast, processes perdure – a process does not have a fixed identity at any time, it is not wholly present at any single point in time (see Goldie, 2011; Hofweber & Velleman, 2011). Processes cannot be constituted at a point in time by relata that are inherently temporal because processes do not wholly exist at any single point in time. One immediate suspicion some readers might harbour is that the notion of diachronic constitution does not leave any space for competitor views to EM, such as the influential embedded cognition thesis championed by Rupert (2009). The embedded view for instance takes cognition to sometimes be explainable in the terms of dynamical systems theory. Proponents of this view allow human cognition to depend heavily on actively engaging with structures in the environment (Rupert, 2009; Sterelny, 2010). Yet it stops short of endorsing EM, for it takes cognition, in the vast majority of cases, to be constitutively bounded by the organism. The claim that cognition is organism-bound that is central to the embedded theory, requires that only what lies inside of the boundary of the organism can be treated as wholly constitutive of the mind. Anything external to this boundary must make a merely causal contribution. Proponents of the embedded theory have advanced a number of considerations for taking the boundary of the organism to be privileged. Rupert (2009) has argued for instance that X counts as a cognitive state only if X is the state of a component of a cognitive system. He defines a cognitive system as an integrated set of mechanisms, a persisting architecture that causes intelligent behaviour across a variety of circumstances. Putative cases of extended cognition by contrast are softly assembled on the fly as we have seen above. They do not persist and are not integrated in the right sort of way according to Rupert to count as cognitive systems. Rupert does not deny the environment can make important, even necessary, causal contributions to a cognitive system. However, he denies such contributions can ever be constitutive of cognition. Rupert’s embedded theory of cognition relies on a synchronic understanding of constitution insofar as it is premised on a distinction between causation and constitution. We have argued that a synchronic conception of constitution is ill-suited for understanding putative cases of extended cognition. In arguing for a diachronic

166

J. Kiverstein and M. Kirchhoff

constitution we take ourselves to have also provided considerations against the embedded view.7 The objection is correct that diachronic constitution leaves no space for a weaker embedded theory of cognition. However, it does so without begging any questions by pointing to the ill-fittingness of the synchronic view of the constitution which the embedded theory of cognition is premised upon. We recognise of course that our notion of diachronic constitution will not go unchallenged. Indeed, by leveraging new work in the literature on mechanistic explanation, Gallagher (2018) argues that E- approaches (ecological, embodied, enactive and extended) to cognition require a mix of causal (diachronic) and constitutive (synchronic) relations. Moreover, Rupert, insofar as he defines cognitive systems as persisting sets of mechanisms (Rupert, 2009), can also be seen as developing a mechanistic account of constitution that he could appeal to in pushing back on the argument of this paper. In the next section we reconstruct Gallagher’s (2018) account of constitutive mechanistic explanation. We then proceed to look at specific problems with this particular approach to constitution in the 4E discussions. We conclude that there is no reason to prefer a mechanistic account of constitution over diachronic constitution. Indeed, Gallagher’s (2018) proposal does not in the end take us beyond the straightjacket of the alleged CC-fallacy.

10.5 Saving the Causal-Constitution Distinction Gallagher (2018) starts with what we think is the right initial observation. He says: “Some versions of embodied cognition, including enactivist and extended mind approaches, […] require a diachronic conception of constitution that includes reciprocal causal relations in order to match the dynamical and holistic conception of cognition that they defend” (2018, p. 2). Yet Gallagher stops short of endorsing our view of diachronic constitution. He is explicit about this, saying: “Rather than giving up the notion of constitution […] enactivism needs to conceive of constitution  Rupert will likely object that his account of cognitive systems as integrated and persisting sets of mechanisms provides additional considerations for endorsing a synchronic understanding of constitution. Rupert puts his account of cognitive systems to work to argue that the boundary of an organism is fixed and non-negotiable. We have argued elsewhere that organismic boundaries are always fragile, negotiable and hard-won (see Author’s articles). Thus we would dispute Rupert’s characterisation of cognitive systems but here is not the place to get into this disagreement. Sprevak (2010) has argued that the scientific findings do not decide between embedded and extended theories. He has argued that inference to the best explanation arguments for the extended mind or against it fail. We agree with Sprevak that inference to the best explanation arguments fail. However, we disagree that the science cannot be used to decide between embedded and extended theories of the mind. We have argued that the dynamical cognitive science used in support of EM is best understood in terms of diachronic and not synchronic constitution. Attempts to interpret dynamical cognitive science in terms of a weaker embedded theory of mind rely upon a synchronic understanding of constitution that may apply to material objects but that is not apt to describe selforganising processes that exhibit reciprocal causality. Our thanks to an anonymous reviewer for discussion of this point. 7

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

167

as including causal relations without ruling out compositional relations. One could thus propose two forms of constitution: compositional constitution and causal constitution” (2018, p. 9). Gallagher’s enactivism is therefore committed to a diachronic and temporal view of constitution (effectively what we call diachronic constitution) and an atemporal and non-causal account of constitution (what he calls compositional constitution). We think this attempt to keep intact an atemporal notion of constitution while at the same time recognising that constitution is diachronic is unstable. However, before we turn to illustrate the problem with Gallagher’s proposal, we need to get his proposal on the table. Gallagher is inspired by the new mechanists in philosophy of science. Building on the work of Krickel (2017), he suggests that the mutual manipulability account of constitutive relevance keeps intact a clear distinction between causation and constitution; yet, without this account falling prey to the CC-fallacy. In brief, the mutual manipulability (MM) criterion defines the notion of constitution in terms of two necessary conditions: (i) the constituents are materially and spatially co-located with the constituted phenomenon; and (ii) the constituents parts and the phenomenon are mutually manipulable (Krickel, 2019). The MM criterion is due to Craver (2007). He defines it in the following fashion: “a component is relevant to the behavior of a mechanism as a whole when one can wiggle the behavior of the whole by wiggling the behavior of the component and one can wiggle the behavior of the component by wiggling the behavior as a whole. The two are related as part to whole and they are mutually manipulable” (2007, p. 153; italics in original). The MM criterion is designed to illustrate how components of mechanisms and their boundaries are both defined and determined across certain empirical sciences such as neuroscience and biology (Kaplan, 2012). Thus, the mutual manipulability criterion is “effectively a boundary demarcation criterion” (Kaplan, 2012, p. 557). Prima facie, the MM criterion provides a means of addressing the distinction between causal and constitutive relevance, and it does so without the typical question begging philosophical intuitions surrounding the debates about how to draw the boundaries of the mind in 4E cognitive science.8 Gallagher (2018) does not adopt the MM criterion without modification. The reason turns on a specific problem with the traditional formulation of the MM criterion. On the one hand, the MM criterion defines the constitutive relation between parts and whole as synchronic (atemporal). On the other hand, Craver (2007) develops his account of mutual manipulability by appeal to the interventionist account of causation developed by Woodward (2003). The problem is then very clear: the MM

 Applying this criterion for constitutive relevance to the discussion of 4E cognition is promising, for unlike other rivals for ‘marks of the mental’, the MM criterion has the virtue of avoiding begging the question in the discussion of how to demarcate the boundaries of the mind. Adams and Aizawa (2008) suggests that the mark of the mental has to be with non-derived content. Yet this has been argued to beg the question against EM (Clark, 2010; Menary, 2007). Furthermore, Rupert (2009) sought to overthrow arguments for EM by appealing to functional differences between the internal dynamics of neuronal functioning and external technologies. However, this has been argued to imply a form of biochauvinism (Chalmers, 2019; Sprevak, 2009). 8

168

J. Kiverstein and M. Kirchhoff

criterion cannot salvage a clear-cut distinction between causation and constitution (for similar critiques see Kirchhoff, 2017 and Leuridan, 2012). Indeed, as Craver (2007) sets it up, the MM criterion cannot be an account of constitutive relevance, for it picks out relations of causal relevance between parts and whole. To suggest otherwise would be to conflate causation with constitution. It would be a case of committing the CC-fallacy. We agree with this initial assessment. Gallagher (2018) adopts a modification to Craver’s (2007) account of the MM criterion developed by Krickel (2017). On this account it is possible to preserve the view that the part-whole (compositional) relation between parts and constituted whole is synchronically constitutive, while keeping intact the idea that inter-level relations are causal (diachronic) relations that can be captured by the MM criterion. Krickel (2018) puts this as follows: “This approach starts with Craver’s original definition of constitutive relevance, but interprets mutual manipulability in causal terms while maintaining that causation and constitution are mutually exclusive relations. The core idea of my approach is that manipulability consists in a causal relation between what I call a temporal […] part (Krickel, 2017) of the phenomenon and a component of the mechanism” (Krickel, 2018, p. 1). See Fig. 10.1 for an illustration of the idea. As we understand this model of mechanistic constitution, there are three relevant variables: the phenomenon (S), a spatial part of a mechanism, i.e., an entity and its activity (X1), and a temporal part of the phenomenon S. The idea is that entities and their activities can be defined at both the level of the mechanism and at the level of the phenomenon as a whole. So what is the problem with Gallagher’s proposal? First, this modified account of the MM criterion cannot overcome the CC-fallacy. The reason for this is: the constitutive relation holds atemporally between the sum of micro-level parts and the

Fig. 10.1  The MM criterion and inter-level, diachronic causal relations (from Gallagher, 2018)

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

169

macro-level temporal parts. Hence, at any given time, employing the MM criterion as a criterion of constitutive relevance will only identify relations between microscale and macroscale processes in causal and diachronic terms. By retaining a distinction between constitution and causation, where constitution is atemporal (synchronic) and causation is temporal (diachronic), it is hard to see how one can apply the MM-criterion at some point in time without conflating causation and constitution. This is not a problem for our account of diachronic constitution, for it does not seek to retain the distinction between causation and constitution. It is however a problem for Gallagher who seeks to retain this distinction. Second, note that the MM criterion was intended to transform philosophical intuitions about systemic parts and boundaries into tractable, empirical issues (Kaplan, 2012). Indeed, it should be entirely acceptable to both defenders and opponents of EM given its broader explanatory success in other areas of science and philosophy of science (Krickel, 2019). But, the application of MM by Gallagher implies that the ‘whole’ cannot be intervened upon. In the end, Gallagher’s proposal commits him to a version of four-dimensionalism, where the parts of the whole or phenomenon at some time t are atemporally part of the larger whole or phenomenon over t … tn. We do not dispute that one can defend four-dimensionalism on metaphysical grounds. The problem with four-dimensionalism in this particular context is that it turns what was intended to be a proposal about empirical tractability into a pure theoretical abstraction. Indeed, one cannot wiggle the whole system, for the whole or the phenomenon does not exist at any given time. On a four-dimensionalist view the system is not wholly present at each moment of its existence. The mechanistic view of constitution ends up positing the existence of a phenomenon that is empirically intractable: the whole exists as a four-dimensional entity; yet there are no scientific means by which to intervene on it. There is no such problem for our account of diachronic constitution; indeed, it is underwritten by formal principles in the sciences such as the slaving principle in statistical physics, and one can use these principles to gain empirical tractability on the issue of where to draw the boundary of the mind. Finally, by retaining the distinction between causation and constitution, the formulation of mechanistic constitution results in reaffirming constitution as synchronic and causation as diachronic. This reveals a tension in Gallagher’s version of mechanistic constitution. In offering a view of constitution tailored for 4E (embodied, embedded, enactive, extended) cognition, he aims to provide a metaphysics of mind that is thoroughly temporal. In retaining the causation-constitution distinction through endorsing a composition view of constitution, he is required to understand constitution in atemporal or synchronic terms. Yet, it is this precise way of conceiving of the causal-constitutive relation that we have shown to be problematic in the context of dynamical systems. It finds no home in the natural world of dynamical systems, and positing the usual distinction between causation and constitution reaffirms the exact same assumptions that fuels the intuitions behind the CC-fallacy. In the end, then, the MM criterion picks out reciprocal causal relations between micro- and macroscale processes. We have argued that, in the case of emotion co-­ regulation, macroscale processes, which evolve from microscale processes, can be shown to feed back down and constrain the dynamics unfolding over faster,

170

J. Kiverstein and M. Kirchhoff

microscopic timescales. The dynamics unfolding over slower timescales constrain the degrees of freedom of the dynamics unfolding over faster timescales, even if it is the dynamics over faster timescales that give rise to the observed ensemble behavior. Our argument for a diachronic conception of constitution lends support to the use of the MM criterion to demarcate the boundaries of mind (cf. Kirchhoff & Kiverstein, 2019b). For example, it becomes possible to lesion or break the operations of the entire process by manipulating or intervening on component processual parts, and vice versa. Indeed, it turns out to support the way in which Clark and Chalmers (1998) originally motivated the concept of an extended cognitive system as forming out of a two way causal interaction between an agent and their environment (cf. Palermos, 2014).9 Diachronic constitution applies to this original definition of EM, and an important upshot is that diachronic constitution breaks with the idea that the constitution relation is a relation that holds synchronically between parts and whole. A final remark on Gallagher’s proposal is needed. Gallagher (2018) moves on to make it clear, without doubt, that “[c]ognitive processes occur on several timescales.” He says that it is “important to grasp how these timescales interrelate in order to understand how constitution can be, not only diachronic, but also non-­ linearly dynamic” (2018, p. 10). In making his case, Gallagher relies on a “threefold distinction proposed by Varela (1999) among elementary, integrative and narrative timescales – a distinction based on neurobiology and phenomenology. In fact, however, only the first two will be relevant for the following analysis. –– The elementary scale (varying roughly between 10s and 100s of milliseconds) –– The integrative scale (varying roughly from 0.5 to 3 seconds) –– The narrative scale involving memory (above 3 seconds).” (2018, p. 10) We entirely agree with Gallagher on these points. In fact, this is precisely what our account of diachronic constitution is designed to account for. We showed in our discussion of emotion co-regulation the importance of attending to different scales of dynamics, and we argued that a central feature determining these dynamics is temporality: the different timescales over which dynamics unfold and enslave one another. However, once you take diachronicity as an essential feature for what it is for something to be a cognitive process, and you seek to show how different cognitive processes unfold over different timescales, the additional inclination to hang onto a synchronic (compositional) notion of constitution makes little sense. What we hope to have established here is that for all involved in developing the metaphysics of 4E philosophy of cognitive science, one should stop assuming that there is a

 There is an ongoing discussion in the mechanistic literature on the possibility of applying the MM criterion to justify EM. Krickel (2019) suggests that it leads to what she calls ‘trivial extendedness’. Others such as Hewitson et al. (2018) seek to overcome this triviality problem for EM via the MM criterion by appealing to the notion of ‘causal specificity’. Others still argue that it is impossible to empirically vindicate EM by appeal to the MM criterion (Baumgartner & Wilutzky, 2017). We mention these issues here, only to set them aside. A proper discussion of these issues will be a task for a follow-up paper. 9

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

171

unique marker separating mere causation from constitution. One should ask instead: what plays a relevant and salient causal role in the generation of some cognitive phenomenon. Diachronic constitution allows one to address such a question in a metaphysical way, whilst adhering to our best sciences of the mind.

10.6 Concluding Remarks The argument of this paper is somewhat unusual in the context of previous responses to the CC-fallacy. We did not attempt a defence of EM premised on (some variety of) functionalism (cf. Clark, 2008; Wheeler, 2010) or by adopting an alternative framework for thinking about EM such as complementarity (cf. Sutton, 2010) or cognitive integration (cf. Menary, 2010). The debate concerning EM has so far been concerned with how to draw the causal-constitution distinction. We have offered a defence of EM that calls into question this very distinction. If there is no causal-­ constitution distinction  that can be drawn for cognitive systems, then there is no CC-fallacy. Adopting a diachronic perspective on constitution still enables one to say that “P is a constitutive element if P is part of the processes that produce X” (De Jaegher et al., 2010, p. 443). Yet it rejects both the synchronicity and asymmetry assumptions of the standard account of constitution in the context of coupled dynamical systems. Diachronic constitution is a notion of constitution premised on reciprocal (non- linear) causation between coupled (open) dynamical systems. Reciprocal causality is ubiquitous in the natural world. We observe it wherever there is generalised synchrony in multiscale systems. We have argued, it is the notion of constitution at the heart of EM – not synchronic constitution. Crucially, the solution proposed in this paper generalises to the entire 4E (embodied, embedded, enactive, extended) movement in the philosophy of cognitive science. All these frameworks seek to establish that cognition is constituted by dynamics that break across the brain, body and environment. Yet they all move from facts about causal relations to facts about constitutive relations. This opens all the arguments for 4E cognition to the CC-fallacy, a fact that critics of the 4E movement have not been slow to pounce upon (see Adams & Aizawa, 2008; Rupert, 2009). Diachronic constitution is a key concept needed for developing a metaphysics of the extended (embodied, embedded, enactive) mind.

References Adams, F., & Aizawa, K. (2001). The bounds of cognition. Philosophical Psychology, 14(1), 43–64. Adams, F., & Aizawa, K. (2008). The bounds of cognition. Blackwell. Adams, F., & Aizawa, K. (2010). Defending the bounds of cognition. In R. Menary (Ed.), The extended mind (pp. 67–80). The MIT Press.

172

J. Kiverstein and M. Kirchhoff

Aizawa, K. (2010). The coupling-constitution fallacy revisited. Cognitive Systems Research, 11, 332–342. Anderson, M. L., Richardson, M. J., & Chemero, A. (2012). Eroding the boundaries of cognition: Implications of embodiment. Topics in Cognitive Science, 4(4), 717–730. Baker, L. B. (2000). Persons and bodies: A constitution view. Cambridge University Press. Baumgartner, M., & Wilutzky, W. (2017). Is it possible to experimentally determine the extension of cognition? Philosophical Psychology, 30(8), 1104–1125. Block, N. (2005). Review of Alva Noë’s action in perception. Journal of Philosophy, 102, 259–272. Chalmers, D. (2019). Extended cognition and extended consciousness. In M. Colombo, E. Irvine & M. Stapleton (Eds.), Andy Clark and his Critics (pp. 9–21). Oxford University Press. Chemero, A. (2009). Radical embodied cognitive science. MIT Press. Clark, A. (2008). Supersizing the mind. Oxford University Press. Clark, A. (2010). Coupling, constitution and the cognitive kind: A reply to Adams and Aizawa. In R. Menary (Ed.), The extended mind (pp. 81–99). MIT Press. Clark, A., & Chalmers, D. (1998). The extended mind. Analysis, 58, 7–19. Colombetti, G., & Krueger, J. (2015). Scaffoldings of the affective mind. Philosophical Psychology, 28(8), 1157–1176. Colombetti, G., & Roberts, T. (2014). Extending the extended mind: The case for extended affectivity. Philosophical Studies, 172, 1243–1263. Craver, C.  F. (2007). Explaining the brain: Mechanisms and the mosaic unity of neuroscience. Oxford University Press. De Jaegher, H., Di Paolo, E., & Gallagher, S. (2010). Can social interaction constitute social cognition? Trends in Cognitive Science, 14(10), 441–447. Friston, K., & Frith, C. (2015). A duet for one. Consciousness and Cognition, 36, 390–405. Gallagher, S. (2018). New mechanisms and the enactivist concept of constitution. In M. P. Guta (Ed.), The metaphysics of consciousness (pp. 207–220). Routledge. Gillett, C. (2007). Understanding the new reductionism: The metaphysics of science and compositional reduction. The Journal of Philosophy, 104(4), 193–216. Goldie, P. (2011). Grief: A narrative approach. Ratio, 2, 119–137. Haugeland, J. (1998). Mind embodied and embedded. In J.  Haugeland (Ed.), Having thought (pp. 207–237). Harvard University Press. Hawley, K. (2006). Principles of composition and criteria of identity. Australasian Journal of Philosophy, 84(4), 481–493. Hewitson, C. L., Kaplan, D. M., & Sutton, J. (2018). Yesterday the earwig, today man, tomorrow the earwig? Comparative Cognition & Behavior Reviews, 13, 25–30. Hofweber, T., & Velleman, D. (2011). How to endure. The Philosophical Quarterly, 61(242), 37–57. Hurley, S. L. (1998). Consciousness in action. Harvard University Press. Jurgens, A., & Kirchhoff, M. D. (2019). Enactive social cognition: Diachronic constitution and coupled anticipation. Consciousness and Cognition, 70, 1–10. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology and Philosophy, 27, 545–570. Kirchhoff, M.  D. (2015). Extended cognition & the causal-constitutive fallacy: In search for a diachronic and dynamical conception of constitution. Philosophy and Phenomenological Research, 90(2), 320–360. Kirchhoff, M. D. (2017). From mutual manipulation to cognitive extension: Challenges and implications. Phenomenology and the Cognitive Sciences, 16, 863–878. Kirchhoff, M. D., & Kiverstein, J. (2019a). Extended consciousness and predictive processing: A third-wave view. Routledge. Kirchhoff, M. D., & Kiverstein, J. (2019b). How to determine the boundaries of mind: A Markov blanket proposal. Synthese, 198(5), 4791–4810. Kirchhoff, M. & Kiverstein, J. (2020). Attuning to the world: the diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11. https://doi.org/10.3389/ fpsyg.2020.01966

10  Dissolving the Causal-Constitution Fallacy: Diachronic Constitution…

173

Krickel, B. (2017). Making sense of interlevel causation in mechanisms from a metaphysical perspective. Journal for General Philosophy of Science, 48(3), 453–468. Krickel, B. (2018). Saving the mutual manipulability account of cognitive relevance. Studies in History and Philosophy of Science, 68, 58–67. Krickel, B. (2019). Extended cognition, the new mechanists’ mutual manipulability criterion, and the challenge of trivial extendedness. Mind & Language, 35(4), 539–561. Krueger, J. (2011). Extended cognition and the space of social interaction. Consciousness and Cognition, 20(3), 643–657. Ladyman, J., & Ross, D. (2007). Every thing must go: Metaphysics naturalized. Oxford University Press. Leuridan, B. (2012). Three problems for the mutual manipulability account of constitutive relevance in mechanisms. British Journal for the Philosophy of Science, 63, 399–427. Leuridan, B., & Lodewyckx, T. (2020). Diachronic causal constitutive relations. Synthese, 198, 9035–9065. Menary, R. (2007). Cognitive integration: Mind and cognition unbounded. Palgrave Macmillan. Menary, R. (2010). Cognitive integration and the extended mind. In R. Menary (Ed.), The extended mind (pp. 227–243). The MIT Press. Noë, A. (2004). Action in perception. The MIT Press. Noë, A. (2006). Experience of the world in time. Analysis, 66(1), 26–32. Palermos, O. (2014). Loops, constitution and cognitive extension. Cognitive Systems, 27, 25–41. Pessoa, L. (2013). The cognitive-emotional brain: From interactions to integration. MIT Press. Polger, T. (2004). Natural minds. MIT Press. Prinz, J. (2004). Gut reactions: A perceptual theory of the emotions. Oxford University Press. Ross, D., & Ladyman, J. (2010). The alleged coupling-constitution fallacy and the mature sciences. In R. Menary (Ed.), The extended mind (pp. 155–166). The MIT Press. Rowlands, M. (2009). Enactivism and the extended mind. Topoi, 28, 53–62. Rupert, R. (2009). Cognitive systems and the extended mind. MIT Press. Salmon, W. (1998). Causality and explanation. Oxford University Press. Spivey, M. (2007). The continuity of mind. Oxford University Press. Sprevak, M. (2009). Extended cognition and functionalism. Journal of Philosophy, 106(9), 503–527. Sprevak, M. (2010). Inference to the hypothesis of extended cognition. Studies in the History and Philosophy of Science, 41(4), 353–362. Sterelny, K. (2010). Minds: Scaffolded or extended? Phenomenology and the Cognitive Sciences, 9, 465–481. 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. Van Gelder, T., & Port, R. (1995). It’s about time: An overview of the dynamical approach to cognition. In R. Port & T. van Gelder (Eds.), Mind as motion: Explorations in the dynamics of cognition (pp. 1–44). MIT Press. Varela, F. (1999). The specious present: a neurophenomenology of time consciousness. In J. Petitot, F.  J. Varela, B.  Pacoud & J-M.  Roy (Eds.), Naturalizing Phenomenology (pp.  266–314). Stanford University Press. Varga, S. (2016). Interaction and extended cognition. Synthese, 193, 2469–2496. Wasserman, R. (2004). The constitution question. Nous, 38(4), 693–710. Wheeler, M. (2010). In defense of extended functionalism. In R. Menary (Ed.), The extended mind (pp. 245–270). The MIT Press. Wilson, R. (2009). The transitivity of material constitution. Noûs, 43(2), 363–377. Woodward, J. (2003). Making Things Happen: A Theory of Causal Explanation. Oxford University Press.

Chapter 11

Why Diachronic Constitution Won’t Help. Commentary on “Dissolving the Causal-­Constitution Fallacy” Beate Krickel

Abstract  In this short commentary, I will first show why Kiverstein & Kirchhoff's (this volume) analysis of the CC-fallacy is inadequate in an important way and show why their strategy for avoiding the CC-fallacy based on diachronic constitution is problematic. Second, I will reply to their criticism of the mechanistic account of constitution.

11.1 Introduction Kiverstein and Kirchhoff’s (K&K’s) chapter can be divided into two parts. In the first, K&K present their own positive account. They explain what diachronic constitution is, how it differs from what they take the standard view of constitution to be, i.e., material constitution, and how applying diachronic constitution in the defense of extended cognition (EC) avoids the CC-fallacy. In the second part, they discuss the mechanistic account of constitution and argue that it does not provide an adequate analysis of constitution in the context of EC. The first part  of their chapter is mainly a summary of what K&K (2020) and especially Kirchhoff (2015, 2017) have argued elsewhere. I have raised some worries concerning Kirchhoff’s account in my article in this volume. The main worry expressed there is that diachronic constitution is incompatible with the co-location principle and, thus, cannot be used to defend EC. I will not repeat this objection here but highlight further problems for K&K’s account. In the following, I will, first, discuss the notion of diachronic constitution. I will show why K&K’s analysis of the CC-fallacy is inadequate in an important way and show why K&K’s strategy for avoiding the CC-fallacy based on diachronic constitution is problematic. Second, I will reply to their criticism of the mechanistic account of constitution. B. Krickel (*) Institute of History and Philosophy of Science, Technology, and Literature, Technical University Berlin, Berlin, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_11

175

176

B. Krickel

11.2 The CC-Fallacy and Diachronic Constitution In the first part of their paper, K&K repeat Kirchhoff’s (2015) core argument against the so-called coupling-constitution (CC) fallacy: the CC fallacy objection against extended cognition (EC) is question-begging. According to K&K, the CC fallacy rests on assumptions about constitution that defenders of EC reject. Adams and Aizawa, they argue, presuppose a notion of constitution that is known from traditional metaphysics  (also known as “material constitution”). This notion must be rejected for two reasons. First, the relata of material constitution  are taken to be static objects such as statues and lumps of clay. In contrast to that, EC deals with processes, which are temporally extended. Defenders of EC are interested in whether the processes of remembering, language processing, or navigating are constituted by processes such as eye movements, gesturing, or consulting a notebook. The second reason for rejecting the material notion of constitution, according to K&K, is that material constitution is taken to be a synchronic relation. Constitution between processes, however, cannot by synchronic but must be thought of as diachronic. They go on arguing that if the synchronous nature of constitution is rejected, the contrast between causation and constitution disappears. K&K’s central claim is that constitution is reciprocal causation. The CC fallacy holds only if causation (or causal coupling) and constitution are taken to be mutually exclusive relations. If constitution just is a specific form of causation, the inference from causation to constitution can (at least sometimes) be justified. When taking a closer look at Adams and Aizawa’s reasoning, one can see that K&K’s analysis is not adequate. First, Adams and Aizawa’s reasoning does not rely on material constitution. They do not talk about objects but about processes as well (see, for example the fourth chapter “Cognitive Processes” of Adams & Aizawa, 2008). Second, Adams and Aizawa’s reasoning does not rest on any specific notion of constitution at all. It does not even rest on a clear understanding of what distinguishes causation from constitution. It simply relies on the logical structure of the underlying arguments (“X and Y are causally related. Therefore, X and Y are constitutively related”) and the intuitive judgements that, e.g., a heater does not constitute the extension of the bimetallic strip in a thermostat (Adams & Aizawa, 2001), or that sleep does not constitute tiredness (Krickel, 2020). Adams and Aizawa make use of the standard procedure for showing that an argument is logically invalid: find an argument with the same logical structure that clearly has a false conclusion (see, for example, Rosenberg, 1996). Their argument does not require any commitment to what constitution actually is. K&K’s strategy for avoiding the CC-fallacy is  the following: they avoid the CC-fallacy because they do not infer directly from the “X and Y are causally related” premise to “Therefore, X and Y are constitutively related”. Rather, K&K  add a premise: 1. X and Y are related by reciprocal causation. 2. Reciprocal causation is (diachronic) constitution. 3. Thus, X and Y are constitutively related.

11  Why Diachronic Constitution Won’t Help. Commentary on “Dissolving…

177

Clearly, this argument is deductively valid. However, the argument is problematic for three reasons. First, one problem is that the intuitive judgements that, e.g., sleep does not constitute tiredness although they are reciprocally causally coupled still hold. Of course, K&K might simply argue that intuitions are not relevant here and that tiredness indeed does constitute sleep (and vice versa). This indicates that diachronic constitution does not capture what one would intuitively want the concept of constitution to express—which is why the introduction of diachronic constitution might be seen as a change of topic. On a more optimistic interpretation, this shows that we need to go beyond intuitions and clarify what the notion of constitution is supposed to mean, generally as well as in the context of EC (see, e.g., my article in this volume). Second, another problem is that K&K’s argument does not capture the examples they use to illustrate their own account. The ant and the beach are reciprocally causally coupled when the ant moves on the sandy beach; the infant and the caregiver are reciprocally causally coupled when they regulate their emotions. But no one claims that the ant constitutes the beach, or vice versa; similarly, no one claims that the infant constitutes the caregiver, or vice versa. Rather, the claim is that the causally coupled items form a new cognitive (or emotional) system. However, the relation between the ant and the cognitive system and that between the infant and the emotional system is not that of reciprocal causation. It is unclear, whether K&K want to keep the claim that the relata of reciprocal causation are the relata of constitution, or whether they want to keep the claim that things that are reciprocally causally coupled together form cognitive/emotional systems. If they want to keep the first, their account makes odd claims about what constitutes what; if they want to keep the second, constitution cannot be reciprocal causation (I raise that worry in my article in this volume when discussing Menary’s account). The third problem is that the argument for premise (2) is not convincing. The argument for (2) seems to be the following: 4. Constitution in the context of EC relates processes. 5. Processes cannot be related synchronically. 6. Thus, constitution in the context of constitution must be diachronic. 7. Diachronic constitution just is reciprocal causation. The problematic premise here is premise (5). The claim seems to be that synchronous relations can hold only between point-like relata and that, therefore, processes cannot be related synchronically. However, it is not clear why this should be the case (see my article in this volume). A synchronous relation requires relata that occur at the same time—which may be a point in time or a time interval. Accounts relying on constitution as standardly understood (i.e., as synchronous) are not committed to an ontology of point-like entities. Processes can occur at the same time intervals even if they are continuous and not point-like. In a nutshell: K&K’s analysis of the justification of the CC-fallacy is misguided as its justification does not rely on material constitution. Furthermore, their own account avoids the CC-fallacy by adding a crucial premise, i.e., that constitution is reciprocal causation—which lacks argumentative support and leads to several inconsistencies.

178

B. Krickel

11.3 Mechanistic Constitution K&K discuss Gallagher’s application of mechanistic constitution to EC that relies on the interpretation of mechanistic constitution developed in (Krickel, 2018a).1 In my 2018-article, I defend what I call “causation-based constitutive relevance” (Krickel, 2018a, p. 64)2: (Causation-based CR) X’s ϕ-ing is constitutively relevant for S’s ψ -ing iff. (i) X’s ϕ-ing is a spatial EIO-part of S’s ψ-ing, (ii) there is a temporal EIO-part of S’s ψ -ing that is a cause of X’s ϕ-ing, and (iii) there is a temporal EIO-part of S’s ψ-ing that it is an effect of X’s ϕ-ing. The notion of a spatial- or temporal EIO-part need not worry us here (for an explanation see Krickel, 2018a, b). Roughly, for example, my neuron’s firing is a spatial EIO-part of my typing because the firing happens during the typing and the neuron is a part of me. My typing “t” is a temporal EIO-part of my typing because the typing of the “t” happens during the typing and in both cases the entity performing the activities is the same (me). Causation-based CR makes clear that while constitution is partly defined in terms of causation, it is not itself a causal relation. The relata of the relevant causal relations ((ii) and (iii)) are not the relata of mechanistic constitutive relevance. The relata of mechanistic constitutive relevance are X’s ϕ-ing and S’s ψ -ing; the relata of the causal relation are temporal EIO-parts of S’s ψ -ing and X’s ϕ-ing. Roughly, Causation-based CR expresses the idea that temporally extended things (in my terminology “entity-involving occurrents” or “EIOs”) are constitutively related if and only if they occur in the same space-time region and the temporal phases of the one cause later temporal phases of the other—if you want, you can describe this as “reciprocally causing” the temporal phases of each other. K&K formulate three objections against Gallagher’s application of my account of mechanistic constitution to EC. First, they doubt that the mutual manipulability account can be applied without conflating causation and constitution. As I have explained above, the interpretation of the mutual manipulability account in terms of causation-based constitutive relevance does not conflate causation and constitution.

 It is surprising that they do not discuss other authors who apply mechanistic constitution to EC, such as Kaplan (2012), Krickel (2020), or Abramova and Slors (2019). They discuss (Krickel, 2020) in a footnote (fn. 9)—but summarize it incorrectly. In my 2020 paper, I follow Hewitson et al. (2018) and adopt causal specificity as a further necessary criterion for constitutive relevance. The main point of my 2020-paper is that we still run into what I call the “challenge of trivial extendedness” (see also my article in this volume). 2  Note that as I argue in my 2020 paper, the condition of causal specificity must be added in order to account for several counterexamples. 1

11  Why Diachronic Constitution Won’t Help. Commentary on “Dissolving…

179

Second, K&K argue that the application of the mutual manipulability account implies that one cannot intervene into the ‘whole’ and that Gallagher is committed to metaphysical four-dimensionalism. It is not clear to me how the commitment to four-dimensionalism arises. I am not sure whether the consequence that there are no interventions into the whole is supposed to follow from the specificities of Gallaghers account or from the mechanistic account/my interpretation of it. According to defenders of the mutual manipulability account, to intervene into the phenomenon (i.e., the ‘whole’) is to intervene into the input of the mechanism (see e.g., Craver et  al. (2021) or it is to intervene into a temporal EIO-part of the phenomenon (Krickel, 2018a, b). It is not clear to me, why Gallagher couldn’t adopt one of these views. The third objection raised by K&K is that adopting the mechanistic account of constitution forces Gallagher to retain an atemporal, synchronic view of constitution—which according to K&K is problematic in the context of dynamical cognitive systems. While it is correct that the mechanistic account of constitution takes constitution to be distinct from causation and takes it to hold between relata that occur at the same time intervals, this objection is convincing only if it were true that the mechanistic account is incompatible with constitution in dynamical systems. This, however, is not the case. As I have argued in the previous section and in my article in this volume, processes can be synchronously related. My negative evaluation of K&K’s account and their arguments can be reformulated as a challenge: K&K must provide a convincing argument for why processes cannot be synchronously related, i.e., they must show that synchronous relations require point-like relata.

References Abramova, E., & Slors, M. (2019). Mechanistic explanation for enactive sociality. Phenomenology and the Cognitive Sciences, 18, 401–424. Adams, F., & Aizawa, K. (2001). The bounds of cognition. Philosophical Psychology, 14, 43–64. Adams, F., & Aizawa, K. (2008). The bounds of cognition. Blackwell Publishing Ltd.. Craver, C. F., Glennan, S., & Povich, M. (2021). Constitutive relevance & mutual manipulability revisited. Synthese, 199, 8807–8828. Hewitson, C. L., Kaplan, D. M., & Sutton, J. (2018). Yesterday the earwig, today man, tomorrow the earwig? Comparative Cognition & Behavior Reviews, 13, 25–30. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology and Philosophy, 27, 545–570. Kirchhoff, M.  D. (2015). Extended cognition & the causal-constitutive fallacy: In search for a diachronic and dynamical conception of constitution. Philosophy and Phenomenological Research, 90, 320–360. Kirchhoff, M. D. (2017). From mutual manipulation to cognitive extension: Challenges and implications. Phenomenology and the Cognitive Sciences, 16, 863–878.

180

B. Krickel

Kirchhoff, M. D., & Kiverstein, J. (2020). Attuning to the world: The diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11, 1966. Krickel, B. (2018a). Saving the mutual manipulability account of constitutive relevance. Studies in History and Philosophy of Science Part A, 68, 58–67. Krickel, B. (2018b). The mechanical world (Vol. 13). Springer International Publishing. Krickel, B. (2020). Extended cognition, the new mechanists’ mutual manipulability criterion, and the challenge of trivial extendedness. Mind & Language, 35, 539–561. Rosenberg, J.  F. (1996). The practice of philosophy  – A handbook for beginners (3rd ed.). Prentice Hall.

Chapter 12

Predictive Processing and Extended Consciousness: Why the Machinery of Consciousness Is (Probably) Still in the Head and the DEUTS Argument Won’t Let It Leak Outside Marco Facchin and Niccolò Negro Abstract  Recently, Kirchhoff and Kiverstein have argued that the extended consciousness thesis, namely the claim that the material vehicles of consciousness extend beyond our heads, is entirely compatible with, and actually mandated by, a correct interpretation of the predictive processing framework. To do so, they rely on a potent argument in favor of the extended consciousness thesis, namely the Dynamical Entanglement and Unique Temporal Signature (DEUTS) argument. Here, we will critically examine Kirchhoff and Kiverstein’s endeavor, arguing for the following three claims. First, we will claim that Kirchhoff and Kiverstein’s emphasis on culture and cultural practices does not help them substantiate the extended consciousness thesis. Secondly, we will argue that the way in which Kirchhoff and Kiverstein formalize the boundaries of a subject’s conscious mind is inadequate, as it yields conclusions running counter some of their assumptions. Lastly, we will argue that the DEUTS argument does not establish the extended consciousness thesis, as it licenses a phenomenal bloat objection which is exactly analogous to the “cognitive bloat” objection to the extended mind thesis. We will thus conclude that Kirchhoff and Kiverstein’s proposed marriage between the extended consciousness thesis and predictive processing fails, and that, contrary to a popular opinion, DEUTS is not a strong argument in favor of the extended consciousness thesis. Keywords  Predictive processing · Free-energy principle · Markov blankets · Extended consciousness · Sensorimotor Enactivism

M. Facchin (*) Linguistics & Philosophy IUSS Center, Scuola Universitaria Superiore Pavia, Pavia, Italy e-mail: [email protected] N. Negro Department of Philosophy, Monash University, Melbourne, VIC, Australia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_12

181

182

M. Facchin and N. Negro

12.1 Introduction The extended consciousness thesis, or consciousness vehicle externalism (CVE) claims that the material vehicles of a subject’s consciousness at least sometimes include the subject’s active body and/or some appropriate environmental props (Hurley, 2010; Vold, 2015). Notice that CVE is a claim concerning phenomenal consciousness, the elusive “what-it-is-like” of lived experiences (cf. Nagel, 1974). Sensorimotor enactivists endorse CVE. They claim that phenomenally conscious perception is achieved by an embodied agent by interacting sensomotorically with the environment; for instance, by saccading over a visual target so as to explore its visual profile. They argue that the phenomenal qualities of perception are determined by sensorimotor contingencies: law-like relations holding between bodily movements and changes in sensory stimulation (O’Regan & Noë, 2001a, b; Hurley & Noë, 2003; O’Regan, 2011). On their view, perceivers possess a tacit body of knowledge concerning such relations (called “sensorimotor mastery”) which they exert so as to enact their perceptual experiences; that is, perceptually explore the environment so as to reveal it phenomenally. The vehicles of perceptual phenomenology thus include, alongside patterns of neuronal activation, bodily acts of perceptual exploration guided by the perceiver sensorimotor mastery (Noë, 2004, 2009; Kiverstein & Farina, 2012; Pepper, 2014). This formulation of CVE has recently been put under significant pressure by the neurocomputational framework of predictive processing (PP). This is because PP operationalizes sensorimotor contingencies in purely neural terms, as expectations concerning the incoming inputs encoded in a generative model (e.g. Pezzulo et al., 2017; Baltieri & Buckley, 2019). The perceiver mastery of sensorimotor contingencies is thus “pushed inside the brain”, and ends up regulating only neuronal message-­ passing, leaving no reason to endorse CVE (Clark, 2012; Seth, 2014).1 Recently, Kirchhoff and Kiverstein (2019a, 2019b, 2020) set off to revise this dialectical situation, showing that the PP framework, properly understood, mandates CVE (Kirchhoff, 2018; Kirchhoff & Kiverstein, 2019a, pp. 57–59, 83–86). To do so, they carefully analyze PP, using it to formulate the most potent argument in favor of CVE in the sensorimotor enactivists’ arsenal; namely the Dynamical Entanglement and Unique Temporal Signature (DEUTS, see § 3 below) argument. Here, we diagnose three problems in Kirchhoff and Kiverstein’s proposal. First, their proposal aims to defend CVE appealing to cultural practices. Yet, cultural practices do little to support CVE. Secondly, we argue that their usage of Markov Blankets (more on which in §2) to formalize the boundaries of a subject’s conscious mind yields results that clash with some of their theoretical commitments. Lastly, we argue that the DEUTS argument generates a phenomenal bloat problem analogous to the “cognitive bloat” problem affecting the extended mind thesis (see Sprevak, 2009; Vold, 2021). We will thus conclude that Kirchhoff and Kiverstein’s  Importantly, when issues regarding CVE are left aside, PP and sensorimotor enactivism have a far less hostile relation both conceptually (e.g. Vázquez, 2020) and empirically (e.g. Laflaquiere, 2017; Leinweber et al., 2017). 1

12  Predictive Processing and Extended Consciousness: Why the Machinery…

183

endeavor fails to secure a happy marriage between CVE and PP, and that, contrary to a popular opinion (Clark, 2009, 2013; Kirchhoff & Kiverstein, 2019a), the DEUTS argument is not a strong argument in favor of CVE. Here is our plan. In the next two sections, we will sketch PP and Kirchhoff and Kiverstein’s position respectively. In section four, we will turn from exposition to criticism, articulating our three claims. A brief concluding paragraph will then close the paper.

12.2 Predictive Processing and the Free-Energy Principle: A Quick Introduction PP designates a family of neurocomputational frameworks based on predictive coding (Tani, 2016; Spratling, 2016, 2017). On its own, predictive coding is “just” a data-compression strategy (see Shi & Sun, 2008, Ch. 3). If one adopts it, one does no longer transmit the full signal one wishes to transmit (e.g. “I am wearing exclusively a shirt, jeans, panties, shoes and nothing else”), but only the way in which the signal deviates from some prior expectation (e.g. “I am not wearing socks”), typically saving bandwidth. Various PP algorithms build differently on this basic assumption. The most popular PP variant, which will be discussed here, (see Hohwy, 2013; Clark, 2016; Wiese & Metzinger, 2017), is also associated with the ambitious theoretical framework of the Free-energy Principle (Friston & Stephan, 2007; Hohwy, 2020). In particular, PP is taken to be a “process theory” for the Free-energy Principle: that is, a more or less detailed story spelling out how (some) systems abide by the Principle. Notice that, even in this variant, PP and the Principle are not the same thing. The Principle is compatible with other process theories (see Parr et al., 2022) and the falsity of PP (in any variant) does not entail the falsity of the Principle (assuming that the Principle is truth-apt; see Andrews, 2021). So, here we have two theoretical objects to introduce: PP and the Free-Energy Principle. Readers already familiar with them might wish to skip ahead to §3. The free-energy principle states that biological self-organization is centered around the avoidance of surprisal: an information-theoretic measure of the unexpectedness of sensory states, given a model (implicitly realized by the organism’s embodiment) of the sensory states compatible with the organism’s prolonged existence, which the organism should expect to occupy (Friston, 2012a, 2013). Organisms, however, cannot track surprisal directly. They can only track its upper bound, which is (variational) free-energy (Friston, 2009). Free-energy is an upper bound on surprisal because it can be understood as surprisal plus a second, always positive, quantity, which measures how much a system’s guesses concerning the causes of its sensory states are aligned with reality. This second quantity is the Kullback-Leibler Divergence (DKL).2  For the most up-to-date review of the formal aspects of the Free-energy Principle, see Parr et al. (2022) 2

184

M. Facchin and N. Negro

Surprisal is also the complement of model evidence (Friston, 2019, p. 177), minimizing it amounts to producing the evidence in favor of one’s model of one’s prolonged existence. Surprisal-minimizing systems are thus self-evidencing systems (Hohwy, 2016): systems striving to bring about the evidence favoring the hypothesis that they exist, thereby prolonging their existence. In the context of the free-energy principle, the relevant self-evidencing models are understood as the states enclosed by a Markov Blanket (Hohwy, 2017; Friston et al., 2020). Markov Blankets are formal boundaries that separate a biological system from its niche in a statistical sense, while allowing the two to causally interact (Friston, 2013; Friston et al., 2020).3 The interaction is allowed by the internal partition of the Markov Blanket: active states allow the organism to influence its niche, whereas sensory states allow the niche to influence the organism. In this way, the Markov Blanket allows a two-way causal interaction between organism and niche, enabling the coupling between the two (Hohwy, 2017; Kirchhoff & Kiverstein, 2019a, pp. 65–67). Importantly, according to the free-energy principle, Markov Blankets are multiple and nested, and can be found at each and every level of organization (Kirchhoff et al., 2018). A biological system such as a wolf, for instance, can be decomposed in biological sub-systems (e.g. cells) and can partake in larger biological systems (e.g. a pack), each busy minimizing free-energy (and thus, equipped with Markov Blankets of their own). The free-energy principle can be related to PP by noticing that free-energy can be equated (under some assumptions, see Buckley et  al., 2017) with prediction error: a well-known neural signal posited by predictive coding accounts of neural functioning (e.g. Rao & Ballard, 1999; Friston, 2005). In the context of PP, prediction error names the mismatch between the sensory states an agent expects or predicts and the actually received ones. On the view of the mind PP proposes, the primary function of brains is to minimize prediction errors, and cognition unfolds as a consequence of error minimization. Prediction error can be minimized in two ways (Hohwy, 2013; Clark, 2016). One is by revising the expectations, fitting them to the received sensory signals. This is perceptual recognition, which, in free-energy terms, corresponds to decreasing the DKL. Another way to minimize prediction error is by changing the received sensory states through movement, so as to bring about the expected ones. This is active inference, which, in free-energy terms, corresponds to a direct minimization of surprisal.

 There is now a vast philosophical literature on Markov Blankets, investigating how this conception of Markov Blankets is removed from the one relevant in machine learning and whether Markov Blankets should be conceived as objective features of biological systems or just as modeling tools (Andrews, 2021; Raja et al., 2021; Bruineberg et al., 2022). Here, we will assume for the sake of discussion that the realistic reading is correct, as it seems the one espoused by Kirchhoff and Kiverstein (2019a, 2019b). See also (Facchin, 2021, 2022; Menary & Gillett, 2020, 2022) for lengthy and sustained critical analyses of the role Markov Blankets may play in the vehicle externalism debate. 3

12  Predictive Processing and Extended Consciousness: Why the Machinery…

185

Importantly, prediction errors are always weighted based on their expected precision: roughly, an estimate of the signal to noise ratio of the sensory signals giving rise to prediction errors. In this way, sensory signals generated by highly noisy sensory states will be dampened, thereby allowing only informative prediction errors to dominate neural processing (Feldman & Friston, 2010). Active inference puts PP in direct contact with sensorimotor enactivism. This is because active inference is performed by predicting the sensory consequences of one’s movement, and then canceling out the error they generate through movement (Friston, 2011; Adams et al., 2013). Hence, on the account of action predictive processing offers, to perform a movement m an agent must first predict the sensory outcomes m, which are the sensory states the agent would encounter, were m performed. So, on the account of action PP offers, the agent must know how actions systematically impact the incoming sensory stream; that is, the agent must know the relevant sensorimotor contingencies, and exert their mastery by predicting a “desirable” stream of sensory inputs.4 Notice how this makes an agent’s mastery of sensorimotor contingencies a purely neural affair (Seth, 2014), pushing us towards an indirect and (vehicle) internalist view of perception (Wiese, 2018). If the perceiver’s mastery of sensorimotor contingencies is neurally realized, and sensorimotor contingencies only mediate the message passing in a neurally realized generative model, then perception naturally appears an internal affair. Notice, lastly, that PP is not a theory of consciousness (phenomenal or otherwise), even if it seems able to account for at least some structural aspects of consciousness and might be relevant for the empirical research on consciousness (see Seth & Hohwy, 2020; Hohwy, 2021; Seth, 2021; Seth & Bayne, 2022). Kirchhoff and Kiverstein, however, seem to take PP as a theory of consciousness. For instance, they write: Predictive processing tells us what the parts of the system must be doing such that when these parts are organised in the right way, they constitute consciousness. The parts of the system will include, for instance, components that perform predictions, error calculation, precision estimation, and so on. (Kirchhoff & Kiverstein, 2019a, p. 104).

We concede the point for the sake of argument. Crucially, however, Kirchhoff and Kiverstein hold that only temporally thick generative models qualify as consciousness-­supporting (Kirchhoff & Kiverstein, 2019a, pp.  106–108, see also Hobson & Friston, 2014). Here, we will assess the temporal thickness of generative models using the following heuristic: the more the model allows a system to make larger loops in the space of all its possible states, the temporally thicker it is (Friston, 2018, pp. 5–6). For instance, a model that allows a subject to celebrate her birthday on an annual basis is considerably thicker than the model of an agent that loops

 Notice that in more traditional theories of motor control such a role is played by forward models (also called motor emulators), which have already been used to operationalize sensorimotor contingencies (e.g. Maye & Engel, 2013). 4

186

M. Facchin and N. Negro

through its state space every few seconds. This way of proceeding seems to fit quite nicely with Kirchhoff and Kiverstein’s (2019a, pp. 55–57) understanding of models, according to which models are first and foremost entire organisms.5 But what are temporally thick models made of? What is the machinery constituting them? PP suggests that this machinery is squarely located within brains. Kirchhoff and Kiverstein disagree. Let us see why.

12.3 The DEUTS Argument, Twenty(Ish) Years Later DEUTS is often considered the strongest argument for CVE (Clark, 2009, 2013; Kirchhoff & Kiverstein, 2019a). Its origins trace back to (Hurley, 1998, Ch. 8). There, she makes a two-stepped argument. Here, we briefly present each step in its original variant, followed by Kirchhoff and Kiverstein’s PP rendition of it.

12.3.1 The First Step: Dynamical Entanglement The first step claims that cognitive processing sometimes weaves agent and environment in a single coupled system. This claim stems from dynamical approaches to cognition, according to which cognitive processing is not “sandwiched” between perception and action (Hurley, 2001), but rather constituted by cyclical sensorimotor interactions. These interactions are best explained using the formal tools of dynamical systems theory (e.g. Hurley, 1998; Chemero, 2009) which allow us to quantitatively model and predict them. This explanatory methodology, however, often forces one to model agent and environment as a single (non-decomposable) coupled system, whose behavior can be accounted for only by using its order parameters,6 and whose dynamics accounts for the production of cognitive outputs (Favela & Martin, 2017; Lamb & Chemero, 2018). Hence, in those cases, agent and environment form a single cognitive system, which vindicates a form of vehicle  There are other ways to formalize the temporal thickness of a model. One is simply that of observing its hierarchical structure, in which hierarchically higher layers generate predictions over longer timescales, and are thus temporally thicker than hierarchically lower ones. But this way of proceeding seems to apply only to hierarchically structured internal (brain-like) models (e.g. Tani, 2016, Ch. 9–10). Hence, it would beg the question against Kirchhoff and Kiverstein (2019a, p. 55) understanding of models. Another way to formalize the temporal thickness of a model may be in terms of expected free energy (Corcoran et  al., 2020): roughly, the free-energy expected in the future, after active inference has been performed. However, expected free-energy might not accurately model the free-energy expected in the future, and hence it might prove inadequate to capture the temporal thickness of a model (Millidge et al., 2021). We rely on Friston’s (2018) heuristic to avoid these problems. 6  An order parameter (or collective variable) is a variable describing the behavior of the components of the entire system. See (Kelso, 1995, Ch. 1 and 2) for a useful introduction. 5

12  Predictive Processing and Extended Consciousness: Why the Machinery…

187

externalism about cognition (Palermos, 2014; Kiverstein, 2018). How can this dynamical image of (extended) cognition be related to PP?

12.3.2 Dynamical Entanglement and Predictive Processing Kirchhoff and Kiverstein commence by noticing that prediction error minimization is a tool for surprisal avoidance. But surprisal can be avoided only through active inference7; that is, embodied action. Thus within PP, real, embodied action is central to cognition (Kirchhoff & Kiverstein, 2019a, pp. 57–59). Secondly, Kirchhoff and Kiverstein notice that albeit Markov Blankets statistically separate agent and environment, they also enable the coupling of the two (Kirchhoff & Kiverstein, 2019a, pp. 65–67; see also Fabry, 2017). This is due to the interplay of the active and sensory states that jointly constitute the blanket. Recall: active states influence sensory and external states, and are influenced by internal states. Conversely, sensory states influence active and internal states, and are influenced by external states.8 Thus, together, active and sensory states enable internal and external states to interlock in a two way interaction, which is a form of coupling. In third place, Kirchhoff and Kiverstein stress that Markov Blankets are not just multiple and nested, but also malleable and plastic. They do so by relying on Clark’s (2017) metamorphosis argument: Clark invites us to consider metamorphic insects, and the boundary (i.e. the Markov Blanket) separating them from the environment. As the insect undergoes the metamorphic process, it re-negotiates that boundary, shifting the set of states that separates it from the environment (trivially, the silk a cocoon is made of is not the exoskeleton of the caterpillar). Hence, Markov Blankets are not fixed: self-evidencing models such as metamorphic insects may undergo dramatic changes, thereby shifting the Blanket separating them from their niche. To this, Kirchhoff and Kiverstein add that, at least sometimes, these shifts can “push inside the Blanket” some environmental prop essential to the self-evidencing of an agent. This is the case of a spider and its web, or of a mildly amnesic patient and the notebook she uses to compensate for her amnesia (Kirchhoff & Kiverstein, 2019a, pp. 73–76, b). Consider this point in the light of the coupling Markov Blankets enable. If an agent can be dynamically entangled with an external resource, they form a single coupled dynamical system. And if it avoids surprisal (e.g. Bruineberg, Rietveld, et al., 2018b), it will be a free-energy/prediction error minimizing system in its own

 Notice that here we are dealing with surprisal. Free-energy can be minimized in other ways too, as indicated above. 8  Notice, importantly, that sensory and active states influence each other, and are thus coupled. This, on Kirchhoff and Kiverstein’s (2019a, p. 69) view, allows Markov Blankets to capture the idea of sensorimotor contingencies. 7

188

M. Facchin and N. Negro

right. So it will have its own Markov Blanket.9 In such a case, the coupled system will be identified through a “wider” Markov Blanket, encompassing “smaller” coupled Markov Blankets. The “wider” Markov Blanket will act as an order parameter on the “smaller” Blankets it contains, constraining10 their degrees of freedom and coalescing them in a single, possibly extended, system (Kirchhoff & Kiverstein, 2019a, pp. 80–81). In Kirchhoff and Kiverstein’s view, such a Blanket identifies the relevant cognitive machinery, namely, the relevant self-evidencing model engaged in prediction error/free-energy minimization. They hold that, by default, such a model encompasses the entire organism (Kircchoff and Kiverstein 2019a, pp. 55–57), but, as the metamorphosis argument purportedly shows, it can extend to incorporate the environmental props the system is coupled to, allowing the creation “on the spot” of extended free-energy/prediction error minimizing systems. In this way, Kirchhoff and Kiverstein resort to Markov Blankets to formalize the boundaries of (extended) cognitive systems. Here, we summarized Kirchhoff and Kiverstein’s PP inspired rendition of the first step of the DEUTS argument – the “Dynamical Entanglement” bit. Yet, thus far, the argument seems only an argument for vehicle externalism about cognition. Where does consciousness come into play? To answer this question, we need to look into the second step of the DEUTS argument  – the “Unique Temporal Signature” bit.

12.3.3 The Second Step: Unique Temporal Signature The idea behind the second step of the argument is that a careful consideration of the temporal development of our experiences (their “unique temporal signature”) shows that the dynamical avenue to vehicle externalism about cognition entails CVE. To claim so, Hurley (1998, Ch. 8) starts by granting the familiar vehicle internalist intuition that if the machinery of consciousness is purely neural, then keeping a subject’s neural state constant will keep the subject’s phenomenology constant, whatever the environment. Hence, if the machinery of consciousness is purely neural, two subjects can be neural and phenomenal duplicates without being environmental duplicates. To attack it, Hurley (1998, Ch.8) argues that two dynamically entangled subjects cannot be neural and phenomenal duplicates without being also environmental duplicates. To see why, consider a simplified rendering of one of Hurley’s (1998,  To be fair to Bruineberg et al. (2018b) we wish to point out that they do not make a similar conclusion about Markov Blankets. Still, it seems to us that such a conclusion follows, if one accepts that free-energy minimizing systems must be enclosed by Markov Blankets. 10  A possible worry here is that constraints are typically described as non-causal (e.g. Lange, 2017). Were constraints to be non-causal, Kirchhoff and Kiverstein’s coupling based argument would be in danger. However, it is possible to understand constraints as causal (cf. Ross, 2020), and that seems sufficient to rescue Kirchhoff and Kiverstein’s argument. 9

12  Predictive Processing and Extended Consciousness: Why the Machinery…

189

pp. 303–314) thought experiments. On earth, subject S is in an entirely white room, containing only S and a black ball on S’s right. On twin earth, S’s physical duplicate TS is in the exact same situation. Since S and TS are physical duplicates, it seems correct to say S and TS are experiencing the same thing: what it feels like to be in a white room with a black ball at one’s right. Here, S and TS are phenomenal, neural and environmental duplicates. But (and this is the vehicle internalist intuition) S and TS could be phenomenal duplicates just by being neural duplicates: we could switch the place of TS’s ball from right to left and insert in TS’s eyes left-to-right inverting lenses, so as to keep the visual input TS receives constant (Hurley, 1998, p. 304). In this case, it seems correct to say that TS and S will be in the same neural state, undergoing the same experience. Hurley (1998, p. 327) suggests this is possible only because S and TS are passive perceivers, and thus they are not coupled with the environment in any meaningful way. Were S and TS dynamically entangled with their (different) environments, they would cease to be phenomenal duplicates. Suppose, for instance, that S and TS try to touch the ball when they’re not environmental duplicates. They will both move their right arms towards the ball (which they both see at their right), causing their neural and phenomenal states to diverge. For S will touch the ball, whereas TS will not. So only S’n neural states will be modified by the reafferent signals. As a result, only S will experience what it is like to touch a ball. Thus, when taking into account the temporal trajectory of the experiences of two dynamically entangled subjects S and TS, phenomenal duplication fails. To duplicate the experience of the dynamical entangled S in TS, one needs to make S and TS environmental duplicates; that is, one has to remove the lenses from TS’s eyes and displace TS’s ball in its original position. Only in this case S’s temporally extended experience can be duplicated in TS. So, in order for the phenomenology of a dynamically entangled subject to be duplicated, it is not sufficient that the subject and their twin are neural duplicates. They also need to be environmental duplicates.11 But, according to Hurley, what needs to be the case in order for a phenomenal state to occur is the vehicle of that state (Hurley, 1998, pp. 330–331). And, in the example just considered, what needs to be the case for the relevant phenomenal states to occur includes at least some environmental factors. Hence, if a subject is dynamically entangled, environmental factors are vehicles of its consciousness (Hurley, 1998, pp. 330–335). Reflections on the unique temporal development of experiences thus show that dynamical entanglement entails CVE. Importantly, such a procedure to identify the external vehicle of consciousness is supposed to be discriminating (Hurley, 1998, pp. 330–331); that is, it is supposed to be able to tell apart genuine external vehicles of a subject’s consciousness from external factors merely impacting it.

 Importantly, the environmental duplication need not be total:only the environmental factors with which S is dynamically entangled need to be duplicated in TS.  Non-phenomenology-shaping aspects of the environment might still systematically differ. 11

190

M. Facchin and N. Negro

12.3.4 Unique Temporal Signature and Predictive Processing Kirchhoff and Kiverstein (2019a, pp. 36, 112–115; 2020, pp. 5–9) build essentially on the same point. Yet they would emphasize the role of the cultural environment in the constitution of one’s experience (Kirchhoff & Kiverstein, 2019a, Ch. 5 and 6; 2020). On the view Kirchhoff and Kiverstein offer, the machinery of consciousness has no fixed properties, and its properties are constantly transformed by the cultural practices12 the subject masters, and these properties have to be constantly negotiated by engaging with the surrounding cultural niche (Kirchhoff & Kiverstein, 2019a, Ch. 5; 2020). Subjects are thus enculturated, in the sense that their participation in cultural practices and their attunement to their cultural niches transforms them and alters their properties. Kirchhoff and Kiverstein hold that cultural practices are so important that they can be said, in a sense, to be vehicles of a subject’s phenomenal consciousness. Their claim seems to be that cultural practices play a constitutive role in determining the expected precision of the incoming sensory signal. That is, they contribute to creating a system’s expectation for certain very reliable streams of prediction error, which enable an agent to quickly deploy its own embodied skills to effectively cope with some relevant environmental contingency (Kirchhoff & Kiverstein, 2019a, pp. 94–100; 2020), thereby shaping one’s subjective experience. Kirchhoff and Kiverstein provide a variety of examples of this, ranging from phoneme recognition (see Roepstorff et  al., 2010, cited in Kirchhoff & Kiverstein, 2019a, pp. 99–100) to color discrimination (see Thierry et al., 2009, cited in Kirchhoff & Kiverstein, 2019a, p. 98). The example they discuss the most, however, is the culture shock Eva Hofmann felt when moving from Poland to Canada (Kirchhoff & Kiverstein, 2019a, pp. 110–112; 2020). They argue that to account for the sense of estrangement Eva felt when moving, we need to refer to her active engagement with the cultural practices of her native cultural (Polish) environment, and her inability to enact such practices in the new (Canadian) cultural environment. This makes Eva’s sensory states in her new cultural niche are surprisal-inducing, as they are not the ones she learned to predict. Her neural duplicate in the appropriate (Polish) environment would instead register no surprisal, courtesy of a compliant cultural environment (Kirchhoff & Kiverstein, 2019a, pp. 113–114). Hence, in Kirchhoff and Kiverstein view, the cultural practices Eva took part in and learned while in Poland qualify as vehicles of her phenomenal experience. Importantly, Kirchhoff and Kiverstein (2020, pp.  7–9)  To our knowledge, Kirchhoff and Kiverstein do not provide an explicit definition of cultural practices. However, it seems that they interpret cultural practices as relatively stable, socially regimented, ways of interaction, which induce a relatively stable sensory flow to which an agent can attune its expectations (cfr. Kirchhoff & Kiverstein, 2020). This way of understanding cultural practices might derive from Kiverstein’s past work on the skilled intentionality framework, which explicitly understands practices as “stable ways of doing things” (Rietveld & Kiverstein, 2014; see also Rietveld et al., 2018 for a presentation of the framework). Many thanks to a colleague for having pointed this out. 12

12  Predictive Processing and Extended Consciousness: Why the Machinery…

191

would add that these practices constitute Eva’s phenomenology diachronically. This means, roughly, that the culture shock Eva felt is not exclusively constituted by factors present in the here-and-now (i.e., the moment when the culture shock is experienced). It is also partially constituted by the factors present in Eva’s past that account for her experience now. The vehicles of Eva’s experience, diachronic constitution suggest, are not just spread out in space. They are also spread out in time. Crucially, Kirchhoff and Kiverstein argue that these culturally-laden modifications of conscious experience are not due to the acquisition of specific neural representations. Rather, they are due to the constant agent-environment interaction; and cultural practices should be seen as elements regulating the behavior of the agent-­ environment coupled system, like macroscopic order parameters organizing the behavior of a dynamical system over very long timescales (Kirchhoff & Kiverstein, 2020, p. 7).13 In their view, brains are not equipped with culturally learned representations; rather, they are nodes in a complex web of loopy causal relations, which are constantly transformed by the culturally regimented loops traversing them. We end our exposition of Kirchhoff and Kiverstein’s complex position here, owing to space limitations. We acknowledge that our summary is incomplete,14 and that it does not convey the entire depth of Kirchhoff and Kiverstein’s overall position. But it is now time for us to articulate our claims.

12.4 Extended Consciousness in Predictive Processing: Three Problems 12.4.1 Cultural Practices Do Not Seem to Support Consciousness Vehicle Externalism As sketched above, Kirchhoff and Kiverstein (2019a, 2020) take cultural practices to be material vehicles of a subject’s phenomenal consciousness: given the process ontology they assume (e. g. Kirchhoff, 2015), the vehicles of consciousness are not physical states but rather physical processes extended in time. Among these processes, there are cultural practices, which determine (at least partially) the expected precision of incoming error signals. Hence, they play a role in error minimization, thereby partially determining a subject’s phenomenally conscious states. Yet, the account of expected precision PP offers is purely neural. PP suggests that neurons reporting prediction errors with a high expected precision have their postsynaptic gain increased, so as to make their prediction error dominate subsequent

 Notice that, according to Kirchhoff and Kiverstein, this is exactly the role that the “wider” Markov Blanket should play in cases of dynamical entanglement. 14  We won’t, for example, discuss in details Kirchhoff and Kiverstein’s complex proposal of a diachronic account of constitution, and we have glossed over a variety of themes explored in Kirchhoff and Kiverstein’s book. 13

192

M. Facchin and N. Negro

processing. Conversely, neurons reporting prediction errors with low expected precision get silenced by decreasing their post synaptic gain. Other mechanisms contribute to precision estimation too: dopamine can be used to quickly shift the post synaptic gain of error neurons, and synchronization of error reporting neurons might further increase their ability to influence cortical processing (see Feldman & Friston, 2010; Friston, 2012b; Friston et al., 2012). Collectively, these mechanisms allow our brains to “attend” only the most significant prediction error streams, thereby simplifying the solution of various cognitive tasks. Yet these mechanisms seem to be neural mechanisms. No external vehicle is present in the picture of precision estimation PP proposes. And given that the mechanisms for precision estimation that PP identifies operate at the neuronal level, it is hard to see just how cultural practices could be integrated in such a picture.15 It could be argued that Kirchhoff and Kiverstein (2020, pp.  7–9) provide an answer to this question, suggesting that cultural practices act as macroscopic order parameters on the agent-environment system. Yet, we see two problems with this suggestion. First, order parameters in dynamical systems typically represent physical magnitudes, such as the temperature gradient controlling the rolling motions of a heated liquid (e.g. Kelso, 1995, pp. 6–8), the frequency of operation of a central pattern generator (e.g. Kiebel et al., 2009) or the relative angle phase of two oscillators (e.g. Schmidt & Richardson, 2008, p. 283). Hence, it is not clear whether the concept of an order parameter could be rightfully applied to cultural practices. Secondly, even allowing cultural practices to play a constraining role on neuronal dynamics similar to that of order parameters, it is hard to see in what sense such constraints would qualify as external vehicles of consciousness. Order parameters are not physical components of a dynamical system. They are mathematical values that perspicuously describe the state of a dynamical system in a dynamical model. Hence, even assuming that cultural practices can be cast as order parameters (or something analogous to them), they wouldn’t be components of the coupled dynamical system they describe (i.e. a subject dynamically entangled with its environment). Thus, they wouldn’t be vehicles inside that system.16 Perhaps focusing on a concrete case might clarify how cultural practices might function as external vehicles of subjective experience. Kirchhoff and Kiverstein (2019a, pp. 98–99) take the experiment performed by Thierry et al. (2009) as evidence of the fact that cultural practices support CVE. Since the experiment is clearly described, we will consider it at length.

 One could challenge this verdict, arguing that many external signs (e.g. traffic lights, stop signs, etc.) do play a role in settling our expectations about the precision of the incoming sensory signals, thereby functioning like a vehicle controlling our precision estimates (see Clark, 2016, Ch. 8 and 9). We doubt, however, that Kirchhoff and Kiverstein can leverage such a challenge against us. This is because this case for vehicle externalism about precision estimation is built upon a parity argument for vehicle externalism (cfr. Constant et  al., 2019a, pp.  16–18). But Kirchhoff and Kiverstein (2019a, Ch. 1) do not endorse such a parity principle based form of vehicle externalism, and are in fact quite critical of it. 16  Many thanks to two colleagues for discussion on these issues. 15

12  Predictive Processing and Extended Consciousness: Why the Machinery…

193

The study (Thierry et al., 2009) involved two groups of participants speaking two different native languages; namely English and Greek. Participants of both groups were required to perform an odball stimulus discrimination. The stimulus was a sequence of squares, and the oddball was a circle. Participants had to press a button as soon as they noticed the circle. Notice that the stimuli were presented at a fixed rate, hence participants could not influence the way stimuli were presented. The neural activity of each participant was captured through an EEG cap. Crucially, albeit participants were required to discriminate shapes, stimuli could also vary in color. In total, four colors were used: light and deep blue and light and deep green. Variations in color, however, were task irrelevant: participants had to press the button only if the oddball (circular) stimulus appeared, independently from its color. Now, there is a crucial difference between English and Greek native speakers when it comes to the colors. Whereas both English and Greek use a single word for green (whether deep or light), only English uses a single word for blue. Greek uses two: galázio (light blue) and ble (deep blue). Thierry and colleagues found that the early visual cortex of Greek (and only Greek) native speakers responded differently to the two task irrelevant shades of blue. Both English and Greek native speakers respond in the same way to the two shades of green. The researchers concluded that their data support the claim that color terminology can influence early visual processing. We really do not see how this experiment is supposed to bolster the case for CVE. Let us start with consciousness. The experiment does not establish that Greek and English native speakers experience color differently. The data speaks only of a “[…] relationship between native language and unconscious, preattentive color discrimination rather than simply conscious, overt color categorization” (Thierry et al., 2009, p. 4568; emphasis added). Moreover, whether early visual cortices qualify as neural correlates of consciousness is still a matter of debate (Chalmers, 2000; Blake et al., 2014; Koch et al., 2016), so it is at least in principle possible that color terminology only influences non-conscious visual processing. Now, suppose (for the sake of argument) that color terminology influences color phenomenology. Still, in the case at hand, it is not clear what the relevant external vehicles that should “extend” a subject’s color phenomenology  are. There are empirical studies that try to assess the impact of external objects on cognitive processing (e.g. Vallée-Tourangeau et al., 2016; Bocanegra, 2019), and these studies can be used to offer empirical support to vehicle externalism. But in these studies experimenters carefully describe the “external vehicles” involved, and what sort of effects they have on cognition. Thierry and colleagues do nothing of that sort. They neither describe the external vehicles that might “extend” their subject’s consciousness, nor they try to describe the impact of such vehicles on subjective experience. And rightfully so: as described in their paper, their experimental procedure just does not involve any external vehicle. Moreover, the experimental subjects do not appear to be dynamically entangled to anything in the environment. Participants were only required to press a button when the oddball was detected, and stimuli were presented every 800 ms and flashed for 200  ms, regardless of the subjects’ responses. There is no closed causal or

194

M. Facchin and N. Negro

sensorimotor loop knitting together participants and environment in a single system. There just seems to be no instance of continuous reciprocal causation or coupling (Palermos, 2014). And, in fact, contra dynamical views of cognition (Hurley, 2001) the participant’s task is easily decomposable in a linear sequence of input (stimulus reception) – cognition (discrimination) – action (eventual button pressing). Kirchhoff and Kiverstein (2020, pp. 5–7) might object that mastering a cultural practice is an essentially extended affair since a subject’s mastery of a cultural practice requires that the subject constantly attunes to her cultural environment, interacting with it.17 Hence, cultural practices cannot be fully internalized in a set of culturally-shaped neural representations: mastering them requires constant agent(cultural) environment interactions. But this argument appears unconvincing. In general, the fact that an agent’s mastery in a domain is achieved and maintained by means of constant environmental interactions does not pull the vehicles realizing that mastery out of the agent’s body. Amy’s mastery in weightlifting, for instance, has been achieved (and must constantly be maintained) by engaging with environmental props and tools (e.g. barbell weights). Yet, the material vehicles realizing her mastery (e.g. her muscle tone and her practical know-how on how to lift heavy weights) seem entirely internal to her body. And the same seems to hold for the mastery of cultural practices: one’s mastery of a language depends critically on one’s continued linguistic engagement with a certain linguistic environment. But the vehicles realizing one’s linguistic mastery (e.g. one’s Broca’s area) sit squarely in one’s skull. Importantly, vehicle internalism allows cultural practices to have an effect on perceptual consciousness. The case of phonetic recognition (i.e. how our perception of linguistic stimuli changes based on the languages we know; see Roepstorff et al., 2010) persuasively exemplifies the effects of culture on our phenomenology. Indeed, the theoretical apparatus of PP seems almost ideally suited to account for such effects (see Lupyan et  al., 2020 for a recent review). This is because PP heavily stresses the role of prior expectations (both about the incoming sensory inputs and their precision) in perception, and in particular in determining perceptual content. Hence, if cultural practices tailor a subject’s estimate of precision, and these estimates impact (or concur to impact) a subject’s perceptual content, then cultural practices can impact the content of a subject’s consciousness. Notice that a form of content externalism would be broadly consonant with Kirchhoff and Kiverstein’s overall position on cultural practices. As noted by Clark and Chalmers (1998), content (as opposed to vehicle) externalism tends to stress factors lying in a subject’s past history, such as the subject’s embedding in a linguistic community (e.g. Putnam, 1975). And these seem to be the factors that Kirchhoff and Kiverstein emphasize when discussing the case of Eva’s culture shock and the diachronic constitution of her experience (see Kirchhoff & Kiverstein, 2019a, pp. 110–112; 2020, pp. 9–10). Moreover, in at least some cases (e.g. Kirchhoff &

 Similar arguments for the essential extendedness of cognitive capacities are provided in (Noë, 2004, pp. 110–115) and (Hurley, 2010, pp. 138–143). 17

12  Predictive Processing and Extended Consciousness: Why the Machinery…

195

Kiverstein, 2019a, p.  111) they speak of cultural practices as settling our prior expectations. But, on the face of it, expectations are contentful states, whose content might be at least partially determined by some relevant cultural practice. For instance, a person born in a British English speaking cultural niche might expect to read “colour” instead of “color”. So, perhaps Kirchhoff and Kiverstein’s emphasis on cultural practices can vindicate (or motivate) some form of externalism about perceptual content. However, content externalism does not entail CVE. To begin with, contents and vehicles should not be conflated; and, in fact, their conflation is in general fallacious (Dennett, 1991; Hurley, 1998). Secondly, content and vehicle externalism are logically independent, and do not entail each other (Hurley, 2010; Rowlands, 2020). The fact that conscious content is (perhaps) determined by extraneural factors does not, in and by itself, entail that the vehicle of said content is extraneural.

12.4.2 The Markov Blankets of the Conscious Mind In the PP literature, the issues concerning vehicle externalism are typically framed in terms of Markov Blankets. Given that everyone agrees that Markov Blankets are multiple and nested, the relevant issues concern how to determine which Markov Blanket delineates the boundary of a subject’s mind (e.g. Hohwy, 2016, 2017; Clark, 2017). Kirchhoff and Kiverstein propose a crisp answer: the relevant Markov Blanket identifying the vehicles of a subject’s phenomenally conscious mind is the “wider” Markov Blanket surrounding “smaller” coupled Markov Blankets and acting as an order parameter on their synchronization (Kirchhoff & Kiverstein, 2019a, pp. 78–81). This is because the self-evidencing behavior of the “wider” Blanket constraints and determines the self-evidencing behavior of the smaller Blankets, forming “a prediction error minimizing whole” (Kirchhoff & Kiverstein, 2019a, p. 81). Here, we wish to argue that this way of identifying the Markov Blanket surrounding the vehicle of a subject’s phenomenal consciousness leads to very unpalatable results, which also clash with Kirchhoff and Kiverstein’s claim that only temporally thick generative models can support phenomenally conscious mental states. To see why, consider first the relevant notion of coupling at play. In the literature on Markov Blankets, the relevant notion of coupling is typically operationalized in terms of generalized synchrony (e.g. Friston, 2013; Bruineberg & Rietveld, 2014; Bruineberg, Kiverstein, & Rietveld, 2018a; Kirchhoff & Kiverstein, 2019a, pp. 108–110; 2020, fn. 3; Palacios et al., 2019).18 Generalized synchrony refers to a  This holds true also in some dynamical models (e.g. Schmidt & Richardson, 2008; Chemero, 2009, pp. 85–98). Intriguingly, (Kirchhoff et al., 2018) seem to point to a different, more demanding, notion of coupling. However, as far as we can see, such a notion of coupling is still lacking an operationalization. 18

196

M. Facchin and N. Negro

physical relation holding between a set of oscillators. If two (or more) oscillators interact (and influence each other’s oscillations), they will eventually start to oscillate either in phase (i.e. with the same rate) or in anti-phase (i.e. with opposite rates). When one of these two conditions is reached, then generalized synchrony among the oscillators has been established, and the two oscillators can be considered as a single coupled system. Secondly, consider the (seemingly well-established) claim that any two PP systems busy modelling each other rapidly fall into generalized synchrony (e.g. Friston & Frith, 2015a, 2015b; Palacios et al., 2019). The reason as to why this is the case can be clearly expressed as follows. Suppose that two agents (A and B) partake in a turn based activity, and suppose that they are both busy predicting what the other agent will do in its own turn. Now, when it’s A’s turn, B will try to predict the sensory consequences of A’s moves. But, and this is the crucial point, A is predicting them too. For, if PP is correct, in order for A to take a move, A is bound to engage in active inference; and it is thus forced to predict the sensory consequences of its own moves. This is why the activity of two mutually predicting PP systems will tend to synchronize. Different sources of empirical evidence reinforce the point, suggesting that interacting subjects tend to synchronize in a number of ways (see Wheatley et al., 2012; Coey et al., 2012; Tognoli et al., 2020 for reviews). Behavioral experiments suggest that interacting subjects spontaneously synchronize their limb movements (e.g. Schmidt et al., 1990; Richardson et al., 2005; see also Schmidt & Richardson, 2008 for a review), postural sway (Shockley et al., 2003; Goodman et al., 2005; Richardson et  al., 2007) and autonomic responses (Kang & Wheatley, 2017). More recently, neuroimaging experiments deploying hyperscanning techniques have highlighted that when two subjects interact, their neuronal activity synchronizes in various regions of interest relevant for social cognition (e.g. Stephens et al., 2010; Liu et al., 2015; Jiang et al., 2015; Liu et al., 2016; see Valencia & Froese, 2020 for a nice review), thereby providing strong evidence in favor of the claim that the neural activity of interacting subjects synchronizes. Hence, interacting subjects are coupled (according to the relevant notion of coupling at play here) at multiple levels. Consider now the computational simulations presented in (Palacios et al., 2020) and (Friston et al., 2015). If correct, these simulations show that when two or more free-energy minimizing systems interact with each other, they naturally tend to form a wider system, with its own Markov Blanket, provided the interacting (i.e. “smaller”) systems have at least some prior expectation in common. But human subjects surely share at least some priors. Some of those are prior expectations regarding the statistics of the environment, to which all perceptual systems must be attuned (Orlandi, 2014). For instance, humans expect natural light to illuminate objects from above and slightly on the left (e.g. Mamassian et  al., 2002). Other shared priors regard our bodily dynamics, and the ways in which we can behave to achieve our goals (e.g. Kilner et al., 2007, 2011; Donnarumma et al., 2017). Consider further culturally established prior expectations. These practices are said to establish regimes of shared expectations among the members of a culture (Constant et al., 2019b; Kirchhoff & Kiverstein, 2019a, Ch. 5; 2020). Taken together, all these threads of evidence strongly suggest that humans have at least some

12  Predictive Processing and Extended Consciousness: Why the Machinery…

197

common priors. Hence, if the results of the simulations provided by Friston et al. (2015) and Palacios et  al. (2020) are correct, the interaction of human subjects (which, if PP is correct, are free-energy minimizing systems) will naturally let a new system, with a “wider” Markov Blanket, emerge. It thus appears that, anytime two subjects interact, they can be modeled as two synchronized Markov Blankets surrounded by a “wider” Blanket. And the internal dynamic of such a “wider” seems to act as an order parameter on the dynamics of the “smaller” Blankets it contains, after all, in many cases of synchronization, the order parameter is the relative phase of the two oscillators (e.g. Haken et al., 1985; Kelso, 1995), which seems precisely what the “wider” Blanket enshrouds. Notice that such a wider Blanket seems precisely the Markov Blanket that, according to Kirchhoff and Kiverstein, identifies the vehicles of a subject’s phenomenal consciousness. Hence, if Kirchhoff and Kiverstein’s Markov Blanket proposal on how to identify the vehicles of a subject’s phenomenology is accepted, it seems that interacting subjects end up being vehicles of each other’s phenomenology. This would surely strike many as intuitively unappealing. But intuitively unappealing claims are not necessarily wrong. Thus, the fact that Kirchhoff and Kiverstein’s defense of CVE ends up delivering intuitively unappealing results is not necessarily a problem for them. What is a problem for them, however, is that such a “multi-subject” Markov Blanket does not appear to identify a temporally thick model.19 Yet, Kirchhoff and Kiverstein claim that only temporally thick models can be vehicles of phenomenal consciousness (Kirchhoff & Kiverstein, 2019a, pp. 53, 106–107). Hence, the way in which Kirchhoff and Kiverstein resort to Markov Blankets to individuate the boundaries of the conscious mind clashes with their theoretical commitments concerning the fact that only temporally thick models can give rise to subjective experience. So, either the way in which Kirchhoff and Kiverstein resort to Markov Blankets to identify the boundaries of a subject consciousness should be abandoned, or it is false that only temporally thick models can be vehicles of consciousness. At any rate, Kirchhoff and Kiverstein’s position on the matter seems to lead to an inconsistency. Something must go.20 Why do we claim that the multi-subject model identified by the “wider” Blanket is not temporally thick? Recall the heuristic proposed by Friston (2018, pp. 5–6) to determine the temporal thickness of a model it is sufficient to consider the time lapsed between successive “visits” to some particular states. The longer that time,

 Recall that, in this context, “model” simply refers to the internal states of a Markov Blanket.  One might object that Kirchhoff and Kiverstein resort to Markov Blankets only to formalize the boundaries of the mind, which need not coincide with the boundaries of the conscious mind. Yet, Kirchhoff and Kiverstein (2019a, p.  104) take the machinery PP describes to be the physical machinery of consciousness. And Markov Blankets presumably individuate the boundaries of that machinery. It is also worth noting that Kirchhoff and Kiverstein (2019a, pp.  62–63) introduce Markov Blankets to deal with Chalmers’s (2019) argument against CVE. It thus seems natural to think that, in Kirchhoff and Kiverstein’s view, Markov Blankets should identify the boundaries of the conscious mind. 19 20

198

M. Facchin and N. Negro

the thicker the model. A bacterium, for instance, might revisit a state x every half an hour. Its model is thus significantly shallower than the model a typical human possesses, given that a typical human visits certain states only once a year (e.g. throwing a party for one’s own birthday). A temporally thick generative model loops through its state space; and the temporal trajectory of that loop is an indicator of the temporal thickness of the model (or so Friston suggests). And if a model traces no loop in its state-space, then it is not temporally thick. But, and we believe the following point is relatively uncontroversial, not all human interactions have the required loopy structure. Surely some human interactions have it, a group of friends can meet, say, once every year to commemorate some particular event. But many other human interactions are one-shot interactions. Consider, for instance, an applicant’s interaction with the interviewer during a job interview. Both the interviewer and the applicant share many prior expectations, and both are prediction-error minimizing systems. So, if the simulations and the empirical evidence discussed above are correct, it seems that their interaction will tend to make them synchronize in various ways; and there will be a Markov Blanket “surrounding” them both. But the model identified through such a Blanket will not re-­ visit any of the states it happens to visit in the future  – typically, applicants and interviewers do not periodically meet to re-enact job interviews. So, the model they jointly instantiate during the interview will not be, according to the relevant heuristic Friston proposes, temporally thick. The same, it seems to us, holds true even if a more regimented notion of temporal thickness is deployed. For instance, temporal thickness is often understood in terms of the model’s hierarchical structure, in which hierarchically higher levels predict (and postdict) the incoming inputs at a higher temporal scale (Tani, 2016, Ch. 9–10; Friston et al., 2017). It is not at all clear whether multi-subject interactions have the required hierarchical structure, or in which sense two interacting subjects compose a functional hierarchy.

12.4.3 DEUTS and the Phenomenal Bloat One popular objection to vehicle externalism about cognition is the so-called “cognitive bloat” objection (Sprevak, 2009; Rowlands, 2010, 2020; Allen-Hermanson, 2013). The objection basically points out that the criteria vehicle externalism offers to identify external vehicles are too easily satisfied, thereby implying that all sorts of things are vehicles of a subject’s cognitive state. This is undesirable (if not a reductio of vehicle externalism) because it disrupts the explanatory and scientific credential of vehicle externalism. An unruly amount of factors plays a causal-­ explanatory role in accounting for an agent’s cognitive performance, and surely not all these factors are vehicles of cognition – and defenders of vehicle externalism do not wish to claim otherwise (Clark, 2008, pp.  76–81). Here, we claim that the DEUTS argument licenses a similar “bloat” worry when it comes to the vehicles of a subject’s phenomenal consciousness (see also Vold, 2021).

12  Predictive Processing and Extended Consciousness: Why the Machinery…

199

To see why, recall the general structure of the DEUTS argument. The first step is a commitment to dynamicism: cognitive processes are often constituted by agent-­ environment sensorimotor interactions. The best way to explain these interactions is through the tools of dynamical systems theory. Once these tools are deployed, one is often forced to model the agent and the environment as a single coupled system, whose joint behavior accounts for the production of cognitive outputs. Hence, cognitive outputs are produced by an “extended” coupled system; just as cognitive vehicle externalism requires (Chemero, 2009; Palermos, 2014; Lamb & Chemero, 2018; Kiverstein, 2018). The second step consists in showing that dynamically entangled subjects cannot be phenomenal duplicates only by being neural duplicates. If one closely scrutinizes how the experience of a dynamically entangled subject evolves over time, one notices that it is necessary, in order for a given temporally sequence of phenomenal states to be experienced, that agent and environment interact in a certain way; and thus that certain environmental features qualify as external vehicles of experience (Hurley, 1998, Ch. 8; Kirchhoff & Kiverstein, 2019a, p. 36). Crucially, this line of thought is supposed to provide a discriminating way to appeal to the causal spread of cognitive processing in order to establish CVE (Hurley, 1998, p. 330). Every agent is a node in a massive causal network connecting it to an unruly manifold of environmental features (e.g. the oxygen an agent breathes), not all of which are putative external constituents of the agent’s phenomenal machinery. To make her account discriminate between mere environmental causes and genuine external vehicles, Hurley (1998, pp. 329–332) reasoned as follows: vehicles explain the obtaining of any particular mental state. The tokening of an appropriate vehicle is what in virtue of which each and every mental state obtains. Hence, what needs to obtain in order for a phenomenal state to obtain is the vehicle of that state. But not every environmental feature needs to obtain in order for a phenomenal state to obtain. So the argument discriminates between putative external constituents and environmental factors merely causally impacting upon a subject’s consciousness. We believe that Hurley’s line of thought is not discriminatory enough. With enough ingenuity, one can always cook up a scenario in which the DEUTS argument ends up counting the oxygen in the atmosphere or the paint on the walls one is seeing (or whatever) as vehicles of a subject’s phenomenal experience. Or so, at least, we contend. Consider the following twist on Hurley’s thought experiment. Suppose that a subject S and S’s twin TS are in entirely white rooms, both with a ball at their right. Suppose that the layout of their environments are identical. The only difference is the amount of oxygen21 present in the rooms: whereas S is in a normally oxygenated environment, TS isn’t. In fact, there is no oxygen in TS’s environment. Lastly, let both S and TS interact with the balls, so as to get dynamically entangled with the environment.

 We chose this example because Kirchhoff and Kiverstein (2020, p. 12) clearly state that they do not wish to consider oxygen as a constituent of the phenomenal machinery. 21

200

M. Facchin and N. Negro

It seems clear that, given the proposed setup, the phenomenology experienced by S and TS, as well as their neural states, will diverge drastically. Whereas S will feel what it is like to play with a ball, TS will feel what it is like to choke to death. Whereas S’s brain will instantiate rich patterns of activity, TS’s brain will not (at least, not for long). Hence, we cannot duplicate S’s temporally extended phenomenology in TS. But, as Hurley (1998, p. 331) writes in regard to mental states: “if not duplicable, then not the whole vehicle”. Hence, the entire vehicle of S’s phenomenology has not been duplicated in TS’s scenario. But, by stipulation, the only difference between S and TS’s environments is the presence of oxygen. And surely, were oxygen present in TS’s room, TS too would feel what it is like to play with the ball: adding oxygen would thus allow us to duplicate S’s experience in TS. But then the DEUTS argument forces us to conclude that oxygen is a constituent of S’s phenomenal machinery, something that surely Kirchhoff and Kiverstein (2020, p. 12), as well as many other philosophers, wish to deny. Similar unappealing results proliferate easily. Consider what would happen if the color of the walls in S and TS’s rooms were different. Or if S and TS’s balls had a different color. Or if TS were made to wear medieval armor. Or if the rooms had different light bulbs emitting differently colored light. It seems correct to say that, in all those cases, the temporally extended phenomenology of S (and the way in which S’s neural states evolve through time) could not be duplicated in TS. And it seems equally correct to say that, in all those cases, eliminating the environmental feature acting as a difference-maker in between the two cases would allow S’s phenomenology to be duplicated. But if this is correct, then a systematic application of the DEUTS argument seemingly leads us to conclude that any (or almost any) environmental parameter contributing to the temporal unfolding of a dynamically entangled subject’s experience is a vehicle of that experience. This, it seems to us, is an unwelcome result, for it suggests that the DEUTS argument leaves CVE open to a nasty phenomenal bloat objection. Perhaps the problem could be avoided by supplementing DEUTS with some further criteria enabling us to discriminate, among the candidate vehicles DEUTS identifies, genuine vehicles of consciousness from spurious ones. And, if we understand them correctly, Kirchhoff and Kiverstein (2019b, 2020) propose two such criteria. We examine them in turn. The first criterion is proposed in Kirchhoff and Kiverstein (2020, pp. 11–12). The criterion revolves around counterfactual manipulations. To understand it, recall first that free-energy (which, under simplificatory assumptions, corresponds to prediction error) is the sum of two quantities: the surprisal of a sensory state and the DKL; where the DKL measures of how much the probability distributions encoded in an agent’s expectations differ from the actual probability distributions defined over environmental causes. Provided this, Kirchhoff and Kiverstein suggest that we should identify as external vehicles of a subject’s consciousness only the elements upon which a counterfactual intervention would change the subject’s DKL and thus the subject’s phenomenology. In their view, this simple test is sufficient to tell apart the external

12  Predictive Processing and Extended Consciousness: Why the Machinery…

201

factors which are part of a subject’s phenomenal machinery from the ones that merely causally interact with that machinery (Kirchhoff & Kiverstein, 2020, p. 12).22 We must confess that we do not see how this simple test can help Kirchhoff and Kiverstein’s cause, as changes in the DKL are typically associated with perceptual inference. If a subject correctly determines the external cause of the incoming sensory signals, the subject’s DKL will be small. Conversely, if a subject mis-infers the cause of the incoming signal, the subject’s DKL will be sizable. If this is correct, it is obvious that any intervention on an external cause of the sensory signal will change a subject’s DKL, and presumably the subject’s phenomenology. Notice that such interventions will alter the DKL of active and passive perceivers alike: even if a subject is “perfectly still”, intervening on the environmental causes of their sensory inputs will alter the subject’s DKL. Thus, the criterion proposed by Kirchhoff and Kiverstein (2020) seems even less discriminating than the original DEUTS argument. For, according to the DEUTS argument, something qualifies as an external vehicle of consciousness only if a subject is dynamically coupled to it. But this requirement is not present in Kirchhoff and Kiverstein’s (2020) criterion. Hence, it is more liberal than the original DEUTS argument, and thus cannot be used to solve the liberality problems afflicting DEUTS. Maybe, then, the second criterion will fare better. It is proposed in Kirchhoff and Kiverstein (2019b, pp. 16–18).23 The idea seems to be the following: an external candidate vehicle of consciousness really qualifies as a vehicle only if it contributes to an agent’s free-energy/prediction error minimization over time. As they write: The self-evidencing nature of biological agents blocks the threat from cognitive bloat. External resources form a part of an agent’s mind when they are poised to play a part in the processes of active inference that keep surprise to a minimum over time (i.e. that minimise free-energy). (Kirchhoff & Kiverstein, 2019b, p. 17)

We do not think that this effectively blocks the threat bloat argument poses. To see why, consider interoceptive active inference. If PP is correct, brains are busy predicting the incoming input in all modalities. Importantly, this means that brains will not just try to guess the incoming sensory signal in the exteroceptive modalities (bluntly put, the five senses traditionally understood); they will also try to guess the proprioceptive signals (i.e. the “sense” of kinesthesia and self-movement) and the interoceptive signals (i.e. the “sense” of one internal bodily state). Albeit typically

 Kaplan’s (2012) mutual manipulability criterion is another criterion that relies on counterfactual interventions to tell apart the genuine constituents of a subject’s mental machinery from factors that merely causally impacing it. Importantly, however, Kaplan’s criterion requires at least two counterfactual interventions: a “bottom-up” intervention on the putative vehicle and a “top-down” intervention on the relevant (allegedly extended) cognitive phenomenon. Kirchhoff and Kiverstein, in contrast, seem only to require “bottom up” interventions on the putative vehicle. So, Kirchhoff and Kiverstein’s criterion is distinct from Kaplan’s. 23  To be fair, in that paper Kirchhoff and Kiverstein do not deal with CVE directly, but only with cognitive vehicle externalism. However, since Kirchhoff and Kiverstein (2019a, p. 104) take the PP explanations of cognitive phenomena to be also an account of phenomenal consciousness, the argument they offer to block the cognitive bloat should also block the consciousness bloat. 22

202

M. Facchin and N. Negro

associated with emotional responses (e.g. Seth, 2013; Pezzulo, 2014; Seth & Friston, 2016) prediction of interoceptive signals is functionally on a par with the prediction of exteroceptive and proprioceptive signals.24 There are thus two general ways to minimize error relative to interoceptive predictions: changing the predictions to make them fit the incoming interoceptive signal or changing the incoming interoceptive signal to make it fit the predictions. The latters is interoceptive active inference. A concrete example of interoceptive active inference occurs when humans try to reduce the error relative to the prediction of their bodily temperature being around 36.6° (Bruineberg, 2017, p. 3). One way to reduce that prediction error is to wear appropriate clothes, given the environmental temperature (e.g. wearing a sweater during the winter). But, most times, humans do wear appropriate clothes, given the environmental temperature. So it seems that clothes are well poised to play the desired error minimizing role over time, just as Kirchhoff and Kiverstein ask. It thus seems correct to say that clothes are part of the machinery that minimizes (interoceptive) prediction error on average and in the long run. And given that Kirchhoff and Kiverstein take that machinery to be the machinery of consciousness (Kirchhoff & Kiverstein, 2019a, p.  104; 2020), it seems correct to conclude that, on their account, clothes are cogs in our phenomenal machinery. This, to us, seems sufficient to conclude that the consciousness bloat objection is not avoided.25 Notice, further, that focusing on the external vehicles that minimize surprise overtime seems to clash with Clark’s (2017) metamorphosis argument, upon which Kirchhoff and Kiverstein rely. The reason is this: presumably, the question whether some candidate vehicle plays an error/free-energy minimizing role overtime has a crisp yes or no answer. Either something plays such a role often enough, or it doesn’t. Hence, if the suggestion put forth by Kirchhoff and Kiverstein (2019b) is correct, an agent’s Markov Blanket is somewhat rigidly fixed. Yet, the metamorphosis argument stresses the plasticity of an agent’s Markov Blanket, suggesting that it can easily be altered. We thus conclude Kirchhoff and Kiverstein additional criteria do not succeed in rescuing the DEUTS argument.

12.5 Conclusion In this paper, we have examined some aspects of Kirchhoff and Kiverstein’s DEUTS-based marriage of PP and CVE, arguing that Kirchhoff and Kiverstein’s position is susceptible to a nasty phenomenal bloat objection, and that their emphasis on cultural practices does not contribute to establishing the truth of CVE.  So much so, that there can be interoceptive perceptual illusions; see (Iodice et al., 2019) for a nice example. 25  Facchin (2021, pp. 15–24) expands on this line of argument in the context of vehicle externalism about cognition, and deflects several intuitive objections to it. 24

12  Predictive Processing and Extended Consciousness: Why the Machinery…

203

Importantly, if the arguments we have presented here are correct, then, contrary to a popular opinion (e.g. Clark, 2009, 2013; Kirchhoff & Kiverstein, 2019a), DEUTS is not a strong argument in favor of CVE, as it is extremely susceptible to the phenomenal bloat objection, which is inbuilt in its argumentative fabric. Notice that we are not implying that CVE is false. There are other arguments in favor of CVE (e.g. Vold, 2015) which might succeed where DEUTS fails. But, thus far, they have received very little attention: DEUTS has always been the focus of the debate on CVE. We thus suggest that time is ripe to put DEUTS in retirement, and find some new argument in favor of CVE.

References Adams, R. A., Shipp, S., & Friston, K. J. (2013). Predictions, not commands. Active inference in the motor cortex. Brain Structure and Function, 218(3), 611–643. Allen-Hermanson, S. (2013). Superdupersizing the mind: Extended cognition and the persistence of cognitive bloat. Philosophical Studies, 164(3), 791–806. Andrews, M. (2021). The math is not the territory. Biology and Philosophy, 36(6), 1–19. Baltieri, M., & Buckley, C.  L. (2019). Generative models as parsimonious descriptions of sensorimotor loops. Behavioral and Brain Sciences, 42, e218. https://doi.org/10.1017/ S0140525X19001353 Blake, R., Brascamp, J., & Heeger, D. J. (2014). Can binocular rivalry reveal the neural correlates of consciousness? Philosophical Transactions of the Royal Society B: Biological Sciences, 369(1641), 20130211. Bocanegra, B.  R. (2019). Intelligent problem solvers externalize cognitive operations. Nature Human Behavior, 3(2), 136–142. Bruineberg, J. (2017). Active inference and the primacy of the “I can”. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing (Vol. 5, pp. 1–18). The MIND Group. Bruineberg, J., & Rietveld, E. (2014). Self-organization, free-energy minimization, and an optimal grip on a field of affordances. Frontiers in Human Neuroscience, 8, 599. Bruineberg, J., Kiverstein, K., & Rietveld, E. (2018a). The anticipating brain is not a scientist: The free-energy principle from an ecological-enactive perspective. Synthese, 195(6), 2417–2444. Bruineberg, J., Rietveld, E., Parr, T., van Maanen, L., & Friston, K. J. (2018b). Free-energy minimization in joint agent environment systems: A niche construction perspective. Journal of Theoretical Biology, 455, 161–178. Bruineberg, J., Dolega, K., Dewhurst, J., & Baltieri, M. (2022). The emperor’s new Markov blankets. Behavioral and Brain Sciences, 45(183), 1–76. Buckley, C.  L., Kim, C.  S., McGregor, S., & Seth, A.  K. (2017). The free-energy principle for action and perception: A mathematical review. Journal of Mathematical Psychology, 81, 55–79. Chalmers, D. J. (2000). What is a neural correlate of consciousness? In T. Metzinger (Ed.), Neural correlates of consciousness (pp. 17–39). MIT Press. Chalmers, D. (2019). Extended cognition and extended consciousness. In M. Colombo, E. Irvine, & M. Stapleton (Eds.), Andy Clark and his critics (pp. 7–20). Oxford University Press. Chemero, A. (2009). Radical embodied cognitive science. The MIT Press. Clark, A. (2008). Supersizing the Mind. Oxford University Press. Clark, A. (2009). Spreading the joy? Why the machinery of consciousness is (probably) still in the head. Mind, 118(472), 963–993. Clark, A. (2012). Dreaming the whole cat. Mind, 121(483), 753–771. Clark, A. (2013). Mindware (2nd ed.). Oxford University Press. Clark, A. (2016). Surfing uncertainty. Oxford University press.

204

M. Facchin and N. Negro

Clark, A. (2017). How to knit your Markov blanket. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing (Vol. 3, pp. 1–19). The MIND Group. Clark, A., & Chalmers, D. J. (1998). The extended mind. Analysis, 58(1), 7–19. Coey, C. A., Varlet, M., & Richardson, M. J. (2012). Coordination dynamics in socially situated nervous systems. Frontiers in Human Neuroscience, 6, 1–16. Constant, A., Clark, A., Kirchhoff, M., & Friston, K.  J. (2019a). Extended active inference: Constructing predictive cognition beyond skulls. Mind and Language, 37(3), 373–394. Constant, A., Clark, A., Kirchhoff, M., & Friston, K. J. (2019b). Regimes of expectations: An active inference model of social conformity and human decision making. Frontiers in Psychology, 10, 679. Corcoran, A., Pezzulo, G., & Hohwy, J. (2020). From allostatic agents to counterfactual cognizers: Active inference, biological regulation, and the origins of cognition. Biology and Philosophy, 35(3), 1–45. Dennett, D. (1991). Consciousness explained. Little Brown. Donnarumma, F., Costantini, M., Ambrosini, E., Friston, K.  J., & Pezzulo, G. (2017). Action understanding as hypothesis testing. Cortex, 89, 45–60. Fabry, R.  E. (2017). Transcending the evidentiary boundary: Prediction error minimization, embodied interaction, and explanatory pluralism. Philosophical Psychology, 30(4), 395–414. Facchin, M. (2021). Extended predictive minds: do Markov Blankets matter? Review of Philosophy and Psychology. https://doi.org/10.1007/s13164-­021-­00607-­9 Facchin, M. (2022). There is no “inference within a model”. Behavioral and Brain Sciences, 45, e 138, 34–e 138, 35. Favela, L., & Martin, J. (2017). “Cognition” and dynamical cognitive science. Minds and Machines, 27(2), 331–355. Feldman, H., & Friston, K. J. (2010). Attention, uncertainty and free-energy. Frontiers in Human Neuroscience, 4, 215. Friston, K.  J. (2005). A theory of cortical responses. Philosophical Transaction of the Royal Society B: Biological Sciences, 360(1456), 815–836. Friston, K. J. (2009). The free-energy principle: A rough guide to the brain? Trends in Cognitive Sciences, 13(7), 293–301. Friston, K. J. (2011). What is optimal about optimal motor control? Neuron, 72(3), 488–498. Friston, K. J. (2012a). A free-energy principle for biological systems. Entropy, 14(11), 2100–2121. Friston, K.  J. (2012b). Predictive coding, precision and synchrony. Cognitive Neuroscience, 3(3–4), 238–239. Friston, K. J. (2013). Life as we know it. Journal of the Royal Society Interface, 10(86), 20130475. Friston, K. J. (2018). Am I self-conscious? (or: Does self-organization entail self-consciousness?). Frontiers in Psychology, 9, 579. Friston, K.  J. (2019). Beyond the desert landscape. In M.  Colombo, E.  Irvine, & M.  Stapleton (Eds.), Andy Clark and his critics (pp. 174–190). Oxford University Press. Friston, K. J., & Frith, C. (2015a). A duet for one. Consciousness and Cognition, 36, 390–405. Friston, K. J., & Frith, C. (2015b). Active inference, communication and hermeneutics. Cortex, 68, 129–163. Friston, K. J., & Stephan, K. (2007). Free-energy and the brain. Synthese, 159(3), 417–458. Friston, K. J., Adams, R. A., Perrinett, L., & Breakspear, M. (2012). Perception as hypothesis, saccades as experiments. Frontiers in Psychology, 3, 151. Friston, K. J., Levin, M., Sengupta, B., & Pezzullo, G. (2015). Knowing one’s place: A free-energy approach to pattern regulation. Journal of the Royal Society Interface, 12(105), 20141383. Friston, K. J., Rosch, R., Parr, T., Price, C., & Bowman, H. (2017). Deep temporal models and active inference. Neuroscience and Biobehavioral Reviews, 77, 388–402. Friston, K. J., Wiese, W., & Hobson, J. A. (2020). Sentience and the origin of consciousness: From Cartesian duality to Markovian monism. Entropy, 22(5), 516.

12  Predictive Processing and Extended Consciousness: Why the Machinery…

205

Goodman, J. R. L., et al. (2005). The interpersonal phase entrainment of rocking chair movements. In H. Heft & K. L. Marsh (Eds.), Studies in perception and action VIII: Thirteenth international conference on perception and action (pp. 49–53). Erlbaum. Haken, H., Kelso, J. S., & Bunz, K. (1985). A theoretical model of phase transitions in human hand movements. Biological Cybernetics, 51(5), 347–356. Hobson, J. A., & Friston, K. J. (2014). Waking and dreaming consciousness: Neurobiological and functional considerations. Progress in Neurobiology, 98(1), 82–98. Hohwy, J. (2013). The predictive mind. Oxford University Press. Hohwy, J. (2016). The self evidencing brain. Noûs, 50(2), 259–282. Hohwy, J. (2017). How to entrain your evil demon. In T. Metzinger & W. Wiese (Eds.), Philosophy and predictive processing (Vol. 2, pp. 1–15). The MIND Group. Hohwy, J. (2020). Self-supervision, normativity and the free-energy principle. Synthese, 199(1–2), 29–53. Hohwy, J. (2021). Conscious self-evidencing. Review of Psychology and Philosophy, 13(4), 809–828. Hurley, S. (1998). Consciousness in action. Harvard University Press. Hurley, S. (2001). Perception and action: Alternative views. Synthese, 129(1), 3–40. Hurley, S. (2010). The varieties of externalism. In R.  Menary (Ed.), The extended mind (pp. 101–153). MIT Press. Hurley, S., & Noe, A. (2003). Neural plasticity and consciousness. Biology and Philosophy, 18(1), 131–168. Iodice, P., Porciello, G., Bufalari, I., Barca, L., & Pezzulo, G. (2019). An interoceptive illusion of effort induced by false heart-rate feedback. Proceedings of the National Academy of Sciences, 116(28), 13897–13902. Jiang, J., Chen, C., Dai, B., Shi, G., Ding, G., Liu, L., & Lu, C. (2015). Leader emergence through interpersonal neuronal synchronization. Proceedings of the National Academy of Sciences, 112(14), 4274–4279. Kang, O., & Wheatley, T. (2017). Pupil dilatation patterns spontaneously synchronize across individuals during shared attention. Journal of Experimental Psychology: General, 146(4), 569–576. Kaplan, D. M. (2012). How to demarcate the boundaries of cognition. Biology and Philosophy, 27(4), 545–570. Kelso, S. (1995). Dynamic patterns. MIT Press. Kiebel, S. J., von Kriegstein, K., Daunizeau, J., & Friston, K. J. (2009). Recognizing sequences of sequences. PLoS Computational Biology, 5(8), e1000464. Kilner, J. M., et al. (2007). Predictive coding: An account of the mirror neuron system. Cognitive Processing, 8(3), 159–166. Kilner, J.  M., et  al. (2011). Action understanding and active inference. Biological Cybernetics, 104(1–2), 137–160. Kirchhoff, M.  D. (2015). Experiential fantasies, prediction, and enactive minds. Journal of Consciousness Studies., 22(3–4), 68–92. Kirchhoff, M. D. (2018). Predictive processing, perceiving and imagining: Is to perceive to imagine, or something close to it? Philosophical Studies, 175(3), 751–767. Kirchhoff, M.  D., & Kiverstein, J. (2019a). Extended consciousness and predictive processing. Routledge. Kirchhoff, M.  D., & Kiverstein, J. (2019b). How to determine the boundaries of the mind: A Markov blanket proposal. Synthese, 198(5), 4791–4810. Kirchhoff, M. D., & Kiverstein, J. (2020). Attuning to the world: The diachronic constitution of the extended conscious mind. Frontiers in Psychology, 11, 1966. Kirchhoff, M. D., Parr, T., Palacios, E., Friston, K. J., & Kiverstein, J. (2018). The Markov blankets of life: Autonomy, active inference and the free-energy principle. Journal of the Royal Society Interface, 15(138), 20170792.

206

M. Facchin and N. Negro

Kiverstein, J. (2018). Extended cognition. In A. Newen, L. de Bruin, & S. Gallagher (Eds.), The Oxford handbook of 4E cognition (pp. 19–40). Oxford University Press. Kiverstein, J., & Farina, M. (2012). Do sensory substitution devices extend the conscious mind? In F.  Paglieri (Ed.), Consciousness in interaction (pp.  19–40). John Benjamins Publishing Company. Koch, C., Massimini, M., Boly, M., & Tononi, G. (2016). Neural correlates of consciousness: Progress and problems. Nature Reviews Neuroscience, 17(5), 307–321. Laflaquiere, A. (2017). Grounding the experience of a visual field through sensorimotor contingencies. Neurocomputing, 268, 142–152. Lamb, M., & Chemero, A. (2018). Interaction in the open: Where dynamical systems become extended and embodied. In A. Newen, L. de Bruin, & S. Gallagher (Eds.), The Oxford handbook of 4E cognition (pp. 147–162). Oxford University Press. Lange, M. (2017). Because without cause: Non-causal explanations in science and mathematics. Oxford University Press. Leinweber, M., Leinweber, M., Ward, D. R., Sobczak, J. M., Attinger, A., & Keller, G. B. (2017). A sensorimotor circuit in the mouse cortex for visual flow prediction. Neuron, 95(6), 1420–1432. Liu, T., Saito, H., & Oi, M. (2015). Role of the right inferior frontal gyrus in turn-based cooperation and competition: A near-infrared spectroscopy study. Brain and Cognition, 99, 17–23. Liu, N., Mok, C., Witt, E.  E., Pradhan, A.  H., Chen, J.  E., & Reiss, A.  L. (2016). NIRS-based hyperscanning reveals inter-brain neural synchronization during cooperative Jenga game with face-to-face communication. Frontiers in Human Neuroscience, 10, 82. Lupyan, G., Rahman, R.  A., Boroditsky, L., & Clark, A. (2020). Effects of language on visual perception. Trends in Cognitive Sciences, 24(11), 930–944. Mamassian, P., Landy, M., & Maloney, L. T. (2002). Bayesian modelling of visual perception. In I. R. Rao, B. Olshausen, & M. Lewicki (Eds.), Probabilistic models of the brain: Perception and neural functioning (pp. 13–36). MIT Press. Maye, A., & Engel, A. K. (2013). Extending sensorimotor contingency theory: Prediction, planning, and action generation. Adaptive Behavior, 21(6), 423–436. Menary, R., & Gillett, A. (2020). Are markov blankets real and does it matter? In D. Mendoça, M.  Curado, & S.  Gouveia (Eds.), The philosophy and science of predictive processing (pp. 39–58). Bloomsbury Academic. Menary, R., & Gillett, A. (2022). Markov blankets do not demarcate the boundaries of the mind. Behavioral and Brain Sciences, 45(136), 44–45. Millidge, B., Tschantz, A., & Buckley, C.  L. (2021). Whence expected free-energy? Neural Computation, 33(2), 447–482. Nagel, T. (1974). What is it like to be a bat? Readings in Philosophy and Psychology, 1, 159–168. Noë, A. (2004). Action in perception. MIT Press. Noë, A. (2009). Out of our heads. Macmillan. O’Regan, J. K. (2011). Why red doesn’t sound like a bell: Understanding the feeling of consciousness. Oxford University Press. O’Regan, J. K., & Noë, A. (2001a). A sensorimotor account of vision and visual consciousness. Behavioral and Brain Sciences, 24(5), 939–973. O’Regan, J. K., & Noë, A. (2001b). What it is like to see. A sensorimotor theory of perceptual experience. Synthese, 192(1), 79–103. Orlandi, N. (2014). The innocent eye. Oxford University Press. Palacios, E. R., Isomura, T., Parr, T., & Friston, K. J. (2019). The emergence of synchrony in networks of mutually inferring neurons. Scientific Reports, 9(1), 1–14. Palacios, E. R., Razi, A., Parr, T., Kirchhoff, M., & Friston, K. (2020). On Markov blanket and hierarchical self-organization. Journal of Theoretical Biology, 486, 110089. Palermos, O. (2014). Loops constitution, and cognitive extension. Cognitive Systems Research, 27, 25–41. Parr, T., Pezzulo, G., & Friston, K. (2022). Active inference. MIT Press.

12  Predictive Processing and Extended Consciousness: Why the Machinery…

207

Pepper, K. (2014). Do sensorimotor dynamics extend the conscious mind? Adaptive Behavior, 22(2), 99–108. Pezzulo, G. (2014). Why do you fear the bogeyman? An embodied predictive coding model of perceptual inference. Cognitive, Affective, & Behavioral Neuroscience, 14(3), 902–911. Pezzulo, G., Donnarumma, F., Iodice, P., Maisto, D., & Stoianov, I. (2017). Model-based approaches to active perception and control. Entropy, 19(6), 266. Putnam, H. (1975). The meaning of “meaning”. In H. Putnam (Ed.), Philosophical Papers Vol. II: Mind, Language and Reality (pp. 215–271). Cambridge University Press. Raja, V., Valluri, D., Baggs, E., Chemero, A., & Anderson, M. L. (2021). The Markov blanket trick. Physics of Life Review, 39, 49–72. Rao, R., & Ballard, D. (1999). Predictive coding in the visual cortex: A functional interpretation of some extra-classical receptive field effects. Nature Neuroscience, 2(1), 79–87. Richardson, M. J., Marsh, K. L., & Schmidt, R. C. (2005). Effects of visual and verbal interaction on unintentional interpersonal coordination. Journal of Experimental Psychology: Human Perception and Performance, 31(1), 62–79. Richardson, M. J., Marsh, K. L., Isenhower, R. W., Goodman, J. R., & Schmidt, R. C. (2007). Rocking together: Dynamics of intentional and unintentional social coordination. Human Movement Science, 26, 867–891. Rietveld, E., & Kiverstein, J. (2014). A rich landscape of affordances. Ecological Psychology, 26(4), 325–352. Rietveld, E., Denys, D., & Van Westen, M. (2018). Ecological-enactive cognition as engaging with a field of relevant affordances: The skilled-intentionality framework (SIF). In A. Newen, L. de Bruin, & S. Gallagher (Eds.), The Oxford handbook of 4E cognition (pp. 41–70). Oxford University Press. Roepstorff, A., Niewöhner, J., & Beck, S. (2010). Enculturating brains through patterned practices. Neural Networks, 23, 1051–1059. Ross, L. (2020). The explanatory nature of constraints: Law-based, mathematical, and causal. Phil-­ Sci Archive, 18504. Rowlands, M. (2010). The New Mind Sciences. The MIT Press. Rowlands, M. (2020). Externalism about the mind. In E. Zalta (Ed.), The Stanford encyclopedia of philosophy (winter 2020 edition). Retrieved from https://plato.stanford.edu/archives/win2020/ entries/content-­externalism Schmidt, R. C., & Richardson, M. J. (2008). Dynamics of interpersonal coordination. In A. Fuchs & V.  K. Jirsa (Eds.), Coordination: Neural, behavioral and social dynamics (pp.  281–308). Springer. Schmidt, R.  C., Carello, C., & Turvey, M.  T. (1990). Phase transitions and critical fluctuations in the visual coordination of rhythmic movements between people. Journal of Experimental Psychology: Human Perception and Performance, 16(2), 227–247. Seth, A. K. (2013). Interoceptive inference, emotion, and the embodied self. Trends in Cognitive Sciences, 17(11), 565–573. Seth, A.  K. (2014). A predictive processing theory of sensorimotor contingencies: Explaining the puzzle of perceptual presence and its absence in synesthesia. Cognitive Neuroscience, 5(2), 97–188. Seth, A. K. (2021). Being you. Faber & Faber. Seth, A. K., & Bayne, T. (2022). Theories of consciousness. Nature Reviews Neuroscience, 23(7), 439–452. Seth, A.  K., & Friston, K. (2016). Active interoceptive inference and the emotional brain. Philosophical Transactions of the Royal Society B: Biological Sciences, 371(1708), 2016007. Seth, A. K., & Hohwy, J. (2020). Predictive processing as a systematic basis for identifying the neural correlates of consciousness. Philosophy and the Mind Sciences, 1(2), 3. Shi, Y.  Y., & Sun, H. (2008). Image and video compression for multimedia engineering. Fundamentals, algorithms and standards (2nd ed.). CRC Press.

208

M. Facchin and N. Negro

Shockley, K., Santana, M. V., & Fowler, C. A. (2003). Mutual interpersonal postural constraints are involved in cooperative conversation. Journal of Experimental Psychology: Human Perception and Performance, 29(2), 326–332. Spratling, M. W. (2016). Predictive coding as a model of cognition. Cognitive Processing, 17(3), 279–305. Spratling, M.  W. (2017). A review of predictive coding algorithms. Brain and Cognition, 112, 92–97. Sprevak, M. (2009). Extended cognition and functionalism. The Journal of Philosophy, 106(9), 503–527. Stephens, G.  J., Silbert, L.  J., & Hasson, U. (2010). Speaker-listener neural coupling underlies successful communication. Proceedings of the National Academy of Sciences, 107(32), 14425–14430. Tani, J. (2016). Exploring robotic minds. Oxford University Press. Thierry, G., Athanasopoulos, P., Wiggett, A., Dering, B., & Kuipers, J. R. (2009). Unconscious effects of language specific terminology on preattentive color perception. Proceedings of the National Academy of Sciences, 106(11), 4567–4570. Tognoli, E., Zhang, M., Fuchs, A., Beetle, C., & Kelso, J. S. (2020). Coordination dynamics: A foundation for understanding social behavior. Frontiers in Human Neuroscience, 14, 317. Valencia, A.  L., & Froese, T. (2020). What binds us? Interbrain neural synchronization and its implications for theories of human consciousness. Neuroscience of Consciousness, 2020(1), niaa010. Vallée-Tourangeau, F., Steffensen, S. V., Vallée-Tourangeau, G., & Sirota, M. (2016). Insight with hands and things. Acta Psychologica, 170, 195–205. Vázquez, M. J. C. (2020). A match made in heaven: Predictive approaches to (an unorthodox) sensorimotor enactivism. Phenomenology and the Cognitive Sciences, 19, 653–684. Vold, K. (2015). The parity argument for extended consciousness. Journal of Consciousness Studies, 22(3–4), 16–33. Vold, K. (2021). Can consciousness extend? Philosophical Topics, 48(1), 243–264. Wheatley, T., Kang, O., Parkinson, C., & Looser, C. E. (2012). From mind perception to mental connection: Synchrony as a mechanism for social understanding. Social and Personality Psychology Compass, 6(8), 589–606. Wiese, W. (2018). Experienced wholeness. Integrating insights from gestalt theory, cognitive neuroscience, and predictive processing. MIT Press. Wiese, W., & Metzinger, T. (2017). Vanilla PP for philosophers. In T.  Metzinger & W.  Wiese (Eds.), Philosophy and predictive processing: 1 (pp. 1–18). The MIND Group.

Chapter 13

Commentary on “Predictive Processing and Extended Consciousness” Julian Kiverstein and Michael Kirchhoff

Abstract  We are grateful to Facchin and Negro (henceforth F&N) for their rich and generous engagement with our arguments for the hypothesis of the extended conscious mind (ECM). They offer a careful and insightful reconstruction of the key arguments from our 2019 monograph (Kirchhoff & Kiverstein, 2019a). In the end however they are not persuaded by the arguments of our book and raise a number of intriguing and puzzling challenges. We deal with each of their challenges and conclude that these challenges  pose no problems for our arguments for the ECM in Kirchhoff & Kiverstein (2019a).  We are grateful to Facchin and Negro (henceforth F&N) for their rich and generous engagement with our arguments for the hypothesis of the extended conscious mind (ECM). They offer a careful and insightful reconstruction of the key arguments from our 2019 monograph (Kirchhoff & Kiverstein, 2019a). In the end however they are not persuaded by the arguments of our book and raise a number of intriguing and puzzling challenges. In our reply we will focus on the following three issues from their chapter: 1 . The relationship between predictive processing and sensorimotor enactivism. 2. The role of cultural practices in the constitution of phenomenal experience. 3. The threat of conscious bloat for ECM.

J. Kiverstein (*) Department of Psychiatry, Amsterdam University Medical Research, Amsterdam, The Netherlands e-mail: [email protected] M. Kirchhoff School of Liberal Arts, University of Wollongong, Wollongong, NSW, Australia © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_13

209

210

J. Kiverstein and M. Kirchhoff

13.1 Sensorimotor Enactivism and Predictive Processing We start with a small terminological correction. F&N characterise our position as conscious vehicle externalism (CVE). The terminology of ‘vehicles’ of experience is generally introduced as a part of a package of ideas that take phenomenal experiences to be, or to depend upon, representational states that carry contents that semantically refer to objects and properties in the world. We wish to remain neutral on the question of the representational status of phenomenal experience. We therefore prefer to characterise our position using the terminology we employed in our book. We will refer to the hypothesis of the extended conscious mind (ECM). ECM claims that the biological system that is constitutive of at least some of a person’s phenomenal experiences of the world is an extended cognitive system. An extended cognitive system is a system composed of elements that are neural, bodily and environmental. ECM denies that the skin and skull of the individual is a “magical membrane” (Hurley, 2010) such that only what is inside of the skin can produce phenomenally conscious experience. The dynamical processes that constitute our phenomenal experiences are spread out across space and time, extending across the boundary that separates the brain from the body and the rest of the world. In our book we set out to defend ECM by combining two theories that might not be thought to be natural partners: the predictive processing theory and sensorimotor enactivism (henceforth SME). We take predictive processing (PP) to be a theory of how biological agents regulate their sensorimotor interactions with the environment, proactively orchestrating the sensory states they sample so as to remain well-­ adapted to their environments. From SME, we borrow the idea that phenomenal experience is constituted by a field of “causal flows characterised through time by a tangle of multiple feedback loops of varying orbits” (Hurley, 1998, p. 2). This field of causal flows is constituted out of the same cycles of perception and action that are described and modelled in the predictive processing theory in terms of active inference. We argue that the complex tangle of feedback loops, described in SME in terms of sensorimotor contingencies, is sometimes closed by the world in ways that allow phenomenal experience to extend into the world. F&N take the predictive processing theory (henceforth PP) to present an insuperable challenge to our argument for ECM. They argue that PP “operationalizes sensorimotor contingencies in purely neural terms, as expectations concerning the incoming inputs encoded in a generative model” (Chap. 12, Sect. 12.1). We do not deny that sensorimotor contingencies can be interpreted in brain-bound terms and explicitly discuss and reject such a challenge in our book. Such a brain-bound reading of SME would claim that the system that constitutes consciousness is a systems whose elements are exclusively neurophysiological and neurochemical. We deny however that SME must be interpreted in brain-bound terms. Sensorimotor enactivism claims instead that sensorimotor contingencies that fix the phenomenal qualities of experience consist of feedback loops that form between perception and action, and that are closed by the agents interaction with the world. We will call this the ECM interpretation of sensorimotor contingencies.

13  Commentary on “Predictive Processing and Extended Consciousness”

211

The ECM interpretation of sensorimotor contingencies is defended by us, and other proponents of SME, on the basis of the DEUTS argument (Hurley, 1998, 2010; Cosmelli & Thompson, 2010; Noë, 2004, 2009; Ward, 2012). The DEUTS argument comprises two theses (first explicitly stated in these terms and critiqued in Clark, 2009): 1. The Dynamic Entanglement Thesis: sensorimotor contingencies sometimes loop through the world in a way that yields the dynamic entanglement of the brain, body and world. 2. The Unique Temporal Signature Thesis: sensorimotor contingencies are characterised by unique temporal signatures that entail that the brain cannot be unplugged from the body and world while keeping the phenomenal character of experience fixed. F&N take PP to challenge the ECM interpretation of sensorimotor contingencies. They offer a plausible reading of PP as generalising theories of motor control that take the brain to use predictions of the sensory consequences of movement to ensure that the outcomes of an action match as closely possible with an agent’s goals. F&N are right to note the similarities between PP and so-called “forward-model” or emulation-­based theories of motor control. However, as Allen and Friston (2018) note, PP goes well beyond these models of motor control in taking inference to be embodied, and the possibilities for action (i.e. environmental affordances) to be correlated with embodied prior beliefs that guide embodied processes of inference (cf. Bruineberg et  al., 2018; Kirchhoff, 2018; Kirchhoff & Robertson, 2020; Ramstead et al., 2019). The sensorimotor contingencies that SME takes to fix the qualities of phenomenal experience have been argued to reflect an agent’s self-sustaining habitual modes of acting in the environment (Di Paolo et al., 2017). We have suggested that processes of prediction-error minimisation may account for the maintaining and sustaining of these habitual modes of engagement (Kirchhoff et al., 2018; Kiverstein et al., 2022). This proposal is in line both with recent work on sensorimotor contingencies (Di Paolo et al., 2017) and predictive processing. Thus, far from PP presenting a challenge to SME, as F&N attempt to argue, we believe ECM offers a perspective on consciousness that smoothly integrates SME and PP. PP provides a computational framework for formally modelling the account of phenomenal experience provided by SME. SME is in turn required to account for the phenomenal qualities and meaningfulness of experience (Kiverstein et al., 2022).

13.2 Cultural Practices and the Constitution of Phenomenal Experience ECM argues that perception and action form a causally circular process that couples the agent to the environment. The agent’s active and sensory states form a closed circuit (modelled as a Markov blanket) only in and through the agent’s sensorimotor

212

J. Kiverstein and M. Kirchhoff

dynamic coupling with the environment. In our book we argue that there is no end-­ point after which the agent can decouple from the world and rely on the generative model for the construction of its perceptual experiences. The necessity of coupling with the environment for the construction of phenomenal experience is ongoing. We gave as an example of this “unplugability” of the brain from the rest of the body and world (Hurley, 2010), the case of culture shock. Culture shock is a real-world example of the differences in phenomenal experience that can arise when you take an individual out of a cultural environment to which they have become attuned and transport them to a new and unfamiliar cultural environment. The result can be sustained experiences of distress and alienation from the world. We take the case of culture shock to highlight that attunement to the cultural environment is an integral part of the phenomenology of our everyday conscious experiences. In culture shock this attunement to the everyday world is lost and this leads to a deep disturbance of lived experience. But this attunement or misattunement relies on the ongoing coupling of the agent to the cultural world through cycles of perception and action. F&N object that the “fact that an agent’s mastery in a domain is achieved and maintained by means of constant environmental interactions does not pull the vehicles realising that mastery out of the agent’s body” (Chap. 12, Sect. 12.4.1). They give as an example: Amy the weightlifter whose mastery of weightlifting is maintained by her regular visits to the gym. F&N suggest, without argument, that Amy’s practical know-how is realised entirely “internal to her body” (Ibid). We wonder however why a proponent of ECM should agree to F&N’s description of Amy’s practical know-how. Amy’s mastery consists in a sensitivity and bodily responsiveness to the affordance of the liftability of weights. Why think of this sensitivity and responsiveness as wholly internally constituted, and not as constituted by the dynamics of the brain, body, environment system she forms when exercising her mastery of weightlifting? F&N take us to argue that cultural practices are parts of the vehicles that constitute subjective experience (Chap. 12, Sect. 12.4.1). We do indeed claim that the environment from which the subject cannot be unplugged has historically been shaped and structured by cultural practices. However, we stop short of claiming that cultural practices “function as external vehicles of subjective experience”. Cultural practices enter our account of ECM through our discussion of attention. Attention is understood in PP in terms of precision-weighting on prediction errors relative to prior expectations. Precision is set high on prediction errors that are estimated to be reliable and salient or newsworthy relative to what is already known. Thus in PP attention serves the function of selecting the prediction errors that get to influence further processing. We argue in our book that precision estimation plays an important role in maintaining the attunement of a person’s experience to the cultural world. We speculated that one mechanism by which this might be achieved is the shaping of precision expectations by cultural practices. We give as an example of this tuning of precision expectations by culture, a study by Thierry and colleagues, discussed by F&N, in which a visual mismatch negativity response (VMNR) was found in Greek speakers but not in English speakers when a light blue circle was introduced into a series of

13  Commentary on “Predictive Processing and Extended Consciousness”

213

dark blue coloured circles. We suppress the details of this study here (but for a good summary see F&N, Chap. 12, Sect. 12.4.1). F&N wonder how this research bolsters the case for ECM (Ibid.). They quote Thierry and colleagues as stating that their data speaks only “to a relationship between native language and unconscious, preattentive colour discrimination”, not to “conscious overt colour categorisation” (Thierry et al., 2009, p. 4568). We discussed this research as an example of how cultural practices can tune precision expectations, not as an example of how cultural practices can extend the vehicles of phenomenal colour experience. Nor did we claim that Greek and English speakers visually experience colours differently. The VMNR is an example of a prediction-error signal. We use the Thierry and colleagues research as an illustration of how a prediction error can be given a precision-­ weighted boost depending on the language a subject speaks. Having precision set in part by linguistic practices will make it likely that individuals that speak the same language will come to share experiences and ways of making sense of the cultural worlds they inhabit. F&N may agree with us that language can tune precision while doubting the relevance of this claim for ECM. We suggest reading our argument in the context of the Unique Temporal Signature Thesis stated above. Attunement to the cultural world will have a unique temporal signature in different contexts that SME claims are constitutive of the phenomenal qualities of experience. We point to the role of cultural practices in bringing about this unique temporal signature through the tuning of precision expectations.

13.3 The Threat of Conscious Bloat The DEUTS argument entails that the field of causal flows that are constitutive of phenomenal experience leak out into the environment. F&N ask how to distinguish the causes that are constitutive of phenomenal experience from those that are background conditions (such as the presence of oxygen in the air that the person breathes). They write: “Every agent is a node in a massive causal network connecting it to an unruly manifold of environmental features  […]  not all of which are putative external constituents of the agent’s phenomenal machinery” (Chap. 12, Sect. 12.4.3). F&N argue that the failure of ECM to distinguish causes that are background conditions from causes that are constitutive of phenomenal experience leads to the threat of conscious bloat. The threat they identify is that the machinery that is constitutive of phenomenal experience uncontrollably expands to include all of the causal networks the individual forms when interacting with the environment. ECM, they argue, implies a bloated consciousness whose material base risks expanding to include the many external causal factors that are better thought of as background conditions. Thus, F&N take ECM to be insufficiently discriminating in the field of causal flows it takes to be constitutive of phenomenal experience. But is this correct? Recall again the two theses that make up the DEUTS argument. The Dynamic

214

J. Kiverstein and M. Kirchhoff

Entanglement Thesis points to sensorimotor contingencies in which perception and action form a loop that is closed by the world. Thus consider an example by Noë: the flavour of wine. He suggests that the precise flavour sensation depends in part on the rolling of the liquid on the tongue. The liquid, the tongue and the rolling action are all parts, he suggests, of “the physical substrate for the experience’s occurrence” (Noë, 2004, p. 220). There is a dynamic entanglement that the rolling action of the tongue forms with the wine – a tangle of feedback loops that include the interaction of the liquid with the movement of the tongue. What is explanatory of the flavour sensation, in this example, is partly the feedback loop that forms between sensation and movement. This feedback loop also has a unique temporal signature. SME claims that the flavour sensation depends on the exact details of the rolling movement of the tongue. We suggest the DEUTS argument blocks F&N’s threat of conscious bloat. The two theses of the DEUTS argument imply that the causes that are constitutive of phenomenal experience are those that pertain to the sensorimotor contingencies that form in the particular token interaction of an agent with its environment. This does not imply, so far as we can tell, that causal background conditions, such as the presence of oxygen, must be treated as constitutive of phenomenal experience. Background conditions are not part of the particular dynamics of the interaction between agent and environment that are relevant to explaining the qualities of phenomenal experience. To return to the wine example, the presence of oxygen is not causally relevant in the same way as the rolling motion of the tongue for the explanation of the quality of the flavour sensation. It is causally relevant as a background condition that is relevant for the explanation of every experience a subject can enjoy so long as they are alive. The rolling motion of the tongue by contrast is a part of a sensorimotor dynamic that is causally relevant for the explanation of this particular experience of taking a sip of wine. F&N might still object that SME conflates causes of phenomenal experience with processes that are constitutive of the material base – the vehicles – of phenomenal experience. Standing behind this objection is the charge that arguments for the extended mind in general are guilty of conflating causation with constitution. How do we go about defining a boundary for the cognitive system that is constitutive of phenomenal experience such that we can determine what causal processes are constitutive of this system, and what causes lie outside of the system? We have proposed to use PP to settle this question (Kirchhoff & Kiverstein, 2019a, 2019b). In PP, agents are taken to be models of their environments. It follows that an agent that samples sensory states in its sensorimotor interactions with the environment that maximise the evidence for its model, will also thereby maximise evidence for its own existence. Maximising evidence for a model is equivalent to minimising prediction error in the long run. A system that selects actions that minimise expected or long-term prediction errors will thereby produce and maintain a nesting of boundaries, each composed of sensory and active states, that distinguish the agent as a cognitive system from its environment. This nesting of boundaries can be formally modelled as a nesting of Markov blankets. We have therefore proposed to apply the

13  Commentary on “Predictive Processing and Extended Consciousness”

215

Markov blanket formalism to systems that minimise long-term prediction errors as a method for identifying the boundaries of a cognitive system. F&N object again that this proposal to use the Markov blanket formalism to individuate cognitive systems is insufficiently discriminating. They point out, correctly, that any two subjects that enter into a state of generalised synchrony can be modelled as forming a system with its own Markov blanket separating the dyad from the rest of the environment. However, we are in no way committed to the claim that such a dyad is thereby an example of a conscious extended mind. We follow Friston (2017, 2018) in arguing that consciousness emerges only in prediction error minimising systems with a deep temporal hierarchical structure. Such an agent will be able to make inferences not only about the present but also about its past and future projecting itself through time so as to engage in “proactive, purposeful inference about its own future” (Friston, 2017). We will call this the “temporal thickness” condition. How does the temporal thickness condition relate to our proposal to use Markov blankets to identify the boundaries of cognitive systems? F&N seem to think of these as relatively independent proposals. They rightly provide the examples of systems satisfying generalised synchrony as a single system with its own Markov blanket that may lack temporal depth. Thus, Markov blankets, they contend, do not necessarily serve as a formal means of identifying the boundaries of the conscious mind. This is correct. However, when we are interested as scientists in identifying the boundaries of the conscious mind we contend the formalism will apply to systems that satisfy the temporal thickness condition. The nesting of Markov blankets in such a system will map onto local or fast temporal dynamics in lower layers of the model that are constrained by the slower global dynamics at higher layers of the model. Hobson and Friston (2014) have argued that it is precisely this time-­ dependent, circular causal dynamics in which the global slower dynamics enslaves the faster local dynamics that is constitutive of consciousness. Thus, if Hobson and Friston are right, the nesting of Markov blankets that we propose to use to identify the boundaries of the conscious mind, and temporal depth of the generative model are necessarily connected. We conclude then that predictive processing and sensorimotor enactivism can be smoothly integrated in ways that support ECM. The machinery of the phenomenally conscious mind probably is not only inside of the head.

References Allen, M., & Friston, K.  J. (2018). From cognitivism to autopoiesis: Towards a computational framework for the embodied mind. Synthese, 195(6), 2459–2482. Bruineberg, J., Kiverstein, J., & Rietveld, E. (2018). The anticipating brain is not a scientist: The free-energy principle from an ecological-enactive perspective. Synthese, 195(6), 2417–2444. Clark, A. (2009). Spreading the joy? Why the machinery of consciousness is (probably) still in the head. Mind, 118(472), 963–993.

216

J. Kiverstein and M. Kirchhoff

Cosmelli, D., & Thompson, E. (2010). Embodiment or envatment? Reflections on the bodily basis of consciousness. In J. Stewart, O. Gapenne, & E. Di Paolo (Eds.), Enaction: Towards a new paradigm in cognitive science (pp. 361–386). MIT Press. Di Paolo, E., Buhrmann, T., & Barandiaran, X. (2017). Sensorimotor life. Oxford University Press. Friston, K. J. (2017). Consciousness is not a thing but a process of inference. AEON, Retrieved from https://aeon.co/essays/consciousness-­is-­not-­a-­thing-­but-­a-­process-­of-­inference Friston, K. J. (2018). Am I self-conscious? (Or does self-organisation entail self-consciousness?). Frontiers in Psychology, 9, 579. Hobson, J., & Friston, K. J. (2014). Consciousness, dreams, and inference: The Cartesian theatre revisited. Journal of Consciousness Studies, 21(1–2), 6–32. Hurley, S. L. (1998). Consciousness in action. Harvard University Press. Hurley, S.  L. (2010). The varieties of externalism. In R.  Menary (Ed.), The extended mind (pp. 101–154). MIT Press. Kirchhoff, M. D. (2018). Autopoiesis, free energy, and the life-mind continuity thesis. Synthese, 195, 2519–2540. Kirchhoff, M., & Kiverstein, J. (2019a). Extended consciousness and predictive processing: A third-wave view. Routledge. Kirchhoff, M., & Kiverstein, J. (2019b). How to determine the boundaries of the mind. Synthese, 198, 4791–4810. Kirchhoff, M.  D., & Robertson, I. (2020). Enactivism and predictive processing: A non-­ representational view. Philosophical Explorations, 21(2), 264–281. Kirchhoff, M. D., Parr, T., Palacios, E., Friston, K. J., & Kiverstein, J. (2018). The Markov blankets of life: Active inference, autonomy and the free energy principle. The Journal of the Royal Society Interface, 15(138), 20170792. Kiverstein, J., Kirchhoff, M. D., & Froese, T. (2022). The problem of meaning: The free energy principle and artificial agency. Frontiers in Neurorobotics, 16, 844773. Noë, A. (2004). Action in perception. MIT Press. Noë, A. (2009). Out of our heads: Why you are not your brain, and other lessons from the biology of consciousness. Hill and Wang. Ramstead, M., Kirchhoff, M. D., & Friston, K. (2019). A tale of two densities: Active inference is enactive inference. Adaptive Behaviour, 28(4), 225–239. Thierry, G., Athanasopoulos, P., Wiggett, A., Dering, B., & Kuipers, J. R. (2009). Unconscious effects of language-specific terminology on preattentive color perception. Proceedings of the National Academy of Sciences, 106(11), 4567–4570. Ward, D. (2012). Enjoying the spread. Conscious externalism reconsidered. Mind, 121(483), 731–751.

Chapter 14

Plural Methods for Plural Ontologies: A Case Study from the Life Sciences Luis H. Favela

and Anthony Chemero

Abstract  As with much contemporary philosophical and scientific research, the predominant metaphysics of situatedness is monism, particularly, physicalism. Here, we claim that while monism is the proper metaphysical thesis, empirically-­ supported theories of situated phenomena require ontological pluralism as well. We defend this position via the example of bird flocks, which are situated systems that exhibit ontologically plural features, namely, component dominance and interaction dominance. The description of these features will illustrate that understanding these phenomena requires a coevolution of conceptual and methodological development. Specifically, ontological features are partially identified and evaluated by way of the analyses applied to them. Both descriptive and normative lessons are drawn. Descriptively, research on bird flocks demonstrate that natural phenomena may not be readily cast via a monistic ontology (e.g., component dominant), even at the same scale of investigation. The normative consequence is that while research on situated systems need not reject metaphysical monism, it ought to begin from a pluralistic position concerning ontology. Keywords  Bird flocking · Component dominant · Interaction dominant · Ontological pluralism

L. H. Favela (*) Department of Philosophy, University of Central Florida, Orlando, FL, USA Cognitive Sciences Program, University of Central Florida, Orlando, FL, USA e-mail: [email protected] A. Chemero Department of Philosophy, University of Cincinnati, Cincinnati, OH, USA Department of Psychology, University of Cincinnati, Cincinnati, OH, USA © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_14

217

218

L. H. Favela and A. Chemero The world we inhabit is abundant beyond our wildest imagination. There are trees, dreams, sunrises; there are thunderstorms, shadows, rivers; there are wars, flea bites, love affairs; there are the lives of people, Gods, entire galaxies. (Feyerabend, 1999, p. 3)

14.1 Introduction Whether practitioners explicitly state it or not, much contemporary philosophical and scientific research, including work in situated cognition, is committed to metaphysical naturalism and physicalism, which are theses about the nature of existence. It is common for these ideas to be interrelated along the lines of understanding that the natural world is physical and what is physical is natural (cf. Papineau, 2021). Claims about what exists in the natural world or what is physical are typically cashed out in terms consistent with the natural/physical sciences, such as physics and chemistry (cf. Stoljar, 2022). The metaphysical core of naturalism and physicalism is a form of monism, namely, that existence is constituted by a single substance (cf. Schaffer, 2018). Such a view of existence contrasts with metaphysical theses like Platonic forms and Cartesian dualism that advance views of existence comprised of more than one substance. The claim by proponents of situated cognition that thinkers are necessarily in and of the environment is fully in line with this monism. A related topic in discussions of existence is ontology. While there is debate about how to exactly define ‘ontology’ and what its relation to metaphysics is (e.g., Hofweber, 2021), one way to understand it is as the study of what kinds of things there are after you have made a metaphysical commitment. Along these lines, it is consistent to be a metaphysical monist and an ontological pluralist at the same time. For example, living entities can be viewed as ontologically distinct from nonliving entities and both can fall under the same naturalist/physicalist metaphysics. It is arguable that while the metaphysics of science since at least the twentieth century was predominantly naturalist/physicalist, its ontology was commonly monist as well. This claim is motivated by the dominance of reductionism as the leading scientific methodology. There are generally three ways to view reductionism in scientific practice (Brigandt & Love, 2022): One is ontological reductionism, which is the idea that phenomena are principally constituted by the lowest level, such as water being hydrogen and oxygen molecules organized and interacting in certain ways. Another is epistemological reductionism, which is the idea that knowledge in one domain can be explained by knowledge at a lower domain, such as the psychological phenomena explained by chemical phenomena. The third is methodological reductionism, which is the idea that a phenomenon is best investigated at the lowest possible levels. A number of the greatest scientific achievements of the twentieth century showcased the power of methodological reductionism, for instance, the discovery of the structure of DNA by James Watson and Francis Crick.

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

219

In order to get to the lowest level, methodological reductionism commonly requires taking a phenomenon apart, localizing its parts, and understanding the parts’ functions and how to put them back together. Consider the example of a mechanical wristwatch. Assuming you have not studied one before and are not provided with an instruction manual, how would you approach understanding it? It would be reasonable to start identifying its parts, how they fit together, and what they do. While this is clearly an instance of methodological reductionism, we think such investigative strategies have commonly been viewed as supporting ontological reductionism as well. The line of thought is as follows: If a phenomenon, like a mechanical wristwatch, is fruitfully understood by taking it apart and identifying the parts’ functions, then that phenomenon is, ontologically speaking, just the collection of those parts and nothing above and beyond them. The move from ontological reductionism to methodological reductionism can be quite reasonable without much defense, that is, if an entity is understood as just being a collection of lower-level parts, then understanding it as such requires identifying those parts and their functions. However, the move from methodological reductionism to ontological reductionism is not as uncontroversial. That is to say, while it may be fruitful to take a thing apart to understand it more, it does not necessarily follow that the target of investigation is just comprised of ontologically-­ distinct parts. Still, there should be no doubting the fruitfulness of doing science with the starting assumption that a phenomenon can be fruitfully investigated by applying a strategy that aimed at taking it apart to attempt to see how the parts work together. Whether as an unspoken heuristic or as an explicit investigative methodology, the aspect of reductionism that aims to decompose phenomena and localize functions came to be taken quite seriously by philosophers of science starting in the mid-1990s. This orientation toward the subject matter of science came to be known as the new mechanism (Craver & Tabery, 2019). For nearly 30 years, a number of philosophers have defended the claim that the search for mechanisms is the most fundamental kind of explanatory strategy, particularly in the life sciences. This claim is said to be both historically and normatively true (e.g., Bechtel & Abrahamsen, 2005; Craver, 2005, 2007; Craver & Bechtel, 2007; Craver & Darden, 2005, 2013; Darden & Craver, 2002; Machamer et al., 2000; Piccinini & Craver, 2011). The reach of mechanistic explanations, it is claimed, not only applies to the biological sciences but also other special sciences such as economics (Craver & Alexandrova, 2008; cf. Reiss, 2007). In keeping with this claim are arguments that purported alternative kinds of explanations actually only augment mechanistic explanations but do not provide full explanations themselves (Bechtel & Abrahamsen, 2010, 2011; Kaplan & Bechtel, 2011; Kaplan & Craver, 2011). Many responses have been made in defense of non-mechanistic approaches being full explanations in themselves (e.g., Beer, 2000; Chemero, 2009; Favela, 2020a; Lange, 2017; Ross, 2022). However, we think there is an issue in the debate that has not received enough attention, namely, the restrictive nature of adhering to “one size fits all” ontological commitment.

220

L. H. Favela and A. Chemero

Whether explicitly or implicitly, in scientific practice the mechanistic approach seems to entail an unnecessarily limited ontology. That is to say, if mechanistic strategies are the proper strategies for explaining the natural world, then it follows that the natural world is populated by mechanisms (Mitchell, 2002; Wright & Bechtel, 2006, pp.  59–61). From the traditional mechanistic approach sketched above that directly follows from the successful reductionist successes of the twentieth century, that would be an ontology entailed by the strategy of decomposition and localization. Such an ontology would be exclusively of the kind we call component dominance. We will say more about component dominance below. For now, the important point concerns the number of ontological postulates had by explanatory approaches. It is worth reiterating here that the terms of our discussion do not center on an issue of “singular ontology” as opposed to, say, “multiple ontologies” like substance dualism. That would be to shift the discussion away from ontology to metaphysics. The issue at hand is the ontological categories that exist within a particular metaphysics. Again, as proponents of situated, embodied cognition, we are assuming a naturalist/physicalist metaphysics. Thus, we are concerned with the ontological categories of scientific metaphysics, which is monist, that is, naturalist/ physicalist through and through. We start our discussion of ontological pluralism with an interpretation of the mechanist approach as entailing a singular ontology. This is because the new mechanism has been the predominant approach for nearly 30 years. But our point about pluralism in matters of scientific ontology applies to all other explanatory orientations as well, such as causal mechanical, deductive-nomological, non-causal mathematical, or unificationist. Our claim that a singular ontology is an unnecessarily restrictive starting position for science is both descriptive and normative. Descriptively, much actual scientific research is not readily cast under a singular ontology that approaches natural phenomena as if they are mechanisms. For example, multidisciplinary scientist Dante Chialvo (2018) reviews explicitly non-­mechanistic approaches to phenomena such as biological organisms, glassy states, language, proteins, and turbulence. There are many natural phenomena that benefit from an orientation that does not assume one ontology and the application of multiple explanatory strategies, even at the same scale of investigation. Thus, normatively speaking, scientific research ought to begin from a pluralistic position concerning ontological kinds and explanatory style (e.g., Suppes, 1978). The world is tremendously complex. Moreover, there are various explanatory goals that guide and motivate investigations. In light of these facts, we believe that investigations in the life and other special sciences ought to begin from a more agnostic position concerning the ontological kinds constituting the phenomena under inquiry (cf. Chemero & Silberstein, 2008; Dale, 2008; Dale et  al., 2009; Godfrey-Smith, 2010; Love, 2012; Sterelny, 1996). To start from an agnostic position is to be open to pluralism. If we are to be pluralistic, then we should not exclude a priori the prospect of multiple ontological kinds constituting a phenomenon and, thus, multiple empirically reasonable theoretical frameworks. Of course, all investigations must start somewhere. However, investigations need not all start from the same place, nor do all investigations need to start with the same theoretical

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

221

commitments. Moreover, in some cases there is no single kind of explanation that will account for all ontological kinds contributing to the structure of a single phenomenon. In other words, there are cases in which no single path to discovery will be able to account for the total nature of a phenomenon. We demonstrate the need for ontological pluralism in scientific explanations by means of a case study from the life sciences. The case study is that of bird flocking, specifically, starling murmurations. Additionally, we focus on mechanistic and dynamical accounts. The mechanistic approach will be shown to have a commitment to a singular ontology in the form of component dominance. The dynamical approach will be shown to have a commitment to a singular ontological in the form of interaction dominance. We proceed by first giving an account of mechanistic and dynamical approaches, with emphases on component and interaction dominance typical of their respective ontological commitments. Then, we present the case study of bird flocking. After, we describe mechanistic and dynamical explanations of different aspects of the ontological features of bird flocking at the same scale of investigation. Finally, we elucidate why this case supports the notion of ontological pluralism. In particular, we make clear how a devotion to ontological monism prevents one from being able to account for certain features of a natural, biological phenomenon. Being an ontological pluralist allows one to account for those features.

14.2 Mechanistic and Dynamical Explanations What follows are accounts of mechanistic and dynamical explanations. Given the enormous size of the relevant literature, we do not claim to provide all encompassing descriptions of mechanistic or dynamical explanations. Instead, we draw attention to those aspects that indicate ontological commitments to component dominance and interaction dominance. In doing so, we highlight some of the typical characteristics of those explanatory strategies, particularly those that we understand as being difficult to reconcile and which we take as underlying reasons why the two approaches are often viewed as at odds with each other. Moreover, we focus on those aspects of these two explanatory strategies that are explicitly utilized in scientific practice.

14.2.1 Mechanistic Explanations There are numerous versions of “mechanistic explanations” (for a very small sample see Bechtel, 2008; Bechtel & Richardson, 1993/2010; Craver, 2007; Glennan, 2005; Machamer et  al., 2000; etc.). Though there are a number of sophisticated philosophical accounts of mechanistic explanation in recent years (for review see Glennan & Illari, 2018), here we limit ourselves to that which can broadly be conceived as the early work in the “new mechanical philosophy” that appeared in the

222

L. H. Favela and A. Chemero

1990s and early 2000s (i.e., Bechtel & Richardson, 1993/2010; Machamer et al., 2000). We limit our discussion to that conception of mechanistic explanation for two reasons. First, attempting to synthesize the literature into a single “mechanistic framework,” while properly treating the relevant work, is far beyond the scope of the current work. Second, we take the early work by people such as Bechtel and Richardson to be attempts at understanding heuristics utilized by scientists in their research (1993/2010, e.g., pp. xiii–xiv, xix, 37, 236); heuristics that we take to still be central to practice in the life sciences. Accordingly, we utilize “mechanistic explanation” as referring to three main features: linearity, localization, and component dominance. Before explaining each of those features, we begin with a general description of a mechanistic explanatory strategy. In very broad terms, a mechanistic explanation is one in which the behavior of a system is accounted for in terms of the functions and interactions of its parts (Bechtel & Richardson, 1993/2010, p. 17). “Mechanisms” are the entities and activities whose organization exhibits the explained phenomenon (Craver, 2007). The organization of these entities and activities produce regular changes from start or set-up to finish or termination conditions (Craver & Darden, 2013, p. 15; Machamer et al., 2000). The methods utilized to bring about these explanations are those of decomposition and localization. As Bechtel and Richardson describe it, decomposition allows for “the activity of a whole system [to be treated] as the product of a set of subordinate functions” (1993/2010, p.  23). Moreover, these interactions are accounted for in a manner that is additive and linear (1993/2010, p. 23). Localization is the investigative process by which “the different activities proposed in a task decomposition [are identified] with the behavior or capacities of specific components” (1993/2010, p. 26). A mechanistic explanation is achieved when a phenomenon’s mechanisms are described (Machamer et al., 2000). With this general picture of mechanistic frameworks and explanations in mind, we now focus on three of the main commitments of mechanistic explanations: linearity, localization, and component dominance. The first commitment underlying mechanistic frameworks is linearity. Machamer, Darden, and Craver state that, “Mechanisms are entities and activities such that they are productive of regular change from start or setup to finish or termination conditions” (2000, p. 3). Such a conception of mechanisms should not be surprising since one of the heuristic assumptions of human bounded rationality is that humans often think of events as occurring linearly (Bechtel & Richardson, 1993/2010, p.  18). In other words, humans think about events in sequential and step-by-step terms: “Why did D happen? Because first A, then B and C, and then D.” In order to reveal the significance such commitments to linearity have, it is informative to look at how linearity constrains the way data is analyzed. Linearity can be understood as a temporal factor in that phenomena are the result of sequential steps. As long as the steps in a particular linear process are followed, then particular outcomes are guaranteed. When conceived of as a temporal factor, linearity can be understood structurally, where “structure” refers to the relations of parts. In that way linearity can be understood as a kind of additivity

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

223

(Sternberg, 1969), and can be summed up in the following equation (Eq.  14.1; Ivancevic & Ivancevic, 2007, p. 52):

f ( x + y) = f ( x) + f ( y)



(14.1)

The idea that the equation captures is that a phenomenon, f, can be characterized as composed of components x and y, or f can be characterized as components x and y added together. The phenomenon f is nothing more or less if it is conceptualized as being a unit comprised of parts x and y or as a phenomenon comprised of part x added to part y. If a phenomenon is linear, and if it is additive, then it should follow that the effects of perturbations on the system are localized in individual components (Holden et al., 2009, p. 319). Systems that are the result of additive dynamics can be described as having weak interactions (Holden et  al., 2009). They are “weak” because system-level behaviors are primarily the result of their parts added together. Consider a system S comprised of parts X, Y, and Z. If the properties of S are the result of additive relations among X, Y, Z, and if X = 1, Y = 2, and Z = 3, then S will have property 6 (SX,Y,Z), specifically, the result of an additive relationship among X, Y, and Z (1 + 2 + 3 = 6). Thus, if X were perturbed such that property 1 was “knocked out” (e.g., when a neuron receives a “block” and is cut off from inputs from other neurons), then S should have a resulting property 5 (Sy,z). System S goes from property 6 to property 5 because perturbing X alone does not affect the properties of Y or Z. Such a system can be understood as being component dominant. Component dominance refers to cases where a system’s behaviors result from additively and linearly organized parts with rigid architecture and predetermined functions. A component-dominant system is one that can be decomposed into distinct functional units of a larger system (Favela & Chemero, 2016). Where the relevant components “bottom out” is a matter of debate (e.g., Kuhlmann & Glennan, 2014). Though for current purposes, we can understand the “lowest level” components as those that are experimentally revealed to be the spatial (e.g., size), temporal (e.g., duration), and causal (e.g., feedback) parts of the mechanism under investigation (Povich & Craver, 2018, p. 106). One way to identify components is to first assume the phenomenon under investigation is additive. In order to reveal additivity, the phenomenon is decomposed in order to localize specific capacities. When the data from the parts are analyzed, they should demonstrate component-dominant dynamics, that is, weakly interacting parts, with local affects that can be summed to reveal the whole behavior. If the parts of the system can be summed to reveal the whole behavior, then the commitments to linearity, additivity, localization, and component dominance was the correct one. Both artifacts and natural phenomena can exhibit additivity, linearity, and component dominance. Artifacts include combustion engines, pocket calculators, and toilets. Natural phenomena include Shepherds’ account of chemical transmission at synapses (Machamer et al., 2000, pp. 8–13), long-term potentiation (Craver, 2005), and human digestion (Fig. 14.1). Again, as we stated at the start of this section, our

224

L. H. Favela and A. Chemero

Fig. 14.1  Human digestive system. Human digestion is a biological phenomenon that can be understood as a mechanism that exhibits additivity, linearity, and component dominance. Here, biological mechanism represented by a realistic anatomical diagram (left) and as an abstract diagram (right). (Left figure modified and reproduced with permission from Buchanan, 1933. Public domain)

aim was not to synthesize the large and sophisticated philosophical literature on mechanisms. Instead, we merely aimed to present a form of mechanistic explanation that we take to be common in scientific practice. In the following section, we take the same approach with dynamical explanations.

14.2.2 Dynamical Explanations The philosophical literature on dynamical explanations is not nearly as vast or controversial as that for mechanisms; this is especially true for special sciences (for discussion, see Favela, 2020b, 2021). This may be due to philosophers inheriting a clearer articulation of the concepts and methods from engineers and scientists who broadly agree on their definition and usage (for review see Strogatz, 2018). Dynamical explanatory strategies typically center on accounting for the relevant variables within a set of differential equations and depicting the phenomenon in a state space, which, following the variables in the equations, depicts all possible states of the system and how the system evolves over time. Although there are linear dynamical systems, much of the appeal and power of dynamical systems theory

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

225

stems from its ability to capture nonlinear phenomena. Dynamical explanations tend to focus on the principles responsible for a system’s spatial and temporal structure (e.g., catastrophe theory, laws of coordination, and universality classes; Favela, 2020a). Such explanations are similar, though not equivalent, to covering law explanations, are counter-factual supporting, and provide evidence for unification of apparently distinct phenomena (Stepp et al., 2011). When thinking about the nature of dynamical explanations, the first thing to keep in mind is that dynamical explanations seek to accomplish something very different than mechanistic explanations. As discussed in the previous section, mechanistic explanations tend to focus on identifying the constitutive components and properties of the components of the phenomenon. Dynamicists tend to be interested in discovering principles, how widely they apply in nature, and how they can be predicted and manipulated in order to facilitate understanding of complex phenomena. Instead of the constituents of a phenomenon, dynamical explanations are about discovering how the states of a system evolve over time according to a principle (Favela, 2020b). These principles are often written captured via differential equations, which are suited due to their ability to express temporal features of a system. Nonetheless, for an investigator to understand how the system evolves, having access to a model and solving the equations is not always enough. That is why it is typical for an investigator to plot the equation in a state space, which contains all possible states of the system over time (Kelso, 1997). Having a qualitative interpretation of a system (i.e., a state space, phase space portrait, etc.) facilitates understanding for an investigator in a way that equations alone do not. Dynamical explanation is, in part, graphical. As mentioned above, the dynamical systems treatment of a phenomenon includes capturing the phenomenon in a state space that represents all possible states of the system and how the system evolves over time. Such pictorial depictions of a system can facilitate understanding in a way that abstract equations might not. Take the example of the FitzHugh-Nagumo model of single-neuron excitability (i.e., the readiness of a cell to respond to stimulus; FitzHugh, 1961; Nagumo et al., 1962). The aim of this model was to simplify the classic Hodgkin-Huxley model (Hodgkin & Huxley, 1952) of single-neuron action potential by isolating the necessary mathematical properties required to produce only physiologically-relevant activity. To do this, the FitzHugh-Nagumo model quantifies just the electrochemical properties of potassium and sodium ion flows to capture the full range of neuronal excitation and propagation (Izhikevich & FitzHugh, 2006). The result is a simplified two-­ dimensional differential equations model (Eqs. 14.2 and 14.3): 3



V −W + I V = V − 3



W = 0.08 (V + 0.7 − 0.8W )

(14.2)

(14.3)

226

L. H. Favela and A. Chemero

where I is the stimulus current magnitude, V is the cell membrane potential, and W is the recovery variable. The full range of biologically-realistic activity is depicted by plotting the trajectories produced by the equations in a state space across two dimensions (Fig. 14.2). Although such biologically realistic models are relatively simple, solving (i.e., understanding) the model is difficult to achieve analytically. In this situation, “analytic” means obtaining knowledge about the behavior of the system by actually solving the equations. Solving differential equations is fundamental to the utilization of such methods. In order to solve a differential equation, one must find a function that specifies a rule that determines values of the dependent variable and does not change the dependent variables (Brown, 2007, p. 13). Thus, in order to solve or understand the model, it is necessary to plot it in a state space (Fig. 14.2). This state space portrait allows the qualitative behavior of a complex system (single neurons) to be analyzed via a model with two differential equations (Eqs. 14.2 and 14.3), all in terms of a simple two-dimensional plane that shows the entire range of behaviors of the system over time. Understanding is facilitated by limiting the causally relevant factors to two dimensions: cell membrane potential (V) and recovery variable (W). This strategy is well-suited for investigating phenomena with two particular characteristics: nonlinearity and interaction dominance. Nonlinearity refers to the dynamics of systems whereby the output is not directly proportional to the input. This commonly comes in the form of exponential or multiplicative dynamics (Lam, 1998) and contrasts with linear systems that are additive 2.0

v

1.5 1.0 0.5 0.0 -0.5 -1.0 -1.5 -2.0

-0.5

w

0.5

1.0

1.5

I=0.73080

1 v

0.0

0 -1 -2

0

100

200

time (ms)

300

400

Fig. 14.2  FitzHugh-Nagumo model. Time series (bottom) and state space (top) of FitzHugh-­ Nagumo model ramp up and stabilized spiking. As in the two-dimensional differential equations model, I is the stimulus current magnitude (here 0.73080), V is the cell membrane potential, and W is the recovery variable. (Reproduced with permission from D.  Müller-Komorowska, 2020. Retrieved from https://gist.github.com/danielmk/2dfbc694e9c47916a079f3a4d34d6d97. CC BY 3.0)

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

227

in that outputs are directly proportional to inputs. As discussed in the previous section, if a phenomenon is linear, and if it is additive, then it should follow that the effects of perturbations on the system will reveal “weak interactions” localized in individual components (Holden et al., 2009, p. 319). Systems that are nonlinear, on the other hand, are not typically comprised of weakly interacting components. Thus, instead of being component dominant, nonlinear systems can be interaction dominant. Interaction dominance refers to a kind of dynamics exhibited by a system. Systems are interaction dominant when the dynamics or properties of the interactions among the parts supersede those that the parts would have separately or if the parts contributed to a system via additive interactions (Holden et  al., 2009; Silberstein & Chemero, 2013; Silberstein, 2021). Interaction-dominant systems can also be described as non-modular or non-decomposable. When a system is modular, decomposable, and component dominant, then the properties of the system result from the properties of the component parts “added” together. In this way, the composition and capacities of parts can be the primary cause of the dynamics of the system. Mechanistic explanation is appropriate in these cases because when a system is component dominant, then the strategy of decomposition and localization works quite well. However, in the case of systems with interaction-dominant dynamics, the method of decomposition and localization is not usually effective, as the properties of interaction-dominant systems result primarily from the dynamic interactions of the components. It is possible to figure out how the properties of the components contribute to the dynamics of the system, but only when examined post-hoc. In other words, if an investigator were presented with a set of components, and if she were not permitted to observe how those components act as a system, then it is unlikely—if not impossible—to predict what the system’s behavior would be. In some cases, this is due to the fact that components of interaction-dominant systems can have properties that result from non-additive interactions (Carello & Moreno, 2005; Sternberg, 1969, 2011). In the previous section, we described a linear and additive system, S. Let us look at a nonlinear and non-additive system, S*. If S* is comprised of A, B, C, where A = 1, B = 2, and C = 3, but the interactions among A, B, C are nonlinear, then swapping out A for D (D = 4) could alter the interactions of the entire system. So, if S* is a nonlinear and non-additive system, then the system-level activity could be exponential or multiplicative, for example: S* = 24 (where S*D,B,C [4 × 2 × 3 = 24]) or S* = 96 (where S*D,B,C [42 × 2 × 3 = 96]). In this way, we can also see what happens to a system that is interaction dominant. Interaction-dominant systems have system-level properties that override the properties of the components. In a component-dominant system, where the interactions of the components are additive, the properties of the whole system are no more or less than the properties of the components: System SX,Y,Z (X = 1, Y = 2, and Z = 3) has property 6 if the interactions are additive (1  +  2  +  3  =  6). In an interaction-­ dominant system where the interactions are nonlinear (i.e., exponential or multiplicative), the interactions of the components are altered based on the nature of the system as a whole: System S*D,B,C (D = 4, B = 2, and C = 3) has property 39 if the

228

L. H. Favela and A. Chemero

interactions are additive ([4 + 2]2 + 3 = 39) or property 192 if the interactions are multiplicative ([4 × 2]2 × 3 = 192). In the previous two examples, the brackets and exponent are intended to capture the nonlinear effects upon an interaction-dominant system, that is, how the place within a system can alter the properties within that system. With all that said, do not let the attempt to convey the flavor or interaction dominance in a mathematical way detract from the fact that in nature interaction-­dominant systems are quite mundane. Returning to component dominance and linearity for a moment: While it may be easier to understand linear relationships and systems that can be decomposed and their functions localized, outside of artifacts, such systems are rare in nature. In fact, interaction dominance and linearity “are surely the rule, not the exception … Not only in research, but also in the everyday world” (May, 1976, p. 467). Consider, for example, the human heart (Fig. 14.3). Even though it can be easy to apply linear thinking and decompose its anatomy into parts (e.g., aorta, ventricle, etc.), to do so would mischaracterize the phenomenon’s ontology. While brief observation makes clear that the heart is comprised of visually distinguishable parts, the functioning of the heart requires that such parts constitute an ontologically single system whereby the whole unit (anatomical heart) enables and constrains the activities of the parts and the parts enable and constrain the whole. Modifying one part will have repercussions throughout the whole and vice versa. As such, a heart is a single ontological, interaction-dominant system par excellence. This concludes our descriptions of mechanistic and dynamical explanations. As we mentioned at the start of this section, given the enormous size of the relevant literature, we have not intended to provide comprehensive accounts of mechanistic

Fig. 14.3  Human heart. Human heart is a biological phenomenon properly characterized as an interaction-dominant system. Here, biological phenomenon represented by a realistic anatomical diagram (left) and as an abstract diagram (right). The abstract diagram depicts the heart as a single system (top) that results from the interactions of parts (bottom), where the larger black arrows indicate mutual constraints from the whole to parts and parts to whole. (Left figure modified and reproduced with permission from Buchanan, 1933. Public domain)

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

229

or dynamical explanations. Instead, we focused on the characteristics that most clearly differentiate them, specifically, the emphasis on linearity and component dominance typical to mechanistic explanations, and the emphasis on nonlinearity and interaction dominance of dynamical explanations. Moreover, we focused on conceptual and methodological commitments that we take scientific practitioners to adhere to when their work expresses commitments to mechanistic or dynamical features. With these accounts given, we can proceed to our case study of the ontological pluralism that we think is appropriate to situated, embodied cognition.

14.3 A Case Study Starlings (sturnus vulgaris) are a family of small to medium-sized birds found in many places around the world including Asia, Europe, and North America. Among other things, starlings are known for their murmurations. A “murmuration” is a flock of starlings. Starling flocks can number in the hundreds to thousands. What makes these flocks noteworthy is their globally unified behavior. Murmurations are a coordination phenomenon in which interactions between individuals produce collective, large-scale patterns (Bialek et al., 2012, p. 4787). To watch a flock of starlings during the day is to see a single, dark, and large blob-like form moving about and taking on various shapes in the sky (Fig.  14.4). No matter the directions or shapes, the flock maintains movements that appear collective, synchronous, and Fig. 14.4  Starling flock with nearby predator. (Modified and reproduced with permission from M. Meraji, 2014. Retrieved https://commons. wikimedia.org/wiki/ File:Starling_flock_with_ nearby_predator.jpg. CC-BY-SA-4.0)

230

L. H. Favela and A. Chemero

ordered in a global manner (Cavagna et al., 2010, p. 11865). Explaining such collective animal behavior is very difficult, particularly if you want to explain how each bird has the nearest neighbors that it has at a particular time. This problem is further confounded if you want your explanation to be done with a few simple rules and variables. There are two explanations that have come to the forefront in the literature on bird flocking (Bialek et al., 2012; Cavagna et al., 2010; Cavagna, Queiros, et al., 2013a; Cavagna, Giardina, & Ginelli, 2013b; Kattas et al., 2012; Procaccini et al., 2011). One explanation is mechanistic and accounts for scale-free behavior and bird orientation. The other explanation is dynamical and accounts for scale-free behavior and bird velocity. Each will be presented in turn. It is crucial to keep in mind that both are applied at the same level of analysis, that is, the biobehavioral scale at which flocking occurs.

14.3.1 Mechanistic Explanation of Orientation in Bird Flocking When attempting to account for bird flocking behavior, the mechanistic explanation attempts to answer the following question: How does the behavior of bird X navigating amongst its nearest neighbors, Y1-Y7, result in scale-free correlations? “Scale-­ free” refers to the behavior of the flock as a whole such that the “behavioral change of one individual influences and is influenced by the behavioral changes of all other individuals in the group” (Cavagna et al., 2010, p. 11866). One aspect of this question involves providing an explanation for the orientation of each bird. The mechanistic explanation for this feature of the phenomenon posits the maximum entropy model (Bialek et al., 2012). This model accounts for the location of nearest neighbors by starting from the individual bird and explaining outwards towards the flock. The model also attempts to account for the effect of one bird’s movements as dispersing throughout the flock. As a result, the flock is understood primarily as a group of individual birds. In utilizing the maximum entropy model, the mechanist need only posit “[s]imple behavioral rules based on imitation” to account for coordination among nearest neighbors (Cavagna et al., 2010, p. 11869). The behavior based on imitation that the maximum entropy model ascribes to individual members of flocks are treated as tantamount to the behavior of a bar magnet. This is the kind of magnetic force that most people are familiar with, for example, the forces responsible for magnets sticking to refrigerator doors and two magnets repelling from one another when both of their positive charged sides are facing each other (Newman, 2008). Similarly, individual birds are treated as individual bar magnets: birds have a front and back end, and bar magnets have a positive and negative end. If individual birds are accurately treated as individual bar magnets, then a model of the individual birds based on such magnetic properties should be consistent with actual flock movements (i.e., maximum entropy distribution; Bialek et  al., 2012, p. 4786). Bialek et al. (2012) provide the following model (Eq. 14.4),

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

  P ( Si =

{ }



 J N     1 exp   ∑∑ si ⋅ s j   Z ( J ,nc )   2 i =1 j∈nci  

231

(14.4)

where j ∈ nci refers to individual bird j belonging to group nc with nearest neighbors i. As Bialek et  al. (2012) demonstrate, the model provides simulations that are indeed consistent with experimental correlations in large flocks of starlings. From this core model, further models exhibiting similar behavior as bar magnets, which underlie maximum entropy distribution, can provide other specific information, such as the orientation of multiple birds (Eq. 14.5), Cint =

1 N

N

1

   

∑ n ∑S ⋅S i =1

c j∈nic

i

j

≈

1 N

N

1

   

∑ n ∑S ⋅S . i =1

c j∈nci

i

(14.5)

j



where nc refers to orientations of neighborhood of birds. A powerful feature of the basic model in Eq. 14.4 is that in addition to providing the orientation of individual birds in a flock, it also exhibits the scale-free structure of the flock as a whole. Thus, modifications to the model, such as Eq. 14.5, allow for the comparison of flocks of various sizes without compromising the experimentally consistent scale-free behavior among flock orientations (Bialek et al., 2012). Whether tracking individual birds (e.g., j in Eq. 14.4) or two or more birds (e.g., nc in Eq. 14.5), it is the influence of the individual upon the group that is captured via the models. Since the models are consistent with experimental data (i.e., movements of real bird flocks), it is reasonable to understand that individual bird orientation is driven via a local (i.e., individual bird) process. Moreover, such local processes also contribute to the scale-free structure of the flock as a whole. By modeling the flocks as being composed of birds with magnet-like rules of behavior, the maximum entropy model accounts for the location of each nearest neighbor, with the added benefit of also capturing the flock’s apparent scale-free coordination. The explanation is mechanistic because the basic tenets of a mechanistic explanation remain, namely, that of linear decomposition and localization: the coordination behavior of the whole flock can be treated as the product of decomposable linear activities that are localizable in the individual birds. Once decomposed and localized, the coordination of the individual birds in the flock can be modeled in a manner similar to the behavior of bar magnets.

14.3.2 Dynamical Explanation of Velocity in Bird Flocking As with the mechanistic explanation, when attempting to account for bird flocking behavior, the dynamical explanation attempts to answer the following question: How does the behavior of bird X navigating amongst its nearest neighbors, Y1-Y7, result in scale-free correlations? As stated above, “scale-free” refers to the

232

L. H. Favela and A. Chemero

behavioral change of individuals influencing the flock as a whole. One aspect of this question involves providing an explanation for the velocity of each bird. The dynamical explanation for this feature of the phenomenon posits the superdiffusion model (Cavagna, Queiros, et al., 2013a). This model accounts for the velocity of each bird by starting from the flock as a whole and explaining inwards towards the individual. In other words, this approach treats the flock as the primary system to be explained, whereas the mechanistic model above treated the individual birds as requiring the primary explanation. The superdiffusion model also accounts for the effect of the velocity of the flock as a whole cascading throughout to affect the velocity of each individual bird. The dynamical explanation accounts for the velocity of a bird and its nearest neighbors in terms of their behavior resulting from diffusion taking its course: at each instant of time each bird is interacting with whatever birds have been brought there by their superdiffusive wandering throughout the flock (Cavagna, Queiros, et al., 2013a, p. 5). In physics, diffusion is the process by which particles seem to spread randomly, often to the point of equilibrium if environmental conditions suffice (Newman, 2008, pp. 15–16, 33–36). When superdiffusion occurs in scale-free systems, particles (or birds) will explore space via nonlinear processes (ben-­ Avraham & Havlin, 2004, pp. 57–72; Li et al., 2011). If bird flocks do indeed move via superdiffusion, then the model should demonstrate individual bird movements that are consistent with actual flock movements. Cavagna, Queiros, et al. (2013a) provide the following model (Eq. 14.6) to capture the diffusion processes of the flock,

δ R 2 ( t ) = Dt a



(14.6)



where δ is the change of the position of birds R at time t, a is the diffusion exponent, and D is the diffusion coefficient. Simulations of the model are consistent with experimental observations of bird flocks. The following model (Eq. 14.7) is derived from Eq. 14.6 in order to calculate bird velocities,

δ r 2 (t ) ≡

2 1 1 T −t −1 N  ∑ ∑ ri ( t0 + t ) − ri ( t0 ) T − t N t0 = 0 i =1 

(14.7)

where RCM(t) refers to the position of the center of the flock at time t and ri(t) = Ri(t) − RCM(t) refers to the position of a bird i in the flock. Along with the number of birds in the flock N and the time series length T, the velocity of the flock can be calculated. A powerful feature of Eq. 14.7 is that in addition to providing the velocity of the flock, it also maintains the scale-free structure of the flock as whole. This is accounted for by treating the individual birds as relating to the whole flock as if they are individual particles within a body of fluid. Moreover, the relationship of bird to flock and vice versa—as particle to fluid and vice versa—is nonlinear. Specifically, the velocity affects the flock via cascades that are not additive (i.e., linear), but multiplicative (i.e., nonlinear). Such a nonlinear relationship contributes

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

233

to explanations of velocity resisting the methods of decomposition and localization, namely, methods more appropriate to phenomena comprised of additively/linearly related components. By modeling the flock as a single superdiffusive system, the superdiffusion model accounts not only for the velocity of each bird, but also for the scale-free coordination of the flock. Moreover, accurate predictions can be made of the velocities of the birds within the flock, thereby negating the need to decompose the flock and localize individual birds (Cavagna, Queiros, et al., 2013a, pp. 7–8).

14.4 Ontological Pluralism In the preceding sections, we presented bird flocking as a case in the life sciences that is justifiably given both a mechanistic and dynamical explanation for various ontological features at the same biobehavioral scale. The mechanistic strategy treats the flock as a component-dominant system, namely, a collection of individual units. Bird orientation and scale-free coordination are explained via the maximum entropy model, which regards individual birds as following simple rules similar to individual magnets. The dynamical strategy treats the flock as an interaction-dominant system, namely, single complex system that emerges from the interaction of individual units. Bird velocity and scale-free coordination are explained via the superdiffusion model, which regards the flock as a single mass where individual birds are moved by and move the system via a diffusive process. Each explanation aims at the same general question: How does the behavior of bird X navigating amongst its nearest neighbors, Y1-Y7, result in scale-free correlations? However, each strategy provides a different kind of explanation for different features of the same phenomenon: bird location and bird velocity. All of the explanations (i.e., models) occur at the same scale of investigation, namely, the biobehavioral level at which murmurations occur. Is there a single explanatory strategy that will account for all aspects of this single scale? Specifically, is there justifiable reason to attempt to provide a mechanist account of velocity or a dynamical account of location? We do not believe so. Though there is overlap in some of the explained behavior (i.e., scale-free coordination), there are essential features of the phenomenon that are excluded by the explanations of each explanatory strategy. Consequently, the features of the bird flock at the biobehavioral scale are not all appropriately conceptualized as mechanisms (e.g., component dominant) or dynamical systems (e.g., interaction dominant). This is because the flock both has mechanisms and is a dynamical system at the same scale of investigation. It is reasonable to see hints of the predicament of underdetermination in science as an underlying issue (Stanford, 2021). Specifically, one could claim there is not yet enough evidence to arbitrate the issue. Perhaps a mechanist can say, “My strategy explains bird orientation as a product of component dominance, it will eventually explain bird location in that way too!” Perhaps a dynamicist can say, “My strategy explains bird velocity as a product of interaction dominance, it will eventually explain bird orientation in that way too!” Nonetheless, underdetermination

234

L. H. Favela and A. Chemero

is not an issue here, nor are issues concerning empirically equivalent theories (van Fraassen, 1980). When placed within different investigative frameworks, bird flocking can be conceptualized in different ways. For example, if the bird flock is a component-dominant system, then all of its features are additively and linearly related. On the other hand, if the bird flock is an interaction-dominant system, then all of its features are nonlinearly related. The challenge with applying one set of ontological commitments (e.g., component or interaction dominance) to this case is that there is empirical evidence for both being applicable to different features. It seems that there is not empirical equivalence among mechanistic and dynamical strategies because each strategy is applied to different features, namely, orientation and velocity. If certain ontological commitments are accurate, namely, if certain features of the flock are properly characterized as component dominant and others as interaction dominant, then it would be unproductive to attempt to fit the square peg of component-dominant phenomena into the round hole of dynamical explanations or the round peg of interaction-dominant phenomena into the square hole of mechanistic explanations. With that said, in light of the possibility of giving two equally viable explanations of different aspects of the same phenomenon, we believe that starting investigations from positions other than pluralism can result in the unjustifiable application of ontological commitments. Starting from a commitment to pluralism allows one to utilize various explanatory tools without the limitations that come with stringent adherence to single strategies. A lesson to take from a commitment to pluralism is that the study of a single phenomenon may simultaneously require multiple kinds of explanatory frameworks. This is not because the data is underdetermined or there is empirical equivalency. This is because the data from a single phenomenon are the result of very different kinds of ontology, for example, causes due to certain ontological structures that in one instance are best treated mechanistically and in another best treated dynamically. Insisting on just one style of explanation will leave us in the position of being able to explain bird orientation or bird velocity, but not both. We have shown that mechanistic explanations appropriately account for the ontology of some natural phenomena, here, bird orientation, which is a phenomenon not readily reducible to another explanatory framework, such as a dynamical one. We have shown that dynamical explanations appropriately account for the ontology of some natural phenomena, here, bird velocity, which is a phenomenon not readily reducible to another explanatory framework, such as a mechanistic one. Starling murmurations are a natural phenomenon that exhibit various ontological features amenable to different kinds of explanatory strategies: mechanistic approach for bird orientation and dynamical approach for bird velocity. Due to key ontological commitments of each, mechanistic strategies (e.g., component dominance) do not reduce to dynamical approaches, and dynamical strategies (e.g., interaction dominance) do not reduce to mechanistic approaches. Thus, starling murmurations require multiple explanatory strategies to explain multiple ontological kinds at the same biobehavioral scale of investigation. Moreover, the ontological constitution of starling murmurations require both mechanistic (i.e., orientation) and dynamical

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

235

(i.e., velocity) explanatory strategies to more fully understand their multiple ontological kinds.

14.5 Conclusion We view considerations of ontological pluralism in scientific investigative strategies as a virtue. Starting from a position of pluralism is a virtue not only in scientific inquiry but other forms of inquiry as well. It is likely that some see ontological pluralism as a bad position to subscribe to. Examples include those who adhere to the mechanism-only or dynamical-only explanatory framework points of view, as well as those who think that what makes dynamical explanations interesting can readily be subsumed by mechanistic approaches. Others likely not to be satisfied with ontological pluralism include those who want necessary and sufficient conditions for such concepts as “science,” “cause,” and “explanation.” However, those who are interested in the practice of science will see that there is no matter of fact as to which kind of explanation holds the position of primacy over all others (cf. Feyerabend, 1975/2010). Furthermore, those interested in the practice of science will also see that there is no adherence to a single theoretical commitment, come what may (cf. Quine, 1951). An honest investigator, be they a biologist, philosopher, or physicist, will accept the primacy of phenomena over commitment to ideology. Moreover, pluralism is not just the beginning, but also sometimes the end. That is to say, that there are different kinds of explanations of the same system; or, as we show here, different kinds of explanations—with some opposing claims; e.g., component vs. interaction dominance—of the different but closely related ontological aspects of the same system. This is not necessarily a case of underdetermination, or a failure to find the right mechanistic or dynamical explanation. This can be the end state. In this case, the orientations of individual birds in the flock are mechanistically explained, while their velocity is dynamically explained. There is no reason to think that starling murmurations are unique in this respect. The situated, embodied approach in the cognitive sciences utilizes both explanatory styles and ought to be committed to pluralistic ontology within its monistic metaphysics.

References Bechtel, W. (2008). Mental mechanisms: Philosophical perspectives on cognitive neuroscience. Routledge. Bechtel, W., & Abrahamsen, A. (2005). Explanation: A mechanist alternative. Studies in History and Philosophy of Biological and Biomedical Sciences, 36, 421–441. Bechtel, W., & Abrahamsen, A. (2010). Dynamic mechanistic explanation: Computational modeling of circadian rhythms as an exemplar for cognitive science. Studies in History and Philosophy of Science, 41, 321–333.

236

L. H. Favela and A. Chemero

Bechtel, W., & Abrahamsen, A. (2011). Complex biological mechanisms: Cyclic, oscillatory, and autonomous. In C.  Hooker (Ed.), Handbook of the philosophy of science. Volume 10: Philosophy of complex systems (pp. 257–285). North Holland. Bechtel, W., & Richardson, R. C. (1993/2010). Discovering complexity: Decomposition and localization as strategies in scientific research (2nd ed.). MIT Press. Beer, R.  D. (2000). Dynamical approaches to cognitive science. Trends in Cognitive Sciences, 4, 91–98. ben-Avraham, D., & Havlin, S. (2004). Diffusion and reactions in fractals and disordered systems. Cambridge University Press. Bialek, W., Cavagna, A., Giardina, I., Mora, T., Silvestri, E., Viale, M., & Walczak, A. M. (2012). Statistical mechanics for natural flocks of birds. Proceedings of the National Academy of Sciences, 109, 4786–4791. Brigandt, I., & Love, A. (2022). Reductionism in biology. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2022 ed.). Stanford University. Retrieved from https://plato. stanford.edu/archives/sum2022/entries/reduction-­biology/ Brown, C. (2007). Differential equations: A modeling approach. SAGE Publications. Buchanan, J. W. (1933). Elements of biology: With special reference to their role in the lives of animals. Harper & Brothers. Carello, C., & Moreno, M. A. (2005). Why nonlinear methods? In M. A. Riley & G. C. Van Orden (Eds.), Tutorials in contemporary nonlinear methods for the behavioral sciences (pp. 1–25). National Science Foundation. Retrieved from https://www.nsf.gov/pubs/2005/nsf05057/nmbs/ nmbs.pdf Cavagna, A., Cimarelli, A., Giardina, I., Parisi, G., Santagati, R., Stefanini, F., & Viale, M. (2010). Scale-free correlations in Starling flocks. Proceedings of the National Academy of Sciences, 107, 11865–11870. Cavagna, A., Queiros, S. M. D., Giardina, I., Stefanini, F., & Viale, M. (2013a). Diffusion of individual birds in starling flocks. Proceedings of the Royal Society B, 280, 20122484. Cavagna, A., Giardina, I., & Ginelli, F. (2013b). Boundary information inflow enhances correlation in flocking. Physical Review Letters, 110, 168107. Chemero, A. (2009). Radical embodied cognitive science. MIT Press. Chemero, A., & Silberstein, M. (2008). After the philosophy of mind: Replacing scholasticism with science. Philosophy of Science, 75, 1–27. Chialvo, D. R. (2018). Life at the edge: Complexity and criticality in biological function. Acta Physica Polonica B, 49(12), 1955–1979. Craver, C.  F. (2005). Beyond reduction: Mechanisms, multifield integration and the unity of neuroscience. Studies in History and Philosophy of Biological and Biomedical Sciences, 36, 373–395. Craver, C. F. (2007). Explaining the brain. Oxford University Press. Craver, C. F., & Alexandrova, A. (2008). No revolution necessary: Neural mechanisms for economics. Economics and Philosophy, 24, 381–406. Craver, C. F., & Bechtel, W. (2007). Top-down causation without top-down causes. Biology and Philosophy, 22, 547–563. Craver, C. F., & Darden, L. (2005). Introduction. Studies in History and Philosophy of Biological and Biomedical Sciences, 36, 233–244. Craver, C. F., & Darden, L. (2013). In search of mechanisms: Discoveries across the life sciences. University of Chicago Press. Craver, C., & Tabery, J. (2019). Mechanisms in science. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2019 ed.). Stanford University. Retrieved from https://plato. stanford.edu/archives/sum2019/entries/science-­mechanisms/ Dale, R. (2008). The possibility of a pluralist cognitive science. Journal of Experimental and Theoretical Artificial Intelligence, 20, 155–179. Dale, R., Dietrich, E., & Chemero, A. (2009). Explanatory pluralism in cognitive science. Cognitive Science, 33, 739–742.

14  Plural Methods for Plural Ontologies: A Case Study from the Life Sciences

237

Darden, L., & Craver, C. (2002). Strategies in the interfield discovery of the mechanism of protein synthesis. Studies in History and Philosophy of Biological and Biomedical Sciences, 33, 1–28. Favela, L. H. (2020a). Cognitive science as complexity science. Wiley Interdisciplinary Reviews: Cognitive Science, 11(4), e1525, 1–e1525,24. Favela, L. H. (2020b). Dynamical systems theory in cognitive science and neuroscience. Philosophy Compass, 15(8), e12695, 1–e12695,16. Favela, L. H. (2021). The dynamical renaissance in neuroscience. Synthese, 199(1–2), 2103–2127. Favela, L. H., & Chemero, A. (2016). The animal-environment system. In Y. Coelllo & M. H. Fischer (Eds.), Foundations of embodied cognition: Volume 1: Perceptual and emotional embodiment (pp. 59–74). Routledge. Feyerabend, P. (1975/2010). Against method (4th ed.). Vereso. Feyerabend, P. (1999). In B. Terpstra (Ed.), Conquest of abundance: A tale of abstraction versus the richness of being. The University of Chicago Press. FitzHugh, R. (1961). Impulses and physiological states in theoretical models of nerve membrane. Biophysical Journal, 1(6), 445–466. Glennan, S. (2005). Modeling mechanisms. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences, 36(2), 443–464. Glennan, S., & Illari, P. (Eds.). (2018). The Routledge handbook of mechanisms and mechanical philosophy. Routledge. Godfrey-Smith, P. (2010). Causal pluralism. In H. Beebee, C. Hitchcock, & P. Menzies (Eds.), The Oxford handbook of causation (pp. 326–337). Oxford University Press. Hodgkin, A. L., & Huxley, A. F. (1952). A quantitative description of membrane current and its application to conduction and excitation in nerve. The Journal of Physiology, 117(4), 500–544. Hofweber, T. (2021). Logic and ontology. In E.  N. Zalta (Ed.), The Stanford encyclopedia of philosophy (spring 2021 ed.). Stanford University. Retrieved from https://plato.stanford.edu/ archives/spr2021/entries/logic-­ontology/ Holden, J. G., Van Orden, G. C., & Turvey, M. T. (2009). Dispersion of response times reveals cognitive dynamics. Psychological Review, 116, 318–342. Ivancevic, V. G., & Ivancevic, T. T. (2007). Computational mind: A complex dynamics perspective. Springer. Izhikevich, E. M., & FitzHugh, R. (2006). FitzHugh-Nagumo model. Scholarpedia, 1(9), 1349. Kaplan, D. M., & Bechtel, W. (2011). Dynamical models: An alternative or complement to mechanistic explanations? Topics in Cognitive Science, 3, 438–444. Kaplan, D. M., & Craver, C. F. (2011). The explanatory force of dynamical and mathematical models in neuroscience: A mechanistic perspective. Philosophy of Science, 78, 601–627. Kattas, G.  D., Xu, X.-K., & Small, M. (2012). Dynamical modeling of collective behavior from pigeon flight data: Flock cohesion and dispersion. PLoS Computational Biology, 8(3), e1002449, 1–e1002449,15. Kelso, J. A. S. (1997). Dynamic patterns: The self-organization of brain and behavior. MIT Press. Kuhlmann, M., & Glennan, S. (2014). On the relation between quantum mechanical and neo-­ mechanistic ontologies and explanatory strategies. European Journal for Philosophy of Science, 4, 337–359. Lam, L. (1998). Nonlinear physics for beginners: Fractals, chaos, solitons, pattern formation, cellular automata and complex systems. World Scientific. Lange, M. (2017). Because without cause: Non-causal explanations in science and mathematics. Oxford University Press. Li, C., Zhao, Z., & Chen, Y.  Q. (2011). Numerical approximation of nonlinear fractional differential equations with subdiffusion and superdiffusion. Computers and Mathematics with Applications, 62, 855–875. Love, A.  C. (2012). Hierarchy, causation and explanation: Ubiquity, locality and pluralism. Interface Focus, 2, 115–125. Machamer, P., Darden, L., & Craver, C.  F. (2000). Thinking about mechanisms. Philosophy of Science, 67, 1–25.

238

L. H. Favela and A. Chemero

May, R. M. (1976). Simple mathematical models with very complicated dynamics. Nature, 261, 459–467. Mitchell, S. D. (2002). Integrative pluralism. Biology and Philosophy, 17, 55–70. Nagumo, J., Arimoto, S., & Yoshizawa, S. (1962). An active pulse transmission line simulating nerve axon. Proceedings of the IRE, 50(10), 2061–2070. Newman, J. (2008). Physics of the life sciences. Springer. Papineau, D. (2021). Naturalism. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2021 ed.). Stanford University. Retrieved from https://plato.stanford.edu/archives/ sum2021/entries/naturalism/ Piccinini, G., & Craver, C. (2011). Integrating psychology and neuroscience: Functional analyses as mechanism sketches. Synthese, 183, 283–311. Povich, M., & Craver, C. F. (2018). Mechanistic levels, reduction, and emergence. In S. Glennan & P.  Illari (Eds.), The Routledge handbook of mechanisms and mechanical philosophy (pp. 185–197). Routledge. Procaccini, A., Orlandi, A., Cavagna, A., Giardina, I., Zoratto, F., Santucci, D., Chiarotti, F., Hemelrijk, C. K., Alleva, E., Parisi, G., & Carere, C. (2011). Propagating waves in Starling, Sturnus Vulgaris, flocks under predation. Animal Behaviour, 82, 759–765. Quine, W. V. (1951). Two dogmas of empiricism. The Philosophical Review, 60, 20–43. Reiss, J. (2007). Do we need mechanisms in social science? Philosophy of the Social Sciences, 37, 163–184. Ross, L. N. (2022). Cascade versus mechanism: The diversity of causal structure in science. The British Journal for the Philosophy of Science. https://doi.org/10.1086/723623 Schaffer, J. (2018). Monism. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Winter 2018 ed.). Stanford University. Retrieved from https://plato.stanford.edu/archives/win2018/ entries/monism/ Silberstein, M. (2021). Constraints on localization and decomposition as explanatory strategies in the biological sciences 2.0. In F. Calzavarini & M. Viola (Eds.), Neural mechanisms: New challenges in the philosophy of neuroscience (pp. 363–393). Springer. Silberstein, M., & Chemero, A. (2013). Constraints on localization and decomposition as explanatory strategies in the biological sciences. Philosophy of Science, 80, 958–970. Stanford, K. (2021). Underdetermination of scientific theory. In E. N. Zalta (Ed.), The Stanford encyclopedia of philosophy (winter 2021 ed.). Stanford University. Retrieved from https:// plato.stanford.edu/archives/win2021/entries/scientific-­underdetermination/ Stepp, N., Chemero, A., & Turvey, M.  T. (2011). Philosophy for the rest of cognitive science. Topics in Cognitive Science, 3, 425–437. Sterelny, K. (1996). Explanatory pluralism in evolutionary biology. Biology and Philosophy, 11, 193–214. Sternberg, S. (1969). The discovery of processing stages: Extensions of Donders’ method. Acta Psychologica, 30, 276–315. Sternberg, S. (2011). Modular processes in mind and brain. Cognitive Neuropsychology, 28, 156–208. Stoljar, D. (2022). Physicalism. In E.  N. Zalta (Ed.), The Stanford encyclopedia of philosophy (Summer 2022 ed.). Stanford University. Retrieved from https://plato.stanford.edu/archives/ sum2022/entries/physicalism/ Strogatz, S.  H. (2018). Nonlinear dynamics and chaos: With applications to physics, biology, chemistry, and engineering (2nd ed.). CRC Press. Suppes, P. (1978). The plurality of science. In P. Asquith & I. Hacking (Eds.), PSA: Proceedings of the biennial meeting of the philosophy of science association, volume two: Symposia and invited papers (pp. 3–16). Philosophy of Science Association. van Fraassen, B. C. (1980). The scientific image. Oxford University Press. Wright, C. D., & Bechtel, W. (2006). Mechanisms and psychological explanation. In D. M. Gabbay, P. Thagard, & J. Woods (Eds.), Handbook of the philosophy of science. Volume 12: Philosophy and psychology (pp. 31–79). Elsevier.

Part III

Applications

Chapter 15

Chronic Pain, Enactivism, & the Challenges of Integration Sabrina Coninx and Peter Stilwell

Abstract  Chronic pain is one of the most disabling conditions globally, yet we are still missing a satisfying theoretical framework to guide research and clinical practice. This is highly relevant as research and practice are not taking place in a vacuum but are always shaped by a particular philosophy of pain, that is, a set of implicitly or explicitly prevailing assumptions about what chronic pain is and how it is to be addressed. In looking at recent history, we identify a promising trend from neuro-­ centrism to the application of the biopsychosocial model. Unfortunately, due to its limited theoretical foundation, the biopsychosocial model is too often implemented in a reductionist, fragmented, and linear manner. In particular, it remains too vague concerning the relationship between involved biological, psychological, and social processes. Sanneke de Haan prominently labeled this the integration problem. In this paper, we introduce five different facets of the integration problem that every philosophy of pain needs to address: (i) ontological, (ii) conceptual, (iii) explanatory, (iv) methodological, and (v) therapeutic. We develop an enactive theory of chronic pain and outline how far it provides solutions to these different integration challenges.

15.1 Introduction Pains are a fundamental part of our everyday life. Many of us know that striking awful feeling when stubbing a toe or the slowly increasing pressure of a headache after a day of work in front of a screen. A central characteristic of pains is that they S. Coninx (*) Department of Philosophy, VU Amsterdam, Amsterdam, The Netherlands e-mail: [email protected] P. Stilwell Faculty of Medicine and Health Sciences, School of Physical and Occupational Therapy, McGill University, Québec, Canada e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_15

241

242

S. Coninx and P. Stilwell

are identified through the first-person perspective. As Klein aptly formulates, “if there’s anything that’s typical of all pains – I say – it is that they feel like pains. That is how you identify them” (2022, p. 4). Interestingly, pain is a condition characterized exclusively in terms of what it is like to undergo such experience. This is reflected in the current definition of pain provided by the International Association for the Study of Pain (IASP) characterizing pain as a personal experience and giving epistemic authority to the person who regards and reports their experience as pain (Raja et al., 2020). Thus, researchers commonly refer to pain as a group of subjective experiences that feel the same or at least similar enough (Aydede, 2017). One central goal of pain science is to better understand the generation and maintenance of pain and to develop effective methods of treatment. So far, our knowledge on this matter must be considered insufficient in the face of the current social and personal challenges that persisting and reoccurring pains pose. It still holds that “millions of people – despite all our research – find themselves alone, disabled, and dispossessed by pain” (Morris, 1993). This proves pressing in the light of the high prevalence of chronic pains – commonly defined as pains lasting longer than 3 or 6 months – and their impact on quality of life (Breivik et al., 2006; Dahlhamer et al., 2018). As a particularly striking example, musculoskeletal pain is one of the most prevalent and disabling conditions worldwide (Hay et al., 2017; Vos et al., 2012) and one of the main reasons people seek care (Koleva, 2005; Mäntyselkä et  al., 2001). Yet, people living with chronic pain are often unsatisfied with the care they receive and left feeling frustrated, hopeless, and stigmatized (De Ruddere & Craig, 2016; Ojala et al., 2015; Toye et al., 2017). On that basis, more and more researchers agree that we need a new approach to chronic pain. As a first step, there is a need for a critical examination of the theoretical assumptions that currently prevail, implicitly and explicitly, about pain and its chronification in research and clinical practice. This is a question of which philosophy of pain dominates, understood as a consistent or inconsistent set of beliefs about what pain is, how it is caused, how it is to be investigated, and how it could be effectively treated. Realized or not, labeled or not, researchers and practitioners always have a philosophy of pain that influences the way they conceptualize, explain, assess, and treat pain. “Health sciences and healthcare practice are enabled, shaped and restricted by some tacit philosophical assumptions, of which practitioners should be aware” (Anjum et al., 2020, p. 3). Thus, the decisive question is not whether we have a philosophy of pain, but what this philosophy of pain looks like, especially in the context of understanding and treating chronic pain. Recently, the biopsychosocial (BPS) model (Engel, 1977, 1980) has become more and more popular in its application to chronic pain as it is purported to be a multifactorial approach that overcomes the reductive and dualist shortcomings of alternative pain theories. However, the BPS model is commonly interpreted and applied in reductionist, fragmented, and linear terms; with serious implications for research and clinical practice (Coninx & Stilwell, 2021; Mescouto et  al., 2020; Stilwell & Harman, 2019). These misinterpretations and misapplications are unsurprising as the original version of the BPS model was imprecise in its theoretical foundation, in particular concerning the relation between the involved biological,

15  Chronic Pain, Enactivism, & the Challenges of Integration

243

psychological, and social processes (Cormack et  al., 2022; Stilwell & Harman, 2019). Sanneke de Haan (2020a, b, c) prominently labeled this as the integration problem. In this paper, we develop an enactive approach to chronic pain with regard to the question of what a philosophy of pain might look like that satisfyingly addresses the integration problem. The paper proceeds as follows: in Sect. 15.2, we outline different manners in which chronic pain has been conceptualized and explained. This leads us to the BPS model as one of the most promising multifactorial approaches. We examine its application, understood as an indicator for the common (and often problematic) philosophy of pain that permeates research and clinical practice. Further, we introduce different facets of the integration problem: (i) ontological, (ii) conceptual, (iii) explanatory, (iv) methodological, and (v) therapeutic challenges. In Sect. 15.3, we develop an enactive approach to chronic pain. We first focus on the ontological assumptions of enactivism concerning the relation of biological, psychological, and social processes and apply them to the particular case of chronic pain. Subsequently, we revisit the remaining challenges of integration and outline the implication of an enactive approach on how we may conceptualize, explain, study, and treat chronic pain. Section 15.4 summarizes our results.

15.2 The Philosophy of Chronic Pain This section aims to provide the background knowledge concerning the philosophy of (chronic) pain that permeates current clinical research and practice, including its challenges. In Sect. 15.2.1, we provide a brief overview concerning the development of the definition and classification of chronic pain. In Sect. 15.2.2, we outline the historic development of some of the most central pain theories and their implications. In Sect. 15.2.3., we discuss the limitations of a neuro-centric perspective on chronic pain which motivates a shift towards more complex multifactorial approaches. In Sect. 15.2.4., we discuss the BPS model of pain as one of the most promising philosophies of pain; however, we outline problems that have developed partly due to its vagueness and limited theoretical underpinnings. In Sect. 15.2.5, we analyze the integration problem that lies at the bottom of those problems. In doing so, we introduce five challenges of integration: (i) ontological, (ii) conceptual, (iii) explanatory, (iv) methodological, and (v) therapeutic.

15.2.1 Chronic Pain: Definition & Classification In this section, we provide a more in-depth analysis of how chronic pains are commonly defined and classified. This enables us to provide a better understanding of why chronic pain constitutes such a relevant though challenging subject matter.

244

S. Coninx and P. Stilwell

In 1979, the IASP Subcommittee on Taxonomy sought a classification of chronic pain to make it more visible as a distinct phenomenon in comparison with acute pain, enable consistent use of terminology, and improve communication in clinical research and practice (Merskey & Bogduk, 1994). In reference to Bonica (1953), chronic pain is commonly defined in terms of its temporal persistence. Pain is considered chronic if it persists over a certain period of time which is understood as a normal healing phase. In this case, pain turns out to be rather maladaptive than beneficial as it loses its protective biological function. As Melzack and Katz put it: “Pain may be the warning signal that saves the lives of some people, but it destroys the lives of countless others” (2013, p. 1). Usually, pain is considered chronic when it persists for about three to 6 months; however, the exact time span might vary and be flexibly adapted (Merskey & Bogduk, 1994). While the work of the Subcommittee on Taxonomy constituted a milestone, its definition of chronic pain solely in terms of persistence still leaves much to be desired. First, there are in principle multiple time spans that could be chosen. It might seem that such distinctions are “purely functional and relatively arbitrary time posts that have little relation to underlying mechanisms” (Apkarian et  al., 2009, p. 81). Second, chronic pains are not just acute pains that have failed to ease (Hardcastle, 2014). In the process of chronicity, the lived experience of those concerned changes profoundly (Coninx & Stilwell, 2021; Svenaeus, 2015) and their quality of life is strongly affected (Breivik et  al., 2006; Rice et  al., 2016). Unfortunately, central aspects of the suffering of patients as well as the resulting social and economic challenges are often neglected (Goldberg & Summer, 2011). Along similar lines, researchers have argued that chronic pain should be considered a disease in its own right and receive a unified definition to facilitate adequate research, diagnosis, and care as well as to acknowledge the impact of chronic pain on patients (Raffaeli & Arnaudo, 2017). In response, an IASP Task Force set out to provide a new classification of chronic pain (Treede et al., 2015, 2019). By this new classification, chronic pain is still characterized in terms of temporal persistence as such criterion allows for a clear and operationalized definition (Treede et al., 2015, p. 1004). What is new is the classification of chronic primary pain in contrast to chronic secondary pain, such as persisting cancer, visceral, musculoskeletal, or neuropathic pains for which we can identify an underlying pathology (e.g. injury, lesion, disease, inflammation, obstruction): Chronic primary pain is pain in 1 or more anatomic regions that persists or recurs for longer than 3 months and is associated with significant emotional distress or significant functional disability (interference with activities of daily life and participation in social roles) and that cannot be better explained by another chronic pain condition. This is a new phenomenological definition, created because the etiology is unknown for many forms of chronic pain. (Treede et al., 2015, p. 1004)

The introduction of chronic primary pain as an independent diagnostic entity beyond a purely temporal and etiological characterization has two decisive advantages. First, it indicates that the best medical approaches to chronic pain do not necessarily rely on the identification and removal of underlying pathophysiology but might rather focus on the varying factors involved in the causation and maintenance of pains and their disabling consequences for patients (Siddall & Cousins, 2004).

15  Chronic Pain, Enactivism, & the Challenges of Integration

245

Second, the introduction of chronic primary pain may help validate patients’ experiences of pain in the absence of clear underlying pathophysiological cause, reduce their stigmatization as psychologically disturbed, and improve their access to and quality of treatment (Melzack & Katz, 2013; Raffaeli & Arnaudo, 2017; Treede et al., 2019). In the next section, we address the question of how pain has been modeled in different pain theories and the extent to which these theories may account for the existence of chronic pains, especially in the absence of definite underlying pathophysiology.

15.2.2 A Brief History of Pain Theories In this section, we aim to provide a brief, and by no means exhaustive, overview of the history of pain theories. Although it might seem that some of these pain theories should be long outdated, they still influence the current philosophy of pain. We outline two central shifts from (a) mechanistic and linear approaches that primarily concern neural structures in the periphery to (b) inclusion of multiple mechanisms interacting at the spinal cord, and (c) finally more recent approaches that investigate the complexity and plasticity of the brain. We respectively indicate the motivation for the further development of these theories. Descartes (1972) famously conceptualized the human body in analogy to a machine and in contrast to the immaterial mind. Although he had some nuanced ideas about pain and the interaction between the mind and body, researchers primarily latched onto his mechanistic writings that portrayed pain as the result of a sufficiently severe disturbance that is passed along physical tubes to the brain where pain is experienced; similar to a bell that rings in the brain when a string is pulled in the body, indicating the presence of a bodily condition. This view implies that there is a linear relationship between an originary disturbance and the resulting pain: the stimulation equals the pain. Further, the brain is ascribed only the role of a receiver of signals from a hardwired and specialized sensory projection system. This Cartesian heritage can be identified in the view of the pain system as a straightforward specialized communication system between body and brain that remained prevalent until the middle of the twentieth century (Melzack, 1973; Moayedi & Davis, 2013). As we shall see, the Cartesian heritage still shapes the philosophy of pain today. The Cartesian view has guided the research on pain with a focus on receptors and fibers involved in peripheral processing of physiological disturbances. Nociceptors have been identified as high-threshold sensory receptors of the peripheral somatosensory nervous system that are capable of transducing and encoding noxious stimuli that are damaging or threaten damage to normal tissues. Yet, there is no specialized and directly transmitting pain system with pain receptors and pain fibers; pain does not equal bodily disturbance (Wall & McMahon, 1986). There is no linear relation between the intensity, location, and quality of pain, on the one hand, and the severity, location, and type of stimulation, on the other (Coninx, 2020a, b, 2021).

246

S. Coninx and P. Stilwell

In 1965, Melzack and Wall introduced the Gate Control Theory. One central element is that peripheral input other than nociceptive activity can act upon neurons in the spinal cord and close a ‘gate’ to limit the transmission of nociceptive information via ascending pathways to the brain. This accounts for the pain-reducing effects of mechanical stimulation (Hillman, 1969) or vibration (Melzack et al., 1963). The traffic of nociceptive signals is also controlled by descending pathways projecting from the brain back to the spinal cord. That is, the transmission of nociceptive signals is only possible when the integration of all ascending and descending pathways exceeds a critical level in the dorsal horn (Melzack & Wall, 1965, 1982). In that, the Gate Control Theory proves successful to account for the absence of pain in the presence of nociceptive activity or the modulation of nociceptive information by means of diverse sensory and psychological factors such as attention, anxiety, stress, or social context (Melzack & Wall, 1965, 1982). Further, it ascribes a more active role to the brain (Melzack & Katz, 2013; Melzack & Wall, 1982). By itself, the Gate Control Theory however fails to account for cases in which pain is experienced in the absence of any nociceptive activity (Melzack, 1989). Complementing the Gate Control Theory, Melzack (1989, 2001) introduced the concept of a body-self matrix, i.e. a widely distributed network of dynamically interacting brain structures that receive sensory, affective, and cognitive input and produce a multidimensional pain experience, action patterns, and stress-regulatory reactions. That is, pain is associated with the ongoing dynamic coordination and reconstruction of neural activity across a non-specific brain system (Apkarian, 2017; Garcia-Larrea & Peyron, 2013; Iannetti & Mouraux, 2010). This constitutes a clear rejection of the traditional Cartesian view considering pain as the result of a linear system that simply transmits sensory signals about tissue pathologies (Melzack & Katz, 2013). Regarding chronic primary pain, the concept of the body-self matrix has created space to better account for the complexity and variability of involved neural processes. Acute and chronic pain may correlate with different activity patterns in the brain (Lee et al., 2021; Ploner et al., 2017; Wager et al., 2013) while alterations of neuroanatomy and connectivity appear to play a crucial role in chronification (Hashmi et al., 2013; Mansour et al., 2014; Schmidt-Wilcke, 2015). Further, such approach allows in principle for the influence of a variety factors on the activity and plasticity of the body-self matrix (Melzack & Katz, 2013). This development of pain theories has found its way into various areas of pain research as well as clinical practice and patient education (most prominently see Moseley & Butler, 2017).

15.2.3 Neuro-Centrism The historical overview presented leads us to a philosophy of pain in which the brain is considered to play a crucial active role. Without diminishing the great contribution of neuroscientific research in the field of pain research and management, we should remain skeptical about employing on that basis a neuro-centric

15  Chronic Pain, Enactivism, & the Challenges of Integration

247

perspective that ascribes an exclusive or at least prioritized role to the brain in understanding and treating chronic pain (Stilwell & Harman, 2019). In this section, we consider five reasons for such skepticism. First, in focusing on the neural activity of the brain, we may overlook or downplay other biological factors that significantly contribute to the causation and maintenance of pain. This involves genetic (Denk et  al., 2014; Vehof et  al., 2014), epigenetic (Descalzi et al., 2015; Stenz et al., 2021), as well as endocrine and immunological processes (Melzack, 1999) which can also function as useful indicators in pain prevention, diagnosis, and treatment (Reckziegel et al., 2019; Vachon-Presseau et  al., 2019). All of these processes dynamically interact along the brain, spinal cord, and periphery – cross-cutting boundaries between brain-bound and non-brain-­ bound systems (Chapman et al., 2009; Melzack & Katz, 2013). Thus, an exclusive study of the brain is too narrow a focus and might lead to a loss of relevant knowledge about the physiology of chronic pain, in particular concerning the intimate relation between neural and non-neural processes that are not as strictly separable as it might seem at first. Second, we have good reasons to reject a neuro-reductionist perspective, which might be encouraged by neuro-centrism, for similar reasons that speak more generally against a bio-reductionist perspective. Defenders of reductionism may argue for the type identity of chronic pain with specific pathology, for example, a disease of the brain (Borsook et al., 2010). This perspective is related to the general effort in medicine to reduce clinical conditions to a single underlying abnormality that can be effectively addressed by highly specialized, and at best selective, interventions.1 While this reductionist agenda might appear attractive due to its coherence and simplicity, it has proven unsuccessful in the application to various conditions (Adam, 2013; Fullana et al., 2020; Kendler, 2005, 2012), including chronic pains (Stegenga, 2018, pp. 61–68; Sullivan, 2014). There are hardly any successful mono-­ causal explanations for chronic pain conditions and there are no highly specialized interventions that can be effectively applied to them (Melzack & Katz, 2013). That is because pains are often not associated with concrete pathophysiological alterations in the body (Kosek et al., 2021) and, to date, researchers are unable to identify a distinct neural correlate of pain (Coninx, 2021; Corns, 2020). Thus, reductionism in terms of type identity presupposes an oversimplification of how chronic pain is brought about that is incompatible with their multiple realization/causation.

 A paradigmatic success story of this approach is the case of neurosyphilis which has been identified with a specific bacterial infection effectively targeted by the administration of arsphenamine. For many conditions, the focus is to identify a corresponding genetic or neurophysiological marker. This constitutes a general shift from the identification of symptom patterns to the search for pathology understood as the common underlying cause or constitutive basis of the phenomenon in question (Goodkind et al., 2015; Insel & Cuthbert, 2015; Kandel, 2018). While neuro-reductionism presupposes the identification of abnormalities in the brain, bio-reductionism might also account for deficits in a larger neuro-endocrine-immunological system, for example. The decisive commonality is that a single disease state must be found that is necessary and sufficient for the respective clinical condition to occur.

1

248

S. Coninx and P. Stilwell

Third, a neuro-centric perspective seems to ignore or minimize the influence of psychological (Martinez-Calderon et al., 2020; Meulders, 2019) and social factors (Rossettini et al., 2018; Slade et al., 2009) in the chronification of pain. In principle, neuro-centrism might allow for other factors to be involved and one might even reconcile the impact of non-biological factors with a reductionist perspective, “as long as one maintains that the brain is the real mechanism of change here” (de Haan, 2020a, p. 28). According to this view, chronic pains are primarily grounded in physiological processes while psychological and social factors are considered modulatory factors, distal causes, or the results of pain. While various studies have shown that alterations in brain anatomy and connectivity correlate with chronic pain, these neural correlates could in principle be realizations, causes, or effects of chronic pain and it remains up for debate of why we should privilege the role of brain processes in contrast to psychological or social factors (Borsboom et al., 2018; de Haan, 2020a, pp. 27–30; Varga, 2015, p. 34).2 Fourth, Stilwell and Harman (2019) argue that neuro-centrism in pain research and practice commits a mereological fallacy: it falsely attributes a property of a whole system to one of its parts. Neuro-centrism seems to suggest that it is the brain that suffers from pain; by contrast, it is a person in a certain context who experiences pain (for a similar argument see Kendler, 2005). We are not treating brains but people who are located in a certain environment. Further, exclusively referring to a neurophysiological process is insufficient to determine that such process constitutes a disorder or pathology (Stegenga, 2018) and to explain why chronic pain often constitutes such a disabling condition for those who suffer from it (Sullivan et al., 2013). That is, even if we could identify a type-identical neurophysiological state, corresponding neuro-centric approaches may still fall short to explain how these states relate to the broader clinical phenomena that chronic pain constitutes and the alterations in patients’ experience, self-concept, and agency that come with it (Hoffman & Zachar, 2017). Fifth, neuro-centrism has serious pragmatic implications in clinical practice and research. If we have an overly strong focus on the brain, we may overlook certain treatment options, paradoxically stigmatize patients, and minimize their engagement. Even if we can identify an underlying brain abnormality, the targeting of psychological and social factors does not lose relevance (Eronen, 2021). The brain is not a particularly useful target for direct intervention and we lack a good reason to believe that more biologically-focused treatment, such as pharmaceuticals, provide us with the best outcomes (Stegenga, 2018), as reflected in clinical practice guidelines for common pain conditions, such as chronic back pain (Qaseem et al.,

 One might reply that brain processes should be prioritized because they are constitutive of pains. We do not aim to defend any position concerning the constitution of pain. However, even if we accept that the minimally sufficient constitutive basis for singular pain experiences are brain processes, it still remains up for debate whether changes in brain anatomy, connectivity, and chemistry are realizations, causes, or effects of the chronification of pain. It is exactly the relation between biological, psychological, and social aspects in the processes of chronification that we are interested in, and in this regard, there is no principle reason of why brain processes should be privileged. 2

15  Chronic Pain, Enactivism, & the Challenges of Integration

249

2017). Further, the trend towards giving neuro-centric explanations for patients’ experiences may have unwanted, paradoxical outcomes. There is some evidence that brain- and biologically-focused explanations and beliefs may increase rather than decrease stigma and negative beliefs towards health conditions (Berent & Platt, 2021; Larkings & Brown, 2018) and negatively impact patient outcomes (Schroder et al., 2020). Patients might also feel that they are not taken seriously or adequately represented in social dialog (Sullivan et al., 2013) as they are reduced to passive targets of treatment (Rocca & Anjum, 2020). Finally, neuro-centrism as a general heuristic can affect progress in pain research. In assuming that only the brain (or body) plays a decisive role, alternative explanations and targets for intervention are ignored from the outset. For example, this may distract from the consideration of socioeconomic and sociocultural aspects and the critical examination of the health care system itself (Mescouto et al., 2020). Defenders of neuro-centrism might surely provide a reply to some of the outlined challenges. For example, they might argue that future science will potentially reveal a single process – cranial or extra-cranial – underlying chronic pain. While being rather pessimistic about this particular aspect, we have provided general arguments for an at least skeptical attitude towards brain-based and biologically focused approaches. The previous considerations indicate that instead of reducing the complexity of chronic pain and the variability of involved biological, psychological, and social factors, we need an account that embraces these aspects. This leads us in the next section to more complex multifactorial pain theories, with the BPS model as the most prominent and promising.

15.2.4 The Biopsychosocial Model In this section, we outline basic assumptions of the BPS model and discuss problematic tendencies in its interpretation. We aim to highlight general biases in clinical research and practice and show that the potential for corresponding misapplications of the BPS model is grounded in its vagueness and limited theoretical grounding. Engel (1977, 1980) introduced the BPS model arguing for a more humanistic approach to patient care and that complex health issues cannot be reduced to pathology or disease as they are determined by biological, psychological, and social processes. In reference to General Systems Theory, he conceptualized these processes as located at vertically ordered levels, either considered as ‘layered’ (see Fig. 15.1 for illustration) or ‘nested’ domains (see Fig. 15.2 for illustration).3 Lower levels concern biological processes (e.g. molecular, cellular, or neural activity). Middle  Engel used the term ‘levels’ to relate biological, psychological, and social processes. We prefer to use the term ‘domain’. This is because the notion ‘level’ is more strongly associated with a particular ontological interpretation, that is, of a vertical order as depicted in Fig.  15.1 and Fig.  15.2, whereas we aim to employ the term ‘domain’ in a more neutral way. 3

250

S. Coninx and P. Stilwell

Fig. 15.1  Vertical hierarchy of layered domains: schematic and simplified illustration of the hierarchy of biological, psychological, and social systems as ‘layered’ domains. (Inspired by Engel (1980), Fig. 1 ‘Hierarchy of Natural Systems’) Fig. 15.2 Vertical hierarchy of nested domains: schematic and simplified illustration of the hierarchy of biological psychological, and social systems as ‘nested’ domains. (Inspired by Engel (1980), Fig. 2 ‘Continuum of Natural Systems’)

levels concern person-related processes (e.g. experiential, cognitive, or behavioral changes). Higher levels concern social processes (e.g. practices in families, subcultures, or nations). Engel viewed this as a ‘vertical stacking’ where larger (higher) units are more complex and superordinate to less complex smaller units (Engel,

15  Chronic Pain, Enactivism, & the Challenges of Integration

251

1982, p. 803). Systems at higher levels are composed of systems of lower levels while each level requires unique methodological access. In scientific work, the investigator generally is obliged to select one system level on which to concentrate, or at least at which to begin, their efforts. For the physician that system level is always person, i.e., a patient (Engel, 1980, p.  537). Further, Engel suggested that the hierarchical levels are always interconnected by “material and information flow” (Engel, 1980, p. 537) across their boundaries. The BPS model was not developed with particular consideration of chronic pain, but it is now widely applied in this context (Cohen et al., 2021; Foster et al., 2018; Gatchel et al., 2007). In its original version, Engel presented the BPS model as a genuinely humanistic and multifactorial framework. However, many applications in the pain literature do not align with this, as they tend towards a reductionist, fragmented, and linear interpretation (Cormack et al., 2022). We take this to characterize a common bias in the current philosophy of pain, once again, understood as the set of beliefs about chronic pain that prevail implicitly or explicitly in research and clinical practice. First, the BPS model is still too often conflated with a bio-reductionist or neuro-­ reductionist model, as evident in prevalent measurement choices, diagnostic focus, and treatment selection (Mescouto et al., 2020; Stilwell & Harman, 2019). In closer examination, the multi-domain approach inherent to the BPS model frequently collapses into a uni-domain approach which, again, prioritizes the search for pathology and corresponding specialized interventions. This implies a reduction of the psychological and social to the biological. The BPS model is thus reinterpreted in that the vertical hierarchy of higher and lower levels is accepted, and that the biological domain is more fundamental – providing the ‘real’ or ‘root’ cause of chronic pain. Initial issues of such reductionist approach are indicated in the previous section. Second, even if the biological, psychological, and social domains are considered equally relevant, their examination is often fragmented. That is, a trichotomization arises between apparently separate domains that become the subject of independent research, diagnosis, and treatment while it remains largely unclear how they are connected. Engel’s characterization of ‘material and information flow’ proves hardly informative (de Haan, 2020b; Svenaeus, 2021). This can lead to an artificially splitting of patients into their physiological, psychological, and social characteristics, a plurality of potentially contradictory diagnoses, and thus a patchwork of independent approaches in pain research and health care (Anjum et  al., 2020; Cabaniss et  al., 2015; Stilwell & Harman, 2019). What would be needed is an approach that is not only complex, in that it rejects reductionist assumptions concerning the prioritization of biological processes, but also integrative, in that it seeks to examine the relationship between biological, psychological, and social aspects rather than aiming for multiple parallel approaches (Bolton & Gillett, 2019). Third, the BPS model is often implemented in a linear manner. That is, uni-­ directional causal pathways are assumed between involved factors, which themselves are considered as insensitive to context or temporal unfolding and, thus, strictly decomposable (Rocca & Anjum, 2020; Stilwell & Harman, 2019). This implies that we can understand a complex phenomenon, such as chronic pain, by

252

S. Coninx and P. Stilwell

analyzing the involved factors in isolation. This picture seems empirically inadequate concerning the relation of processes within and across domains as they are dynamically intertwined along different time spans (Kendler, 2012). As a consequence, the involved processes themselves cannot be easily differentiated and there are limitations to their decomposability (Boer et al., 2021). Further, it seems misguided to start from a static picture of ‘layered’ or ‘nested’ domains that at best interact in sequential order. Instead, we need a more dynamic model that creates space for context- and time-sensitive interaction along multiple feedback-loops (Bolton & Gillett, 2019; Lehman et al., 2017). Taken together, the recent and widespread philosophy of pain in research and practice reveals reductionistic, fragmented, and/or linear tendencies. Engel’s work on its own does not offer strict principles that prevent such interpretations and applications, as central aspects concerning the relationship between biological, psychological, and social factors are kept too vague or do not offer sufficiently detailed and coherent theoretical orientation (Benning, 2015; Bolton & Gillett, 2019; Ghaemi, 2009; Kendler, 2010). Instead, we need a truly complex, integrative, and dynamic model. In the following section, we discuss these findings in the context of the integration problem. Before doing so, a few words shall be said about multifactorial alternatives to the BPS model, in particular, network models. Network models aim to account, among others, for the complex web of interconnected processes involved in psychopathologies (Borsboom, 2017; Borsboom et al., 2018; Cramer et al., 2016). Mental disorders are conceptualized as arising from the dynamic interaction between a network of symptoms causally connected through biological, psychological, and social processes. Network models rely on already existing lists of symptoms provided by diagnostic frameworks and also allow non-­ symptoms, such as environmental factors, to become part of the relevant network architecture (Borsboom et al., 2018). To suffer from disorder means to be trapped in a relatively stable, self-sustaining network of causally interacting factors. For present purposes, it might be sufficient to note that, despite their advantages, network models are, by themselves, insufficient to provide a complex, integrative, and dynamic approach to chronic pain. Network models fail to show how the symptoms and non-symptoms that are considered part of a network architecture are precisely related (e.g. common cause vs. uni-directional causation vs. bi-directional causation vs. (partial) realization) (Boer et  al., 2021; de Haan, 2020a, pp.  41–43; Kästner, 2023). That is, they fail to adequately address the integration problem.

15.2.5 Five Facets of Integration In this section, we take a closer look at the integration problem. Sanneke de Haan (de Haan, 2020a, pp. 36–43; 2020b, c) introduces the integration problem as one of the most central challenges that multifactorial approaches in clinical medicine face. In particular, de Haan focuses on the integration problem in the field of psychiatry. In a nutshell, to address the integration problem we need to characterize the relation between those factors involved in the generation and maintenance of a clinical

15  Chronic Pain, Enactivism, & the Challenges of Integration

253

condition, assuming that there is a variety of relevant factors of different kinds – biological, psychological, and social. Given the investigations outlined in the previous sections, it is obvious that such integration problem equally applies to chronic pain. What we aim to highlight is that there are different interrelated facets to the integration problem that require a more fine-grained examination. (i) Ontological Challenge: At its core, the integration problem is an ontological matter raising the question of how processes of different kinds relate to each other. Can biological, psychological, and social factors be reduced to each other? Are these factors connected in linear ways or do they reciprocally affect each other? Are they sensitive to contexts and temporally unfolding dynamics? Do the investigated factors partially overlap, or can they be modelled as perfectly mapping layers? Do the biological, psychological, and social domains characterize independent ontological spaces, or are they perspectives on one and the same process? (ii) Conceptual Challenge: This challenge raises the question of how to conceptualize chronic pain in the light of the integration problem. Is chronic pain a brain disease or a psychological disorder? And what role is attributed to the social context in the definition of chronic pain? What is chronic pain assuming that temporal persistence does not provide on its own an informative and exhaustive understanding? (iii) Explanatory Challenge: This challenge addresses how we explain the occurrence of chronic pain, given the plurality of biological, psychological, and social factors involved. Do we need to rely on multiple independent explanations for different scientific purposes? Or, can we reach a single coherent explanation of chronic pain that unifies insights from all disciplines at once? (iv) Methodological Challenge: This challenge addresses the constraints of how we should study chronic pain. Which methods are best suited to investigate chronic pain? Is there any method to be prioritized? How can we integrate insights from multiple research areas which employ different methods while targeting the same phenomenon? And, how could a resulting complex set of data be of practical value? (v) Therapeutic Challenge: The therapeutic challenge concerns how we can best treat chronic pain. What is the best target for intervention? What role does the relationship between patients, practitioners, and institutional structures play for the success of treatment? How do we account for the uniqueness of each patient and still do justice to the ideal of science where clinicians base assessment and treatment decisions on commonalities and statistical averages? The integration challenge is typically formulated in terms of the ontological issues, targeting the BPS model as originally developed by Engel. Aftab and Nielsen (2021) argue that Engel’s main concern was not to establish an ontology but to prove psychological and social aspects as worthy of integration into the scientific realm, that is, as relevant for our understanding and explanation of health and disease, selection of methodological tools, and choice of treatment. They further argue that addressing the ontological challenge does not provide an answer to the question of how biological, psychological, and social factors are to be integrated, for example, in

254

S. Coninx and P. Stilwell

coherent explanations or therapeutic practice. Indeed, Engel hardly explicitly addresses ontological issues in the development of the BPS model and an answer to the ontological challenge does not immediately provide an answer to the others, as this might require to also consider epistemic, normative, or pragmatic aspects. However, ontological issues play a crucial role in the background (Bolton, 2021) and we surely need to address them somehow to address the other facets of the integration problem. That is, we need an approach that “defines the foundational theoretical constructs  – the ontology of the biological, the psychological and the social  – and especially the causal relations within and between these domains” (Bolton & Gillett, 2019). In our view, the ontological, explanatory, conceptual, methodological, and therapeutic facets of the integration problem are not independent. In particular, the ontological assumptions that define our philosophy of pain shape how we explain, conceptualize, study, and treat pain. Thus, a central step in initiating far-reaching and stable change in different areas of research and clinical practice is to change how we think about how biological, psychological, and social processes relate from an ontological perspective (de Haan, 2020b; Rocca & Anjum, 2020). Again, clinical research and practice do not take place in a vacuum, but they are shaped by theoretical assumptions and ideals.

15.3 An Enactive Approach to Chronic Pain: Revisiting the Five Challenges of Integration Our aim in the following is to introduce an enactive approach to chronic pain and indicate how it might fruitfully contribute to a change in the philosophy of pain that permeates all facets of the integration problem. In Sect. 15.3.1, we provide a brief introduction to enactivism. In Sect. 15.3.2, we outline what we consider the minimal assumptions of an enactive ontology. In Sect. 15.3.3, we apply these assumptions to chronic pain and explore their implications for addressing the ontological challenge. We argue that an enactive ontology can provide the needed theoretical foundation for a complex, integrative, and dynamic approach. In Sect. 15.3.4, we revisit on that basis the remaining four challenges of integration – explanatory, conceptual, methodological, and therapeutic.

15.3.1 Main Pillars of Enactivism In this section, we briefly outline the main pillars of enactivism. To provide a universal definition of enactivism that does justice to all the nuances of those accounts that are members of this theory family is hardly possible. Enactivism is a rapidly evolving movement with many strands, such as autopoietic enactivism (Di Paolo

15  Chronic Pain, Enactivism, & the Challenges of Integration

255

et al., 2018; Di Paolo & Thompson, 2014; Thompson, 2007; Varela et al., 1991), sensorimotor enactivism (Noë, 2004; O’Regan & Noë, 2001; O’Regan, 2011), and radical enactivism (Hutto & Myin, 2013, 2017). We do not aim to provide a systematic historical overview or detailed analysis of the various strands of enactivism and other related traditions (see Baggs & Chemero, 2021; Käufer & Chemero, 2021; Popova & Rączaszek-Leonardi, 2020; Ward et al., 2017; Ward & Stapleton, 2012). What we consider most central for present purposes is that enactivism rejects neuro-centrism and motivates to employ a broader perspective taking into account the entire person, including brain and body, in interaction with their environment (Di Paolo & Thompson, 2014; Fuchs, 2017; Varela et al., 1991). Concerning our target phenomenon, this means that (chronic) pain is (a) enabled and constrained by the dynamic coupling of the neural and non-neural body (embodiment), (b) crucially dependent on the bi-directional relation between person and environment (embeddedness), and (c) bound and brought forward by a person’s activity and actions in the environment (enactment) (Coninx & Stilwell, 2021; Miyahara, 2019; Stilwell & Harman, 2019; Tabor et al., 2017).4 Enactive approaches typically focus on the first-person perspective and are deeply intertwined with the phenomenological tradition. This is of particular relevance to chronic pain as there is a need to better understand the lived experiences of patients (Stilwell & Harman, 2021) and how chronic pain affects the manner in which people attune to their environment, including changes in the perception of themselves and their environment (Coninx & Stilwell, 2021). However, in this paper, the primary aim is to address the relation between the biological, psychological, and social factors in the process of pain chronification, rather than to characterize in detail the changes in experience that paradigmatically occur during such process and profoundly affect the lives of those concerned (for more detail see Coninx & Stilwell, 2021).

15.3.2 Minimal Ontological Assumptions In the following, we outline five closely related aspects that we consider the minimal assumptions of enactive ontology: naturalism, emergence, asymmetric determination, dynamic organization, and spatio-temporal scaling. We discuss these concepts here in the broader context of the relationship between biological, psychological, and social processes to be translated into the more specific context of chronic pain in the subsequent section. We focus the discussion mainly on the work of de Haan (2020a, b, c, d) as “a genuinely novel and outstanding branch on the enactivist tree” (Bruineberg, 2021, p. 1). Further connections are made to the work of Aftab and Nielsen (2021), Fuchs (2020), and Thompson and Varela (2001).  We do not consider extendedness as a main pillar of enactivism, although some authors in the enactive debate commit to the extendedness of certain phenomena, with partly deviating definitions of extendedness. 4

256

S. Coninx and P. Stilwell

Naturalism   Enactivism is a naturalistic theory in that it is compatible with the assumption that everything is made of natural components. All properties, behaviors, and structures that can be found in our world are instantiated by systems composed of the same matter. There are no supernatural entities, mystic powers, or vital forces involved. There is no tripartite structure of ontology as biological, psychological, and social processes are all located in the same ontological realm (Aftab & Nielsen, 2021). Thus, enactivism rejects the dualistic assumption of a further substance over and above matter. It is compatible with monistic versions of naturalism, but at the same time incompatible with certain reductionist versions of monism as it entails a commitment to emergence (de Haan, 2020a, pp. 65–66). Emergence  Enactivism entails emergence in that new properties, behaviors, and structures may occur in complex systems given the particular organization and interactions of their components. This kind of novelty presupposes ‘qualitative emergence’: a system instantiates characteristics that are not present in any of its parts and, thus, may count as qualitatively new (Gillett, 2016, p. 176). In general, the notion of ‘emergence’ is highly debated and a full exploration of its diverse meanings would exceed the scope of this paper (Gillett, 2002, 2016; O’Connor & Hong, 2020; Stephan, 2006). Our provided characterization of emergence merely implies that the constellations and interactions of components can bring forth properties, behaviors, and structures in a complex system that do not exist in its parts. This is compatible but does not necessarily imply further ontological commitments of ‘stronger’ variants of emergence theories (e.g., de Haan, 2020a, pp. 113–121). This idea is implemented in the enactive tradition in different ways (e.g., Di Paolo & Thompson, 2014; Froese & Di Paolo, 2009; Thompson, 2007; Varela et al., 1991). Most prominently, organisms are considered self-organizing units which interact with their environment to maintain themselves and show characteristics that their physiological components, such as neurons, genes, and hormones, do not instantiate. Only the arrangements of and relations between these components bring about the emergent characteristics of the organism as a whole. Similarly, we may consider the emergence of new properties, behaviors, and structures in social systems that involve the interaction of multiple organisms, although they are not present in the individual organisms themselves. The biological, psychological, and social domains characterize systems of growing complexity, such as brains, organisms, or societies, with properties, behaviors, and structures emerging in the specific constellation of matter. These systems can be described in more local or global manners as we can zoom in or out, employing a more or less spatially and temporally extended perspective (de Haan, 2020a, pp. 97–104). Thus, we may address characteristics of more local systems (e.g. neural activity patterns), the person (e.g. expectations about the future), and more global systems (e.g. socio-cultural practices). At the same time, these systems cannot be reduced in the sense that they implement properties, behaviors, and structures that are not found in more local systems. This entails a rejection of what we might label as naïve reductionism, i.e. the assumption that all things just differ in terms of the

15  Chronic Pain, Enactivism, & the Challenges of Integration

257

number, arrangement, and movement of the natural components they are made of. In contrast, we may assume following an enactive framework that although there is nothing over and above matter, this does not imply that everything is nothing but matter (de Haan, 2020a, p.  106). That is, although everything is supposed to be made out of the same natural component, we cannot reduce chronic pain – an emergent experience – to just biological processes.5 Asymmetric Determination  As a naturalistic approach, enactivism needs to assume that the emergent properties, behaviors, and structures of global systems depend or at least supervene on the characteristics of the simpler components of which the system is composed (Stephan, 2002). All other things being equal, a system with the exact same organization of components will reveal the exact same emergent characteristics. There can be no difference in the system without corresponding differences in the involved parts, their arrangements, and interactions. Further, it is not excluded that different components can bring about systems with the same emergent characteristics. The same characteristics of a global system can be realized by different constellations of more local systems. The characteristics of the components can in turn be determined by the emergent characteristics of the system as a whole. Local systems may derive their properties and configuration, and in parts their existence, from being part of more global systems (de Haan, 2020a, p. 95). The brain functions the way it does because it is part of a more complex organism. Psychological processes are bound to be instantiated by a person engaging with their environment. This implies that the components of a system are not the same when they occur in isolation or when they are part of a larger system that differs in terms of the involved components, their constellations, and interactions. This is a rejection of an atomistic view assuming that the parts of a system could be individuated solely in terms of their intrinsic properties, independent of the context in which they are embedded (Burnston, 2021). By contrast, the components of a system do not remain unchanged when being part of such system. They are distinct but not separate and as such not strictly composable and decomposable. It follows that there are two directions of determination: local-to-global and global-to-local (Thompson & Varela, 2001). The properties, behaviors, and structures of a more global system are determined by the properties and organizations of the more local systems that it is composed of. The global properties, behaviors, and structures govern, constrain, or enslave local characteristics and interactions. Interestingly, there exists an asymmetry in the determination between local and

 Two terminological remarks are needed. First, we use de Haan’s distinction of ‘global’ and ‘local’ systems, in contrast to ‘higher-level’ and ‘lower-level’ systems. This is because we want to draw a clear distinction with respect to the vertically stacked levels, understood as universal layers (Fig. 15.1) or perfectly mapping parts and wholes (Fig. 15.2), as originally introduced by Engel. Second, we often refer to ‘more global’ and ‘more local’ systems, instead of simply ‘global’ and ‘local’ systems, to emphasize that the distinction of what counts as the global or local system depends on our epistemic perspective and scientific interest. 5

258

S. Coninx and P. Stilwell

global systems (de Haan, 2020a, pp. 101–102). Global changes necessarily involve changes in  local systems while not all local changes ‘add up’ to global effects. Physiological processes necessarily change with psychological processes as the former are part of a global system that instantiates the latter. At the same time, not every change in physiological activity leads to a qualitative change in what a person thinks or feels. Similarly, social changes necessarily rely on changes in the members of the respective group and their relations while not every change on the side of individual group members adds up to a social upheaval. That is, an asymmetric determination is to be understood in that global and local systems determine each other, however, changes in more global systems necessarily imply changes in more local systems while this relation does not hold the other way around. Dynamic Organization  The emergent properties, behaviors, and structures of global systems substantially rely on the dynamic interaction between the involved components. It is the positive and negative feedback loops between more local systems that give rise to the emergent characteristics of more global systems in the absence of an external control mechanism (Thompson & Varela, 2001). As such, the considered systems cannot be understood in reference to components interacting sequentially, like gears in a clockwork (de Haan, 2020d). A machinal apparatus has pre-fixed rules along which the components interact while the components typically do not change in terms of their intrinsic properties. In a rigid sequence, one gear moves the next and we can replace a single gear (e.g., because of structural damage) without affecting other gears or the overall functioning of the clockwork. In contrast, in a dynamic system, the components interact and influence each other over more or less extended time spans, and the components themselves thereby often change. Suitable comparisons that illustrate the context-sensitivity and temporal unfolding of complex systems in the interaction of their components are rather found in chemistry than in relation to classical mechanics and the interaction of macroscopic objects (Varela et al., 1991). We should also keep in mind that the interaction between the components of a system does not necessarily lead to a change in that system as a whole. Only in a trivial sense, does a change in local components imply a change in the global system. However, when certain negative and positive feedback loops add up in a certain manner, they can lead to a shift, “tipping the system over from one stable state into another” (de Haan, 2020a, p. 101). When the respective threshold is reached might depend on the connections between the involved components as well as the constraining properties, behaviors, and structures of the global system. That is, the effect of a single component is always to be considered against the background of the dynamic interaction with other components in the overall organization of the system. At the same time, how the rather local components and their organization are affected by the system as a whole also depends on the broader context in which the system itself is embedded.

15  Chronic Pain, Enactivism, & the Challenges of Integration

259

Spatio-Temporal Scaling  It follows that the relation between the biological, psychological, and social domains is to be considered in terms of their complex, dynamic organization (de Haan, 2020a, pp. 114–116), rather than in terms of their reciprocal determination (e.g., Thompson & Varela, 2001). Speaking of reciprocal relations between biological, psychological, and social processes might suggest that they are instantiated by ontologically distinct systems that are related to each other in terms of a two-way dependence. This might still foster a false trichotomy of opposed systems that apparently interact in a sequential manner, reciprocally switching processes in the other systems on and off. This is problematic as it is incompatible with what is actually happening. The biological, psychological, and social aspects characterize aspects of the same matter with a narrower or wider spatio-temporal focus, showing distinct emergent properties, behaviors, and structures in their increasing constitutional complexity (Aftab & Nielsen, 2021; Fuchs, 2020). It follows that when speaking about biological, psychological, and social processes we can do so in an insightful manner, but we are not referring to three separate ontological buckets along which natural things can be strictly categorized. What we do when referring to biological, psychological, or social aspects, we zoom in or out on the same process. This allows for the same target system to be investigated from different spatio-temporal perspectives while the respective distinctions of biological, psychological, and social are potentially fuzzy (Eronen, 2021). Furthermore, we should not expect that the biological, psychological, and social processes that we typically address from different perspectives perfectly map in that they are spatio-temporally cross-cutting and overlapping (Potochnik & Sanches de Oliveira, 2019). Based on these minimal ontological assumptions, it follows that the enactive view rejects the idea that we can uniformly cut all nature into horizontal layers across which material and information flow back and forth (Fig. 15.1). While such approach of ‘layered’ domains might account for the emergence of new characteristics at higher levels, it facilitates a fragmented and linear reading. The view of ‘nested’ domains (Fig. 15.2) might do better in so far as it emphasizes that biological, psychological, and social processes do not exist in independent ontological realms but relate as parts and wholes with increasing complexity. Still, this view is misleading as it fails to account for the dynamic entanglement of involved processes within and across domains that unfolds over time and allows for their partial overlap and crosscutting. An enactive view demands to embrace complexity in terms of the emergence of qualitatively new characteristics in more global systems, integration in terms of the asymmetric determination of global and local aspects as spatio-­ temporally crosscutting and overlapping, and dynamics in terms of the looping relations between the local parts of a system and the coupling between the global system and its environment. That is, biological, psychological, and social processes cannot be reduced to each other and stand in asymmetric local-to-global and

260

S. Coninx and P. Stilwell

global-to-local relations as more or less spatio-temporally extended excerpts of the same natural process (Aftab & Nielsen, 2021; de Haan, 2020a; Fuchs, 2020).6 The previous considerations focused primarily on the general relation between biological, psychological, and social processes as instantiated by more local and global systems. What we chose to be our target system might depend on our scientific interests as there is no universally valid division of what counts as parts and wholes; rather these distinctions are best understood as useful heuristic idealizations (Eronen, 2021). From an enactive perspective, the central unit of analysis is the embodied person in dynamic interaction with their environment. Considering the person as the central unit of analysis does not mean addressing only psychological aspects. We also need to look down and consider the dynamic interaction of more local physiological systems, look around and consider the organization of the involved parts in their co-determining relation to the person as a whole, and look up and consider the broader (social) context in which the person is located in and dynamically coupled to (Bechtel, 2009). Taken together, this provides a picture of a hierarchy best characterized as ‘organizational’ (de Haan, 2020a, b, c) or ‘enmeshed’ (Thompson & Varela, 2001) (for an illustration see Fig. 15.3). According to this ontological architecture, biological, psychological, or social aspects are located in the same material realm: they are the more or less spatio-temporally extended aspects of the dynamic development of the same person-environment-system, or brain-body-environment-system, relating as partially overlapping and cross-cutting parts and wholes (Aftab & Nielsen, 2021; de Haan, 2020a, b, c).

15.3.3 Application to Chronic Pain In this section, we further unpack the minimal assumptions of an enactive ontology in the context of chronic pain. We implement the previously introduced approach of an organizational or enmeshed hierarchy by highlighting five central implications for our understanding of the relationship between biological, psychological, and social factors in the process of pain chronification.7 We do so with the help of an  One may ask whether the assumptions previously specified are not in some way still compatible with more sophisticated versions of reductionism. For example, Dan Burnston (2021) argues that a certain approach to reduction is compatible with properties, such as context dependence and dynamic interaction, as highlighted by enactivism. We acknowledge that many aspects of the debate between reductionism and emergentism, for example, depend on how exactly reduction and emergence are understood. Addressing these issues would clearly exceed the scope and purpose of this paper. What is primarily relevant at this point is that any promising approach should meet the indicated minimal ontological assumptions, which is clearly not the case with respect to naïve or atomistic variants of reductionism. 7  For a more detailed analysis of empirical data concerning the relation of biological, psychological, and social factors in the process of pain chronification see Coninx and Stilwell (2021), Stilwell and Harman (2019), and Cormack et al. (2022). 6

15  Chronic Pain, Enactivism, & the Challenges of Integration

261

Fig. 15.3  Organizational or enmeshed hierarchy. (inspired by Krickel (2018), Fig. 7.2): the figure illustrates the relation of biological, psychological, and social processes as more or less spatio-­ temporally extended aspects of the same person-environment-system. The research target (pain and the process of its chronification) is a property of the person and their unfolding dynamics over time whose understanding requires not only the consideration of the psychological aspects but also of more local biological and more global social aspects. The arrows indicate that biological, psychological, and social processes co-determine each other which behave asymmetrically in their global-to-local and local-to-global relation. The breaks between sections are not meant to indicate that the biological, psychological, or social domain temporally cease to exist, but that specific biological, psychological, and social processes can start or stop and spatiotemporally intersect. Finally, it might be noted that this model also allows for the influence of processes taken place in the even more global biosphere (e.g. climate, biodiversity) constituting the background of the person-environment-system and potentially permeating processes at all domains

illustrative analogy. The enactive view rejects misleading comparisons of complex systems with clockworks involving neatly separable parts that are linearly connected. Instead, we find in the literature different references to the process of baking. Feldman Barrett (2017) uses the analogy of baking bread to exemplify how different brain systems contribute to psychological phenomena. de Haan (2020a, pp. 97–104, b, c, d) applies the analogy of baking cakes to the interaction of physiological and psychological aspects in psychopathologies. This analogy has clear limitations as we are about to outline in the following. Nonetheless, it may help to illustrate the relation between physiological and psychological processes in chronic pain and it provides a more compelling picture of the complex, integrative, and dynamic approach of an enactive philosophy of pain. First, consider the process of baking bread or cake. The baking product (e.g. banana bread or chocolate cake) represents a more global system, in our case, the person suffering from chronic pain. The ingredients of the baking product, such as flour, water, sugar, or salt, are the more local components of which it is made, standing in for physiological processes, including (epi)genetic, immunological, endocrine, or neural aspects. What such analogy illustrates is that we cannot reduce chronic pain to a single physiological factor in the sense that it is the person suffering from pain, not their brain, hormones, or genes. Chronic pain is an emergent characteristic of the person; just as being crusty or fluffy are not properties present

262

S. Coninx and P. Stilwell

in any of the ingredients, but the baking product. At the same time, the more global experiential, behavioral, or cognitive alterations instantiated by a person in chronic pain are determined by the processes of more local physiological systems. Persisting pain as experienced by a person is nothing over and above the physiological (re) organization of their brain and body without being reducible to it; just like a bread or cake is nothing over and above flour, water, salt, or sugar without being reducible to any such ingredient. Second, we may understand how physiological processes contribute to pain chronification only in their interaction and as being part of a more complex organism. “To understand how salt transforms a recipe of bread, you must watch it work in context” (Feldman Barrett, 2017). The physiological processes involved in the generation and maintenance of pain cannot be individuated according to what they do intrinsically and chronic pain cannot be analyzed as the mere collection of what these physiological systems do in isolation or other systems. Individual brain-bound processes depend on their broader neural contexts and their interaction with non-­ brain-­bound processes while these biological processes themselves are constrained by psychological and social aspects. That is, not all physiological events, such as an injury, necessarily result in pain. Further, a shift in a more global system may take place and persist, even if the local components that were initially involved have changed. That is, the inciting issue linked to the onset of pain may not be the sustaining factor. Third, psychological characteristics are instantiated by global systems and equal in our analogy to properties of the bread or cake as a whole. At this point, the first limitations of the baking analogy become evident, as these psychological processes themselves interact in looping manners which can hardly be illustrated by the relation between ‘crustiness’ or ‘fluffiness’. For example, positive expectations about treatment outcome, negative thoughts about one’s body, and increasing pain may reveal ongoing faciliatory and inhibitory effects on each other. Further, psychological processes constrain physiological processes in terms of global-to-local determination. Most importantly, psychological processes do not affect physiological processes in a sequential manner, but to intervene on the former is to intervene on the latter. It is not that a subject holds negative thoughts about their body which in turn causes certain physiological alterations that may further contribute to pain chronification. Psychological changes always involve physiological changes; just like changes to a cake always involve changes to its ingredients. At the same time, changes in the whole might mold changes in the involved components in varying manners. For example, psychological processes (e.g. negative expectations) might influence some physiological systems faster than others (e.g. neural vs endocrine system) (de Haan, 2020a, pp. 101–102). Fourth, in the processes of chronification, organism and environment continuously determine and constrain each other. In our analogy, the oven that provides heat or the baker that continuously moistures a bread with water might stand in for the relevant social factors in the process of chronification, such as stigmatization, social support, or messages of rest and avoidance. This is where the introduced analogy reveals further limitations (Bruineberg, 2021): oven and baker only have a

15  Chronic Pain, Enactivism, & the Challenges of Integration

263

uni-­directional relation to the baking product which is typically finished at some point while a person interacts with their environment in an ongoing dynamic manner. However, what the enactive view emphasizes is that the processes contributing to pain are not restricted to the patient, as chronic pain is partly a social and iatrogenic phenomenon. Further, none of the domains functions as a uni-directional switch, in that social factors only trigger psychological or physiological changes, rather social factors permeate the person and their experiences, beliefs, and behavior and in that their physiological constitution. Fifth, it follows that chronic pain does not emerge only because of certain physiological, psychological, or social processes, as they are not static and fragmented elements in a collection of heterogeneous factors. It is not simply relevant which factors are involved but how they relate and unfold over time. We can hardly find any ‘root causes’ for chronic pain as the involved biological, psychological, and social processes continuously co-determine each other as more or less local or global excerpts of the same process. We cannot say that the persistence of pain is the result of a single factor; just like we cannot say that bread or cake is the sole result of any particular ingredient or the heat of the oven. Further, we cannot decompose a person in pain into clear-cut biological, psychological, and social elements. In our analogy, we might study bread and determine that a certain ingredient or external condition has been involved, for example, because it tastes salty or is fully baked. However, we cannot open the bread and neatly dissect it into the different ingredients and procedural steps to unravel their contribution in isolation of all other factors involved. In that, enactivism guarantees the rejection of naïve reductionism, fragmentation, and linearity in relation to the generation and maintenance of chronic pain and provides at the same time a complex, integrative, and dynamic alternative. As a crucial final note, it should be highlighted that even if the outlined assumptions might seem hardly controversial or radical to some philosophers, their acceptance might have a significant influence on how we conceptualize, explain, assess, and treat pain; especially as these assumptions are not yet fully embraced in pain research and clinical practice. That is, even if the minimal assumptions of an enactive ontology reflect common sense in certain philosophical debates, they do not characterize the implicit or explicit philosophy of pain implemented in clinical research and practice. As such, enactivism offers a theoretical foundation for a complex, integrative, and dynamic view of chronic pain. It provides conceptual tools and illustrations that might prove useful for the cascading interaction of philosophers, researchers, clinicians, and patients, especially to avoid misguided interpretations and applications of the BPS model. We are aware that enactive frameworks often entail stronger ontological assumptions than so far presented while some aspects are, in our view, still up for debate. First, it seems necessary to further investigate how an enactive ontology relates to strong and weak emergence and different epistemic or ontological interpretations of the global-to-local determination (e.g., Gillett, 2016; O’Connor & Hong, 2020; Stephan, 2006). Second, while an enactive framework is not compatible with the idea of a vertical hierarchy, it remains questionable whether this entails a commitment to a horizontal hierarchy (e.g., de Haan, 2020a, p. 119). Third, it remains to be

264

S. Coninx and P. Stilwell

seen whether enactivism implies a certain interpretation of (synchronic or diachronic) constitution (e.g., Gallagher, 2017). In our view, the outlined minimal assumptions themselves do not commit to a particular position regarding these more controversial issues. While further investigation is needed, it might be the case that at least some of those assumptions going beyond the outlined minimal commitments make no difference to the outstanding conceptual, explanatory, methodological, and therapeutic challenges in the context of chronic pain.

15.3.4 Conceptual, Explanatory, Methodological, & Therapeutic Challenges In the following, we revisit the remaining challenges of integration and outline which implications the previous considerations may have on how we conceptualize, explain, study, and treat chronic pain. We are aware that these facets of the integration problem require an in-depth analysis as they connect to extensive ongoing debates in the philosophical literature. We aim to primarily indicate ways in which an enactive approach to chronic pain may potentially contribute to discussions on conceptual, explanatory, methodological, and therapeutic issues and explore how far it is compatible with existing positions in corresponding debates. Conceptual Challenge   This challenge raises the question of how to conceptualize chronic pain in the light of the integration problem. It follows from the previously introduced framework that chronic pain cannot be conceptualized as a disease of the body or brain, at least not exclusively. Chronic pain is not identical or primarily grounded in physiological abnormalities. However, it is neither adequate to conceptualize chronic pain only as a psychological or only as a social disorder. It seems misguided from the beginning to discuss the conceptual challenge in terms of such trichotomy. From an enactive perspective, chronic pain is best understood as a disruption of the interactive embodied relation between person and environment that permeates all domains in their dynamic interplay as more local or global excerpts of the same process (Coninx & Stilwell, 2021; Cormack et al., 2022).8 Pain is identified in reference to the first-person perspective. Analyzing the lived experience of patients shows that they are often unable to flexibly attune to the requirements of different situations as they are stuck in a particular pattern of perceiving themselves and their interactive relation to the environment (Coninx & Stilwell, 2021). This is not to say that the transition to chronicity is somehow the individual’s fault or that they can simply think away their pain or act differently. It is the interplay of biological, psychological, and social processes over time that

 This enactive consideration still aligns with the IASP definition of pain, acknowledging that pain is understood as a psychological phenomenon in the sense that it is a personal (subjective) sensory and emotional experience that is best identified from a first-person perspective. 8

15  Chronic Pain, Enactivism, & the Challenges of Integration

265

brings forth chronic pain and the corresponding changes in lived experience. Thus, we need to conceptualize chronic pain as a condition of the person as a whole that we can fully understand only when considering the ongoing interaction of brain, body, and environment (Cormack et al., 2022). In the conceptualization of chronic pain, references to temporal persistence are useful only in as much as the dynamics of chronification typically unfold over a certain time span in the co-determination of biological, psychological, and social processes. Explanatory Challenge   This challenge addresses the question of how we can explain chronic pain given the variety and heterogeneity of involved factors and their dynamic entanglement. An enactive approach suggests the rejection of explanatory reductionism, according to which only biological explanations are ‘real’ explanations (Bickle, 2003), or at least to be preferred (Insel & Cuthbert, 2015; Kandel, 2018). In contrast, enactivism favors explanatory pluralism in that none of the three domains are considered more fundamental (Borsboom et  al., 2018). Biological, psychological, and social explanations rely on different perspectives on one and the same process, while there is no principle reason for why any of them should be considered more ‘real’ or superior (Kästner, 2018). While accepting explanatory pluralism, it remains challenging to determine how biological, psychological, and social explanations can be brought together (Kästner, 2018; Potochnik & Sanches de Oliveira, 2019). Within and across disciplines, there are different approaches to what the relevant global and local systems under investigation might be and the corresponding explanatory approaches might frame chronic pain in terms of varying spatio-temporal complexity. Further, the processes they refer to prove partially overlapping and crosscutting, connected through multiple more or less extended feedback-loops. That is, we receive a complex picture of different explanatory perspectives on chronic pain suited for varying scientific purposes which come with a different focus. In principle, explanatory pluralism allows for multiple interpretations (Eronen, 2021; Miłkowski & Hohol, 2021). First, we may argue that researchers need to rely on a set of incompatible explanations concerning more local or global aspects of pain chronification. This provides a non-reductionist, although fragmented picture. Second, we may aim for a single unifying explanation of chronic pain that brings together all scientific perspectives at once. This however seems to neglect that our explanations may indeed vary with our scientific interests. A third option emerges when thinking about the previous two interpretations as two extremes with moderate versions of explanatory pluralism in-between. For example, integrative pluralism considers biological, psychological, and social explanations as complementary and mutually informative. Still, its aim is not necessarily to provide a single unified explanation but to integrate ‘bit-by-bit’ in different contexts different aspects of explaining the emergence of a certain phenomenon (Kendler, 2005, 2008). Enactivism fits well with integrative pluralism. First, enactivism takes a particular perspective on chronic pain, focusing on patients in their environment, which requires the combination of multiple explanatory perspectives. This seems most suitable for the aims of clinical research and practice while enactivism might accept

266

S. Coninx and P. Stilwell

that these are not the only purposes of pain science (Coninx, 2021). Second, even if a full multi-domain integration could never be reached, this might still function as a useful heuristic ideal. That is, integration is to be understood as a process rather than a goal, promoting local integrative efforts in terms of interdisciplinary exchange (Eronen, 2021; Kendler, 2005). Third, integrative pluralism aligns with the enactive claim that the targets of many research projects are only excerpts of a more complex process. This should result in more modest interpretations in that a single discipline might hardly provide the explanation of chronic pain. For example, an explanation of chronic pain in terms of neural reorganization can be most useful in certain scientific circumstances but is not a ‘full’ explanation as it provides only one piece of the puzzle. Methodological Challenge   Concerning the methodological challenge, an enactive approach faces similar issues as with respect to the previous explanatory challenge. In a nutshell, there is no particular instrument of investigation that is to be prioritized on principle over others, as they provide access to different aspects of dynamically developing processes.9 At the same time, we should also not expect to identify a single unifying methodology for all pain science. Aspects of varying spatio-­ temporal complexity require different methodological approaches to be studied. No method is superior but it may only prove more or less suited for certain scientific purposes. As such, an enactive approach allows for a flexible choice of methods depending on the more local or global focus of a particular investigation while fostering overall an integrative multi-domain perspective (Stilwell & Harman, 2021). Physiological investigations are on their own inadequate to assess the generation and maintenance of chronic pain. We also need to consider methods addressing the lived experience of patients and how social processes co-determine psychological and physiological aspects (Mescouto et  al., 2020). One might argue that certain ‘objective’ methods are to be preferred. By contrast, all methods are to some degree dependent on the scientific context in which they are employed, coming with their own constraints and idealizations. For example, interpretations of neural data always depend on research context as these are sensitive to interest-driven decisions of which groups of research participants and features are respectively selected for study and comparison (e.g., Viola, 2021; Ward, 2019). At the same time, the study of how chronic pain affects and is affected by psychological and social processes is in principle not more ‘subjective’. For example, we should be careful not to conflate a subjective approach to studying pain with an approach to studying the subjective experience of pain (Gallagher & Zahavi, 2022). One might still critically ask how we can make a multi-domain approach pragmatically manageable. As an initial reply, integration of a variety of methodological  While multiple instruments might be of relevance for the study of the process of chronification of pain, with none of them being in principle more valuable than others, there are differences when it comes to the identification of pain cases. Not least for ethical, legal, and methodological reasons, first-person pain reports are to be considered the best available way to identify whether a person is experiencing pain or not and how it feels.

9

15  Chronic Pain, Enactivism, & the Challenges of Integration

267

perspectives is not a goal that needs to be actively pursued at all times, but it might rather provide the overarching framework within which research studies and clinical efforts are to be conducted and interpreted. As a strand of future research, it might also prove fruitful to investigate the complementarity of the enactive framework with multiplexes, that is, complex multi-layered network models that have the potential to account for the dynamic interaction between multiple factors of varying spatio-temporal complexity (Boer et al., 2021; Kästner, 2023). Multiplexes could provide a statistically grounded model to make an integrative approach to chronic pain manageable while enactivism offers a theoretical foundation of how biological, psychological, and social processes relate, making ontological and conceptual assumptions that are not inherent to network models themselves (de Haan, 2020a, p. 43). Therapeutic Challenge   This challenge concerns the question of how we can best treat chronic pain. Once again, an enactive approach indicates some kind of integrative pluralism. There are many routes to change and no target or method of intervention is to be prioritized only because they address biological, psychological, or social processes (Coninx & Stilwell, 2021; Stilwell & Harman, 2019; Cormack et al., 2022). This means that treatment should not limit patients to passive targets of physiological interventions. By contrast, a central element of pain management can be to reinforce the patient’s self-efficacy by training problem-solving skills and providing support in the re-assessment of goals (Kongsted et al., 2021). Improving self-management strategies and active self-care can be an effective measure to increase the patient’s autonomy and sense of agency, as well as to increase their ability to flexibly adapt to the requirements of a situation in and outside of clinical contexts (Coninx & Stilwell, 2021). We often do not find ‘magic bullets’ that allow us to cure medical conditions or fully eliminate pain; instead, we might rather focus on improving quality of life and addressing social conditions, such as reducing stigmatization and facilitating access to healthcare (Stegenga, 2018). Further, we need to reflect on the role that clinical practitioners themselves play in this process by transmitting to patients a particular philosophy of pain that may not be helpful or aligned with current pain science (Setchell et al., 2017; Stilwell & Harman, 2017). Enactivism emphasizes that it is the dynamic interaction between biological, psychological, and social aspects that is most relevant. A pluralistic approach is misguided if it results in fragmented health care neglecting the connections and interactions across domains, including inhibitory and facilitatory effects. By contrast, an enactive approach considers multidisciplinary approaches most promising that take into account the interplay of interventions (Coninx & Stilwell, 2021; Kamper et  al., 2014; Low, 2017). Further, we must recognize that there are no purely psychological or purely physiological interventions in pain management, for example, as psychological and physiological processes asymmetrically co-determine each other. More global or local interventions address aspects of varying spatio-temporal complexity, but they all concern the person living with chronic pain in their environmental context. They only offer different points of intervention and trajectories of the same target system (de Haan, 2020b, 2020c). This implies that the

268

S. Coninx and P. Stilwell

effects of treatment are non-linear and can be difficult to predict. Treatments that are effective concerning one patient might not work for another. Even interventions that have been effective for the same patient before might no longer prove helpful, indicating the need for care that is sensitive to the patient and their context. Acknowledging the complexities and dynamics of chronification poses a serious challenge. Each patient in their particular situation reveals idiosyncratic characteristics which means that we cannot make – with certainty – any assumptions concerning the effectiveness of future interventions (Coninx, 2021; Corns, 2020). This raises the question of how we could ever make informed therapeutic choices. Even if there are no strict regularities, it remains possible to make probabilistic assumptions about the success of certain interventions based on the similarities between patients and their contexts (Coninx, 2021). What we need to identify are the most relevant comparative classes for the respective cases. For example, concerning multiple groups of patients, it might show that changing more local aspects is less effective than intervening on more global aspects. The enactive framework indicates that subgrouping of patients can be most useful when not only one but multiple factors are considered and respectively addressed in prediction and treatment (Cholewicki et al., 2019). Furthermore, it is not necessarily the static set of involved factors but their dynamic interplay that might prove most relevant in this consideration. At the same time, it does not follow that ‘everything goes’. In the study of similarities and dissimilarities, an evidence-based approach is needed to identify the most promising intervention targets and methods for the relevant comparison classes, while also being aware of the respective limitations of sub-grouping (Saragiotto et al., 2017). For example, so far, there is no clear answer to the question of which sub-groupings are most predictive of therapeutic success given the idiosyncrasies of chronic pains. Finally, it is an empirical question of which comparison classes prove most useful for certain scientific purposes and which are the most useful intervention targets and methods for these comparison classes. An enactive framework indicates that team-based approaches, involving different disciplines intervening on different aspects of the person-environment system at (potentially) different times during the development of pain conditions, are most promising. Further, independent of our scientific progress, we should always expect individual and situational differences to impact therapeutic outcomes. Thus, a certain degree of uncertainty might be inevitable. It should therefore be part of clinical practice to address such uncertainties and potentially explore them in communication with patients, instead of searching for simple solutions to ‘fix’ the patient (Coninx & Stilwell, 2021; Cormack et al., 2022).

15.4 Conclusion In studying the recent development in pain sciences, we have good reason to take a critical view of neuro-reductionist and bio-reductionist tendencies in research and clinical practice. Instead, to meet the challenges of chronic pain, multifactorial

15  Chronic Pain, Enactivism, & the Challenges of Integration

269

approaches prove more promising. While the enactive approach is considered a further development of the BPS model, it provides a framework avoiding reductionist, fragmented, or linear misinterpretations and misapplications due to its more comprehensive theoretical foundation. At the same time, the enactive philosophy of pain enables us to address the different challenges of integration. Following enactive theory, the biological, psychological, and social processes involved in the chronification of pain are located in the same ontological realm without being reducible. They relate to each other as spatio-temporally overlapping and cross-cutting excerpts that characterize the same matter with a narrower or wider focus, showing qualitatively new properties, behaviors, and structures in their increasing complexity. Thus, chronic pain is understood as an emergent property of the person in their context that cannot be fully understood in the consideration of isolated factors, but only when their asymmetric co-determination and unfolding dynamic are taken into account. Chronic pain is best understood as a disruption of the relationship between person and environment that permeates all domains in their ongoing interplay as more local or global processes. This motivates an integrative pluralism in terms of the explanation of chronic pain as well as the selection of methodological instruments or therapeutic interventions. Finally, one might criticize the outlined approach as too abstract as it does not make concrete hypotheses to be directly tested. It also remains vague in that it does not prioritize or exclude particular research approaches and allows for the application of different tools and treatments. In reply, the enactive philosophy of pain should not be understood as a specific research program within a specific scientific discipline. On the contrary, we see it as a ‘philosophy of nature’, that is, a philosophical stance that connects the results from multiple research strands, provides an overarching framework for their interpretation, and motivates new scientific endeavors (Käufer & Chemero, 2021, pp. 181–182). We hope to have shown how fruitful this philosophical stance can be in relation to chronic pain and the challenges of integration. Acknowledgments  Many thanks to Beate Krickel and Lena Kästner as well as Somogy Varga and the students of the Master Course ‘Philosophy of Medicine’ (Spring 2022, University of Aarhus) for their helpful comments on different versions of the paper. Furthermore, we are grateful for the feedback of organizers and participants of the workshop ‘Methodology of Situated Cognition Research’ (Spring 2021, University of Kassel), helping us to initiate this project. The research of Sabrina Coninx was funded by the German Research Foundation (DFG), project number GRK-2185/1 (Research Training Group Situated Cognition).

References Adam, D. (2013). On the spectrum. Nature, 496, 416–418. Aftab, A., & Nielsen, K. (2021). From Engel to enactivism: Conceptualizing the biopsychosocial model. EuJAP, 17(2), (M2)5-23.

270

S. Coninx and P. Stilwell

Anjum, R.  L., Copeland, S., & Rocca, E. (2020). Introduction: Why is philosophy relevant for clinical practice? In R. L. Anjum, S. Copeland, & E. Rocca (Eds.), Rethinking causality, complexity and evidence for the unique patient (pp. 3–11). Springer Open. Apkarian, A. V. (2017). Advances in the neuroscience of pain. In J. Corns (Ed.), Routledge handbook of philosophy of pain (pp. 73–86). Routledge. Apkarian, A. V., Baliki, M. N., & Geha, P. Y. (2009). Towards a theory of chronic pain. Progress in Neurobiology, 87(2), 81–97. Aydede, M. (2017). Defending the IASP definition of pain. The Monist, 100, 439–464. Baggs, E., & Chemero, A. (2021). Radical embodiment in two directions. Synthese, 198, 2175–2190. Bechtel, W. (2009). Looking down, around, and up: Mechanistic explanation in psychology. Philosophical Psychology, 22(5), 543–564. Benning, T. (2015). Limitations of the biopsychosocial model in psychiatry. AMEP, 6, 347–352. Berent, I., & Platt, M. (2021). Essentialist biases toward psychiatric disorders: Brain disorders are presumed innate. Cognitive Science, 45(4), e12970. Bickle, J. (2003). Philosophy and neuroscience: A ruthlessly reductive account. Kluwer Academic Publishers. Boer, N. S., de Bruin, L. C., de Geurts, J. J. G., & Glas, G. (2021). The network theory of psychiatric disorders: A critical assessment of the inclusion of environmental factors. Frontiers in Psychology, 12, 623970. Bolton, D. (2021). The biopsychosocial model of health and disease: Responses to the 4 commentaries. EuJAP, 17(2), 5–26. Bolton, D., & Gillett, G. (2019). The biopsychosocial model of health and disease: New philosophical and scientific developments. Palgrave Macmillan. Bonica, J. J. (1953). The management of pain. Lea & Febige. Borsboom, D. (2017). A network theory of mental disorders. World Psychiatry, 16, 5–13. Borsboom, D., Cramer, A.  O. J., & Kalis, A. (2018). Brain disorders? Not really: Why network structures block reductionism in psychopathology research. The Behavioral and Brain Sciences, 42, e2. Borsook, D., Sava, S., & Becerra, L. R. (2010). The pain imaging revolution: Advancing pain into the 21st century. The Neuroscientist, 16(2), 171–185. Breivik, H., Collett, B., Ventafridda, V., Cohen, R., & Gallacher, D. (2006). Survey of chronic pain in Europe: Prevalence, impact on daily life, and treatment. European Journal of Pain, 10(4), 287–333. Bruineberg, J. (2021). Review of Sanneke de Haan ‘enactive psychiatry’ (2020). Phenomenology and the Cognitive Sciences, 22, 541–547. Burnston, D. C. (2021). Getting over atomism: Functional decomposition in complex neural systems. The British Journal for the Philosophy of Science, 72(3), 743–772. Cabaniss, D. L., Moga, D. E., & Oquendo, M. A. (2015). Rethinking the biopsychosocial formulation. The Lancet Psychiatry, 2(7), 579–581. Chapman, C. R., Tuckett, R. P., & Song, C. W. (2009). Pain and stress in a system’s perspective: Reciprocal neural, endocrine and immune interactions. The Journal of Pain, 9(2), 122–145. Cholewicki, J., Pathak, P. K., Reeves, N. P., & Popovich, J. M. J. (2019). Model simulations challenge reductionist research approaches to studying chronic low back pain. The Journal of Orthopaedic and Sports Physical Therapy, 49(6), 477–448. Cohen, S. P., Vase, L., & Hooten, W. M. (2021). Chronic pain: An update on burden, best practices, and new advances. Lancet, 397(10289), 2082–2097. Coninx, S. (2020a). Experiencing pain: A scientific enigma and its philosophical solution. de Gruyter. Coninx, S. (2020b). Strong representationalism and bodily sensations: Reliable causal covariance and biological function. Philosophical Psychology, 34(2), 210–232. Coninx, S. (2021). The notorious neurophilosophy of pain: A family resemblance approach to idiosyncrasy and generalizability. Mind & Language, 8(1), 178–197.

15  Chronic Pain, Enactivism, & the Challenges of Integration

271

Coninx, S., & Stilwell, P. (2021). Pain and the field of affordances: An enactive approach to acute and chronic pain. Synthese, 199, 7835–7863. Cormack, B., Stilwell, P., Coninx, S., & Gibson, J. (2022). The biopsychosocial model is lost in translation: From misrepresentation to an enactive modernization. Physiotherapy: Theory and Practice, 28, 1–16. Corns, J. (2020). The complex reality of pain. Routledge. Cramer, A. O. J., van Borkulo, C. D., Giltay, E. J., van der Maas, H. L. J., Kendler, K. S., Scheffer, M., & Borsboom, D. (2016). Major depression as a complex dynamic system. PLoS One, 11, e0167490. Dahlhamer, J., Lucas, J., Zelaya, C., Nahin, R., Mackey, S., DeBar, L., Kerns, B., Von Korff, M., Porter, L., & Helmick, C. (2018). Prevalence of chronic pain and high-impact chronic pain among adults — United States, 2016. MMWR, 67(36), 1001–1006. de Haan, S. (2020a). Enactive psychiatry. Cambridge University Press. de Haan, S. (2020b). An enactive approach to psychiatry. Philosophy, Psychiatry, & Psychology, 27(1), 3–25. de Haan, S. (2020c). Bio-psycho-social interaction: An enactive perspective. International Review of Psychiatry, 33(5), 471–477. de Haan, S. (2020d). Enactive causality: Interventions, cakes, and clockworks  – A reply to Gallagher and Donovan and Murphy. Philosophy, Psychiatry, & Psychology, 27(1), 31–33. De Ruddere, L., & Craig, K. D. (2016). Understanding stigma and chronic pain: A state-of-the-art review. Pain, 157(8), 1607–1610. Denk, F., MacMahon, S. B., & Tracey, I. (2014). Pain vulnerability: A neurobiological perspective. Nature Neuroscience, 17(2), 192–200. Descalzi, G., Ikegami, D., Ushijima, T., Nestler, E.  J., Zachariou, V., & Narita, M. (2015). Epigenetic mechanisms of chronic pain. Trends in Neurosciences, 38(4), 237–246. Descartes, R. (1972). In T. T. S. Hall (Ed.), Treatise of man. Harvard University Press. Di Paolo, E.  A., & Thompson, E. (2014). The enactive approach. In L.  A. Shapiro (Ed.), The Routledge handbook of embodied cognition (pp. 68–78). Routledge. Di Paolo, E. A., Cuffari, E. C., & De Jaegher, H. (2018). Linguistic bodies: The continuity between life and language. MIT Press. Engel, G. L. (1977). The need for a new medical model: A challenge for biomedicine. Science, 196(4286), 129–136. Engel, G. L. (1980). The clinical application of the biopsychosocial model. The American Journal of Psychiatry, 137(5), 535–544. Engel, G. L. (1982). The biopsychosocial model and medical education. The New England Journal of Medicine, 306(13), 802–805. Eronen, M.  I. (2021). The levels problem in psychopathology. Psychological Medicine, 51(6), 927–933. Feldman Barrett, L. (2017). How emotions are made: The secret life of the brain. Pan Macmillan. Foster, N.  E., Anema, J.  R., Cherkin, D., Chou, R., Cohen, S.  P., Gross, D.  P., Ferreira, P.  H., Fritz, J. M., Koes, B. W., Peul, W., Turner, J. D., & Maher, C. G. (2018). Prevention and treatment of low back pain: Evidence, challenges, and promising directions. Lancet, 391(10137), 2368–2383. Froese, T., & Di Paolo, E. A. (2009). Sociality and the life–mind continuity thesis. Phenomenology and the Cognitive Sciences, 8(4), 439–463. Fuchs, T. (2017). Ecology of the brain: The phenomenology and biology of the embodied mind. Oxford University Press. Fuchs, T. (2020). The circularity of the embodied mind. Frontiers in Psychology, 11, 1707. Fullana, M.  A., Abramovitch, A., Via, E., López-Sola, C., Goldberg, X., Reina, N., Fortea, L., Solanes, A., Buckely, M. J., Remella-Cavraro, V., Carvahlo, A. F., Tortella-Feliu, M., Vieta, E., Soriano-Mas, C., Lázaro, L., Stein, D. J., Fernández de la Cruz, L., Mataix-Cols, D., & Radua, J. (2020). Diagnostic biomarkers for obsessive-compulsive disorder: A reasonable quest or ignis fatuus? Neuroscience and Biobehavioral Reviews, 118, 504–513.

272

S. Coninx and P. Stilwell

Gallagher, S. (2017). Enactivist interventions: Rethinking the mind. Oxford University Press. Gallagher, S., & Zahavi, D. (2022). The phenomenal mind (2nd ed.). Routledge. Garcia-Larrea, L., & Peyron, R. (2013). Pain matrices and neuropathic pain matrices: A review. Pain, 154(1), 29–43. Gatchel, R. J., Peng, Y. B., Peters, M. L., Fuchs, P. N., & Turk, D. C. (2007). The biopsychosocial approach to chronic pain: Scientific advances and future directions. Psychological Bulletin, 133(4), 581–624. Ghaemi, S. N. (2009). The rise and fall of the biopsychosocial model. British Journal of Psychiatry, 195(01), 3–4. Gillett, C. (2002). The varieties of emergence: Their purpose, obligations and importance. Grazer Philosophische Studien, 65, 95–121. Gillett, C. (2016). Reduction and emergence in science and philosophy. Cambridge University Press. Goldberg, D. S., & Summer, J. M. (2011). Pain as a global public health priority. BMC Public Health, 11, 770. Goodkind, M., Eickhoff, S. B., Oathes, D. J., Jiang, Y., Chang, A., Jones-Hagata, L. B., Ortega, B. N., Zaiko, Y. V., Roach, E. L., Korgaonkar, M. S., Grieve, S. M., Galatzer-Levy, I., Fox, P. T., & Etkin, A. (2015). Identification of a common neurobiological substrate for mental illness. JAMA Psychiatry, 72, 305–315. Hardcastle, V. G. (2014). Pleasure gone awry? A new conceptualization of chronic pain and addiction. Review of Philosophy and Psychology, 5(1), 71–85. Hashmi, J. A., Baliki, M. N., Huang, L., Baria, A. T., Torbey, S., Hermann, K. M., Schnitzer, T. J., & Apkarian, A. V. (2013). Shape shifting pain: Chronification of back pain shifts brain representation from nociceptive to emotional circuits. Brain, 136, 2751–2768. Hay, S. I., Abajobir, A. A., Abate, K. H., Abbafati, C., Abbas, K. M., Abd-Allah, F., et al. (2017). Global, regional, and national disability-adjusted life-years (DALYs) for 333 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990–2016: A systematic analysis for the global burden of disease study 2016. Lancet, 390(10100), 1260–1344. Hillman, P., & Wall, P. D. (1969). Inhibitory and excitatory factors influencing the receptive fields of lamina 5 spinal cord cells. Experimental Brain Research, 9(4), 284–306. Hoffman, G. A., & Zachar, P. (2017). RDoC’s metaphysical assumptions: Problems and promises. In J. Poland & S. Tekin (Eds.), Extraordinary science and psychiatry: Responses to the crisis in mental health research (pp. 59–86). MIT Press. Hutto, D. D., & Myin, E. (2013). Radicalizing enactivism: Basic minds without content. MIT Press. Hutto, D. D., & Myin, E. (2017). Evolving enactivism: Basic minds meet content. MIT Press. Iannetti, G.  D., & Mouraux, A. (2010). From the neuromatrix to the pain matrix (and back). Experimental Brain Research, 205(1), 1–12. Insel, T., & Cuthbert, B. N. (2015). Brain disorders? Precisely: Precision medicine comes to psychiatry. Science, 348, 499–500. Kamper, S. J., Apeldoorn, A. T., Chiarotto, A., Smeets, R. J. E. M., Ostelo, R. W. J. G., Guzman, J., & van Tulder, M. W. (2014). Multidisciplinary biopsychosocial rehabilitation for chronic low back pain. Cochrane Database of Systematic Reviews, 9, CD000963. Kandel, E. (2018). The disordered mind: What unusual brains tell us about ourselves. Farrar. Kästner, L. (2018). Integrating mechanistic explanations through epistemic perspectives. Studies in History and Philosophy of Science, 68, 68–79. Kästner, L. (2023). Modeling psychopathology: 4D multiplexes to the rescue. Synthese, 201, 9. Käufer, S., & Chemero, A. (2021). Phenomenology: An introduction (2nd ed.). Polity Press. Kendler, K. S. (2005). Toward a philosophical structure for psychiatry. The American Journal of Psychiatry, 162(3), 433–440. Kendler, K.  S. (2008). Explanatory models for psychiatric illness. The American Journal of Psychiatry, 165(6), 695–702. Kendler, K. S. (2010). The rise and fall of the biopsychosocial model: Reconciling art and science in psychiatry. The American Journal of Psychiatry, 167(8), 999–1000.

15  Chronic Pain, Enactivism, & the Challenges of Integration

273

Kendler, K. S. (2012). The dappled nature of causes of psychiatric illness: Replacing the organic-­ functional/hardware-software dichotomy with empirically based pluralism. Molecular Psychiatry, 17, 377–388. Klein, C. (2022). Review of ‘the complex reality of pain’ by Jennifer Corns. Mind, 131, 988–997. Koleva, D. (2005). Pain in primary care: An Italian survey. European Journal of Public Health, 15, 475–479. Kongsted, A., Risa, I., Kjaera, P., & Hartvigsen, J. (2021). Self-management at the core of back pain care: 10 key points for clinicians. Brazilian Journal of Physical Therapy, 25(4), 396–406. Kosek, E., Clauw, D., Nijs, J., Baron, R., Gilron, I., Harris, R.  E., Mico, J., Rice, A.  S. C., & Sterling, M. (2021). Chronic nociplastic pain affecting the musculoskeletal system: Clinical criteria and grading system. Pain, 162(11), 2629–2634. Krickel, B. (2018). The mechanical world: The metaphysical commitments of the new mechanistic approach. Springer. Larkings, J.  S., & Brown, P.  M. (2018). Do biogenetic causal beliefs reduce mental illness stigma in people with mental illness and in mental health professionals? A systematic review. International Journal of Mental Health Nursing, 27(3), 928–941. Lee, J. J., Kim, H. J., Čeko, M., Park, B., Lee, S. A., Park, H., Roy, M., Kim, S., Wager, T. D., & Woo, C. (2021). A neuroimaging biomarker for sustained experimental and clinical pain. Nature Medicine, 27, 174–182. Lehman, B.  J., David, D.  M., & Gruber, J.  A. (2017). Rethinking the biopsychosocial model of health: Understanding health as a dynamic system. Social and Personality Psychology Compass, 11(8), e12328. Low, M. (2017). A novel clinical framework: The use of dispositions in clinical practice. A person centred approach. Journal of Evaluation in Clinical Practice, 23(5), 1062–1070. Mansour, A. R., Farmer, M. A., Baliki, M. N., & Apkarian, A. V. (2014). Chronic pain: The role of learning and brain plasticity. Restorative Neurology and Neuroscience, 32(1), 129–139. Mäntyselkä, P., Kumpusalo, E., Ahonen, R., Kumpusalo, A., Kauhanen, J., Viinamäki, H., Halonen, P., & Takala, J. (2001). Pain as a reason to visit the doctor: A study in Finnish primary health care. Pain, 89, 175–178. Martinez-Calderon, J., Flores-Cortes, M., Morales-Asencio, J.  M., & Luque-Suarez, A. (2020). Which psychological factors are involved in the onset and/or persistence of musculoskeletal pain? An umbrella review of systematic reviews and meta-analyses of prospective cohort studies. The Clinical Journal of Pain, 36(8), 626–637. Melzack, R. (1973). The puzzle of pain. Basic Books. Melzack, R. (1989). Phantom limbs, the self and the brain. Canadian Psychologist, 30(1), 1–16. Melzack, R. (1999). Pain and stress: A new perspective. In R.  J. Gatchel & D.  C. Turk (Eds.), Psychological factors in pain (pp. 89–106). Guilford Press. Melzack, R. (2001). Pain and the neuromatrix in the brain. Journal of Dental Education, 65(12), 1378–1382. Melzack, R., & Katz, J. (2013). Pain. Interdisciplinary Reviews: Cognitive Science, 4(1), 1–15. Melzack, R., & Wall, P. D. (1965). Pain mechanisms: A new theory. Science, 150(3699), 971–979. Melzack, R., & Wall, P. D. (1982). The challenge of pain (2nd ed.). Penguin Books. Melzack, R., Wall, P. D., & Weisz, A. Z. (1963). Masking and metacontrast phenomena in the skin sensory system. Experimental Neurology, 8(1), 35–46. Merskey, H., & Bogduk, N. (1994). Classification of chronic pain: Descriptions of chronic pain syndromes and definitions of pain terms (2nd ed.). IASP Press. Mescouto, K., Olson, R. E., Hodges, P. W., & Setchell, J. (2020). A critical review of the biopsychosocial model of low back pain care: Time for a new approach? Disability and Rehabilitation, 7, 1–15. Meulders, A. (2019). From fear of movement-related pain and avoidance to chronic pain disability: A state-of-the-art review. Current Opinion in Behavioral Sciences, 26, 130–136. Miłkowski, M., & Hohol, M. (2021). Explanations in cognitive science: Unification versus pluralism. Synthese, 199(1), 1–17.

274

S. Coninx and P. Stilwell

Miyahara, K. (2019). Enactive pain and its sociocultural embeddedness. Phenomenology and the Cognitive Sciences, 20, 871–886. Moayedi, M., & Davis, K. D. (2013). Theories of pain: From specificity to gate control. Journal of Neurophysiology, 109(2), 5–12. Morris, D. B. (1993). The culture of pain (Vol. 9, p. 61). University of California Press. Moseley, G.  L., & Butler, D.  S. (2017). Explain pain supercharged: The Clinician’s manual. Noigroup Publications. Noë, A. (2004). Action in perception. MIT Press. O’Connor, T., & Hong, Y.  W. (2020). Emergent properties. In E.N.  Zalta (ed.), Stanford encyclopedia of philosophy, Winter 2021 ed. https://plato.stanford.edu/archives/win2021/entries/ properties-­emergent/. Accessed 20 Feb 2022. O’Regan, J. K. (2011). Why red doesn’t sound like a bell. Oxford University Press. O’Regan, J., & Noë, A. (2001). A sensorimotor account of visual consciousness. The Behavioral and Brain Sciences, 24, 936–1031. Ojala, T., Häkkinen, A., Karppinen, J., Sipilä, K., Suutama, T., & Piirainen, A. (2015). Chronic pain affects the whole person: A phenomenological study. Disability and Rehabilitation, 37(4), 363–371. Ploner, M., Sorg, C., & Gross, J. (2017). Brain rhythms of pain. Trends in Cognitive Sciences, 21(2), 100–110. Popova, Y. B., & Rączaszek-Leonardi, J. (2020). Enactivism and ecological psychology: The role of bodily experience in agency. Frontiers in Psychology, 11, 1–16. Potochnik, A., & Sanches de Oliveira, G. (2019). Patterns in cognitive phenomena and pluralism of explanatory styles. Topics in Cognitive Science, 12(4), 1306–1320. Qaseem, A., Wilt, T.  J., McLean, R.  M., & Forciea, M.  A. (2017). Noninvasive treatments for acute, subacute, and chronic low back pain: A clinical practice guideline from the American college of physicians. Annals of Internal Medicine, 166(7), 514–530. Raffaeli, W., & Arnaudo, E. (2017). Pain as a disease: An overview. Journal of Pain Research, 1, 2003–2008. Raja, S. N., Carr, D. B., Cohen, M., Finnerup, N. B., Flor, H., Gibson, S., et al. (2020). The revised International Association for the Study of Pain definition of pain: Concepts, challenges, and compromises. Pain, 161(9), 1976–1982. Reckziegel, D., Vachon-Presseau, E., Petre, B., Schnitzer, T.  J., Baliki, M.  N., & Apkarian, A. V. (2019). Deconstructing biomarkers for chronic pain: Context- and hypothesis-dependent biomarker types in relation to chronic pain. Pain, 160(1), 37–48. Rice, A.  S., Smith, B.  H., & Blyth, F.  M. (2016). Pain and the global burden of disease. Pain, 157(4), 791–796. Rocca, E., & Anjum, R.  L. (2020). Complexity, reductionism and the biomedical model. In R. L. Anjum, S. Copeland, & E. Rocca (Eds.), Rethinking causality, complexity and evidence for the unique patient (pp. 75–94). Springer Open. Rossettini, G., Carlino, E., & Testa, M. (2018). Clinical relevance of contextual factors as triggers of placebo and nocebo effects in musculoskeletal pain. BMC Musculoskeletal Disorders, 19, 19–27. Saragiotto, B. T., Maher, C. G., Hancock, M. J., & Koes, B. W. (2017). Subgrouping patients with nonspecific low back pain: Hope or hype? The Journal of Orthopaedic and Sports Physical Therapy, 47(2), 44–129. Schmidt-Wilcke, T. (2015). Neuroimaging of chronic pain. Best Practice & Research: Clinical Rheumatology, 29(1), 29–41. Schroder, H. S., Duda, J. M., Christensen, K., Beard, C., & Björgvinsson, T. (2020). Stressors and chemical imbalances: Beliefs about the causes of depression in an acute psychiatric treatment sample. Journal of Affective Disorders, 276, 537–545. Setchell, J., Costa, N., Ferreira, M., Makovey, J., Nielsen, M., & Hodges, P. W. (2017). Individuals’ explanations for their persistent or recurrent low back pain: A cross-sectional survey. BMC Musculoskeletal Disorders, 18(1), 466.

15  Chronic Pain, Enactivism, & the Challenges of Integration

275

Siddall, P. J., & Cousins, M. J. (2004). Persistent pain as a disease entity: Implications for clinical management. Anesthesia and Analgesia, 99(2), 510–520. Slade, S. C., Molloy, E., & Keating, J. L. (2009). Stigma experienced by people with nonspecific chronic Low back pain: A qualitative study. Pain Medicine, 10(1), 143–154. Stegenga, J. (2018). Medical nihilism. Oxford University Press. Stenz, L., Le, C. J., Luthi, F., Vuistiner, P., Burrus, C., Paoloni-Giacobino, A., & Léger, B. (2021). Genome-wide epigenomic analyses in patients with nociceptive and neuropathic chronic pain subtypes reveals alterations in methylation of genes involved in the neuro-musculoskeletal system. The Journal of Pain, 23(2), 326–336. Stephan, A. (2002). Emergentism, irreducibility, and downward causation. Grazer Philos Stud, 65, 77–93. Stephan, A. (2006). The dual role of ‘emergence’ in the philosophy of mind and in cognitive science. Synthese, 151, 485–498. Stilwell, P., & Harman, K. (2017). ‘I didn’t pay her to teach me how to fix my back’: A focused ethnographic study exploring chiropractors’ and chiropractic patients’ experiences and beliefs regarding exercise adherence. J Can Chiropr Assoc, 61(3), 219–230. Stilwell, P., & Harman, K. (2019). An enactive approach to pain: Beyond the biopsychosocial model. Phenomenology and the Cognitive Sciences, 18(4), 637–665. Stilwell, P., & Harman, K. (2021). Phenomenological research needs to be renewed: Time to integrate enactivism as a flexible resource. International Journal of Qualitative Methods, 20, 1–15. Sullivan, J.  A. (2014). Stabilizing mental disorders: Prospects and problems. In H.  C. Kincaid & J. A. (Eds.), Classifying psychopathology: Mental kinds and natural kinds (pp. 257–281). MIT Press. Sullivan, M. D., Cahana, A., Derbyshire, S., & Loeser, J. D. (2013). What does it mean to call chronic pain a brain disease? The Journal of Pain, 14(4), 37–322. Svenaeus, F. (2015). The phenomenology of chronic pain: Embodiment and alienation. Continental Philosophy Review, 48(2), 107–122. Svenaeus, F. (2021). Health and illness as enacted phenomena. Topoi, 41, 373–382. Tabor, A., Keogh, E., & Eccleston, C. (2017). Embodied pain: Negotiating the boundaries of possible action. Pain, 158(6), 1007–1011. Thompson, E. (2007). Mind in life: Biology, phenomenology, and the sciences of mind. Harvard University Press. Thompson, E., & Varela, F. (2001). Radical embodiment: Neural dynamics and consciousness. Trends in Cognitive Sciences, 5(10), 418–425. Toye, F., Seers, K., Hannink, E., & Barker, K. (2017). A mega-ethnography of eleven qualitative evidence syntheses exploring the experience of living with chronic non-malignant pain. BMC Medical Research Methodology, 17, 116. Treede, R. D., Rief, W., Barke, A., Aziz, Q., Bennett, M. I., Benoliel, R., et al. (2015). A classification of chronic pain for ICD-11. Pain, 156(6), 1003–1007. Treede, R. D., Rief, W., Barke, A., Aziz, Q., Bennett, M. I., Benoliel, R., et al. (2019). Chronic pain as a symptom or a disease: The IASP classification of chronic pain for the international classification of diseases (ICD-11). Pain, 160(1), 19–27. Vachon-Presseau, E., Berger, S.  E., Abdullah, T.  B., Griffith, J.  W., Schnitzer, T.  J., & Vania Apkarian, A. (2019). Identification of traits and functional connectivity-based neurotraits of chronic pain. PLoS Biology, 17(8), e3000349. Varela, F., Thompson, E., & Rosch, E. (1991). The embodied mind: Cognitive science and human experience. MIT Press. Varga, S. (2015). Naturalism, interpretation, and mental disorder. Oxford University Press. Vehof, J., Zavos, H. M., Lachance, G., Hammond, C. J., & Williams, F. M. (2014). Shared genetic factors underlie chronic pain syndromes. Pain, 155(8), 1562–1568. Viola, M. (2021). Beyond the platonic brain: Facing the challenge of individual differences in function-structure mapping. Synthese, 199, 2129–2155.

276

S. Coninx and P. Stilwell

Vos, T., Flaxman, A. D., Naghavi, M., Lozano, R., Michaud, C., Ezzati, P. M., et al. (2012). Years lived with disability (YLDs) for 1160 sequelae of 289 diseases and injuries 1990–2010: A systematic analysis for the global burden of disease study 2010. Lancet, 380(9859), 263–2196. Wager, T. D., Atlas, L. Y., Lindquist, M. A., Roy, M., Woo, C., & Kross, E. (2013). An fMRI-­ based neurologic signature of physical pain. The New England Journal of Medicine, 368(15), 1388–1397. Wall, P.  D., & McMahon, S.  B. (1986). The relationship of perceived pain to afferent nerve impulses. Trends in Neurosciences, 9, 254–255. Ward, Z.  B. (2019). Registration pluralism and the cartographic approach to data aggregation across brains. British Journal for the Philosophy of Science, 73, 47–72. Ward, D., & Stapleton, M. (2012). Es are good: Cognition as enacted, embodied, embedded, affective and extended. In F. Paglieri (Ed.), Consciousness in interaction: The role of the natural and social context in shaping consciousness (pp. 89–104). John Benjamins Publishing. Ward, D., Silverman, D., & Villalobos, M. (2017). Introduction: The varieties of enactivism. Topoi, 36(3), 365–375.

Chapter 16

Affordances, the Social Environment, and the Notion of Field: State of the Debate and Methodological Insights Giuseppe Flavio Artese

Do you not believe that one can first speak of symbols only after the differentiation between the sacred and the profane has occurred, or is starting to occur? Before this differentiation, the sacred is not symbolized, but rather manifests itself in that, what from the perspective of our standpoint is a symbol – from ours and not from that of the participants […]. It is “self-­ evident” to us that they are facts and events, and so we interpret them as symbols. September 10, 1952. Aron Gurwitsch, Letter to Alfred Schutz,

Abstract  In his writings, Gibson firmly claimed that cultural or social factors could never distort perception. Cultural artefacts, social norms, language and signs were instead described as influencing perception and behavior only indirectly. At the same time, in his last monumental monograph, Gibson introduced the concept of affordance as applicable to the “whole realm of social significance”. Unfortunately, Gibson did not elaborate further on the relationship between the notion of affordance and the socio-cultural organization of niches like ours. This issue divided Gibson’s followers into two sides. Some of them followed Gibson, claiming that socio-cultural factors can drive behavior but never permeate perceptual experience to its bottom layers. A second group assumes instead that the presence of a structured socio-cultural context permeates the meaning of affordances already at the perceptual level. The aim of this chapter is twofold. The first part of the chapter aims to reconstruct Gibson’s view on the role of culture and sociality and then highlight the reasoning that animates both groups of theorists. In particular, a large part of the discussion is based on motivating the problematics that led the second group of theorists to disagree with Gibson and his idea that the individual’s social ­background G. F. Artese (*) Department of Philosophy, University of Kassel, Kassel, Germany e-mail: [email protected] © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8_16

277

278

G. F. Artese

never permeates sensory perception. The second aim is to throw the seeds to develop a methodological tool available to these theorists in order to characterize the experience of affordances as shaped by the social and cultural context in which they are embedded; the notion of field. This notion has a long history in psychology but its methodological consequences are often overlooked. Notably, one of the risks of the concept of field is, as scholars of the Gestalt school like Parlett and Lee claimed, that the notion could be framed so broadly to include “anything and everything”, making it theoretically and scientifically trivial and hence unable to provide a rigorous characterization of what is encountered in individual experiences. To avoid the trivialization of this concept, I conclude by developing a recent proposal that suggests endorsing the notion of field specifically elaborated by the phenomenologist Aron Gurwitsch.

16.1 Introduction James J. Gibson notoriously proposed the notion of affordance to refer to the possibilities for action directly perceivable in the environment. However, Gibson also believed that the notion of affordance had not just the scope of explaining behavior as generally understood but that it was also applicable to the “whole realm of social significance” (Gibson, 1979, p. 128). Overcoming the culture vs. nature dichotomy, despite not being discussed with the same frequency as other themes in ecological psychology, has played an essential role in Gibson’s last two books (1966, 1979) and, even today, remains a central issue for the new generations of ecological psychologists. For Gibson, the distinction between a natural environment that can be contrasted with an artificial one is a theoretical mistake and ultimately an abstraction generally endorsed to characterize the environmental surroundings scaffolded by human activities. Following Gibson, what is commonly defined as the cultural or the artificial environment “is not a new environment […] but the same environment modified by men” (1979, p. 122). Despite the important rejections of all forms of dualism endorsed in “The Ecological Approach to Visual Perception”, in the last 30 years, it has been forcefully discussed whether Gibson actually managed to do justice to the role that culture and sociality play in affordance perception. This debate stemmed from Gibson’s belief that language and cultural conventions can strongly affect how behavior is constrained and unfolded, but they nonetheless never “distort our perception of the world” (Gibson, 1966, p. 282). As Costall and Still (1989) noticed, Gibson argued that the perception of ecological values (i.e. affordances) is a radically different phenomenon compared to the perception of cultural artifacts, signs, and social conventions. Independently of the reasons that motivated his stance,1 Gibson disagreed  Following Costal and Still (1989, p.  482), Gibson’s differentiation between the perception of affordances and the perception of cultural artefacts was primarily motivated by ethical and not just ontological concerns, mostly developed during the World War II. Gibson was worried about the 1

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

279

with the so-called social perceptionists who theorized perception as inherently social from its early pre-reflective stages. Indeed, Gibson’s position might appear at odds for readers that are familiar with the rejection of the natural vs. cultural domain firmly advocated in his last book. The tension between, on the one hand, Gibson’s emphasis on the indirect character of perception when directed to cultural artefacts and social norms and, on the other one, his aversion towards all sorts of dualisms, including the natural vs. cultural one, divided his followers into two sides. A first group has argued that ecological psychology, as originally conceived by Gibson, can do justice to the role of social and cultural factors on perception and behavior. Scholars that defend Gibson’s characterization of language and socio-cultural practices reject the thesis that these phenomena can affect perception to its very bottom layers (see Reed, 1996a). As Heras-Escribano puts it (2019, p. 108): “you perceive the same as everyone, but you act differently given your social background”. A second group has instead argued that, despite his rejection of the cultural/natural distinction, Gibson nonetheless conceptualized affordances in an asocial and solipsistic manner (Costall, 1995). By reconceptualising some of Gibson’s central ideas, such as his notion of animal-­ environment complementarity, or the same concept of affordance, these theorists have tried to offer a more relational ontology such as to properly overcome the dichotomy between the social and the natural domains (Ingold, 1992; Still & Good, 1992; Rietveld & Kiverstein, 2014). Obviously, the separation of Gibson’s followers into two groups is a generalization since contemporary ecological psychology represents a highly heterogeneous field in which a variegated number of concepts, methods and specific theoretical stances have been differently used, developed and endorsed. This generalization can nonetheless be useful to shed light on the disagreements regarding the role that culture and sociality might play in relation to affordance perception. The aim of this paper is twofold. The first one is to highlight the motivations and the strategies that animated both groups of theorists. In particular, it will be attempted to motivate the problematics that led some ecological psychologists to disagree with Gibson and some his followers in thinking of the perceivers’ social backgrounds as not permeating sensory perception. The second aim is to throw the seeds to develop a methodological tool available for theorists who decided not to follow Gibson’s stance on culture and sociality. In particular, in the last sections, a great deal of it will be dedicated to the concept of field, a notion often used to characterize the pragmatic experience of a situation as encountered from the perspective of the individual perceiver. The notion of field has a relatively long history in psychology and its usage is not uncommon in old as in recent psychological writings. However, the methodological consequences of this concept have been often overlooked and rarely acknowledged. I will conclude by proposing a notion of field that I argue to be ethical outcomes of cultural relativism and the cynicism of the social scientists of his time towards the rise of anti-Semitism. He refused to accept the idea that Germans related to the Jews perceptually or in an immediate fashion, claiming instead that they were purely guided by social stereotypes and non-perceptual habits.

280

G. F. Artese

adequate to capture the experience of affordances situated in socially structured settings. My work will be an extension of what was argued in Artese (2022) and will draw extensively on some central ideas proposed by the phenomenologist Aron Gurwitsch (1974, 1979, 2010).

16.2 Gibson’s Direct Realism and the Role Assigned to Language and Socio-cultural Practices Because of its firm realist commitments, ecological psychology can be said, with no hesitation, to represent a radical departure from cognitivist, inferentialist and associationist theories of perception. Gibson’s thesis of direct perception, often just defined under the label of ecological realism, can be summarized as the idea that the environment can be directly perceived without the necessity of invoking the existence of entities such as mental representations, innate ideas or mechanisms responsible for the manipulation of raw sense data. In contrast, Gibson thought that perception was something organisms specifically evolved to prepare the adequate behavioural responses to ecological information. Rather than producing a map of the environment, Gibson’s theorized perception as a process necessary for selecting, picking up or resonating with the ecological information in the ambient optic array of the perceiver. The ambient optic array is a concept particularly important in ecological psychology and deserves some clarification. The ambient optic array is characterized by Gibson as a bundle of visual solid angles at a point of observation in a component-to-component correspondence with the environment (Gibson, 1972). Ecological or “optic-array” information, which Gibson carefully differentiated from the notion of Shannon information used in cognitivist psychology, makes it possible for the perceiver, if supported by the right background conditions, to perceive the structural invariants of the environment. The notion of structural invariants was instead introduced to refer to the immanent properties of the environment that do not change while the perceiver is actively exploring its surroundings. Perception is thus not something happening inside the brain but rather an achievement of the agent that emerges from the “retino-neuro-muscular system as an activity of the whole system” (Ivi, p. 79) during its attempt to attend to structural invariants. While Gibson devoted several pages to describe how direct perception works, he also recognized that it is possible to instantiate forms of indirect perception or awareness through cultural artifacts such as paintings, sound-reproducing devices, drawings, and language. To account for these phenomena, in “The Senses Considered as a Perceptual System” (1966), he distinguished between first-hand and second-­ hand forms of perception. While the first one refers to the information directly picked up in the optic ambient array, the second refers to the information mediated by cultural devices or language. As argued by Kiverstein & van Dijk (2021), the Gibsonian notions of first and second-hand information are generally understood by ecological psychologists in a hierarchical sense. In other words, second-hand

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

281

perception depends on the information directly perceived in the animal’s immediate surroundings. Gibson expresses this idea straightforwardly: “No symbol exists except as it is realized in sound, projected light, mechanical contact, or the like. All knowledge rests on sensitivity” (1966, p.  26). What follows from these considerations is that the vehicles responsible for second-hand forms of perception are considered as mere surrogates of other sources of information. For Gibson, the meaning behind these vehicles is something that needs to be learned through associative learning (in the case of symbols) or projection (in the case of images) necessarily involving, on one side, a cultural artefact and, on the other, a “natural” object represented by the first one. In the case of images, the more the artefact does justice to the correspondent natural object in the world, the less associative learning is required. Notably, Gibson formulated the distinction between direct and indirect perception in different ways during his career. For example, in “The senses considered as a perceptual system”, by relying on the ideas proposed by William James, he also used the notion of “knowledge by acquaintance” to refer to direct experience and the one of “knowledge about” which only observable only in humans and that can be transmitted through symbols and language. In a similar vein, in one of his early works (1951), he proposed to distinguish the notion of schematic world (that is perceived through cultural artefacts and conventions) from the literal world (which is instead directly given in perception). While these concepts can be read as synonyms, in the current text, for reasons of practicality, I will just refer to the concepts of first and second-hand perception. The consequence of Gibson’s sharp separation between these two main modes of perception has important implications in the investigation and usage of the concept of affordance. As noticed by Michael Turvey, affordances are, for the vast majority of ecological psychologists, exclusively related to first-hand perception and perceived as such in virtue of their material configurations. Turvey’s materialist definition of affordances as dispositional properties can be seen an example of such a stance: An affordance is an invariant combination of properties of substance and surface taken with reference to an animal. One invariant combination of properties affords grasping, another affords support for upright posture, another affords catching, and so on. […] My analysis is of the affordances for actions. I take these as fundamental and their explication as propaedeutic to any extension of affordances to other domains. (Turvey, 1992, p. 174, italics added)

Ecological values and the perception of action possibilities depend then on the combinations of properties, textures, and layouts that constitute the niche of an animal. Perceivers need to resonate with optic-array information and respond to what is offered by the physical properties of their surroundings. Ecological information and values need to be thus thought of as outside in the world and not inside the perceiver. The agent must possess the capacity to attend or resonate with ecological values, but their existence remains well-grounded in the ecological structure of the environment. For Gibson, information is something that “need[s] only to be attended to” (Gibson, 1972, p. 79) and cannot be constructed either internally or emerge through the agent’s interaction with the environment. He makes crystal clear that

282

G. F. Artese

Orthodox theories assume that there is always an “objective contribution” to perception (the sensations) and a “subjective contribution” to perception (innate ideas, or memories, or field-forces in the brain), the two contributions being combined in various proportions. I reject this assumption. If unequivocal stimulus information is made available to an observer in an experiment, his perception will be determined by it and by nothing else. (Ivi, p. 88, italics added)

It is not surprising that many neo-Gibsonians generally interpret the claim that language might affect perceptions as “the antithesis of the entire program of direct perception Gibson advocated” (Heft, 2017, p. 135). It is important to be clear that Gibson never denied that language makes it possible to direct our attention towards certain environmental features, to refer to what is not momentarily present or making abstractions. But, as has been emphasized already, ecological meanings and values are thought of as present in the environment independently of our linguistic categories, cultural practices and social conventions. By describing first-hand perception as a primordial mode of engagement with the world, Gibson tried to demystify the conception that culture and individual social backgrounds had to be considered as “extra ingredients” to be added in the study of perceptual processes. At the same time, the rejection of the natural vs. cultural dualism advocated in his last book suggests that both forms of perception are in our everyday life enmeshed enough to be equally able to drive and elicit all possible forms of skilled and unreflective behavior. In other words, if we follow Gibson in recognizing first-hand perception as the most fundamental mode of perception, we are left with the crucial task of providing an adequate characterization of the capacities and processes that make it possible for cultural practices and artefacts to drive behavior in the same smooth fashion of the ecological information directly attended in the ambient optic array. Unfortunately, this challenge has not been directly tackled by Gibson but has not remained unnoticed by his followers.2  Importantly, this chapter deals with Gibson’s followers that endorsed the central ideas at the heart of the ecological approach, i.e., the thesis of direct perception, the coupling between perception and action, the complementarity between organism and environment, etc. At the same time, it should also be recognized that the popularity of the concept of affordance has surpassed that of the ecological research program and has often been appropriated by conventional cognitive science. On many occasions, the term has been extrapolated from the role it plays in ecological psychology and has been either read in a representationalist key or, even if theorized as a world-involving phenomenon, that could nonetheless be grasped by neurocentric and representationalist theories of cognition (see Bickhard & Richie, 1983; Orlandi, 2016; Scarantino, 2003; Dennett, 2017 just to name a few). The theoretical issues concerning the relationship between the socio-cultural environment and individual affordances would certainly be approached differently from the perspective of representationalist theories, in which, as one reviewer kindly noticed, the notion of mental representation would play a key role in determining the normative character of a situated action. A position that may summarise the representationalist position on this issue was articulated by Chow (1989) in his debate with Heft, in which he argued that theorizing affordances in representational terms is necessary to make sense of how the meaning of the same object may be differently perceived when situated in different social contexts (for a response see Heft, 1989). Including the limitations and strategies available for non-ecological approaches that decided to rely on the term affordance is unfortunately beyond the scope of this chapter. I would like to thank the reviewer for bringing out this important point to light. 2

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

283

16.3 Neo-Gibsonian Interpretations of the Role of Sociality and Cultural Practices In their respective works, Reed (1993, 1996a) and more recently Heras-Escribano (2019), have tried to frame in Gibsonian terms how it is possible for different agents to directly perceive the environment in the same way (and thus maintaining the role Gibson assigned to first-hand perception) while producing different forms of (social) behavior (that are only possible through the usage of artefacts and practices responsible for second-hand forms of perception). Through the years, Edward Reed  has made significant progress in integrating ecological psychology with the social sciences. His main achievement consists in the introduction of the concept of “field of promoted action”. The latter concept is used by Heft to refer to the aspects of the environment an infant is trained to attend since the age of 3 months. More precisely, the field of promoted action can be thought of as a behavioural space constituted by a set of constraints intellectually inherited by the subject through education mostly received during her developmental years. Reed, as it is particularly clear in his book “The Necessity of Experience” (1996b) follows and develops Gibson’s distinction between first and second-hand forms of perception.3 Reed claims that, through social learning, it becomes possible for infants to start distinguishing between first-hand forms of perception from the second-hand ones. The difference between the two is that, while first-hand experience is direct and autonomous, second-hand experience is necessarily selected by somebody else. Similarly to Gibson, Reed highlights that second-hand perception is generally instantiated through mediums such as language, images, artefacts, etc. However, in contrast to the father of ecological psychology, Reed devoted much more attention to the role that other people, in particular caregivers, play for infants and young children in the process of learning about the different forms of second-­hand perception. While activities such as reading or interpreting symbols are not immediately possible for infants, the first forms of second-hand experiences an infant generally encounters are produced by caregivers who, during face-to-face interactions, make extensive use of gestures and linguistic communication. It is through the continuous exposure to language and gestures that the field of promoted action takes shape. In this context, it is important not to confuse the latter with the notion of perceptual field often found in the work of Gestalt psychologists or phenomenologists but rather as an cognitive force encouraging the perceiver to react to specific affordances based on her participation in communal practices. Without furtherly explore Reed’s proposal, what is central to understand is that, similarly to Gibson, the field of promoted action influences our actions but not the way the environment looks like. It is not a coincidence that, on several occasions, Reed forcefully pointed out that learning about affordances is radically different compared to learning about social norms. When we learn about a new affordance, we learn about the properties of “objects, events, places with respect one’s own actions” (1993, p.  52). In contrast, learning about  It is worth noticing that in this book Reed substituted the word “perception” with “experience”.

3

284

G. F. Artese

norms implies becoming aware of the consequences of our actions in respect to some socially established rules and customs. Following Reed, both forms of learning are required in most of our everyday interactions but they nonetheless represent two different and non-intersecting processes. It is indeed important to recognize the value of Reed’s work in relation to our understanding of how subjects learn to coordinate, directs their attention and ultimately deal with affordances in accordance to the conventions and rules typical of their community. However, a fair criticism that is possible to raise is that Reed’s field of promoted action only provides a developmental story of how certain capacities are acquired but then does not provide a clear description of how these same capacities, once developed, are (or are not) actively reflected in perception and in the unreflective behavior of a social agent. In this context, what has been proposed by Heras-Escribano (2019) is particularly relevant. Heras-Escribano starts his considerations by arguing, like Reed and Gibson, that social norms can only influence how the subject carries her actions but not on how she perceives. Probably inspired by a dispositionalist reading of Gibson (Turvey, 1992), Heras-Escribano argues that if affordances are properly perceived, they never fail to be actualized. In contrast to social behavior that is inherently normative, affordances leave no space for failure. Therefore, while affordances exist as dispositions or properties in the environment, norms and social practices exist only in human communities with linguistic practices that makes it possible to conceptualize the correct/incorrect conditions for the proper behavior to be unfolded in a related social context. While Reed offered a developmental story of how social behavior can be achieved, Heras-Escribano aims instead to explain what are the concrete capacities that make it possible for the perceiver to be directed towards a field of promoted action. In this regard, for Heras-Escribano the solution to has to be found in some of the central ideas proposed by McDowell (2007a) and Ryle (1949/2009). Following these authors, a practical and situated form of rationality always accompanies human’s behavior. This situated form of rationality, as has been argued by McDowell in his debate against Hubert Dreyfus (2007b, p. 366), does not have to be understood as a general maxim or an abstract rule, but as the idea that unreflective behavior is always permeated and guided by “linguistic concepts” that remain implicitly operative during the agent’s perceptual engagement with the world. By implementing the considerations made by McDowell on the role of linguistic and social norms, Heras-Escribano ultimately aimed to complement Reed’s concept of field of promoted action and provide a compelling story of how it is possible for agents to behave according to social norms in everyday contexts. However, in contrast to McDowell, who also originally conceived sensory experience as permeated by his notion of normative and conceptual rationality, HerasEscribano follows Reed in arguing that socio-cultural practices constrain human behavior but leave unaltered the perceptual appearance of affordances. In the case of both authors, it is not crystalline whether it if is really possible to justify, simply by postulating the existence of a field of promoted action, or the presence of a situated form of rationality, that perception is not inherently influenced by the perceiver’s developmental history and participation to communal practices.

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

285

In the next sections of this work, reasons for scepticisms and alternatives that assign a more fundamental role to social and cultural practices in the context of perceptual experience will be examined.

16.4 The Organism-Environment Mutuality and the Social Context: Is a Reconceptualization Necessary? During the years, serious criticisms against the orthodox neo-Gibsonians view to culture and sociality have been raised. By distinguishing first from second-hand perception, by arguing that direct perception is impenetrable by the individual’s social background, and, in some cases, that is possible to differentiate social from non-social affordances, following the critics, Gibson and some of his followers failed in accounting for the irreducible intertwinement between materiality and culture. In contrast, a proper overcome of the nature-culture dichotomy requires, as Costal suggested (1995), to assume that the social simply squares with reality. It is, however, important to acknowledge that this criticism is not new. For example, Tomasello (1999) criticized ecological psychologists for prioritizing how perceivers manage to get a grip on “cultural” affordances and social norms while ignoring the “social matrix” in which these objects and activities are embedded in the first place. More recently and armed with similar concerns, McGann has pointed out that the necessity of considering culture not as something introducing a bias or an addition to our behavioural and perceptual capacities but rather as something that inherently constitute the surroundings in which our same capacities are acquired and developed in the first place (2014). These criticisms led several authors to recognize the necessity of reconceptualising the organism and environment mutuality initially sketched by Gibson in such a way to provide an understanding of the ecological niche as inherently cultural and social in its material configuration. Gibson, because of his constant concern of showing that affordances exist in a physical, and not phenomenal environment, can be reasonably accused of prioritizing the environment over the perceiver. As a matter of fact, in order to differentiate the ecological environment from Koffka and Lewin’s phenomenal one, he readily admitted that, while animals inherently depend on the environment to survive, the organism does not depend on any living form for its existence (1979, p. 121). This idea has been inherited in Reed’s work when he claims: [The] organism requires the environment for its sustained existence, but the environment does not require any given organism for its sustained existence. […] affordances in particular, and eco-niches in general, are aspects of the environment of all organisms, not just features of the environment of one creature. (1996a, p. 26)

The differentiation between first and second-hand perception or between schematic and literal world becomes  then a necessary move to account for the diversified forms of cultural behavior and artifact use in the human’s niche. Problematically, what seems to follow from the distinction between first and second-hand perception

286

G. F. Artese

is that cultural and social meaning cannot said to exist if not as something projected by the perceiver on an environment originally empty of social significance. Different attempts to recast Gibson’s non-dualistic ontology in a such a way to reject the premise that the cultural characteristics of the environment are something projected by the perceiver have nonetheless been proposed. For example, Ingold has argued the environment is not perceived as inherently cultural because we unreflectively rely on forms of associations, projections or linguistic categories. While we can indeed make use of these capacities when necessary, sociality “is rather given from the start, prior to objectification of experience in cultural categories, in the direct perceptual involvement of fellow subjects immersed in joint action in the same environment” (Ingold, 1992, p. 47). Embracing the intertwinement between sociality and materiality implies accepting that agents are not only perceptually sensitive to structural invariants as generally understood but also to social and pragmatic constrains embodied within an ecological niche scaffolded by others of our kind as well as by other species. This does imply giving up on ecological research program as such or implying the existence of theoretical entities (e.g. mental representations, mental maps, etc.) in contrast with Gibson’s assumptions. It rather implies appreciating how culture and social norms, once established, are enmeshed and inseparable from materiality in virtue of the activities of agents that continuously scaffold their niches. Through their histories of interactions and participation to communal life, agents sustain the existence of meaningful affordances and, in the case of human beings, manage to be attuned towards the relevant aspects of culturally organized surroundings. Culture and social practices need thus to be conceptualized neither as something projected from the head of the perceiver to a culturally meaningfulness environment nor exclusively in the world. Culture can be instead thought relationally and emerging from a long term dialectic of mutual symbiosis between the social environment and its inhabitants (Donald, 2000). Despite their individual differences, scholars coming from different traditions and disciplines have drawn very similar conclusions regarding the intertwinement between social practices and materiality. A clear example can be seen in the so-­ called mutualists or authors particularly inspired by the pragmatist tradition (Still & Good, 1992; Good, 2007). These authors have often highlighted that, to make sense of everyday behavior, it is necessary to accept that we can directly perceive affordances through signs, symbols, artifacts and the social practices usually considered by orthodox ecological psychologists as simple medium for perception (Noble, 1981; Windsor, 2004; Van Dijk & Kiverstein, 2020). A similar line of reasoning can also be seen in authors that are particularly close to the niche construction theory such as Walsh (2015), in which a relation of mutual co-determination and inseparability between organisms and environment or, in the specific case of the author, between organisms and the landscape of affordances is advocated. Finally, the most recent proposals that seem to follow this direction have been proposed by the approaches that nowadays defend a relational ontology of affordances (Chemero, 2009; Rietveld & Kiverstein, 2014). Kiverstein (2020) has provided an example that clarifies how it is possible to think of affordances through an explicit relational ontology in which the existence of affordances and of the

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

287

organisms that rely on them is one of mutual determination. Affordances actually depend on the material existence of environmental surroundings. However, at the same time, the material surroundings and the ecological meaning they embody in a specific niche cannot be understood independently from the  form of life that makes regularly use of them and that collectively scaffolded the environment in the first place.4 It is because of continuous processes of social scaffolding that it possible for the members of the same form of life to perceive a meaningful and shared landscape of affordances. The example used by Kiverstein is the one of a bird and its nest. On the one hand, it is undeniable that the nest exists in virtue of its material configurations. At the same time, its same components and features depend on the past activities of the members of the same group of birds and the abilities that derived from their phylogenetic and ontogenetic histories. However, if the birds would stop using the next, its affordances would decay and started being used by other species. If we think instead of the case of human beings living in rich but continuously changing socio-cultural surroundings, a similar example can be seen in the affordances of a Japanese sword. The sword offered specific activities if perceived by a swordsman in the Edo period. However, today, the same weapon, even by maintaining the same identical material configuration, offers very different action possibilities when perceived by a common citizen inside a museum. The affordances offered by the nest or by the Japanese sword, simply decay or disappear if a species or a form of life do not contribute to their maintenance through their activities.

16.5 Empirical Evidence and Further Challenges The necessity of reconciling the material dimension of affordance perception with the social one is an issue mostly highligted by scholars looking for theoretical parsimony. However, the scepticism towards the distinction between first from second-­ hand perception also  finds feedback in the recent advancements of empirical affordance research and embodied cognitive science. In the last 10 years, the idea that affordance perception is inherently filtered by the socio-cultural context has animated a large part of laboratory work. As a matter of fact, if on the one hand it is undeniable, as Reed and Heras-Escribano claim, that affordances drive our behavior in a way that is conformed to our socio-cultural practices, on the other hand, empirical researchers rarely deny the idea that affordances are pre-reflectively and spontaneously experienced as emerging from the constellation of practices where they are typically used. Borghi (2018), for example, reviewed significant evidence that in part challenges the hypothesis that perception is not permeated by the social and cultural  The notion of form of life is in this context borrowed from the late-Wittgenstein and is used by Kiverstein and, in general by defenders of the relational approach to affordances known as the Skilled Intentionality Framework to grasp the stable and regular way of doing things together of a group of animals or individuals (see also Rietveld & Kiverstein, 2014) 4

288

G. F. Artese

background of the perceiver. In the first place, she noticed that, in recent years, it has been possible to observe a fundamental paradigm shift in affordance research. Action possibilities have slowly begun to be considered as necessarily activated in a highly task, goal and context-dependent manner rather than independently of the situation or the surroundings in which they are embedded. Starting from a relational perspective on affordances (see Chemero, 2009), she discussed an impressive number of studies showing the influence of the social, physical, and linguistic context on affordance perception and behavior. Similarly to Cisek and Kalaska (2010), she suggests that affordance perception non-trivially depends by an ongoing competition among different affordances continuously driven and filtered by the social context and the functional relationships between the objects perceived. Socially and functionally related contexts are well known for facilitating the recognition of individual objects (Bar, 2004; Boyce et al., 1989). However, in a recent study, Borghi et  al. (2012), inspired by similar findings reported by Yoon et  al. (2010), also showed that similar effects can be observed with affordance perception. In these two studies, participants were asked to press a button when the objects perceived, which were always presented in pairs, were functionally related or used in similar contexts. Results clearly showed that faster responses were elicited based on the location, functional relationships (e.g., scissor-paper, butter-knife), and the typical contexts in which the utensils are generally used. These studies indicate that functional relations and social contexts make it possible for participants to react faster and provide more immediate responses in contrast to the perception of two unrelated objects (or affordances) embedded in atypical settings. A reasonable interpretation of these studies is that perceivers are not just sensitive to individual affordances, but to the structured social context in which several interconnected affordances are embedded. Importantly, studies as the ones above do not represent direct evidence against the hypothesis that the social context does not influence perception since they rely on categorization tasks rather than actual perceptual or behavioural ones. At the same time, the results seem to justify inferring that the context and the relationship between functionally related utensils pre-reflectively influence and facilitate how their related action possibilities can be detected in everyday social settings. One of the versions of the experiment conducted in the study by Yoon and colleagues is particularly in line with this interpretation. They replicated the same study and found no significant effects when words were used instead of pictures. The interpretation of Yoon and colleagues is that these results are observable because, in contrast to the abstract semantic level, perceiving objects that are familiar between each other directly activates a visual route to action (2010, p. 822). A more ecologically valid study has been instead conducted by Wokke et al. (2016). In their study, they recreated two ecologically valid behavior settings, a kitchen and a workshop. Participants took part in a go/no-go task in which they had to press a spacebar when an object appeared on the screen and hold themselves from doing it when it was obscured by a black X. Importantly, participants were instructed to perform the task as fast as possible. The objects perceived were respectively items congruent or incongruent with one context or the other. In the first place, the study showed that reaction

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

289

times were faster when tools and utensils were perceived in their familiar environment. In the second place, an EEG analysis revealed an enhanced N2 component, generally associated with a response to conflictual situations, when the tool was not congruent to its typical social context. Following the interpretation of the authors, but also of Borghi (2018), “the findings of faster responses and an enhanced N2 on context congruent trials suggest that the environment influences the strength of the perception–action coupling” (Wokke et al., 2016, p. 145). While it is particularly hard to make fully explicit how third-person studies can contribute to our understanding of the experience of affordances, the hypothesis that social and contextual factors directly permeate perceptual processes is not too surprising, especially in light of the results of embodied and cognitive neuroscience. As embodied theorists such as Simone Schnall pointed out (2017), nowadays, numerous studies and meta-reviews reported that early visual areas are continuously modulated by motor areas, reward areas and from past experiences more in general (see O’Callahan et al., 2017; Shuler & Bear, 2006). The most striking evidence reported by Schnall in this regard is a study made by Fujiwara et al. (2017) that demonstrated that even in blind fruit flies, visual areas are altered by motor feedbacks associated with leg movements. At the same time, the investigation of motor areas in primates, and in particular of the mirror neuron system and the mu wave suppression, demonstrated that these regions are not simply related to the execution and observation of actions but are also selective towards the (socially meaningful) context in which these same actions are performed (Iacoboni et al., 2005). It is well-known that these areas are sensitive to social significance, the acquired social capacities of the perceiver (Oberman et al., 2007) and are, more generally, continuously shaped by her individual history of learning (Orgs et al., 2008). In light of similar studies, but also behavioural ones, Fox et  al. (2020) have suggested that there are valid scientific reasons to embrace the possibility that cultural factors and social practices shape on the long term the neural activation patterns related to the observation and execution of situated actions. Notably, neuroscientific findings have historically found some resistance in ecological psychology and in radical embodied cognitive science more in general. This partly depends on the fact that cognitivist and weak embodied approaches often interpret these studies in terms of internal models of some states-of-affairs of the world. A paradigmatic case can be seen in the mirror neuron system activation that is often interpreted in terms of motor representations or simulations of other individuals’ actions. While these neuro-centric interpretations can be said with no hesitation to be unattainable from a non-representationalist and ecological perspective, these study are nonetheless important in helping us thinking how affordance perception is continuously shaped and modulated by the agents’ phylogenetic and ontogenetic histories. As argued by Shaun Gallagher, mirror neurons, and neural dynamics more in general, need to be understood as contributing to action and perception not as simulating the actions of others but as “by being part of larger sensory-motor processes that respond to different interaction to affordances” (Gallagher, 2017, p. 12). Ultimately, despite the difficulties and the limits of the empirical investigations today available, the studies here mentioned raise some important challenges against

290

G. F. Artese

the idea that the socio-cultural aspects of our environment cannot influence affordance perception if not through situated forms of rationality and reason. On the contrary, the immediacy and automaticity with which the social context influences the instrumentality of the objects perceived seem to imply that the socio-cultural background may permeate affordance perception without necessarily assuming the intervention of rationality or linguistic concepts.

16.6 The Value of Phenomenological Analyses and the Concept of Solicitation in the Study of Perception-Action Cycles Probably, the main complaint that can be raised against the claim that culture and sociality influence behavior but leave perception unaffected rests on its phenomenological implausibility. It is worth acknowledging that most of the discussions developed so far have been partly anticipated by the post-Husserlian phenomenological tradition that focused on the social structure of everyday experience. However, fairly recently, some of the central concepts that stemmed from phenomenological analyses, especially in the work of Merleau-Ponty and Dreyfus, have been integrated and transformed into concepts available for affordance theorists. In this section, the phenomenological work of authors who have addressed the problem of social perception and the implementation of their phenomenological concepts in contemporary affordance research are going to be examined. These latter ideas provide a preliminary starting point to support the methodological proposal made in the last section. The idea that socio-cultural factors are not reflected in the perceptual appearance of affordances is contradicted by most of the phenomenological investigations of our everyday social world as encountered in pre-reflective experience.5 For example, authors like Alfred Schutz (1945) and Aron Gurwitsch (1979) argued that the most primordial way of perceiving our surroundings is fundamentally cultural. In everyday perception, for Gurwitsch we do not perceive brute material objects but we directly deal with: works of art, buildings which serve specific purposes like abodes, places for work, schools, libraries, churches…tools, instruments, and utensils related to human needs and desires” (Gurwitsch, 1974, p. 143). The cultural meaning of these objects is not something projected over their material configurations but directly given in experience in virtue of their pragmatic usages and the roles played in the system of purposes and activities of a society. Another fundamental contribution on this issue can be seen in the work of Hubert Dreyfus that, inspired by Merleau-Ponty and Heidegger’s existential phenomenology, for long has argued that experience is not something that can be explained by considering perception as a neutral process subsequently enriched by extra non-­ perceptual cognitive processes. In contrast, phenomenological facts are already  See Moran (2017) for an excursus on the phenomenological tradition on this issue.

5

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

291

given to the perceiver as meaningfully organized and permeated by what he defines as “situational significance”. He claims that the everyday world is given as already organized in terms of significance and relevance, and that significance and relevance, and that significance can’t be constructed by giving meaning to brute facts—both because we don’t normally experience brute facts and, even if we did, no value predicate could do the job of giving them situational significance. (2007, p. 1150)

Perceptual experience is never given to the perceiver as a mere set of affordances that will be then selected and interpreted according to the individual’s social background. What is directly perceived are rather the valences and roles that affordances play within concrete and socially meaningful situations. Phenomenologically speaking, perception is never just the detection of affordances, but a direct awareness of meaningful “Gestalten” in which some action possibilities are perceived as more relevant and appropriate given the circumstances than others. It is not a coincidence that some of the central figures of the phenomenological tradition such as Heidegger, Schutz, Merleau-Ponty and Gurwitsch, despite their differences, converged in arguing that while absorbed and pragmatically engaged in everyday situations, subjects do not necessarily need to rely on the usage of concepts or thoughts to deal with their social surroundings. In everyday interactions, Dreyfus and the phenomenological tradition describe agents as able to skilfully and unreflectively deal with the environment in virtue of their capacity of being guided by the structure and the pragmatics of the social context rather than from concepts or explicit rationality. Gurwitsch cannot express this idea more clearly: [W]hat is imposed on us to do is not determined by us as someone standing outside the situation simply looking on at it; what occurs and is imposed are rather prescribed by the situation and its own structure; and we do more and greater justice to it the more we let ourselves be guided by it, i.e., the less reserved we are in immersing ourselves in it and subordinating ourselves to it. We find ourselves in a situation and are interwoven with it, encompassed by it, indeed just “absorbed” into it. (1979, p. 67)

To frame how agents are actively driven by the social context in which they are geared, Dreyfus and Kelly (2007) proposed distinguishing between simple affordances and solicitations. The idea behind the notion of solicitation can be traced back to the early theories of gestalt psychologists of Lewin and Koffka that, on several occasions, pointed out that objects can temporarily invite and solicit specific activities. A solicitation is an affordance experienced as permeated by a demanding character that invites the perceiver to act without the necessity of deliberating, deciding or speculating about the possible outcomes of her behavior. The experience of a solicitation (or motivation if we wish to maintain the language used by Merleau-Ponty) necessarily manifests a normative dimension that does not have to be understood as a psychological phenomenon but rather as a perceptual one (Dreyfus, 2005, p. 56). A solicitation is thus described as the experience of an affective tension, or a deviation from the norm, that can be reduced only through an appropriate action. The notion of “normative” does not need to be understood in terms of some truth-conditions of a mental representation of a future action, but

292

G. F. Artese

aimed at grasping the experential character of a future behavior that is felt as adequate or correct in respect to the situation. For Dreyfus, intentionality, if understood as a mental representation of the eventual outcomes (understood in terms of success or accuracy conditions) of the agent’s actions, is not required and often not relevant for everyday behavior. Solicitations, on the other side, are related to what Dreyfus defines as the “conditions of improvements” or “satisfaction conditions” of an intentional action required by the pragmatic motifs of the social context. In the case of solicitations, what is experienced is enough for the agent “to move as to reduce a sense of deviation from a satisfactory gestalt without the agent knowing what that satisfactory gestalt will be like in advance of achieving it” (Dreyfus, 1999, p. 4). Over the past decade, the need to differentiate affordances from solicitations has moved from being just a phenomenological distinction to a potential concept to be integrated into ecological experimental contexts. (Withagen et  al., 2012, 2017; Käufer & Chemero, 2021; Rietveld & Kiverstein, 2014). Even more recently, the first attempts to highlight the main qualitative features of their perceptual appearances have also been formulated (Dings, 2018). At present, the use and deployment of the notion of solicitation represents a central concept for those who assume that sociality permeates perception all the way through, characterizing the action possibilities imposed by the social situation as having a demand character that is directly and pre-reflectively felt by the agent. However, a question that spontaneously follows from these observations is how the notion of solicitation is related to socio-­ cultural practices. While the figures of the phenomenological tradition just presupposed the existence of a social and intersubjective environment, the one of solicitation seems to be a concept apparently just related to the subjective concerns and needs of the perceiver. However, Rietveld and colleagues, by suggesting that solicitations are deeply intertwined with the agent’s participation in communal practices (Rietveld, 2008; Klaassen et al., 2010; Van Dijk & Rietveld, 2017), have offered a contemporary reading and an implementation in affordance research of the insights of the phenomenological tradition that focussed on a characterization of experience of our everyday social surroundings. In contrast to the work of neo-Gibsonians discussed in the previous sections, Rietveld and colleagues propose a less intellectualist reading of how social norms constrain action and perception. Inspired by Dreyfus and the late Wittgenstein, they argued that there is a direct link between the agent’s unreflective response to solicitations and socially constituted norms. The source of normativity experienced while being “invited” by a solicitation depends on the participation in the socio-cultural practices that are crystalized in the structure of our shared and scaffolded surroundings. In this context, social norms do not have to be interpreted as formal rules or concepts but rather as something emerging from our stable and regular way of doing things together, or, the same Wittgenstein puts it, from the implicit “background of massive agreements in actions” (Rietveld, 2008, p. 984). Experiencing a solicitation is not equivalent to making explicit a social norm from a third-person perspective. What is relevant for a solicitation to be manifested cannot be captured by general abstractions but needs instead to be contextualized in the light of the histories of the individual members of a social group and in their  regular and

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

293

collective behavior. Experience and social norms are intrinsically related and reflect the existence of an environment that is inherently social and cultural. On the one hand, the deviations from what is required from social norms are reflected in the agent’s phenomenology as an affective tension demanding her to perform a specific action. On the other hand, the practices, norms, and customs can be said to exist because of the collective activities of the members of a socio-cultural community that maintain the existence of affordances by continuously responding to meaningful social contexts. Still, despite the phenomenological support, empirical evidence and the theoretical advancements achieved by these theorists, a proper methodology to fully analyse and study solicitations in experimental settings has not been fully consolidated yet. Among the toolbox of concepts that might be required to integrate the social and cultural dynamics of affordance perception to the already existing methods of radical embodied cognitive science, the rest of the article will focus on the frequently used notion, but not necessarily fully articulated, of “field”. A concept that, I claim, is necessary, for both methodological and historical reasons, to a characterization of affordance experience in socially structured niches like ours.

16.7 The Notion of Field: A Brief History and Its Implications If perception, and not just behavior, needs to be investigated as an activity situated and unfolded in variegated socio-cultural contexts, an increase of the methodological resources available in affordance research is necessary. In other words, it is important to be explicit on the methodological tools that need to be applied at the immediate perceptual level in order to describe the perceptual appearance of affordances situated in social contexts. In this regard, a crucial concept that plays a central role in contemporary debates is the one of field. While the latter has a long history in psychological and philosophical disciplines, in recent times, its usage became very popular to characterize the experience of socio-cultural settings in which different affordances are present at the same time. Despite the clear and renovated interest that the notion of field is receiving, the notion has been employed in various ways. Among the different examples it is possible to mention Bruineberg and Rietveld (2014) that, with their Skilled Intentionality Framework, argue that agents always deal with a “field of relevant affordances”. In this context, the notion of field is used to make explicit that affordances are always perceived as relevant for a particular individual and in relation to a specific situation. Kadar and Effken (1994) have instead used the notion of field as a primary ontological entity that was supposed to represent an alternative to the objectified notion of ‘thing’ introduced in Turvey’s work on the ontology of affordances (1992). For Kadar and Effken, the notion of field was described as necessary to make space for the one of intentionality in ecological psychology and to do justice to a

294

G. F. Artese

conception of affordances as inseparable from a dynamic and ever-changing environment. Despite the  differences between the works mentioned, something that immediately emerges from these considerations is that the notion of field is generally used to highlight a dynamical relation of mutual interdependence between affordances and the respective environmental context in which the subject of experience is situated. At the same time, the more fine-grained theoretical implications of the term, as currently used in contemporary debates, can vary and are either not exhaustively developed or often taken for granted. However, the fact that different meanings have been attributed to the notion of field is not a novelty. This concept has a way longer history than the debate about affordance perception in ecological psychology. The notion of field finds its roots in physics and very often, authors that rely on this notion directly take inspiration from the features and characteristics observed in electromagnetic fields. The metaphor of the electromagnetic field is nicely grasped by the example of an electrically charged body approached with a metal stick. If this phenomenon takes place, it is possible to influence the local repartition of electricity in the charged body. However, these changes will not only alter the electrical repartition of small and circumscribed areas. In contrast, what will be observed will be a reorganization of the structure of the whole field. In other words, a local change in the field will establish a new structured field with a different electrical distribution (Gurwitsch, 1966). Importantly, although the notion of field has been made popular by Gestalt theorists,6 Kadar and Shaw (2000) rightly noticed that the notion has also found space in the work of behaviorists such as Hull and even in the same Gibson in one of his early works with Crooks (1938) in which they tried to characterize the valences and values perceived from the perspective of a car driver. Finally yet importantly, the concept of field has also been employed in the work of pragmatists like James and in the phenomenological tradition initiated by Husserl and later on developed by Gurwitsch, Sartre and Merleau-Ponty. Regarding its more specific usages in psychological disciplines, the notion of field has been developed in greater detail by the Berlin Gestalt school of psychology. Wolfgang Köhler (1947), for example, argued in favour of the existence of some magnetic electrical fields in the central nervous system that were supposed to be isomorphically congruent with the invariant structure of all subjective experiences. Another author that made popular the notion of field is the Gestalt psychologist Kurt Lewin. He used the notion of field to characterize the “life-space” or the “psychological environment” of the perceiver. Lewin’s work (1951), and in a similar way the one of Koffka with his distinction between geographical and behavioral environment (1935), was constantly guided by the concern of catching the differences between a physical and subject-independent reality from the subjective world in which valences and vectors of forces are experienced. Inspired by the notion of Umwelt (or subjective environment) proposed by the ethologist Jakob von Uexküll,  It is also interesting to notice that Lewin’s Field theory, as much as the work of Koffka, have directly inspired Roger Barker’s Behavior Setting theory; an approach to study social behavior in an ecological fashion that in recent years has received a large attention in 4E and ecological theories (see Heft, 2001; McGann, 2014) 6

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

295

Lewin described the life-space as inherently meaningful in virtue of a multitude of factors related to the needs, concerns and goals of the perceiver. While the general commitments of the notion of field are generally shared across all these authors and traditions, some others can drastically change based on the approach or theory in which it is used. The general idea behind its usage is to indicate a phenomenon, in our case sensory experience, in which phenomena such as forces, vectors and valences need to be approached in a non-atomistic and interdependent fashion. As Kadar and Shaw claim, in psychology, the general definition that seem to be shared among all these theorists and tradition is the one offered by Lewin. He claims: “A totality of coexisting facts which are conceived of as mutually interdependent is called a field” (2000, p. 143). Framed in these terms, the value of the notion of field should be self-evident in relation to what has been discussed so far. The field metaphor highly resembles the phenomenon of social context since the latter, just as a electromagnetic field, cannot be simply reduced to its single parts or elements. Gurwitsch nicely illustrates the intertwined relation between an active agent and her environmental context through some of the core concepts typical of field theories. We are never […] in the presences of multiplicities of isolated sensations, volitions, impulses to act; we are always placed in a field of perception, of action, etc. These fields have their proper articulation: they have a center with regard to which they are organized, they have privileged or “vectorial” directions tending towards the center, and they have slopes. If an animal is placed in front of food, the animal is placed in a field of action whose center is the food and whose vectorial directions are constituted by the most direct accesses to the food. The animal follows these directions when rushing to the food. (Gurwitsch, 1966, pp. 62-63)

Even if these considerations are framed just generally and related to the valences and the action drives experienced by an animal, the interdependency and holistic character embodied in the notion of field may turn to be important to formalize how a structured socio-cultural setting can drive a perceiver into performing some actions instead of others. Importantly, understood from a radical embodied and non-­dualistic perspective, the same perceiver needs to be considered within the field and as an actor participating in a situation taking place in a physical and cultural environment. At the same time, the field also reflects the needs, concerns and behavioural tendencies of the agent changing over time. By appreciating this ambivalence, in which a field is considered as inherently material but also as cultural and relative to the experiences of the perceiver, it becomes possible to do justice to Gibson’s attempt of overcoming the natural vs. cultural dichotomy. The field is then not a description of the private and phenomenal reality of the perceiver, but of her individual perspective on a cultural and shared environment in which she acts and participates. While this interpretation might appear controversial, especially in light of its internalist and private connotations that Gestalt psychologists assigned to it, this understanding of the concept of field is instead in line with many of the ideas embraced by the authors and positions that showed some resistance towards Gibson’s take on the role of culture and sociality. Given the above, some conclusive considerations about the notion of field need to be added. Because of its large number of formulations and superficial usages, the concept of field has often been accused of being way too obscure and imprecise and

296

G. F. Artese

thus unable to represent a concept rigorous and precise enough to be employed in scientific practice. As noticed by Staemmler (2005) and Parlett and Lee (2005, p. 45), one of the risks of the notion of field is that, on several occasions, it has been so liberally and broadly framed such as to include “anything and everything”. The risk here highlighted is that the strong interdependence that is assumed to exist among the individual components that constitute a field might make it difficult for affordance theorists to individuate the aspects of the social context that make it possible for an action possibility to be perceived as relevant and socially meaningful. In order to represent a reliable tool to study situated affordance perception, the notion of field should clearly not include “anything and everything” but rather be relevant to describe the aspects of the environment that lead the perceiver to experience a solicitation as such in the first place. In the next section, I will argue that, among the different options that can be observed in the history of psychology, the understanding of field provided by the phenomenologist Aron Gurwitsch with its differentiation between Theme, Thematic Field and Margin has been formalized rigorously enough to fulfil this role.

16.8 Gurwitsch’s Field of Consciousness and Its Methodological Implications: A Phenomenologically-Oriented Proposal The concerns stemming from the concept of field mentioned at the end of the previous section are not too different from some of the criticisms that have also been raised in recent years against embodied and ecological approaches in cognitive science. More specifically, dynamical and non-representational approaches such as ecological psychology and enactivism, a further non-representational approach in situated cognition research, have been accused of not providing a clear distinction between causal and constitutive factors in the study of cognition. Quite similarly, they have also been criticized for focussing on an overbroad and unnecessary number of variables in the study of action-perception cycles and, consequently, of not being capable of providing some clear boundaries for their phenomena of investigation (see Adams & Aizawa, 2008 among the others). Because of these criticisms, radical embodied approaches have often been called for scientific rigor and required to develop methodological tools capable of dealing with these issues. It is here suggested that the notion of field of consciousness proposed by the phenomenologist Aron Gurwitsch can fulfill this role and represent a suitable starting point to characterize the experience of affordances situated in their social contexts without relying upon an overly liberal concept of field. The affinities of the thought of Aron Gurwitsch with the themes discussed in this article are not superficial. Gurwitsch is considered, with his friend Alfred Schutz, one of the first post-Husserlian social phenomenologists, and is often remembered for having emphasized how our fundamental and daily mode of engagement with

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

297

reality is inherently cultural. In our daily life, Gurwitsch suggests that the instrumentality of tools is always perceived in relation: “to a particular society, the circumstances and conditions under which that society lives and a system of purposes recognized by it” (Gurwitsch, 1974, p. 93). However, while the role of culture and social interactions had been described in several articles (1966, 1974) and his dissertation “Human Encounters in the Social World” (1979), it is in his monography “The Field of Consciousness”, (2010) that he described a notion of field complementary to these considerations. Gurwitsch proposed what could be considered as the first gestalt-inspired phenomenological notion of field by creatively combining some of the most interesting ideas of Husserl, Goldstein, Koffka, Piaget, Schutz and James, among others. Gurwitsch’s field of consciousness does not ultimately have the aim to characterize individual experiential episodes but rather to grasp the invariant features and structural domains of all conscious experiences. As a student of Goldstein and Kelb, he was in strict contact with Gestalt psychologists and was influenced by their scepticism towards the atomistic theories of sensory perception of their times. He was also a careful reader of William James that inspired him in theorizing experential episodes as sharing a field structure. However, in contrast to James, who argued that experience always presents a central focus that emerges out of a chaotic and indefinite background of fringes, Gurwitsch defended the idea that experience is always structurally organized and autochthonously given in a gestalt fashion. For Gurwitsch, the field of consciousness (sometimes just defined as the field of attention, experience or awareness) necessarily presents three main domains that remain constant throughout all our conscious life. The first domain is the Theme. The Theme overlaps with the focus of attention of the perceiver or, as Gurwitsch puts it, with “what engrosses the mind of the experiencing subject” (2010, p. 58). Similarly to James’ notion of focus, Gurwitsch claims that, in all their conscious experiences, perceivers are always intentionally directed at something that occupies a privileged place in the field. Adjacent to the Theme, Gurwitsch mentions the existence of a second domain, the Thematic Field. The latter is a domain of relevance and has the role of determining the perspective, attitude, and context from which the objects in the Theme are experienced. The connection of the Thematic Field with the Theme can be described in terms of a figure-background relationship like the one observed in the Rubin’s vase. While the Theme is independent from the Thematic Field, the existence of the latter plays a pivotal and necessary condition for the former to be perceived as belonging to a meaningful context. Finally, Gurwitsch also mentions the existence of the last domain, the Margin. The Margin is a domain of irrelevance that is related to neither the Theme nor the Thematic Field, but nonetheless always dimly present in our field of attention. The Margin is functionally detached from the rest of the field and perceived just because of mere temporal reasons. In light of the discussion developed so far, the three domains of Gurwitsch’s field of consciousness need now to be examined one by one and contextualized in relation to the debate on affordance perception and the social context. As has been claimed, the Theme overlaps with what we might define as the focus of attention. The existence of what is experienced in the Theme is constitutively independent of all the other domains of the field. Just as the Rubin’s vase can be

298

G. F. Artese

experienced if juxtaposed over several backgrounds of different colours, a Theme can be experienced over a different Thematic Field. The same cannot however be said for the internal elements within it. The elements that constitute a Theme mutually point at each other in such a way to exist as a meaningful ensemble. These constituents thus exhibit a certain functional weight and degree of gestalt coherence towards each other in such a way to make it possible for the perceiver to deal with a unified Theme. If one of the elements is removed, the overall structure of the Theme dramatically changes. At the same time, the Theme is independent of the rest of the field, but it is not outside of it. Following Gurwitsch, we never perceive isolated sensations or raw thoughts. However, we never deal with individual items and objects of perception either. The Theme, for Gurwitsch, “carries a field along with it so as not to appear and be present to consciousness except as being in, and pointing to, the field” (2010, p.  133). In other words, while the elements within the Theme need each other to meaningfully constitute what is inside their domain, it is actually the overall structure of the field, and in particular the Thematic Field, that determines the contextual meaning, the salience and perspective from which the Theme is perceived. In making these considerations, Gurwitsch is here concerned with the conditions of possibility for a Theme to be considered as such and be manifested in experience. While the discussion over the characteristics of the field’s domains might appear too formal or overly abstract for an affordance theorist, it is also crucial to keep in mind that Gurwitsch eloquently characterized everyday perception in an action-oriented fashion by anticipating contemporary embodied approaches in cognitive science. In particular, he relied on the concept of “functional character” proposed by Koffka and Köhler to grasp the sense of instrumentality with which objects present themselves in perception. In pre-reflective experience, we do not perceive a hammer as a t-shaped object of a particular colour and size but as a proper tool perceived as useful for certain purposes and to fulfill my aims and intentions. It is also interesting to notice that for Gurwitsch, the teleological and functional character of utensils, tools, and social roles is precisely what determines their role in a cultural group or a society (Gurwitsch, 1974). However, in contrast to the affordance theorists that, on most occasions, just focused on the relationship between agents and individual affordances, Gurwitsch assumed a cultural and developmental perspective and thought of functional character as developed and emerging in the context of concrete action and always experienced in relation to socially structured contexts. He claimed: Functional characters accrue to objects in situations of concrete action in which the subject manipulates the object, learns to handle it in a determinate manner, to use it for a certain purpose in connection with other objects, and thus acquires a certain mode of action. Such acquisition will henceforth codetermine future perception. (Gurwitsch, 2010, p.  96, italics added)

The claim that the functional character of a tool is developed in virtue of how it is used in connection with other objects is an important aspect that is necessary to

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

299

keep in mind while moving to the second domain of the field of consciousness, the Thematic Field.7 The elements experienced in the Thematic Field guarantee that what is inside the Theme is perceived as pertinent to a specific context. The Thematic Field is, by definition, always experienced as functionally and materially relevant to the Theme. The elements experienced in the Theme and the Thematic Fields are reciprocally connected by their functional values and are likely to be found in similar situations and contexts. As a matter of fact, what has been defined as “functional character” by Gurwitsch is not an individual property of an object but something dependent on the practical situation and on the interrelation between utensils, activities and individuals that are geared and constitute the same social context (Gurwitsch, 1974). As Gurwitsch puts it, the elements inside the Theme and the Thematic Field belong to the same “order of existence”. Gurwitsch uses the notion of order of existence to refer to systematic contexts of objects, activities, and situations that are functionally related to each and that generally reflect the “natural groupings” and organization of pre-reflective experience. Roughly, an order of existence can be understood as the wider (socio-cultural) context directly perceived altogether with the meaning offered by individual affordances. All orders of existence exhibit their own functional and relevancy principles and, at least for what concerns the ones related to our everyday reality, are described by Gurwitsch as spheres of actions delimiting one another like our professional activity, the several public spheres in which we act as citizens, our family life etc. (2010, p.  373). If these considerations are

 It is now possible to compare it with Gibson’s asocial conception of affordance perception. In the passage mentioned in Sect. 16.2 (1972, p. 88), Gibson thought of affordances as exclusively determined by the stimulus “and nothing else”. Gurwitsch, because of the influence of Piaget and Gestalt psychology (in particular Koffka and Köhler), assumed an opposite stance and characterized perception as a developmental and historical process. The Lithuanian phenomenologist follows Gestalt psychologists in claiming that perception cannot be understood independently of the equally relevant contribution of the perceiver’s developmental history and environment. Gurwitsch defended the idea that the contributions of internal (organismic) and external (environmental) factors are both necessary and dependent conditions to understand how the objects’ functional connotations can be acquired, modified, and maintained. The dependency on internal and external conditions has been sometimes formalized through the formula P = f (xe, xi) (2010, p. 94). Here, a percept P is modified even if just one of the two constants goes through any change. These modifications might include the subject’s individual developmental history (internal conditions; xe) and the presence of new environmental stimuli (external conditions; xi). Very reasonably, by not offloading to the environment the role of fully determining the stimulus meaning, Gurwitsch managed to open a window towards the developmental and historical factors necessary to consider the socio-cultural dimension at the roots of affordance perception. This stance, however, does not need to be interpreted as the idea that reality is perceived through abstract concepts, rules or memories (that actually consisted in the view of Benussi, a Gestalt psychologist from the School of Graz that Gurwitsch heavily criticized and labeled for the dualistic commitments of his approach) but rather seen as a stronger form of relationalism. Gurwitsch followed the main idea promoted by Gestalt psychology that, as Piaget puts it, is s to explain intelligence by perception” (Ivi., p. 48). The field of consciousness, as much as the same existence of the orders of existence constituted by the activities of the members of a society needs thus to be seen as going through continuous re-organizations and changes as the same notion of field implies. 7

300

G. F. Artese

contextualized in relation to Gurwitsch’s field of consciousness, in the Theme and the Thematic Field, perceivers deal only with delimited subsections of the orders of existence that include the activities, social roles, objects and situations typical of their community or social group. For example, a coloured chalk exhibits different relevancy principles in respect to its usual surroundings in the educational or school order of existence compared with the same object perceived by some children in a playground. Ultimately, understood as a part of a broader order of existence, the Thematic Field needs to be conceptualized as having the role of determining the situational context, the attitude, the meaning, and the salience of the Theme’s content. The Theme instead always occupies a privileged location in respect to the Thematic Field that, in turn, defines the positional index of the former. If now the example of the Japanese sword previously mentioned is reconceptualised through Gurwitsch’s three domains of experience, it is possible to understand how the same object can belong to different orders of existence and, in virtue of the latter, affording completely different activities. The experience of a museum visitor that is perceiving the functional character of the sword will be accompanied by a Thematic Field belonging to an order of existence (one related to museum activities) in which its relevancy principles do not lead the agent to perceive the instrumentality of the object inside the Theme as a weapon. In contrast, the Theme-Thematic Field relationship perceived by a swordsman of the Edo period would be based on activities and further elements of an order of existence (one related to Edo swordsmen activities) generally connected to his social role and the typical usages attributed to the sword in that specific historical context. Finally, it is also important to clarify that the experience of the Theme-Thematic Field relationship does not only depend on the agent’s abilities and attunement towards the individual action possibilities generally associated to a specific object. The perspective from which the Theme is perceived also depends on the functional relationship between the items experienced in the here-and-now of the situation. Perceiving a writing utensil as belonging to a writing-related order of existence like the academic one solicits the perceiver to act in different ways if the same utensil is not experienced in its usual surroundings. In this regard, a handy example made by Gurwitsch that highly recalls what demonstrated by the studies illustrated by Borghi (2018) in Sect. 16.4, is the one of an inkwell. An inkwell, understood as a utensil related to writing, can be said, with no hesitation, to be part of well-circumscribed orders of existence usually related to the familiar, professional or bureaucratic life. The inkwell often shares its order of existence with objects that are likely to be found in the kind of activities and social settings in which it is generally used. Generically speaking, an inkwell is likely to be functionally and materially related to a desk, an office, papers, books, colleagues, certain recurrent situations and further writing-related activities. The objects, social roles and activities just mentioned are generally connected between each other through an intertwined web of functional relationships in such a way to form, following Gurwitsch, the “objective background” or the typical “utensil-totality” of the inkwell. This implies that a perceiver is more likely to be spontaneously solicited towards certain actions if the inkwell is perceived on a desk rather than in an unusual scenario like a piano.

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

301

The inkwell looks differently on the desk when I put it on the piano. It is displaced, “does not belong there”, is not found is in authentic milieu […] It looks differently, and its “functions” changes with the Thematic Field. (Gurwitsch, 1966, pp. 206–207)

Finally, the last domain of the field is represented by the Margin. As claimed already, in contrast to the Theme and the Thematic Field, the Margin represents a domain of irrelevance that is manifested in the field of consciousness just because of mere temporality. While using my computer to write this chapter, I could be vaguely aware of an aeroplane crossing the sky outside my window. This dim experience, since it is not in any way connected to affordances offered by my computer, remains in the Margin and is not likely to influence my writing flow as long as I am thematizing the affordances offered by my computer. Interestingly, for Gurwitsch there are at least three components that are always present in the Margin: (1) a minimal sense of our own embodiment; (2) a basic sense of the perceptual world; an (3) our own stream of thoughts. While these factors are always present and constant within the field of consciousness, they are not necessarily relevant, especially when relegated in the Margin to the affordances we are thematizing in our experiences. While the Margin might superficially appear as a domain of poor importance for the discussion here developed, it is because of its existence that Gurwitsch’s field of consciousness can represent a methodological resource for embodied theorists that aim to grasp the importance of the social context in everyday action and perception cycles. By taking into consideration the existence of the Margin, it becomes possible to take advantage of the concept of field and, at the same time, recognize and exclude what is functionally relevant in a given social context from what is not.

16.9 Conclusion Historically, as recognized by Kiverstein and Van Dijk, ecological psychologists have often hesitated in recognizing the effects of language and socio-cultural practices on direct perception (2021, p. 11). However, in this article, I have tried to give space and reconstruct the main considerations and proposals of the authors that consider the opposite possibility. After having reconstructed Gibson’s stance on the role that culture and sociality play for perceptual processes, I have discussed the works of some of his followers that tried to develop and put into practice his considerations. Immediately after, I have showed that the treatment that these authors reserve for the role of culture in perception has been criticized based on theoretical, empirical or phenomenological reasons. In parallel, I also reconstructed the ideas that animated the approaches hostile to the claim that culture and sociality do not permeate perception to its very bottom layers. Finally, I have concluded by developing the implications of one of the methodological tools used by characterizing the experience of affordances embedded in social contexts; the notion of field. Based on what was argued in Artese (2022), I argued in favor of communion between ecological psychology and some of the central ideas, in particular his notion of field,

302

G. F. Artese

proposed by the phenomenologist Aron Gurwitsch. If the considerations made are correct, Gurwitsch’s three domains of experience can help make explicit how individual affordances can solicit the perceiver in virtue of contextual factors by establishing, as Heft (1989) puts it, a higher-order relationship between a focal object and some surroundings that specify their functional significance. The methodological resources found in Gurwitsch’s field of consciousness can help us systematically think about how affordances are always perceived in relation socio-materially organized surroundings.

References Adams, F., & Aizawa, K. (2008). The bounds of cognition. Blackwell. Artese, G. F. (2022). Gurwitsch’s field of consciousness and radical embodied cognitive science: A case of mutual enlightenment. Journal of the British Society for Phenomenology, 53(2), 177–192. Bar, M. (2004). Visual objects in context. Nature Reviews. Neuroscience, 5, 617–629. Bickhard, M. H., & Richie, D. M. (1983). On the nature of representation: A case study of James J. Gibson’s theory of perception. Praege. Borghi, A. (2018). Affordances, context and sociality. Synthese, 199(5–6), 12485–12515. Borghi, A. M., Flumini, A., Natraj, N., & Wheaton, L. A. (2012). One hand, two objects: Emergence of affordance in contexts. Brain and Cognition, 80(1), 64–73. Boyce, S.  J., Pollatsek, A., & Rayner, K. (1989). Effect of background information on object identification. Journal of Experimental Psychology. Human Perception and Performance, 15, 556–566. Bruineberg, J., & Rietveld, E. (2014). Self-organization, free energy minimization, and optimal grip on a field of affordances. Frontiers in Human Neuroscience, 8, 599. Chemero, A. (2009). Radical embodied cognitive science. MIT Press. Chow, S. I. (1989). An intentional analysis of “affordance” revisited. Journal for the Theory of Social Behavior, 19, 357–365. Cisek, P., & Kalaska, J. F. (2010). Neural mechanisms for interacting with a world full of action choices. Annual Review of Neuroscience, 33, 269–298. Costall, A. (1995). Socializing affordances. Theory & Psychology, 5(4), 467–481. Costall, A., & Still, A. (1989). Gibson’s theory of direct perception and the problem of cultural relativism. Journal for the Theory of Social Behavior, 19, 433–441. Dennett, D. C. (2017). From bacteria to Bach and back: The evolution of minds. WW Norton & Company. Dings, R. (2018). Understanding phenomenological differences in how affordances solicit action. An exploration. Phenomenology and the Cognitive Sciences, 17, 681–699. Donald, M. (2000). The central role of culture in cognitive evolution on the myth of the “isolated mind”. In L. P. Nucci & G. B. Saxe (Eds.), Culture, thought, and development (pp. 19–38). Lawrence Erlbaum Associates. Dreyfus, H. (1999). The primacy of phenomenology over logical analysis. Philosophical Topics, 27(2), 3–24. Dreyfus, H. (2005). Overcoming the myth of the mental: How philosophers can profit from the phenomenology of everyday expertise. Proceedings and Addresses of the American Philosophical Association, 79(2), 47–65. Dreyfus, H. (2007). Why Heideggerian AI failed and how fixing it would require making it more Heideggerian. Philosophical Psychology, 20(2), 247–268.

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

303

Dreyfus, H., & Kelly, S.  D. (2007). Heterophenomenology: Heavy-handed sleight-of hand. Phenomenology and the Cognitive Sciences, 6(1–2), 45–55. Fox, N., Salo, V., Debnath, R., Morales, S., & Smith, E. (2020). The mirror neuron system and social cognition. In J.  Lockman & C.  Tamis-LeMonda (Eds.), The Cambridge handbook of infant development: Brain, behavior, and cultural context (pp. 495–519). Cambridge University Pres. Fujiwara, T., Cruz, T. L., Bohnslav, J. P., & Chiappe, M. E. (2017). A faithful internal representation of walking movements in the drosophila visual system. Nature Neuroscience, 20, 72–81. Gallagher, S. (2017). Enactivist interventions: Rethinking the mind. Oxford University Press. Gibson, J. J. (1951). Theories of perception. In W. Dennis (Ed.), Current trends in psychological theory (pp. 85–110). University of Pittsburgh Press. Gibson, J. J. (1966). The senses considered as perceptual systems. Houghton Mifflin. Gibson, J. J. (1972/2002). A theory of direct visual perception. In A. Noë, & E. Thompson (Eds.), Vision and mind: Selected readings in the philosophy of perception (pp. 77–89). MIT Press. Gibson, J.J. (1979/2015). The ecological approach to visual perception. Psychology Press. Gibson, J.  J., & Crooks, L.  E. (1938). A theoretical field-analysis of automobile-driving. The American Journal of Psychology, 51(3), 453–471. Good, J. M. (2007). The affordances for social psychology of the ecological approach to social knowing. Theory & Psychology, 17(2), 265–295. Gurwitsch, A. (1966). Studies in phenomenology and psychology. Northwestern University Press. Gurwitsch, A. (1974). Phenomenology and theory of science. Northwestern University Press. Gurwitsch, A. (1979). Human encounters in the social world. Duquesne University Press. Gurwitsch, A. (2010). The collected works of Aron Gurwitsch (1901-1973) Vol. III the field of consciousness: Theme, thematic field, and margin. Springer. Heft, H. (1989). Perceiving affordances in context: A reply to Chow. Journal for the Theory of Social Behaviour, 20(3), 277–284. Heft, H. (2001). Ecological psychology in context: James Gibson, Roger Barker, and the legacy of William James’s radical empiricism. Lawrence Erlbaum Associates, Inc. Heft, H. (2017). Perceptual information of “an entirely different order”: The “cultural environment” in the senses considered as perceptual systems. Ecological Psychology, 29(2), 122–145. Heras-Escribano, M. (2019). The philosophy of affordances. Palgrave Macmillan. Iacoboni, M., Molnar-Szakacs, I., Gallese, V., Buccino, G., Mazziotta, J.  C., & Rizzolatti, G. (2005). Grasping the intentions of others with one’s own mirror neuron system. PLoS Biology, 3(3), e79. Ingold, T. (1992). Culture and the perception of the environment. In E. Croll & D. Parkin (Eds.), Bush Base, Forest farm: Culture, environment and development (pp. 38–56). Routledge. Kadar, E., & Effken, J. (1994). Heideggerian meditations on an alternative ontology for ecological psychology: A response to Turvey’s (1992) proposal. Ecological Psychology, 6, 297–341. Kadar, E., & Shaw, R. E. (2000). Toward an ecological field theory of perceptual control of locomotion. Ecological Psychology, 12, 141–180. Käufer, S., & Chemero, A. (2021). Phenomenology: An introduction (2nd ed.). Polity. Kiverstein, J. (2020). In defence of a relational ontology of affordances. Constructivist Foundations., 15(3), 226–229. Kiverstein, J., & Van Dijk, L. (2021). Language without representation: Gibson’s first- and second-­ hand perception on a pragmatic continuum. Language Sciences, 85, 101380. Klaassen, P., Rietveld, E., & Topal, J. (2010). Inviting complementary perspectives on situated normativity in everyday life. Phenomenology and the Cognitive Sciences, 9, 53–73. Koffka, K. (1935). Principles of Gestalt psychology. Harcourt Brace. Köhler, W. (1947). Gestalt psychology. Liveright Publishing. Lewin, K. (1951). Field theory in social science. Harper & Row. McDowell, J. (2007a). What myth? Inquiry, 50(4), 338–351. McDowell, J. (2007b). Response to Dreyfus. Inquiry, 50(4), 366–370. McGann, M. (2014). Enacting a social ecology: Radically embodied intersubjectivity. Frontiers in Psychology, 5, 1321.

304

G. F. Artese

Moran, D. (2017). The phenomenology of the social word: Husserl on Mitsein as Ineienandersein and Füreinandersein. Metodo International Studies in Phenomenology and Philosophy, 5, 99–142. Noble, W. (1981). Gibsonian theory and the pragmatist perspective. Journal for the Theory of Social Behaviour, 11(1), 65–85. O’Callahan, C., Kveraga, K., Shine, J.  M., Adams, R.  B., & Bar, M. (2017). Predictions penetrate perception: Converging insights from brain, behaviour and disorder. Consciousness and Cognition, 47, 63–74. Oberman, L. M., Pineda, J. A., & Ramachandran, V. S. (2007). The human mirror neuron system: A link between action observation and social skills. Social Cognitive and Affective Neuroscience, 2(1), 62–66. Orgs, G., Dombrowski, J.-H., Heil, M., & Jansen-Osmann, P. (2008). Expertise in dance modulates alphabeta event-related desynchronization during action observation. European Journal of Neuroscience, 27(12), 3380–3384. Orlandi, N. (2016). Bayesian perception is ecological perception. Philosophical Topics, 44(2), 327–352. Parlett, M., & Lee, R.  G. (2005). Contemporary gestalt therapy: Field theory. In A.  L. Woldt & S.  M. Toman (Eds.), Gestalt therapy: History, theory, and practice (pp.  41–63). Sage Publications. Reed, E. S. (1993). The intention to use a specific affordance: A conceptual framework for psychology. In R. M. Wozniak & K. W. Fischer (Eds.), Development in context: Acting and thinking in specific environments (pp. 61–92). Erlbaum. Reed, E.  S. (1996a). Encountering the world: Toward an ecological psychology. Oxford University Press. Reed, E. S. (1996b). The necessity of experience. Yale University Press. Rietveld, E. (2008). Situated normativity: The normative aspect of embodied cognition in unreflective action. Mind, 117(468), 973–1001. Rietveld, E., & Kiverstein, J. (2014). A rich landscape of affordances. Ecological Psychology, 26(4), 325–352. Ryle, G. (1949/2009). The concept of mind. Routledge. Scarantino, A. (2003). Affordances explained. Philosophy of Science, 70(5), 949–961. Schnall, S. (2017). No magic bullet in sight: A reply to firestone and Scholl (2017) and Durgin (2017). Perspectives on Psychological Science, 12(2), 347–349. Schutz, A. (1945). On multiple realities. Philosophy and Phenomenological Research, 5, 533–576. Shuler, M. G., & Bear, M. F. (2006). Neurons in the primary visual cortex respond differently to a flash of light after it has been paired with a reward, unexpectedly showing that cognitive information is coded at this level in the cortex. Science, 311, 1606–1609. Staemmler, F.-M. (2005). A Babylonian confusion? On the uses and meanings of the term “field”. British Gestalt journal, 15, 64–83. Still, A. W., & Good, J. M. M. (1992). Mutualism in the human science: Towards the implementation of a theory. Journal for the Theory of Social Behaviour, 22, 105–128. Tomasello, M. (1999). The cultural ecology of young children’s interactions with objects and artifacts. In E. Winograd, R. Fivush, & W. Hirst (Eds.), Ecological approaches to cognition: Essays in honor of Ulric Neisser (pp. 153–170). Hove Psychology Press. Turvey, M. (1992). Affordances and prospective control: An outline of the ontology. Ecological Psychology, 4, 173–187. Van Dijk, L., & Kiverstein, J. (2020). Direct perception in context: Radical empiricist reflections on the medium. Synthese, 198(9), 8389–8411. Van Dijk, L., & Rietveld, E. (2017). Foregrounding sociomaterial practice in our understanding of affordances: The skilled intentionality framework. Frontiers in Psychology, 7, 1969. Walsh, D. M. (2015). Organisms, agency, and evolution. Cambridge University Press. Windsor, W.  L. (2004). An ecological approach to semiotics. Journal for the Theory of Social Behaviour, 34, 179–198.

16  Affordances, the Social Environment, and the Notion of Field: State of the Debate…

305

Withagen, R., de Poel, H. J., Araújo, D., & Pepping, G. (2012). Affordances can invite behavior: Reconsidering the relationship between affordances and agency. New Ideas in Psychology, 30, 250–258. Withagen, R., Araújo, D., & de Poel, H. J. (2017). Inviting affordances and agency. New Ideas in Psychology, 45, 11–18. Wokke, M.  E., Knot, S.  L., Fouad, A., & Ridderinkhof, K.  R. (2016). Conflict in the kitchen: Contextual modulation of responsiveness to affordances. Consciousness and Cognition, 40, 141–146. Yoon, E. Y., Humphreys, G. W., & Riddoch, M. J. (2010). The paired-object affordance effect. Journal of Experimental Psychology: Human Perception and Performance, 36(4), 812.

Index

A Active inference, 184, 185, 187, 196, 201, 202, 210 Adaptive, 4 Adaptivity, 3, 92, 222–224 Additivity, 222–224 Affordance, 10, 67, 85, 86, 94–97, 105, 211, 212, 278–302 Alva Noë, 88, 89 Anti-reductionism, 56 Anti-reductionist, 6, 56 Asymmetric determination, 255, 257–259 Attunement, 112, 190, 212, 213, 300 Autopoiesis, 67, 87, 93 B Bechtel, W., 7, 20, 23, 35, 55, 164, 219, 222, 260 Biopsychosocial model (BPS), 242, 243, 249–254, 263, 269 Bird flocking, 221, 230, 231, 233, 234 Bird flock, 10, 231–234 Birds, 114, 152, 229–235, 287 Body-self matrix, 246 Brandom, R., 6, 33, 109, 112 Broca’s area, 114, 115, 194 C Causal-constitution (CC), 156–171 Causal-constitution fallacy (CC-fallacy), 175–179 Chalmers, D., 3, 133, 156, 158, 159, 167, 170, 193–195, 197

Chronic pain, 10, 242–269 Chronic primary pain, 244–246 Chronic secondary pain, 244 Clark, A., 133, 156, 158, 159, 163, 170, 171 Cognitive bloat objection, 2 Component dominant, 223, 227, 233, 234 Component-dominant dynamics, 223 Consciousness, 10, 84, 97, 182–203, 210, 211, 213, 215, 296–302 Consciousness vehicle externalism (CVE), 182, 183, 186, 188, 189, 192, 193, 195, 197, 199, 200, 202, 203, 210 Coupling-constitution fallacy (CC fallacy), 2, 9, 166–169, 171, 175–177 Craver, C., 18, 22, 23, 28, 29, 33, 34, 55, 59, 138–140, 164, 167, 168, 179, 219, 221–223 Cultural practices, 104, 182, 190–192, 194, 195, 202, 209, 211–213, 282–285 D Davidson, D., 109, 113 Decomposition, 36, 37, 114, 220, 222, 227, 231, 233 Dewey, J., 1, 48, 64–66 Diachronic, 9, 90, 135–137, 140, 148, 149, 151, 152, 157, 159, 160, 162, 166, 167, 169–171, 176, 177, 191 Diachronic constitution, 9, 20, 130, 133, 135–137, 140, 141, 144, 148–152, 156–171, 175–179, 191, 194 Di Paolo, E., 6, 19, 55, 64, 71, 86, 87, 89, 92, 113, 211, 254, 256 Diversity, 45, 46, 51, 56, 57, 60, 61, 68, 114

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 M.-O. Casper, G. F. Artese (eds.), Situated Cognition Research, Studies in Brain and Mind 23, https://doi.org/10.1007/978-3-031-39744-8

307

308 Dreyfus, H., 1, 96, 284, 290–292 Dynamical, 5, 18, 52, 114, 157, 179, 186, 210, 221, 246, 294 Dynamical Entanglement and Unique Temporal Signature (DEUTS), 182–203, 211, 213, 214 Dynamic organization, 255, 258, 259 Dynamics, 3–5, 8, 35, 56, 88, 92, 97, 150, 156, 160–162, 164, 167, 170, 171, 186, 196, 197, 211, 212, 214, 215, 223, 246, 252–254, 258–261, 263–265, 267–269, 289, 293, 294 Dynamic system, 258 E Ecological, 4, 52, 55, 67, 70, 71, 83–85, 88, 92, 95–98, 166, 278, 280–282, 285, 287, 289, 292, 294, 296 Ecological psychology, 5, 8, 9, 52, 55, 67, 70, 83–92, 94–98, 278–280, 282, 283, 289, 293, 294, 296, 301 Embedded view, 165, 166 Emergence, 3, 85, 108, 110, 255, 256, 259, 260, 263, 265 Enaction, 83, 86–94, 98 Enactive approach, 86–89, 243, 254–269 Enactivism, 5–9, 21, 52, 55, 62, 70, 80, 83, 84, 86–91, 93, 94, 109, 110, 166, 182, 241–269, 296 Ethology, 6, 33 Etiology, 115, 121, 244 Experimental phenomenology, 6, 94, 292 Explaining, 23, 27, 30, 36, 47, 49, 51–53, 60, 62, 69, 77–80, 87, 95, 97, 105, 110–112, 118, 120–122, 214, 220, 222, 230, 232, 265, 278 Explanandum, 105, 110, 116, 117, 122 Explanans, 116 Explanantia, 105, 122 Explanation types, 8, 18–21, 23–35, 37–39, 45, 52 Explanations, 5–8, 18, 22–30, 33–39, 45–61, 63, 67, 69, 70, 77–80, 86, 91, 92, 94, 95, 97, 98, 105, 107, 110–113, 120, 122, 138, 152, 156, 157, 164, 166, 178, 214, 219, 221–235, 247, 249, 253, 254, 265, 266, 269 Explanatory asymmetry, 29, 30 Explanatory diversity, 46, 49, 51–72, 77–81 Explanatory integrations, 58 Explanatory pluralism, 7, 8, 10, 18, 19, 22, 25–28, 30, 31, 35, 38, 59, 69, 77, 265 Explanatory plurality, 8, 22, 77–79, 81

Index Explanatory power, 6, 57, 120 Explanatory reductionism, 265 Explanatory style, 8, 47–49, 53, 59–60, 67–72, 78–81, 220, 235 Explanatory unification, 7, 8, 18, 19, 22–25, 28, 30, 77 Extended conscious mind (ECM), 209–213, 215 Extended consciousness, 182 Extended mind, 89, 90, 147, 153, 156, 158, 159, 163, 166, 182, 214, 215 F Field, 24, 45, 57, 84, 156, 210, 246, 279 Field of promoted action, 283, 284 Field theory, 295 Flocks, 10, 229–233 Folk psychology, 110–123 Forward model, 211 4E cognition, 83, 89, 90, 104–111, 113–115, 117, 118, 121–123, 167, 171 Fragmentation, 2, 21, 46, 263 Fragmentation worry, 20–22, 50 Free-energy principle, 183, 184 Friston, K., 114–117, 161, 183–186, 192, 196–198, 211, 215 Front-loaded phenomenology, 6 G Gate control theory, 246 Gestalt psychology, 299 Gibson, J.J., 1, 55, 84–86, 95, 278–286, 294, 295, 299, 301 Global to local, 257, 260–263 Godfrey-Smith, P., 83, 91, 220 Gurwitsch, A., 280, 290, 291, 294–302 H Heidegger, M., 291 Heideggerian, 56 Holt, E., 84 Husserl, E., 1, 294, 297 Husserlian, E.H., 93 I Integration, 7, 18, 20, 22, 23, 25, 31, 32, 34, 35, 37–39, 45, 50, 54, 57, 60, 70, 110, 113, 171, 241–269 Integrationism, 57–60 Integration problem, 243, 252–254, 264

Index Interaction dominance, 10, 221, 226–229, 233–235 Interaction dominant, 56, 160, 162, 227, 228, 233, 234 Interlevel relation, 27, 164 Interoceptive signals, 201, 202 Intralevel relation, 164 J James, W., 10, 47, 48, 297 K Kirchhoff, M., 20, 131, 148, 156, 175, 209 Kiverstein, J., 3, 19, 70, 86, 131, 148, 156, 175, 182, 209, 279 Koffka, K., 285, 291, 294, 297–299 Köhler, W., 294, 298, 299 L Langfeld, H., 84 Lewin, K., 285, 291, 294 Liberal naturalism, 107 Linearity, 222–224, 228, 229, 263 Literalism, 121 Local to global, 257, 259, 261 M Margin, 10, 296, 297, 301 Mark of the cognitive, 2, 9, 130, 134, 142, 144 Markov blanket, 182, 184, 187, 188, 195–198, 202, 211, 214, 215 Maximum entropy model, 230, 231, 233 Mechanisms, 6, 23, 54, 104, 138, 165, 179, 192, 212, 219, 244, 280 Mechanist, 35, 220, 230, 233 Mechanistic, 7, 9, 22–25, 27–29, 33–37, 48, 55, 84, 92, 130, 133, 135, 137–144, 148, 157, 166, 168–170, 175, 178–179, 219–235, 245 Mereological fallacy, 119, 248 Metaphysic, 7, 156–171, 175–179, 218, 220, 235 Metaphysical, 9, 10, 59, 106, 107, 117, 122, 147–149, 157–160, 162, 169, 171, 176, 179, 218 Method, 2, 18, 45, 56, 77, 83, 111, 215, 222, 242, 279 Methodical, 3, 6, 7, 10, 18–22, 25, 29, 30, 39, 77 Methodical fragmentation, 20

309 Methodological, 1–10, 18, 19, 22, 39, 52, 53, 56, 62, 67, 68, 77, 80, 81, 97, 105, 106, 118, 122, 162, 218, 229, 243, 251, 253, 264, 266, 269 Methodological fragmentation, 2, 5 Methodology, 2, 4, 8–10, 18–21, 28–31, 37, 39, 46, 49, 50, 77, 78, 80, 81, 88–90, 94, 96, 105, 114, 115, 123, 186, 218, 219, 266, 269, 293 Mirror neurons, 289 Motley-crew argument, 2, 3, 5, 19–21, 31, 80 Multiple ontologies, 220 Mutual manipulability (MM), 138–142, 167–170, 178, 179 Mutual manipulability criterion, 138, 167, 201 N Naturalism, 107, 218, 255, 256 Naturalistic, 61, 110, 256, 257 Network models, 36, 252, 267 Neuro-centrism, 246–249, 255 Neurophenomenology, 6, 56, 93 Neuro-reductionist perspective, 247 New mechanists, 6, 18, 167 Non-additive interaction, 227 Non-additive system, 227 Nonlinear, 136, 157, 159, 225, 227, 232 Non linear behavior, 3 Nonlinear dynamical models, 49 Nonlinear dynamics, 50 Nonlinearity, 226, 229 Normativism, 9 Normativist approaches, 6 Normativity, 93, 103–123, 292 Norms, 59, 103–108, 110–113, 118–120, 122, 130, 284, 291–293 O Ontological, 9, 54, 56, 84, 110, 150, 160, 218–221, 228, 233–235, 243, 249, 253–261, 263, 267, 269, 293 Ontological pluralism, 10, 220, 221, 229, 233–235 Ontology, 94, 95, 110, 115, 117, 123, 160, 162, 177, 191, 218, 220, 234, 235, 253–255, 263, 279, 286, 293 P Pain, 10, 148, 158, 241 Participatory sense-making, 87 Phenomenal bloat, 10, 182, 200, 202, 203

310 Phenomenal consciousness, 182, 190, 191, 195, 197, 198, 201, 209–215 Phenomenology, 6–8, 20, 52, 84, 87, 90, 93, 118, 170, 182, 188, 189, 191, 193, 194, 197, 200, 201, 212, 290, 293 Philosophy of nature, 8, 19, 83, 91, 93–96, 98, 269 Philosophy of pain, 242, 243, 245, 246, 249–252, 254, 261, 263, 267, 269 Pluralism, 7, 25, 46, 53, 78, 122, 220, 265 Plurality, 7, 21, 26, 30, 52, 57, 59, 77, 121, 251, 253 Post-cognitivism, 34, 89 Pragmatism, 8, 20, 84, 85, 90 Prediction error, 184, 185, 187, 188, 190–192, 195, 198, 200, 202, 211–215 Prediction error minimization, 187, 201 Predictive processing (PP), 28, 29, 182–188, 190–192, 194–197, 201, 202, 209–212, 214, 215 Problem of choice, 8, 17–39, 45–50, 77–81 Problem of replicability, 2 Problem of triviality, 2 Proprioceptive signal, 201, 202 R Reduction, 7, 23, 27–29, 34, 38, 39, 46, 52, 54, 55, 57, 60, 92, 251, 260 Reductionism, 22, 27, 54–58, 69, 70, 218, 219, 247, 260, 263 Reductionists, 54, 56, 57, 69, 70, 220, 242, 247, 248, 251, 256, 269 Reed, E.S., 84–86, 96, 279, 283–285, 287 Reflexivity, 51–72, 77–81 Representationalism, 7, 20–22, 79, 88, 106, 109, 110 Research program, 1, 93, 94, 96, 98 Rosch, E., 55, 87 S Scaffolding, 287 Scaling-up problem, 2, 71 Self-assembly, 87, 92 Self-organization, 62, 87, 183 Self-organizing system, 35 Self-regulation, 87 Sellars, W., 109, 112 Sense-making, 67, 87, 92, 121 Sensorimotor contingencies, 67, 87, 88, 182, 185, 210, 211, 214 Sensorimotor enactivism, 5, 89, 185, 209–211, 215, 255 Singular ontology, 220, 221

Index Situated, 1–10, 46–49, 52, 55, 64–72, 78, 80, 83–98, 104–106, 109, 220, 235, 280, 282, 284, 289, 293, 296 Situated cognition, 1–5, 8, 18–20, 22, 24, 28–30, 34, 38, 39, 81, 83, 85, 89, 129, 218, 269, 296 Situatedness, 71, 89–91, 98 Situation, 3, 4, 18, 20, 25, 45, 47, 49, 50, 64–72, 78–81, 105, 115, 120, 130, 144, 182, 189, 226, 264, 267, 268, 279, 288, 289, 291–293, 295, 298–300 Social norms, 104, 106, 108, 111, 120, 279, 283–286, 292, 293 Socio-cultural norms, 105 Spatio-temporal scaling, 255, 259 Spectrum of integration, 19, 30–39, 46, 47 Subpersonal mechanisms, 120 Superdiffusion model, 232, 233 Superdiffusive, 232, 233 Surprisal, 183, 184, 187, 190, 200 Synchronic, 9, 20, 148, 149, 151, 157, 158, 162, 163, 165, 167, 169, 170, 176, 179, 264 Synchronic constitution, 149, 157, 160, 163, 166, 171 Synergetics, 5 T Temporal persistence, 244, 253, 265 Thematic field, 10, 296–301 Theme, 6, 10, 278, 296–301 Thompson, E., 3, 55, 87, 93, 94, 121, 211, 254–257, 259, 260 Titchener, E., 84 U Unification, 10, 22–26, 28–31, 34, 35, 46, 54 Unity, 22, 24, 25, 34, 38, 39, 53, 54, 56, 108 Utensil totality, 300 V Varela, F., 3, 5, 55, 62–64, 87, 92, 254, 256–260 Variables, 3, 19, 26, 29, 36, 97, 115, 116, 140, 168, 224, 226, 230, 296 W Wittgenstein, L., 1, 292 Wund, W., 84