Culture, Society, and Cognition: Collective Goals, Values, Action, and Knowledge 9783110211481, 9783110206074

Citation preview

Culture, Society, and Cognition

≥

Mouton Series in Pragmatics 3 Editor Istvan Kecskes

Editorial Board Reinhard Blutner Universiteit van Amsterdam The Netherlands N. J. Enfield Max-Planck-Institute for Psycholinguistics Nijmegen The Netherlands

Ferenc Kiefer Hungarian Academy of Sciences Budapest Hungary Lluı´s Payrato´ University of Barcelona Spain

Raymond W. Gibbs University of California Santa Cruz USA

Franc¸ois Recanati Institut Jean-Nicod Paris France

Laurence R. Horn Yale University USA

John Searle University of California Berkeley USA

Boaz Keysar University of Chicago USA

Mouton de Gruyter Berlin · New York

Deirdre Wilson University College London Great Britain

Culture, Society, and Cognition Collective Goals, Values, Action, and Knowledge by

David B. Kronenfeld

Mouton de Gruyter Berlin · New York

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter GmbH & Co. KG, Berlin.

앝 Printed on acid-free paper which falls within the guidelines 앪 of the ANSI to ensure permanence and durability.

Library of Congress Cataloging-in-Publication Data Kronenfeld, David B., 1941⫺ Culture, society, and cognition : collective goals, values, action, and knowledge / David B. Kronenfeld. p. cm. ⫺ (Mouton series in pragmatics ; 3) Includes bibliographical references and index. ISBN 978-3-11-020607-4 (cloth : alk. paper) 1. Culture. 2. Cognition and culture. 3. Distributed cognition. I. Title. HM621.K76 2008 306.41201⫺dc22 2008037495

Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.d-nb.de.

ISBN 978-3-11-020607-4 ISSN 1864-6409 쑔 Copyright 2008 by Walter de Gruyter GmbH & Co. KG, D-10785 Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Cover design: Martin Zech, Bremen. Printed in Germany.

To Judy, who has never stopped trying to teach me how to understand and appreciate the richness and subtlety of human culture

Acknowledgements

I owe thanks to a number of people without whom this work would not have achieved fruition. Giovanni Bennardo, Kathi Beratan, Michael Fischer, Theodor Gordon, Jerrold Kronenfeld, Judy Z. Kronenfeld, Robert L. Moore, Dwight Read, Halvard Vike, and Douglas White all provided useful feedback on one or another part of this work. Benjamin N. Colby, Roy G. D’Andrade, and Charles O. Frake were former teachers who, through both their teaching and their research, have greatly influenced the present project. Victor de Munck, Pace Lubinsky, and Kim Hedrick all were major participants in research reported on here. I want to acknowledge the research assistance provided by Intramural Research Grants from the Academic Senate of the University of California. I owe thanks also to anonymous readers of the manuscript for the Series – who provided much helpful guidance. Finally I want to thank Victor de Munck, F. K. Lehman (F. K. L. Chit Hlaing), and Eugene Anderson for their ongoing comments, criticism, suggestions, and encouragement. None of the above, or course, are to be held responsible for my use (or misuse) of their advice.

Contents

Chapter 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1. 2. 3. 3.1. 3.2. 3.3. 3.3.1. 3.3.2. 3.4. 3.4.1. 3.4.2. 3.4.3. 3.4.4. 4. 4.1 4.2. 4.3. 4.4. 4.5. 4.6.

Culture as a system of distributed cognition. . . . . . . . . . . . . . . . . . . . . . . 3 Contemporary anthropological concerns . . . . . . . . . . . . . . . . . . . . . . . . . 3 Framing notions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 The language-culture interface. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 The social nature of language and culture. . . . . . . . . . . . . . . . . . . . . . . . . 5 Pragmatics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Meaning: semantics and pragmatics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Pragmatics proper. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Academic context. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Putnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Wittgenstein . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Berger and Luckman . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Sperber. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Cognitive anthropology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Systems of cultural knowledge including cultural models, cultural conceptual systems, and others . . . . . . . . . . . . . . . . . . . . . . . . . 27 Historical frame . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Cultural models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Strauss and Quinn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Other context and background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 A note on individual structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

Chapter 2 Background and history . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 1.

Durkheimian collective representations: A distributed cognition perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1. Appendix: selected quotations from Durkheim’s The Rules of Sociological Method (1938) . . . . . . . . . . . . . . . . . . . . . . . 1.1.1. A system of mental entities – as in Saussure. . . . . . . . . . . . . . . . . . . . 1.1.2. System of generative patterns. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1.3. The crowd as a minimal simple example, a microcosm, of a society . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

40 45 45 45 45

x 1.1.4. 1.1.5. 1.1.6. 2. 2.1. 2.2. 2.3. 2.4. 2.5. 3. 4. 4.1. 4.2. 4.3. 4.4. 5. 5.1. 5.2. 6. 7. 8. 9.

Contents

A collective emotion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Universality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regarding the possibility of a “Social Psychology” . . . . . . . . . . . . . . Anthropological studies of routine decision making. . . . . . . . . . . . . . Christina and Hugh Gladwin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . James Young. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Carol Mukhopadhyay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Robert Randall . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Stuart Plattner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Comparison: Decision theory with simulation approaches. . . . . . . . General cognitive background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mode of intellectual functioning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cognitive schemas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Piagetian stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Gombrichian conventions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Piaget’s paradigm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Basic framework. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The stages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Piaget vs. other cognitive psychology . . . . . . . . . . . . . . . . . . . . . . . . . . . Gombrich’s approach: Art and Illusion. . . . . . . . . . . . . . . . . . . . . . . . . . Schank & Abelson 1977: Scripts, plans, goals, and understanding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Hutchins: Culture and Inference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

46 46 46 47 47 49 50 51 53 54 56 56 57 57 58 60 60 62 64 65 68 72

Chapter 3 Language to culture – building from Kronenfeld’s semantic theory . 74 1.

Culture and society. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76

Chapter 4 Culture as Distributed Cognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 1. 2. 3. 4. 5. 6. 7. 8. 9.

Representations – collective and individual . . . . . . . . . . . . . . . . . . . . . 83 Culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 The issue is not “internalization” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Culture and language. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Parallel distributed processing and decentralization . . . . . . . . . . . . . . 93 The role of culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Biological bases, mammalian sociability, and the social origins of human intelligence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 What culture gives us . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 What makes culture socially systematic . . . . . . . . . . . . . . . . . . . . . . . . 102

xi

Contents

Chapter 5 An agent-based approach to cultural (and linguistic) change: Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 1. 2. 3. 4.

“PEN”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 “COUSIN” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 “SIBLING”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 “JEW”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114

Chapter 6 Society (with a note on the self) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 1. 2. 3. 3.1. 3.2. 3.3.

Society: issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Society: Kronenfeld view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130 Identity and self, loyalty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 A general overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Self . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 Loyalty. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136

Chapter 7 Ethnicity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 1.

Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

Chapter 8 The social construction of ethnicity: Intuition, authenticity, authenticators – the Sami example . . . . . . . . . . 149 1. 2. 3. 4.

The Sami example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Other Sami thoughts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Problems and issues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Distributed cognition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159

Chapter 9 Some kinds of cultural knowledge – a non-exhaustive list . . . . . . . . . . . 162 1. 2. 2.1. 2.2. 2.3. 2.4. 2.5.

The current state of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Cultural models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 Empirical validation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Form . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 Structure and content. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Function. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Existence status. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169

xii Contents 2.6. 2.7. 2.8. 2.9. 2.10. 2.11. 3. 3.1. 3.2. 3.3. 3.4. 3.5. 3.6. 3.7. 3.8. 3.9.

How constructed . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 How used . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 Cultural models vs. systems of classification . . . . . . . . . . . . . . . . . . . 172 Cultural models, language, and communication . . . . . . . . . . . . . . . . . 173 More on the content of cultural models. . . . . . . . . . . . . . . . . . . . . . . . . 175 Cultural models and culture. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Collective knowledge . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Collective representations and social “others” . . . . . . . . . . . . . . . . . . 176 Learned without being taught. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Cognitive hierarchy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Social complexity. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Presupposition of sharing. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 Productivity, form, and application . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 Function and use. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 184 “Application” or “invocation” . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 188

Chapter 10 Illustrative examples. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 1. 1.1. 1.2. 2. 3.

Calculating and applying kinterms in Fanti . . . . . . . . . . . . . . . . . . . . . 189 The example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 Implications of the example. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 190 The stages of instantiation and English kin. . . . . . . . . . . . . . . . . . . . . . 194 Cultural models and romantic love. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 195

Chapter 11 Problems – messages vs. codes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 198 1. 2. 2.1. 2.2. 2.3. 3. 3.1. 3.2. 3.3. 3.4. 3.5.

Shared “codes” and individual “messages” . . . . . . . . . . . . . . . . . . . . . 198 Nature of cultural models, and their relationship to psychological schemas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 200 Cultural models are social . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 202 Cultural models are shared and distributed. . . . . . . . . . . . . . . . . . . . . . 203 Role, nature, and functioning of cultural values. . . . . . . . . . . . . . . . . 204 Aspects and attributes of cultural models . . . . . . . . . . . . . . . . . . . . . . . 205 Prototypes and prototypicality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 205 Alternative or conflicting cultural models . . . . . . . . . . . . . . . . . . . . . . 208 Degrees and aspects of specificity of cultural models . . . . . . . . . . . 209 Where we have good vs. weak vs. no cultural models . . . . . . . . . . . 210 Rules for breaking rules . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 210

xiii Contents 3.6. 3.7. 3.8. 3.9. 4. 5. 6.

Variability in cultural models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Instantiation issues in apparent variability or vagueness of cultural models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Variability in cultural models vs. in individuals’ knowledge of them. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 212 Conformity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 Collective representations reprise . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214 Looking back . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 Whither now . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217

Chapter 12 Other theoretical issues and relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 1. 2. 2.1. 2.2. 2.3. 3. 4. 5.

Cultural models – truth status. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Theoretical models and empirical tests . . . . . . . . . . . . . . . . . . . . . . . . . 221 Culture as a mixed structure: empirical implications . . . . . . . . . . . . 221 Behavioral models and analytic regularities. . . . . . . . . . . . . . . . . . . . . 222 Empirical tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Some further methodological thoughts . . . . . . . . . . . . . . . . . . . . . . . . . 224 What seem pressing empirical issues. . . . . . . . . . . . . . . . . . . . . . . . . . . 225 Neural networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225

Chapter 13 Illustrative examples: cultural models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 1. 2. 3. 3.1. 3.2. 3.3. 3.4. 3.5. 3.6. 3.7.

Cultural models: empirical examples . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Image studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 An extended example: Ranches, rangeland, and environmental cultural models . . . . . . . . . 237 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 237 Empirical testing of cultural models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 238 What is a ranch? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 239 Environmental categories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 241 A use category . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 244 Environmental health and conflict. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 Forms of cultural models and action . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247

Chapter 14 Gregory Bateson: Pulling it all together . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 1. 2.

Bateson’s Naven . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 250 Bateson’s system . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 251

xiv Contents 3. 4. 4.1. 4.2. 5. 5.1. 5.2.

Cultural models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 Some lessons from Bateson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 How it all goes together . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 257 Eidos and cultural models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 258 Some concluding reactions to Naven. . . . . . . . . . . . . . . . . . . . . . . . . . . 259 Sociology and culture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 Levels of analysis – a demurrer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 260

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 275

Chapter 1 Introduction

Americans know that uncles should be warm and caring, that corn can be a vegetable, that homes include bathrooms which contain (in a single room) sinks, toilets, and showers or bathtubs, that successful marriages are rooted in love, and that love is a mystery. These are a few examples of cultural knowledge – knowledge that is learned (vs. innate or biological) but not normally taught and that is often but not always summarized in (and referred to by) language. This knowledge can form complex systems, as in the mental map one has of a familiar city; the system of Micronesian navigation as described by Thomas Gladwin among others; the set of expectations (regarding duties and obligations, feelings, shared perceptions and beliefs) that characterize one’s universe of relatives, friends, and colleagues; the everyday dress codes of modern life; or the conventions regarding acceptable vs. unacceptable violence and/or rule breaking in various sports. A central concern of anthropology has been the description and analysis of such knowledge and of the behavior in which it is expressed. Issues concerning the make-up and structuring of cultural knowledge and its relationship to behavior have also been important to fields as diverse as history and literary studies on the one side and cognitive sciences on the other. More recently we anthropologists have become more concerned with how cultural knowledge is learned, how uniformly shared it is (or is not), how it changes and adapts, and how it is created. Related anthropological and sociological theory (going back at least to Durkheim and Saussure) has been concerned with how best to understand the nature of shared or collective knowledge and the relationship of individual knowledge (and behavior) to shared or collective forms. Insofar as cultural knowledge represents a special characteristic – whether of humans, of anthropoids or primates, or of social mammals – one would like to know what payoffs rendered worthwhile the selection of the combination of capabilities and predilections that bring it about. Building on a variety of strands in contemporary cognitive anthropology, and drawing on work in cognitive sciences, pragmatics, and cognitive psychology, this book offers an integrated conception of culturally shared cognition that systematically addresses these questions, and provides a

2 Chapter 1 – Introduction theoretically motivated, integrated, and empirically useful interpretation of basic anthropological concepts such as “culture”, “society”, and “collective representations”. The thrust of the book aims at some of the most basic and traditional anthropological questions, including: what is culture, how does it work, how is it shaped and transmitted, how does it adapt, and to what does it adapt. This discussion not only addresses the central concerns of anthropology but treats basic pragmatic issues and speaks importantly to contemporary cognitive sciences. The approach builds on my prior semantics work (see Plastic Glasses and Church Fathers, Kronenfeld 1996), and has been worked out and elaborated over the past 10 or so years. This book embraces insights from parallel distributed processing work in computer science, work on cultural knowledge systems including cultural conceptual systems (such as ethnobotanical and kinterm systems), and cultural models for action; it builds on theoretical insights and methods from linguistics, cognitive sciences, and cognitive psychology – and attends carefully to the distinction (including basic functional differences) between individual cognition and collective cognitive systems. This book extends my approach from the semantics of word meanings to the pragmatics of those meanings and to the pragmatic meaning of other cultural concepts as well – especially cultural concepts involving action and the relationship of action to knowledge, goals, values, emotions, interactive and communicative effects, context, credibility, social groupings and affiliations, and so forth. This treatment takes culture as a framework of understandings and knowledge and considers how individuals (as in individual “agency”) make use of this framework. My Plastic Glasses and Church Fathers (Kronenfeld 1996) provides an important starting point in its analytic treatment of language (Saussure’s langue) as a putative collective representation which is instantiated in the separate individual representations of it (in the form of Piagetian-type schemas) which each speaker forms on the basis of his or her interaction with other speakers through an adaptive process of assimilation and accommodation (of the sort outlined by Piaget). A combination of constraints on communicative effectiveness and shared natural predilections keeps these various individual representations close enough to one another to form, in effect, for natural language a single tightly structured collective system. Other shared cognitive systems may be less extensive, less tightly structured, and/or less widely shared.

Culture as a system of distributed cognition

3

1. Culture as a system of distributed cognition This book’s approach focuses on culture as a system of distributed cognition. Attempts to link collections of small computers together in order to simulate larger ones have shown that it is hard to program the division of labor among the little computers directly – in an effective way; the resulting system is too vulnerable to unforeseen problems (delays by one machine in producing its output, another machine breaking down, unforeseen calculating problems, etc.). An alternative, embodied in the idea of parallel distributed processing, is to give the various little machines all some substantial, though varying, information about the wider project and about the intended division of labor, and then let them negotiate actual task allocation and coordination among themselves. The downside of this approach is the “overhead” cost – the amount of each machine’s memory and calculating ability which is taken up by the “social” coordination tasks – but the upside is that such systems are flexible and responsive. One research task in computer science has been to figure out how to make such coordination more efficient but less costly. I want to suggest that society can be seen as such a network of individual computational entities (people) that, via the division of labor, are engaged in a myriad of collective enterprises. I want to suggest that “culture” represents the varyingly shared information (skills, knowledge, expectations, goals, values, etc.) – the overhead – what enables effective cooperation. “Cultural models”, in this context, represent a claim about how the relevant information is packaged, transmitted, shared, and brought to bear on problems.

2. Contemporary anthropological concerns This book is concerned with cultural codes and cultural forms, and thus it does not fall within the central focus of much of contemporary anthropology: power and interpretation. But cultural forms play an important foundational role in these areas. Power is expressed via forms and codes and understandings. It never exists simply in itself, but always depends on socio-cultural constructs such as identity, loyalty, and motivation, and on technological constructs such as communication media, weaponry, base of production, and so forth – which themselves depend on society and culture. The relationship between this approach and questions of power and politics is addressed in several places – see, for semantics in particular, the ex-

4 Chapter 1 – Introduction amples in Chapter 11 of Kronenfeld (1996) and see Kronenfeld and Vike (2002). Other areas of contemporary anthropological concern, such as agency, some aspects of interpretation, and the construction of social categories, are more directly dealt with. The goal is not to ignore, deny, or undermine the importance of these and related concerns, but to explain how they manage to work within the apparently chaotic world in which we seem to be living. The claim is that it is culture that gives us the tools, scaffolding, and shared code that we need in order to act in this world.

3.

Framing notions

This book is constructed within two framing concerns: the interface of culture with language and the social nature of both language and culture.

3.1. The Language-culture interface The language-culture interface shows up, on the one hand, in the use of language to index other knowledge – thus leading to language’s reliance on presuppositions of cultural knowledge – and, on the other hand, in the very different functions of language (to distinguish what is being talked of or asserted from other alternative assertions or topics) and cultural knowledge (to represent complexes of goals, values, action, knowledge, and emotion). The distinguishing function means that semantic knowledge has, in essence, a taxonomic structure – i.e., is a classification of the world into larger categories which are, in turn, in various ways subdivided into ever finer categories, in a context in which general knowledge is fairly widely dispersed, while more specific knowledge may, to a greater or lesser degree, be concentrated, and while there exist ways in conversation to use shared general categories to communicate about and query more specific categories that may not be known by everyone in a given conversation. The most general linguistic categories may not be actually realized in language – if/where they are too general to be of any conversational use. Cultural knowledge, on the other hand, seems more about what does what – about how the world works and about how we (and others) get done what we (they) do. Cultural knowledge – here seen, particularly, as cultural models – is not taxonomic but action oriented – and, as such, seems more like action schemas (as described in psychology) or systems analysis (from the computational world).

Framing notions

5

The interface exists in the indexing by language of cultural knowledge – indexing the models, but also the components of the models (the actors, the actions, the paraphernalia, the goals, the values, the feelings of the actors, the mechanical and social and conceptual connections, decision devices, and so forth). 3.2. The social nature of language and culture The social nature shows up in the idea of collective knowledge (or representations) and in the idea of language and culture as systems of distributed cognition. The social nature of culture (and language) in turn raises two important issues. The first concerns the ways in which collective knowledge (representations) are structures – as collective cognitive entities – and how these collective entities map onto, sometimes structure, and sometimes are structured by individual cognitive structures. The suggestion in this book is that the basic collective cognitive entities of action are “cultural models”1, but such models are construed somewhat differently here than by many others elsewhere in cognitive anthropology. In the present view, cultural models – as shared cognitive structures – have to pass many of the same constraints as do the structures of language (Saussure’s langue). These include (in most cases) being easily learned/learnable from the experience of ordinary, everyday interaction and thus being frequently enough evoked and used in everyday life to produce “in passing” learning; being abstractly enough formed to apply productively to a range and variety of actual situations; and being flexible enough in application and interpretation to allow users sufficient freedom of individual construction and action and to allow effective productive extension to new situations. Unlike langue (and grammar), however, cultural models are not empty containers into which specific meaning is poured, but instead are prototypic formulations of cultural content, to which layers of interpretative content are added through successive degrees of putative and then actual instantiation (application). “Interpretative content” 1

Other kinds of cultural knowledge systems, including “cultural conceptual systems” and “cultural modes of thought”, are described later – and then contrasted with “cultural models”. But my focus will be on cultural models and cultural patterns of action. Much of what I say will apply to all kinds of cultural knowledge systems while other parts will apply more to cultural models; I will try to be clear about the intended application of what I say.

6 Chapter 1 – Introduction here refers to shared and distributed knowledge concerning typical applications of a given model (where and why it is expected to be invoked), typical social associations (who is expected to use it, about whom or what), typical moral associations (is it seen as usually invoked by good or by bad people, is its invocation seen as laudatory or reprehensible, etc.), typical deictic associations (is it typically associated with oneself or with some kind of alter), and so forth. Thus, a network of interpretative knowledge exists around cultural models and informs their application by members of a culture to some particular specific situation – whether as a model for generating one’s own behavior or as a model for interpreting the behavior of others. Cultural models, by this view, then, are not in any sense simply mechanically or automatically followed or instantiated by members of a culture; they are, instead, utilized in the construction of an understanding of a situation that enables a person to interpret what is going on and respond both appropriately and effectively. This abductive use would seem to preclude cultural models from being deeply “internalized” into individuals’ constructions of self, but, rather, suggests that they are used more externally (as is the grammar of a language) to construct and interpret social situations; as is the case with language, any given individual may participate in a variety of cultural systems (with, perhaps varying degrees of expertise) and may “code switch” among them. The second issue raised by the social nature of culture concerns the functional role of (or benefits conferred by) such a system. Culture is a system that we humans spend much time and energy learning; it is part of the reason, presumably, for our drastically drawn-out period of childhood dependency. That is, it is a biologically expensive system, and expensive biological systems are typically seen as producing some benefit that outweighs their cost. The benefit of culture as conceived in this book is the enablement of the complex society – with its distributed knowledge, action, decision-making, and so forth – that we live in. Social knowledge, thus, provides the “overhead” which enables the distributed system to work. It provides the shared understanding of actions, plans, and motives which allow us each to do our bit in coordination with all the bits being done by those around us – whether the coordination involves cooperation, competition, or non-interference. It also provides the special knowledge that each of us needs to perform our own particular deeds, and a way of linking general knowledge to the specific end of bringing new individuals into whatever specific knowledge changing roles and circumstances may require of them. Culture – as a cognitive construct – thus consists of an active interplay between consensus (within society as a whole and within varying segments of society) and

Framing notions

7

variation (as required by the division of labor – both economic and intellectual – among varying segments of society as well as among individuals). The levels of sharing problem – the fact that both sharing and differentiation apply not only or simply to individuals, but equally to segments of society – requires that we have some cognitive devices or means for conceptualizing and “understanding” such segments (including their relationship to our individual selves). Our means of conceptualizing and dealing with social segments have to be flexible, though, since while – some kinds of segments can be seen as quite stable (even if there is turnover in the individuals that make them up) – others, at the other extreme, can be totally accidental, adventitious, and transitory. 3.3.

Pragmatics

This book is, in an important sense, about pragmatics, initially and focally the pragmatics of word meanings, but from there it extends to the pragmatics of other cultural conceptual constructs as well. 3.3.1. Meaning: semantics and pragmatics Semantic meaning, as of words and other concepts, is commonly split into some version of “sense” vs. “reference”. “Sense” refers to the relationship of conceptual entities (in language re words or other terms – “segregates” in the sense of Frake (1962) – the “signified” side of the Saussurean “sign”) to one another. “Reference” refers to the things (objects, concepts, actions, or whatever; whether real, imagined, or hypothetical) that words appear to label or “point to”. We need also to consider the difference between intensional and extensional definitions of the reference of terms or concepts (see, for example, Putnam 2003). “Intensional” definitions – ideally – specify the necessary and sufficient conditions for being truly a referent of the term in question, and thus address what might be seen as the essential make-up of a concept – what the concept really means. Sometimes this definition is specified via some if-and-only-if attribute, and sometimes by a decision rule of some sort. “Extensional” definitions most simply consist of a list of all the referents of the term in question. This can be done by specifying or pointing to instances. Extensional definitions often have the weaknesses of being, on the one hand, incomplete (especially when one has only a finite set of instances

8 Chapter 1 – Introduction for a concept that is in any sense infinite in its reference), while, on the other hand, potentially including (via irrelevant attributes of the instances) much irrelevant information. Use of the word “really” above is a way of noting that, even with intensional definitions, if we are not careful, and if we get too concrete or specific, we can contaminate our definitions with irrelevant material. This is the problem that caused the linguist Trubetzkoy, in his treatment of phonology, to distinguish the abstract definition of a phoneme (e.g., /p/ in English – as an unvoiced bilabial stop) from its “realization” in any particular speech act – where the pronunciation of /p/ acquires attributes (such as the presence or absence of aspiration or the location of the “release” of air, if any) which are irrelevant to its phonemic status. And for similar reasons the theory underlying this book makes use of Trubetzkoy’s conceptual distinction between an abstract entity and its realization in any concrete instance. Levinson (1983) worries much about the precise definition of “pragmatics” and of its relationship to semantics. He aims at a general definition that applies across the full range of relevant theories (even though he acknowledges his inability to find such – and details the issues which he sees as blocking such generality). But such a precise definition (and the precise classification that it implies) seems impossible in the absence of a specific theory – that is, all such precision is theory dependent (even if the underlying notions are much more robust); the reasoning behind this claim has been laid out in Kronenfeld (1985) – there, in relation to problems relating to the use of “numerical taxonomy” techniques in biological classification and “lexico-statistics”. Some individual writers, discussed by Levinson, such as Morris (see 1938), Carnap (see, e.g., 1938), Montague (1968), and Grice (see, e.g., 1957) have theories in mind that do lead to motivated definitions of “pragmatics”, but these definitions are tied to theories and focal concerns very different from mine, and so are not directly relevant to the present book. Others, such as Lyons (1977) or Searle et al. (1980) say in effect that there is no “there” there – leaving the meaning of “pragmatics” to be something analogous to what a number of people (including me in the past) have taken “culture’s” meaning as: “whatever it is that anthropologists happen to be studying”. My approach in this book grows directly out of my prior work on the semantics of words (Kronenfeld 1996), and both works are rooted in my extensive and rigorous empirical work on kinship (published in a long series of articles, and now collected in Kronenfeld 2008). This first means that the present discussion does not deal with (or make any assertions about) syntax. The semantic approach and theory developed out of a Saussurean approach

Framing notions

9

wherein semantics was the part of langue that concerned the linguistic structuring of meaning in terms of contrast and inclusion, and of the features that structured these relations. In that book the meshing of those linguistic relations with our non-linguistic knowledge of the world – i.e., the non-linguistic relations among linguistically labeled concepts that we rely upon in our use of language – was treated. The present volume examines that non-linguistic knowledge – focally in terms of what our use of words relies upon, but, on that model, also what cultural concepts in general rely upon – whether linguistically labelled or delimited, or not. It should still be noted, though, that, whatever pragmatics in general be understood to cover, the treatment in the present volume can at most only cover a segment of it, since there is much more to meaning than simply words or concepts. 3.3.2. Pragmatics proper Pragmatics, taken as a topic of study, is the major focus of this volume – even though the present theory (with its associated concerns and discussion) puts its own twist on the definition of “pragmatics”. Pragmatic meanings of concepts involve the same problems of sense vs. reference and intensional vs. extensional definitions that have been noted above in connection with semantic meanings. Philosophers (see Putnam 2003) often seem to write as if there exists a kind of absolute meaning for terms, apart from the incidentals of individuals’ understanding or knowledge – that is, as if there exists a true meaning for a term or category that people may only imperfectly know. This view has language existing in a kind of non-human never-never land. Such philosophic approaches to pragmatics seem to fall into deep problems because they ignore a) the social nature of language, and b) the incorrectness of language (and culture) users’ presupposition of the existence of an absolute form of language or culture. Language users presume such an absolute which they then try to approximate in their usage or behavior. But these users themselves only indirectly experience the presumed absolute forms – via (and only via) their experience of other approximators. All that actually exists of the systems of language and culture are these presuppositions and approximations; the absolutes themselves exist nowhere. One goal of the present volume is to explain how people’s presuppositions regarding such systems are formed and then used, and thus account for their role and functioning in systems of collective representations such as those that make up culture.

10 Chapter 1 – Introduction Thus, by the view of the present work, and by the theory presented in it, language is not only (if still importantly) a social construct, but is also necessarily an imperfect one – in the sense that the mechanisms which produce individual convergence (on shared communal meanings) are only satisficing rather than absolute, because all that learners experience is other convergers, and never the thing – language – itself. 3.4.

Academic context

3.4.1. Putnam Two related issues of some relevance to the present work arise out of Putnam’s Twin Earth thought experiment and his discussions of it. One issue concerns the relationship of terms’ meanings to people’s knowledge of those meanings, while the other concerns the division of intellectual labor entailed by different degrees of expertise. Putnam’s article is aimed at somewhat different (though related) issues from those of this book: his disagreement with views holding that a sameness of intension entails a sameness of extension and with the view that extension is tied to psychological states. On both issues, I should note, I strongly agree with him. The division of intellectual labor, with the variations in knowledge – including technical knowledge – that go into it, is a central theme of this book. In this context, Putnam’s notion that a chemist knows more about the technical definition of water than does a lay person – in the context of, say, some process that depends on the absence of contaminants – is totally built into the present approach, even if deriving from different sources than Putnam’s. A major problem with Putnam’s argument has to do with his notion of “same” or sameness – as in whether or not the liquid in one glass is “the same” as the liquid in another (see, e.g., Putnam 2008: 73). He treats this as a problem simply of the make-up of the stuff in the glasses and of the extent of our knowledge of that make-up. But by my lights sameness is defined relative to some purpose and by some cultural or linguistic community. That determination can be immediate and direct, as in “we understand an ‘automobile’ vs. a ‘truck’ to be defined by …” or it can be indirect relative to some standard, as in “water defined by its chemical structure”, or even progressive in the sense of allowing for further emendations in the future, as where our understanding of “chemical structure” has gradually progressed from simple chemical properties to a characterization of the atomic make-up of the material’s molecules, and then to the sub-atomic make-up of those atoms (as “water” vs. “heavy water”).

Framing notions

11

The progressive definition of water does allow for a constancy of meaning such that stuff we now regard as water may turn out in hindsight not to have been water. But this feature is not an intrinsic absolute property of water, but, instead, flows from how we have chosen to define “water”. The privileged position of the chemist’s definition is dependent on the context of what is at issue. Various speech communities may have different definitions based on different standards – how it tastes, how it is to cook with, how it works with crops, or whatever. By those standards, anything that fits the given standard is (good) “water”, and other attributes of the water in question are irrelevant; this is the semantic version of the phonological problem that Trubetzkoy addressed with his notion of “realization” – which is discussed later. That is, in the North Carolina of my youth, “pure mountain spring water” was not in fact water free of contaminants (however I and my friends might have thought of it), but instead water for which the highly valued characteristic taste depended on the presence of certain trace contaminants. Chemists can determine the basis of that flavor (as, indeed, they have – which is why I now know this [!]), but the defining standard remains the flavor favored by a particular local community – not some chemically based Platonic abstraction. In Putnam’s terms, there would seem to exist for “water”, in this example, some variation in both intension and extension. It should also be noted that other kinds of indirect standards exist too – as when members of a community cede definitional authority to some expert (what exactly is a “miter saw”) or panel of experts (as to a given doctoral committee for the determination “does this dissertation make an ‘original contribution to knowledge’?”). In many communities we freely cede to experts the definitive definitions of concepts such as a proper exorcism, the Christian “Mass”, “turbo-charged” (regarding an engine), and so forth. These are all terms or concepts that we freely use in our everyday lives as if we know their meaning, but about which when pushed we will more or less freely admit that we lack “technical expertise” and so could be “wrong” in our definitions. What in such a case might make our definitions wrong is not any absolute property of the concept or of its referents but is our social convention concerning relevant expertise and authority. For some concepts, for instance, “am I (really) sick”, the question of definitional authority may be contested: doctors are the presumed authorities but sometimes they don’t pay adequate attention to what we tell them or show them or they don’t understand our own bodies the way we do, or they suffer from a kind of educational or institutional myopia. For other concepts, such as “good person”, in my culture we admit no technical expertise; anyone’s opinion is as good

12 Chapter 1 – Introduction as anyone else’s – even if in some particular situation a community consensus might emerge and dominate. Other words of cultural importance show similar variation across communities – as in the “pen” example explored in Kronenfeld (1996) or the “Jew” example of the present book – and across contexts – as we can see in the classification of, say, “tomatoes”. Botanically tomatoes are, obviously, a fruit, but in the store they mostly get displayed as salad vegetables (never as fruit), and on the table they are eaten either (if raw) as a salad constituent or (if cooked) a vegetable. Other anomalous foods such as avocados and corn show similar inconsistencies. Thus Kronenfeld’s (1996) claim is that the definitions of terms – both intensions and extensions – depend on communal and usage context. Some of these definitions or definitional contexts imply a criterion that depends on special expertise – as when our definition of “water” relates to its chemical properties – and in these contexts ordinary folks like me will cede the ‘proper’ definition to the experts – as H20. But, even when – in a situation like that of “pure mountain spring water” – we cede a definition (say, that of the Fanti kinterm, egya) to members of that community, we also recognize that outside experts (here, anthropologists) may in some sense help explicate underlying aspects of that definition. The “indirect standards” above does represent what Putnam refers to as the “division of linguistic labor”(2003: 76–77). The immediate difference between us is that he ties it directly to the mundane division of labor – a consequence of differences in training and skill wherein only some of us actually know how to tell whether or not something really is made of gold (p. 75) while I tie it to a community ‘decision’ concerning relevant expertise or bases of interpretation. A richer example can perhaps be seen in the inter-related areas of biosystematics and ethnobotany. Our underlying system for classifying living things goes back well before Darwin to Linnaeus. At first flush this botanical example seems much like what Putnam describes for his water example: a folk system was regularized and systematized, and then, with Darwin, we gained an understanding of the process which drove and explained the classification, and thus a way for correctly resolving apparently ambiguous cases. Brent Berlin (1972, 1992) has contributed to this view by pointing out how deeply similar other folk botanical and zoological taxonomic systems around the world are to our own, while Eugene Hunn (1976, 1977, 1985, 1987) has explained how the processes Darwin described account for these

Framing notions

13

similarities. If we look more closely at the details we see some anomalies that might seem to represent “errors” analogous to Putnam’s characterization (2003: 72–73) of 18th century understandings of water: for example, some systems include palms with trees while others do not, and some include bats with either insects or birds, as opposed to the “proper” mammalian inclusion. But a view of these variants as “errors” makes the unwarranted assumption that the purpose of all these classifications is to understand the evolutionary history of the included organisms. Hunn’s mechanism depends only on relative gross similarities – not a technical criterion. In general people’s folk classifications seem driven by some systematized interaction of Hunn’s mechanism with what is relevant to their lives, interactions with the biotic world, and their observations of that world. Thus, the key fact for one group may not be the internal anatomy of bats, but either their bird-like life style or their more insect-like close-up appearance. And there exist, even for us in the modern, Western world, clear, if not (in the above sense) general, purposes (relating to the functional requirements of a free-ranging pelagic adaptation) relative to which sharks and (mammalian) dolphins go closely together. For more on this issue of classification – as well as related issues concerning our representations of classifications – see Kronenfeld (1985 – and also Kronenfeld and Thomas 1983 and Kronenfeld 2007). To continue. I emphatically agree with Putnam that “the extension of a term is not fixed by a concept that the individual speaker has in his head, and this is true […] because extension is, in general, determined socially – there is division of linguistic labor as much as of ‘real labor’”. But this comment is followed by “ – and because extension is, in part determined indexically.”; the statement is reasonable, but I do want to query his immediate application of it (pp. 80–81). He appeals to indexicality to explain why “natural-kind words” have “meanings” which “are [not] concepts, or, indeed, mental entities of any kind” and connects this use of indexicality to Kripke’s “rigid designators”. From my experience in ethnography and in linguistics, I am not convinced that “natural kinds” do constitute any sort of natural kind. That is, I see the underlying semantic and conceptual senses of, say, “house”, “tree”, and “mountain” as not being notably different from one another. In each case there is special information that we know about the category’s intension and about attributes that pertain to its extension. In a simplified, but I think not irrelevant, sense I could say: houses are sort of big, have been built by someone, cannot do anything themselves, and will tend to crumble (perhaps quickly) if not maintained (kept repaired); trees

14 Chapter 1 – Introduction are also sort of big, grow on their own and (though with varying degrees of assistance) are self maintained; mountains are very big, feel like eternal, and do tend to crumble (mostly slowly), though proper maintenance can slow the crumbling down. Stone is a kind of material not unlike concrete, while iron and steel are different metals. In sum, supposed “natural kinds” don’t seem that different from other kinds. 1a) We know that we have varying amounts of control over the creative and some of the destructive processes that apply to various entities. And we know that the material and structural constraints on houses are not unrelated to those that shape mountains and trees. We can turn stone into gravel and then into cement – or iron ore. And, in the other direction, we can turn coal into diamonds. Natural vs. human-made, then, cannot be any key issue for our understanding of concepts or meaning. 1b) And we do have notions of self-acting (as living things) vs. only acted upon. But still these attributes, while important to our thinking and understanding, are only attributes – and, in fact, ones we can play with, as in cartoons in which houses think and talk, or the nursery rhyme in which “the fork ran away with the spoon”, or in the ways we anthropomorphize various vehicles. 2) Similarly, it would seem that we cannot rely on categorization bases to distinguish “natural kinds”. Trees themselves – individual trees, that is – may well be natural entities, but the category (the “kind”, if you will) is a linguistic imposition. Because of the similarities in their lived worlds, most linguistic communities do have a category similar to “tree” in English. But the details – which specific entities or sub-categories of entities are in the category can vary significantly from one to another (e.g., “palms” are kinds of “trees” in English, but not in Spanish). As Saussure had noted, the “treeness” of trees is a conceptual imposition of language; that is, we, the members of a speech community, “decide” which attributes are relevant to a given categorization and which are not.2

2

Since language is a social tool that is constantly readapted to what we feel the need to talk about, the meanings of words referring to things in our experienced world are kept in line with important and salient aspects of that world – and the similarities (relating to general patterns of function and salience) among living things do possess the particular kind of hierarchical pattern produced by Darwinian evolution with the result that most speech communities code such relations in their language (see Hunn 1976, 1985).

Framing notions

15

Through an interesting series of thought experiments and analyses Putnam demonstrates and explores both the intercommunal variability of extension and the intra-communal variation embodied in the division of labor (in language) wherein different people know different parts of what together makes up the communal whole. And he recognizes that this variability implies that concepts are socially constructed. But he still sees scientific expertise as having a privileged position regarding the definition of “water” (in his thought experiment example). He then rescues social construction by deciding that water is a “natural-kind” and that natural kinds are not “concepts”, but rather are a kind of “indexical” (cf. “I”, “you”, “this”, “that”, etc.)3. That is, finally, it amounts to “Water is THIS”, where then science characterizes THIS – with increasing accuracy as knowledge grows – but where the THIS has an existence outside the language/culture/thought game, and thus is always only to be characterized and never to be actually defined. It is regarding “natural-kinds” that I specifically disagree with his analysis. And this difference offers one perspective from which to see how this book’s approach (re semantics in Kronenfeld [1996] and pragmatics in the present volume) differs from (and, by my lights, improves on) his approach. Language always implies categorizations, and word meanings always imply categorizations of the phenomena they refer to (or only putatively refer to). And, as Saussure noted in his discussion of “tree” and “arbre”, words in a language imply a human categorization – even of natural phenomena. Words are tools we use to enable us to talk about things that we most “need” to talk about in one or another context (while also enabling us to talk about the infinity of other things that we sometimes need or want to talk of). “Context” and “need” imply purposes – what relevant discourses are about. Some purposes entail the existence of special expertise, while many others do not. Reasons for such an entailment, within one or another speech community (or wider cultural community), can include scientific accuracy, judicial standards, a proper rendition of tradition, ritual efficacy, and so forth. For concepts regarding which we, as members of a speech community, recognize the relevance of expertise, we cede ultimate definitional decisions and criteria to the (communally recognized) relevant experts. Thus, we cede 3

Putnam speaks of “indexical, or token-reflexive – i.e., to have an extension which varies from context to context or token to token.” This seems to be a version of what linguists speak of as deixis or deictics.

16 Chapter 1 – Introduction the chemical definition of water to the chemists – as we cede the biological classification of species (in terms of evolutionary descent) to biologists (or taxonomists), and, outside science, we cede the delineation of the proper performance and interpretation of the Mass or of an exorcism to priests or church officials (i.e., experts). Exorcisms and Masses do, of course, get us back to communities, since there exist many communities in wider Western society who do not buy into the given interpretation of either activity (and thus the implied definitional standard). But evolutionary biological classification does also get us back to such non-accepting communities, since there exist communities that do not buy into Darwinian evolution and thus into the specifics of a classification based on that model. A related problem raised by Putnam’s discussion of his water example, one implicit in the whole idea of semantic extension, concerns the relationship between a concept (including attributes which govern its extension to referents) and the actual referents that members of the relevant speech community directly experience. This is an area with which linguists have long had to deal – in delineating the relationship between a given phoneme (say, the English /p/) and the form it takes in the various speech acts in which it appears. One early answer involved the idea of “allophones” or “allophonic variation” by which the aspirated /p/ of “pit” was distinguished from the unaspirated /p/ of “spit”. The definitive solution to this problem, however, came with Trubetzkoy’s distinguishing of the concrete attributes that define a phoneme (in the abstract) from its “realization” in specific speech acts – that is, from the concrete shape the phoneme takes in any particular speech context (through the addition of other contextual attributes of one sort or another). As members of a speech community all we experience of phonemes are their range of realizations; from this experience we each infer the shape of the phonemes of the given language – the dimensions on which the phonemes contrast with one another. Linguists trying to describe and analyze a language face the same task that native learners face – going from observed acts to the underlying patterns of concepts, distinctions, and relations on which the observed forms (there, linguistic) are based. A similar approach seems needed for any good understanding of the conceptual system of any language or culture and of how that system relates to the actual stuff that we see members of relevant linguistic and cultural communities actually doing – what Pike (1967) spoke of as “emic” analysis4. That is, in these wider tasks 4

Even if Pike’s particular attempt at the generalization seems not to have worked.

Framing notions

17

we need similarly to aim at generalizations that are at once simple, clean, and abstract and still very concrete and specific in their implications regarding the data of actual experience – that is, that can account for the apparently messy detail of relevant parts of ‘real life’ in a coherent and straightforward way. 3.4.2. Wittgenstein The essentially social nature of language was in a sense the problem that Wittgenstein – giving up on any absolute or essential standard for the extension of concepts – responded to with his idea of “family resemblances”. But his treatment, and the understanding implied by that treatment, misses the facts of a) the distinction of prototypic (or core or focal) vs. extended meanings, and b) the convergence (in the sense discussed just above) not just of usage but, particularly, of posited prototypes. My views on Wittgenstein’s approach are developed and explicated in Kronenfeld (1996: 21–22, and Chapter 10).

3.4.3. Berger and Luckman The kind of approach to cultural knowledge taken in the present volume comes out of a linguistic tradition rooted in the work of Saussure joined to insights from systems theory and cybernetics. Somewhat similar insights have come, as well, out of other traditions: e.g., in philosophy – as we just saw in Putnam’s notion of “the division of linguistic labor” – and in sociology – as in Berger and Luckman’s “social construction of reality”. I get to a similar place (regarding social cognition and the division of intellectual labor) from a different source tradition than the philosophic sources of Putnam and the salient sociologic context and tradition of Berger and Luckman – and thus independently of their work. It is interesting to note the convergence. The Berger and Luckman book contains impressive insights regarding the social construction of knowledge and why it matters; its major drawback, from the present perspective, is its insufficient attention to the systematic nature of culture (to the ways in which culture and society constitute each other, and to the complexity of the units and their relations that make

18 Chapter 1 – Introduction up each) and to the actual ways in which learning takes place (involving not just inference and induction, but also constructive leaps and trial generalizations). Given the date of their work, the limitations are not to be faulted, but are ones that we now hopefully can move beyond. I do want to suggest some ways in which the combination of my linguistic perspective with my computational and systems orientation forces attention to kinds of questions and detail that the other approaches do not – and thus that, hopefully, enables me to go beyond their contributions. The present approach is a linguistics-based system coming out of the need to understand the structure (concepts and axioms – and what derives from them) of unfamiliar conceptual systems – cf., in particular, Sapir (1925, 1933) and Trubetzkoy ([1939] 1969). Of particular importance to the present work is Trubetzkoy’s idea of “realization” – his delineation of, and handling of, the implicational relationship that exists between the concretely specific abstract concept of a given phoneme and the actual concrete situations within which it always is perceived. The systems perspective contributes a sense of how a variety of different sub-systems might fit together. Computer programming attunes one to a level of detail in specification and interactive effects that can be hard to get elsewhere – and opens one to new ways of imagining the structure and operation of systems, including the role of iteration and feedback.

3.4.4. Sperber In Sperber’s work we see a related approach to culture, including its status as a representation and its relationship to individual cognition, that comes more directly out of cultural anthropology. A brief discussion of his approach, in comparison with that of the present work, will help foreground the particular thrust and contributions of the theoretical approach proposed here. Sperber’s approach in many ways parallels mine in the present volume, but it also contrasts in significant ways. And, apart from contrasts within what we both look at, the focal concerns of this book are somewhat different in important ways from his. The concern here is with the pragmatics of how we actually communicate with language – and with the inseparably related question of how we understand other cultural concepts. This approach leads to a view of what culture is and of how it works. This view of culture is not deeply at odds with much of what Sperber says, but it goes beyond Sperber’s formulation in important ways. Relying upon a view of

Framing notions

19

the role of systematicity and generative productivity in our innate predilections combined with systematic social feedback, it takes the content and sharing of linguistic and cultural categories more seriously than Sperber appears to admit or be comfortable with. In the context of a critique of traditional functionalist analyses (p. 47 5), Sperber does nicely speak of the importance of feedback in keeping a cultural phenomenon tied to its effects and of the fact that one cannot use its future function as an explanation of the emergence of the phenomenon. His observations, and my agreement with them, are important because, in this volume, I will be speaking often of “function”, functions, and their importance, but not in the teleological, just-so-story sense of traditional functionalism. We agree that macro forms are produced by a cumulation of micro processes, and we agree on the importance of cognition and cognitive science for an understanding of culture. I share his view of the importance of “the relationship between a central concern of anthropology, the causal explanation of cultural facts and a central concern of psychology, the study of conceptual thought processes” (p. 56). We agree also on the importance of representations, and on the distinction between “individual representations” and other more public ones. However this volume’s contrast of “individual representations” with “collective representations” differs from his contrast of them with “public representations”. He regards “public representations” as material creations of individuals while in this volume “collective representations” are regarded as non-existent but posited and relied upon cognitive structures putatively belonging to collectivities. We both see the importance of what he calls “meta-representations”: individual representations of other individual representations and of public ones (p. 71). But I would add to his list of payoffs the crucial role such native representations play in enabling linguistic and cultural systems to be organized and function as systems – vs. simply as directly reinforced patterns.6 We both think that “anthropologists interpret behaviours – verbal behaviours in particular – by attributing beliefs, desires, and intentions to individual or collective actors”. My sense is that everyone, not just anthropologists,

5 6

Page references in this section are all to Sperber (1996). For a similar view in the context of human computer interactions, see Perry (2003: 206).

20 Chapter 1 – Introduction uses such attributions normally and routinely in their interpretations of behavior. We agree that interpretation of representations (by members of the relevant culture or by anthropologists) depends on shared cultural representations – as part of the relevant context even though not directly expressed (p. 39). That is, interpretations are context dependent. And this means that our anthropological or comparative understanding of something from some given culture has “to go beyond mere translation” (p. 39). He characterizes “society” and “culture” as interchangeable and equivalent terms (p.15); I sharply disagree. The present volume makes a case for seeing them as representing respectively the groupings of people that hold common cultural understandings and the cultural understandings that enable groups to be felt and acted in terms of. As such, the two are mutually constitutive, but still quite distinct. My view is more functional (in the sense spoken of above) than is his. Thus, in the theory developed in this volume, culture plays an important functional role re society, itself. And, unlike the thrust of Sperber’s discussion, I see much of cultural content – especially content that is expensive in time, resources, or psychological investment – as buying its holders something – something valuable enough to justify the cost.7 Sperber does not totally ignore such functional concerns but they do seem of only minor concern to him (for instance, see his discussion of the character of representations on pp. 33–34). The problems involved in backgrounding function come to the fore where, as on p. 34, Sperber talks about public representations as representations of the individual mental representations of their creators – and then speaks of those responding to the public representation as forming their own individual mental representations of it – where each link in this chain of representations is characterized by significant interpretation. What he leaves out is any consideration of, for example, the goal or purpose served by the initial mental representation (governing what went into it vs. what was left out, what attitude went into it, to what other representations it was 7

But, note, this is not a functional prerequisite argument, not equivalent to a demonstration in any particular case and/or relative to any particular posited payoff. It is more a suggestion about where to look for an answer to the “why” question – and where to explore the relationship of “what” and “how” to other considerations. And, yes, it is similar to modern Darwinian arguments, though involving very different mechanisms.

Framing notions

21

linked, and so forth), and leaves out, comparably, any consideration of the uses of the public representation (its intended use, the uses to which such constructs were commonly put, the way they were read in one or another context, and so forth) as well as of the functional considerations involving those responding to the public representation. This is important not just regarding fine details of content that may be added or lost over the process, but also because content itself may not be the issue for many of the individuals involved; phatic communication may be the focus for some, while for others it may be some attitude toward the content that matters, and sometimes what drives a communicative or constructive event may be some conscious aim while at other times the reasons may seem more subconscious. All of these considerations affect what gets transmitted, when and where it gets transmitted, how it gets transmitted, and how it gets reshaped (“interpreted”) in transmission – and thus they have to be relevant to any epidemiological approach. The need to understand these considerations and the mechanisms by which they work is one reason for thinking that we need a better understanding of what culture is and does before focusing on how it spreads. Sperber’s discussion of “explaining cultural representations” (pp. 41–55) is good. The first sense he discusses, the common one of “explain” as “to make [something] intelligible – that is, to interpret it” is not the sense he is concerned with – and is not mine. Like him, I want to “show how [a cultural representation] results from relatively general mechanisms at work in a given specific situation”, and I affirm the importance of theory. He goes through three common anthropological approaches to theory-based explanations (interpretive generalizations, structuralist explanations, and [traditional] functionalist explanations) and effectively disposes of each. He then offers his own choice, epidemiological models, as a fourth, actually empirical, alternative. His alternative seems potentially a reasonable one, even if there still remains much development needed to make it work. But I strongly disagree with his implication that epidemiological models are the only (or, maybe, the best) practical way to go. I think that with a better understanding of what we mean by culture we can frame questions more like anthropology’s traditional ethnological and communicative ones in a clear, meaningful, and soluble empirical manner. Offering such a view of culture is one aim for the present volume. Thus my goals for explanation are somewhat different from Sperber’s. First, as the preceding discussion of function suggests, I am interested in the explanation of why some particular cultural forms (e.g., kinship terminolo-

22 Chapter 1 – Introduction gies, attitudes toward sharing and ownership, elaboration of bio-taxonomies or parts of taxonomies, particular kinds of political structures, particular folklore and religious foci, and so forth) occur in specific ecological contexts or with other cultural forms. And I see these questions as empirically answerable – even if, in most cases, the explanation will involve a complex combination of not intrinsically (axiomatically) related factors (such as history, function, goals, areas of strong interest, and so forth). This goal is not foregrounded in the present work, but does form an important part of the intellectual matrix that it comes out of. Second, more immediately relevant is my aim at an explanation of how we actually use the limited and relatively stable resources of language to communicate the infinite and shifting variety of messages and shadings of meaning as effectively as we seem able to do. This focus involves linked concerns with how we construct the content of messages and how we read both the content and the communicative intent of messages constructed by others (syntax and phonology are also important here, but lie well outside our current purview.) A similar and linked problem concerns how we use non-linguistic behavior and material creations in communication – sometimes intentionally, sometimes more in passing, and sometimes even against our intentions. Sperber is one of the many who buy Wittgenstein’s dichotomization of semantic possibilities into Bertrand Russell’s all-and-only feature definitions of the referents of a term8 vs. Wittgenstein’s own “family resemblances” view in which the set of referents of a given term may have nothing in common, but may be interrelated only via various chains of locally particular similarities. “Dichotomization” because neither Wittgenstein nor most of those who follow him (including Sperber) allow for any third possibility. Kronenfeld (1996) offers a worked-out “extensionist” or “prototype-extension” third alternative. This is important because Sperber uses the Wittgenstein view in support of his notion that neither the cultural concepts of members of any given culture nor the cultural analytic concepts of anthropologists are systematic enough to support any notion of actually shared (vs. incidentally locally similar) concepts. The argument both in Kronenfeld 1996 and in the current volume is that the extensionist view characterizes 8

Referents of a term are (or, really, “should be”) defined by the intersection of all of a given set of defining features, and only those features. Russell did note – and was unhappy with – the fact that natural language semantics did not do very well by this measure.

Framing notions

23

not just ordinary semantic usage but also, more widely, the pragmatics of language use and cultural concepts as well, and it applies as much to the language and concepts of our anthropological culture as to those of the cultures we study. Sperber similarly buys into the related dichotomy between seeing culture and society as bounded, unitary things and seeing them as epiphenomenal clumps of individual similarity. A third view is to see them as systems (productive, generative conceptual systems in the case of culture). In this third view, “cultural systems” include a variety of hierarchies of subcultural systems where the systems of related sub-cultures share many propositions. All individuals belong to a variety and range of such sub-cultural systems, and it is what these individuals carry across from one to the other that keeps related ones close together. The difference from Sperber’s position is that the variability is not across individuals but across the subsystems (cultural systems of given groups) which individuals presume the existence of and to which individuals see themselves and others as belonging. In a related contrast, Sperber sees public representations (his label for non-individual, non-mental representations) as particular items while I take collective representations to be systems of conceptual items including the relations among those items. He sees communication as based on “inferential processes” as opposed to “codes”. The dichotomization is an artificial one. In the approach described in this volume, inference is important, but it is the code-like regularities that enable productivity (or generative capacity) that are inferred, rather than simply messages or thoughts. We are not born with knowledge of linguistic or cultural codes, but we are strongly predisposed to pull whatever regularities we can out of our experience, to join such regularities into systems, and then to use such systems – as if they were codes – to make sense of what others do and say and to shape our own speech and actions according to our goals, understandings, and values. In the theory offered in this volume, these understandings are our individual representations of the collective representations that make up culture. It is these systematic understandings that are the building blocks of culture and language. What we do and say is mediated by (and shaped by) these understandings – our understandings of presumptively socially based systems – rather than being any direct expression of our individual intentions or representations. That is, in plain language, we infer from our early experience how to construct messages or actions of different sorts for different purposes within different social contexts; we use this knowledge relative to social units; and it is the sharing of such understandings that constitutes social entities.

24 Chapter 1 – Introduction Sperber (p. 20) suggests that ethnological comparison requires a universal definition of the cultural entity (here, marriage) being considered. But really, all one needs is comparability regarding aspects that are relevant to the problem under consideration – as, for example, what is entailed by the local variant of marriage across a range of societies or what attributes of marriage (as it exists in one or another given society) are relevant to some particular theoretical or practical issue or claim. In this connection Sperber confuses a communicative vocabulary (where core-extension considerations govern communicative effectiveness) with the kind of technical vocabulary (all and only category definitions) that we need to test some given proposition or theory. Within cultural anthropology we do have a few of the technical type of situations and vocabularies (e.g., kinship terminological systems, implicational relations among pronominal systems, rules of succession), but very few. And, as suggested just above, the relevant technical definitions need not be of the “institution” itself (as “marriage”) but can apply only to particular aspects of it (as, “if marriagelike phenomena in a given society entail X then Y is predicted”). Such a piecemeal approach might in turn give us a more dynamic picture of what explains the range of marriage types that we see cross-culturally. That is, there may exist mutually contradictory pressures (or impulses) to which a given culture’s institutionalized form may be responding. Differing local conditions may produce different trade-offs among such pressures and thus different institutional outcomes. This might amount to something akin to the “balanced polymorphism” that one sees in comparable genetic situations (cf. the operation of the sickling gene in some malarial areas). The ethnological usefulness of a term such as “marriage” (cf. p. 22) does not depend on inter-cultural/inter-social comparisons alone (that is, on some universal positive definition of “marriage”) but also depends on comparison of the range of human relations with what one finds for other relevant species. (We do, of course, have to be careful not to confound the latter comparison with the kinds of potential artifacts that result from our being able to converse with and query other humans but not individuals from other species.) Such comparisons allow us to see much that human “marriage” (or “pair-bonding” or “sexual liaisons” or “concubinage” or …) is not – and thus help us to begin to home in on the attributes that do seem intrinsically human. Sperber says (p. 38) “anthropologists interpret behaviours – verbal behaviours in particular – by attributing beliefs, desires, and intentions to individual or collective actors, in a manner that makes these behaviours appear rational.”

Framing notions

25

Up until “in a manner…” all seems fine. But there are, in anthropology as in other social sciences, two views of rationality. In one view rationality has to do with an empirically and logically reasonable view of the world, which particular people at particular times may or may not have; that is, people can be, and maybe often are, irrational and/or can (often) act irrationally. In the other view, people are assumed always to be rational, and the empirical task for the social scientist is to figure out what they are being rational about and how they are doing it. While Sperber is closer to the latter view, he does not seem to recognize the distinction, and seems in his particular claims sometimes/somewhat to muddy the distinction. His ambiguity (or confusion?) about “rationality” is relevant to our discussion – even though the present book does not address “rationality” in any of its versions – because it muddies his attempt to characterize the empirical feedback that shapes some cultural content. The role of empirical and social feedback in shaping conceptual systems is a concern of this volume. In his discussion of the explanations offered by members of a culture for an outcome (here, for one kind of misfortune) and the practices they use ostensibly to control that outcome relative to those people’s “rational” 9 interpretation of the evidence of events (see pp. 50–51), he apparently assumes that their goal is the overt one – control of the outcome. But one can imagine other options, such as, for example, the possibility that the practices are not “really” aimed at controlling the event but at giving the players involved some sense that they are being active and involved in struggling – as opposed to simply passively accepting the inevitable outcome – or, as another alternative, that the practices are aimed at dealing with the social and emotional effects of the event rather than its control. Sperber does not seem even to consider such alternative possibilities – where such a consideration might 9

Sperber has a somewhat strange view of what “rational” means (see, for example, pp. 38, 55). His view is not outside the range of what one or another anthropologist has said before, and is not directly inconsistent with what comes out of the union of its ordinary language sense with contemporary anthropological, psychological, and economic knowledge and usage, but it has a focus that seems a little misleading. He sees “rational” intentions as following from underlying desires and beliefs (p. 38) and rationality of underlying beliefs themselves as a combination of mutual consistency (a stand-in for logical consistency) and their compatibility with the believers’ experience (a stand-in for empirical consistency) (p. 38). But he does not really insist on any stronger logical consistency with what either they, we, or scientists know of the empirical world, and he does not discuss how the purposes served by beliefs (their function, if you will) fit with those beliefs (i.e., any kind of psychological rationality).

26 Chapter 1 – Introduction greatly complicate his explanatory story line. As he says elsewhere, in other contexts, multiple independent factors seem likely all to be in play. There is a further problem with his view of empirical rationality. He says that strong rationality (involving a tight relationship to feedback from the external world) applies to (but only to) fully understood ideas (see pp. 71–73). But he points out, reasonably, that “half-understood” ideas are not susceptible to logical feedback – because no statement about an untrue proposition can be proven to be false! The problem is that there is maybe nothing in this world that we “fully” understand, and thus the distinction would seem to have no real force. Throughout his discussion (e.g., pp. 32, 61, 78) runs a kind of Peircean thread regarding the elements involved in representation – as “a three-place predicate: something represents something for someone.” In Kronenfeld and Decker (1976: 511–512) the weakness of Peirce’s own formulation (of the symbol) are explained – both absolutely (for what it leaves out) and relative to the Saussurean sign. The issue involves the theory dependent nature of verbal (or other symbolic) categories themselves. Take “tree”: even the members of such common and commonsense categories differ from one another in myriad ways, and so the categorical grouping entailed by a word represents an analytic imposition regarding which differences matter and which do not. This cognitive classifying aspect, both on the part of the person sending a message and on the part of the person receiving it, is not included in the Peircean approach. As Berlin (1992), Hunn (1976, 1977), and others have shown, such groupings, at least of biota, are not random or happenstantial, but represent the interaction of a strong apparent structuring in the world around us with pre-existing organizational and recognitory predilections of our minds probably evolved in part in response to such structuring. But still what we get in specific languages is in no way simply innate. And thus the idea of “an innate schema for zoological concepts” (p. 68) seems too strong a formulation. Similarly, given Hunn’s (1976, 1985) explanation of how our conceptual structures of living things are shaped by the similarity structure of the world we encounter, one has serious doubts about Sperber’s positing of innately based domain specific schemas. Thus it certainly seems true, as Sperber notes (e.g., on pp. 68–69) that the human mind is “susceptible” or “disposed” to culture, and thus that culture is a “catching” condition. But it is hard to say how that leads directly to an understanding of culture’s change over time or across space or to the interactive effects that follow from cultural encounters of one sort or another. Some lessons from linguistics are relevant here:

Cognitive anthropology

27

a) What is being changed or adapted or borrowed into is a system, not just a collection of independent items. b) All adaptation and change is from some given state to something else – as opposed to any straight selection of any absolutely optimal solution. c) Language is complex enough and old enough that making one thing better always seems to make something else worse. That is, there is a constant tension between “easier” cognitive organization and “easier” mapping on to the experienced world, so improving the one tends to come at the expense of the other. Culture is not linguistics. It is more a congeries of systems than a single tight system, even if some of its component systems may be tight. Much of culture has a direct functional relationship to the experienced world, as opposed to the more mediated relationships of language. But still, these lessons from linguistics seem worth at least thinking about regarding culture and cultural systems.

4.

Cognitive anthropology

4.1. Systems of cultural knowledge including cultural models, cultural conceptual systems, and others In this book the previously discussed pragmatic and communicative problems are approached from the perspective of cognitive anthropology, and so I would like to explain a little about that perspective. My earlier book (Kronenfeld 1996) focused on language – particularly on the semantics of word meanings and on how words mean. There semantic relations were clearly distinguished from other cognitive (or pragmatic) meanings. Within that distinguishing, the book described how semantic knowledge and usage related to, and relied upon, presumed pragmatic knowledge – knowledge represented by what was spoken of there as “world schemas”, but which in the present context seem best characterized as “cultural models”. However, that discussion took such schemas or models as givens, without offering any analysis or discussion of them or their properties. In that book’s introduction, the semantic work was related to prior work on cultural models only by way of showing a) its compatibility with such work and b) its clear non-equivalence to it – i.e., the non-isomorphism between the semantic categories delimited by language and the cognitive categories represented in other areas of thought and action. This present book addresses cultural

28 Chapter 1 – Introduction knowledge systems, including the other, pragmatic, side of that languageculture relationship, but with particular attention to cultural knowledge which is not directly coded linguistically. It will begin with classification, and with the contrast between semantic and functional classifications (of the sort noted so clearly by Keller and Keller 1996), and then proceed to more dynamic and active cognitive structures, such as those involved in routine decision making (as studied by Christina Gladwin 1975, 1976, 1980; C. Gladwin and Butler 1984; H. Gladwin and C. Gladwin 1971; James Young 1981; Carol Mukhopadhyay 1980, 1984; and Stuart Plattner 1984; among others) before moving on to the consideration of cultural models. 4.2. Historical frame Cognitive Anthropology can be seen as having passed through three stages, which are sometimes seen as alternatives, but which seem more accurately to be successively wider, cumulative developments. The stages are classification, decision models, and cultural models. The discussion of the classification phase of cognitive anthropology, is aimed not just at summarizing what work was done but at further drawing out and clarifying some issues that only seem to have appeared with the benefit of hindsight. These issues include, saliently, a clear distinction between semantic classifications and pragmatic (particularly cultural) ones. The first stage – classification – started out as the Ethnoscience concern with semantic classification, but broadened out (at first, perhaps, without those doing the work being aware of the difference) into non-linguistic cultural classifications. “Semantic classification” refers to classification aimed at determining the semantic relations of contrast and inclusion among terms (“segregates”, often lexemes) and at determining the relationship between semantic concepts (Saussure’s “signifieds” – the conceptual side of “signs”) and external (to language) referents of the terms (i.e., what the “signs” call to mind or point to in the world). The relationship of signifieds to referents serves the functional purpose of enabling speakers to distinguish those things in the world (whether objects, actions, attributes, or whatever) that they wish to talk about from others – even as the role of “signifiers” is to distinguish signs from one another.10 This semantic classification 10

It is in this sense, or from this perspective, that one can speak of language as essentially taxonomic.

Cognitive anthropology

29

includes the attributes by which contrasting entities are distinguished from one another.11 Cultural classifications (aka “pragmatic” classifications in a linguistic context) represent what we know about the functional (but also aesthetic, social, and other) relationships among things in the world – including (but not limited to) things that receive linguistic labels and are semantically distinguished. Classifications are entailed by our knowledge of what things co-occur, of what things are functional alternatives, of what things represent successive stages in some process, of what things constitute related parts of some larger process, of what things are ingredients of something else, and so forth. The idea of building classifications out of such relationships was intimated by Metzger and Williams’ analysis of Tzeltal Firewood (1966), was systematically laid out in Frake’s discussion of Notes and Queries (1964a, and see Frake 1964b), and was offered as a general outgrowth of narrower classifications by Colby, Fernandez, and Kronenfeld (1981). But these publications did not make the sort of distinction between semantic and cultural knowledge (or semantic vs. pragmatic classifications 12) that is now being proposed. The next significant development came with Keller and Keller’s (1996) discussion of classification in smithies (blacksmith work areas). They proposed such functional classifications (i.e., the grouping by smiths together of the tools that were used together in one or another specific operation – a specific kind of hammer with a specific type of tongs, etc., in a specific location) as an alternative to the traditional semantic anthropology taxonomic grouping (by type of tool – hammers subdivided by types, tongs subdivided by types, etc.) but did not consider how the two approaches went together, nor what different kinds of contrasting functions the two kinds of classification might serve. The stages are cumulative because each successive one makes significant use of the knowledge and constructs developed in the earlier ones. Thus decision models depend on both semantic and cultural classifications of the 11

12

Partinomies can perhaps be seen as semantic by virtue of constituting attributes of the whole entities they subdivide. But they also can be seen as part of our pragmatic knowledge of things. While distinguishing semantic knowledge from pragmatic knowledge is old news, I am not aware of much prior interest in distinguishing semantic from pragmatic classifications, and I am not aware of much of a tendency to see cultural classifiations as instances of pragmantic classification. My approach to the analytic understanding of semantic knowledge is offered by Kronenfeld (1996), where the case is made for analytically distinguishing semantic relations and structures from pragmatic ones. The present discussion continues from that base.

30 Chapter 1 – Introduction entities, attributes, and actions out of which and for which they are constructed. The kind of approach to decisions considered here was early noted by Quinn (1975) and has been most elaborated by C. Gladwin (1975, 1977, 1980). But Gladwin’s elaborations have joined the ethnographic representation considerations that are of concern in this book to psychological issues that, while relevant to us, are more specifically targeted at information processing issues than fits our needs, and to economic theory issues that are outside our concern. Important other elaborations have been by Young (1981) and Mukhopadhyay (1980, 1984); Mukhopadhyay’s (1980) dissertation is most relevant to our present perspective. Decision theory models aimed at discovering native criteria for routine decisions among alternatives for reasons defined within native classifications – and could be said to be “emic” 13 in that sense. But they were addressed to problems defined from the outside – whether that “outside” be ethnological comparison, ethnological theory, deductive theory (perhaps from an outside discipline, such as economics), or practical problems (delivering better health care or better agricultural advice). Such models have proved exceedingly powerful, often accounting for ca. 90% of decisions within some specified universe – where the decisions were non-trivial and where the sample exhibited substantial variability. The problem with them, though, is that they have not led to broader understandings of cultural systems – even the ones within which the given decisions were embedded. They don’t seem to have much helped us understand the process by which the given decision structure (vs. some other) came to be, the wider values and goals which it serves, and so forth. A partial exception to the preceding generalization has been Randall’s hierarchical use of default options (i.e, marking theory) to relate local fishing decisions to a wider framework; Mukhopadhyay’s use of “cultural precedents” is also important in this regard. 4.3. Cultural models My recent research has been focused on cultural models – the distributed cognitive models which I presume to embody cultural knowledge of how to behave (including how to make things, motivate others to act in some way, convey one’s feelings, and so forth) and how to interpret the behavior of others, and which I presume to define group membership and thus social 13

Pike’s “emic” (see Pike 1967), not Marvin Harris’!

Cognitive anthropology

31

entities. Of particular interest and concern has been the empirical and experimental problem of trying to figure out the specific nature (shape, structure, boundaries, content) of cultural models and then trying to find empirical means of actually pinning these down in detail. Important in this work has been a consideration of the implications for cultural models of the fact that they are socially constructed. But in trying to think out how to attack these issues I found that some other analytic questions kept getting in the way. In my semantic work I have been very concerned with function, and with how function shapes structure (though not the tautological Functionalism of classic Radcliffe-Brownian just-so stories). And so it occurred to me that I really should try to approach the problem of what cultural models are from that perspective. Part of the claim about cultural models is that they contrast in important functionally based ways with what are sometimes called “cultural conceptual systems” (the products of ethnographic studies of specific meaning domains such as kinship, ethnobotanical, and disease terminologies) – and so some but not all of this book’s observations about cultural models are by way of contrasting the two. The effects of functional constraints posed by social construction seem dependent, to some degree at least, on this prior understanding of what cultural models might be – and that is why I am discussing issues regarding the social construction and distribution of cultural knowledge first.

4.4. Strauss and Quinn The major developers of the idea of cultural models have been Naomi Quinn and Claudia Strauss, along with Roy D’Andrade. It was their work which brought the idea to my attention, and led me to think about its potentialities. At the same time, in considering their work, in thinking about examples developed by me and by students and colleagues working with me, and in relating the cultural models idea to my prior work in kinship and semantics, I came to realize that my view differs in some significant ways from their views. I would like now briefly to summarize some of these differences; more extended treatments of the issues involved will come in the body of the text. But I do want to emphasize my appreciation of the originality and importance of their work and my dependence on it. For the purpose of underlining differences I want to focus on Strauss and Quinn’s (1997: Chapter 2) discussion of “Anthropological Resistance”. I pick this focus because in their discussion they make their case for the internalization of culture and thus of cultural models. In the course of that

32 Chapter 1 – Introduction discussion they contrast their understandings and emphases with those of some leading theoreticians by way of underlining their own positions. The contrasts are, they say, not always either-or ones, but often matters of emphasis or attention. I disagree strongly with their claim about internalization, and I think their discussion misconstrues alternative possibilities – though not necessarily the specific positions of the theoreticians they discuss. I should note that I am distinguishing “internalization” from “learning”, as I assume do Strauss and Quinn. I presume that internalization implies some kind of planting deep within the psyche (that is, where it becomes some basic part of the self), while learning can be more superficial – even if – when rehearsed, automated, and chunked – still relegated to the subconscious (as, e.g., are the “rules” of language). By “superficial” I mean that the same person can, in different cultural contexts, take on the values, presuppositions, and way of viewing the world that pertain to the different cultural communities involved. This ability to shift is analogous to what linguists describe as “code switching” among languages, and does presuppose adequate familiarity with and command of the alternative systems to be credible in each – even if credibility may be something of a shifting standard, depending on context. In their discussion of Geertzian Interpretivism, with its notion that meanings are public, they properly note that all meaning and thought has to be located somewhere, and that since there exists no collective “somewhere”, the location has to be in individual minds (pp. 13–23). The claim both in this book and in Kronenfeld (1996 – there re semantics), following Saussure, is that as members of cultural and language communities, we presuppose the existence of some collective knowledge (culture or language [langue]) and see our own knowledge as existing relative to that posited communal knowledge. In Durkheimian terms, “collective representations”14 exist nowhere apart from our individual representations of them, but they seem to us real, and they do in fact have an emergent, systemic reality – from a combination of our distinguishing our own individual representation of stuff in the world from our understandings of these presumed collective representations, 14

By “representation” here I refer only to some still unspecified kind of cognitive structure. Later in the book I will provide some further specification of what I mean, but there will still remain individual psychological aspects of form and mode of functioning that I will leave unspecified and, even, undiscussed. Note that while my usage clearly echoes Durkheim and derives from him, I am not in any sense claiming to explicate or use Durkheim’s own particular sense of the term.

Cognitive anthropology

33

our predication of much of our behavior on these presumed collective representations, and the fact that our both shared and interactive experience causes our various individual representations of these posited collective representations to converge (whether as similarities or complementarities). Thus one can reconcile the Geertzian assertion with Strauss and Quinn’s criticism. Sharing of knowledge (see Strauss and Quinn pp. 15–16) seems to depend on function and experience. That is, to the degree that we each (as members of some group) have similar experiences, including similar interactive experiences, we will form similar representations. And, likewise, to the degree that our actual interactions depend on (or rely upon) some sort of sharing, we will come to share. But interactive sharing can have two aspects: in similar situations we may produce similar representations, while in complementary situations we may come up with complementary representations – producing the sort of complementarity that the division of labor depends upon. My argument against any deep internalization in part refers to the constant code switching that we all do – among the systems that characterize different subgroups of the wider speech and cultural communities that we belong to even when not among the major linguistic and cultural groups. We each, as individuals, have in our own minds representations or models of the cultural models that inhere in the various communities to which we belong and in which we participate. For this reason (cf. Strauss and Quinn, p. 23), I have been insistent about separating individual schemas from their collective analogs (“cultural models” in this book’s usage). My separation of individual representations (individual schemas, knowledge, understandings, or whatever) from individual representations of collective representations requires a delineation of the differences between individual and collective representations of the sort that Strauss and Quinn have not felt the need of. Such differences stem from differences in function, and thus in the shape and structure required by those functions and of the kind of relationship that each kind of representation (i.e., cognitive structure) has to action; such differences stem also from the different kinds of experience and feedback relations that shape each kind of representation. Relative to their discussion on p. 24 of culture as invented and inclusive of internal variation, I do see culture as a congeries of collectively invented systems. As in my Saussurean view of language (langue) – I see the systems of culture as received passively by individuals, but then actively applied and adapted by those individuals to their ends (whether selfish or selfless).

34 Chapter 1 – Introduction In this view, a repeated pattern of similar individual creations will be learned by ensuing learners as part of the received pattern – and thus culture changes in a manner that tracks shared interests and foci. I thus share their view that culture does not consist of any “superorganic, cohesive, bounded, timeless entity” – and I have complained in the past about older ethnographic approaches that seemed to present a kind of extra-individual, totalizing, uniform “dead hand” of culture. Culture is constructed and continually adapted, but it is always a collective construct – which, then, is always individually applied. Judgments of “authenticity” and cultural “credibility” are real and important, but always depend on communities and contexts. Regarding their discussion of the “constructed” “self” on pp. 26–34 (concerning the degree to which a “self” might exist, might be bounded and might [in what sense] be “constructed”), I have no real contribution to offer. That is, I will have some observations regarding the self in Chapter 6, but no real theory or analysis. But I do note that my view of culture does require that there be something there (in the individual) that picks out which cultural tools to use, how, when, and where. I also want to note that, from this perspective, cultural views of the self (see p. 29) – whether collective, autonomous, or whatever – are simply cultural constructions, and not themselves the thing – the self – itself. We are all each individuals – whether our individual aim (alone or together) is to aggrandize the self or submerge it. I do further note, though, that the present social emphasis does suggest that much of what we commonly refer to as personality is really an interactive dyadic construction between self and some “other” (based on the roles played, context, history, and aims of a given relationship) rather than any simple manifestation of the “self” itself. In this area I find Strauss and Quinn’s arguments on p. 27 quite reasonable and relevant. Strauss and Quinn address critiques of their position which might come from an historical materialist perspective (p. 36). I do not want to get into these issues beyond suggesting that they are not immediately relevant to the present work. That is, since by this volume’s view, culture offers an array of tools which individuals can choose to use or adapt – as they choose and as their societies will let them; culture itself is neither hegemonic nor coercive. Individual human agency uses culture, and political and economic constraints work through individuals’ use of culture and of the social groups individuals belong to or create.

Cognitive anthropology

35

4.5. Other context and background Cultural models involve not just decisions, but goals (with subgoals, subsubgoals, etc.), values, affect, and so forth. Describing cultural models involves an attempt to construct not just emic decision processes (as done in routine decision making studies such as those of C. Gladwin) but fuller emic action plans that involve emic ends and emic constructs. Cultural models can look like stereotypes or like “just so” stories. But, unlike some of our stereotypes, cultural models are not simply pictures (or caricatures), but are dynamic models which involve someone doing something. Stereotypes are basic to our dealing with the world, and not always bad – cf. Roger Brown (1965: Chapter 4). They are abstractions that reflect common elements of experience – but the experience can be of various sorts, including not just our direct experience with the categories of stuff (people, actions, …) involved, but also our indirect experience via what we are (repeatedly) told about the categories or, even, via what we infer from the way in which we hear about the stuff. Many of our destructive stereotypes derive particularly from indirect experience. Part of the evidence for the existence of cultural models, for their prototypic and abstract form, and for the specific content of specific cultural models comes from the errors that people make in various sorts of memory or attribution tasks. That is, memory drift is in the direction of culturally normal, or standard forms (cf. Cancian 1965; Kronenfeld [experiments reported in Bernard et al 1984]; Freeman, Romney, and Freeman 1987). Experiments with folktales carried out by Dr. Leann Martin (then a graduate student in anthropology at UCR) showed that the standardization had to be based on productive (i.e., rule-like) understandings rather than on any simple memorization of any kind of list of normally experienced events.15 15

Martin told Bushman folktales to students in her San Bernardino Valley College anthropology classes, and had them tell the tales back to her. As I remember it (I do not have a copy of the paper she wrote describing this experiment), in her research design she compared the accuracy of the tell-backs for groups of 1, 2, 4, and 8 students. The major finding was that the larger the number of students in the group, the more accurate was the tell-back – indicating that students could (at least in that relatively short-term memory context) distinguish between what they actually remembered hearing and what they had automatically filled in; without such a distinction, accurate and inaccurate recalls would have had an equal probability of getting into the collective re-tell, and so the accuracy of the group tell-backs would have, on average, been the same as the individual ones. My

36 Chapter 1 – Introduction A comparable conclusion comes from examination of folk etymologies and word form trajectories (Rundblad and Kronenfeld 2000, 2003; Kronenfeld and Rundblad 2003). Folk etymologies – that is, false etymologies shared by members of a community – represent constructions that are, at the minimum, meaningfully interpretable on the basis of routine cultural knowledge, and often constructed on the basis of such cultural knowledge; such knowledge is, again, too complex to have been simply memorized, and so has to have been reasoned from some productive structure or model. An example is Edwin Ardener’s account of a folk etymologization in a Welsh community of “asphalt” as “ash felt” (vs. the true, but constructed, derivation from Greek “a sphaltos” [“not hard”]). The folk etymology, since it is NOT true, cannot represent any simple rendering of any nonunderstood or meaningless fact (as an account of the true etymology might be), but has to represent a communal understanding, and the communal understanding has to involve an understanding of (experience with?) the process of making felt cloth and an appreciation of the abstract parallel between that felting process and the process by which an asphalt or black-top road is made. By “word form trajectories” I refer to a gradual reshaping of the morphological shape of a word or set of words in order to make that shape accord with some cultural model – as in the process by which Latin “masculus” and “femina”, through Old French into Middle English “male/ masle” and “femelle”, became our modern “male” and “female” – suggestive of an unmarked male form and a marked, derivative female form, when the true history shows no such relationship (Rundblad and Kronenfeld 2000; also there see a comparable situation with “man” and “woman”). The preceding schematic cultural models amount to a cultural and behavioral analog of the grammar of language – both in terms of grammar as the native’s sense of the regularities which guide production and in terms of grammar as the formal device by which those regularities are modeled by linguists or anthropologists or sociologists. It is modern computer science, in its cognitive sciences guise, that has finally given us an effective means for describing proposed grammatical-like regularities of socio-cultural behavior and for evaluating the success of such modelings. Among other examples, Edwin Hutchins’ study of Trobriand litigation (Culture and Inference, 1980) and Shank and Abelson’s study of restaurant behavior (Scripts, present point, though, is slightly different, though related – and is one that Martin made in her paper; it is that the errors were all in the direction of making the stories (their structure and events) more like comparable American (or Western European) stories.

Cognitive anthropology

37

Plans, Goals, and Understanding, 1977) stand out. With such models, here too, comes the temptation to confound the model with the reality it describes – as Roger Schank showed too clearly for my comfort in the account of his daughter’s behavior which he appended to his and Abelson’s book (1977: Chapter 9, and see Kronenfeld 1996: 28). To understand how these grammatical regularities are learned one needs to turn to the cognitive psychology component of cognitive sciences. Since these patterns, like those of langue, are not explicitly taught or even directly dealt with in their analyzed form, individuals mostly (though here, one would guess, less so than for grammatical regularities of language) have to infer them from their experience and from their understanding of the experience of others. These inferences are made not as full-blown logical leaps, but rather through a circular feedback process of the sort described by Piaget (see Flavell 1963: 47–68 and Kronenfeld 1996: 237–242) – via his linked adaptive notions of “assimilation” and “accommodation” – for the learning of other representations. The process is one in which the learner successively leaps to (over) generalizations, tries them out on the relevant piece of the world, focuses on where they don’t work, and uses that information to leap to a new generalization. It is a kind of homing-in process that converges (asymptotically) on an ideally successful model – converging (in a satisficing manner) as well as time permits and relevant behavioral needs require. This learning model, incidentally, allows for the potentiality that different aspects of our collective life may exhibit very different degrees of actual uniformity. In Los Angeles I suspect we agree more on the appearance (and behavior) of dogs than of marmots, let alone unicorns!

4.6. A note on individual structures One reader of an early version of this work offered the following: “If you really think that distributed cognition (and of course linguistic division of labor) play central roles in the creation and reproduction of culture … , then one more thing follows: there has to be a way for people to be exchanging or at least partially exchanging this un-shared information. And, unfortunately, at this point the book does not seem to have a particular discussion (as far as I can tell) of interaction, cross-person signaling, and all the other pragmatic ways of ‘dealing with’ the distributed nature of knowledge.” Part of learning a cultural system always necessarily seems to involve ways of linking one’s individual knowledge and aims to wider socially shared systems.

38 Chapter 1 – Introduction Cultural models, as conceived in this book, are more or less routinized (vs. ad hoc “on the fly” creations), and do embody reciprocal roles and activities. Cultural models are continually being adapted and extended by their users, and some of these changes happen frequently enough to be experienced by new learners as part of the givens of the received system. Other kinds of shared cognitive structures also are adapted by their communities of users in ways that are consistent with their forms and functions. Ethnobotanical systems in Berlin’s presentation always involve some potentiality for growth and re-organization – which begins as individual specialized knowledge or observations. We have, amongst our shared systems, various systems for communicating attitudes about our actions or our communications. These can be verbal, such as hedges (as, “I sort of mean …”) or self-effacement locutions (“I don’t really know, but it seems to me that …”), or serious, no-nonsense ways of talking (e.g., slowly and carefully enunciating each word in an instruction), or our choice of register and diction (e.g., breezy, slangy, formal, etc.). Non-verbal, body-language equivalents also exist – as in (in my world) accompanying a message or action with elaborately raised eyebrows (signaling a questioning distance) or a shrug (signalling a “who knows?” lack of certainty) or a focused glare (signaling that “I really mean this!”). These can be the sociolinguistic conventions that shape our use of language (see Duranti 1997, 2003). And, among our non-language communicative systems, we have the full range of culturally tuned proxemic behavior – as described first by Edward Hall (see 1966 and 1959) and subsequently by others. Beyond commentary on messages, we have non-verbal messages – e.g., our array of insulting finger gestures, and the variants that we each individually can ring on them; the signal systems used in different sports and games, and all of the presentation of self devices described by Goffman (1959, 1963, 1974, 1981) and others. And so forth. I see all of the preceding as part of culture, and see the pragmatics of the use of these systems as not unlike the pragmatics of the mostly more language oriented examples that I am discussing in this book. I do not much discuss the specifics of these systems simply because they are areas of culture that I have not myself much studied. Non-sharing (“un-shared information”) has at least two aspects: individual idiosyncrasy and complementarity. By “idiosyncrasy” I mean stuff that we as individuals do that is unique to each individual. This kind of stuff comes to be understood by those we frequently interact with, and those people learn

Cognitive anthropology

39

how to read, interpret, and respond to such actions or messages – which leads to what one might call a “micro-culture” involving the focal actor and those interacted with in the given manner. If the relevant interactions are shared or repeated and public, the micro-culture may involve a group; if they are unique to a given dyad, then the emergent culture is limited to that dyad. “Complementarity” refers to the differential knowledge that characterizes and enables the intellectual division of labor frequently spoken of in this book. We can learn such knowledge in part from working with others who share our role, but we also learn this material from our interactions with those playing complementary roles; if we know no one who shares our role, then the learning comes totally out of the complementarity. This complementary learning seems part of the same process as shared learning – the kind of convergence on an abstract understanding that can be seen as the social version of Piagetian adaptation (see Kronenfeld 1996: 237–242). It is what Bateson (cf. 1958), among others, has noted – that a lot of role learning is interactive and reciprocal. We learn how to fill a role by learning how to interact with reciprocal roles – how to behave as a “son” by adapting to the behavior of “father”. My sense is that to some degree the learning of the non-shared component of social/distributed knowledge is a function of a combination of shared goals (both immediate and more distant), interaction and feedback from interaction re effectiveness, triangulation from “similar” situations, and individual creativity. That is, there is some joining of sources, constraints, and feedback – and then it is a question of what works.

Chapter 2 Background and history

This chapter describes some of the sources that the positions described in this book have evolved out of – a sort of literature review, but one which is very selective and very focused. The review begins with a brief description of the Durkheimian basis of the present theory. It then move on to more recent anthropological background before finally discussing some psychological sources. 1. Durkheimian collective representations: A distributed cognition perspective The focus of this discussion is on a view of society rather than any exegesis of Durkheim or his views on religion. The view is one, however, which has been very much informed by my reading of Durkheim, and one that I claim to be consistent with his conception of society. As context, I should explain that my approach to Durkheim’s conception of society has evolved out of a previous exploration of Saussure’s conception of langue and parole (Kronenfeld and Decker 1979, and see Kronenfeld 1996: Ch. 1). In that work I noted the parallels and connections between Durkheim’s work and Saussure’s. Many of the social and cognitive ideas that are only being sketched here are more fully explicated in that discussion of Saussure and of the semantic theory that has been developed from his base. The Saussurean parallel (including Saussure’s insights) continues to be useful, and it was through my reading of Durkheim that I came to realize the wide applicability and importance of those Saussurean insights. Durkheim’s major works, together, address the basic theoretical issues that underlie the present distributed cognition approach – and thus comprise a reason for linking this book’s approach to Durkheim and for suggesting that this approach was what he was trying to describe in the inadequate language of his day – before the advent of systems theory and cybernetics, and before feedback and its coordinative capabilities were known. Culture, then, is being examined here as a cognitive avatar of the familiar Durk-

Durkheimian collective representations 41

heimian division of labor in society. The “labor” here divided is that of cognition (i.e., of knowledge linked to action). Durkheim while not explicitly recognizing this avatar of the division of labor, does accept cognition (consider his continuing reference to “representations”) as a key constituent of behavior, including social behavior (as in 1938: xliii, xliv, xlix, 93). In Suicide and in the The Rules of Sociological Method Durkheim considered how individuals link their varying particular knowledge, goals, and feelings (including motivations, fears, etc.) to their society’s collective representations, and how collective representations are distinguished, created, and maintained within the same individual consciousnesses that also embody each person’s individual knowledge, goals, and feelings. And the present cognitive approach attends seriously to religion’s role in providing the overarching values and motivation which enable people to know what is understood to be the common good and to act on such knowledge even at the expense of aspects of their individual good – questions which Durkheim focuses on in The Elementary Forms of Religious Life. Durkheim (1938: xliii, 14) was right in speaking of the analytic need to view societies (and sub-societies) as “things”, but he was, at the same time, misled by his insight. Because of our difficulty in viewing our own society in such a manner, he thought that this perspective was a difficult one to attain (see 1938: 14–131 [Section I of Chapter II]). He seems sometimes to have confounded our ease in so viewing other societies with presumptions about evolutionary stage and its effects, and sometimes to have confounded our inside view of our own society with his understanding of “advanced” society, and our outside view of other societies with his understanding of more “primitive” societies. In spite of various contrary insights, he seems generally to have accepted “our” view of “them” as different from us – in the sense that their social things are more readily apparent than are ours. He did, however, correctly by modern views, understand the difference as one not of mentality but of the social response to political-economic underpinnings. The preceding ambiguities notwithstanding, Durkheim did gain the important insight that we hold collective representations within us – as our personal representations of the collective knowledge (including collective beliefs, imperatives, will, moral standards, and so forth) – and that, thus, within us is the only place wherein they reside. He understood that we acquire our representations of collective knowledge through our experience of our society and its standards – that is, through our interactions with our fellows, and our attempts to coherently organize and understand (i.e., make

42 Chapter 2 – Background and history sense of) what they do, say, and say should be done. This is not a mystical view – and, indeed, he eschews any mysticism – but it is a view which was hard to express and explain before the development of modern cybernetic models. It is here – regarding the conceptualization of systems of collective representations – that Saussure, looking at the more constrained, structured, and limited system of langue had the advantage. He inferred langue as the shared, passively received regularities upon which speakers relied in their individual, active construction of acts of parole. Building on Saussure’s insights, subsequent linguists went on to construct various characterizations of the grammar of language’s phonology and then syntax. The attempt at a characterization of the social “grammar” which must presumptively underlie and enable social acts has only little been tried. The problem of such construction seems harder for society than for language because of the apparently greater looseness of much of our socio-cultural life (vs. of our linguistic usage), because of the problem of disentangling social regularities and forms from the linguistic forms by which we represent them in conversation and in much of our conscious thought, and because of our lack of good forms for the expression of such regularities as we find. By “grammar” here, even for language, I refer not to the lists of rules which linguists have come up with (resorted to) in order to describe the regularities of language, but to the regularities of patterning, reference, inference and so forth which we each subconsciously infer as we learn our native language, and on which we rely in our construction of the messages that constitute our communication. The emergence (and recognition) of linguistics as a science was greatly aided by the existence of quasi-mathematical forms capable of describing the regularities which linguists found, but the history of the discipline has, at the same time, been bedeviled by the naive leap that many linguists have repeatedly then made – to the belief that their rules directly represented the form in which speakers held and applied the information in question. The collective representations of, for example, some ethnic group held by these various groups within the wider society can vary from one another to some significant degree – see Chapters 5 and 8 of the present work for extended presentations. At the same time, as Saussure noted, the demands of effective communication (across such group membership) require that the various representations not be too divergent – or at least not produce substantially differing conversational usage and referents. It is in this sense that Durkheim regards the collective representations of sub-societies to be con-

Durkheimian collective representations 43

strained by (subordinate to) those of the wider society.16 An equivalent constraint, but in a form that is more theoretically challenging, is that the collective representations within the various groups to which sets of given individuals belong not be so divergent as to cause great cognitive dissonance for those individuals with overlapping membership. This last is a way of pointing out that, while the groups in question are made up wholly and only of individuals – and thus defined by the collective force of those individuals, the individuals in question can (and usually do) belong concomitantly to a number of such alternative groups (sometimes, even, simultaneously to directly contrasting groups – as the child of a Serb-Croat marriage in Bosnia or the child of a Jewish-Italian Catholic marriage in the US). If collective representations are indeed “things”, as Durkheim asserts and as I believe, then we are left with the problem of wherein exist these things, how are they to be observed, and how do they exercise their force. This perspective highlights the apparent paradox of the Durkheimian conception: our thought and behavior are constrained and shaped by collectivities to which we do not intrinsically belong – regarding each of which we each see our own selves as standing outside and looking in. I am a non-religious Jew, an anthropologist, an agnostic, a native North Carolinian, an escaped Southerner, and many other things – each of which represents some sort of self identification. I belong to other groupings – such as Riverside residents, Californians, White Americans, young seniors (the over 55 crowd) – that do not so much form my self-identification but which loom large among my collective memberships, and within which collective representations surely inhere. It is this problem regarding the de-

16

Who or what are the “real” X is and can only be the combination (whether it be the logical sum, logical intersection, or something in between) of core collective conceptualizations held by the various social groups (societies and sub-societies) concerned with the X. The relationship among such alternative conceptualizations is itself up for negotiation. That is, insofar as the various groups share some arena of discourse their conceptualizations have to relate to one another in some systematic (translatable) way, but which conceptualization (definition) assumes primacy (is acceded to by other interested groups) varies according to the socio-political situation. Sometimes, as with the Sami, the authoritative voice is that of the focal instantiators of the group being labelled. In other cases, as with Blacks in the classic American South or Jews in Nazi Germany, it is the opponents who have such primacy. In still other cases, as with Jews in modern America, there exists no clearly defined definitional position.

44 Chapter 2 – Background and history fining locus of collective representations which has repeatedly led scholars to accuse Durkheim of mysticism. 17,18 The view of society (with culture) presented in this book is consistent with the Durkheimian plan. It is based on the division of labor in society, and fleshes out the cognitive and conceptual side of that division. It recognizes and lays the basis for an explanation of the socially bound yet transcendent role of religion. It accounts for how collective representations can exist and be distinctive from individual knowledge and understandings even while sharing with those individual desiderata their location within the individual human mind. It accounts for the role of collective representations in individual thought, values, goals, and action. More strongly, I suggest that this view is very close to that which Durkheim himself was groping for, given the minimal availability in his time of useful analytic models for thinking about such distributed and projected systems and thus the minimal availability of relevant linguistic concepts for their description. This claim is based especially on his discussion of relevant issues in the Rules… (see the quotations appended to this discussion), and I am relying heavily on the analysis of the Saussurean paradigm offered by Kronenfeld (1996), including the links between Saussure and Durkheim discussed there.

17

18

Durkheim did not specify such a locus, but, in his study of suicide, used statistical norms for different social units as indices of relevant collective representations; however, such regularities seem more to be evidence for the existence of such representations than descriptions of their form. In a sense we could back up a little and say that society is constructed of roles, which individuals then fill. But this analytic construct still does not resolve the problem, since we still, as individuals, move in and out of roles, and to varying degrees in variable situations choose to read those around us as filling one or another role. Furthermore my kinship research has shown conclusively that whatever associations we have of behavioral regularities with kinship roles, at least, is very hypothetical and idealized; kinship roles are clearly not the units in terms of which kinship behavior is structured, and I strongly suspect the same kind of study would produce the same results for any other set of role terms.

Durkheimian collective representations 45

1.1.

Appendix: selected quotations from Durkheim’s The Rules of Sociological Method (1938)

1.1.1. A system of mental entities – as in Saussure “The system of signs I use to express my thought, the system of currency I employ to pay my debts…” (p. 2). cf. Saussure’s characterization of language as a system of signs, and his comparison of the system of linguistic value to that of the system of economic value. 1.1.2. System of generative patterns The idea of a system of patterns (aka rules) that generate the regularities we see in collective representations – cf. the grammar of a language. “None of these [collective habits] can be found entirely reproduced in the applications made of them by individuals, since they can exist even without being actually applied.” (p. 7; emphasis DBK)

1.1.3. The crowd as a minimal simple example, a microcosm, of a society. This pertains to the relevance of Kronenfeld and Kaus’s (1993) simulation of the development of “emergent properties” via feedback relations among individuals in a group. “So, even if we ourselves have spontaneously contributed to the production of the common emotion, the impression we have received differs markedly from that which we would have experienced if we had been alone.” (p. 5; cf. pp. 4, 9, 104, 105, 106). We each are coerced by the crowd – which consists only of us. And cf. the characterization of the hoard on p. 82 as having no segments or internal parts and no internal aggregates. Society is distinguished from what is produced by individuals in isolation, not so much from what produced by individuals per se. “…the only elements making up society are individuals…” (p. 102); “Of course, nothing collective can be produced if individual consciousnesses are not assumed; but this necessary condition is by itself insufficient. These consciousnesses must be combined in a certain way; social life results from this combination and is, consequently, explained by it. Individual minds, forming groups by mingling and fusing, give birth to a being, psychological if you will, but constituting a psychic individuality of a new sort. {fn 18: In

46 Chapter 2 – Background and history this sense, and for these reasons, one can, and must, speak of a collective consciousness distinct from individual consciousnesses. In order to justify this distinction, it is not necessary to posit for the former a separate personal existence; it is something special and must be designated by a special term because the states which constitute it differ specifically from those which constitute the individual consciousnesses. This specificity comes from the fact that they are not formed from the same elements. The latter result from the nature of the organicopsychological being taken in isolation, while the former from the combination of a plurality of beings of this kind. …} … The group thinks, feels, and acts quite differently form the way in which its members would were they isolated” (pp. 103–104; emphasis DBK) 1.1.4. A collective emotion “A collective emotion …a product of the actions and reactions which take place between individual consciousnesses… If all hearts beat in unison, this is not the result of a spontaneous and pre-established harmony but rather because an identical force propels them in the same direction. Each is carried along by all.” (pp. 9–10). Regarding a feedback relationship, though between different kinds of variables, “…the method of concomitant variations or correlation…shows [two facts] as mutually influencing each other in a continuous manner…” (p. 130). 1.1.5. Universality Regarding universality as being neither sufficient for the existence of a social fact (p. 6) nor necessary for it (p. 7). “As for the forms that the collective states assume when refracted in the individual, these are things of another sort” from the truly social collective states themselves.” 1.1.6. Regarding the possibility of a “Social Psychology” Regarding the possibility of a “Social Psychology” (concerned with collective representations) and of an overarching psychology that deals with the common features of all “representations”, social and individual (p. xlix),

Anthropological studies of routine decision making

47

…they are … also mental after a fashion, since they all consist of way of thinking or behaving. But the states of the collective consciousness … are ‘representations’ of another type. The mentality of groups is not the same as that of individuals; it has its own laws.” (p. xlix), and on pp. l– li, With this distinction in mind one can ask whether individual and social representations do not, nevertheless, resemble each other in that both are equally ‘representations’; and whether in consequence of these resemblances, certain abstract laws would not be common to the two realms … but the way in which they attract and repel each other, unite or separate, may nevertheless be independent of their content and may depend uniquely on their general quality as representations… . Thus we arrive at the idea of the possibility of an entirely formal psychology which would be a sort of common ground for individual psychology and for sociology… all we know about the way in which individual ideas combine reduces itself to those few very general and vague propositions commonly called the ‘laws of the association of ideas.’ As for the laws of collective thinking, they are still more completely unknown. Social psychology, whose task it is to determine these laws is scarcely more than a name, without a definite subject matter, and including all sorts of generalities, diverse and inexact.

And see p. 8, where the “individual manifestations” of social phenomena are discussed, and where the possibility of a “hybrid science” of “social psychology” is raised. And see p. 111, where “it is indisputable that social facts are produced by action on psychological factors.” Note the fact that Durkheim is not resisting psychology per se, but the individual psychology of his (and, often, our) day – and, more particularly, the “associative” variety of psychology which then was dominant. 2.

Anthropological studies of routine decision making

The first more recent source is a research area that was salient in the 1970s and 1980s – the ethnographic study of routine decisions and their modeling. Naomi Quinn and others played an early role, but key early contributions were those of Christina and Hugh Gladwin. 2.1. Christina and Hugh Gladwin First, an ethnographic insight – similar to that of Hutchins (see below), and relating to traditional ethnoscience: If you want to collect good, reliable

48 Chapter 2 – Background and history data about cognitive categories or operations of your informants from them, you have to find out how they structure and communicate about these things. In collecting economic data from Ghanaian fishmongers in Cape Coast the Gladwins made some initial eliciting mistakes that turned out to be revealing. They initially surveyed a sample of informants, asking questions about volume of sales, cost, margin, profit, etc. for each day. The numbers did not add up. The fishmongers looked dumb. But, later, as their ethnography proceeded, they learned how the fishmongers (who were illiterate) structured and stored this kind of data (by lots, each of which was dealt with over a spread of several days), in traditional units of baskets and hundreds. When they eventually started surveying the same women with questions and categories that were meaningful to them, the numbers did add up. This was not simply a question of interviewing in Fante, but more importantly depended on an understanding of how the Fante conceptually structured the data in question and how they processed that data. With these ethnographic insights, they collected market choices, pricing, etc. for one whole fishing season from a sample of fishmongers. Second, having in mind the aim to model the marketing choices that fishmongers made, the Gladwins – in addition to normal ethnographic participant-observation, generalized loose interviewing, and more formal ethnoscience-based formal elicitation, used a particular eliciting approach, paired comparison eliciting frames, to elicit data on the conditions that would lead to one particular choice vs. another – for all relevant pairs. That is, they asked questions such as “under what conditions do you X instead of Y?” and “Under what conditions Y instead of X?”. In approaching the analysis of their decision data they were strongly aware of studies of constraints on human information-processing (such as Miller [1956] and various studies of Kahneman and Tversky [see Kahneman Slovic and Tversky 1982 for a later summary]) and of ways people used to get around these constraints. They considered the information processing involved in the fishmongers’ marketing decisions from this perspective. Important among these information-processing adaptations were a) transforming continuous variables to a small set of contrasting categories, b) transforming complex simultaneous calculations into sequences of simple calculations, and c) building in routine “rules of thumb” wherever possible.19 (Later, Christina Gladwin organized these insights into a generalized version 19

This is a topic to which Jean Lave’s study of Liberian tailors (Lave 1988) also contributes importantly.

Anthropological studies of routine decision making

49

of “elimination by aspects” (from Tversky) which involved ordering the relevant attributes by importance, and then reducing the complexity of action choices by first eliminating action alternatives that failed on relatively important attributes, before moving down the list to a consideration of less important attributes.) In their analysis of their actual marketing data, they, especially Christina Gladwin, tried several approaches. They applied a variety of formal economic models directly to their data – and they applied an ethnographic flowchart model of the decision process (based on their paired comparison and other interview data) which they then interpreted economically (i.e., by applying formal economic models to the successful flowchart – rather than to the raw economic data). They found that the simple flow chart model did much better at predicting marketing decisions (r-squares of ca. .9) than did the economic models (at best, r-squares of ca. .6) – even though, ultimately, the decision process for the fishmongers was found to be driven by an economic consideration, “satisficing”. The reason for the difference in models’ success seems to be that, since the decision was a routine one, informants simplified the decision process and turned it into a kind of check-off list with cut-off points for decision forks – and thus transformed underlying economic continua into small numbers of discrete categories. Their actual decisions then were driven by the simplified decision plan (represented by the Gladwins’ flow chart) rather than directly by the relevant (continuous) economic variables. I hasten to add, though, that this simplification process was in no sense mindless. During the period of the Gladwins’ study Ghana went through a major devaluation of its currency (particularly affecting fishmongers’ transport costs) and the change from pounds/shillings/pence to a decimal system. The fishmongers dealt with these (and other changes in relevant conditions) successfully and without missing a beat. 2.2. James Young Young (1981) offers a study of medical treatment choice in rural Mexico that is impressive in its insightfulness, its rigor, and in the clarity with which it makes its case. Its major substantive point concerns the dynamic considerations (available resources, seriousness of the illness, past experience with different treatment options) that shape decisions. The study attends carefully to “native” categories, and then carefully models them, their interactions, and their implications in a decision model. It effectively (and nicely) treats inter-

50 Chapter 2 – Background and history informant differences within a general decision model. Its statistical test is clean and convincing, and shows the tested model to be extremely powerful.

2.3. Carol Mukhopadhyay Mukhopadhyay in her dissertation found that more complex decisions involving actions that could not be totally routinized were harder to construct any simple model of. The example she studied was the household division of labor between spouses in a blue collar Los Angeles suburb in the 1970s. In the routine course of events, the shifting demands of young children, special maintenance problems, changing weather, etc. made decisions regarding which spouse actually did which tasks hard to routinize. This difficulty became even larger when the patterns that the spouses had grown up in were impacted by substantially changed conditions – as when both spouses worked outside the house (a relatively new condition for the people in her study population). She found that she could not model (or, thus, predict) exactly what tasks would be done at any given time by which spouse (or by others, such as children, other relatives, or hired help). What she found she could do was delineate “task areas” which one or the other spouse was in charge of – where “in charge of” entailed making sure the task got done (finding someone to do it, making sure needed materials were available, and so forth). One spouse might well wind up doing substantial work in a task area of the other spouse; similarly, depending on circumstances, specific tasks might be assigned (whether on a more or less permanent basis or temporarily, whether as an obligation or as a favor) to a child, to some relative or friend not in the household, or to hired help. The basis of task area assignments, and thus for many default actual task performances, was what Mukhopadhyay called “cultural precedents” – which here mostly amounted to how the couple’s parents did it. But sometimes the two parental family households did “it” differently – and so the couple formed by their children had to negotiate how it would be done in their new household; mostly these differences showed up at marriage, but relevant situations could arise later – as children came along or as housing conditions changed, or etc. And sometimes other conditions might lead a couple to diverge from relevant parental precedents – and thus to have to renegotiate their division of labor; such conditions could result from personal tastes, particular personal skills, or the changed household framework created by wives working outside the home. By “negotiate” I sometimes do

Anthropological studies of routine decision making

51

mean actual discussions (even arguments), but sometimes the process was a more subtle and indirect one involving feedback loops between the spouses that included comments such as “I’m too tired…”, “I have a meeting…”, or “I got assigned to the swing shift…” – or ones like “that’s really nice of you…” or “gee, you’re really good at …” (these are not quotes from Mukhopadhyay, but are my rendering of what I understood from her work). Mukhopadhyay contributed to our understanding of culture in several ways. She provided a general model for a specific cultural arena that subsumes (and is instantiatable in) a class of specific decision events in that arena. She clarified the contrast between individual decision-making routines and the actual decisions that came to be made – and the contrast between routines based on cultural precedents (i.e., tradition) and individually evolved routines. She therefore shares with Randall (below) a concern with difference between routinized (non-calculated, “non-“decision) decisions and actually calculated real decisions. Randall’s approach handles the difference via its distinction between unmarked defaults and marked deviation from the defaults. Mukhopadhyay’s approach uses cultural precedents and basic defaults in individual decision-making routines as local defaults and then addresses novel actual decisions and/or active decisions about which of a set of alternative routines are to be applied to a given specific situation. She thus, also, addresses the problem of deciding among competing alternative routines (or models). Important too is her distinction between general responsibility for a task area and the actual doing of actual tasks – and the ambiguity on the issue (sometimes intentional) in informant responses to survey or interview questions about who does what, and the efficiency payoffs entailed by such an arrangement. Mukhopadhyay’s combination of problem, findings, and approach (including both her ways of specifying parts of the problem and her approach to modeling processes) move us substantially away from the simpler framework of routine decisions and in the direction of a more integrated understanding of how culture actually works in the day-to-day activities of ordinary life. For this reason, and in this context, her study has considerably influenced my view of culture and of cultural models. 2.4. Robert Randall Robert Randall in a study of fishing among the Samal of southern Mindanao considered the decisions that fishermen made in their day-to-day work. But

52 Chapter 2 – Background and history instead of framing specific decision analyses for each type of routine action and instead of treating non-routine situations and decisions differently, he brought the whole of Samal fishing together in a single analysis and embedded that in the larger frame of “making a living”. In the most ambitious and thorough application of Greenberg’s notion of marked vs. unmarked categories that I have seen, he treated the whole of fisheries – and, indeed, of making-a-living choices – as a single hierarchy of activities wherein each node was characterized by a null or “default” unmarked choice and various exceptional or marked alternatives. The unmarked choices represented what one did automatically in normal circumstances – without special thought or attention (and, so, without any conscious decision). Abnormal circumstances could force one to raise the relevant decision tree node to consciousness and make an actual decision. Often, for commonly encountered non-canonical circumstances, there would be secondary marking defaults – of the “if you can’t do x then do y” sort or of the “if x condition occurs then do y”. Thus, commonly encountered conditions were all factored into this elaborate but functional hierarchy of defaults. But, again, as we saw with the Gladwin fishmongers, the decisions and their arrangement were not some sort of mindlessly followed cultural habit. They were the active products of experience coupled with goals, capabilities, tools and so forth. As any of these changed the structure of defaults could be changed quickly to adapt. The marking hierarchy was a communal cognitive structure which individuals learned through participation, interaction, and conversation with others. Individuals used the collective models as a basis for their actions, but each adapted it to their own particular needs. Randall’s contributions, from our present perspective start with his framing of all of life in a decision/action structure, and then continue with his use, at all levels, of unmarked default base choices and marked exceptions with given conditions for non-canonical choices. Important to the development of his decision/action structure was his use of base line – default – eliciting. For example: Q: A: Q: A: Q:

where do you fish? location X. If you did not fish at X, where would you fish? location Y. Why not-X? (Why then Y?)

The function of such simplifying structures as the Gladwins and Randall each portrayed, if somewhat differently, was a labor-saving one – a way to

Anthropological studies of routine decision making

53

let people focus on the job at hand while guaranteeing that the preferable job was being done in the right manner. Randall’s use of marking was powerful and creative – and seems to me to have captured something important about cultural patterns of behavior. His conception has much influenced my understanding of culturally based and shared cognition. 2.5. Stuart Plattner Stuart Plattner, in his “Economic Decision Making of Marketplace Merchants: an Ethnographic Model” (1964) provided us with an exciting new model for building and testing models of some important (but fast-moving and largely subconsciously carried-out) cultural activity. Plattner’s ethnographic problem (or, at least, one such problem) was how to elicit precise, accurate information regarding reasons for making a large number of routine decisions – stuff that is habitual and about which people are, when removed from the fact, largely inarticulate. That is, when they are engaged in the activity and actually making the decisions of interest, they are too busy to talk, and when not engaged in it they can’t summon up the focus or the thought train. Platter’s answer was to create an interactive computer program to simulate the market situation and allow market stall operators to imitate (or, actually, to recreate) their own behavior realistically. The program was set up as a game that venders (or would-be venders) could play, where playing involved running a stall. 1) In this situation his market venders were in their normal marketing thought train, and could give the ethnographer a description of, or commentary on, what was going through their minds at any particular moment in any particular situation. And the situation could be frozen for in-action exploration of points without loss of other information or losing touch with the flow of action. The computer program was in many ways like a verbal model or description. But it had a great advantage over verbal models by being dynamic – and thus allowing Plattner to look at real-time feedback effects among venders and between venders and buyers.. 2) The process of writing the program, itself, proved also crucial. Plattner worked with market venders as he developed the game program, having them play successive versions and tell him how realistic they were, what in them worked and didn’t, and so forth. The program’s development process provided Plattner with constant feedback concerning how well the program was embodying actual market conditions and interlinkages, including the economic and social processes that were relevant to buy/sell price deci-

54 Chapter 2 – Background and history sions, and other vender decisions and the nature, form, and content of the included variables. The feedback, then, spoke to the question of how well he understood the relevant market mechanisms. 3) The program/game, finally, became a great teaching tool – for bringing home to students the activities, constraints, decisions, and pressures that were involved in marketing activity – and that had to be mastered by a successful market vender.

3. Comparison: Decision theory with simulation approaches We can conclude this discussion of anthropological and related background with a generic contrast between Decision Theory approach and Simulationbased approach. “Generic” because I am picking out tendencies that I see as being salient and important, but that may not apply to all examples of the indicated approach. The major contrast is between the criteria which are used to decide what ethnographic subsystems are to be included – that is, the standards of relevance and success (i.e., how you tell a successful model from an unsuccessful one). Both, in principle, share a range of ethnographic techniques based in structural/descriptive linguistics – even if the precise techniques do vary – and even if some practitioners got them from sources other than linguistics. Decision theory normally is more concerned with getting some sample of “objectively” described cases to evaluate the model against – while the simulation based approach deals, typically, with only a few, well-workedthrough cases. However the major issue is not the “few” but the “wellworked-through”. There is another difference that is more basic to the concerns of the present volume. Decision Theory wants to deal with ethnographically meaningful (“emic”) entities and concepts, but analyzes them in the context of a comparative (ethnological) context – and, ideally, uses them as input into non-culture-specific (“etic”) analytic theories. The general task was not unlike what was involved in typological studies in linguistics such as those of Greenberg (and it seems possible that some of these approaches were influenced by Greenberg’s work since the researchers involved were students in Greenberg’s classes or were students of students of Greenberg). The Gladwins were using their ethnographic sophistication to answer questions deriving from economic theory. Similarly, Plattner’s study was concerned with economic issues, though, perhaps, with more of a specifi-

Comparison: Decision theory with simulation approaches

55

cally anthropological flavor. Mukhopadhyay’s concerns, while involving economic issues and pressures, and while richly attentive to cultural forms, came fairly directly from traditional ethnological theory. Young’s work addressed central issues in the body of theory associated with Medical Anthropology. Randall is more of a bridge – or more ambiguous. He shared the Simulation-based notion of locating lower level ethnographic chunks in higher level ones, but shared in the Decision Theory predilection to frame these chunks in an analytic frame rather than in a user model type of frame. Simulation-based approaches analyze lower level (specific) “emic” 20 ethnographic entities and concepts in terms of higher level (still “emic”) descriptive entities. That is, the standard for what goes into a description of a scene is the determination of what aspects of that scene are relevant to the role that scene plays in the higher level (still “emic”) scene that includes it as a part or an input. The process here is like what one sees in the construction of a description or model of the grammar of a language. Frake, in his “A Structural Analysis of Subanun Religious Behavior” (1964b), most directly ties such a study to the linguistic model. His use of the linguistic model, incidentally, is represented not by a superficial use of linguistic descriptive or analytic forms (as was Metzger and Williams’ Tenejapa Marriage study [1963]), but, rather, embodies an adaptation of the goal and approach of linguistic study to the cultural enterprise. Like linguists, incidentally, he does not limit himself simply to the native’s conscious story line, but also includes functionally indicated covert categories, and the like (i.e., contra what might be imagined from any simplistic thoughts about Harrisian – as in Marvin Harris, the anthropologist – “emic” presentations); that is, we have not a folk theory but a descriptive theory of specific folk knowledge. Since his study, and he, are from the pre-computer age, he, in this article, is not actually modeling the cultural knowledge in question, but, rather, is only characterizing what such a model might look like. At the same time, such a schematic characterization makes it easier for us to see what the game is than do denser reports of actual simulations. Schank and Abelson’s, Scripts, Plans, Goals and Understanding … (1977), best epitomizes the Simulation-based approach. They model some real chunks of real life. Their measure of success is how realistic the output looks – they do not worry about the wider theoretical findings concerning 20

“Emic” is used here in the Pikean (see Pike 1967) – linguistics-based – sense of being an analytical account based on native categories and observations which the analyst has elicited. Marvin Harris’ notion of a folk theory is NOT intended.

56 Chapter 2 – Background and history knowledge systems, grammatical systems, and the like which their work relates to. They do address the issue of what we might call descriptive theory – the computer modeling equivalent of anthropology’s ethnoscience. They do make attempts to generalize about knowledge systems, but these generalizations are too tied to their particular analytic apparatus, research design, and analytic simplifications to have the general import that they would like. Hutchins’ Culture and Inference (1980) embodies a more analytically rigorous (in the mathematical sense of clearly defined units and operations – vs. the more ad hoc components of a computer program) model of a system of cultural action and the knowledge that is embodied in it. In some important ways Hutchins’ work can be seen as an instantiation of the Frakean program, though Hutchins also has some fish of his own to fry regarding the cross-cultural consistency/nature of logic and reasoning. In some ways his work can be seen as transitional to our next, schemabased, approach to cognitive structures: he has the formality but not the selfcontained functioning cognitive “organisms” that we will explore there. 4.

General cognitive background

This book is not the place to get into theories of cognition – they are too many and too rich. Kronenfeld (1996) offers brief discussions of “situated” vs. “symbolic” cognition (pp. 40–41), schemas (pp. 17–20, 27–30), and Piaget (pp. 237–242), among others, in the context of the language and thought. My general approach to individual cognition comes out of my reading of Piaget, Gombrich, Bartlett, and Kohler, among others. The brief discussion below summarizes what I have come up with, and what thus underlies this book’s approach to collective cognition. 4.1. Mode of intellectual functioning Our mode of intellectual functioning is tied in a particular way to the “real” world. We are not animals that carefully find out about the world, but, rather, animals that keep trying to outguess it. But we do learn from our mistakes – and hence home in on that part/aspect of the world that we are concerned with. But such learning is a function always of action – trying and trying, and each time getting feedback. See also notes on Gombrich below.

General cognitive background

57

4.2. Cognitive schemas Cognitive schemas are models – with the normal limits of models (regarding the partial nature of the representation of reality, and the possibly rough nature of the approximation – depending on what is “needed”). In the Piagetian world children construct an internal representation of the world. But it’s not just a “picture”; it’s more of a model – a model that has actually to work and, depending on the subject matter of the schema, has to work in the actual world that we actually live in (Euclidian geometry, proximate causes and effects, solids/liquids/gasses, folks with various motivations but similar make-ups, etc.). The model, then, has to be a kind of operating system. Again see Gombrich notes below. 4.3. Piagetian stages The famous Piagetian Stages are epiphenomenal – joint functions of a) the level of generality of a schema and b) the joining of a schema for talking about action (involving self-consciousness, facility at abstracting, learning the words, etc.) to the (sensory-motor) schema of the action itself. The child’s development entails learning sensory-motor schemas, and then developing (some of) these into concrete operational ones, and some of these into formal operational ones. So, in general, and initially, abstract knowledge presumes and depends on concrete experience. But this relationship too, between abstractions and concrete instantiations, is one we learn about. We can and do learn to presume the existence of a concrete instantiation when we encounter appropriate abstractions – which, as Sperber points out, is what allows us to communicate about stuff we haven’t yet encountered, and what, thus, can guide our acquiring of the relevant experience. Note, such abstractions do not necessarily represent mature structures (operational schemas) that have achieved equilibrium (and the kind of internal consistency and coherence implied by Piagetian equilibrium) and that have been refined against the world of interactive experience. At the same time, this ability to act as if an abstraction has an instantiation when we don’t know the instantiation enables us to “believe in” imaginary beings, events, and situations (whether myth/fantasy or prediction). This presumption of the “where there’s smoke there must be fire” sort can produce reification – the notion that, just because someone has come up with an idea and defined it, the thing posited by the idea must actually exist. This backward reasoning from abstractions to concreteness is tied to our

58 Chapter 2 – Background and history ability to define something we want abstractly and then posit its existence – as in religion. It enables us to construct paradoxes (e.g., “my god is both human and superhuman”) and posit the existence of something that fits the conditions. Being clever critters, we can, when challenged on such things, come up with very ingenious “solutions” (“my god is human in her ability to empathize but superhuman in her ability to avoid being distracted in her aims by such empathy”).

4.4. Gombrichian conventions In connection with Gombrichian conventions (see below), note the Piagetian interplay of expecting (what to see and/or represent in the art), learning (to see or include a little more than what is looked for), construction of representations (or representation rules and conventions), and/but constraints on freedom of representation entailed by the particular nature of the part or aspect of the world being represented and by the particular nature of our perceiving and computing machinery. The choice of what to represent in art via artistic conventions is a relatively free one, as is the choice of by what representational conventions it is to be pictured. But there do seem to exist constraints of nature within the conventions: you do not have to use size to represent social importance; but if you do so choose, big will be more important than small. Or if you choose to use size to represent distance; big will be close. Location on the page may represent heaven or earth or a time line (Chinese); up will be heaven or time will be a line (vs. hop scotching around). The choice of what to include in a picture vs. leave out (cf. Gombrich’s discussion of the silent traveler [1960: 84–85]) is schema driven, too – in the sense that it is the artist’s schema (relating to cultural conventions) which defines what in a scene is intrinsic to it and what is incidental – an accident, perhaps, of the moment. The issue concerns what is seen as noise and what as message/signal. We can see Western art, at least in periods, as (among other things) a gradual elaboration of a convention of representation – moving toward successively more immediate (momentary) reality – through the Greeks and Romans, then freezing for a while before being picked back up in the Renaissance. The Dutch later moved it in the direction of super detail, while the impressionists aimed more at the momentary appearance or feel of the thing pictured; the expressionists shifted the focus more to the feel of the

General cognitive background

59

thing/situation relative to the appearance, and abstractionism dropped the appearance of something totally in favor of the feel. Intersecting, but evolving with, this convention moving toward immediacy and perceptual accuracy was maybe a (particularly French) convention moving sort of schismogenically (Bateson 1958: 171–197, 285–292) toward ever more heroic heroism – culminating in Napoleon’s tomb; and the modern Soviet variant embodied in Socialist Realism. These “conventions” seem to operate in a schema-like manner within the minds and actions of individual artists. But the consistency of patterns running across individual artists, while focused within constraints of space and time and dependent often on teacher-student relations suggests extraindividual cultural forms. The word “schema” is sometimes used for these extra-individual cultural forms (as well as for individual forms), but – for reasons central to this book – I prefer to keep the two kinds of forms clearly distinguished. Hence I am here using the term “convention” for the cultural forms – and want to suggest that these conventions will turn out to be forms of the “cultural models” on which much of this book is focused. 4.5. Conventions in literature Similar conventions exist in literature. Daphnis and Chloe offers an example of a single specific conventional story (a version of the royal foundling) embedded in different wider cultural social conventions. Both cultures involved in our comparison (2nd Century AD Greek and 16 th Century Elizabethan) knew that “blood will out”, and thus that a royal foundling’s royalty would always express itself (see, also, for example “The Princess and the Pea” tale). But the two cultures differed on what form the “expression” would take, a difference which seems to hinge on different conceptions of the relations among the various social classes (or of the skills and predilections of each). For Longus’s late Greek version of the story (Longus 1968 [orig. 2nd Century AD]), the royal foundling raised among shepherds and cowherds reveals his status by being more skilled at caring for sheep than those born to such work (that is, he is better at whatever he tries – which justifies his exalted rank).21 For Day’s Elizabethan version (Longus 1890 [orig. 1587] based on Longus’s, but via Amyot’s French version, and with some recrea21

We have a little of this attitude in our present society, too – in connection with some of the skills or capabilities we attribute to class differences such as those obtaining between managers and workers, etc.

60 Chapter 2 – Background and history tion of missing material) the royal foundling reveals his status by doing different (and nobler!) things than ordinary cowherds and shepherds. He’s one of those shepherds who plays beautiful music on pan pipes and writes poetry as opposed to those shepherds who wrestle with sheep and smell of sheep dip. That is, in his world, each estate is born to its own work.22 The same story gets fleshed out in different (even though historically related) cultures according to wider cultural schemas concerning social class relations.23

5.

Piaget’s paradigm

Much of this discussion, including unlabeled quotes, is taken from Flavell 1963; other parts are my own contribution. Much of my understanding of and interpretation of Piaget’s paradigm is laid out in Kronenfeld (1996: 237–242). 5.1. Basic framework We get from our biological inheritance a “mode of intellectual functioning” which generates our cognitive structures and which remains constant throughout our lives. Its fundamental properties are the functional invariants of organization and adaptation: Organization. An organism is organized, is a system. Adaptation, a given. The organism-environment interchange has the effect of modifying the organism in such a way that further interchanges, favorable to its preservation, are enhanced. Adaptation has two intimately related, but distinct aspects – assimilation and accommodation. Assimilation. Adaptation of the “aliments” (substances taken in): the process of transforming the “substances” taken in (food, objects for action, information, …) into a form in which the organism (as it is constituted and organized) can use them. It is a “coming to grips with the properties of the thing apprehended.” 22

23

Indian castes maybe have a something of this attitude, and some of our own American folk views of supposed racial differences in smarts, muscle, rhythm, and the like also have something of this flavor. My thanks to Judy Z. Kronenfeld who originally called my attention to the differences between the Hellenistic and Renaissance versions of this story in her work on Renaissance pastoral.

Piaget’s paradigm

61

Accommodation. Adaptation of the organism: the process by which the organism changes itself in ways which allow it to make fuller use of more “aliments” (substances). As the organism comes to grips with properties of a newly apprehended thing, it reshapes itself a little to them. Schemas (or schemata) is the name we give to cognitive structures. Schemas are like little organisms: they have structure and they function – they do things. They, then, are organized and show adaptive change. “A schema is a cognitive structure which has reference to a class of similar action sequences… strong bounded totalities in which the constituent behavioral elements are tightly interrelated.” “Schemas are labeled by the behavior sequences to which they refer.” They are “behavioral totalities”, functional (psychological) organs. Schemas can be embedded in larger schemas. Schemas are action, but much of the action can be internalized (i.e., occur as interior thought). Repetition, generalization, differentiation. Schemas come with an intrinsic tendency toward repetition and generalization – they are, at least after each adaptive change, “hungry” for new trials and “hungry” to gobble up all the variety of “aliments” that they can process. The generalization process rides on their ability to take in (“eat”) a little broader category of inputs than they specifically were adapted to (an instance of assimilation); so the world of their application keeps growing. At the same time, the process of generalization and growth allows the schema to begin internally adapting itself to deal more precisely with each specific variety of input; thus a process of internal differentiation (related to recognizing the variation among the new inputs) produces a family of sub-schemas, each adapted to a different input/situation (an instance of accommodation). Equilibration. Each time something new comes into a schema, it produces a disequilibrium. Each successive disequilibrium is brought under control, into a newer and more complex or embracing equilibrium, through a process of equilibration. Each successive equilibrium period is a period of temporary stability/closure – before new assimilations produce new disequilibrium. Equilibrium is dynamic, and entails feedback (as in a thermostat). Successive equilibria are more general and embracing of wider ranges of application than those they grow out of and encompass. The various conservations we come to understand (e.g., conservation of volume of a liquid regardless of the shape of the container into which it is poured) depend on such equilibria – including the experience of reversibility involving feedback actions. Group and lattice structures (from mathematics) represent the formal properties that structures (of operations) based on equalities (involving

62 Chapter 2 – Background and history quantitative operations) and ones based on inequalities (such as inclusion relations among sets), respectively, must have to be in equilibrium. The mathematics here is not a metaphor, but is a literal and precise summary of the relevant properties that must be present in the schema for it to be a contained system with the requisite feedback. Groups are defined by the properties of composition, associativity, reversibility, and the existence of an identity operator; lattices are more complex. A particular group structure that looms large in Piaget’s work is the “four group” (INRC: identity, negation, reciprocal, correlative) which embodies the structural knowledge we need to be able to move around in our local Euclidian space. 5.2. The stages Sensory-motor schemas are those involving unverbalized actions in the world. Concrete operational schemas refer to those actions for which we can give specific verbal characterizations. Formal operational schemas refer to actions for which we can give formal analytic representations. Piaget treats these as successive specific stages in a developmental sequence. He presents them in a way that often leads people mistakenly to see them as levels of generalized intelligence keyed to specific ages (and, in terms of which, various people are more or less advanced or retarded). But a) The stages represent successive levels of understanding. In general, we have to be able to do a thing, and understand behaviorally how to do it, before we can talk intelligently about it, and we have to be able to talk accurately about a bunch of specific instances before we can generalize effectively across the instances and abstract out the underlying axioms. But being able to do a thing does not automatically entail being able to talk about it – we have to learn the lingo, the rules for representing various actions in the lingo, and ways of reasoning within the lingo. A similar learning process is involved in moving from concrete operations to formal operations. And b) It follows that we are not at the same level for each of our areas of knowledge. We can have formal operational schemas in one area (say geometry), while still having only sensory-motor ones in another (say, biking – and the notions of balance and compensation involved there). These inconsistencies are called decalages (horizontal for different skills with reference to the same stage, vertical for same skill at different stages). When we are given ages for the onset of equilibrium structures (developed schemas) at one or the other level, these ages represent the earliest that structures of

Piaget’s paradigm

63

the given sort show up in some given learning context – not the age at which all thinking, or thinking in general, is supposed to be at that level. The ages are fairly standard, to some degree, because Piaget’s examples came from a very culturally homogeneous universe. They are also standard, to some degree, because each of the kinds of learning required takes time, and thus the result is, at least in principle, the necessary length of time it takes to get there from zero. Much writing presumes or assumes the ages are standard because of some intrinsic maturational process (similar to those postulated by ethologists) – i.e., that, regardless of experience, there is something you can do at age 5 that you could not do at 4. While we can not rule out some influence by such maturational factors (nor, I think, does Piaget rule such out), we need to be clear that the ages are primarily a result of the needed cumulation of relevant experience – and thus are intrinsically relative (vs. absolute). The illusion of absoluteness comes from the absolute amount of experience (of various sorts) that it takes to get from zero to the first instance of each stage – or to some other such milestone. Piaget’s use of finely graded absolute ages in his discussions results from his having used cross-sectional comparative data to study longitudinal developmental sequences; his use of such data works because of the great cultural and experiential homogeneity present in his sample population. c) Since an essential part of the theory is learning through action, and learning includes learning about the properties of the world, our learning also includes learning what kinds of understandings are possible and learning how to go about finding/forming these. Thus, the more experience we have forming equilibrated (successful) concrete operational schemas, the easier it gets to form each successive one and the more we tend actively to look for such. So, while our cognitive functioning at any given moment is not all of a stage, and we (therefore) cannot be said, simply, to be at some given stage (e.g., concrete operations), there is this meta-learning sense in which our stage of operational understanding is more general than merely a list of the schemas that we have at the given stage. This kind of learning how to learn is the kind of operation that Bateson (1972) has labeled “deutero-learning” and that is often called “meta-learning.” Piaget’s version is especially noteworthy because it is embedded in a coherent theory of epistemology and of learning – and because that theory speaks to the kind of natural world, experience-based learning that is our everyday experience (as opposed to the more specialized pedagogical theories that can underlie our schooling).

64 Chapter 2 – Background and history 6. Piaget vs. other cognitive psychology Schemas constitute the basic cognitive structures of Piagetian psychology. Schemas, similarly though not identically characterized, constitute the basic units of much American cognitive psychology. Gombrich (convincingly and usefully, to me) applies the schema idea to art genres; his usage is psychologically informed and sophisticated. My focus is on Piagetian schemas because he embeds them in a general theory of development (adaptive ontogeny – here, particularly, how an individual’s schemas develop and adapt as a function of that individual’s experiences over that individual’s lifetime, but also how schemas, looked at more broadly across individuals in a culture, grow), and ties them to a coherent view of epistemology (i.e., of the nature of knowing). In particular, Piaget’s notion of schemas as functional (doing something – something useful in some sense to the organism), his view of how they grow and adapt, and his discussion of schemas (including definitions, instantiations, and experiments) in terms of natural contexts, make his metatheory more attractive for anthropological purposes than typical American models which, while often more rigorously characterized in a formal sense, seem more tied to artificial (and artificially isolated) laboratory situations. Why schemas are useful to us, how they grow out of our experience, how they can be flexible enough to embrace new experience, how they can fit themselves to the variety of natural and cultural contexts in which we find ourselves, etc. are kinds of questions which seem crucial to any application in an anthropological context, and which, in general, American psychology either eschews or puts off until the indefinite future. Similarly, Piaget’s view of the holistic nature of intellectual functioning (as a single integrated system of subsystems) is attractive to anthropologists. The preceding is not to say that Piaget cannot be wrong on many details – both regarding what happens when and regarding the particular mix of genetic programming and experience-based construction that goes into various particular childhood actions and abilities. It is also not to say that Piaget himself (vs., by implication, his theoretical paradigm) has any particularly useful insights to offer us about cross-cultural differences (or, thus, about cross-cultural uniformities which might underlie such differences) or about the peculiar attributes of the kind of distributed computing system that culture or society seems to constitute – i.e., the world of Saussurean/Durkheimian “collective representations”.

Gombrich’s approach: Art and Illusion

65

7. Gombrich’s approach: Art and Illusion The art historian E. H. Gombrich, especially in his Art and Illusion (1960), provides an interesting and worked out application of the psychological views of gestalt psychologists and of cognitive psychologists like Piaget to cultural patterns of representation. What I want to do here is to characterize some of the lessons and/or insights regarding cultural representations that I have drawn from his work (and provide references to the lesson’s source). Early on (pp. 5–6) he reminds us of an important fact regarding optical illusions: we can switch back and forth between the alternatives; we can remember the one while seeing the other, but we can not experience both pictures at the same time. This implies that we use our perceptual input to recognize a picture rather than simply to see what’s there. With a messy or difficult picture, there may be a period of non-recognition; in such a period we see only a jumble of features, a kind of semi-structured chaos, not a picture; when we finally see the picture, we then lose the capacity to recover the chaos. Our perception is guided, from the beginning, by knowledge. The dependency of perception on prior knowledge gets us to the role of convention in art and the discovery of conceptual devices (such as systems for representing perspective) which can come to be integrated into the convention, and thus add to the capabilities (pp. 11–12). Here, we can see the Piagetian interplay of expectation and stimulus. A successful “device” is needed in order to represent some piece or aspect of “reality”. But we can’t just simply will ourselves to see whatever aspect of “reality” is at issue, and, within the range of imaginable devices, some work better than others. The contrast between relatively easy vs. hard representational devices relates to what I have spoken of as “natural conventions” (see below). Also relevant is Sperber’s idea of relatively “catching” vs. non-“catching” representations. Gombrich discusses the role of prior models and conventions in drawings purportedly “from life” on pp. 68–83. He shows the need for representational conventions, and shows that the “from life” part consists not in the whole picture per se, but in the local adjustments added to the pre-existing model. In this interpretative rendering of Piaget: we can only see (deal with, represent) a little more than we already know to expect (understand, represent). Also relevant here is Part III of Gombrich’s book, “The Beholder’s Share”, regarding the role of ‘trained’ viewers.24 24

On p. 88 the word “schema” appears – in a related but not identical sense to our Piagetian one. Here, it is more literally the mind’s picture rather than the set of operations used by us/painters to produce external (to our minds) pictures. Also

66 Chapter 2 – Background and history He notes (see p. 90) that representations in art are, necessarily, only models of some reality – and thus necessarily representations of only selectively limited and abstracted aspects of that reality. Conventions provide the “mapping” rules for relating aspects of a representation to the exterior reality that it is understood to represent. In learning artistic conventions (see p. 100) we do not learn to “generalize” (i.e., to abstract general features from particular instances), but, instead, to find distinctions in what had before been only been an undifferentiated mass. That is, we learn ever greater specificity of concepts through our acquisition of increasingly individuating distinctions. This relates in part to the way in which Piagetian schemas home in through the adaptive process of accommodation and assimilation, but also echoes a Saussurean and Trubetzkoyian attitude toward the crucial role of contrast in determining structure. On p. 99 Gombrich shows a Piagetian-like emphasis on operations – vs. pure serendipity or intellectual creation. Thus, on p. 52, he addresses the nature of, the role played by (and the importance of) “constancy” in correcting in our minds for actually experienced (and, misleading) variation, e.g., of perspective, ambient light, etc. A strange perspective produces first a moment of shock, but then a period of adjustment as we adapt our understanding of relevant constancies to fit the new perspective. And, on p. 127, he calls attention to the importance of variation and change in the function of art – regarding both its “task” (what it is supposed to accomplish) and what (about the pictured subject) it was intended to portray, illustrate, or demonstrate. This observation comes in the context of Gombrich’s (p. 125) comparison of the function of Egyptian art as portraying timeless truth with the Greek (p. 129) concern with narration and thus change. He goes on (p. 145) to note the post Greek and Rome return, in the the East, to the Byzantine focus on timeless awe, and he compares these views with (p. 150) a Chinese focus on poetic evocation. He offers (p. 101) a suggestion of what one might call the “satisficing” nature of recognition – that is, we learn just whatever it is that is needed to distinguish a given something from what it is likely (based on past experience) to be confused with. His examples from nature in favor of this limited form of on pp. 100, 117 we see “schema” as a general cultural model, to be individuated (to the degree needed) for the task at hand. I want note that his use of the term “schema” is like that of many cognitive anthropologists (such as D’Andrade and Quinn – see Chapter 11, Section 2), and thus sharply contrasting with my own distinction of collective “models” from individual “schemas”.

Gombrich’s approach: Art and Illusion

67

recognition (as opposed to a more global form) include gulls taking rocks for eggs (in unusual situations) and Tinbergen’s male sticklebacks (little fish in England) confusing of inert objects such as fishing floats with sexually relevant stickleback others on the basis of the big red spots on the undersides (bellies) which, in the natural situation, signal the sexual relevance. His discussion on p. 121 of the idea of seeing “the conceptual, diagrammatic character of Egyptian images” not as primitive and not as the product of some strange Egyptian mentality, but as having to do with the function of the representations – what about the pictured people was being represented – seems insightful and important. As evidence he points (p. 112) to where the ancient Egyptians ignored their stylized conventions, and produced good (to our minds) “natural” portrayals – when the people being pictured were of too low a status to matter. On p.125, he gives us the Egyptian conventions: a stylized, “typical” (what part shown from what viewpoint) presentation of those people who mattered – for eternity, to be timeless (vs. momentary) truth – over the “illusionist” view of the moment. He notes on p.126 that the “timeless” Egyptian view was Plato’s preference. This Egyptian rendering of the timeless is contrasted with the (p. 134) Greek aim at “convincing” art. One part of the ancient Egyptian convention was the use of the size of a figure. They used relative size to show relative importance (p. 135). My observation is that a) at first the result of the convention looks strange (capricious, senseless) to us, but that b) we rapidly realize what is being shown, and then see it with no trouble. That is, our convention is to use the size of an object in a piece of art (or in a photograph) to represent a combination of the actual size of the object relative to other objects in the scene and the distance of the object from the viewer relative to that of other objects in the scene (i.e., to render three-dimensional depth). But we have no great trouble learning and accepting the Egyptian conception. The connection of size with importance is not unfamiliar to us (as in, re some politician, “He’s a big man” or “He casts a big shadow”) and we often are surprised when we get up close to an “important” person and find that person to be physically small. I think we would have much more trouble coming to grips with a convention in which bigger signalled less important – or one in which bigger signalled further away! Both the ancient Egyptian convention and our own convention regarding the size of figures in art seem rooted in the physiology of the eye and the ways in which we infer information from the image that hits our retinas – which is in turn rooted in the way the physical and social world we live in is constructed. There is nothing in the world which forces us to use size to represent either distance or importance, but

68 Chapter 2 – Background and history once we chose one or the other of these representations the meaning of “big” and “small” seem forced. These are examples of what I call “natural conventions”.

8. Schank & Abelson: Scripts, plans, goals, and understanding Schank and Abelson’s study (1977) remains the best example that I am aware of that actually works out and works through all of the minutiae that are involved in the pragmatics of an example of ordinary everyday conversation. While not based on their formalisms or their specific concepts, my approach has been much influenced by their simulation. Their task requires them to include all the knowledge and reasoning one needs to really “understand” what is going on in a conversation about some ordinary, but real, cultural activity. Their book’s primary conversational domain is a restaurant visit. In the restaurant universe, one needs both actual specific knowledge – e.g.: what is a restaurant; what happens there – and relevant implicit knowledge of the nature of the world – e.g.: that to eat you have to get food to your mouth, and thus that the food has to be within reach, and thus that there has to exist a mechanism for conveyance. The mechanism here entails a device to convey (fork, fingers, or whatever) and an actor (a person to mentally decide, who has the physical means – a free hand). In a similar vein, in a school universe, to “go to school” you have to arrive there and then to be there, but you don’t particularly need much of an internal representation of the school. On the other hand, to “solve the math problem” you do not need to be at any particular place, but you do need a representation of the problem. Their simulation makes much use of implicit “marking”(my term, not theirs)25 – that is, of presuppositions about default actions (in the absence of specific information) combined with the noting of exceptional events that require alternative actions and with the noting of exceptionally pivotal events that are needed to carry the story along (i.e., non-normal terminators of scripts). 25

The idea of marking comes from Trubetzkoy’s handling of “privative oppositions” in phonology (1969). Greenberg (e.g. 1966, and see 1968) much expanded and extended the idea. Marking is essentially the linguistic equivalent of the “default option” in computer science – an observation I first heard from Roy D’Andrade when I was a student.

Schank & Abelson: Scripts, plans, goals, and understanding

69

They distinguish scripts from plans. In their simulation, a script offers a precise description of a course of action for a well-known situation, while a plan provides a mechanism for reasoning from what is known (goals, capabilities, interdependencies, etc.) to a course of action in less-well known situations. Schank and Abelson treat the two as absolute alternatives, as opposed to the ends of some kind of continuum. They act as if such a dichotomization is basic to the human knowledge cum action that they are modeling, while I think the dichotomy is more an epiphenomenal artifact of their computer programming constraints. I think – and will discuss at length later in this volume – that our normal cognitive functioning involves both a continuum from script-like to plan-like, and various combinations of the two in different situations. My reasons involve the actual fluidity of common, everyday real life scripts (i.e., no two restaurant visits are quite the same – and so what we do can not be quite automatic), and the variable levels of the script which can be (are) raised to consciousness and altered at one time or another (either in response to restaurants that exhibit a strange mix of attributes or in response to unusual aspects of a given visit). In this particular situation I much prefer Robert Randall’s approach (described earlier) to the difference between how members of a culture handle relatively routine problems vs. relatively novel ones. But Randall did not attempt to construct an actual simulation, where the problem, of course, is that the flexibility we see in real life can be much harder to program. Thus programs always contain shortcuts – which is why we always have to be alive to the differences between our actual theories (or theoretical claims or assertions) and the programs that instantiate them. On p. 10 the relationship of Schank & Abelson’s ideas to schemas is made clear. Their programs are like our “schemas” (in the present volume) in the sense that they produce actual behavior; but they are like our “conventions” or “cultural models” in the sense that they represent general shared knowledge. Thus, Schank and Abelson do not make any distinction between collective representations and individual ones. This is another situation (as was the plans/scripts situation above) in which even a wonderfully powerful and insightful program still takes (and has to take) simplifying shortcuts vis-a-vis what actually happens in the modeled world. I would next like to direct attention to particular aspects of their simulation and discussion that are pertinent to the present volume. On p. 9 they make clear the importance for their simulation task of semantics and pragmatics, including the relevance of context – all implying knowledge of the concrete situation. That is, interpretation of a message depends on the situational and

70 Chapter 2 – Background and history conversational contexts, as well as on what is “known” (in the conversation) about the context. The simulation depends on the conversion of implicit native knowledge into explicit propositions (p. 11) – and thus shows the importance of such raising to consciousness for any effective cognitive modeling. “Conceptual Dependency Theory (pp. 12–15) comes in here, by which they refer to their analysis of the content of actions. On page 22 they show (discuss and illustrate) the problem of understanding normal text. A lot is normally left out of our normal descriptions of some situation – to be inferred (often, by us, unconsciously) from our knowledge of the situation. A simulation model needs to have the knowledge and procedures (explicitly in the program!) to make such inferences. They provide a formal characterization of the kinds of causal relations, causal chains, logically/physically necessary links, etc. that go into the program’s reasoning (p. 24 and following). On p. 34 they cover Tversky & Kahneman (see Kahneman, Slovic, and Tversky 1982) type predilections of people to make various kinds of false inferences – and the need to model these. The simulation relies upon episodic (as opposed, e.g., to semantic) memory (p. 17–19). (Note how this kind of information storage parallels what Dougherty & Keller (1985) found in their “taskonomy” analysis of the organization of a blacksmithy). The version in the program is propositional in the sense that definitions are functional – relating to actions. For a given situation Schank and Abelson provide, first, generalized propositions (note that “generalized” is still a kind of default, and still, as such, is specific to a [possibly only implied] context – i.e., in real life, we do play a very complex relativity game.). Second they provide a list of all specific propositions relating to specific instances. The “generalized” aspect includes two relevant specifics. First, the program’s knowledge of an action or situation includes a physical description of the normally encountered one (cf. “prototypic” referent in Kronenfeld 1996). Second, a notion of a script as a generalized standardized episode is built into the study. The distinction between basic general knowledge (which produces a plan) and specific knowledge (which produces a script) is introduced on p. 37. Scripts are top down processes, often unconscious and routine, while plans are bottom up, consciously reasoned, procedures (pp. 98–100). How repeated plans might become scripts is not dealt with. They offer examples of script based inferences – how we use what we know must have happened to fill in the story (pp. 37–42). On p. 42 they ad-

Schank & Abelson: Scripts, plans, goals, and understanding

71

dress the role specific nature of scripts – even if several different role-based scripts might feed into our (or the program’s) conception of the wider event. They offer a discussion of the important features of the restaurant script (p. 42). On p. 46 they address the hierarchy of script inclusions (in and/or of other scripts, plans, etc.), and address cues for script invocation (i.e., in their actual simulation two lines from the script must appear for the script to be invoked) vs. only a fleeting reference to the script. This discussion also contains examples of where they get too literal minded in their ascription of the relevance of their specific computational procedures to the human behavior being modeled. They address script headers – conceptualizations which invoke scripts: precondition headers, instrumental headers, locale headers, etc. – which invoke other information packages on p. 48. They consider interferences (obstacles and errors) and distractions – things which derail or alter a script, which signal (in conversation) noncompletion of the default path (p. 51). Script interactions – e.g., between train and restaurant scripts for a dining car are illustrated on p. 59. Types of scripts – personal role, attitude, function, etc. are introduced on p. 61. – and related to the inter-script interactions which must obtain in any specific event. Script based understanding – as necessarily the product of, and abstraction from, past experience – is discussed. Plans are described on p. 69, including the deduction of action needs from goals and the inferring of presumed goals from observed actions, contexts, and general knowledge. On pp. 73–74 we see that understanding implies ascertainment of relevant goals and then a method to be used to realize each goal. On p. 75 they introduce sets of possible actions, called “planboxes” – composed of a desired act, relevant preconditions (controllable, uncontrollable, mediating) for the act, and the likely result. The simulated actor creates planboxes and reasons with them more or less as with scripts (p. 90). On p. 74 they treat change-of-state goals (“D-goals”) as they relate to achieving standard sub-goals (starting on p. 83). These are things like “get control of”, “get close to”, etc. The planning stuff distinguishes physical actions from mental ones, and relates each to the task within which the reasoning is proceeding – as described above. In a discussion beginning on p. 131, they consider what lies behind goals, i.e., themes – the background knowledge that enable us to anticipate goals. Types include role themes, interpersonal themes, life themes, and so forth. On p. 151 They discuss levels of representation (cf. what – following George Miller (1956) – Kronenfeld [1996: Ch, 8] speaks of as “chunks” or

72 Chapter 2 – Background and history “chunking”). Conceptual Dependency is not always needed there; alternative (not so deep) knowledge structures are also treated. Schank and Abelson’s discussion on p. 160 deals with macroscopic vs. microscopic representation. They use only two levels, but this situation clearly can entail a whole nested hierarchy of levels. 9. Hutchins: Culture and Inference In order to be able to simulate some interesting cultural process – such as, in Hutchins’ (1980) case, the reasoning by which a traditional Trobriand court arrived at a decision concerning disputed ownership of some farmland – we need a great deal of specific cultural knowledge. But, with that knowledge, a simulation of the decision process is possible – which predicts the correct outcome, includes many of the intermediate statements of participants in the trial, and which gives us a rich sense of how land/farm ownership, interpersonal relations, and public reasoning work in the Trobriands. Hutchins’ work attacks traditional anthropological problems in new ways that are powerful, formal, rigorous, and effective – and shows the important relevance of cognition and cognitive processes to cultural constructions and intercultural variation. A. The relevant kinds of cultural knowledge include: 1. kinds of land/farm rights (intersection of these, after pruning of impossible types, gives 4 kinds of title to farm). a. ownership – land, by clan, unalienable b. use rights – individual has right to farm it c. allocation rights – right to say who farms it 2. kinds of farms, including a. kinds of farming routines b. sequences of farming actions c. normal reasons for doing each action 3. kinds of prestations/gifts – pokala a. katuyumali (return) – gift to user to return R-use to deceased R-alloc holder’s family b. katumata – gift by user to deceased R-alloc. holder’s family to keep R-use c. kaivatum – production of an exchange garden in yams – a gift designed to elicit a reciprocal gift d. yolova – a kind of loan – perhaps of use rights

Hutchins: Culture and Inference

73

4. kinds of ways rights are transferred a. kasali allocation of rights (in return for pokala) 5. the kinds of transactions that are likely to observed and or observable vs. not. 6. kinds of relationships that exist among people including kin, friends, contractual, etc. 7. kinds of rights linked to kinds of relatives a. inheritance b. exchanges with affines B. We need to know reasoning procedures – especially those embodied in the language: 1. logical connectives – strong implications of statements or actions in the local language a. kidamwa – “if” as in “if, then”, “as if”, and “in order to” – that is, as a kind of conditional b. mwada – “if only…” – a desired hypothetical; presupposes the desired entity not at time possessed c. kaina – “or the other” “or not” d. mitaga – a kind of negative intensifier, somewhat like “¿¡Como no!?” and somewhat like “or else” e. pela – “goal” “end” “reason for” f. uula – “basis” “source” “cause” g. – – – conjunction (“and”) is expressed in several ad hoc ways, but not through a special term 2. probabilistic inferences – default presumptions concerning what some action is likely to imply or entail, with whatever degrees of certainty measures that obtain. For example, say A dies, in apparent possession of a car that had been registered to B. A’s heir claims A bought the car, but hadn’t yet filed the title transfer. B’s heir (A and B are both dead) claims the car was only rented to A. We know, and all agree, A gave B $50. If the car was a 15 year old Chevy sedan, we might tend to believe A’s heir; on the other hand, if it was a year old Nissan Stanza, we’d be more likely to believe B’s folks. The issue is one of plausibility: B could have sold the Stanza (or leased the Chevy) for $50, but such seems unlikely (because the values are too far out of line) while the competing assertion of leasing the Stanza (or selling the Chevy) seems much more reasonable. On the other hand, if A’s heir and B’s heir both agreed the Stanza was sold for $50, and no one else posed a counter claim, we’d probably accept it – as certainly not impossible.

Chapter 3 Language to culture – building from Kronenfeld’s semantic theory

People’s individual representations of cultural phenomena are based on the collective representations that they impute to the various local (generally, face to face) groups, both explicit and implicit, that they belong to. In the case of language, the various groups can be seen as areal, social, occupational, ethnic, etc. dialects. It is people’s overlapping memberships, and the (non-dialect-distinctive) regularities that they carry over from one context to the other, that keep the wider system together. The fact that all people participate in a wide variety of such local groups means that no one is intrinsically a member of any one group and thus that no one’s self is constructed as a simple instantiation of any one such group or representation. All individuals “code switch” as they move from group to group, even if the code switching of monolinguals may seem less dramatic than that of multilinguals. Parts of a language (especially of a language spoken by members of a culture with a relatively complicated division of labor), such as a technical vocabulary, may be much less widely shared than is the language’s basic structure or its general vocabulary. Mass media facilitate sharing and overlap, but do not undermine the importance of local groups as the initial formative loci of language (and culture). Similarly, individual representations of non-linguistic culture (other shared cognitive systems) pertain most directly to local cultural groups, and are variably spread though the wider society. Functional factors which shape sharing include what people deal with in their daily lives and occupations, what social divisions are actively maintained (for whatever reason), what social and linguistic shadings people want to impart to their communications and actions, and so forth. In the semantic theory of Kronenfeld (1996) some other notions are developed which are relevant to a broader cognitive view of culture. These include – a particular view of the relationship between a focal prototypic referent of a category and various alternative extensions of the category to other referents,

Introduction

75

– a distinction of the semantic system of contrast and inclusion from the prototypic schematic “picture” (based on prototypic referents) to which the terms intrinsically refer (cf. Saussure’s “signifieds”), – a distinction of both semantic system and prototypic schematic “picture” from an independent schematic understanding of the specific target situation to which a term is being applied in a given conversation and context, – an active reliance on context for semantic interpretation with the recognition of sub-prototypes (and pertinent schematics) dependent on context, and – a generalized application of hierarchically structured marking theory defaults as “presuppositions” regarding referents, contexts, conversational goals, and so forth, with a use of discordant elements as triggers for application of relatively marked alternatives. Semantic extension is not of a term in isolation (say, “chair”), but of a term relative to the alternatives in the relevant contrast set (“chair” vs. “table”, “sofa”, etc.). In usage, the choice among alternative extension possibilities is not driven by some automatic rule, but rather by a more ad hoc determination about which alternative will best accomplish the given communicative aims – given, inter alia, the conversational context, the reference context, and presuppositions about what knowledge is shared by speaker and hearer. Cognitive schematic “pictures” (including ones of the represented world), form an important component of the Kronenfeld (1996) semantic theory, and they form a comparable component in the present theory of other shared cognitive systems, or systems of “collective representations”. Drawing on early insights of Saussure and Durkheim, Kronenfeld (1996) offered an interpretation of “collective representations” that distinguished them from individual representations or schemas and that explored their nature and the loci of their existence. This analytic perspective is concerned not only with how shared knowledge is learned and applied, but also with how it is created and how it adapts to the lives and needs of its holders. The understanding of the dynamic interaction between individual representations and putative collective representations and the resulting felt-to-be-real but empirically epiphenomenal status of collective representations that is sketched out in Kronenfeld (1996) offers a useful way of understanding other culturally shared, or collective cognitive phenomena. In particular, that theory offers a clear and non-mystical way of understanding and interpreting cultural cognitive constructs and allows a clear analytic distinction between such collective schema-like cultural models and the individual, personal schemas (however these may best be defined in cognitive psy-

76 Chapter 3 – Language to Culture chology) that actually drive our actual behavior. This perspective gives us a way of understanding the motivational force of cultural models – one that involves them not themselves directly producing behavior (or responding to or producing affective responses, intensity, and the like), but that has them providing potential (and potentially socially sanctioned – or otherwise interpreted) models for individual instantiation in the personal schemas which do produce behavior (and which, inter alia, relate affective states and other motivational factors to behaviors and to other affective output). The code switching among alternative local cultural communities offers an explanation of how individuals can behave in different contexts (or at different times) as if they had very different internal make-ups driving their actions. In this view cultural representations or systems of knowledge are, like language (and its parts), collective cognitive structures. These representations are often accessed by language (even named – when commonly communicated about), but they are not themselves linguistic entities (not equivalent to language and not part of language). They have direct representations in the mind which are not constructed out of language and which are often not, as representations, mediated by language. The evidence for this distinction between linguistic and cultural-behavioral cognitive constructs is based on prior research of mine (see Kronenfeld 2008: Chapter 8, and Kronenfeld 1975) in which was shown a clear lack of isomorphism between semantic kin categories and the categorization of kin implicit in a variety of behavioral and attitudinal conceptualizations – and in which, furthermore, the two kinds of categorization were shown to rely on very different kinds of structures. 1. Culture and society The underlying approach of this volume is a functional one – in the sense of worrying about why we do what we do. The starting point is “culture”: what it is and why it might be useful; what benefits make it worth its costs, and why it might take one form vs. another.26 Two salient things about us re other animals provide a starting place: we are social and we are smart. But our intelligence also involves doing things – and doing things collectively – the Durkheimian division of labor in society that we have spoken 26

This is not the “Functionalism” of days gone by. No “just so” stories, no arguments re functional necessity, etc. It is more of a perspective and an approach – to empirical questions looking for empirical solutions.

Culture and society

77

of. The division of labor requires sophisticated coordination – matching different individuals’ contributions, timing these appropriately (re when what are needed, what is their shelf life, what labor exists for maintenance, etc.), distributing motivation (rewards, punishment, etc.). Specialized knowledge and coordination are required. But knowledge comes with a cost (in nonproductive learning time, and maybe in limited storage capacity), so it seems inefficient to have everyone learn everything – while the possibility of unexpected glitches or dropouts necessitates some overlap and redundancy. Coordination requires both knowledge and some mechanism for getting the right “who” to do the right “what”; it can be authority or some sort of negotiation (or possibly something else). Going back to early ethnoscience, one perspective has been to consider culture to be the knowledge that it takes to behave appropriately in a given society. A presupposition has been that this cultural knowledge will generally be shared, even if many parts may be less well shared than others. It now seems useful to think about what is shared and what is not, and why such a mixture of sharing and not sharing might come about. One model that comes to mind is that of parallel distributed processing. Culture, from this perspective, represents the shared content which enables society; and, conversely, society represents the interactions which enable culture. It seems likely that this cultural knowledge (or much of it, at least) is constituted as some hierarchically structured array of cultural knowledge, as suggested in Frake’s “A Structural Description of Subanun ‘Religious Behavior’” (1964 – also see Colby’s work on cultural grammars [1973, 1975]) – involving models which are in various ways indexed by language. Bateson’s Naven (1958, and see Chapter 14 of the present volume) remains the classic model for how to integrate the roles of cognition and affect across collective understandings and individual understandings and behavior. Cultural knowledge then represents the shared knowledge structures of human societies, where the hierarchically organized levels of detail provide the differential degrees of sharing needed, with the knowledge specific to each particular group (level of social organization) tied to its contribution to the whole. At each level the culturally specific knowledge of that level group is tied to the special actions of that group and is built on the more general knowledge that is shared with higher level groups. As children – or other new members of a group – we learn the shared patterns of behavior that characterize members of that group and/or the knowledge that seems presumed by members, and we learn the kinds of internal

78 Chapter 3 – Language to Culture internal variation that characterize the group. Within that variation, depending on the size, nature, etc. of the group, we learn some of the variants better than others – some well enough to ourselves act them out, others well enough at least to properly interact with the other actors, and some just enough to file away. Such groups vary from ad hoc to stably long-term, from local to national and international, from informal to formal, from explicitly named to implicit, and so forth. Each of these groups embodies a kind of subcultural variant of the wider culture; some of the groups are part of a hierarchy of inclusion (e.g., neighborhood, city, state, nation) while others are not. This means that all of us, often and frequently, do a lot of “code switching” among at least subcultural variants. “Society”, in the social sciences, refers to patterns of grouping and interaction through which a collection of individuals constitute an extra-individual entity. The mere proximal juxtaposition of several individuals is not enough; the individuals have to behave in some manner such that their behaviors are interdependent and organized. There must, therefore, exist a means by which goals, emotional states, intended actions, and so forth are shared, and by which members of the given social unit and/or system are distinguished from non-members. For some biological species this needed information seems to be innately given, but for human societies the role of such innate bases is more minimal and less direct. “Culture” refers to the shared system of knowledge, feelings, and behavior (and, sometimes, their products) that characterizes one human community versus another. Culture, thus, is not directly innate (though it may well be responsive to various innate dispositions) but is learned. At the same time it is not explicitly or formally taught. We glean its regularities from our experience with others, and we each construct our representations of the shared system on the basis of our experience. Shared experiences plus exigencies of communication – whether direct or indirect (via the requirements of shared or interactive behavior) – keep the individual representations of members of any given community adequately similar. An important aspect of the sharing and communication is the requirement that individual representations be productive and apply flexibly to novel situations and contexts; such productivity implies some sort of axiomatic system through which behavioral outcomes are intensionally defined.27

27

Note that the claim of a need for well-defined behavioral outcomes is not equivalent to a claim that particular behavior events can or should be predictable. The issue is one of systematicity and interpretability, not detailed predictability.

Culture and society

79

Either included within culture, or standing as a major parallel learned system, is language. Language is in many ways tightly structured and susceptible to rigorous formal and even mathematical characterization. Numerous attempts have been made to pattern understandings and representations of culture on the properties of language. The problem is that culture – taken as a whole – is nowhere nearly as tightly structured, but is much more, as we have seen, Lowie’s “thing of shreds and patches” – or a kind of patchwork quilt. Like the pieces in a patchwork quilt, parts of culture (such as kinship, ethnobotany, and – most extremely – language) are tightly and coherently organized, while other parts seem less so; and like a patchwork quilt, the whole has a shape and form but no overarching structure. Culture and society, then, are mutually constitutive: Culture provides the shared knowledge system which enables members of a society to recognize fellow members and to coordinate their actions with one another, while society provides the communities, and thus the patterned interactions and experiences, out of which individuals construct their representations of culture. Since culture (like language) is intrinsically social and only exists as a social device, it cannot be identical to what is in any single head, but has to consist of socially shared forms. But since culture has no existence outside of our individual representations of it, and since these representations are variable, there exists no single place where the whole of any culture is stored or represented. Thus culture is necessarily and intrinsically a distributed system. At the same time, culture’s various sub-systems – the various patches – consist not in directly memorized actual behaviors or fixed specific knowledge but in productive representations (based on flexible and adaptively growing knowledge systems) that are capable of generating novel responses to novel situations that are still systematic enough to be understood and appropriately responded to by others in the given social system. As indicated above, interpretable productivity requires a largely shared or equivalent formal system for defining behavioral entities (whether these entities be specific behaviors themselves or patterns, qualities, or aspects of behavior). The “largely” hedge allows for the kind of gradual slippage that we see socially between one group and another and diachronically from one generation to the next that will be discussed later. The “shared” or “equivalent” alternative is an indication that alternative formal systems are possible so long as their relevant output is largely the same. Such alternatives, in asymmetric situations, can be different enough to be complementary rather than similar. More complex and complexly interactive systems such as natural language constrain individuals’ internal representations more tightly than do looser systems.

80 Chapter 3 – Language to Culture We hold some of these logical systems (and/or their constituent axioms) self-consciously – as, for example, the rules of a ball game – but mostly – as with language itself – we do not. That is, many of our shared formal linguistic or cultural systems are necessarily “transparent” (meaning that they work sub-consciously, outside of our conscious awareness) in order that reasoning within them not get in the way of our more – often necessarily – conscious use of the resources of these ‘background’ systems to accomplish our specific social, communicative, or material goals in different contexts. I am referring here to the kind of interference that occurs, for example, when one is speaking a foreign language one does not know well. In this situation the attention one pays to knowing “how to express what I want to say” interferes significantly with focusing on the content of the intended communication – “what is it that I want to say”. The social system is similarly complex. We neither belong only to a few social groups nor only to fixed ones – but to many groups that range in size from a few to millions of people, that range in durability from evanescent to centuries long, that range in awareness from totally subconscious to publicly and legally inscribed. The groups include occupational groups, ethnic groups, recreational groups, neighborhood groups, etc. Many of these groups participate in hierarchical levels of inclusion (as where a neighborhood is part of a town that is part of a region). We share many of our memberships with those around us but we share all of them with no one and no two of us share the same, precise set of memberships. The preceding suggests a mode of cognitive functioning that is oriented toward rapid leaps (versus careful, inductive generalization) into tentative generalizations (that is, categorizations, including the kinds of patterned relationships that we commonly speak of as “rules”) – and rapid reassessment of generalizations that do not appear to work. The generalizations are part of a process of feedback and construction of the sort outlined by, among others, Piaget (see Flavell [1963: Chapters 2–7] for an overview of Piaget’s explicitly mathematical system).28 Work in cognitive psychology concerning information processing has offered insights into how this process works. 28

Following Piaget, I mean that the cognitive systems produced by these generalizations are intrinsically mathematical. It follows then, in regard to their study and description, that these generalizations represent systems that are in principle mathematically model-able, and that when less well analytically understood are still susceptible to computational modeling – where experiments on computational models can help in the process of refinement which leads to better mathematical models.

Culture and society

81

Many of these generalizations relate to the physical world, but others have to do with the social world of individuals and groups of individuals. We seem disposed to distinguish, inter alia, between animate or willful beings and other things. In understanding the behavior of animate beings we refer to our own motivations and linkages and to what would produce the same behavior in us. Prototypically, this reference pertains to other humans, but I suggest it also extends to other animate objects – animals, cartoon characters (as in the dish that is running away with the spoon), and to apparently capricious machines. I suggest also that this self-referenced understanding forms our basis for understanding the behavior of collectivities or groups; we think of them as if they were individuals and assess their behavior in terms of motives, instrumentalities, etc. accordingly. The claim being made here is not simply that emergent properties of one kind or another come out of the cumulated and interactive behavior of individual agents, but more significantly that emergent systems – with their own systemic properties – are produced. These systems neither are just the product of the repeated behavior of individuals nor do they simply arise by adding feedback to relations among individuals with repeated behavior. To understand these emergent systems we must include, along with repeated, interactive behavior of agents and feedback relations among agents, a propensity of individual agents to infer structure (systemic properties) from the behavior of other agents or categories of agents and to structure their own acts in terms of that inferred structure, and then to continue to adjust their inferred structures on the basis of feedback from members of the categories. This “constructivist” approach comes out of various branches of cognitive psychology, especially (for the present case) Piaget, but is extended to the social world via the assumption that categories of people are perceived and treated as if they were individual entities – the psychological equivalent of what corporations are economically and legally.29 Shared knowledge is essential to this view of culture – not a single monolithic sharing but a whole set of interacting hierarchies of relative sharing. 29

This approach to culture and society, while itself reductionist, does contrast in important ways with traditional reductionist views such as those of George C. Homans (e.g., 1950) or Fredrik Barth (e.g., 1992). The difference lies in the constructivist approach to emergent structures. The same difference distinguishes this approach from the kinds of neural network models now in vogue – even if the kind of connections they focus on seem necessarily a significant part of the wider picture.

82 Chapter 3 – Language to Culture Important is not just the sharing itself, but the limits of sharing – and the distinction between the differentiation (among individuals) that depends on prior sharing and the differentiation that has no such dependency and that thus instead signals (at least relatively) membership in different worlds.

Chapter 4 Culture as distributed cognition Unpacking the idea of “Culture as Distributed Cognition”, is perhaps a good place to begin. What does it mean and why is it important? This approach provides a useful way of thinking about decentralization in the context of complex human institutions – and thus to our understanding of the very nature and function of culture itself. 1. Representations – collective and individual In this book the term “representations” will appear as both “individual representations” and “collective representations” and so a brief explanation of the usage might be useful. The idea of “collective representations” is taken from Durkheim – and I think this book’s usage and understanding are consistent with (implicit in) what he says. But, the ideas on representations that are discussed here seem useful and important – regardless of whether or not I am right about what Durkheim meant. I am not offering any exegesis of Durkheim’s thought or usage (see Chapter 2), and if anyone does not accept my ascription to him, then I am prepared to accept responsibility myself for the thought and usage offered here. A “representation” is an understood mental model of something. A representation can be a model of other representations. In particular we are concerned, here, with models of aspects of the world we live in – including social aspects. The default representation is an “individual representation”. This is, normally, a person’s internal mental model of something. Such a model may well not be a “picture” – what forms are possible seems still open to debate and exploration. And it probably is the case that such representations are not in themselves complete, but depend on information pulled out of the modeler’s environmental context. Note that individuals can have in their minds models of the understandings (that is, individual representations) of various of their fellows. A “collective representation” is a shared (for our purposes, cultural) mental model. Since people do not share minds, collective representations cannot literally exist, but, insofar as individuals have in their minds individual

84 Chapter 4 – Culture as distributed cognition representations of presumed or posited collective representations, and insofar as these individual representations converge, then we can speak of the set of converged representations as an apparent (and maybe effective) collective representation. Two related kinds of convergence are involved. There exist both a social convergence among different individuals – as they approach increasingly similar representations – and a conceptual convergence within each individual – as the internal representation in question held by each converges on an ever cleaner and more abstract version. Culture, in one sense, can be seen as the set of asymptotes toward which the latter convergence moves. Thus, the only actual locus of such collective representations is in our individual representations of them. There exists no separate repository of cultural or linguistic knowledge outside of the minds of members of the relevant cultural or linguistic communities. At the same time, however, a) we each, as individuals, belong equally to other different communities; b) in constructing our own individual schemas or behavior, we each add our own individual detail to what we perceive as the relevant general cultural norms or forms; and, c) importantly, we each are aware of our own selves (including desires, perceptions, feelings, etc.) existing apart from and in contradistinction to these communities – the communities to which we impute the norms and forms of language and culture. And, as is the case with langue, other collective representations (here, those of culture) cannot be perceived directly either by natives or by ethnographers, but can only be inferred from usage. That is, representations of them are constructed from the shared elements seen in the behavior (including verbal behavior) of members of the given group, and from the logic of the system these elements imply. These representations are constructed on the basis of the individuals’ conversations and interactions with others in their various communities. They seem drawn as prototypic inferences from common and recurrent elements of the set of conversations and interactions one experiences. Representations do not seem limited to humans, and thus do not seem necessarily tied to language. That is, other social mammals and birds, but especially social carnivores, behave in ways toward other group members that are individuated enough with regard to the relevant “alter” to suggest that they each have separate understandings (or behavioral conceptions) of the various members of the group – that is, individual representations of these other individuals. Such understandings would have to be learned since the array of personalities and dispositions exhibited by various alters can be

Representations – collective and individual

85

quite variable, and seems not easily predictable from any obvious context or role. There is at least some evidence, as well, that individuals (at least in some canids, primates, and bird species) are able to distinguish members of one group (pack, troop, or flock) from members of other and to learn behaviors appropriate to the given group – suggesting at least some minimal collective representations. Representations, including collective representations, then, do not seem necessarily mediated by language, even if human language becomes a powerful tool for building and shaping them. Even where used to summarize cultural action, language is only the “transparent” medium by which a posited reality is presented, not the essence of that reality.30 Remember (from Kronenfeld 1996), words are vehicles, to be seen through. So, the presence of a verbal abstraction presupposes the existence of a non-verbal something – even if the presupposed something may not in fact actually exist. Since, in the case of a cultural model, the referenced (presupposed) something is a collective representation, it does NOT in fact exist. Instead, people act as if it exists, and share some common understanding of what the “it” is – based on their public interactive experience with instantiations of “it” or with references to “it”.

2. Culture From the perspective of the anthropological tradition I come out of, “culture” refers primarily to knowledge – all the stuff that we know which enables us to act appropriately within a community, to construct the tools, objects, designs, etc. of a community, to recognize the environment experienced by members of a community, to read and manipulate others within the 30

I want also to note, as I pointed out in Plastic Glasses and Church Fathers (1996), that the nature of natural language leads often to complex and internally inconsistent sets of referents for important or commonly used words. The combination of the frequency it takes to maintain a word in active vocabulary, our often appearing need to use existing vocabulary to speak of novel referents, our reliance on context to disambiguate reference, and the subsequent stabilization in a speech community of some of these secondary referents means that the words which we use to express our thought can, in isolation, easily (and often) have only a loose and indeterminate relationship to the thoughts they are used to express. This observation means that even in the case of verbally coded cultural models, an analysis based too closely on the words used to express the model is seriously at risk.

86 Chapter 4 – Culture as distributed cognition community, and so forth31. Furthermore, this knowledge is primarily stuff that we have learned without having been overtly or explicitly taught. It is knowledge that is to varying degrees systematic – in some cases impressively so – but where we have largely not ever been taught the relevant systems, and where, mostly, we are not even consciously aware of the systems. The proof of the systematicity lies in the susceptibility of parts of culture to tight formal modeling. The varying degrees with which the various parts of culture can be formally modeled, however, and the variability with which different parts are shared, suggest that culture itself does not form a tight system, but rather is a looser congeries of components of various sorts. Culture, in this view, is intrinsically social – as represented by constant references to “community”; that is, it does not consist of what you know or what I know, but on what we as a group (within one community or another) not only know but rely (whether implicitly or explicitly) on others knowing. The fact that the stuff we are talking about varies from one community to another means that it cannot, at least in any simple or direct sense, be innate. I do not mean to say that there are not biological foundations to our knowledge and to our learning, but only that the actual known systems are often substantially and significantly variable from one community to another – in ways that have mostly proved hard to deduce or simulate from any posited biological base. From this perspective, then, I see no opposition between biology and culture, but claim that our culture is rooted in our biology – a topic to which I shall return later. If the knowledge is not innate, it has to be learned. If it is complex, systematic, and flexible, then it cannot be learned in any simple, rote memorization fashion, but has to be learned as a productive system – that is, as something that can be represented as a relatively small set of axiomatic basics from which one or the other of the given systems can be deduced (however these systems actually be formed in the “wetware”). The problem, if one does not believe in some sort of superorganic collective mind, as I do not (!), is that the collective system can never be directly observed, either by “natives” or by social scientists; it can only be seen, indirectly, in the internal representation of it that each one of us has formed – or, rather, even more indirectly, in the products of our separate internal representations. 31

Included also, though possibly less relevant to our present concerns, are similarly learned, shared, non-innate emotional aspects of our interaction, including, inter alia, our ability to read the affect and emotions of others, to adequately convey our own emotional states, and to properly tune our emotional states to a given situation.

Culture

87

Since we, each and all, belong to many different communities which overlap with one another to varying degrees in membership, purpose, and context of relevance, and since in any given situation a variety of these communities may to varying degrees come into play, we – whether as social scientists or as “natives” – have no automatic way of linking an observed action or behavior to any particular culturally relevant community or reference frame32. Our overlapping community membership, further, means that none of us is simply or intrinsically a member of – representative of or, more importantly, an instantiation of – some particular or specific cultural or culture-bearing community. This dense overlapping and intergrading of actual cultural-bearing social units joined with the differing social distributions of various parts of the congeries of subsystems that make up culture, together suggest that we should not expect to find clearly bounded large scale cultural units. This suggestion is supported by recent cross-cultural comparative work (e.g., De Munck 2000, Gatewood 2000, 2001) where it has been found that there exist no all purpose general units for cultural comparison (and, thus, that comparisons, to be informative and valid, have to be based on problem-specific units). Given the learned nature of culture and the absence of any direct experience of it by learners or bearers, we have to conclude that it is inferred 33 – pretty much separately and individually by each of us – from our experience observing and interacting with those around us. Our understanding of the different communities to which each experience links is similarly inferred – from patterns of repetition, from our growing sense of what fits together and what doesn’t, and from our developing sense of the logic that drives events within each community. This last is also to say that our native understanding of culture is always a work in progress – never complete, finished, and closed. Note that if culture itself only exists in our separate representations of it, and if these representations depend on our experience and are changing, then culture becomes only a kind of epiphenomenon. Its reality – and its 32

33

“Reference frame” represents the observation that even within any given community an action entails presuppositions concerning the context that an actor is responding to, how the situation is understood, the role that the actor is filling, other communities whose members may be “watching” or “participating in” the given situation, and so forth. By “inferred” I do not refer to anything like formal inductive logic, but instead to a rougher process of successive approximations (of the sort described in Piagetian psychology), in which tentative understandings are tested against new experience, and, when failing, serve as the basis for intuitive leaps to new tentative understandings.

88 Chapter 4 – Culture as distributed cognition force – lies not in any kind of independent, objective existence but instead in, and only in, our acting as if it had such existence. What enables it to work, then, is the degree to which our separate representations of relevant communities and of the relevant cultural regularities that pertain to each match one another. This matching depends on the degree to which we have similar experiences and – usually with the aid of inter-individual feedback – draw similar inferences from them. Since the experiences from which we are known to make such inferences can clearly be seen to vary in a great many apparently irrelevant ways, the operative similarities in perception and cognition clearly must depend, in part, on shared modes of perception, cognition, and inference.

3. The issue is not “internalization” The preceding means that I think D’Andrade is wrong (1995: 227–229 [at the end of his book on cognitive anthropology], and also see Strauss and Quinn 1997) in speaking of the big remaining cognitive anthropology problem as one of “internalization”. I think the whole idea of seeing cultural values (or cultural forms, cultural models, etc.) as being internalized in individuals is wrong. Instead, I claim that, as with language itself, the process is the Piagetian one (see the Appendix in Kronenfeld 1996 for a fuller view of what Piagetian means in this context) in which individuals through their experience form internal representations of entities in the world they experience. These internal representations do NOT form part of the individuals’ core selves, as “internalization” would suggest. Just to be clear, I do not mean to deny that culture is learned nor that what is learned resides within the mind; I do mean, though, that such learned cultural content does not become part of the person’s inner “self”. Part of the evidence that they are not “internalized” is that people pick and choose among alternatives (sometimes associated with different, maybe contrasting, immediate communities), and use different ones for different purposes at different times. That is, cultural models are inferred, utilized and manipulated by the self – and hence seem unlikely to form any basic part of that self. Instead, they seem to represent a kind of reference library to which the self can turn for cognitive, emotional, and behavioral guidance. The similarities across individuals that we, as anthropologists, experience within a culture we are studying, and that lead us to speak of things like “cultural norms” or to describe culturally specific forms of behavior, do not come from some internalized core (which members of that culture share)

The issue is not “internalization”

89

but come instead from the fact that almost all members of the given culture have each experienced and interacted with similar sets of social “others” and all have, in parallel, arrived at similar representations of those others (whether individual “others” or collective ones) which they learn to apply in often similar ways – where these represented “others” include not just the entities but also the norms and values that purportedly guide their behavior, the mechanisms by which they influence outcomes, and so forth. Mine is a more communicative and interactive (you can’t have interactive without communicative) view of culture. Kids (and, decreasingly, adults) learn the cultural forms which they experience – perhaps inter alia – sometimes via responses to their own actions, sometimes via conversations about action or interaction, and sometimes non-verbally through the interactions themselves. They form internal representations of these models, and link these representations to the “others” that they experience them to be associated with – whether these “others” be individuals, social groups, cultural categories, collective mechanisms (as in “the stock market”), or whatever. They use these internal representations in conversations (as in Mukhopadhyay’s reference to the use of “cultural precedents”) but also similarly in reasoning to themselves re some action or situation. My claim is that neither culture nor cultural elements are inserted deeply into the self (see discussion in Chapter 6), but, instead, are external to the self, and thus available for use by the self as tools. But there are ways in which some forms can be deeply embedded in culture such that they seem psychologically deep. One way applies to concepts or orientations that seem to run pervasively through a culture. Such entities seem deeply embedded in the premises on which the given cultural system is built. This seems a kind of logical or conceptual depth. In a related vein, some of what we learn early (whether absolutely early as children or relatively early as when we first encounter something radically new), seems harder for us to un-learn, replace, or shake off than other stuff learned later. One example can be, for immigrants, the accent that goes with their childhood language. But it seems possible that some attitudes or behavioral patterns could also be examples. And, even when we are grown, our first serious encounter with, say, an academic discipline can be hard for us to shake – even when what we learn later about the discipline is somewhat discordant with that first learning. My sense is that, for much of language and culture (certainly the parts that are not explicitly taught), what we learn before we know the relevant

90 Chapter 4 – Culture as distributed cognition system functions as the stuff from which we infer the system. Such items (whether an accent, a word meaning, a cultural premise, a non-verbal concept, or whatever) acquire a kind of protean quality – which makes them hard to pin down and thus hard to supplant. But this kind of depth still seems to me to be a property of the given cultural or linguistic system rather than of the user’s psyche. What does sometimes make some of these basic forms of language or culture seem more personal is the trouble we sometimes have switching out of them – as with the childhood accent spoken of above. I think the answer here is that much of our learning aims at an economy of effort – which is to say that, if at all possible, we would rather learn new forms as variants of old ones rather than attempt to construct our representations of the new forms de novo. We learn the new forms well enough for them to function correctly, but not necessarily perfectly34 – especially where the new forms are part of a new system.

4. Culture and language This view of culture derives in important ways from the view of language first taught by Saussure (the founder of modern structural linguistics), though lost by many of the structuralists who followed him. It is a view that interacted importantly with Durkheimian sociology, possibly through the agency of the Indo-Europeanist linguist (and Saussure student) Antoine Meillet who participated in the Annee Sociologique group. It provides a particular perspective on how one might understand “collective representations” and how one might conceptualize the framework in which the “emergent properties” of society are seen to emerge. The analog to culture, in language, is grammar. In the case of tightly organized language, linguists can show that we behave as if we have in our heads a very complex set of rules which we follow (in some ways) with great rigor. But these rules, the code behind our messages by some accounts, are never directly experienced by speakers – especially speakers learning their native language. Instead, each of us infers our own representation of the linguistic regularities which we experience in the communication of others and which we use to construct our own speech acts. We try to mimic new forms, to generalize tentatively understood ones, and recast ones which don’t work; this process of construction is guided by some innate predispo34

A kind of satisficing solution.

Culture and language 91

sitions – but, to the minds of many of us, much less specific predispositions than are spoken of, for example, by Chomsky (e.g., Language and Mind [1972]). In any event, in language, an area of social experience that is important to us, we rapidly form rigorous representations of complex structures, where the structures involved vary impressively from one language community to another. From one perspective one might consider language to be just one particular part of culture – the particular system within the congeries that enables us to construct, understand, and interpret speech acts. The difference between language and culture is that language does form a system – a system tight enough to lend itself to mathematically precise formal characterizations – as opposed to culture’s looser congeries. The “congeries” notion is a way of saying that culture itself, in the whole, is not any tight or indissoluble system, but is instead a collection of parts that can to greater or lesser degrees be independently altered or exchanged for comparable parts from other systems – that is, in the words familiarly associated with Robert Lowie,35 culture is “a thing of shreds and patches”, a collection of separately structured pieces joined together to make a kind of whole fabric, as in a patchwork quilt. Language is not the only well structured system within the cultural fabric, though it is certainly the largest, most distinct, and most complex36; kinship systems, various folk systems of farming and trade, con35

36

Lowie’s reference was more to our civilization than to culture itself, but the common misunderstanding seems to have resonated – because, I think, it captured something that struck many anthropologists as insightful. The phrase was from Gilbert and Sullivan – who perhaps had a similar passage from Hamlet in mind. Because of the clearly systematically patterned nature of language, language and its study (linguistics) have long provided a model for how to do social science and how to make it scientific. The linguistic model dates at least from the PreDarwin days of early Indo-European studies when the history of the development of Indo-European and the methods by which linguists described and validated that history provided a model for the study, presentation, and understanding of the history of development (even “evolution”) of cultural forms – as seen, for example, in Lewis Henry Morgan’s seminal 19th Century work on kinship systems. Durkheim accorded a similar pride of place to language – as a a par excellence instance of a collective representation. More recently first Structural Linguistics (mostly the Bloomfieldian version) and then Transformational Generative Linguistics have played a similar role, particularly in anthropology. However, the culture-language difference is about more than just some greater looseness or ambiguity within cultural systems than within linguistic ones

92 Chapter 4 – Culture as distributed cognition ventions of highway driving behavior, and aspects of religious systems represent some others. The Durkheimian perspective suggests further that the social nature of language is not simply a matter of sharing but is something deeper and more systemically important. That is, if culture represents the knowledge of how to make or do things that are important to us, then we have also to consider the fact that we do not all have the same knowledge. There exists, if you will, a division of intellectual labor in society; we each know some different things from what those around us know – and these differences are both systematic and useful. At the same time, sharing a culture means that we, within a culture, also have much shared knowledge – particularly shared knowledge that links and contextualizes our different pieces of individual knowledge, and that allows us, when it is necessary or desirable, to substitute for one another. We also can see from this perspective that the degree of our sharing of knowledge with any specific other, the degree of (direct) complementarity linking our knowledge with the different knowledge of any specific other, and the degree to which our knowledge is irrelevant to the knowledge of any specific other are all variable – and relative to the connections among us in society. And, note then that it is not the sharing (or non-sharing) per se which ties together our separate individual knowledge into a single cultural system; it is the pattern of inter-relatedness and complementarity37 linking the knowledge of individuals in a social system which makes for a cultural system.

37

(though that surely be present); it is also about important functional differences between culture and language, and important differences between the two regarding how they interact with the phenomenal world. The differences in functional task and functional constraints have often not been sufficiently attended to. Anthropologists often speak of culture as if its being shared were the major issue – which would make culture an instance of Durkheimian mechanical solidarity. But in doing so they are only foregrounding the parts which most impressed them by being most obviously distinctive, and even here they often provide modifiers or hedges (such as limitations relating to age and/or sex, conceptual or action context, goal, and so forth). The problem of how to conceptualize culture has been around as long as has anthropology (cf. Kroeber and Kluckhohn 1952), and this particular aspect of it was foregrounded by Anthony F. C. Wallace (1970) in terms echoing Durkheim’s mechanical and organic solidarity – as a contrast between seeing culture as a “replication of uniformity” and as an “organization of diversity”. But, without any clear idea of what culture did or how it did it, it was hard to find any motivation for picking one conception over another.

Culture and language 93

It is culture thus construed that should be considered as a system of distributed cognition, and that, more particularly, I now want to consider in terms of issues that come out of the construction of parallel distributed processing systems in computer science. And, I want further to suggest, that it is this understanding of culture and its role which will provide a useful analytic perspective on the decentralized state of complex societies. Models of such distributed knowledge and action systems furthermore enable us to explore kinds of phenomena such as information cascades, mis-attributions of causes or agency, over-corrections, and so forth, which sometimes characterize human organizations. 5. Parallel distributed processing and decentralization My appreciation of the power offered by this perspective came when I was looking at research profiles in our local computer science department and realized that one big problem area concerned linking a bunch of small computers together in order that they might effectively simulate a big (or “super”) computer. That is, when a bunch of little computers are tied together and told to act as if they are one big computer, the big trick is how to program them. Particular problems involved the division of the overall problem into subtasks and the efficient allocation of subtasks to the various component computers, the efficient sharing of needed (vs. unneeded) data, and the efficient real-time coordination of activities of component computers that were working on different parts of a single task. One potential approach to this problem area was a kind of centralized, top-down one in which either the human programmer or one (“master”, if you will) computer did the problem analysis, task allocation, and scheduling Similar discussions have concerned the conceptualization of culture as adaptive – whether in the specific sense of one culture being adapted better than another to some particular circumstances or in the general sense of culture itself (or the human capacity for culture) representing a more generic human adaptive response. Of the problem of adaptation, more later. Part of the goal of my larger project in this book is to provide an understanding of the function and functioning of culture (not “Function” in the well and properly discredited sense of – in anthropology, Radcliffe-Brownian – Functionalism or functional prerequisites) that allows motivated, systemic answers to these questions. At the same time I want to offer that my understanding does square well with the sense of culture (and its attributes) that anthropologists have pretty much intuitively worked off of from the time of the emergence of the modern conception of socio-cultural anthropology.

94 Chapter 4 – Culture as distributed cognition for the other (shall we say, “slave”) computers. This approach apparently produced the same kinds of difficulties in the world of task cooperation among networked computers that it produced in the world of economic planning in the former Soviet bloc: no one was able to adequately anticipate all the possible downstream glitches that might come up – problems with real-time coordination, with unexpected hitches and unforeseen problems, with changing environments and resulting constraints and demands, and so forth – and the system was too ponderous and top-heavy to be very good at responding to unanticipated delays, failures to produce, and the like. An alternative approach, the one being pursued by my computer science colleagues at the time, was a decentralized one – in which a central machine did a basic break-down and share-out of the problem, and farmed out subsequent analysis and planning down the line to other machines. Then the various networked machines communicated among themselves in subtask relevant groupings, and worked out their own task sharing and temporal coordination on an ongoing dynamic basis. This organizational basis allowed them to respond flexibly to unanticipated glitches, to things getting done ahead of or behind time, to machines breaking down, and so forth. The down side of this approach – which is to say the research problem that it created – concerned how to keep the various separate machines coherently working together on the ultimate, common problem. This was not any problem of motivation, incidentally, since these were, after all, only dumb machines! It was the problem of making sure that the separate machines each knew enough about each other’s work, operative status, and progress, and enough about the overall problem and solution plan, to enable them to negotiate effectively and thus to jointly work out task coordination in real time. This decentralized coordination problem, then, amounted in part to a kind of overhead problem; each machine had to dedicate a certain amount of its resources to the coordination task (removing these resources from direct application to the substantive problem being attacked. Computer scientists had to figure out how (and then how best) to organize and structure this information sharing, including, for example, “who” needed to know what when, what should be stored vs. recalculated as needed, and so forth. Let us stop the computer story at this point, since we have enough for our culture purposes, and since the computer specifics are getting over my head and beyond my memory. PDP models have been applied in cognitive psychology to individual cognition – see, e.g., Rumelhart et al.’s Parallel Distributed Processing… (1986 – and see Perry 2003: 208–209, and Gatewood, in Press) – but not so much to socially distributed cognition. See Perry (2003) for a nice overview

Parallel distributed processing and decentralization

95

and discussion of PDP and related areas in the context of human computer interaction; his perspective is based primarily on that of Hutchins (1995). I aim at no literal application of either the computer model or the psychological one, but rather a consideration from that perspective of the problems posed by human social life and the role of culture in that social life. Since at least Durkheim we have known that the division of labor was an important aspect of society, and that increasingly complex societies were characterized by increasingly complex divisions of labor. This division depends, obviously, on different individuals carrying out different tasks. The total knowledge required for the carrying out of all the tasks, maybe even in a simple society, seems greater than what any one person is likely to possess, and is clearly greater than what many of those doing various specific tasks (e.g., minimally, new members, youths, etc.) can know. Additionally, many tasks are complex or spread-out enough to require the simultaneous participation of several individuals (as in Hutchins’ Cognition in the Wild [1995] examples). In simple societies, small and shifting group membership can mean that what experts that exist may not be at the right place at the right time while in larger and more complex societies the total knowledge rapidly becomes too big and locally specific for any one person. This situation approximates what we just saw for PDP computing. The suggestion is that culture (as generally understood in anthropology) can be seen as the overhead knowledge I suggest that each of us is a separate thinker (analogous to the separate computers in our simulation of a supercomputer), that unlike Hofstadter’s centralized anthill (see below) we have taken the decentralized route, and that culture or cultural knowledge is the system we have evolved to deal with the “overhead” problem (where “knowledge” involves not just facts or skills but also affect, motivation, the ability to manipulate and read other people, and so forth) – providing and organizing the “overhead” that enables our cooperation within society. That is, culture provides us with the information that we need to be able to play our role in the larger social enterprise – society. Culture, thus understood, involves a range of sharing of knowledge from very general down to very locally detailed. Differing levels of shared knowledge enable differing levels of subgoals and subtasks. In a rough general sense, we each know some stuff – what we are actually working on – really well and in detail and other stuff – what allows us to communicate and coordinate with others – in varyingly less detail. Since there is/was no computer scientist designing the cultural system, it is not as narrowly goal directed as is the supercomputer simulation. On the other hand, it is a system designed (at the systemic properties level) by Dar-

96 Chapter 4 – Culture as distributed cognition winian natural selection, and thus has been shaped by the usual Darwinian constraints – mostly differential reproduction of those having one form re those having another. This cultural system is one that allows us considerable plasticity or flexibility in the solution of particular problems. In such a framework knowledge seems likely to be hierarchically distributed – where everyone shares some basic stuff, large subsets share within them a little more, and successively small subsets share successively more. The shared aspects of culture, at various levels of inclusion, provide the common understanding that enables our general coexistence. The more specific our subgroupings the more specific is the relatively specialized knowledge shared by members of that group primarily to enable coordination of their specific group activities but also to enable recognition of group members in non-group-specific situations. Within the group there will tend to be some differential elaboration of knowledge (and vocabulary) specific to people’s specific tasks or roles, but there will also be the kinds of overlap of experience and communication that enable people to slide over into new tasks or roles. Cultural knowledge in this context will often be linguistically coded, but often also will include significant non-linguistic knowledge38 such as how to do things or how to present oneself. This “decentralization” aspect of culture obviously does not mean that all of the social structures we create via it are equally decentralized; the reference to the Soviet bloc gives the lie to that. And the computer model should not be taken as indicating that we are totally flexible or creative in our decentralized planning. Culture, especially taken in the context of how we have seen that we learn culture, does entail a significant default role for tradition – the patterns that we have already experienced as working. The combination of decentralization and tradition does mean that we are particularly geared toward relatively ad hoc adaptations of prior plans, and that successful adaptations tend to get integrated into future plans (tradition evolves). The computer model as well, as role of culture re the creation and structuring of social institutions, suggests furthermore that there is no contradiction or incompatibility between decentralization and a complex society. In fact, the impression one gets from this line of thought is that the more complex the social system the greater is the need for the kind of flexibility represented by decentralization. But for those of us who sometimes might be reformers decentralization poses the same problem that it poses for wouldbe autocrats – it is hard to get everyone’s attention, hard to get everyone to take the same perspective, and hard to introduce far-reaching change. 38

As in Piaget’s “sensory-motor” knowledge.

The role of culture

97

6. The role of culture Having seen what culture is, we now turn to what it does and how it does it. Human society, indeed mammalian society in general, is much different from the anthill model of society posed by Douglas Hofstadter in Gödel, Escher, Bach: an Eternal Golden Braid. Hofstadter’s ants all share a single, largely genetically structured set of goals and a single plan. By contrast, human knowledge and behavior, obviously, are not so tightly governed by genetic structures. Sometimes culture has been proposed as the human replacement for the genetic plan of the insects – at least in discussions of cultures other than the proposer’s own. But this equation misses a further, essential difference – which seems readily apparent to us as we contemplate our own culture: members of human societies do not all share a single set of goals in support of which they (we) all selflessly strive; instead we each strive for our own individual goals, whether these be selfless or selfish. We strive vis-a-vis other individuals, but also vis-a-vis the various collectivities to which we and others belong. However, this is not any Hobbesian war of all against all. As individuals we can strive to cooperate with one another, to submerge our individual identities and goals in some wider social construction – or we can strive to compete and exalt the self. We have the potential and the ability not just to address our individual perceptions and sense of self to our own individual ends, but also, if we choose, to address them to the task of understanding or empathizing with others. I do see culture as providing the system which coordinates human society, but I see this coordination as being of a very different sort than that suggested by the anthill model. It does not provide us with a set of instructions or “thou shalt”s, but instead provides us with a tool kit or vocabulary of goal instantiations, categories of objects and actions, means toward ends, ways of conveying or reading motives, attitudes, desires, and the like, schematic “pictures” or story lines that link these, and so forth. Each individual one of us uses this cultural knowledge, or “capital”, not just to attain our individual ends, but to flesh out our concrete instantiations, to assess the attitudes and responses of other people, to negotiate collective goals, to induce cooperation, to fend off resistance, to get emotional support and validation, and so forth39. What I am underlining, in part, is the distributed, inter39

Culture, more deeply, but in ways that I am less clear about, also provides us with the material out of which we construct our self-conception – our individual sense of who we each are. Presumably this construction is driven by some intrinsic (innate?, structured by some combination of genetics and developmental

98 Chapter 4 – Culture as distributed cognition active nature of this organization vs. the kind of centralized, top-down nature of Hofstadter’s anthill organization. Since we all participate in various different cultures and/or subcultures, we also pick among these alternative systems (in linguistic terms, “code switch”) according to which provides the best mix – in the given context – of serving the purpose at hand, being credible to relevant other people, conducing the most useful attitudes among relevant others, providing the best self-image, and so forth. The point is that our cooperation as well as our competition, our selflessness as well as our selfishness, are products of the separate individual consciousness and the separate individual goals, decisions, and actions of each of us. Culture does not provide us with some sort of super-organic individual. And none of us seems the least tempted to see it as such when we look around us at our own community (whichever that may be). However, the problem of interpreting culture as a kind of super-organic individual does arise because many of us, anthropologists in particular, have often written about other peoples as if their culture consisted, in part, at least, of a set of rules or norms – or habits – which they automatically and unthinkingly (mindlessly and will-lessly) followed – until they, or their culture, were corrupted by some outside (modern) influence. This fallacy of the “dead hand of culture” is important and dangerous, wherever it exists, both because it dehumanizes those others whom we seek to understand and because, by dehumanizing them, it prevents us from accurately understanding the shape of those people’s cultural forms, the problems to which the cultural forms are addressed, and the use members of the culture make of the forms. I do want to suggest, though, that this tendency we anthropologists sometimes have to think of other societies and cultures as some kind of superorganic individual is not totally strange or inexplicable. I think this tendency to ascribe not just imperatives to culture, but also goals and personalities, comes not from some social scientific (or humanistic, for that matter) analytic insight but, instead, from the fact that we anthropologists are humans, and share some, perhaps innate, human dispositions – including a tendency to reify society into a kind of organic thing. That is, part of the way we as individuals deal with social and cultural forms (including norms and rules) seems to be to treat communities (societies and sub-societies) as if they interactions?) predispositions, but this consideration is beyond both our present scope and my present analytic progress.

The role of culture

99

were individuals – in the sense of having understandings, goals, knowledge, and so forth. We each, as individuals, know that we ourselves are not part of any such anthropomorphized social beings, but we seem often to conceptualize social entities – even ones with which we ourselves as individuals interact – as if they are such. Thus we talk about the stock market not having confidence in the recovery, or the electorate being unhappy with the president’s performance, and so forth. These are, of course, merely figures of speech – as opposed to any beliefs we actually hold. But I do want to suggest that, as figures of speech, they do come out of a mode of thought that comes naturally and easily to us. We do not, typically, use such expressions as shorthands for some statistical assessment – that is, we do not question the proportion of investors (or analysts or brokers) who doubt the recovery and we do not query the variance in beliefs around these posited central tendencies. We social scientists also know, from the work of cognitively oriented social scientists such as Simon or Kahneman and Tversky, that ordinary human intuitive judgments are particularly bad at dealing with statistical issues, and that, instead of thinking statistically (even about clearly statistical problems), we humans (emphatically including trained social scientists!) tend to leap to clear categorical conceptualizations (even when these clearly fly in the face of other knowledge we have about the given problem). Thus, it comes naturally to us, as humans, to think of society (or culture) as a thing – rather than as a statistical distribution of individuals (or of knowledge and beliefs). Our sense of cultural norms, then, would seem to come not from our examination of our own beliefs, but from our reified sense of the collective social other.40

40

This shared sense of the beliefs of the reified other is the cultural norm (within some given social and cultural context), I suspect, that Romney and his associates have found a reasonably objective way to identify. They have worked out a way to evaluate individual representations of cultural knowledge against the standard implicit in the collectivity, and thus to ascertain the cultural centrality of various individuals. I presume it is normal modes of communication that cause individuals to converge (the same thing that shapes up our individual representation of the system of language), and that individuals who have more relevant interactions with relevant other individuals are more likely to converge – and emerge as Romney’s more culturally central individuals.

100 Chapter 4 – Culture as distributed cognition 7. Biological bases, mammalian sociability, and the social origins of human intelligence I would like to conclude this discussion of the nature of society and culture with a brief excursis into biology – first into the social systems of social mammals (and perhaps social birds) and then into biological shaping of modern humans. I want to point out that the general model offered above for human as opposed to ant hill society applies also to many if not all social mammals – if we leave out the information compression and preservation that is enabled by language. That is, social mammals exhibit the same individuality and the same idea of a social structure that emerges from a contending diversity of individuals. Both social apes and social carnivores such as wild canids, at least, show the same ability to work out a collective plan on the run and to coordinate their individual roles flexibly as situations change. They appear to have some notion of social roles and some inclination to scheme based on their current role and some ability to play other roles as the opportunity arises. A lot of what we consider intelligence in animals seems to pertain particularly to social animals and relate to social coordination among them. Some of these animals seem also to exhibit something like culture, although in the absence of language such culture remains very simple and minimal. One current theory regarding the emergence of modern humans starts with the supposition that our intelligence (however it be defined) together with the communicative skills that led to human language arose not in response to hunting, gathering, or tool use, but as a concomitant of social interaction, coordination, and manipulation – much of this work has come under the heading of studies of “Machiavellian Intelligence”. It appears that our most distinctive human attributes may have arisen as an intensification of attributes already present and well developed in social mammals (and perhaps social birds), and that these attributes may particularly have been a response to the kind of coordination problems we saw in our parallel processing computer example. The biological importance of our particular forms of intelligence to our recent evolution can be seen in a couple of ways. First, there are the salient ways in which we contrast with the other ape species – even those such as chimps and bonobos to whom we are particularly closely related. Second, there is the tremendous selective advantage that must necessarily be afforded by our special adaptation in order to offset the great biological price we have paid for it from fairly early on. That price includes the extremely

What culture gives us 101

underdeveloped (and, hence, at risk) state in which our babies are born and, even with that early birth, the great risk that childbirth still poses for human mothers (as opposed to the mothers of all other mammalian species). And the price includes our very long period of childhood dependence and relative non-productivity while all the necessary learning takes place. Insofar as social coordination has been at least part of the early driving force behind our move into the “smart animal” niche, and insofar as coordinated distributed cognition is a component (or a product) of that social coordination, we can see something of how important this mode of decentralized participation in complex social entities is to us. 8. What culture gives us Culture is not, for us as natives, a set of cultural imperatives, but instead is the set of understandings that we share with other members of our social systems concerning what goes with what, what means what, what follows from what, and so forth. We utilize these understandings in calculating our behavior, in interpreting the behavior of others, in coordinating with others, in competing with others, and so forth. Our understandings embrace not just regular behavior and prescribed behavior and proscribed behavior, but also expectations about when such regularities will be adhered to and when they will not, and expectations about the signs of non-adherence and the consequences of non-adherence. By this view culture is in part a form of communication – a sometimes non-verbal means by which we exchange information about what we are doing, what we want, and what we expect; but culture also represents a kit-bag of devices – a set of techniques for manipulating the world of things as well as of people. Culture involves, as well mechanisms by which we, in concert, can change our culture. The kit bag contains, too, other kinds of useful conceptual things, such as potential goals, potential sets of others with whom to identify and to whom to be loyal, potential attributes in terms of which we can each construct our individual concept of self (of who we each are). We as individuals are no more constrained by these culturally proffered values than we are by the culture’s array of beliefs, rules, and actions; but we do need values to guide our actions, and – as social animals who belong to various alternative but overlapping groups that sometimes compete with each other – we do need a way to decide which “we” counts vs. which “they”, and we need to have ways within our group to know whom to follow and whom to lead. As individuals we are not forced to do things as our culture (or our individual

102 Chapter 4 – Culture as distributed cognition understanding of our culture) says – and every culture seems to have some who to greater or lesser degrees resist or rebel – but it is convenient to have available a set of cultural defaults – which most of us do more or less follow for most purposes (producing the uniformity which enables our reification of culture to “sort of” work). Looked at this way, culture is quintessentially human. But we should also remember that it is not necessarily restricted only to humans. Some aspects of our sociability, of our socially held knowledge, and of our relationship as individuals to groups is paralleled among other social mammals – including wild dogs and chimpanzees as well, perhaps, as elephants, dolphins, and others. To the degree that aspects of our cultural apparatus or of the social forms and behaviors to which they relate are phylogenetically shared, the biological issue concerning our evolution of the capacity for culture is made easier to deal with – both as regards the time span and the succession of steps via which we arrived at our present state.

9. What makes culture socially systematic We have so far seen something of the way in which culture is individually inferred and organized, something of what makes the individual representations of a culture cohere as a socially distributed system, and something of what seems the culture’s use to the individual. Next we need to consider the problem of how culture thus conceived addresses society’s problems, produces the “emergent properties” that Durkheim spoke of, and adapts to the changing circumstances that people always, to a greater or lesser degree, experience. Society from this perspective is made up of the groupings of people we form, ranging from local, ad hoc ones (such as the attendees of the academic session at which this argument was first presented) up through various kinds and levels of overlapping groups of varying permanence and self-consciousness (such as the attendees at the Lake Arrowhead Conference which the session was part of, the memberships of our various departments, universities, and companies, the memberships of our various disciplinary groups, the memberships of our children’s PTA or our cactus club or bowling league, the memberships of our churches and political clubs, and so forth), up to larger national or international units or collections of units. Culture in this perspective is the shared knowledge (including appropriate behavior, beliefs, values, attitudes, emotions, etc.) that enables members of the various

What makes culture socially systematic 103

groups to recognize each other, anticipate each other’s behavior, work together on common projects, read each other’s goals and biases, compete effectively with one another for scarce goods, etc. Both society and culture are epiphenomenal in the sense that a) as individual agents we each constantly shift our immediate, working memberships (and attendant operative self identifications) from one to the other of these groups, making us each in our individual essence members of none, while b) the groups are each totally and only made up of people like us – individual agents who lack any essential link to the group but who (varyingly often and varyingly convincingly) act as if they are members. At the same time we (again, each of us, as “natives”) do ascribe reality – including the possession of authentic, intrinsic membership – to many of these groups. Our participation or membership in such groups is not “all or nothing”, but is a matter of varying degrees of credibility, commitment, and apparent authenticity. The groups we know of – and contribute to the wider social representations of – are not just those we belong to, but those we interact with, those our friends and enemies belong to, those we see around us, and so forth. Society in its larger sense, with its cultural content, comes to be seen not as any simple collection of individuals but as a fabric “felted” out of the various varyingly overlapping groups to which individual agents relate. The fabric seems more “felted” than “woven” in the sense that it is formed by the mass of local adhesions – represented by shared individual members – that tie little groups into bigger ones, that tie different kinds of groups (recreational, professional, political, incidental, etc.) together and thereby keep logically or organizationally unrelated groups in synch. Thus, as explained, culture can insightfully be seen as the system of distributed cognition that enables society to function as a system of something like parallel distributed processing. I want to suggest – in a loose but serious fashion – that society (with its various parts) is like the big computer that is being simulated (see below) and that individuals are like the little computers that are being linked together; social structure represents something like the connectedness among these networked little computers while culture represents that shared program overhead (aka knowledge) that allows them to coordinate (whether cooperate, compete, or …). Thus culture consists of some mix of shared and individuated knowledge, but with systematic ways of relating individuated knowledge to the shared stuff. The degree of sharing is variable, depending on how tight the coordination. I want next to note that the mechanism by which culture is transmitted – reconstructed anew as a system by each new learner in response to the par-

104 Chapter 4 – Culture as distributed cognition ticular experiences of that learner – makes it an ideal vehicle for a form of non-genetic natural selection of knowledge in which only relevant knowledge (that which actually comes up in life experiences) gets transmitted, in which potential new knowledge is constantly being created (via the continual striving by a variety of different individuals for new solutions to new and old, big and small problems), and in which the interactive constraints (of effective communication, cooperation, competition, avoidance of collisions, etc.) keep all the various individual representations more or less in synch with one another and thereby filter out all of the innovations or alterations that do not have general and systematic relevance to some community. It is worth reiterating in this context that, while on the one hand culture is dependent on society (that is, evolves in and is a property of social groups), on the other hand culture represents the shared understandings that enable anything beyond the most minimal social interaction to take place and thus that enable society. Culture and society are interdependent – or, rather, social units and cultural (sub)systems are interdependent. Language too, perhaps as a part of culture, has a similar interdependence with communities (speech communities) and thus society 41. The interdependence of culture and society, that is of shared representations and shared interactions suggest the existence of at least two kinds of variability in individual representations within a wider society and culture. The first is the familiar sub-cultural kind. People within different local communities42 will have somewhat different experiences, including both the things or people they interact with and the social feedback they get regarding various of their actions. Each of these communities, as sub-societies, then, entails a sub-culture. At the same time, each of us (at least within any large, modern society 43) belongs to many such local communities and sub41

42

43

This feedback process is discussed in Plastic Glasses and Church Fathers (1996), and this sense of social construction is treated at some length in connection with language. There the Saussurean basis of this view, its Durkheimian connections, and some of the psychological processes and capabilities which might drive it are all considered. “Local” communities here can be based on geographical, occupational, recreational, or other groupings. The idea is not limited to neighborhoods or villages. Note that, from the perspective of the present book, the difference between large, modern societies and smaller and/or more traditional ones regarding internal communities and associated sub-cultures is only a matter of degree, not of kind. That is, the membership of individuals in a variety of varyingly overlapping local communities, and the resulting make-up of wider societies and cultures as amalgams of such overlapping entities seems a general feature of human society.

What makes culture socially systematic 105

cultures. Our overlapping membership and the information and experience we bring from one local community to another serve to keep the culture of the wider community or society more or less consistent; since the mechanism for keeping the various local communities aligned depends on common experiences with some overlapping membership, the degree of resulting consistency will depend on the degree to which experiences in one community are marked off from those in others. See Chapter 5 for an extended example. The second kind of variability in individual representations within a wider culture comes from the relative degree to which social interaction constrains the representations – or aspects or parts of them. For example, our individual conceptualizations of unicorns are somewhat constrained on certain features: we know that they are mythical, white, graceful, and that they have a single (perhaps cork-screw twisted) horn, and an association with feminine chastity; the feminine grace aspect tends to push their body representations in the direction of a slender horse (or, sometimes, a graceful goat or antelope). On the other hand, we are free to represent their diet, their weight, the quality and shape of their teeth, their preference for sleeping places, and so forth however we want. The aspects, attributes, parts of culturally relevant entities are constrained to the degree that they enter into the social interactions that shape culture; the rest is free for us to understand or imagine as we will, perhaps subject to some more general constraints of logical or cultural consistency. Cybernetics and systems theory have given us the answer to the old Durkheimian problem of how individual actors, acting in terms of individual perceptions, knowledge, and motivations, can come to make up a system with the emergent properties which go with such a new level of organization. The essential element is that the various individuals predicate their behavior not only on the physical world around them, but also on the social world; that is, individual calculations are based in part on what other individuals do and on what they are expected to do. Culture as outlined above does embody this form of social feedback. See Kronenfeld and Kaus (1993) for a simple simulation (using “grazers” such as “starlings“) of how the introduction of a parameter representing minimal responsiveness of critters in a group to other critters transforms apparently random or chaotic behavior by the critters into the kind of patterned (here, mob-like) behavior which we associate with the simplest kind of emergent social forms. The added element that comes with culture and thought (and that perhaps reflects some innate predisposition) is for the feedback not simply to be used to modify our immediate behavior, but also – more importantly –

106 Chapter 4 – Culture as distributed cognition the wider and more structured conceptualizations that we use to guide that behavior. That is, as seen in our striving to read grammar-like regularities into the patches of speech behavior that we encounter, we try with everything else we encounter (within both the natural world and the social world of other humans) also to infer patterns and understandings that will enable us to anticipate and manipulate44 – that is, to infer structure. The kind of feedback explored in Kronenfeld and Kaus’s (1993) starling simulation – joined with the process of inferring structure and with our predication of our behavior (including thought) on the behavior (real, anticipated, and imagined) of others in the complex world of human society and language – seems adequate to produce the kind of emergent effects that Durkheim noted and adequate to produce the kinds of adaptiveness that anthropologists find in culture and cultural systems.

44

My view of this process of inferring insightful representations of structure derives in part from the work of Piaget and is outlined in the Appendix of my Plastic Glasses and Church Fathers (1996, OUP).

Chapter 5 An agent-based approach to cultural (and linguistic) change: Examples 45

We can now use some simple examples to explore how culture and language – as decentralized systems of distributed cognition – change and adapt over time, and thus how they are learned and transmitted. My own research interest is with systems of cultural knowledge, especially cultural action models, but still unresolved problems involving the nature of units (including their shape, structure, boundaries, and degree of specificity) make full-blown “cultural models” hard to deal with in any clear and straightforward manner. Instead, for the present exploration I will offer some simple semantic examples involving word meanings in English. These examples will illustrate how actions by independent agents, when linked in a social system, construct and change collective cognitive systems. We will consider examples of the kinds of functional situations that produce such collective effects and the mechanisms by which the effects are produced. I would like you the reader, if you will, to conjure up in your mind’s eye an image of a “pen” – the thing you write with. Describe that image to yourself well enough so that you can refer to it later. Language is either a major constituent of culture or else a closely related parallel system. Either way it is both similar to culture and different from it in regard to its formal properties in ways that have been discussed above. Referential semantics – relating the stuff of culture to the categories of language – represents one interface. When studying change in such collective cognitive systems, everyday words – as parts of language – have the advantage of being overt, public, labeled, clearly delimited or bounded, and assured of some importance, while their referential aspect entails a dependence on something outside of language.

45

I want to thank Dwight Read and an anonymous reviewer of an earlier version of this discussion for their helpful suggestions regarding how to make it more effective.

108 Chapter 5 – An agent-based approach to cultural (and linguistic) change 1. “PEN” To see this, first let us consider the kind of objects suggested immediately by the word “pen” – in connection with my request at the beginning of this discussion. My expectation is that the current, younger generation will come up with some variety of “ball-point” pen, with a few, perhaps, having in mind something closer to a “felt-tip”. People in an older generation may instead think of a “fountain pen”, especially if they are attending more to their “Platonic ideal” than to what they feel to be more mundanely salient. Few will pick out the old “stick pen” of my elementary school youth – a wood stick with a metal nib that one dipped in an inkwell; the system (though I’m not that old) that produced the inkwell into which Tom Sawyer dipped Becky Thatcher’s pigtails. And similarly, I doubt if anyone will describe the primary wing or tail feather of a turkey or goose – a “quill pen”. We have here a word with a clear historical record of past referents.46 The functional basis of the term – a portable, personal writing implement for producing relatively permanent records via ink – has remained constant. But, over time, thanks largely to technology – both as a producer of new materials and material capabilities and as a means to reducing the price of older materials – the primary form of the object denoted by the term has changed from the feather quill up through the other objects listed above. We need two facts about language and reference before we move on. First, language is a productive, socially constituted collective system, and so the pace and nature of semantic change is constrained by the requirement that different members of the community do not lose the ability to communicate with one another via ordinary words. Second, reference seems to split into two levels: a) the primary prototypic (also called core, kernel, or focal) referents of words that we conjure up in our minds eye and upon which we base our understanding of contrastive relations among word categories and to which we apply feature definitions, if any and b) secondary extended referents linked to the term via their similarity or “closeness” to the primary referent (whether such similarity be in denotative form, a connotative function, or a figurative analogy). When we, as children, learn our first language we are not taught the system of language, but infer it from our linguistic experiences. We learn words, with their referents, first as something like proper names and then – rapidly and broadly – we try to apply them to wider categories. We try out 46

Consistent with my prior semantic work, I am distinguishing specific actual referents of terms from the general relationship of reference.

„PEN“ 109

feature and/or relational bases until we hit a basis that works; i.e., one that enables us to use the word in ways that elicit appropriate responses (which is to say, in ways that indicate that those with whom we are talking also understand what we understand by the word). In this process we also learn what related or similar items are NOT the referents of the given words and, instead, what words DO refer to those items. That is, we learn the pattern of contrastive relations for a given word within the system of language. Within language we have a finite and limited (even if eventually, very large) vocabulary to use in talking about an infinitude of potential referents. We first learn which attributes are significant for what terms and second, as part of our productive use of language, we learn how to use existing words for new referents in a way that enables others in the community to understand what we are talking about and what we are saying about those referents. It is this latter stretching of words to new referents which produces semantic extension to secondary referents. As new language learners we are not explicitly taught the primary versus secondary (prototypic versus extended) referents of a word, but we infer that information from our experience. The referents that we most frequently encounter – which seem also best to epitomize the functional basis of the category and best to encompass what we know of the contrastive relationships of the given word with other words – become our prototypic referents for the word in question. Thus, under changing conditions, later learners may infer a different primary or prototypic referent from the one inferred by an earlier learner. The mechanism of extension enables earlier learners to stretch their category to encompass newly developed referents and enables later learners to include earlier referents (especially ones which seem still in active, even if lessened, use); thus changes in inferred prototypicality do not produce breakdowns in communication. Now we can see what has happened with the word “pen”. Due to changing technology the range of material referents for the word has gradually evolved. Since our personal prototypes for a kind of object change slowly and more in response to relatively major (vs. minor) changes in the world of referents, our current individual prototypic referents mostly reflect the referents we encountered during out first conversations concerning the objects in question. Hence we find a wide range of prototypic referents in any community that includes people with a wide range of ages or experience. The range gets even wider when we include past referents that are no longer part of anyone’s prototype. Which ones of the currently available, portable permanent writing devices become prototypes for contemporary learners depends on the ones they and their peers use the most and that depends on

110 Chapter 5 – An agent-based approach to cultural (and linguistic) change the price, aesthetics (the look of the line produced with the various pens as well as the pens’ appearance), convenience, and so on of the alternative pens. It depends as well on the kind of pen that writes most easily and cleanly, and on which aspects of the looks and behavior of text written with a pen contrast most sharply with the looks and behavior of other portable writing or drawing instruments such as pencils, styluses, crayons, etc. and with, e.g., typewriters (viewed as personally operated producers of relatively permanent textual output). Having laid out, hopefully convincingly, the straightforward materialist basis of our sense of what is a “pen”, I would like now to turn to two English kin term examples that illustrate referential changes independent of direct material factors or experience. These examples underscore the central role of communicative intent and force in the experiences from which we learn vocabulary and infer prototypicality.

2. “COUSIN” Our first English kin term example involves the word “cousin”. In this case our concern is not with changing external referents, as these have remained largely constant for the time periods we are considering. Instead, I want to offer two observations about the usage and implicit understanding of “cousin” that reflect on the way in which a pattern of experiences, aims, and actions by individual agents change the collective cultural reality in a systematic way. The first observation concerns the onetime usage of “cousin” (as an address term) in a connotative sense for people (non-kin) of approximately one’s own age to whom one felt close and with whom one felt a sense of solidarity. Today this kind of usage involves the terms “brother” and “sister” (as in “fraternity brother” or “sisters in the movement”). But in the parental generation we use “uncle” and “aunt” for such purposes rather than “mother” and “father” because the parental terms have too specific implications regarding authority, training, and inheritance – implications that would interfere with the solidarity message to be conveyed by the use of kin terms for close non-kin. We do use “mother” and “father” for non-kin, but only when specific authority or training implications are at issue, as in our use of “father” for a priest, or “mother” for an abbess, or when – at least in some European traditions – one speaks of ones academic adviser as one’s academic “father”. My presumption is that when “cousin” was used for nonkin in the way “brother” and “sister” are used today, it was in a situation

“COUSIN”

111

comparable to what occurs today with “uncle”/“aunt” relative to “father”/ “mother”; that is, I suggest that the sibling terms had specific legal and behavioral entailments (in terms of legal responsibilities and shared – if unequally structured – interest in patrimony) that interfered with the solidarity message intended by the extension of a kin term to non-kin.47 The primary difference between cousins and siblings was not solidarity but the legal rights and obligations of siblings to one another versus the comparative absence of such rights and obligations with cousins. If my conjecture is correct, then we have a case where a decrease in the legal and social importance of the extended family – which is formed around “brothers” and “sisters” when they become adults – has shifted the functional significance, and thus connotations, of sibling terms by stripping away most of the legal implications, thereby leaving them as terms that can indicate relatively unencumbered expressions of solidarity. At this point the only difference between “cousin” and “brother” or “sister” is the degree of closeness, and so “cousin” drops out as an expression of solidarity since it implies distant (versus close) kin – which undermines the solidarity message that the use of a kin term for non-kin is intended to transmit. Semantic extension – using a word for something slightly different from its primary or prototypic exemplar – requires that the extended referent have an attribute relevant to the source domain (here, kinfolk), that the extended referent be, in some relevant way, more similar to the source prototype than to the prototypic referents of contrasting categories (other kinsfolk), and that the functional issue that drives the contrast have some perceived relevance to what one wants to say about the extended referent. Presumably, as sibling terms gradually lost their technical legal importance, they were increasingly used to signal closeness and solidarity. A new generation of speakers saw this 47

The term “cousin” is not seen in English (according to the Oxford English Dictionary – OED) until the end of the 13th century when it was borrowed from Latin via French, and its modern usage may well not have stabilized until somewhat later. “Brother” and “sister”, on the other hand, are part of English’s direct heritage from Indo-European. “Brother” and “sister” have long been used in English for non-kin – going back well before the introduction of “cousin” into the language. However those non-kin extensions of the English sibling terms appear often to be in religious contexts and often to be used in a way that connotes obligations, while “cousin,” from its earliest introduction, was extended to non-kin in non-religious contexts and in usage that betokened, in the words of the OED, “intimacy, friendship, or familiarity”. It is this latter usage which has more lately moved over to “brother” and “sister”.

112 Chapter 5 – An agent-based approach to cultural (and linguistic) change extended usage as a fairly routine part of their language. This new generation of speakers grew up hearing both extensions – “cousin” and “brother”/ “sister” – for apparently similar messages, and so had to figure out the distinction (and possibly the confusions promoted by one or the other usage). The repeated pattern of individual decisions apparently had the effect of changing the collective reality: they avoided using “cousin” for solidarity extensions to age-mates because (via the contrast with “brother”/“sister”) it seemed to entail a relatively distant closeness versus the closeness that, if you will, was the whole point of the extension in the first place. What changed was that “cousin” was no longer the closest, same generation kin term available for generalized extension; the sibling terms, by virtue of their loss of their more specific legal associations, had become available for this purpose, and were clearly closer than were cousins.48

3. “SIBLING” Our second English kin example has to do with a word that I already have been using quite a bit, though without any flagging quotation marks: “sibling”. Traditionally the term is not listed among English kin terms, but rather is seen as a technical, analytic category, and, in fact, when used as a cover term for the category embracing both “brother” and “sister” it represents a recent anthropological creation. In contrast, cover terms such as “parent” and “child” have long been part of English.49 The context is the general re-alignment of American (English-speaking) kin roles already addressed in connection with “cousin”. Not only have the specific legal associations of “brother” and “sister” been weakened, but also at the same time (over, say, the last 70 or so years of the 20th Century) the legal and traditional – social, political, and economic – differences between “brothers” and “sisters” have, similarly, largely disappeared. This change 48

49

This example parallels Rundblad and Kronenfeld’s (2003: 129) discussion of Keller’s (1994: 80–83) example of German “englisch” homonyms meaning “English” and “angelic”, and his analysis – using his “invisible hand” maxims – of how cultural and social changes then forced a change for “angelic” to “engelhaft”. Although “parent” and “child” are not technically considered kin terms by some anthropologists, no formal criterion for rejecting them as kin terms has been provided and Read (1984) has analytically demonstrated their centrality in the logic underlying the American kinship terminology.

“SIBLING”

113

means that the sex of a sibling is relevant (or an issue) in far fewer settings than was previously the case and so there now exists an expanded set of conversational settings (e.g., shared interest in parents or in their estate, shared obligations toward parents, shared economic rights or interests, stake in each other’s sex lives or marriages, etc.) in which the sex distinction is essentially irrelevant. We adapt language to our changing needs, and it has become increasingly convenient to have a non-sex marked sibling term – a model for which anthropology happened to provide in the word “sibling”. Individual agents presumably, first encountered it in anthropology (or sociology) classes and then, as exposure spread, others began choosing – individually and creatively – to use the term in their own kinship conversations. Then a generation arose whose members heard the term used commonly enough to learn it as part of their received language. A pattern of parallel and repeated individual agent decisions changed the collective system. The “sibling” example is particularly interesting for another – more political and academic – reason. One big political push connected with Women’s Liberation has been the removal of sexist terms and usage from our language. Part of this push has involved the removal of male (or, more rarely, female) defaults – as in “all men are created equal” – in favor of unisex terms, “all people…” This push has thus spawned a search for, and development of, unisex terms such as “chairperson”, “s/he”, etc. The question arises as to whether the conscious political push is changing our language – a kind of intentional change that linguists have traditionally considered to be impossible – or are other changes in our lives both “causing” the political push and the linguistic change. The case of “sibling” is noteworthy since it does represent a clear and explicit recent unisex creation but one which was neither politically addressed nor pushed and was not part of anyone’s overt political agenda. This suggests that unisex innovations in modern English are not the result of centralized action by powerful or influential leaders but instead are the result of parallel individual actions by individual agents each seeking to be efficient and effective in their own individual communication. The systemic changes in language, though quite important, are only epiphenomenal. The cultural changes that led to them are not policies seeking these outcomes, but instead are changes in the legal (and, derivatively, customary) rights and obligations that kinfolk have to one another. This interpretation is consistent with Floyd Lounsbury’s suggestion, in his path-breaking study of generationally skewed kin terminology types, that legal rights ultimately drive the shape of kin term categories.

114 Chapter 5 – An agent-based approach to cultural (and linguistic) change 4. “JEW” So far I have suggested that our patterned, systematic semantic understandings in the interface between language and culture are the products of decentralized, agent-based innovations, and I have tried to illustrate some of the ways in which agents interact with existing semantic resources (i.e., language) to produce new words and meanings. But these are simplified examples from a world in which people (agents) belong to a single, homogeneous language community. This is unlike the profusion (or cacophony) of dialects, local speech communities, ethnic and occupational subcultures, regional cultures, etc. that we often seem to inhabit. I now turn to another example that illustrates how the present agent-based approach works in a world of variously overlapping and diverging – and sometimes opposed – sub-communities with different histories, economic stakes, cultural resources, and so on – namely a somewhat loaded and in various ways contested ethnic label that applies to me: “Jew” or “Jewish.” We can begin by examining the range of referential meanings for the term “Jew” and then turn to a consideration of its major connotations. Where relevant I will indicate the social grouping from whose perspective the indicated meaning resonates. I will then make several general points. Without going into more semantic theory, we can note that the word “Jew” has many alternative, more or less related, meanings. At the community level these include a) a member of a particular religious faith (with an orthodoxy continuum in the US for which being less orthodox is more common, but being more orthodox is considered more truly Jewish in religion, and where non-orthodox sometimes grades into non-religious) and b) a member of a particular ethnic group (with some complications arising from mixed parentage, whether or not they have embraced an alternative religious faith, and which parent is Jewish). Less obviously, but often in practice, the term refers to a member of “the Jewish community,” which often betokens more membership in a particular ethnically based economic community.50 The economic community in the US is based on the “general store” of ages past and its descendent enterprises 50

Such “stranger communities” (see Kronenfeld 1998) are actual networked groups based on a matrix of kinship and economic interconnections. They function in – sometimes actually organize – some particular economic niche, and typically are endogamous and united by a distinctive religion. Jewish groups offer only one example – though, to be sure, a salient one – of such ethnic-religious economic specialization.

“JEW”

115

such as supermarkets, clothing stores, furniture stores, restaurant equipment suppliers, junk yards, department stores, and so forth. In this community members exchange information, cooperate with buying and “jobbing” stock, sometimes help with capitalization, and so forth. The continued existence of this community has depended upon endogamous marriages.51 To some people, both Jews and non-Jews, it appears as if this distinctive, solidary community has carried its economic role from business over into some of the professions such as medicine, law, or university education, but in the absence of economic payoffs it appears that this apparent extension is based more on where economically mobile people chose to go than on any actual communal structure or coordination. For some people (particular communities or groups of people) the term Jew refers to a Zionist – especially to a member of an international Zionist group. Whether this Zionist connection is considered good or bad depends on the perspective of the group whose members are doing the perceiving. In some Christian communities “Jew” still evokes the “Christ killer” response. To members of some extremist religious and/or political groups the term refers to a member of an international conspiracy to rule the world, or to those helping the Devil rule the world, or those who do evil, ritual things to members of one or another non-Jewish community. The term has rich connotations. It can connote a good business person – as when, back in the North Carolina of my youth, Smoky Levi, my gentile barber, explained that it was his “Jew blood” (gained via the “shot-gun” marriage of his paternal grandfather) that made him good at business – though often with an added sense of sharp or legalistic dealing (Shakespeare’s Shylock comes to mind or the expression, “Jewing someone down”). These perspectives are normally those of non-Jews, but Jews who are deeply enough embedded in a local community or in a larger gentile world can sometimes be heard slipping into them. The term can suggest a member of the professions, especially law, medicine, or academics. In law and medicine it can carry over some of the closed, clannish flavor of the old “stranger community”, and carry over some of the idea of sharp dealing – presumably what President Nixon had in mind when discussing the use51

This community and members’ stores particularly characterized small-town America, though with important presence in major cities; more recently the community’s economic functions seem to have been taken over by mall-based franchise operations. Other such communities, with different ethnic bases, characterize, inter alia, the restaurant and motel worlds. See Kronenfeld (1998) for an analytic discussion of such “stranger communities”.

116 Chapter 5 – An agent-based approach to cultural (and linguistic) change fulness of his “Jew lawyers”. In the academic world it can carry a sense of participation in an old and rich intellectual tradition – which can shade over into socialist radicalism. Extended from the old stranger community conceptualization, the term often betokens social isolation and a lack of effective participation in local communities, especially as seen in a lack of intermarriage within them. We thus have the old “Wandering Jew” image and Stalin’s newer “Rootless Cosmopolitans”. To members of the Jewish community it often connotes a victim of prejudice or an outcast – unloved, often hated, and frequently the target of aggression and oppression. More recently, Israeli military and intelligence successes have conjured up an image among Jewish supporters of heroic and skilled modern-day Davids besting old Goliaths. On the other hand, with Israeli occupation of the West Bank and Gaza, the term “Jew” can symbolize the oppressors of Palestinians and their rights. I could keep going on …and on. Note that most of us are conversant with many of the above meanings. Also note that we have each been participants in social groupings actively adopting several of these perspectives. This can lead to our wearing one hat in one group even though it contradicts the hat we wear when participating in another group52. Some of these contradictions only represent alternative perspectives or definitions, while others can represent head-on, conflicting judgments and valuations. As individuals we do not belong wholly to any particular perspective, and we always (or, at least, mostly!) carry some coloring over from one group to the other. Besides joining in on a group perspective, we (both as individuals and as constituents of a group) adjust our expressions and apparent presuppositions to our audience – to whom we think may be listening and to what they might be thinking. My purpose here is neither to review the sociology of these issues nor the history of these constructions; rather it is to illustrate how an agent-based, semantic approach works in a realm of contested meanings and contesting communities. Initially as kids, but then continuing through our lives, when we encounter a word we try to understand what it applies to, what it connotes, who uses it, and why, and so on. Insofar as we get a variable picture, we try to make sense of towards what issue or context the varying usages are directed, of who is using the word in one way or the other, of what situational attributes account for, or constrain, the differences we see, and so on. Some of this knowledge affects our own usage and our interpretations of the usage 52

See Kronenfeld and Vike (2002) for a very different example of an individual’s participation in such contradictory memberships.

“JEW”

117

of others. We repeatedly select communities to which we attribute usages and align people. These communities include the alternative constructions of the “we” with whom we align ourselves, the “you” with whom we align those to whom we are talking, and the “they” with whom we align those we are talking about. As with our previous examples, people in similar situations make similar creative adjustments to the language regularities they encounter and learn. They create new regularities that new learners experience and learn as regularities in the pattern of word usage. Individual reactions and creations, unless they get picked up and repeated by others, remain idiosyncratic and thus not a part of the system of either language or culture. The conditioning regularities in the framing experiential conditions (which trigger the creative adjustments) are structured by some attribute of the people involved, the situation, or something else. This gets internalized by those who experience the new semantic forms and shapes the definition of persons to whom they ascribe the semantic forms and the entailments that can be read into the forms’ use. Thus, Israeli military successes converted what for many was a furtive and suffering – and perhaps cowardly and/or cerebral – image of Jews into a braver and more physically active and, for some, more aggressive and domineering image. The adjustment was fairly broadly experienced, but the details of how it played out (brave vs. aggressive, sacrificing spiritual high ground for mundane normality, shopkeepers vs. warriors, victims vs. victimizers, etc.) depended on the perspective of those reacting to the change. But we all, as individuals, understand most, if not all, of the above variants (both the old variants already in the system and the more newly created ones) apart from whether we agree, disagree, or remain neutral about any particular variant. None of those evaluations or judgments is intrinsically part of us even if they are associated with communities to which we belong, with situations in which we find ourselves, or with interests that we feel apply to us. As individuals we decide, to varying degrees, to agree or disagree with one or another attribute or claim. And most of us experience shifts in relevant frames (equivalent to “code switching” or dialect shifts in sociolinguistics) at one time or another. These shifts result from our participating in relevant conversations from different perspectives in conjunction with conversations in which we form different “we”s. We understand how to play those different roles. The semantic information belongs to collectivities that we move in and out of rather than those collectivities being anything that we are intrinsically or essentially part of. I do not mean that we know all the perspectives of our society equally well any more than we know all of the dialects of our wider language com-

118 Chapter 5 – An agent-based approach to cultural (and linguistic) change munity equally well. Our views about groups for which we are not members range from viewing them as very abstracted and stereotyped monolithic unities to knowing intimately their variation. In general, the more we engage in interaction the less stereotypic are our views, but we are capable of holding at the same time both a simple and even nasty, stereotypic view of the group and a richer, less nasty view of individual members whom we happen to know. Members of most minorities have had the experience, I’m sure, of having someone they know speak disparagingly of their minority, and then – when their presence or membership is noticed – explain (perhaps with a little embarrassment) “Oh, YOU aren’t one of them …” And, we do have substantial “inside” knowledge of many of the “them”s with whom we interact – that is, knowledge of how to think or act as members of those groups. And we normally move back and forth across a variety of the groups that we are aware of and talk about. And, as we each participate in one or another of these groups or contexts, we form part of the collectivity that others experience.53 So, the agent-based model has every individual acting as an individual but trying to understand collectivities of others and trying to adjust her or his individual behavior to take account of what is known about the relevant collectivities. We each try to generalize as much and as widely as possible, and use feedback (sometimes happily, sometimes grudgingly) to identify overgeneralizations. Just as individuals in similar circumstances behave in similar ways or evince similar understandings, those similarities form part of the array of collectivities that others experience. The internal representations which we form of these various collectivities are not just lists of known “facts” but are productive models that use sets of underlying assumptions to make inferences about behavior or attitudes from observed conditions. These representations belong potentially to the universe of mathematical or computational models. Many of our shared internal representations – the shreds and patches of culture – are candidates for formal modeling by ethnographers, not only for description but also (as Levi-Strauss long ago noted) for exploration and experimentation. This process of tuning individual representations seems the cognitive equivalent of that of Kronenfeld and Kaus’s (1993) computer simulated starlings, where each moved randomly and individually, but with a firm rule to never move more than a given “leash” distance from “the flock” defined as the center of gravity of the rest of the flock. The leash constraint was 53

And thus sometimes we even belong to that part of the group which in someone’s eyes constitute the group’s prototypic exemplars.

“JEW”

119

sufficient to produce emergent properties of the group as a genuine social entity. To really make the simulation a social analog of the cultural system being offered here, it would have to be complicated by the addition of other groups and having the “starlings” live in a multi-dimensional universe wherein they could simultaneously be tied by separate leashes to a variety of otherwise independent groups!54

54

For a nice example of how separate connections to otherwise independent groups can work see Padgett and Ansell’s (1993) discussion of Cosimo de’ Medici’s style of political action. A relevant summary is on p. 1263.

Chapter 6 Society (with a note on the self)

1. Society: issues Our discussion of the various social groups to which ethnic and other group labels are applied and to which the people applying the group labels belong takes us to question of what is the nature of such social groups and through them of society. Society has often been understood to be an emergent phenomenon, a kind of super-individual and superorganic entity that shapes and constrains the behavior of its members. Society in the large is considered, in this context, to be made up of a collection of smaller social groups in part hierarchically interlinked, in part networked to one another, and in part independent (though possibly linked by overlapping membership). Society thus understood can be taken as a kind of structure – as it was by Radcliffe-Brown and Fortes. The problem with the idea of society as a “thing” is that the more that one tries to precisely define – or focus on – it or its parts the more it seems to dissolve into fuzzier patterns of interaction among individual persons. And the related problem with the structure idea, as so richly if inadvertently illustrated by Fortes’ Tallensi books, is that the interactions that are supposed to instantiate and exemplify the structure seem too inconsistent and variable. These considerations take us to Fredrik Barth’s conception of society (1992) as cumulated patterns of individual interactions, shaped and aligned through feedback with the environment (including other individuals) – where consistencies in the environment induce consistencies across the behavior of individuals. Barth contrasts his view, in which “society” itself can be considered an epiphenomenal appearance produced by recurring patterns of individual behavior produced for individual reasons, with the view which he rejects, in which society is seen as some actual entity with clearly delineated boundaries and some clear systemic properties and interactions. Is the “system” of “social system” to be seen as referring to the systematic aspects of the behavior of the individuals who make up the society or is it to be seen as inhering in the organizational properties of the social entity itself? In a similar vein, one can contrast a view in which structure is to be found in the organization of actors’ knowledge, motivations, and actions with

Society: issues

121

one in which structure is to be found in the arrangements of the segments of the social entity. How we conceptualize pieces of society, such as ethnic groups, follows in part at least from how we conceptualize society itself. Our anthropological understanding of society has, certainly since Durkheim, been torn by the disagreement between reductionists who see society and social groups simply as agglomerations of individual actors – each thinking and acting as individuals – and anti-reductionists who see society as an emergent phenomenon of some sort. Since, by most accounts, only individuals actually have consciousness, know things, and make decisions, reductionists have considered any recourse to any “higher” level of collective action to be mystical. On the other hand, as Durkheim, among others showed, social groups lead their members to behave in ways that they would not without the group, and many kinds of action seem predictable from features of groups without any recourse to individual actors, and regardless of who the individuals in the group are. My thoughts on the nature of society come out of my work on kinship and my reflections on the tightly interconnected work of Durkheim and Saussure; they also reflect insights gained in development of a computer simulation of some kinds of simple social systems (Kronenfeld and Kaus 1993 and Kronenfeld and Kronenfeld n.d). Much of the particular discussion that follows was triggered by Fredrik Barth’s essay entitled “Towards greater naturalism in conceptualizing societies” (1992). Let me start by agreeing with Barth’s critique of the idea of seeing “society” as a bounded, discrete thing (vs. as a collection or patterning of individual behaviors), and by agreeing with his list of the kinds of things that society is not. That is, that society is not a “summed up [] aggregate of social relations”, not to be taken as “the aggregate of institutions of a population”, not anything that can “be represented by any schema which depicts it as a whole composed of parts”, not something whereby the whole-part problem can be “escaped by declaring the whole world one society”, not something which “can[] be extracted from the material context…[which can be] separate[d] from ‘environment’ and then show[n to] affect[] or … adapt [to ‘environment’]”, and not a concept which, “as much as ‘culture’, serves to homogenize and essentialize our conception of the social” since “we know that not only interests but also values and realities are contested between persons in stable social interaction with each other” – even if I might have some mild caveats about how some of his denials are phrased. However – I want to attach two “howevers” to my agreement with Barth.

122 Chapter 6 – Society (with a note on the self) My first “however” involves my recognition that there may seem to be something contradictory about my agreement with Barth since my work and thought are tightly tied to Durkheim’s – and particularly to his societal notions of “collective representations” and “emergent properties”. In the discussion that follows I sketch in how I interpret Durkheim (and Saussure) and, thus, of why I see no actual contradiction between this view and my agreement with Barth. At issue is what we precisely understand by “collective representations” and what precise claims about individual behavioral motivation we understand to be entailed by relevant assertions concerning “emergent properties”. I might also add, here, that I am a psychological reductionist in good standing. My computer simulation work (Kronenfeld and Kaus 1993 – and see Kronenfeld and Kronenfeld 2007, n.d.) both shows how I blend these diverse strands and demonstrates the plausibility of the kinds of claims that I am making. In that simulation a small society (of “starlings”) was created that exhibited some simple but clear “emergent properties”; this was done without introducing any actors other than the birds themselves, and without the birds basing their actions on anything beyond their individual perceptions and motivations. At the same time, the birds were able to perceive other birds and were motivated to stay near what they perceived as the collectivity. The simulation did not deal with cognition or language, and so the birds had no internal representations, thoughts, or names for things, and no “folk” explanations of their behavior. The simulation is reductionist, but I think, in a way that Durkheim himself would not have disapproved of. My second “however” is more specific and substantive. I think that there exists a clear sense in which, by and large, Barth is attacking a straw man. That is, most anthropologists who talk about society are too vague and imprecise in what they say (and about what they mean) for the distinction we are making between society as phenomenon and as epiphenomenon to be clearly drawn in their work; we all have our favorite nominees, but I would put Leach in this category. Many of those who are clear, such as EvansPritchard in his Nuer studies (1940, 1951) and Levi-Strauss in his kinship writings (1963), mean something much more cognitive and abstractly systematic than social by their use of terms like “society” – that is, they are analyzing conceptions of society rather than anything to do with society itself. What we are left with are a very few anthropologists of a previous generation – such as, most saliently, Fortes (1945, 1949) – who were clear about what they meant and who were talking of society as an actual social entity. Fortes would be a worthy target for a critique except for the fact that no one any longer takes his position on society seriously.

Society: issues

123

However, even if the idea of society Barth which attacks is not formally and explicitly entertained very often in anthropological theory, the issue he raises is important in another way. My sense is that its real force applies to the unspoken and implicit presuppositions that anthropologists (and others) bring to their theorizing. The problem is that the conceptualization of society as a definite entity springs all too naturally to our minds as “natives”, even when we analytically (as anthropologists) know better. That is, we do feel, as natives, that society is a sort of “thing”, and that that “thing” is what our analytic selves are studying. If the “thing” society does not actually exist – that is, if it is only an epiphenomenon of our perceptions of the actions and statements of individuals, and not any kind of systematically imperative and operative, functioning entity – then two questions are posed: 1) What systems of individual knowledge and action do exist that produce the epiphenomenon? And 2) why do we (even social scientists), persist, as “natives”, in thinking of the epiphenomenon as a phenomenal reality? The remainder of this chapter will be addressed to answering these two questions. 1) As an answer to the question concerning what produces the epiphenomenon, Barth offers exchanges among individuals, and the cumulative effect of past exchanges via feedback loops (producing convergent views and models where aims/goals converge, and maintaining differences where aims/goals suggest such) as his nominee for the system that actually exists. I propose that he is right about the exchange relations among individuals and about the systemic interaction and feedback relations among these individuals. But I also propose that he is too simplistic in his view that the individuals bring no abstracted knowledge and no structures (by which experience, knowledge, and action are organized) to these exchanges. The problem with abstracted knowledge and action structures that troubles him and others is that these seem to slip back in the idea of social entities that exist outside of individuals, and outside of individual knowledge and motivation – the same problem that led to the rejection of “society” as an entity. It seems to me, though, that the difficulty really depends on how one conceptualizes structure. A common model these days of a conceptual structure is the kind that Chomsky offers for language. His structure does capture many of the regularities of language, but it cannot represent any kind of direct description of what speakers (or hearers) actually do in their use of language. Whatever it is that those who assert these structures are “psychologically real” mean by

124 Chapter 6 – Society (with a note on the self) “psychologically real”, it cannot involve the absurd notion that speakers first construct a Chomskyian syntactic structure for a sentence, and only then introduce any semantic input; at the same time, by not specifying what they do mean by psychological reality, and by not even suggesting how structures might concretely relate to the individual’s production of the action the structures describe, the proponents of such structure do leave the matter of the actual nature and status of these structures very much in doubt, safely removed from any empirical criticism or exploration – and smacking just the least bit of mysticism. The same would apply to any equivalent cultural or societal structures, and to the positions of those who advance them. Instead, let us assume that the structures found by Chomsky (or those suggested by proponents of analogous cultural or social structures) are epiphenomenal (in much the same manner that “society” was claimed above to be) – that they are formal accounts of regularities produced by the normal operation of the processes by which we produce speech (or cultural or social action). Fortes himself might have claimed that his Tallensi structures do directly represent actors’ decisions, but, even for those of us who admire him and his ethnographic work, his structures are better seen as entailing claims analogous to those entailed by Chomsky’s structures (as understood here).55 But we still have not spoken of what it is that produces such regularities of speech or action – what are the processes, how do these processes manage to produce the systematic regularities which we see in the productions of a single individual, and how are the productions of separate individuals kept as tightly aligned as they are? That is, one wants to know what kind of cognitive structures we do have, how these are learned, and how they are 55

As an aside, I might note that the kind of structures proposed by Evans-Pritchard for the Nuer are something quite different. They are more of a kind of cognitive template, a kind of conceptualization that people in a culture can impose on events which meet defined application conditions. These conceptualizations can be used to interpret (or makes sense of) events that occur, or frame events in a way that leads to action. They are different, though, from the kinds of structures that we have been discussing because they represent "native" abstractions rather than anything that inheres intrinsically in the events themselves, and because often members of the culture will have a choice among several very different conceptualizations which can fit the same given concrete situation or train of events. A related kind of abstract mental structure is the kind offered by Levi-Strauss (in his work on kinship) which is really a kind of working out of the systemic implications of a set of assumptions, and not in any sense a direct claim about what some actual folks are actually doing.

Society: issues

125

coordinated among all the people who make up a given society or speech community. These cognitive structures have to be systematic enough in their structuring of what we know and how we express ourselves to produce the complex regularities that people like Chomsky (or Lounsbury or Fortes) observe and capture in their structural accounts and they have to be flexible enough and fast enough in their operation to allow our normal “real time” purposive behavior and interaction. The answer to the structural question is implicit in Saussure’s conceptualization of langue as “passive” and “socially” received – that is, as the pattern of passively experienced regularities which a speaker uses, actively, in constructing a new speech act. Saussure’s regularities are not abstracted out as statistical patterns but as mechanical relations which we can use in our own individual active constructions. Saussure speaks in terms of such “passively” experienced regularities or patterns rather than in today’s idiom of language as some kind of “code”. Our notion of “code”, though in some ways useful, has the unfortunate consequence of implying the existence of rules abstracted out from the context of their ordinary application and of some meta-language in which the rules of the code are expressed and via which they are modified. I would offer, as a modern concrete instantiation of Saussure’s conceptualization something like a public and collective version of a Piagetian “schema” – what I have spoken of above sometimes as a “schematic” or “schematic picture” and what I will shortly introduce as the class of collective representations that includes “cultural models”. Since language (and, perhaps, culture) is intrinsically and essentially social, these language (and culture) schematics would be socially shared – via the same vehicles of shared context of learning and interactive context of communication that lies at the heart of Saussure’s conceptualization. In fact, though, unlike Piaget’s schemas, Saussure’s langue is a socially held structure – a Durkheimian collective representation, if you will. This Saussurean collective structure is not what is in any one person’s head nor what directly structures any one person’s action. Instead, it is the parts or aspects of the various individual structures that overlap throughout the relevant speech community – that is, the parts that are shared. Since one person can participate in several different speech communities (at home, at work, in a bar, etc.) a single person may more or less simultaneously participate in several different variants of langue. In the Saussurean system langue is not a structure which directly produces speech. Instead it constitutes the corpus of socially shared regularities which an individual relies upon in his or her active individual construction of a speech act. More precisely, it is her or his own individual internal rep-

126 Chapter 6 – Society (with a note on the self) resentation of langue which the individual uses. It is this individual representation which links directly with Piagetian schemas – since it is the individual representation which is actively and directly involved in the speaker’s production of speech. Piagetian schemas are constantly adapting structures of knowledge linked to action and goals. In adapting the Piagetian schema to the Saussurean concept we need to understand that the environment to which the schemas in question (schemas of individual people) are adapting to include not just regularities encountered in the world of things but also regularities experienced in the behavior (and behavioral responses) of other people and that people pull out of their experience a generic version – which they attribute to the collectivity and on which they then base their on personal schemas. Piaget has shown how the construction of cognitive schemas can produce formally describable regularities in understanding and behavior while being flexible and adaptive. Children construct their individual internal representations of langue out of their attempts to model the regularities they encounter in the speech they hear. Their models or representations are constantly tuned via the feedback they get from members of the community as they attempt to produce speech that is communicatively effective and socially appropriate. The ubiquitousness of language, its importance, and its requirement (for effectiveness) of a close match between speakers’ conceptualizations and hearers’ conceptualizations all act to guarantee that all of the children in a given speech community have more or less the same kinds of langue experiences and thus wind up constructing essentially similar internal representations; but note that – even though linguists often do treat linguistic rules as a nonstatistical, all or nothing thing – there does exist some minor variation from one person to the other. Each child’s individual representation does constitute a real system, but the totality of representations in the community do not, even if the constraints regarding interactive stability lead the separate systems to converge and thus to come close to appearing to constitute a single system. Note also that the child’s constructed internal representation of langue (especially vocabulary, but also grammatical features) does depend on what instances of language (that is, what corpus of speech acts) the child experiences, which, in turn, depends in part on what happens to get talked about in the child’s community. Thus, insofar as in different spatial or temporal situations different objects, relations, and actions get talked about, children growing up in those different situations will wind up experiencing different regularities and constructing different variants of langue. This is part of the process by which language “drifts”.

Society: issues

127

As a first approximation of an answer to our question regarding what it is that produces the regularities and patterning of “society” (even if society be epiphenomenal), then, I would suggest that our “native” participant conceptualizations of social structure, as seen both in our speech and in our and behavior, might be understood as collective representations similar to Saussure’s langue. I would also suggest that something of the sort is what Durkheim himself had in mind in speaking of “collective representations”. Like the systematic patterning of language, the systematic patterning of society is tight enough (even if, perhaps, not as tight as that of language) that we all have to form pretty much the same individual internal representation of it. Such tightly socially constrained systems would contrast, for example, with our representations of the appearance and private lives of angels or unicorns – where we are each free to form individual representations which may be widely at odds with the representations held by those around us. 2) We now can turn to our second Barth-related question, concerning why we persist in thinking of the epiphenomenon as a phenomenal reality – why we persist in mistaking society as a bounded, discrete “thing” even when we know it is not such. First, I think that this is another of those areas in which social scientists have taken folk concepts (under some apparently straightforward interpretation) as the basis for theoretical refinement into technical scientific concepts (on the model of how physicists got “mass” out of “weight”).56 People talk about society as an entity in ordinary discourse – and so social scientists presume it must be such, and try to find refined theoretical definitions of the entity; the object of the chase seems always, though, to turn into some “will-o’th-wisp” or chimera. However, I suggest that57 our “natives” are actually focussing on the nature of their felt interactions with social entities and not on any general or abstract definition of these entities. The reason these cumulations of social interactions (which we, as social animals, constantly participate in) “feel” like interactions with “things” or entities has to do with the way in which we conceptualize (even, perhaps, “domesticate”) “others”, rather than with anything in particular about the collective/social others with whom we happen to be interacting. In brief, 56

57

My idea here comes directly from Roy D’Andrade’s explanation of how personality psychology wound up being about the semantics of their labels rather than systematic patterns in the behavior of the subjects being studied (see D’Andrade 1964). Like D’Andrade’s subjects.

128 Chapter 6 – Society (with a note on the self) the suggestion is that we understand (rationalize, make sense of) the actions of others by ascribing to them the kinds of thoughts, perceptions, and motivations that we “know” govern our own behavior as individuals. Thus, we make sense of the behavior of a pet dog or a filmed otter by “anthropomorphizing” the animal – by ascribing to it a human intellect with human values, goals, and reasoning abilities. We appear to do the same “anthropomorphizing” with more metaphoric creatures as well – that is, we (as “natives”) “understand: the behavior of any supposed social entity (or social collectivity) by seeing it as if it were a single sentient creature. Thus we speak of the bullish “mood” of the stock market or of the consumer “aims” of the new Russian middle class or of the “feelings” of frustration of Bosnian Moslems. It is, often, true that many of the individuals in each of the just mentioned groupings have the feelings just ascribed, but we do not think or say that “many individuals58 feel or think thus”, nor that “the politically dominant faction acts as if it feels or thinks such”; instead in our language and apparently in our very thought we ascribe the feelings and thoughts in question to the social entity itself, and thereby ascribe to this entity the kind of systematic existence that Barth (rightfully) takes issue with. One should note that such collective anthropomorphizing always involves some “other” with which one is interacting. Our perceptions of groups as unitary or collective and our ascription of thoughts, motives, and actions to groups never include ourselves in those perceptions and/or ascriptions – even when the definition of the group in question does include us – and so there is no mechanism by which collective representations can automatically drive our own perceptions or our own behavior. The collectivity only exists in our minds, and it is a “collective representation” only to the degree that some community of us all have pretty much the same one in each of our separate minds. Collective representations only affect our conscious individual behavior in the same way that other individuals affect it – by constituting, insofar as we attend to them or feel their relevance, one of the pressures operating on us as we, as individual actors, decide what to do. But by representing understandings whose sharing we can depend on in our communication and interaction, collective representations do play a crucial role in language and in our social interaction. By giving us categories for 58

In a similar manner, I do not think that our references to groups as if they were individuals are to seen as elided references either to some kind of “best exemplar” of the group or to some kind of focal or idealized individual member of the group.

Society: issues

129

recognizing things in the world, by providing us with a stock of motives and goals, and by giving us sets of default actions for dealing with the world, collective representations do also provide much of the conceptual world within which we do our individual calculating. To recapitulate, society is only real in our “native” minds, but it is always real there. At the same time, we, also, always see our own selves as autonomous individuals who take society into account as we plan our individual actions – rather than as will-less pieces of society acting automatically according to some cultural plan. This gap between our selves and our collectivities exists even where we, in the abstract, define a social unit that includes us, and even when the goal of our individual calculations is socialist collective behavior or Buddhist submergence of the self. Apart from whether my particular explanation of society is right or wrong, I want also to claim, emphatically, that we do have to understand where our (apparent) misconceptions come from before we can successfully clear them up and move beyond them. That is, I see this issue of why we persist in our reification of society as being as basic to any critique of the idea of society as a “thing” and to any further development of our understanding of society as is the issue of what produces social regularities. In conclusion we can now turn to the relationship between our understandings of ethnicity and of society. Our natural “native” disposition, I have suggested, is to “reify” the social groups with which we interact – to treat them as organisms possessing the same kind of perceptions, motivations, and actions that we ourselves, as sentient individuals, have. Particularly subject to this process of anthropomorphization are groups which seem to us to represent some sort of “natural” “kinds of” humans – groups based on shared descent and exhibiting shared appearance and customs. Thus, as natives, we are strongly inclined to see ethnic groups as phenomenal social organisms, as actual “things” which social science should be able to accurately describe and model the properties of. This “native” presumption of phenomenality or “thing-ness” constantly interferes with our analytic ability to see clearly and credit the more epiphenomenally ad hoc communicative uses of ethnic labels and the perceptual and presuppositional facts on which this communicative use relies.

130 Chapter 6 – Society (with a note on the self) 2. Society: Kronenfeld view Society is never an unmediated collection of individuals. It is always mediated by some conceptions of what the relevant grouping is and of who is in it. Such mediation always involves some conception of collective representations – that is, some conception of a conceptual structure that is shared among members of (participants in) the group and that provides a context of meaning for actions by members, and that thus contributes to interpretations by members of the import of actions – interpretations by actors, recipients, and observers. Some societal collections are directly seen as direct collections of individuals, while others are seen as second order collections of collections. Direct collections constitute both the speech communities in which language is, ultimately, rooted (learned, defined, modified) and the communities in which culture, similarly, is rooted. Individuals in any society (not just in “complex” or “modern” societies) all belong to a great many societally relevant groupings. To a greater or lesser degree each person’s total set of groupings is unique and thus unlike that of any other person. Each of these groups has its own mediating conceptions. Some groupings and some mediating conceptions are self-consciously held, while others are not (but still can be shown to exist behaviorally and interactively). This densely cross-cutting membership by individuals contributes to a kind of felt-like structure of society as a whole. Often many (but far from all) of the groupings in a society are hierarchically structured (where local subgroups from different localities are linked into successively higher level subgroups at successively broader levels of society). Where such hierarchically structured groupings are important, they can lead to a perception of society as a kind of fabric in which they constitute the warp and other local or economic or political or … relations constitute the woof. Such hierarchically structured groupings can be seen as social segments. Language represents one such hierarchical grouping – of speech communities into dialects into languages. The multiplicity of primary groups to which individuals each belong and the variety of ways individuals’ various primary groups get linked to other groups explain the looseness of the language-culture fit, especially when looked at from the high level of whole languages and whole cultures. The groundedness 59 of language and culture 59

By “the groundedness of language and culture … in local communities” I refer to a combination of the fact that individuals learn them within such local interactive

Society: Kronenfeld view 131

in the same local communities explains part of why they feel coterminous to us and why we often feel it appropriate to use the one to stand for the other. This partial substitutability feels stronger and more convincing in relatively homogeneous local communities, while it feels less obvious in more cosmopolitan communities. The sense we have of the primary local linkage would seem to be one reason why (or mechanism by which) higher level linguistic groupings can come, politically, to be taken as representing cultural and hence political unities – producing a kind of linguistic nationalism. The mediating conceptions of each group or grouping are what defines the group by constituting the shared frame within which actions and their implications and effects are recognized and interpreted. At the same time, these mediating conceptions are shaped by the experience and actions and desires of the people that make up the relevant groupings. At the most immediate and local sense the mediating conceptions define and are defined by the groupings. People’s overlapping memberships constitute a pressure via a kind of cognitive ease push by individuals for – or , maybe better, tendency by individuals to collapse into – consistency across their various contexts, whenever and insofar as particular group situations or pressures don’t actively force differences. As individuals, we each and all can maintain within our repertoires the cultural analogue of dialectical differences, and can code switch among alternatives; but as with languages and dialects we do tend to drag constructions from the one across into the other. This process of carrying over is one mechanism that produces the wider consistencies of interpretative conceptualizations that, by some definitions, constitute culture. A second mechanism productive of cultural uniformity is the tendency of some conceptualizations to become or be seen as authoritative re society as a whole or re some significant social segment. Such authoritativeness can either be a matter of explicit rules and norms (as tried by religions, schools, and so forth) or a matter of exposure to authoritative or attractive models (as in TV dramas or sitcoms, or in lives of culture heroes, or …). This mechanism can also be involved in the kinds of instances of linguistic nationalism referred to above.

communities (the “rootedness” of the systems) and my sense that individual defaults and presuppositions about these systems get initially defined in and based on the behavioral patterns of such communities – even if later experience may selectively override such defaults and presuppositions.

132 Chapter 6 – Society (with a note on the self) Since individuals each belong to a range and variety of groups, the conceptualizations that define or are defined by each group are dealt with by the individuals from the outside. That is, each individual forms an internal representation of the conceptualizations of each group to which she or he relates – with perhaps varying degrees of completeness and specificity depending on how close and of what sort the relationship is. The conceptualizations that define or are defined by a group can be thought of as the group’s “collective representations”. It is these collective representations which anthropologists aim to be describing when they describe “cultural models” or “cultural knowledge systems” or “culture” (when taken in its broadly cognitive sense – including appropriate affect, motivation, etc.). But, since each and every group consists of nothing but such “outsiders”, none of them has any (and thus there exists no) direct representation of the conceptualizations – no place that the anthropologist can go to directly observe “it”. As Saussure realized in the case of langue, each member of any group or community only knows the conceptualizations (or langue/ language) of that group from a particular point of view and from a particular set of experiences; such knowledge is necessarily always partial, limited, and partaking of some idiosyncrasies. In this sense true/actual collective representations or cultural models can be said (MUST be said) not to actually or truly exist; they have to be only a convenient fiction. What make these collective representations seem real or feel real to us are the facts that a) as individuals we conceptualize each group with which we interact as, in some sense and to some degree, an animate “it” capable of a unitary characterization60 – and hence as social scientists we feel comfortable with looking for a description of such an “it”, b) our informants talk about these its as if they actually exist – both the cultures and subcultures and the beliefs and knowledge that characterize each, and c) within each social entity the conceptualizations which actively form part of that entity’s (group’s) collective conceptualizations are actively present in discourse and/ or interaction within the group and hence are actively involved in the constant communicative interaction by which – driven by our need to effectively 60

My suspicion is that this tendency, related to our anthropomorphization of animals, represents part of the innate mechanism by which we seek to understand others – in each case by trying to project ourselves inside its situation and thinking and then use our own internal sense of what we would respond to and do in that situation as a way of understanding the action of the other. We then constitute these representations of those others as kinds of ethograms of their species and their individual variants on the species.

Society: Kronenfeld view 133

communicate – our separate conceptualizations of the counters we use in that communication are continually brought into alignment with one another. Our informants all act as if they feel these collective representations or cultural models to actually, independently exist, and – because of the operative communicative constraints – the separate “its” (that each member constructs a representation of) do seem pretty much to pertain to and describe the same (apparently external and objective) “it.” In sum and in short, I am trying to explain how it is that I keep talking of collective representations and cultural models even though I firmly don’t believe in their objective actual existence. And I am trying to lay out what are my differences with people such as Fredrik Barth who want to “eliminate the middle man” (of culture or collective representations) and speak directly of patterns of interaction, shared experience, and shared knowledge – that is, to explain why I see a necessity for mediation of these interactions, and why I see “nonexistent” entities as the agents of this mediation.

3.

Identity and self, loyalty

This approach poses some interesting questions regarding the nature of the self and regarding “we’s” into which one sees oneself as inhering – that is, the underlying basis of one’s identification with various alternative social groupings. I have no systematic proposal to make in these areas, but do have a few observations to offer. 3.1. A general overview First, there is a “self” in us somewhere – something that operates in the first person, singular, that has volition and that has something like inchoate and undefined versions of reactions, feelings, and desires. But that “self” needs something from the outside to flesh out the forms that reactions, feelings, and desires take (lust, love, hunger, ambition, fear, awe, exhaustion, nobility, etc.), to determine what counts as instances of one or another form (is it love or lust, fear or exhaustion, etc.), and to distinguish acceptable from unacceptable ways of acting on them, to work out trade-offs among them, to decide which trump which others, to decide what to go public with and what to keep private, and so forth. I think that the major component of that outside something is culture – at least the way I have been defining and describing it. The “self” still has its

134 Chapter 6 – Society (with a note on the self) latitude and volition, but it operates within a social universe that enables it to sense what presentation or “fleshing out” is possible (given the person’s social context, attributes, and knowledge) vs. not, what will be acceptable by others vs. not, what will produce rewards (cash, admiration, power, etc.) vs. not. And it is this culture that the self takes on as social, ecological, stringency, and others kinds of context require or suggest. And what I mean by culture not being deeply “internalized” is that the self can always do its version of “when in Rome do as the Romans”. That is, it can code switch among alternative cultural systems – as circumstances make desirable and as it has the knowledge and other attributes needed to operate effectively in the new socio-cultural milieu. 3.2. Self But the nature of the “self” gets posed by such reflections. As suggested above, I do not see the self as the overlap (i.e., logical intersection) of the communities to which one belongs, but rather tend to see it as the internal entity which responds to the various contexts of participation. The communities to which one belongs are more than just roles. They are the contexts within which roles are defined. Roles exist within them, even if sometimes, in varying ways, the roles also define them. The interactive relationship between the “self” and the communities that define the codes, values, and loyalties of that “self” poses a kind of conundrum (or maybe just a dynamic circling – in the form of the “world snake”): 1) I negotiate my use of (and, thus, membership in and loyalty to) the various communities or categories to which I have (whatever degree of) access in order that I may “win” in the game of life. 2) But “win” implies winning relative to some values (and maybe by some rules). 3) And values (and rules) are social constructions that inhere in communities. At issue here are not only choices among conceptually opposed entities of a single sort (such as ethnic identifications) but also choices concerning which kind of entity to foreground – e.g., “Jew” vs. “Southerner”, “Black” vs. “Middle Class”, “Catholic” vs. “Anthropologist”, and so forth. The notion of “self” breaks apart into a variety of different aspects which appear to represent very different processes of reflection, experience, and construction, and in which the role of society and context is quite divergent. We have, first, the answer to the (often only implicit) question of “who are you?” Answers typically will involve some sorts of social categories –

Identity and self, loyalty 135

such as age, gender, ethnicity, class, occupation, etc. The answer to this question is framed and constrained by the situational context: who is asking the question, why is it being asked, what outcomes or actions are at issue, what kinds of frames are currently salient, and so forth. Next, we have the answer to the “who am I?” question – by which locution I mean to free the conception from immediate tactical situational constraints, and thus to get at a more context free sense of the aspects or features that seem basic to ones self-conception in the privacy of ones own mind’s eye. But such definitions are still cast in terms of socially (and linguistically) constructed concepts, and such answers are still constructed with regard to ones sense of what one sees as relatively constant or salient aspects of interactive context (whether social, political, economic, religious, language-based, or whatever) Further along, we have, not the answer to a question, but the “ego” or the “I” which is seeing, hearing, speaking, etc. This is not a “defined” self (delineated in terms of categories or concerns), but an experiencing one. It is not the answer to “who am I?” but the “I” that considers the question. The first two senses of self are intrinsically social, and, insofar as they are conceptualized in words, they are intrinsically within the domain of langue. The third sense is not social or linguistic, but is intrinsically individual – but it is also, thereby, necessarily an experiencing subject – not (in its construction or self-referenced understanding) an object for categorization or analysis. When we talk about it, as now, we are discussing the kind of thing it is, not the particular nature of some specific “it”, and (obviously) not the descriptive characteristics of some class of “it”s. I do not mean to imply that the selves (in the third sense) of different people are all the same, or, more particularly may not differ in ways that affect the different people’s ability to play different cultural roles or their inclination to utilize or exploit different cultural roles. I do claim, though, that these are differences among the users of culture rather than among its products. I also want to suggest that whatever kinds of variation exist among different people regarding such orientations are probably more or less constant across different human populations – and, thus, across different cultural groups. In talking about core meanings of “signs” that refer to social groups, I am speaking, primarily, about conceptual cores, core meanings. But I do want to point out that since these conceptual cores refer to descriptive attributes, they imply the existence of social cores – that is, of individuals who have the full set of core attributes and who thereby are recognized as prototypic

136 Chapter 6 – Society (with a note on the self) members of the social category in question (even if, for accidental reasons at some point in time I may have trouble finding such an actual person to actually point at!). There exists within a social system a shared dynamic about what forms a given social/cultural contrast (i.e., the social identities that people have available to them) and about the logic of what information (attributes, goals, values, etc.) is entailed by that contrast; the dynamic is shared both within a cultural community, and across socially relevant cultural oppositions. For an example, see Blackfeet inferences regarding who and what is on which side of the full-blood vs. mixed-blood opposition (below and, for more detail, see Kronenfeld 1996: 203–209); but see also in Kronenfeld’s Renaissance England example (1996: 209–218) that the dynamic can focus on default senses of what is presumed-to-be rather than on what really-is or what logically has-to-be. For the Sami case (below), we will focus on how the basic Sami-Norwegian contrast is first foregrounded (note: NOT created, but raised to consciousness and first seen as a significant or important categorical issue); and thus by whom it is raised. We see that this foregrounding process has a lot to do with what train of entailments are felt to follow. 3.3. Loyalty A concern with “self” and “identity” significantly relates to the problem of “loyalty”. How does loyalty work. How does one decide to what person or to what group to be loyal? Here, I am primarily referring to one’s penultimate loyalty – what one sees as the primary loyalty after loyalty to the society/ state itself; but I think the same considerations enter into the loyalty question when it becomes the prior question of loyalty to the state vs. to something else. Does one’s primary loyalty go to a leader (which one) to a local community, to an ethnicity (or race or language), to a religion, to a class? When a (Marxist) analyst thinks one has picked the wrong grouping, then it is called “false consciousness.” Meanwhile, as the Reverend Ike (or Adam Clayton Powell) showed, when members of a group feel their group really denigrated, they can wind up feeling a lot of pride in (and loyalty to) one of theirs who has made it – even if he or she has made it by turning their hardearned gifts into conspicuous consumption. How do competing loyalties get traded off? What happens when I am forced to choose? Do I dodge, mediate, or pick? What factors (locally and/or generally) affect/bias/stack the question of how I deal and (if I pick) who I pick.

Identity and self, loyalty 137

The loyalty question, when limited to groups, starts shading into the identity question. “Who am I?” – i.e., what category of person. Do I see myself primarily as a member/part of group X, or do I see myself as a person belonging to many different groups (and hence defined by none of them).

Chapter 7 Ethnicity

We now turn to a more extended example concerning social groups and cognition. We begin by considering ethnic labels and the groupings they represent in order to see why the semantic issue is foregrounded even by a social treatment of the functioning of ethnic groups, and to see what light the semantic approach of Kronenfeld (1996) can throw on our use of ethnic labels and on the collective cultural concepts that this use depends upon. We will see that ethnic groups are not clearly defined and bounded social entities – even though they often feel to us as if they are such. If segments of society, such as ethnic groups, which appear (both to native participants and to anthropological observers) to form such social units do not in fact do so, then the question is posed concerning what it is that they in fact are. And the same question is posed about society itself – the wider social unit in which such groups are situated. We begin with a brief treatment of Dr. Stan Wilmoth’s (1987) research concerning Blackfeet Indian political factions, and of the semantic theory of mine that he drew on in that work. Dr. Wilmoth went to the Blackfeet with the intention of studying the action of political groupings, not of addressing any semantic issue. He found that much Blackfeet political discussion was couched in terms of a major factional opposition, that between full-bloods and mixed-bloods. He then attempted to delineate the membership of each faction – that is, to draw a line around, each, separating members from nonmembers. His problem was that he could not draw such a line; that is, there appeared to be very little consistency from one occasion in which such labels were used to another concerning which individuals belonged to which faction or concerning which particular issue pertained to which faction. At the same time he noted that participants in such political discourse always seemed to understand what and whom they were referring to. This ambiguity in actual Blackfeet usage, coupled with the apparent meaningfulness of the usage in actual political discourse, forced him to reassess his approach and to initiate a systematic exploration of what Blackfeet meant by their use of these terms and of how they knew what other Blackfeet meant by their use of them.

Ethnicity

139

My semantic approach depends on distinguishing core referents of a term (a Saussurean “sign”) from extended referents. In this approach core referents do not represent some sort of average or typical referent but, rather, a prototypic referent. A “prototypic” referent is one that unites all the features that are in some sense definitional; by way of contrast, “typical” or “average” referents can, and often do, leave out many such features. For the Fanti egya or “father” category, prototypicality depends on the person’s being understood as one’s biological genitor, being in the male parental role toward one, and being married to ones prototypic na or “mother”; for the “cup” category in American English, prototypicality depends on being made of “china”, on being concave and hemispherical in shape, on having a handle, and on having a capacity of ca. 8 ounces. For conceptions of ethnicity attributes of prototypicality can (and often do) include one’s biological ancestry, one’s appearance, one’s language and accent, one’s style of dress, one’s behavioral and gestural mannerisms, and so forth. Particular occupational specializations or particular kinds of (imputed) skills or abilities can also be included in such prototypes. That is, the set of features includes those attributes which come from one’s parents, which one’s parents similarly received from their parents and share with a community whose other members similarly received them, and which differentiate one’s community from other communities. The totality of such features represents what we conjure up in our mind’s eye when we think of the (ethnic) group in question. But, as often happens with words in natural language, our typical application of such a label is generally not to such a prototypic referent. The constitution of language enables us to use the limited number of words in our lexicon flexibly to speak of the infinite variety of things that we deal with in the world and to make the infinite shadings of meaning that we require in discussing these dealings. One device by which we achieve this flexibility is by extending labels from their core referents to other referents that lack separate labels; the label applied to some such non-core referent is that whose core is, in the relevant context, conceptually closer to (or more similar to) the non-core referent than are cores labeled by contrasting terms. The Kronenfeld semantic theory distinguishes between “denotative” extension which is made on the basis of “form” features – the features which enable us to recognize a proper example of the category – and “connotative” extension which is made on the basis of “functional” attributes – the characteristics or uses of the class of things referred to by the label that account for our finding it useful to have a label for the class and for our referring frequently enough to the class in conversation for the label to be maintained in our language. Thus, the Fanti “father” category contains, as denotatively

140 Chapter 7 – Ethnicity “correct” referents, the brothers and parallel cousins of one’s father; the extension is made on the basis of the terminological equation of any father’s “brother” with “father”. The American English “cup” category includes as proper or genuine members cylindrical cup-like containers (which are also called “mugs”) and “cups” made of a variety of other materials (such as plastic, paper, or “tin”). As examples of connotative extension, we can take the common application in English of the “uncle” term by children to close family friends of their parents’ generation; such a person is, by the users’ accounts, not really an “uncle”, but a person who treats one the way an uncle should treat one and to whom one feels as one should feel towards an uncle – that is, such “uncles” are empathetic, concerned senior males. Connotative extension of the American English “cup” category might be exemplified by referring to an emptied “tin can” as a “cup” in some context (such, perhaps, as a camping trip) when one needs a cup (functionally, for a hot drink) and it is the most cup-like thing around. In the case of ethnic labels, I want to suggest that the possession of any of the separate features which (all together) go into defining (or delineating) the core referents can be used for denotatively extending the ethnic label from the prototype to the potential referent which has that feature. I don’t mean that all extensions are equally satisfying or equally convincing when someone chooses to contest the extension – but I do mean that any such extension is at least minimally plausible, and, in the absence of some active contestation, normally accepted. Examples of connotative extension of ethnic labels might include reference to the overseas Chinese as the “Jews” of Asia or 1960’s references to “the student as nigger”. For the moment I have nothing further to say about connotative extensions of ethnic terms – beyond noting that such extensions do pointedly drive home the associations (sometimes nasty and negative!) that extenders of the labels have with the groups to which the labels primarily apply. With that theoretical background in hand, we can now return to the Blackfeet. The basic characterization of the Blackfeet that follows is taken from Plastic Glasses and Church Fathers (Kronenfeld 1996: 204–208). An important semantic, as well as socio-political, context for these concepts (“full blood” and “mixed blood”) is provided by the basic framing contrast between “Indian” and “white” – that is, by the contrast set involving the item (Indian) which includes them. In the universe we are considering (the white controlled world in the United States where Indians exist in relation to

Ethnicity

141

whites) white-ness is the unmarked58 category, the conceptual default that applies to people not otherwise marked for some other category. Indian-ness is defined, implicitly, as a marked contrastive category, and hence relevant attributes or gradations of Indian-ness pertain primarily to locations along an Indian-white continuum. More specific concepts, such as “Blackfeet”, while very important to Blackfeet people and maybe to anthropologists, lose their specific ethnic significance (for whites, and, though perhaps to a lesser degree, for Blackfeet as well) and become almost synonymous with “Indian” in this wider racial/political linguistic context – and even though there exist particular political and economic situations where being Blackfeet vs. some other kind of Indian becomes quite important. Thus, in our Blackfeet situation, the unmarked Indian is a Blackfeet, and core Indian is normally understood as core Blackfeet. The basic role of the Indian vs. white polarity means that not-Indian automatically entails white. The primacy of that opposition means that not being Blackfeet similarly entails being white rather than being some other kind of Indian. Both the Indian-white polarity and the full blood vs. mixed blood one can be used to frame more or less continuous gradations – in the sense that being “less” Indian entails being “more” white, and so forth. … Ancestry, at least to the whites who designed the blood system, referred to “racial” characteristics. Full bloods, most marked as Indians conceptually, were also in principal marked physically. The less white one looked, the more Indian (by implication) one must look, and so dark (“red” to whites, though not to Blackfeet) skin and (straight) hair become attributes of Indian vs. white, and hence of full blood vs. mixed, identity. The conceptual opposition between full blood and mixed blood can be used in an extended sense in reference to these supposedly ancestral features, even in the absence of the denotatively defining information. The biological (here, genetic) vs. cultural (ranging from time in the sun to style of dress) aspects of appearance are often not clearly distinguished by non-anthropologists (both because of confusion concerning their basis and because of the feeling that what you get from your parents is “ancestral” regardless of whether you have been born with it or have acquired it), and it is not surprising that extension of the full blood vs. mixed blood opposition on the basis of appear58

Marking is the phenomenon noted by linguists in which one category operates at two successive levels of contrast. So, for instance, the word “man”, at one level, is opposed to “woman” and includes “boy” – as can be seen in the boy’s use of the “men’s room”. However, at another level, “man” contrasts with “boy”, as can be seen in the expression, “this is a man’s work” or in the seagoing tradition of “women and children first”. In the “man”/”boy” opposition, “man” is spoken of as the unmarked member, and is defined as the one which can represent both terms when the opposition between them is neutralized.

142 Chapter 7 – Ethnicity ance should stretch to include (as bases) attributes such as clothing and grooming style – and maybe more general features of life style as well. Another kind of very important, but non-biological, ancestry is represented by language. The language one speaks is important as a symbolic marker both because of its salience and ease and clarity of recognition and because it cannot be as casually acquired as can clothing or style. Furthermore, native speaker fluency is tied up closely enough with growing up in an appropriate environment for it to feel equivalent to biological ancestry to many Indians and non-Indians. Second language fluency (in Blackfeet or in English) represents an important enough commitment of time and effort, or an important enough period of intimate interaction, to have important identificational implications. The implications regarding cultural contact and interest of even second language use are serious enough to cause a person in a full blood context and/or seeking a full blood identification to be very leery of using English; conversely, whites who use Blackfeet are understood to be asserting a strong pro-Indian stance. Since Indian ancestry was used as a measure of Indian-ness (or, in our particular present case, of Blackfeet-ness), full bloods were thought of as more clearly and thoroughly Indian, and hence more traditionally Indian. Mixed bloods, on the other hand, being of mixed (Indian-white) ancestry, were seen as culturally mixed, and hence more assimilated or acculturated to white society. Thus, the full blood vs. mixed blood contrast connoted traditional Indian vs. assimilated Indian, and one simple and basic extension of the terms (and, therefore, of the contrast set in which they participated, including the implicit third – polar – term, white) was to situations in which the terms were applied on the basis of these connotations rather than on the denotative basis of literal biological ancestry. Such extension is strongly tempting (and hence common and basic) because it makes the concepts more culturally relevant and useful, and because it subtly (and perhaps subconsciously) undermines the authority of the externally created and applied definition of who’s an Indian (or a Blackfeet). This first kind of culturally based extension of the conceptual opposition provides the basis for most other connotative usage. In the area of politics, the “traditional Indian” end of the continuum refers most directly to traditional political forms, means of making collective decisions, and so forth. But, since these traditional Indian forms exist in contrast to a white norm of individualistic “one man one vote” expression of narrow self interest, the Indian or full blood side of the opposition can come (within Blackfeet political discourse) to connote communally motivated (or implemented) values or actions in contrast to individual ones – even in a Plains Indian culture that greatly emphasizes individual experience. And, in the same context, “communal” action can contrast with “democratic”, where the former is Indian or full blood and the latter white or mixed blood.

Ethnicity

143

Since white culture is considered (by most whites and many Indians) to be more modern (technologically, but by inference often also politically and culturally), the idea of traditional Indian gets linked with the idea not just of culturally conservative, but also with the more general sense of conservative that implies opposition to change, adaptation, and modernizing (i.e., “stickin-the-mud”). By this extension, full blood vs. mixed blood can come to entail “opposed to” vs. “supportive of” modernization (whether of political forms – e.g., elected councils – or of economic ones – e.g., newer forms of land management). “Culturally conservative”, in the context in which whites are associated with technological power, implies non-technological and thus natural. Since being an Indian means (to both Indians and whites) having less technological control over nature, Indians get associated with the natural end of the natural vs. man-made opposition. And hence “Indian” (and thus “full blood”) comes to signify (or to be usable as a label for) a pro nature or pro unspoiled wilderness position – in opposition to “white” (or “mixed blood”) which signifies in this context technological intervention. On specific economic or political issues we now have several bases on which the full blood vs. mixed blood opposition can be extended: 1) The mixed blood label can be applied to those in favor of some innovative (or “progressive”) technique or management scheme – as opposed to those, full bloods, who are pushing maintenance of “the traditional way”. 2) The mixed blood label can be applied to those who are supporting some program backed by the whites – vs. “full bloods” who oppose the program. “Whites” can refer either to local whites or to the white establishment – especially as embodied in the government. The Bureau of Indian Affairs, on the one hand, as an arm of the government, can be “white”, but on the other hand, as the Indian representative in government and as the government agency itself most populated with Indians, can be “Indian,” especially in a context in which it is opposed to some other part of the white establishment. Thus, depending on context, backers of the BIA can thereby be labelled as either full bloods or mixed bloods. 3) Any issue which is in fact backed by people generally known as full blood leaders can become a full blood issue; other people taking the same position can, then, by extension, be labelled as full bloods. Those taking the opposite position become, in this context, mixed bloods. The same applies vice versa for issues espoused by clearly recognized mixed blood leaders. In terms of the general idea of core vs. extended concepts, we can say that a core Indian is a person of complete Indian biological ancestry who speaks an Indian language, looks physically like an Indian, and dresses like one. This core Indian espouses traditional Indian values, including respect for nature and respect for traditional Indian forms of leadership and collective action. This core Indian contrasts with a core white, who is of European descent and northern European appearance, who speaks English, is technologi-

144 Chapter 7 – Ethnicity cally adept and school educated, and who believes in individual desires and actions being impersonally amalgamated to produce collective action when necessary. While the preceding features are definitely not of equal weight, in general the more of one or the other set of attributes a person possesses the more unarguably and unnegotiably the person is presumed (in the context of reservation political discourse) to belong to one or the other opposed categories. On the other hand, in an appropriate context, any one of the attributes can put a person (on an ad hoc basis) in one or the other category – in opposition to other persons who have a contrasting attribute on the relevant feature. People who have attributes from both sets are freer to move back and forth between the two identities, but are seen as less convincingly belonging to either. The unmarked, or default, person is a white – and so, in the absence of relevant information about a person, one presumes him or her to be white. If any particular relevant attribute is known (in the absence of others) then an identification is made on that basis and core values appropriate to that identification are presumed. In the context of the preceding opposition, the category of Indians is subdivided into more or less Indian-y Indians, which gets coded as a full blood vs. mixed blood opposition. The default category, of Indian-y Indians, is the full bloods. The mixed blood category is defined by its contrast with (difference from) the full blood one. But since white vs. Indian is the context, and since Indian is defined, in a sense, as maximally non-white, non-Indian-y Indians can only differ from Indian-y ones in the direction of whites, and so mixed bloods become relatively white-y Indians. Core full bloods, then, are core Indians. Core mixed bloods represent a more complicated notion. They are maximally white-like Indians who are still, nonetheless, Indians. That is, they are Indians who have white ancestry and a somewhat white appearance, who speak English, and try to live like whites while claiming Indian-ness. Since Indians are a marked category distinguished racially (and hence by virtue of birth) from whites, and since they are in some ways a lower status (or even stigmatized) group, there is a kind of residual presumption that they cannot actually become white and hence that even the most white-like of Indians will be only imperfectly white (or imperfectly accepted as white – by whites or Indians). The mixed blood core will therefore fall somewhere in between the Indian core and the white core. … The core referents of a term (or concept) represent the truest members of the category, and are the presumed referents of any usage of the term in the absence of further specification or contextual information. If some particular non-core attribute is mentioned, e.g., the Indian is speaking English, the presumption is still that the remaining, non-specified, attributes are core – with the exception of ones implied by the specified attribute in context. If

Ethnicity

145

the Indian is described as dressed like a white and as being at a meeting with important government officials, then English speech might also be inferred by members of the local universe.

The primary point of the Blackfeet example is to illustrate how the framing opposition (Indian vs. white), its definition, and the logic of its related conditions, joined with the way in which the subsidiary opposition – between full blood and mixed blood – was initially defined, closely constrain the ways in which the subsidiary (full blood vs. mixed blood) opposition can be extended or taken. The core full blood has to rest on the core Indian and the opposed concept has to differ in the direction of whiteness; the rest follows. Various other attributes were discussed which signal full blood vs. mixed. They function connotatively – marking someone as acting as if full or mixed in some given context, and providing a basis for the use of the “full” or “mixed” label to communicate or be understood in that context, even in the absence of the denotatively defining attributes. These connotative attributes are not defined by anyone by fiat, and are not learned as any kind of rote list, but rather are derived (as stated) from the pattern of oppositions, the pragmatic givens, the communicative force of the opposition, and the denotative definitions – as well as from the uses of political rhetoric. 1. Summary As a summary of some of the implications of this approach to ethnic categories let me offer the following. First, a given ethnic label can, potentially, be extended to several different individuals who have nothing relevant in common – that is, alternative extensions can be made on the basis of totally non-overlapping feature sets. A “Native American” can be a culturally mainstream person of Native American ancestry – or a person of minimal Native American ancestry who was raised in (and actively participates in) some Native American cultural group. Second, individuals who have ethnically mixed feature sets (that is, a mixture of features coming from the cores of different and contrasting ethnic categories) can receive either/any of the alternative ethnic labels. Such a mix could either be the result of mixed parentage or of participating in an ethnic community different from that of ones ancestry (and, then, from consciously or unconsciously taking on the attributes of others in that community – whether neighbors, co-workers, or whatever). Either “Native Ameri-

146 Chapter 7 – Ethnicity can” described above could also claim some non-Native American ethnicity (Anglo or Hispanic, depending on the person’s relevant other affiliations). Third, a given individual with a mixed feature set can take on, in very short order, very different ethnic identities – even (where the features are such as to enable it) identities that seem deeply opposed and irreconcilable (such as “black” and “white” in the modern US). My daughter’s freshman year roommate at her university had such mixed parentage, and moved back and forth between those two worlds. One’s freedom to take on ethnic identities, however, is not totally unconstrained. The defining features that make up an identity’s core or prototype do represent real limitations; one’s assumption of the ethnic identity requires one’s possession of at least one of these features, that is, at least some overlap with the relevant prototype. The various defining features of a term do not all have the same kinds or degrees of effect, or the same kinds of definitional attributes. One kind of variation concerns the ease with which one can take on or assume one or another feature. Some features, such as skin color or elements of body build are, indeed, biologically given and effectively unchangeable. Others, such as clothing styles, are easily acquired. And others, such as convincing control of a local language or dialect, are acquirable but only with difficulty and some investment of time or effort. A separate kind of variation concerns the sharpness with which the markers of one ethnic group delineate that group – in some given social context – from other groups. Such sharpness can come from a particularly distinctive clothing style or from particularly contrastive “racial” features. This issue affects the salience with which their ethnic group is perceived by the wider community. It also can constrain the ability of members to “pass” or to slide over into neighboring groups. A third way in which ethnicity features or markers vary from one another is in the folk understanding (or theory) of how one comes to have the features in question. Relatively passively acquired features range from things which you automatically get from your biological parents to things you automatically get from your sociological parents (those who claim you and raise you – such as adoptive parents), to things you get (or can get) from those (caretakers or friends) with whom you associated as a child. On the other, more active, side is the question of whether or not (or to what degree) some given feature is something you can choose to adopt or learn. There exists also a range of folk views concerning the degree to which your possession of some given ethnic factor (or of the ethnic identification itself) requires some ongoing active participation on your part: Is it (the

Summary 147

factor or identification) just a part of you, like your eye color, or is it something you have at least to identify with (that is, recognize or affirm your possession of), or is it something that you have actively to participate in (some way or other) in order to maintain? These differences in the nature of various potentially defining ethnic features do affect the convincingness of any extension of an ethnic assignment or label based on these features, but, in turn, in a variety of ways. Hard to acquire features (such as skin color or language) weigh more than easily acquired ones (such as clothing). Biologically given ones (such as skin color or body build) weigh more than learned or assumed ones (such as clothing or manner of speaking). Features that require long and dense community interaction (such as command of a local language or dialect) weigh more than ones that require only little or no community interaction (such as wearing some piece of ethnically marked clothing). And, finally, I would suggest a kind of salience aspect: Features that are considered salient in the wider community – i.e., that have been socially defined as ethnic markers – will carry more weight in extra-ethnic or interethnic contexts than will less salient or marked features, and such features will probably carry more weight as internal assertions or affirmations of ethnic identity. On the other hand, features which are not salient in the wider community may well weigh more within the ethnic group when private interaction or mutual support is the issue. These convincingness factors, while helping members of the wider culture to resolve situations of ambiguous ethnicity, do not totally resolve such issues – and, in fact, themselves introduce some further ambiguities. In particular, it is important to note that the convincingness factors do not necessarily covary, and that some can have effects contradictory to the effects of others. For instance, one can have biological features without having those features that come from community contact. Ethnic groups represent a sort of folk taxonomy of the “kinds of” people that exist, and, as such, our ethnic classification has some of the same characteristics as our folk taxonomies of kinds of animals or plants. Such groups are contrastively perceived relative to other such groups, and lower level groups can be included hierarchically in higher. For example, “Blackfeet” contrast with “Crow”, while both together belong to the category of “Native Americans” – which contrasts with “Whites”. The taxonomy of kinds of people can include marking relations – as in our present example where Blackfeet is being taken (in the Blackfeet context) as the unmarked Indian.

148 Chapter 7 – Ethnicity Still so far unaddressed is the debate about whether ethnicity is to be understood primarily as just a social structure, based on a set of social boundaries, wherein the cultural content merely serves as the vehicle for marking and maintaining the boundaries – or whether it is to be understood primarily as some kind of cultural content, in terms of which, in turn, social boundaries are erected. Is some sort of cultural substratum required for the existence of ethnic groups? My own sense is that this is a kind of “chicken and egg” circle. That is, I suspect that there exists a kind of temporal spiral in the course of which each view is correct, but only one at a time. At any given moment only one kind of information – either social or cultural – forms, in the language of Gestalt psychology, the figure and one the ground. I think this is true whether we are speaking of the minds of native participants or of external analysts – though we can build models that encompass both social and cultural perspectives. The social information we learn and build into our individual internal representations of our society’s collective representations would seem to include both kinds of information: cognitive information about the cultural content that distinguishes our groups and social information about the boundaries and membership of these groups. When we detect a misfit between the two kinds of information, we can take either as our fixed focus and manipulate or modify the other to fit it. That is, the more Blackfeet things one does, the more Blackfeet one is considered to be – but, at the same time, whatever a Blackfeet does becomes a Blackfeet thing. “Natives” use ethnic labels as if they capture some deep and essential fixed reality – even though the categories they label are, in fact, fluid and flexible. Even social scientists seem never to have fully freed themselves/ ourselves from basic folk conceptions of ethnicity and ethnic groups. There is something deeply consonant to us – to our perceptions or to our way of understanding our social world – embedded in our folk views of ethnicity and of social groups. I want to suggest that this consonance goes back to the basic interpersonal devices by which we subjectively understand the actions and thoughts, first of other individuals, and then, derivatively, of collectivities of individuals. This re-raising of the nature of our perception and understanding of collectivities takes us back to our previous discussion concerning the nature of society.

Chapter 8 The social construction of ethnicity: Intuition, 59 authenticity, authenticators – the Sami example

1. The Sami example In this extended example we explore the particular question of how ethnic categories acquire the acceptance and “legitimacy” that makes them so effective in political mobilization – even while they are being visibly created before our very eyes. What kinds of categories constitute potential ethnicities, and what activates the transition from a descriptive category to an active group? What attributes and situations make ethnicities plausible or convincing? Does the distributed cognition perspective throw some useful light on how the process of ethnic mobilization plays itself out? Are communicative constraints a significant part of the way by which “distributed” individuals shape one another’s conceptions? But, additionally, the example should help clarify the role played by the alternative communities (social entities) within which words, including ethnic labels, are constructed and construed. Individuals identify with particular communities, but are not intrinsically defined by any of these communities; as we code switch among languages and dialects so we also switch among various communities to which we belong. Where the wider community of concern is itself made up of smaller local communities the levels of collective construction, and the roles that individuals play in such constructions, get interesting. In the present case the issue concerns the “ownership” of ethnic labels. When there is some sufficient advantage to be gained, whoever has a plausible basis will try to claim some given ethnic label. This relates to the ability of an ethnic category to mobilize people. Consider, for example, the 59

Much of my understanding of ethnicity and many of my ideas about society have come from discussions with Halvard Vike during his visits to the US. My Sami information is drawn from insightful ethnogrphic work by Arild Hovland (n.d., 1995), but he is not to be held responsible for my interpretations of his work.

150 Chapter 8 – The social construction of ethnicity Northern Utes, with whom I worked during the summer of 1962, who had a small population of people claiming the identification before a big court settlement, who had a much larger population during the payout and the lifespan of projects the money brought, and who returned to their low population levels after the money ran out. Kronenfeld (1996 – see the excerpt quoted in the previous chapter) discussed the relationship between Blackfeet Indian ethnicity and a factional opposition between full-bloods and mixed bloods. The opposed categories were specified by the United States government in a different age, with a mixture of motives, and a train of inferences regarding defining or distinctive attributes followed from the nature of that opposition and the larger Indian-White one within which it was embedded. The problem was that the application of the terms to actual referents flipped around considerably. The semantic insight was that the terms were not used absolutely, but situationally – on whatever basis did the job at hand – the job of identifying relevant actors in a situation and accounting for who was on which side. The case of the Sami of Norway – the people we have known as Lapps – offers us a chance to delve more deeply into some of the issues raised by the Blackfeet case. In many ways – in Sami eyes, in Norwegian eyes, and in anthropological eyes – the position of the Sami in Norway is analogous to that of Native American groups in the United States. The situation and processes of Sami ethnicity are similar to those of Native Americans, but have the analytic usefulness of being more recently recognized and formed. Being more recent, the formation of Sami ethnicity has taken place in a world of more modern sensibilities, morality, and concerns, and has thus has been more self-conscious and more public. The Sami case thus seems a good place in which to examine the social processes by which ethnic identities are raised to consciousness, instantiated, and politically structured. Let us explore a slightly hypothetical version of the emergence of Sami ethnicity. The Norwegian government, perhaps in response to what we might call proto-Sami activists, decides that it needs to recognize the Sami people as a special category of citizen – as the indigenous people of the Northland. Some historical context for this action is represented by the fact that Norwegian nationalism itself is recent. The Norwegian nationality is cast in terms of colonial independence – of a colony alternatively of Sweden and Denmark eventually achieving independence. Norwegian nationalism is a nineteenth century movement that only produced independence around the end of that century; the folk, regional, and ethnic expressions /symbols of

The Sami example

151

Norwegian nationalism and of its regional subdivisions are still in the process of being actively elaborated. This means that the Norwegian people are perhaps particularly sensitive to issues of ethnic autonomy and selfdetermination. The Sami language and lifeway existed through this period in various rural areas in connection with people who were seen as poor, primitive, and different. Even though Scandinavians, as part of the Indo-European expansion, have been in Scandinavia for a couple of thousand years, and even though (as far as I know) we have no sense of who, if anyone, they displaced when they moved into the area, and, similarly, we have no idea of when ancestral Sami moved into which areas, their “primitiveness” seems to make Sami appear in Norwegian eyes to be the autochthonous inhabitants of the Northland. At some point, I gather in connection with world-wide civil rights movements, people associated with the Sami language and culture began agitating for an end to anti-Sami prejudice and for some positive recognition of Sami ethnicity. Let me remind you that I am thin on my Norwegian history, and that my lack of detailed knowledge is why I introduced this part of my discussion as a “slightly hypothetical” history. The Norwegian government was not (perhaps, could not be) insensitive to the arguments for Sami recognition. The government also understood that, in the existing national and ideological context, it could not itself be the definer or custodian of Sami identity – that that role had to be filled by Sami. At the same time, since this was a question of political rights for a recognized ethnic entity, there had to be some political representative of that entity. Hence a Sami parliament was created, along with definitions of who could determine the right to vote in this parliament – and thus who could define who (by what criteria and against what standard) was to be considered Sami. The Norwegians wanted Sami to make these decisions, but they had nonetheless to determine who were the initial Sami to whom such decisions would be turned over and with whom negotiations would take place. Presumably activists already pressing for the recognition of Sami ethnicity were part of the “other” with which the Norwegians dealt, but such people don’t automatically make a category – and are not, in any event, a representative cross-section of the population who might potentially, by one or another reasonable potential definition, be considered Sami. Insofar as the Norwegian government picked whom to deal with, it was inclined to pick those people who, by its lights, were most distinctively and unambiguously Sami by both ancestry and life-style (including language, culture, religion, ways of making a living [reindeer herding], ways of dress-

152 Chapter 8 – The social construction of ethnicity ing, and so forth). This recognition of potential Sami-ness did not take place in limbo, but, rather, in the particular context provided by the contrast of “Sami” with “Norwegian”. Since the Norwegians represented the modern world – modern technology, life-style, religion, etc. – Sami-ness was associated with contrasting “non-modern”, “traditional” forms. Since Sami were the victims of oppression, Sami-ness would be associated with members of local underclasses. What this means is that descendants of Sami ancestors who spoke Norwegian, who made their living in ordinary Norwegian ways (say, fishing) or in modern ways (say, manufacturing), who had merged into local Norwegian social and political structure, and/or who had achieved some sort of status in Norwegian society were to a greater or lesser degree removed from the Sami side of the Norwegian-Sami discourse on Sami identity and ethnicity. What, almost inevitably, got recognized was a kind of “nativist” category. Insofar as initial Sami activists were recognized and accepted as the representatives of the emerging Sami identity a similar dynamic with similar content would emerge – because of whom in particular these activists represented and what experiences produced their activism. The main difference would perhaps be a more politically militant and anti-Norwegian cast to the emergent Sami entity that might otherwise have been the case. My point, here is not that this is wrong or strange, but only that there is nothing that automatically makes this categorization the only reasonably imaginable one that one might come up with. All cultures contain people who adapt and change – and thus why should adaptive, modernizing Sami automatically be disadvantaged or disenfranchised relative to adaptive, modernizing Norwegians. We might note – leaping ahead a little – that Sami who did adapt are, in fact, now challenging the conservative, nativist definition. One can easily imagine that a different founding dynamic (based on some valued modern economic niche, for example) might have produced a different discourse with different participants. Evanescence. Once a Sami group gets accepted by the outside world (Norwegian government and world-wide indigenous peoples groups) as the Sami spokesgroup and as the keeper of Saminess, then a new dynamic starts. One of the motivations driving initial pressures for Sami rights was economic and political discrimination – denial of access to the benefits of modern life. So a dynamic begins in which the essence of Sami-ness is increasingly invested in a small (and perhaps narrowing) set of salient symbols – including reindeer herding, special clothing, special religious and folk ethnic designs, a special chant form, and the Sami language. These symbols are increasingly recognized by Sami and Norwegians as special

The Sami example

153

Sami property, and their use (especially clothing, designs, and chants) become, in effect, copyrighted. In the meantime, the Sami cultural artifacts and images are recognized by Norwegian society as good for tourism. Control of the Sami polity and of the rights to be producers of authentic Sami products becomes an important income resource; at the same time, enhanced Sami general status makes more of the range of occupations and benefits of ordinary Norwegian society available to Sami. Thus, effective institutionalization of Sami difference becomes part of the process by which the Sami lifestyle becomes more and more similar to that of surrounding Norwegian (and Western, in general) society. We have the – in some ways anomalous – outcome that a culture which was initially recognized in particular for its “primitiveness” and its non-material values – its differentness from the mainstream society – and which within itself discriminated against relatively assimilated members or potential members, increasingly vests its identity and definition in a few material symbols (and perhaps some specific religious values) while using the benefits provided by ethnic separation and recognition to buy the modern paraphernalia that define mainstream society. Through time the definition of such ethnicity gets more sharply drawn and more generally accepted, while at the same time becoming thinner and less consequential. As it becomes a clearer difference it makes less of a difference. As lifestyles (and personal values) blend and blur the ethnic contrast, the social lines also get blurred. When the group was first formally constituted, the results of intermarriage, alternative forms of education, and alternative forms of participation in national life could be contained and walled off; but thereafter rights flow from parents to children (via official criteria) regardless of attributes of felt authenticity. Social integration follows from the success of the program, and the effect of social integration is to make the group – both by measures of social integration and by measures of cultural sharing – more like a club (within the – here, Norwegian – framing culture) and less like any distinctive culture or ethnicity. This thinning into nothingness can be seen as a kind of ethnic evanescence. This process seems something of what happened to America’s earlier immigrant ethnicities – first Germans, then Scandinavians, Irish, Italians, and Jews – and that is happening now with economically and politically successful Native American groups in places like Oklahoma. Factors such as religion – which slowed down Jewish integration by a generation or so, perhaps – and race – which still interferes with Black integration – do interfere. But still what comes out of all of this is that economic integration produces cultural and social integration. Black Nationalist separatism in the US produced

154 Chapter 8 – The social construction of ethnicity Blacks with middle class values and middle class economic skills; the children of such separatism are now better integrated into the mainstream, on the average, than the children of those who chose not to set themselves apart. Contestation. In northern coastal areas of Norway – where people of Sami ancestry were more linguistically and economically assimilated to Norwegian society than residents of the inland core Sami area, and where they were economically better off – older residents do have some non-foregrounded, non-politicized memory of a history which is now recognized as Sami. Many coastal Sami youth are rediscovering their Sami-ness. Often they are learning the canonical, authoritative version of Sami ethnic identity (symbols, texts, religion, crafts, etc.) at the regional university and then bringing this information back home. At home this authoritative version sometimes get confronted with the significantly different local versions of Saminess. Because of its institutionalization, its control of economic resources, its formalized self consciousness, and its general national acceptance the authoritative version predominates, but one gets a sense from Hovland’s (1995, n.d.) account that there is some pressure toward asserting the legitimacy of the local versions. So far, it appears that the official Sami establishment is open to the recruitment of Sami who have newly embraced their roots, but the establishment seems much less open to alternative conceptions or instantiations of Sami-ness. Different communities of people who were unself-consciously Sami – members of a stigmatized culture at the bottom of the social ladder – in the last century experienced different economic pressures and opportunities and moved in different directions – the coast toward assimilation, integration, and economic development and the interior toward maintaining isolation and poverty. More recently, when the Norwegian government, in response to pressure from local activists, decided to correct the injustice, it invested its resources in those who seemed best to epitomize the culture and those who seemed most to have suffered – those in the interior. This government action formalized the ethnic line between Sami and Norwegian, institutionalized a particular version of Sami culture, and foregrounded a sense of Sami ethnicity; by giving the Sami establishment legal control over Sami symbols the government also created what turned out to be rich economic resource for that establishment – as well as a source of admired cultural capital. So now, instead of local Sami just being whatever they were in their local communities, an overarching Sami entity was created which spoke authori-

The Sami example

155

tatively as to what was properly Sami. Local traditions, especially those of the relatively assimilated coastal communities were denied legitimacy in things Sami – as that became a valuable resource. So, instead of each small community doing its own thing, we now have local communities contesting with other communities over cultural definitions. Since this cultural focus is not so much a legal issue as a contest regarding collective representations, it is not played out in court but in the minds of the individuals who make up the various Sami communities and the wider Norwegian one. The various communities each have their own special interests, but they need to (and have the urge to) speak to – and convince – a wider social universe, in which individuals measure how much each competing claim squares with their own intuitions based on their own experience and their own reasoning. This distributed “reality” constrains the kinds of arguments that members of the different communities make (in ways explored in Kronenfeld 1996) and structures the dynamic by which a coherent broad collective representation emerges from the give and take of contending communities (each of which is, itself, made up of contending individuals). 2. Other Sami thoughts People do not belong to social groups, including ethnic groups and ethnic sub-groups; instead they identify themselves with them (that is, adopt them or espouse them or embrace them or take them on) – with varyingly good “credentials” for doing so (that is, with varying credibility as putative instances of the group in question) and with varying consistency over time (i.e., some people’s identifications change more than others, or appear to be less complete or total than others’). Credibility is a function of how well the identifier’s apparent traits match the traits associated with the group in question; and the relative weight of the various traits can vary – from group to group, from one perspective to another, or from one context/situation to another. This is all by way of saying that no one is simply or really a member (regardless of whether or not the person chooses to join some organization that supposedly represents or instantiates the group in question); everyone involved is choosing to act as if they are a member, and they all can imagine other choices. This means that, in fact, no member can be objectively held up or instanced as a “true” X. The closest you can get to that is that some collection of relevant people can decide (socially and/or politically) to affirm some member (or themselves) as a “true” X (or, relatedly, to affirm some or another trait as a “true” X trait).

156 Chapter 8 – The social construction of ethnicity Who makes up the “collection of relevant people”, and who constitutes it as a social entity, is important for the credibility of its affirmations. Evaluating, or defining, or setting up this collection of relevant people, though, also represents a tricky idea – since we just got through saying that there are no such things as true X’s, putting the collection forward as a presidium of true X’s will not do the job. The credibility of the collection (or the social entity that it can become) cannot depend on “truth”, but can, then, only depend on symbols. It would seem especially to depend on symbolic factors that have clear historical associations with the putative sources of the group, and on factors that are relatively hard to acquire in a casual or incidental manner (which means either biologically given factors or factors, such as language fluency, which are [normally] the products of long and deep cultural experience – or, less convincingly, which are arcane matters which outsiders might be expected not to know [which, of course, anthropologists immediately screw up!]). In short, the validators must, to be convincing, themselves be “validated” – but in a particularly impersonal, but quite non-objective (i.e., biased) way; this second order validation (of validators) rests in the hands not of insiders but of the wider society and its ideas of authenticity. The wider society knows not much of the group in question, but knows the group is “different” and out of the mainstream. The validation of credible validators reflects that knowledge. The preceding process necessarily biases the game in favor of a separatist (non-assimilationist but also boundary building), relatively non-adaptive (i.e., “traditionalist”) viewpoint that embraces strange (to the wider society) symbols. And, then, people who have excellent historical claims on the identity in question, but who have chosen (or whose parents chose) to blend in, to “stay up with the times” or to “get ahead” in the wider society lose not just their claims to authenticity (regarding their ancestry) but also, behind that, their ability to influence the definition of authenticity. In other words, “true” X winds up translating into truly not Y, where X is seen primarily (by members of Y) as being not-Y! The group of validators has to be big enough to look like a culture and not just some strange individual. This means that the actual nature (extent, depth, breadth) of the culturally conservative position (that the dynamic selects for) will be limited by the nature of the most culturally conservative collection of ethnics who can be found when the process is initiated. After the process has been initiated, the recognized group has much more control; it can recreate or revivify selected symbols that had actually passed into disuse, but its cultural “primitivism” is limited by the kind of lives its

Other Sami thoughts

157

members are willing themselves to lead (as exemplars of “true” X-dom). Since, as controllers of the gate, the group of validators now/newly has control of resources, the economic position of its members improves, and their ability to invest in the creature comforts and technology of the wider society increases which means that the bulk of their lives becomes more and more like the lives of everyone else in the dominant/wider society and thus that the cultural identity they represent gradually becomes more and more attenuated and more and more concentrated in a small set of salient “iconic” symbols. There are unlikely to be several such groups of validators (vs. only one) because the wider society will only recognize one – the most “authentic” by its standards – if we are talking about small ethnic minorities which have no power and whose members were in a second class position in the wider society, and where there is some sense of “justice” or guilt assuaging behind the new positive interest in the group. (Only if there is real wealth and power involved will the wider society look for another more suborned/ friendly group of validators to push – and then the presence of such internal power will make it harder for outsiders to effectively assert their control.) Beyond the preceding topic, it is clear that an increasingly salient political issue for ethnic groups such as Sami or American Indian groups, as well as many others, is the control of the use of symbols and the definitions of authentic traits, history, etc. This seems to be part of the current tension between anthropologists and Native American groups in the US. The following seem to be factors which constrain and shape this universe of claims (re Sami-ness) and make for plausibility and convincingness. a) What people (in relevant communities) know about the relevant world: That the Sami are not Norwegian -> “different” in significant and salient ways. I.e., What the Sami are not – what they are contrasted with. b) What people in these communities have experienced as relevant exemplars/exemplifications (combination of actual action and the salience of the ethnic link) – i.e., What the group is, as indicated by the attributes by which its members are recognized: For the Sami: – reindeer herding – (historically) poor, treated badly, “out” -> deserving of aid, protection, and recompense – possessors of a special religion, characterized by special symbols and decorations

158 Chapter 8 – The social construction of ethnicity c) What people in these communities “know” about the nature of (what we call) ethnic category membership: In this case, that it is inherited from ancestors – a kind of birthright. Relevant here is the discussion of Protestants and Catholics in Kronenfeld (1996: 209–218) – which illustrates a semantic field within which opposed groups are contesting the definitions of key concepts, but where such contested versions are tied to a presumed pragmatic reality by particular commonly understood “facts”. Therefore: wealthier Sami on the coast who fish in modern ways, who speak Norwegian and live what looks to be a Norwegian life style are a harder sell as “true Sami” than the interior, poor, reindeer folks – a harder sell both to other Sami and to Norwegians. The jury that adjudicates these claims, ultimately, is a community of communities – even if, unlike for ordinary language, political actions (such as the Norwegian government giving control over political participation in Sami decisions and economic control of things Sami to a particular “Sami” community) do help shape where it comes out.60 The issue relates to the phenomenon we see in the US, where a great many people claim some sort of Native American ancestry without claiming Native American (or some particular variety of Native American) ethnicity. Other Americans, with comparable ancestry, in different political, economic, social circumstances do claim the ethnicity. (An example was seen in an undergraduate research project at the Riverside campus of the University of California by Rudy Cervantes examining the adoption of Native American ethnicity within local prison populations – including, there at least, the adoption of religious/behavioral practices aimed at validating the claim). 3. Problems and issues Regarding the Sami-ness (and the control thereof) of the central Sami group, there does exist a challenge. But the challenge is not going real well, even in the minds/hands of the challengers. Implying that the challengers them60

Examples of strong influence on definitions of ethnic membership by powerful outsiders include Nazi definitions of who was a “Jew” in Germany in the 1930s and white definitions of who was Black in the American South through the 1950s. Comparable strong influence by privileged insiders can be seen in contemporary Israeli legal definitions of who is a “Jew”.

Other Sami thoughts

159

selves are buying into some significant part of the official Sami framework which, while – visibly – historically and socially created, feels objectively commanding enough that the challengers themselves do not feel confident about denying it. But, also, there exists a complication regarding the official core Saminess: That is, the canonical basis of Sami ethnicity – even as it wins cognitively, socially, and politically – is – by virtue of its very success – disappearing. It is “evanescing” away; that is, the very life style which is at the core of the authenticity is disappearing as (newly) successful “traditional” Sami use their new resources to buy/build better (and, thus, less “traditional”) housing, to find less demanding and better paying sources of livelihood than reindeer herding, and to take over an active role in local and national political affairs. Increasingly, all that is left of canonical Sami ethnicity is the store of copyright protected symbols and motifs – and these, as they lose their embeddedness in everyday life and rise more to the fore as objects of power, undergo a thinning of their meaning, a conversion into something more like any other commercial product. Sami authenticity seems possibly on the road to becoming something akin to the Irishness of the wearing of the Shamrock on St Patrick’s Day in the US. The process of evanescence eventually leads to a situation in which different populations come out with different understandings of the essential core features of Sami-ness. That is, for the wider Norwegian population the issue is one of life-style and way of living, while for the economically improving focal Sami population the issue is more of a religious one involving sacred symbols (and perhaps a communal one involving who is one of “us”). Another such a situation of contested cores is described in Kronenfeld’s Renaissance English Protestant / Catholic example (1996: 209 –218). One wonders if the Sami community’s stripping away of the life-style attributes had something to do with the emerging challenge (by non-traditional people of Sami ancestry) to the constituted Sami community’s monopoly of things Sami.

4. Distributed cognition The distributed cognition approach is relevant to the cultural inferences which are made about category membership in the ways that are described by Kronenfeld (1996, especially in Chapter 11). The distributed cognition approach is relevant, also, to our understanding of the collectivity (or presumed collectivity) within which the presumed collective experience has

160 Chapter 8 – The social construction of ethnicity taken place, and to the process by which that “collective” experience is distilled, recognized, and exchanged. This collective experience, presumably, is based on people’s communications of their sense of what they each see as the commonly shared experience of others in the various groups to which they each belong – that is, the relevant communities of knowledge are assumed not to exist directly in people, but to exist indirectly in people’s senses of what others taken collectively know – what I have spoken of above (see Chapters 3, 5, and 7) a kind of anthropomorphism. In general the plausibility of ethnic membership claims, of the desiderata of ethnic membership definitions, etc. are implicitly communal decisions; relevant communities are those in which the claimants participate (as claimants), those of accepted members (if there exist such), and the wider social universe which encompasses these. Unlike the case for the purely subconscious structure of grammar, vocabulary, especially the adult vocabulary of stuff like ethnicity, is conscious and can be pushed by politicians and others. But, as shown for English Renaissance Protestants and Catholics by Kronenfeld (1996), framing cultural givens greatly constrain the range and effectiveness of such pushing – since people always evaluate the degree to which the political claims match with their intuitions (intuitions formed by the confluence of knowledge and experience indicated above). Thus, the logic of the Norwegian government’s decision regarding the conception of “what defines a Sami” which led to the decision regarding who was given formal control of Sami affairs – apparently feels right even to the challengers. They want in, but do not seem to be contesting the key definitions. This lack of ultimate definitional contestation is similar to what happens in my own mind regarding my own Jewish identity. That is, I know that my non-believing and non-observing position makes me less focally a Jew and thus less able/allowed to speak for Jews – regardless of how well I fit by birth and training, and regardless of how large a number of other “Jews” are like me. A similar – but more extreme, more public, and more strongly contested – example of the Jewish definition problem is represented by the Catholic bishop in France who claimed, even while being a high Catholic religious leader, also to be a good Jew (by birth and adherence to principles) – and therefore claimed a legitimacy for his measure of Jewishness that was comparable to that of the Jewish community which denied his claim. No one, Catholic or Jewish, seems to have taken his position very seriously. The frames win not by virtue of power or authority, but by a kind of derivational default – a communally recurrent cognitive process: Everyone

Distributed cognition

161

who does not remember (or credit) what she/he has specifically been told figures it out anew (re-derives it) from the givens. The issue concerns what ethnic conception gets actually talked of in people’s experience – defining what is socially “real”. This is the process, illustrated in Evans-Pritchard’s account (1940) – by which Nuer people’s experience of genealogically defined units in conflict situations defines which apical ancestors are learned (in action) by youth, and thus who else gets socially “forgotten” – what sometimes (as by Gulliver 1998) has been called “structural amnesia”. This effect of recurrent usage is my conception of how grammatical regularities also are learned. At the same time, we should note, definitions of ethnic group membership do shift according to context and community: Who counts as a keeper of the (say, Jewish) faith or idealized exemplar of the ethnic identity is not necessarily – or even likely – to count as the filler of some different kind of ethnically identified slot (say, as one of Nixon’s “Jew lawyers”!). The Norwegian government’s decision did play a specific kind of significant role in the structuring of Sami ethnic identity by helping change the national perception from what had been seen as a cultural continuum of Norwegian-ness (where people thought of as Sami were at the far end – non-Norwegian and primitive) into a set of contrasted categories with a boundary and wherein a defined Sami focus contrasted with the defined Norwegian focus.

Chapter 9 Some kinds of cultural knowledge – a non-exhaustive list

Cultural knowledge seems to be of at least several sorts – though it seems too soon to say exactly what all of these might be. For instance, regarding the domain of love we have a) the concept “love”, b) the semantic fields (contrasts and inclusions) in which “love” participates (as “love” vs. “hate”, “love” vs. “like”, “love” vs. “lust”, etc.) and c) the various cultural knowledge structures that “love” of one sort or another participates in – including ones associated with mothers, with romantic courtship, with teen-age sex, and so forth. These knowledge structures include salient notions such as chicken soup, locales such as idyllic parks, progressions – the look, the kiss, the awkward fumbling, and then, serious involvement. Sometimes, the “serious involvement” becomes a matter of “getting to first base”, “second base”, etc. if it be seen instrumentally – which, of course, is not how it is supposed to be seen, since being “in love” is supposed to be natural, spontaneous, and heart-felt. More specifically, the kinds of cultural knowledge systems that I am aware of include the following. The semantic system of word meanings alluded to earlier represents one kind of cultural knowledge – as in examples under b) above. The pragmatic classification of relations among the entities so labeled represents another kind of knowledge system. By “classification” I refer to what it is about the domain of plants or of kin that structures the interrelationships of entities in that domain. Thus, in English, what is it that allows us to know that if you are my “son” then my brother’s child is your “cousin”? Or, what allows us, again in English, to know that “oak” is more like “aspen” than like “hyacinth”, and to know that we can get wood from the former two (“trees”) but not from the latter. These are cultural conceptual structures, but ones based in part on attributes of the items being classified and in part on what it is about the items that makes them important to people in our culture. The progressions I listed under c) above might be examples (or, at least, fragments of examples). For the sake of convenience I am calling these “cultural conceptual systems”.

The current state of the art 163

Still another kind of cultural knowledge is involved in our interpretive knowledge (in American) that if someone says “You and what army are going to make me do it!” that person means a challenge – and particularly a challenge to one’s “manhood” – and that the culturally appropriate response for some Americans is a counter threat whereas other Americans may see it as something to walk away from. But no one responds with a description of units of the US Army! Cultural knowledge provides us with models of how to act in certain situations – including the implications of alternative actions, expected default actions, and the effects of relevant social variation (gender, class, subculture), personal and cultural values, goals, personal dispositions, and so forth. I refer to such knowledge as “cultural models” or “cultural models of action”. Another kind of cultural knowledge might be more abstract thought orientations – such as a tendency to see phenomena in sets of pairs, or a tendency to see social groups as consisting of a focal leader and followers (vs., say, something more categorically defined). We might call these “cultural modes of thought”. These various systems of cultural knowledge interpenetrate each other in various ways – cultural models make use of the classifications provided by cultural conceptual systems; cultural modes of thought presumably guide the creation of cultural models; semantic systems are used in communicating the information involved – but each kind seems to represent an autonomous system with its own functions, its own kind of structure, and its own forms of conceptualization. It would seem worthwhile eventually to develop a systematic typology of culturally shared cognitive systems, but I suspect that that effort will have to await a better analysis of some of the specific kinds we have already on the table – and to await a better understanding of and delineation of the alternative functions served by such cognitive tools. In the meantime the types that were sketched out would seem to cover the major functions involved in everyday culture.

1. The current state of the art The formal study of the particular cultural conceptual systems mentioned above seems well in hand, though our analytic understanding of other, less complex and/or less clearly bounded ones seems perhaps less so. Attention is now being paid to cultural modes of thought, and some formal under-

164 Chapter 9 – Some Kinds of Cultural Knowledge standing is beginning to emerge. “Cultural models” has recently emerged as the label characterizing models in culture for behaving and interpreting behavior of the sort that have long been included under ethnography. The difference signaled by the new label is the attempt to relate such descriptions both to psychologically described knowledge/action structures (schemas) and to the modeling of knowledge/action structures in the cognitive sciences. The new cognitive perspective implies a commitment to pay attention to formal and explicit characterization of posited knowledge/action structures and to the specific empirical output produced by models of knowledge/action structures. However, in the actual anthropological event the formal focus has turned out mostly to apply only to abstract discussions of presumed ethnographic content; precise definitions, formally explicit modeling, and rigorous empirical testing otherwise seem largely abandoned. The grounding in cognitive psychology and cognitive sciences has been largely reduced to only a metaphor. In the present work I take these cognitive sources seriously and formally – including positing schemas as the form for individual behavioral knowledge and attending to what cognitive scientists have found in their attempts to model such knowledge. I aim then to consider the formal ways in which collective knowledge structures – i.e., ones inhering in a group rather than in a single individual, and that are partially shared and partially distributed across the members of that group – differ from the structures of individual behavioral knowledge. The long-term goal is a theoretical model of collective knowledge systems and their relationship to individual knowledge such that individual people’s use of collective knowledge can be modeled and empirical tests (of both content and form) generated. The theoretical model will attend to the ways members of the group learn a model, generalize its application to novel situations, and – through such generalization – extend and evolve it. What I mean by “cultural models” will be more fully explicated below. Later, I will consider some examples of relevant empirical data, including some more extended applications to Western American models of ranching and rangeland and to models of romantic love, sex, and marriage.

2. Cultural models The development of “cultural models” by Quinn, Holland, D’Andrade, Strauss, and others (see Holland and Quinn 1987, D’Andrade and Strauss 1992, and Holland 1992) has been one of the most important recent advances in cognitive anthropology, and one for which its developers deserve

Cultural models 165

much credit. Cultural models furthermore hold out the promise of resolving major issues in cultural anthropology in general – including, especially, the nature of culture itself. But to fulfill that promise some important clarifications and refinements are needed, and I would like, here, briefly to sketch out my take on the nature of these suggested emendations. 2.1. Empirical validation The concerns that I want to address arise when one considers the empirical validation of proposed cultural models. One finds that the problem is not simply one of testing or of correlating data with a proposed model, but first requires some clearer delineation than we have so far seen of specifically proposed cultural models and their roles vis-a-vis the beliefs and behavior of individual members of the relevant culture. What makes a cultural model something beyond the kinds of “values” statements that we studied 40 or 50 years ago is its organization as a particular kind of shared cognitive structure. Thus, evaluations have to focus not just on the cultural content of the model, but also on its actual structure – as proposed specifically in a given analysis. Since cultural models are shared, we have to separate them out analytically from the specific individual structures (here called “schemas”) that our informants each have and that drive their own individual beliefs and actions. We have to consider the form which shared cultural model structures might take and the manner in which they might be instantiated in order to see how they might have their posited effect on individual beliefs and behavior. These questions about the relationship between collective and individual structures entail a need for both formal and empirical investigation of the relationship between collective, shared cultural models and relevant specific individual schemas – including, within a cultural community, the statistical distribution of key aspects of those individual schemas. That is, we cannot really validate any cultural models until we have a clearer idea of what exactly they are and are not. Thus, this refinement of cultural models entails a need for both formal and empirical work. The empirical part concerns the need for more attempts at describing and constructing cultural models and for the collection of data that addresses not just the values and implications for action embodied in a cultural model but also the ways in which the values are structured and related to action and the ways in which alternative or competing cultural models are played against one another. Other empirical questions concern the degree to which apparent or putative cultural models are shared within

166 Chapter 9 – Some Kinds of Cultural Knowledge relevant cultural groups. “Sharing” here refers both to the range of values, entities, operations, implications, etc. included in a cultural model and to the degree to which the details of those elements are specified. The formal part is concerned with thinking out the relevant logical properties of cultural models and the interconnections of cultural models with individual schemas and with communities of individuals. It includes the questions of the form taken by cultural models, of where they physically reside, and of how they structure/impinge on individual schemas. Behind these questions are ones of how they are learned, how they change, and of how they are kept coordinated (to whatever degree they are in fact coordinated) throughout the relevant cultural community. It is this latter, formal, part of the problem that I would like now to address. 2.2. Form Potentially relevant are the assumptions we make regarding the actual structural form that is presumed for individual knowledge and action – what I am calling “individual schemas”. We need to be clear about what assumptions we are building into our theory of cultural models – but, still, it is worth noting that the fewer and more general these assumptions the less dependent will our proposed cultural models be on the changing specifics of psychological theory. Relevant also are our views of the relationship among language, thought, and culture: what does unpacking the meaning of a key word tell us – or not tell us – about the thought of users of that word or about the cultural content of the speech community from whose language the word comes? Cultural models are – seem basically, at their heart – something like scenarios – that is, action/situation units, as opposed to the classificatory and organizational structure and role of cultural conceptual systems. Cultural conceptual systems are devices for organizing analytic or “sense” (vs. experiential or “referential”) knowledge. Their function is a classificatory one, and they are about relations among meanings – hence they are to be as stable and shared as possible within a population. They are about what the mind knows and/or what we want to communicate to others – rather than about what is actually encountered in the experiential world. Their application to the experiential world is, thus, potentially loose and flexible. But the assertions they make about relationships within that external world are – in principle – sharp and distinct.

Cultural models 167

2.3. Structure and content Like grammatical kernels, cultural models are minimal “syntagms” that are then capable of expansion. “Expandable” means, in part at least, that the “action” part (or some other part ) of one cultural model can be filled in by another cultural model. That is, embedding does seem likely for cultural models. This embedding – as with some views of grammar – explains how individual cultural models (cf. specific grammatical relations or operations) can each be very simple, while filled out instantiations (cf. a particular deep structure of an interesting sentence) can get a lot more complicated. Re embedding: a “true love” scenario, for instance, can include a “go on a date” scenario as its next action. Schank and Abelson talked of such embedding, and had it pretty well worked out. Unlike grammatical units, though, cultural models are not empty formal units but, instead, are minimal substantive units. If the cultural models have the kind of reference library (of instantiatable programs) aspect 61 that we have considered earlier, then they have to have some “semantic” content – certainly not the script of a play, probably not even the shooting script of a movie – which is why I have turned to the “scenario” conceptualization. But even that misses it a little – which is why I sometimes fall back on analogs to, e.g., the commedia dell’arte. Commedia dell’arte is based on a stock of basic generic situations built around stock characters (a little like a more clearly marked version of a soap opera or a horse opera); the recognizatory (i.e., recognition-atory) knowledge was part of the fun for the audience. Those commedia dell’arte plots show up all over the place as the core of more filled out and subtle dramas. Good, creative drama challenges us and gives us unexpected but convincing twists and angles, but sitcoms and such play to our expectations – tease us about them while directly building on them.

2.4. Function Cultural models are not linguistic entities, but are cultural ones. Their function is to provide models for how the world works or for how people work on the world. As models of action – whether models for interpreting events observed or heard of or models for how to behave to achieve some desired end – cultural models bring together knowledge about the world, action, values, goals, affect, etc.. They are models for doing something (e.g., running 61

As opposed to some deeper internalization or embedding in the psyche.

168 Chapter 9 – Some Kinds of Cultural Knowledge a ranch – see Chapter 13) – or for assessing conditions relative to action (is the rangeland we are looking at healthy or not). We seem to some degree inclined to create story lines (cf. psychology’s episodic memory) in order to fix events and their interconnections in our minds and to explain at least some aspect of these events and connections. Insofar as our various individual purposes depend in one way or another on coordination with others – whether it be cooperation, competition, or just non-interference – it helps if our understanding of events relates systematically to the understandings of others. Thus, one interpretation of cultural models is that they are these shared abstracted story lines that we use to coordinate our behavior with that of others – that is, to understand the behavior of others and to construct our own behavior so as be understandable (in desired ways) to others, to decide when and how we can rely on others to do what and when we can not. This social understanding (represented in these cultural models), insofar as it is driven by aims that include aspects of the material or natural non-human world, involves representations of interconnections and processes there too. Cultural models’ social function requires that they be public and overt (vs. psychologically interior), easily recognized, and flexible. Easy recognizability implies a need for clear, gestalt-like patterns. Flexibility implies applicability to a variety and range of concrete situations and amenability to some variety of different perspectives. By different perspectives I refer at least to the points of view of one or more principle players (i.e., roles), of interested observers with one or another stake in the outcome, and of relatively disinterested observers. Cultural models’ status as shared or collective cognitive entities requires that they show up frequently enough in interaction patterns and communication for new learners to experience them as part of the received cultural givens. Their content seems some mix of (a) what consistently comes up and is emphasized in relevant interactions and communication and (b) what links or emendations or conditions are necessary to give the emergent collective representation a clear, coherent, and meaningful story line. The emergent cultural model, then, while dependent on frequency of occurrence is not any kind of simple statistical average of experience – but depends as well on a kind of interpretative coherence. As indicated above, specific cultural models seem constructed relative to some purpose – and are drawn out of frequent experiences relating to the relevant purpose. Purposes imply contexts of applicability and/or relevance. Their tie to purpose and context represents another sense in which they are not to be seen as some simple statistical artifact of interaction patterns.

Cultural models 169

2.5. Existence status Cultural models are external to the individual and hence technically nonexistent in the same way that langue and other collective representations are non-existent. That is, on the one hand, such collective representations are the property not of individuals but of groups of people; they are the shared knowledge that enables coordination and communication among members of a group. As individuals we each belong to a large number and wide variety of relevant groups (i.e., minimal speech communities or cultural communities). Many of these local communities belong to hierarchies of larger and less immediate social groupings, as in the relationship of faceto-face speech communities to successively larger dialect communities and language communities. But, on the other hand, the only actual locus of such collective representations is in our individual representations of them. That is, there exists no separate repository of cultural or linguistic knowledge outside of the minds of members of the relevant cultural or linguistic communities. At the same time, however, a) each of us as individuals belong equally to other different communities; b) in constructing our own individual schemas or behavior, each of us adds our own individual detail to what we perceive as the relevant general cultural norms or forms; and, c) importantly, each of us is aware of our own self (including desires, perceptions, feelings, etc.) existing apart from and in contradistinction to these communities – the communities to which we impute the norms and forms of language and culture. 2.6. How constructed Cultural models are constructed by individuals as abstractions of a certain sort. They are “abstractions” since they are inferred by each individual from a range of interactions, with the detail required for any specific instantiation left to be worked out as they are applied. They are abstract in two ways. First, they do not contain complete sets of information or behavioral prescriptions, but, instead, contain only those elements which – based on what consistently comes up in the relevant interactions and is consistently emphasized in those interactions – are seen as relevant to the task or situation of the cultural model. And second, the elements cultural models do contain are not fully defined or fleshed out behaviorally, but, instead, only include those attributes which are recurrent, important, and/or felt to be logically entailed. This is precisely the kind of abstract quality that Trubetzkoy ascribes

170 Chapter 9 – Some Kinds of Cultural Knowledge to phonemes in his detailed phonological implementation of Saussure’s linguistics program and that he contrasts with the filled out realizations that occur in actual speech events. ([1939] Trubetzkoy 1969: 36). But “of a certain sort” signals that these cultural model abstractions do come with default presuppositions of the additional specifications (of entities, relations, actions, etc.) which serve to make them canonical (or “unmarked”) instances rather than “bare bones” abstractions. This presupposition of unmarked default values (in the absence of additional information) applies as well to Trubetzkoy’s abstract phonemes. The “instantiation” replaces canonical default parameter values with the specific local parameter values of the given situation, while the “realization” fills in all the additional detail of the given local reality. 2.7. Prediction Predicting which cultural models someone might appeal to in a given situation is, of course, difficult. There are lots of “‘dialect” variants (as, for example, the marriage/love variants above) running around to which people can appeal, as the spirit moves them. One then also needs some sort of decision process for explaining (summarizing, predicting) how people actually choose among these variants in one or another actual situation. I don’t know any good elaborated theory for this, but the best approximation or approach I have seen is Carol Mukhopadhyay’s in her Ph. D. dissertation (1980). Her approach suggests a kind of potpourri of factors: what seems right, what comes out the way that’s best for “me”, what am “I” capable of, what am “I” interested in, how will “I”/“we” look to (the various different kinds of) those around “me”/“us”, and so forth. It would also be useful to be able to predict what might constitute wrong ways of reasoning. The research trick, of course, would be to find a kind of “wrong way” that is generically plausible (could be plausible for people somewhere) but “wrong” for the given society.62 But this, of course, may 62

Cross-cultural work might help with this one. De Munck’s (1996) work on love and marriage in Sri Lanka is particularly relevant because the issue in his study was not large scale patterns (e.g. love vs. arranged marriages [cf. “nature” vs. “nurture” in a very different dichotomy]) but the individual reasoning by which outcomes were reached. One needs to find some reasoning paths which are shown to be plausible by working in one place, but which are not accepted in some other place.

Cultural models 171

be more a matter of degree than any absolute; all humans seem capable of imagining all human possibilities (cf. Lévi-Strauss’s old (1964) point regarding why “native” views and “anthropological” views inhabit the same world). That is, e.g., we in the U.S. can easily imagine marrying for money (even if we don’t think it’s really right) and we even can imagine marrying for money a person our parents chose and at their behest – but the scenario is beginning to seem fairly weird (to us). But, of course, this evaluation is from the perspective of the marrying kids; as American parents we can, quite easily and much less weirdly, imagine making a good choice (of a person who would care for them, value them, …) for our kids. And, even in De Munck’s “arranged marriage” cultures (1996), the idea of parents doing the selecting can be a little weird (undesirable) to many kids, as can also be seen in Robert Moore’s (1981, 1998) Hong Kong finding that who got to make the marriage depended simply on whose resources the new couple was to rely (the parental farm or their own jobs). 2.8. How used Individuals refer to cultural models in thinking about and constructing their own actions, in interpreting the actions of others, in positively viewing some behavior, in negatively viewing others, etc. These different purposes can be served in any of a variety of ways: cultural model selection (e.g., “Is it ‘true love’ or is he just a ‘player’?”), variants of a given cultural model (fancy “date” implying “he loves me”, or “cheap” “date” implying a lack of respect, or honest “date” where money and glitter are not the issue but where it’s all about “us” being together, or …), or something more detailed about role specifications or actions within a cultural model. Thus, cultural models can entail or carry points of view – angles of vision, evaluation perspectives, viewing distances, stakes, etc. One’s choice of which cultural model to apply in some given situation then depends in part on what fits, but also depends in part on where one’s internal predispositions lead one and on what outcomes are felt to beckon. That is, the problem of the “self” is posed, but it is a self that exists outside of cultural models – a self that evaluates its own predilections, its stake in alternative outcomes, and the plausibility of various courses of behavior, and then that chooses among potential cultural models in the light of these considerations.

172 Chapter 9 – Some Kinds of Cultural Knowledge Since cultural models are external to the individual, and not directly internalized, they can not be in themselves directly motivational. At best they can only serve as standards or models available to be referenced by individual schemas or self-plans. Actual motivation has to come from within the person, and hence has to be individual. What cultural models can do is provide a socially validated set of motives and feelings relating to a situation and a socially validated reasoning process for relating that situation and pertinent desires to a course of action. Because they are commonly spoken of, and because much of the interpersonal coordination that they often entail is often linguistically mediated, cultural models do involve a linguistic (naming and classifying) component, but this linguistic component is a subordinate part of them, and need not be present at all. That is, systems of classification are tools for use in the application of cultural models, while cultural models are tools for use in planning (or understanding) actions. Cultural models involve means-ends determinations, and other like kinds of reasoning, that do not seem part of linguistic (including classificatory) systems – even if such calculations do sometimes come into play in the application of a classification system to the experiential world.

2.9. Cultural models vs. systems of classification Cultural models, thus, are very different kinds of cognitive structures than are systems of classification. “Cultural Conceptual systems” – including kinship terminologies and ethnobiological systems – are productive systems of classification. Similarly, grammars, too, at least as conceived in transformational linguistics, are, too – in the sense that they define what are sentences and sort out different kinds of sentences according to their structural relations. Classification systems and grammar are all part of the linguistic system by which meanings are communicated. They have to do with what is like and unlike what in which ways. Their function is to distinguish messages from one another. Systems of classification – often, especially in important, rich, and frequently used domains – have to be productive systems. For that reason, because they deal with a single homogeneous kind of cognitive entity, and because they are used for reasoning rather than for action, they can be (and benefit from being) complex, rigorous – even axiomatic – systems. It is the heterogeneity of cultural models and their job of relating other stuff to ac-

Cultural models 173

tion that makes the logic and formal structure of cultural models looser and simpler63 – and less mathematically interesting. Cultural models where relevant and appropriate seem to include reference to various cultural conceptual systems for organizing and classifying cultural knowledge – such as kinship terminologies, ethnobotanical classifications, classifications of types and brands of automobiles, types of tools, and so forth. But, given the role of cultural conceptual systems as mental resources and cultural models as application devices, I would not expect cultural conceptual systems to reference cultural models. Cultural models and systems of classification are both collective cognitive entities, property of groups rather than single individuals – and thus both are subject to all the issues, concerns, assertions, etc. that have begun to emerge in our attempts to understand distributed cognitive systems, including issues of formal shape, of what is shared among whom, of productive use in novel situations, of how they are learned, of how individual representations of them within relevant groups converge and are held together, and so forth. These are issues that have been addressed in earlier chapters of the present work. Cultural content – norms, attitudes, knowledge, word meanings, etc. – is learned by individuals from members of the groups they interact with. This content is not directly learned as a system, but instead each of us individually extracts out the regularities we encounter in our interactions with group members, and then construct from those regularities our individual “internal representations” of the “collective representations” (rules of behavior, regularities of attitude, shared presumptive emotional valence, beliefs, grammar or grammar-like regularities of construction [where appropriate], etc.) that characterize each of these groups. And it seems that it is recognition of these patterns – joined with experience of collective activities – which leads us to recognize the groups as groups. 2.10. Cultural models, language, and communication Since cultural models are social they have in some sense to be intrinsically tied to interaction patterns and communication. They often are linguistically recognized (and, maybe, coded) – but, even if non-verbal (cf. Piaget’s sensory-motor knowledge), they still are social and entail communication. And, 63

At least from our present perspective – in which the description and analysis of cultural models is still far from formal or rigorous.

174 Chapter 9 – Some Kinds of Cultural Knowledge even where used, language is only a “transparent” medium via which a posited reality is presented, not the essence of that reality 64; the cultural reality, then, is not linguistic. But also remember that the communicative constraints which keep the communal representations similar only directly work on the relevant words or communicative actions. That means that the fuller situational understandings, the “it”s behind the words or interactive patterns, are much freer to vary from individual to individual – and this is part of what leaves the instantiations of cultural models free to wander about. These presumed “it”s are still important and relevant, though, because in reasoning to or from specific instantiations users work off these images. It is worth noting, as pointed out in Plastic Glasses and Church Fathers, that the nature of natural language leads often to complex and internally inconsistent sets of referents for important or commonly used words. The combination of the frequency it takes to maintain a word in active vocabulary, our often appearing need to use existing vocabulary to speak of novel referents, our reliance on context to disambiguate reference, and the subsequent stabilization in a speech community of some of these secondary referents means that the words which we use to express our thought can, in isolation, easily (and often) have only a loose and indeterminate relationship to the thoughts they are used to express. This observation means that even in the case of verbally coded cultural models, an analysis based too closely on the words used to express the model is seriously at risk. The interactive and communicative aspect of cultural models extends beyond their specific semantic content, and includes information relative to their evocation in one specific situation or another, to who uses them to/re whom, to their relationship to other models available for application to the given situation, etc.

64

The idea of each “critter” in a group forming an internal representation of the aims, behavior, etc. of others in the group, of the group as an entity, and of other groups as entities seems implicit, at the least, in what we know of the behavior of social carnivores (such as dogs and wolves) and perhaps other social mammals (and maybe birds) as well. This comparative perspective suggests that the primary basis of such representation cannot lie in language, even if language does become a powerful tool for such representation.

Cultural models 175

2.11. More on the content of cultural models Cultural models seem to involve some kinds of basic elements. Colby’s “eidochronic” analysis (see Colby 1973, 1975) is perhaps a place to start – though one can not yet be sure. In any event, some elements of cultural models are maybe more basic or necessary than others. Necessary ones may include a) a topic b) a “cast” of i) basic or necessary roles, and ii) other potentially relevant roles c) an idea of which roles represent potential focal perspectives – points of view from which the scenario/cultural model is to be understood or evaluated. Some, at least, seem to lend themselves to more than one alternative. d) for each role (or, maybe, only each principle role?) one needs to know i. actual goals aimed at ii. motivations – reasons for doing iii. emotional states iv. values v. actions e) one needs to know the dynamics of interaction between the various roles – presuppositions that guide these interactions. (But, remember, we still are dealing with roles in a scenario, not the actual people in a realized instantiation.) f) outcome and scorecard. That is, the default outcome implicit in the basic scenario, and the understanding of which roles (if any) win and which (if any) lose. And winning and losing may well be a matter of “in what sense” and “by what measure”, depending on the situation. Here I mean just within the “fairy tale” or the “player’s story” or whatever as it is seen for the abstract vanilla version of the cultural model in question. An example of an a basic element that might be more optional rather than necessary might be g) how self-conscious one or another roles is about its aims, motivations, etc. The scenario itself – the told story – does not explicitly or overtly display most of that understood background information. In the romantic love area it may just overtly show the actors (roles) and their actions or the feelings;

176 Chapter 9 – Some Kinds of Cultural Knowledge in other areas it might just show actors (roles) and motivation. The rest of the basic information, then, is not spelled out, but is implicit – parts of it can be appealed to by users of a cultural model in explaining what happens in a realized instantiation, or in accounting for why they choose to interpret the observed actions in terms of one cultural model rather than another, or in explaining their own choice of which cultural model to structure their own behavior around, or so forth. 3.

Cultural models and culture

I am taking cultural models as the basic units of cultural knowledge regarding action. 3.1. Collective knowledge Cultural models are one kind of collective knowledge. That means they are shared, and are based in commonly experienced elements of relevant communities. “Relevant communities” can vary – as contrasting alternatives, as non-contrasting overlapping alternatives and as hierarchically structured sets and subsets; people always belong to many such communities more or less at the same time (though, at any given instance, only one or a few may be foregrounded). That is, the sharing is relative to some particular community or set of communities – possibly, but not necessarily, including the widest level of cultural community. Culture, as suggested earlier, necessarily (given its coordination function in a social system of distributed cognition and parallel processing), is – within a given community – some mix of shared stuff, distinctly specialized stuff, and intermeshed or complementary stuff. And under a general culture (such as Modern American or modern Western Culture) can fall a great variety of alternative sub-cultures, including a number of alternative hierarchies of inclusion. This interplay of socially patterned similarity and difference should apply as much to cultural models as to any other aspect of culture. 3.2. Collective representations and social “others” Each of us constructs an (our own individual) internal representation of (what we perceive as) cultural patterns. However, we don’t construct these

Cultural models and culture

177

as abstractions, but, rather, anthropomorphically impute them (as concrete appurtenances) to abstracted “individuals” or alters. We can then judge how similar any real individual is to this “idealized” individual (not “ideal” in the sense of perfect, but in the sense of purely representing…). Since each of us infer our own individual constructions of these anthropomorphized ideal individuals, the abstractions are potentially free to vary a fair amount from one of us to the other – depending on our various individual experiences of the culture around us. At the same time, the same communicative constraints that bind our separate representations of language together also operate here, albeit somewhat more loosely, to keep our various individual representations in some kind of useful alignment with those of others around us. This alignment pertains especially to those parts of our representations which enter commonly into discourse or that rest relatively heavily on direct experience. It applies much less to those aspects which are not much explored – but which still, willy-nilly, exist (because, remember, it was/is a whole person [Golem-like, if you will] that we have called into existence). Thus, such cultural models are among those phenomena for which viewing distance (as well as viewing angle) is crucial; as with a Monet painting, these collective representations tend to dissolve into the white noise of disparate details when examined too closely (or from the wrong angle, as in those psychological experiment games on gestalt perception [where, e.g., the sides of a cube are individually suspended separately at different distances from the viewer, but stacked so that, from the one viewpoint, they look to form an intact cube]). We can (and often do), of course, have lots of these idealized alters running around in our heads – representing the various cultural “dialects” among which we actively “code switch” – with several different planes of variation. In one plane, some of these are moral ideals, some are “realpolitik” ideals, some are sensitive “heart-ful” ideals, some are competent, and so forth. In another plane, some are rich or poor, educated or ignorant, cultured or grubby, leaders or followers, … In yet another plane, some are Anglo, some Black, some Latino (or Chicano or Cubano or …), some Jewish, some WASP, some redneck, some city-wise, some country bumpkins, some country gentlefolk, etc. 3.3. Learned without being taught Cultural models are not, for the most part, explicitly taught. Instead they are picked up via experience, interaction, and communication. Hence they

178 Chapter 9 – Some Kinds of Cultural Knowledge have to be in a form that is “easy” to learn. Since they (at least often) are gradually learned via accretion and/or increased specification (and differentiation), and since people with different levels of knowledge have to be able to communicate and interact via them, cultural models have to be cumulatively or hierarchically structured, wherein rough or general forms each include some variety of more specified or more specific forms (as in the way Berlin’s ethnobotanical systems [1972, 1992] develop and are learned). Communication between people with unequal knowledge involves locating a shared version (conception, form) and then working down from there as needed to the level of needed specificity. Buried in here somewhere is the Piagetian insight that we can only recognize a little more information than we know to look for, and the Gombrichian/Piagetian insight that most description aims not at any direct coding of information but instead at at the recognition of a known generic category with then whatever individuating modulation is needed to evoke the desired specific.

3.4. Cognitive hierarchy The variable levels of sharing and application imply a marking hierarchy (or hierarchy of relative default forms) of successive bases for further elaboration, specification, and/or individuation. That is, each new thing (recognized entity, constructed group action, etc.) is not created de novo, but is modified from some already existing thing. The requirements of flexible adult sharing, of gradual, accretive child learning, and of adaptive creation of new forms all press for the same sort of hierarchical cognitive structure of inclusion (cf. George Miller’s [1956] Magical Number Seven notion of “chunking” and Berlin’s [1992] ethnobotanical systems). 3.5. Social complexity Finally, there are the implications of the fact that cultural situations grade into one another, as also do the communities in which culture is held. We can (in the context of Schank and Abelson’s [1977] study, say) construct a taxonomy of restaurant types, but in our actual experience the extended versions of the types that we experience grade into one another (even if not always in a strictly “continuous” fashion). Similarly, we can construct some variety of intersecting taxonomies to account for the interrelations that exist among the various cultural and subcultural communities to which we each

Cultural models and culture

179

belong, and/but these two can be experienced as a kind of grading. At issue is the variability and fluidity of our cultural/cultural-community associations when looked at in detail. We seem unlikely to come up with completely autonomous cultural model sets for each specific cultural community that we experience (participate in, observe, hear about, …). Instead, it seems, we have to have some way of coding generalities that obtain across sets of communities, while learning/storing the details on which the included communities differ. But since the communities we experience represent some variety of cross-cutting oppositions and hierarchies, our native performer task of keeping track of these commonalities and distinctions while also pulling out coherent cultural models that we can turn to for guidelines for generating or interpreting behavior in actual concrete situations seems interestingly complex. The problem concerns how people in cross-cutting communities (culturebearing groups) keep in mind and keep track of the inter-related sets of cultural models that are linked to cross-cutting communities and that apply to inter-grading situations. Thought of this way, it does seem to go Einstein’s “hitting a moving target from a moving platform” one or two better The problem is sort of shown by the illustration in figure A. The figure was originally constructed for Chapter 7’s presentation on ethnicity, but applies equally well to our present discussion of the general case. We can approach the problem via a list of propositions that seem likely to obtain. a) Entities or activities are always, each, parts of a variety of hierarchies: Restaurants exist in a kind of taxonomy of basic types (fast food, luxury, basic, etc.); but cross-cutting those types is information about ethnicity and region that is often also hierarchically organized, and there exist social class/wealth/expense levels as well; and so forth. These hierarchies often seem expandable. They also are grouped into cross-cutting categories based on organization form, types of employees, modes of behavior, ambience, kind of decor, etc. etc. Many other activities, such as, say, how to play baseball participate in comparable hierarchies and cross-cutting classifications. b) At any given level of contrast we have a bunch of entities or activities that contrast with one another. Often there is room for adding in new items of contrast – such as new kinds of restaurants (such as Sushi bars) or new kinds of social slices (Yuppies, then Dinks, and now whatever). c) The communities within which views of these entities or activities are held, also exhibit the same kinds of variability – both regarding their

180 Chapter 9 – Some Kinds of Cultural Knowledge

Regarding Venn Diagrams and Society The squares represent society. The ‘ice cream cone on it side’ is an eye. The circles represent the communities in which inhere collective representations. Included within the squares, circles, and squiggles, but otherwise unshown, are people. 1. The first diagram suggests that society is what constitutes – and analyzes – society. Or Society is only viewed from within society. Society’s truths are socially constructed. 2. The second diagram queries whether society is to be understood as a unitary THING or as some squiggly amalgam of thingS – the limiting case of which is a mere STATISTICAL FUZZ. 3. The third diagram queries the nature or structured-ness of the squiggly amalgam: a CHAIN of familiy resemblances, variation around some shared CORE, or a MUSH of varied overlaps. 4. For the sake of argument, let us assume that cultural and linguistic concepts (including word meanings) are defined within circles. Let us note a) that contested terms, such as many ethnic labels, will be defined within many circles, but somewhat differently from circle to circle, and b) that only a few of the defining circles will represent communities within the labeled ethnic group.

Figure A. Regarding Venn Diagrams and Society

variety of inclusion relations and their variety of contrast relations. And the understandings that members of each of these communities have of the entities or activities under consideration show some variation from one community/group to another. d) Individuals can move back and forth (“code switch”) among some number of these communities – and thus can each (if at different times) “hold” the structures of different, contrasting communities in mind while recognizing communal differences in relevant understandings.

Cultural models and culture

181

e) It seems both wildly inefficient and thus wildly unlikely for any person to hold independent cognitive knowledge structures of each variant kind of every given entity or activity for each community in which the individual participates or with which the individual interacts. It seems much more likely that at each level and re each comparison the common elements are recognized and made use of and the cognitive detail specific to the particular entity/activity and community is only that which distinguishes contrasting ones from one another and subordinate categories from superordinate ones. Thus, in most (all?) cases the locally specific information (information that applies only to the given group in question) will be fragmentary – will, in effect, only consist of modifications of some slightly more general inclusive knowledge; but that more general inclusive knowledge will itself only be a similar fragment of something further up the line. f) Marked vs. unmarked oppositions – the linguistic and anthropological equivalent of more specified options vs. default options in computer programming – will be common in such nests of knowledge structures – maybe even universal. g) Let us call these cognitive knowledge structures of what things/situations look like or of what behavior is considered appropriate to some situation “cultural models”. Other kinds of cultural knowledge structures exist – and this reasoning may or may not apply to them, depending on what functions they serve. h) The integration of the fragments is governed by the need for a coherent story line in the case of dynamic scenarios and for a coherent, meaningful picture (even if of something “unreal”) in the case of more static scenes. And the contents and organization of the story line or picture are constrained by the function which the cultural model is seen to be addressing. That is, there must be some substantial coherence of content. i) The nature (shape, contents, boundaries, …) of the fragments is governed by their role in the scenario or picture. Thus there would seem to be some sort of functionally driven grammatical-like constraints on the shape and contents of cultural models. This represents a need for some coherence of form. j) Given inclusion relations, the specific cultural model fragment associated with the lower level (more specified) item (e.g., fast food restaurant vs. restaurant) will be specifically whatever detail distinguishes it from the default version of the less specified one. Similarly, in a contrast situation (e.g., ethnic vs. general American), the cultural model fragment associated with a given item will be specifi-

182 Chapter 9 – Some Kinds of Cultural Knowledge cally what distinguishes it from the contrasting item – unless one is unmarked and the other marked, in which case the situation will be the inclusion one. Any complete cultural model – i.e., whole scene or scenario – for some given item/situation/action then will wind up being a mosaic of fragments defined by all of the various categories in which it is included and fragments attributed to it via all of its contrast relations. It follows that, in turn, the fragments associated with the higher level (more inclusive) categories will themselves similarly be a collection defined at each level by operative contrasts. 3.6. Presupposition of sharing Roy D’Andrade speaks of the sharing as an “I know that you know that I know…” game – of the sort that comes out of studies of Machiavellian intelligence. In some literal sense he has to be right, but I think the real cognitive force of sharing of cultural models is something different and more prosaic. I think that we presume the “objective” existence of social forms and cultural (i.e., shared cognitive) forms in the same way as we presume the objective existence of the physical world around us – it simply is there and, unless hidden, available to anyone in the area. It is like a language grammar which in fact is changing, contingent, and only existent as a mental construct in the minds of individuals – but to which (as Saussure noted) we ascribe an external, objective reality. We may ask “Do you know English?” but not “Do you agree with me that the plural of ‘cat’ is ‘cats’?” or “Do you know that I know that the plural of ‘cat’ is ‘cats’?” 3.7. Productivity, form, and application The flexibility and unpredictability of situations to which cultural models are applied means that they cannot be some sort of fixed forms (or scripts65) which are simply memorized and automatically applied. Instead, they have to have some sort of productive generative form. Cultural models are thus, 65

Even where apparently script-like the cultural models have to be query-able and adaptable in a manner equivalent to an on-the-fly rewriting of relevant parts of the script – in a way that still effectively communicates. (The need for such an ability poses one problem with Schank and Abelson’s overly sharp distinction between script and plan).

Cultural models and culture

183

in one sense, like grammatical rules. But, unlike grammatical rules which are concerned with the form (vs. content) of messages, cultural models are about the content – which might perhaps imply significantly different formats (shapes or structures – either in the cognitive entities themselves or in the descriptive devices which we create to represent the cognitive entities). And, even for grammatical descriptions, we utilize a kind of formal descriptive rule to describe the regularities – even where we know the cognitive operations in question cannot take that specific form. (See Kronenfeld [2006b], where X2 + Y2 = a2 is offered as the formal description of any and all circles while assuredly not being how we actually produce circles.) The format of the rule, where one form is set equal in an equation to another form, is partially what is at issue. But also two other attributes seem at issue. One problem concerns the formal abstraction of these grammatical regularities into a closed formal system, when they are actually held and developed in a more piecemeal fashion that is more embedded in other things. The second problem concerns treating categories as closed and bounded by whole category definitions when (at least in many areas) the category definition seems to involved prototypic exemplars (or gestalten) to which other potential exemplars are secondarily and contextually related. So, even for the grammar of language, it seems that the operative cognitive entities are “patterns” of some sort which are applied flexibly to the range of situations that seem at least roughly to fit them. Kids learning don’t just generalize the “official” grammatical patterns but instead try to generalize any pattern they see – until they get negative responses from those around them – e.g., the English strong verb pattern “fling” “flang” “flung” vs. “fling” “flinged”, etc. or regularizing irregular forms, as in “one men” “two men” and “one man” “two mans”. It seems, at least now, that the best way to think about the shape or form of cultural models is as some sorts of scenes, scenarios, or story lines. Our normal mode of memory is episodic (rather than “semantic” – that is, rather than being made up of abstract relations and categories) – it involves a sequence of actions with goals, motives, etc. built in – a kind of dramatic structure. My sense is that these cultural model structures are not bare-bones abstractions, but are filled-in instances – though, with null, generic, or unmarked detail; that is, they are visualizable. The added detail that defines subtypes gets included via the specification of marked variants (within the marking hierarchy). Other relevant detail comes in through the process of instantiation (see below), while other irrelevant detail belongs to the actual realization (see below).

184 Chapter 9 – Some Kinds of Cultural Knowledge Some features of a cultural model are essential – changing any of them moves the story to a different cultural model (or a non-cultural model). Some features of a cultural model are important (even essential) in the sense of shaping how the cultural model is applied and used, but variable – changing them affects how the cultural model story plays out and maybe distinguishes one subtype from another (for instance, the particular mix of tables, booths, and counter in a restaurant may exemplify such a variable), but does not make it a different cultural model. These features are what one can call “variables” within the cultural model. Some features (in the filled out story) may be only there by way of completing a filled out and familiar scene but may be totally irrelevant to application of the cultural model (e.g., any restaurant will have glasses, and any picturing of a restaurant will include glasses, and any pictured glasses will have to have a particular shape – even for a kind or restaurant where the shape of the glasses is totally immaterial). The application of a given cultural model to a particular actual situation seems to me not to be a matter of fitting some “all and only” definition, but instead a matter of assessing what cultural model(s) the situation is most relevantly like. This is by way of saying that the prototype-extension format (described for semantics in Kronenfeld 1996) applies not narrowly to words but more broadly to all of our normal conceptual categories (see “realization” below). 3.8. Function and use Cultural models are not models of abstract knowledge (as in cultural conceptual systems [or systems of classification] such as kinship terminologies or ethnobotanical systems), but are models of culturally appropriate action. (cf. Schank and Abelson’s [1977] restaurant simulations or Frake’s [1964b] “Structural Description of Subanun Religious Behavior”; B. N. Colby’s Ixil diviner work [Colby and Colby 1981] is also relevant here). This distinction is important; a different function implies a different form – and, in this case, different kinds of agreement, variability, complexity, and systematicity. Unlike what some have said, I do not think that Cultural models are immediately and directly motivational. That is, they can have story lines which imply some relationship among values, goals, motives, emotions, and the like, but they don’t actually “do” it; the doing has to be a matter of individual cognitive structures (schemas of whatever sort the psychologists

Cultural models and culture

185

can come up with). Cultural models are thus opposed to “schemata” or “schemas”, which are the individual cognitive structures that actually drive our action. These schemata are never fully available to our consciousness, though parts may be quite conscious. What cultural models do is give us a kind of library of motivation/action packages. These packages may come with motivational ratings such as “I must do X if I want Y.” or “People expect me to do X” or “To be accepted by Group Z I must do X” or so forth. We still – always – have, as individual users, to pick out the specific cultural model or models that serve our purposes – cultural models are tools – and then apply them via incorporation and adaptation in our individual schemas. There exist different ways in which we can apply a given cultural model. We can turn to it as a model for our behavior (e.g., how is a professor supposed to behave in a classroom?) or as a model for our interpretation of someone else’s behavior (e.g., who is this person talking in that classroom?). We can apply a cultural model to a situation we are ourselves in or to one we are observing from without. Since the cultural model is a tool, which is chosen and how it is applied will depend on who is applying it (here, which role in the story – e.g., cook, waiter, or customer in a restaurant). Cultural models proper, by virtue of being general and generic, represent a kind of abstracted and reified knowledge – that then has to be related to individual situations and actual events. And, as implied above, the given cultural model may not match up exactly to a specific situation to which it is applied. The application process seems to have at least three elements: the cultural model, its instantiation, and its realization. 1) A cultural model itself is a basic abstract scenario (containing, itself, only the absolute bare bones minimum of content – mostly in terms of parameters. “Parameters” refers to the fact that the cultural model’s actors and events are more likely to be abstract roles and actions than the rich, filled-in versions that we will see in actual empirical situations. The same may also be true of other aspects, such as goals, motives, values, etc.. The cultural model will contain “interior” motivation – that is, to what moves the action within the story along – but not the more “exterior” framing or situational motivation which would explain why in a given situation a person might choose this cultural model over some other alternative. 2) “Instantiation” (taken from Artificial Intelligence [AI]), where it refers to the application of a model to some specific instance) labels to the mental process of applying the cultural model to some particular situation – including selection of relevant parameter values (e.g., this “restaurant” has tables

186 Chapter 9 – Some Kinds of Cultural Knowledge but no booths, and special waitresses – vs. the regular waiters – serve the booze, etc.). Instantiation involves also a point of view (is the “restaurant” being considered from the viewpoint of customer, waitress, cook, owner, …) and maybe some consideration of which “type” or restaurant (in the marking hierarchy of types: e.g., a generic restaurant, a “coffee shop”, a “cafe”, or a “family style” place). The cultural model – or one or more potentially relevant ones – gets “instantiated” for a particular situation – where instantiation refers to the filling in of the above parameter values, which bring in relevant situationally specific information about the entities in the scenario (players, perspectives, motives, aims, emotions, goals, etc. see below). At this level the instantiated cultural model is still a mental entity (like a computer program, if you will). At this level users of the cultural model will be, in effect, addressing questions such as “Why pick this cultural model over some others?”; such decisions will depend on factors such as how the cultural model scenario comes out in the end, and why that is beneficial or not – for whom. The details of the instantiation (what I have sometimes spoken of as “linking propositions”) are not intrinsic to the cultural model, but are inferred from a comparison of an understanding of the concrete situation at hand with understandings of potentially pertinent cultural models. Since the categories of cultural models are abstract it seems unlikely for concrete behavioral observations to directly feed into the construction of cultural models or to be directly entailed by them. That is, “a loving relationship”, for instance, is not pegged to any specific observable behavior – such as kissing, hugging, washing dishes, buying flowers, or so forth – and so cannot be inferred from actual behavior without some linking mechanism which connects the abstractions of a given cultural model to actual behavior. I suggest that the linking mechanism for cultural models is a process of instantiation wherein generic default attributes – parameter values – of the cultural model are replaced with the parameter values appropriate to the application situation. These instantiations do not seem intrinsic to the cultural model in question, but, rather, seem in a general sense to be reasoned during the process of actual application of any cultural model. This reasoning seems based on a comparison of the goals and structure represented in the given cultural model (and in other already familiar instantiations of it) with the situation at hand. The selection of one cultural model vs. another seems to depend on a combination of the person’s read of the situation and the situation’s fit to available possible cultural models, the person’s stake in the interpretation or the outcome of the event (or read of the stakes of central players), and the person’s own experience of relevant precedents. Recurrent

Cultural models and culture

187

and familiar instantiations can be learned and referred to (vs. reasoned anew with each application), and, as suggested, these familiar instantiations probably play a role in reasoning out new ones. And we should note that “situations”, here, potentially include – as relevant – actors, goals, values, feelings, knowledge levels, knowledge levels, and so forth.66 Thus some instantiations may be constructed in a relatively ad hoc manner (the extreme of which would be Schank and Abelson’s “plans”) while others might be commonly enough encountered to be learned as a common routine (the extreme of which would be Schank and Abelson’s “script”). The differences seem likely to string out along a continuum – rather than form the stark dichotomy utilized by Schank and Abelson. The same physical situation can sometimes lend itself to several alternative instantiations (i.e., conceptualizations), and the user may have to decide which to apply. As users we may simply decide on which seems a better fit to an ambiguous situation, but we also may try to figure out which is more to our advantage, which leads to a nicer outcome, or a better self-image, etc. 3) “Realization” (taken from Trubetzkoy’s discussion of the relationship between the abstract form a phoneme has in language and the physical form it actually takes in some piece of speech) refers to the mapping of the cultural model (a cognitive structure) onto the actual physical events to which it is being applied. Several different instantiations of several different variants of a given cultural model might be equally applicable to some given actually occurrent situation. However, the cognitive interpretation (cultural model/ instantiation) of any given situation at any given moment seems to be only one gestalt – even if, as in the optical illusions of Gestalt psychology (e.g., the wine glass that morphs into two faces facing each other), we can rapidly switch back and forth between one and another. There is a lot of information/detail present in any actual realization which is irrelevant and immaterial to any cultural model which is being instantiated (in the same way that any heard instance of the phoneme /a/ has a particular pitch even though pitch is totally absent from any specification of the phoneme).

66

A different kind of perspective issue from that discussed earlier might concern whether players or observers are looking at a private (in their minds) unrealized instantiation, at a more public discussion of an instantiation in the context of its potential or hypothesized realization, or at a concrete situation for which a cultural model which might serve as a basis for interpretation is being sought.

188 Chapter 9 – Some Kinds of Cultural Knowledge An instantiated cultural model gets “realized” in an actual situation. A given situation can be “read” by observers or participants as realizations of different instantiated cultural models, depending on the usual contextual issues. Of the competing instantiations of potential cultural models, the one that actually gets turned to by some person – for interpretation of observed actions or for structuring the participants’ actions – is taken as that person’s realization. It is possible that different people from different perspectives will see different alternative cultural models as being realized in some given situation. If they talk together or need to decide on some common response, then there will exist some pressure on them to reconcile their interpretive differences; but in the absence of such constraints it seems possible for different people to interpret a given concrete situation in terms of quite different cultural models. In fact it seems possible for the same person to switch back and forth between alternative cultural model-based interpretations – much in the manner that, in one of the classical optical illusions, we can go back and forth between seeing a wine class and seeing two people kissing, but we cannot see both at the same moment and we can not/do not ever average the two images. Such a cultural model example – in the domain of romantic love – might involve “her” interpretation of “his” romantic advances as the actions either of a “player” or of a serious suitor. 3.9. “Application” or “invocation” Since I reserve “instantiation” for the individual’s adaptation of a cultural model to a particular situation via a filling in of specific parameter values (replacing the generic default values coded in the cultural model), I cannot use it (as some have done) for the individual’s actual utilization of a specific cultural model in some particular situation. For the latter case I have used application”, but it is possible that “invocation” might better convey the pertinent cognitive action.67 67

“Instantiation”, which one colleague has suggested for this individual application, would be fine, except that I already am using it in the slightly more abstract sense previously described. In a sense this is just words – and as long as we know what we are talking about it doesn’t matter which words we use. But it is useful, insofar as possible, to have a common vocabulary (if meanings – including mechanisms, relations, etc. – are agreed upon); it prevents a lot of talking past and misconstrual.

Chapter 10 Illustrative Examples

1. Calculating and applying kinterms in Fanti 1.1. The example Let us turn to a simple example embodying much simpler entities than cultural models to illustrate a point concerning the ways in which cognitive structures, including cultural models, do and do not embody automatic instantiation or realization (i.e., application) procedures. The example involves the question of which word to use to address or refer to someone – or the interpretation of observed usage – in the following situation that I encountered in my Fanti kinship study. The principals involved were two men who were cross-cousins to each other. Fanti in the village I studied commonly greeted each other in the morning with a phrase like “Good morning Father”. The situation that triggered my interest was one in which the first man said “Good morning Father” to the second, who responded “Good morning Sibling”. In the Fanti kinterminology, the reciprocal of “father” is “child”, while the reciprocal of “sibling” is “sibling”. (See Kronenfeld 1973, 1980a, 1980b, and 1991 [all reprinted in 2008] for the actual Fanti lexemes and a full explication of their definitions and usage.) One should note that the Fanti had a traditional default of showing as much public respect for seniors as was appropriate and possible – unless there existed some rift and thus a positive decision to “diss” the person. The example might have been taken as an illustration of a system in which traditional (anthropological) reciprocal relations did not hold, or as a case of mis-translation, except for some facts that emerged from a more extensive study of the Fanti terminology. The Fanti kinterminology has three different patterns for extending kinterm reference from a single set of kernel (core, or prototypic) referents. One of these patterns is a Crow-type, matrilineally skewed, pattern. The second pattern is a “Cheyenne-type” unskewed pattern – like a generational or Hawaiian-type pattern, but in which a cross/parallel distinction is made in generation 1, but not in generation 0. The third pattern is a “courtesy” pattern involving the honorific use of kinterms – without a genealogical

190 Chapter 10 – Illustrative Examples basis, but based on relative age and sex. The two genealogical patterns are both considered appropriate ways to designate real kinsfolk, and were both used by most of my informants at one time or another; the courtesy pattern is used commonly, and commonly used for non-kin as well as distant kin in a non-kinship context – much like the use in English of “uncle” and “aunt” for non-relatives or older cousins. Both genealogical patterns were described by informants as technically (denotatively) correct, while the courtesy pattern was described as not technically correct, but used (connotatively) in an honorary or courtesy sense. I have elsewhere (Kronenfeld 1973, 1980a, 1980b, 1991 and 2008) described formal, systematic and empirical, ethnographic reasons for treating the alternative extension patterns as systematic. It is the systematic nature of the patterns, and of their relationship to the shared paradigm of kernel referents, that enables me to speak, above, of normal reciprocal relations, and to say that, in Fanti, the reciprocal of egya or “father” is ba or “child”, while the reciprocal of nua or “sibling” is nua itself. Whatever drawbacks there may be regarding my choice of English translation glosses for the Fanti kinterms, misconstruing reciprocals is not among them. In those analyses I have suggested that the skewed pattern is particularly reflective of lineage relations, while the unskewed pattern more reflects nurturance relations, but I have also noted that I could find no context that consistently elicited either pattern. (In a sense the present discussion addresses why no such context was found.) Thus, in the cross-cousin example we can now see that each man was using the most senior, and hence most respectful available kinterm for the other: “father” vs. “sibling” in the one case, and “sibling” vs. “child” in the other. The system has consistent reciprocals, as does each of the extension patterns, and individuals were totally aware of such reciprocal relations – that’s how I learned them! But individuals did not feel any compulsion to adhere to such systematic constraints in their actual usage – even in a situation in which custom suggested the use of genealogically correct – vs. honorary – kinterms.

1.2. Implications of the example So far I have been treating the terminological classification of kin and the application of kinterms to individuals in a conversation as if these were the same activity. But hopefully the example shows adequately why the two tasks need to be separated. One task, the classification one, asks into what

Calculating and applying kinterms in Fanti 191

kinterm category (by what reasoning) does a given kinsperson fall; the other task, the pragmatic application one, asks – given the relevant classification information (including available alternative systems), given the cultural and communicative implications of the various kinterm categories, and given the implications of choice of one or another of the systems – what term does someone choose to apply to that kinsperson in some given situation. The discussion that follows will be limited to the pragmatic application task. I have extensively treated the kinterm classification task elsewhere (e.g., Kronenfeld 1973, 1980a, 1980b, 1989, 1991, 2004, 2001a, 2001b, 2001c, and 2006a). Past kinship work of mine that is relevant to the application task includes Kronenfeld (1973, 1975, 1991, and 2008: ch. 2). At the end of this discussion we will briefly look at some differences between the two tasks and some important considerations regarding the “cultural models” version of the application task. We can consider the kinterminological system as a kind of native model for the assignment of kinterm labels to kinsfolk (or of kinsfolk to kinterm categories). Such a model is much simpler than what are commonly understood as “cultural models”, and its instantiation – that is, applying labels to people – seems more automatic. The initial complication for the Fanti case is that there are two contrasting genealogical patterns (or models) for extension, and so the user has to make a choice. That choice seems to be made – at least as seen in the given case – for reasons wholly extrinsic to the classificatory models and their instantiation; that is, one picks the outcome one likes and then instantiates the model that produces that outcome. This kind of seemingly backward reasoning is not unique to this Fanti situation; Frake has some illustrations of similar reasoning regarding disease diagnosis in his Subanun Disease paper (1961) and I have seen American examples in decisions concerning whether some given malaise is a bad cold or a mild flu. What the classification model itself provides is an argument linking attributes – in this case, one person’s genealogical links to another person – to an outcome – here, the choice of a kinterm label to use in a greeting. The instantiation of the model is the determination of relevant attributes and the determination then of what would be the appropriate label. But in the decisions just mentioned we see that a separate application decision concerns which instantiated model is actually to be applied (or “realized”) in the actual event – the observed conversation. In the case of the Fanti kinterm labeling decision, the culture does provide some guidelines for model application or choice, though nothing nearly

192 Chapter 10 – Illustrative Examples as stringent as the rules for model instantiation. The guidelines, however, are not so much a matter of a better or automatically more appropriate choice, but, rather, have to do with communicative implications and default presuppositions – and context. That is, the choice among alternative “correct” genealogical labels – assuming a greeting or casual public reference context68 – is understood by Fanti as communicating something about respect, and the default presupposition is that one shows maximal respect by picking the most senior available term. In this context, the choice of a relatively junior term sends an explicit “dissing” message. However, the context is crucial to interpretation. In contexts in which the nature of genealogically based kinterm relationships is the focus, the seniority concern is much less likely to surface, while there will be an increased emphasis on consistency – regarding considerations of both “uniform descent” and “uniform reciprocals”. The preceding discussion is all about genealogically-based usage. But that too represents a decision – a decision with default expectations and communicative implications of non-default choices.69 The default greeting mode in the village when I was there was the use of kinterms (as illustrated above).70 Since the village was small (a few hundred people) with a lot of local marriage, there existed a rich set of traceable kinship links – or, at least, links that I was able to trace, but the default was to use genealogically based kinterms only for relatively close kin. For more distant kin the courtesy pattern of “honorary” kinterm usage was applied. In this pattern (or model), the subset of terms emphasizing nurturance roles was applied to non-kin based on sex and apparent generation (inferred from relative age). The instantiation of this model was not as automatic as was the instantiation of the genealogical models. First, while sex was clear, relative age was a continuum that did not map cleanly onto generation. Second, relative age 68

69

70

I want to avoid the traditional “terms of reference” vs. “terms of address” distinction, since I think it misrepresents what is going on. What that distinction, as commonly taken in anthropology, piggy-backs is something to do with the focus of communication. In what follows I am working off of very old memories re material that I did not explicitly attend to at the time. Thus my ethnographic statements should be taken as hypothetical – even though, in the interest of clarity I am going to omit all the conditionals and hedges that I would otherwise provide. I seem to remember some use of kinterms “down” to juniors as well as “sideways” and “up”, as in the illustration; but I also remember first names being used down.

Calculating and applying kinterms in Fanti 193

was constantly biased by considerations of status (wealth, power, position, etc.). Thus any instantiation of the courtesy pattern always involved some active decisions concerning the attitude that one wanted to convey – from the relatively distant respect shown to senior grandrelatives (nana, the term also used for a chief) through the direct respect accorded to a mother or father (na or egya) to the more equal relationship with a sibling (nua) – and on down to a child (ba)71 – though, as before, the normal default mode was to err on the side of showing respect. My sense is that the use of kinterms in such greeting situations – as opposed to name or other kinds of terms – was a way of emphasizing the kinlike solidarity and closeness of members of the village community – as opposed to the more formal and distant relations that people had with unrelated or unconnected outsiders. In this case, then the use of a non-kinterm option in the greeting context would by itself (over and above any implications of the term actually chosen) be a way of conveying distance and a lack of solidarity. Such distancing would be particularly stark in addressing someone within the normal genealogical range of active kin with a non-kin term; use of the courtesy pattern for such a person would also seem somewhat distancing, though not as aggressive. What we have is a kind of hierarchy or taxonomy of models – kin vs. non-kin, genealogical kin vs. courtesy kin, skewed genealogical kin vs. unskewed – among which a speaker can choose which to apply.72 Presumably the range of choices includes some variety as well on the non-kin side. I did not investigate these very much. I know that abofraba (“boy”) was used by adults for dependent youths, and could be used as a real “put-down” for someone older. I don’t know if nyenko (“friend”) was ever used in such contexts. And, in Fanti as elsewhere, I assume there exist more unusual insults such as our “pig” or “snake in the grass” – and, perhaps, positive terms such as our “angel”. And, while I did not notice Fanti examples at the time, I assume that in any language one always has the option of being creative – say, via figurative language – calling someone something novel, such as a “mountain” or a “cesspool” or a “helicopter”.

71

72

The junior grandrelative, also nana, seems, to my present mind, not to have had the same low status overtones. I would assume that similar considerations affect kinterm choices for dyads who are related through more than one genealogical string (or mapping).

194 Chapter 10 – Illustrative Examples 2. The stages of instantiation and English kin Referring back to the “cousin” and “sibling” examples of Chapter 5, we can now look at how application (or invocation) works. The suggestion is that application is an at least a several stage process: the choice of models (here, classification models) to be considered, the instantiation of those models, the comparison of outcomes (and other relevant features), and then the selection of an instantiated model to be acted on (or “realized”). Many models do carry within them either automatic instantiations or default instantiations, but the choice of model is never simply automatic – and so the actual behavioral response to any given situation is never automatic. To a greater or lesser degree cultures do provide default expectations regarding what model might best apply in one or another standard (or standardized) situation. The choices that people make of which model to invoke in a given situation (or of how to construe the situation itself) do convey important information about their attitudes, goals, knowledge, etc. Hence these decisions often are socially shaped, and thus negotiated in one way or another. While people are not automatically required to follow existing cultural precedents in their choice of models – or in their construction of chosen models – their freedom is effectively constrained (though far from eliminated) by the requirements of social living – that is, by the requirement that relevant others understand what the actors are doing well enough to respond appropriately (and, ideally, as desired). This effectively limits us (as individuals) in most cases to modest variations on prior forms. It should be noted that the models that we are speaking of (immediately, narrow models for applying kinterms to people in some social context, but more broadly, cultural models for action as well) serve several related purposes. They can be prospective models that guide our behavior in some situation – and this seems how we most commonly think of them. But they also can be retrospective models that enable us to reason back from someone else’s behavior to an interpretation of that person’s aims and attitudes. At a more abstract level, they can provide us with conceptual entities which we can use to construct more complex social, economic, or political structures. For example, what role relations would one try to build into a rigorously honest and efficient organization – as opposed to into a solidary and mutually supportive one – and thus what kinds of role terms would one want to encourage the use of in such an organization? We similarly need to note that such models are, at least often, utilized as if they are fixed and timeless even while in fact they are always changing – some more rapidly and others more slowly. The appearance of timelessness

The stages of instantiation and English kin

195

or fixity seems a prerequisite for basing future actions on them and for interpreting past actions through them. But, since our individual construction of our productive individual representations (understandings, if you prefer) of these shared models is based on the experience we each individually have with their applications, and since such experience is a function of the ubiquity of situations and their attributes as much as of the given definitions in some momentary form of the model, successive learners through time learn successively different – if minimally different – versions. The lock that keeps these different versions adequately similar to one another is the communicative or mutual action one, but it is, if you will, a satisficing rather than a maximizing lock. The kind of “drift” that happens with language also happens with other shared cognitive models – and, in many cases, maybe a lot more rapidly. For instance, in English kinship we have the example we have seen – of the shifting connotations of the term “cousin” – from Renaissance times when it was a term of solidarity and apparently could be extended as such to non-kin to more modern times when its connotations are more of distance and thus when its extension to non-kin would be meaningless. More recently we have the taking over in English of the anthropological term “sibling” and its continuing expansion at the expense of “brother” and “sister”!

3. Cultural models and romantic love When we move out from the narrow and relatively tight world of the application of terms to kinsfolk to a broader world of cultural models, some other insights emerge. Cultural models are a device (in part) for the interpretation of experienced events and the planning of future actions in the chain of events. But it would seem that the interpretation of events – the models through which the events are interpreted – can sometimes change in midstream. Several models can contain similar scenes, and so the model or scenario (with its interpretation) inferred from a given act or scene can change as more evidence is experienced.73 73

I have in mind a cultural version of the kind of sentence interpretation example offered by the neuropsychologist Karl Lashley in his 1951 Hixon Symposium discussion of the Problem of Serial Order in Psychology (Lashley 1951). He there anticipated the essence of Chomskyean transformations – but in a form suggesting a much wider applicability and richer interpretation. Charles Frake (1961) also provides some simple semantic examples.

196 Chapter 10 – Illustrative Examples For instance, in the world of romantic love, a deep and full kiss can be suggestive of a genuine and heartfelt loss of self in the love for and admiration of an other. And a person being thus kissed by an appropriate love object in the right circumstances might infer such love and admiration. However, such kisses have also been known to have been performed more coldbloodedly and predatorily as an expression of something more like lust – in an attempt to casually seduce the other. The follow-up to the kiss – or the follow-up to the consummation led to by the kiss – can sometimes lead the receiving party to reinterpret the kiss’s significance from love to lust (or, at least in movies and sitcoms, from lust to love). The kiss itself, and even surrounding events, can be the same for the two scenarios; the information that triggers the re-evaluation can be indirect and from the outside – say, information about the person’s recent history. And, to complicate things more, the initiator (the would be lover or player) may also be unsure of which scenario applies, sensing only the deep desire for the kiss and attendant expressions of affection. Again, as with our kinship examples (at least the non-genealogical ones), the various scenarios (or models) entail74 fairly automatic instantiations. The scenarios and their instantiations are central parts of the culture in question. But there is nothing automatic or ineluctable or firmly predictable about the chain of behavior – because of the choices that exist regarding which scenario to invoke and how specifically to play it out ( – perhaps, sub scenarios: say, for example, go slow, building up warmth and affection vs. get carried away in the white-hot throes of passion). And better psychological studies of the participants do not automatically resolve these problems of understanding or prediction; sometimes the actors themselves – even while aware of the alternative scripts available – are not sure which they are following. It is on the basis of thoughts like the preceding that I suggest that cultural models are best understood not as internalized behavioral imperatives, but instead as a kind of library of procedures, or scenarios or games – much like the libraries of procedures included in computer program compilers – that can be invoked as guides to producing some effect or for interpreting the implication of some actor’s behavior. While kinship and love are not unrelated, it does seem that much of life falls within the range outlined, and thus that attributes of instantiation, invocation, and realization that we are discussing here are fairly general for a wide range of kinds of cultural models – whatever particular shape or formal structure these may eventually be shown to have. 74

if somewhat probabilistically

Cultural models and romantic love

197

As a tentative take home message about instantiation and application (invocation), consider the following. Each cultural model can be considered to be a kind of game, where some games have loose rules while other have tight rules. The models do often entail significant aspects of their instantiations (i.e., the rules – including goals, scoring, penalties, moral vs. immoral fouls, etc. – do tell us much about how to play), and in situations in which actors (or players) are basing their behavior on a given model it should often be the case that their behavior will fit the form of the instantiation entailed by the model. And similarly, where models are used for interpreting the behavior of others it seems likely that behavior will be seen and categorized as maximally consistent with the instantiation implicit in the model. The reason for this is not that the models have formal rules which are legally enforced (though such can sometimes happen), but that social life requires complex shared understanding of shifting situations, and thus requires that inferred models (what we have said cultural models are) be productive representations that are productively applied – as opposed to any memorized or purely habitual behavioral routines. At the same time, it can sometimes be very hard to reason from observed conditions or behavior to future behavior – because of the broad choices actors (or game players) have concerning which game to play, and because of individual variations concerning how to push the rules and how to carry out the plays.

Chapter 11 Problems – messages vs. codes

1. Shared “codes” and individual “messages” A major problem in anthropology and related areas – going back at least to early Structuralism and seen equally in Bourdieu’s (1977) early response to Structuralist excesses – has been a confounding of specific “messages” with the “codes” by which the messages are constructed and understood. This problem has been accentuated or exacerbated by the fact that the “codes” of both language and culture can only be perceived – whether by natives or by scholars, by new learners or by old users – indirectly via the medium of messages constructed in them. This last is to say that our individual internal representations of each of these codes are simply abstractions or regularities which we have pulled out of our experience behaving with and communicating with other people. As such, these abstractions are – in one important sense – similar to, and “of the same order as” all of the other more or less empirical abstractions about the world that we pull out of that same experience – such as what will hurt our hands if we hit it, how much things look to weigh, how much arc to put on a basketball, how long to bake a cake, and so forth. Our different systems of abstractions from that continuum of primary experience simply represent different questions that we ask of the experience. (I do not mean to deny that our biological make-up probably predisposes us toward some kinds of “questions” or suppositions – as with our drive to construct language, as well as our disposition to throw things with some accuracy. But such dispositions still work on, and build on the common experience that I have spoken of.) The “problem” seen by much of anthropology, including the Structuralists and their opponents, has not been that of disentangling the individual from the collective – the Durkheimian and Saussurean problem that is central to this book – but instead has been a search for the “right” code – one that would correctly specify peoples’ actual behavioral choices. And then, when such a right code seemed impossible, opponents of Structuralism and, by extension, of other linguistic-based approaches went on to infer that codes themselves (aka rules, grammars, …) were impossible to construct for human behavior.

Shared “codes” and individual “messages”

199

A confounding problem has been the equation of “cultural models” (in the general sense of our models of some bits of cultural behavior) with cultural “rules” or “grammars” (some logical specification of a set of operations or decisions that produce the behavior in question), and those, in turn, with the “rules” of linguistic description, especially those of early Transformational Linguistics. The psychological operation of linguistic “rules” themselves seems still quite deeply not understood, and insufficiently researched, but there exist good reasons for distinguishing descriptive rules from the operations that natives use to produce the forms in question (remember Chomsky’s [1957] statement that a transformational, generative grammar is not a model of the actual operations of either speaker or hearer and see Kronenfeld 2006). In addition, the different functional role of culture (vs. language), the very different communicative constraints that it meets, and the much more significant role of substantive experience (in addition to, and along side of communicative efficiency) in culture all argue for the possibility that cultural “rules” might well have a very different form and a very different mode of operation than have linguistic “rules”. Possibly something like “culture’s greater systemic looseness” (the idea of culture as a congeries of more or less autonomous subsystems vs. a single coherent system – Lowie’s thing of “shreds and patches”) should be added to the preceding list, but I am not sure whether the looseness might not just be an effect of the listed properties! In particular, anthropologists have seemed often to confound people’s actual internal personal procedures for doing something with the models or patterns or representations that one draws on in ones analytic characterization of these actions – and thus treat actively constructed behavior as if it were the ineluctable following of fixed cultural rules. The models or patterns, here cultural models, seem to be some mix of grammar-like rules and prototypic gestalten; cultural (and sub-cultural) systems seem to include a variety of cultural models. In their selection of models people pick from among the possibilities they are aware of. Individuals within a culture may have differing levels of knowledge of the culture’s cultural models; the amount of variation in levels of knowledge depends, inter alia, on variation among people in frequency and intensity of their encounters with different cultural models, in the importance of the topics of the various cultural models for their lives, and perhaps in the relevance of the given cultural models to higher-level patterns (values, goals, etc.) that individuals work out for their lives. The specific cultural models within the set of cultural models existing within a given culture (even regarding some single topic or issue) may be

200 Chapter 11 – Problems – messages vs. codes logically inconsistent with one another – and in important, problematic, or contested areas seem almost certain to encompass such inconsistencies. The inconsistencies point to the fact that cultural models have a variety of uses. Besides their use as templates (or models or building blocks) for individuals’ constructions of their own behavior, cultural models are also used in individuals’ interpretation of the behavior of other people in their culture or society, in their constructions of counter-factual negative examples (of the sort explored by Lévi-Strauss [e.g., in 1960] among others), in their coding of cultural differences, and so forth. The internal procedures of construction and interpretation are sometimes spoken of, unhelpfully, as “rules”, and are parallel to peoples’ procedures for constructing actual sentences (Saussure’s parole), while the cultural models are parallel to peoples’ grammatical knowledge (better seen in terms of the patterns of Saussure’s langue than in terms of the rules of Chomsky’s grammar).

2. Nature of cultural models, and their relationship to psychological schemas Cultural models have been developed within cognitive anthropology, on the model of schema theory in cognitive psychology and artificial intelligence. Schemas can be seen – perhaps over-simply – as cognitive structures relating values, goals, plans, and knowledge to action (including, inter alia, the process of action, the interpretation of action, and conversations about action). As psychological entities, schemas are created within and used by individual minds, even if much about many of them may be common to many (or, even, all) people – whether for reasons of innate disposition, common experience, structured learning, or whatever. Schemas, thus, are seen as the detailed models that produce the specific and detailed behavior of individual people. At first flush cultural models might be seen simply as schemas that are widely shared. The problems with such a simple extrapolation from individual representation (schema) to collective representation (cultural model) are: 1) Individuals seem to vary a lot from one to another in the details of their particular versions (including actual actions, specific values and preferences, interpretations, uses, and so forth) and 2) There seemingly exists no mechanism by which the existing full schema of one individual (say, a parent) could be directly transferred to another (say, a child). Most of what we see as cultural knowledge is inferred from various kinds of direct and indirect experience without actually being taught, and often without even being raised to consciousness. Individuals, then, are not taught finished cul-

Nature of cultural models, and their relationship to psychological schemas

201

tural models (structures, knowledge, behavioral patterns, means-ends relations, etc.), but have to infer and construct them from some generally incomplete set of experiences. Such construction takes some amount of repeated exposure coupled with feedback from whatever activities to which the learned stuff has been directed – which is to say that only the repeated and salient elements of experience get incorporated into the learner’s emerging individual representation of what people do (know, etc. – the combination of knowledge and action that constitutes their individual schemas). Individuals’ representations of what people do thus are relatively abstract; they also should be similar across people within a given community who have similar experiences, and seem often appealed to in conversations. The set of individual representations – held by a set of people in a community – of what some category of people (or other agents) “do” can serve as a first approximation of what I mean by cultural models. By virtue of being abstract, shared, and received (more or less passively) from ones community, cultural models seem intrinsically different from the active, constructed individual schemas of members of that community – even if cultural models are inferred by their users from such individual schemas, even if anthropological theorizing about cultural models has been based on schema theory from psychology, and even if, in turn, newly enculturated individuals construct their representations of cultural models from their experience of individual schemas. Cultural models are conceptual entities, and ones which people know well enough to recognize (in the same way in which people recognize a person’s face, or a pick-up truck, or a waitress in a restaurant) as opposed to novel or strange entities which people have to carefully describe and analyze if they are to deal with them successfully. As such cultural models simplify perceptual problems and give people one way of dealing with the glut of information that we often are exposed to. They give people pre-packaged, canonical prototypes which they can extend to a variety of variable specific situations – in the manner described for semantics by Kronenfeld (1996). To recapitulate: Individual people and the individual knowledge, abilities, thoughts, goals, feelings, et cetera of those people do, of course, exist. Neither culture nor cultural knowledge structures (such as cultural models) replace or dictate what is in individual minds. Instead cultural structures provide people with sets of conceptual tools that provide standardized linkages among concepts, goals, emotions, context, and so forth and that come with the presupposition that the relevant knowledge and understanding will be shared among and recognized by members of the relevant cultural com-

202 Chapter 11 – Problems – messages vs. codes munity. These tools can be used in communication, in cooperative endeavors, in social manipulation, in mobilization, and in the interpretation of the activities one sees going on around one. 2.1. Cultural models are social Since cultural models are social the following inferences can be made. a) They are collective, and hence have all the various rough edges of individual schemas and of variation among individual schemas get “rubbed off”. The closest they could approximate actual individual schemas is as a kind of average or lowest common denominator, but more likely and often they approximate such schemas as generic abstractions. b) They can not be directly in themselves motivational, but at best can only be standards or models that can be referenced by individual schemas or self-plans. Actual motivation has to come from within, and hence has to be individual. Cultural models thus have a kind of “referential” role, and thus do not embody many of the properties we associate with individual knowledge structures. For instance, a colleague asked me about what happens to the difference between “declarative” and “procedural” knowledge. He said, “Goals can be declarative, motivating, formal, tentative, whatever. Steps in a process are declarative if you’re just talking about the process in an abstract sense, but procedural if you’re doing it. Various slightly-motivating and sort-of-action-oriented concepts bridge the gap. This is interesting and needs discussion. I wonder where it leaves the whole distinction, so basic to so much herein and in psychology.” My sense is that our cultural models are representations (pictures, scenes, scenarios, or whatever) that are in themselves neither declarative nor procedural, and that we refer to these collective representations both in framing what we do and in thinking about what we or others might do. These reference scenarios are like “other worlds” or “imagined worlds” that presume the existence of the given entities and actions with associated outcomes and so forth. We, as individuals in cultural communities, consider alternative cultural models and choose to apply ones which seem relevant to our own potential actions in a given situation (whether these actions be real, discussed, or imagined) or to the actions of others (ditto), and that seem useful and socially plausible. And surely emotions – including posited feelings of participants, feelings of imaginers, etc. – are part of these representations.

Nature of cultural models, and their relationship to psychological schemas

203

c) They have in some sense to be intrinsically tied to communication – often via being actually coded linguistically, but otherwise, even if non-verbal (cf. Piaget’s sensory-motor knowledge), they still are social and entail communication beyond their specific content – i.e., by their evocation in one specific situation or another, by who uses them to/re whom, by their relationship to other models available for application to the given situation, etc. This communication function has to be an important part of them or they would not get used in the ways that are needed for their social reproduction – that is, for new people to learn them and their use. This communication aspect, and the social manipulation necessarily entailed by communication, means that the accuracy or completeness with which cultural models represent individual motivation is always, to a greater or lesser degree, compromised.

2.2. Cultural models are shared and distributed Other conclusions follow from the shared and distributed nature of cultural models. Cultural models represent variously shared and distributed cognitive structures. If culture is indeed the metaprogram that enables our social division of labor (our “organic” “organization of diversity” – in A. F. C. Wallace’s [1970] phase) then to work it must have a couple of properties. Some part of it must represent the different kinds or amounts of knowledge that different ones of us have and that each of us contributes to the larger enterprise. Some other part of it must represent the shared framework within which and via which our individual components are coordinated and interlinked (this is the aspect that Romney’s consensus measures give us a way of getting at – see Romney, Weller, and Batchelder 1986; Romney Batchelder, and Weller 1987). It is cultural knowledge in this interlinked mix of shared and distributed forms that cultural models can best be seen as representing. The function of cultural models is to enable social coordination – of information, of abilities, of interests, of motives, of degree of concern, of goals, of values, and so forth. This gets us back to the idea that cultural models are not directly instantiated in the individual schemas that actually produce our behavior, but instead provide reference points or frameworks via or within which our individual schemas operate. My sense of this kind of referential function of cultural models and its relationship to conversations (as negotiations) between people trying to arrive at a common response to a problematic situation derives from Carol Mukhopadhyay’s discussion of the use

204 Chapter 11 – Problems – messages vs. codes of “cultural precedents” in her Ph. D. dissertation (1980) on the sexual division of domestic labor in an urban American community. One thing we learn through interaction and observation, as children but also later, is these models. We learn them via the same (Piagetian) feedback process of constructivist interaction with the world around us that shapes both our interaction with that physical world and our attempts to form internal representations of it. These representations include representations of the physical world and of other individuals, but also representations of various levels of collectivities. These representations of individuals include modes of behavior, motivations, goals, standards, and so forth; they are like models, which we home in on through the Piagetian adaptive processes of assimilation and accommodation. My presumptive hypothesis is, first, that we understand the behavior and motives of others by reference to – for instance – what would cause us to act in the given way (if we shared the attributes and/or context of the person/s in question), or (in the relevant context) what action some given motivation might lead to – that is, by some version of the old intersubjectivity assumption. As a codicil we can add the notion that “others” include animals and other beings – at least insofar as we can find ways to relate our own experience to what we see these beings doing or communicating; that is, we anthropomorphize them. Second, I presumptively hypothesize that we understand collectivities not as averages or as distributions but as if they were individual cognizing creatures subject to anthropomorphization in the same way as other creatures. 2.3. Role, nature, and functioning of cultural values This approach offers us a new approach to such old and tantalizing anthropological problems as the role and nature of cultural “values”. If values are embedded (along with other relevant stuff relating to action, context, goals, etc.) in cultural models then they do not ever become part of the direct motivation of behavior, but are always only taken as available reference points. In problematic situations we (as individuals, internally, – as well in external social situations) adduce relevant models and “negotiate” among them much in the way that Mukhopadhyay’s informants adduced and negotiated among relevant cultural precedents75. This means that values never 75

For example, one spouse might claim that the husband should prepare the taxes “because it is a heavy job” and men are supposed to do the heavy work, while the other spouse might claim that the tax preparation was part of the wife’s

Aspects and attributes of cultural models

205

become an intrinsic, constitutive part of the self, but are always only part of the compendium of material (on right vs. wrong, on goal-action links, on social context, etc. etc.) that the self considers in arriving at some serious decision. For example, “thou shalt not kill” and “I must defend my country” are not contradictory injunctions for a soldier, but are values tied to cultural models which the soldier consults in relevant situations; in those situations the soldier makes a determination of which model is more apropos to the givens of the situation.

3.

Aspects and attributes of cultural models

3.1. Prototypes and prototypicality The selection of a cultural model to apply to a given situation – say, as a guide to action – depends on the determination of which cultural model’s prototype (a concept) best fits the applier’s mental representation of the relevant situational facts. “Best” would involve some combination of the fit to the situation’s shape and detail, to the applier’s desired outcome, to the applier’s presentation of self, to the available resources, etc. Similar but different (in obvious ways) considerations would shape the selection of a cultural model when used as a guide for the interpretation of the behavior of others. That is, the same immediate situation may elicit different structures based on other issues, as we can see in Carol Mukhopadhyay’s example of one family’s tax preparation. “He” does their taxes because the job is “heavy” (i.e., the man’s job), but that is only because “she” doesn’t want to do them. Otherwise, or in an other family, “she” could do the job on the basis of the claim that the major applicable attribute was “indoor (in the house) work” or “household maintenance” (which would be the woman’s job). As long as the other partner did not particularly want to do the job, any household maintenance realm; neither spouse might bring up another cultural precedent (which might decide the issue in another family) which has men doing major family financial stuff, since neither liked its implications. In another example, the wife might accept the relevance of a maternal nurturance role and be willing to carry a 35 pound child, while arguing that lugging a 25 pound bag of dog food represents heavy lifting – and is her husband’s task. These are adapted from actual examples of Mukhopadhyay’s, and here are only intended to provide a quick sense of what I mean by “negotiat[ing] among relevant cultural precedents.”

206 Chapter 11 – Problems – messages vs. codes credible claim was enough to settle/justify the task allocation. However, if either both wanted to do the job or both did not want to do it, then the discussion would resolve down to a combination of who cared more and who’s claim basis was more direct or credible; in the above tax preparation case the obviously metaphoric nature of the “heaviness” of the job would weaken the claim that he should do the job while something similar might apply to the “indoor” aspect – and so other attributes might get brought into the discussion in order to finally resolve it. Thus, the same situation, in different families whose members have different desires, produces different outcomes. I pretty much think of prototypes as gestalts, but am not certain that feature specification definitions always need to be ruled out. Prototypes are not existential primes, but are items in one or another system – and so what they reflect or result from or do depends on the system. Semantic prototypes of word meanings might differ from cultural prototypes of relevant cultural actions; and individuals might well have their own personal prototypes for these things – based on their own individual experience. Presumably individuals “know” the difference (my quotes are a way of saying that the knowledge may not be self-aware, but may be only implicit or subconscious.). Word meanings and cultural forms are both social, and hence involve communicative and interactive constraints – and so the individual’s prototypes in these areas have to fit those constraints (even if, in other respects, they might vary from one individual to another). There would be no such communicative constraints on an individual’s private prototypes. I have been speaking of cultural models as representing prototypic situations which then are adapted to specific actual instantiations and then realizations in particular events. I want to go a bit further with that line of thought and suggest that the kind of extensionist approach which was applied to word meanings in Kronenfeld (1996) actually applies more generally to our full range of shared/distributed cognitive structures. This line of reasoning suggests that the cultural model prototypes are not to be understood as bare bone abstractions but as filled in “pictures” of unmarked (see Marking Theory as developed by Trubetzkoy, Greenberg, and various others) typical situations. “Typical”, as indicated above, refers to some interaction of “frequently experienced” and “clearly and focally embodying the relevant functional relations”. It is this idea of a scenario totally constructed of default actions, players, values, motivations, etc. that I tried to capture with the “commedia dell’arte” comparison. Instantiation, then, is not a matter of filling in blank parameter values, but, instead, is a matter of replacing default values for relevant parameters with situationally specific ones.

Aspects and attributes of cultural models

207

Prototypes serve an important systemic function. They give us a way of applying precise past knowledge to a potentially infinite variety of unanticipated future stuff (forms, experience, intentions, etc.) in a way that enables sensible and consistent behavior – especially re semantic and cultural systems. The ways by which relatively fixed prototypic meanings (referents, instances, …) are extended (i.e., applied in context) to the range of stuff we actually encounter, talk about, and interact with is one major source of the productivity exhibited by linguistic and cultural systems. I think prototypes can be empirically described, measured, and analyzed. But it can be tricky. This empirical problem regarding “cultural models” is my major present research focus. For special cases, such as kinship, it is all fairly well worked out. But no one yet has come up with a good characterization of the general case – of the sort that would then enable tight empirical characterizations. We can get informants to tell us how some given exemplar of a concept differs from the most general or ideal exemplar, and we can get informants to tell us wherein their own individual views or understandings seem (to themselves) out of the mainstream. The problem is not simply one of empirical description, though, but rests on a theoretical problem: we need the right concepts before we can measure and describe instances of them; this is the problem that physics had with “weight” before eventually physicists figured out about “mass”. The problem is what has sometimes been described as “cutting nature at the joints” – with the proviso that one first has to figure out how to recognize the relevant joints and then find where they are before one can cut at them! If I am right about the prototype/extension conception of cultural categories, and about what I have said about instantiation and realization, then the link between cultural model and experienced event is usually neither automatic nor mechanical. We attend to the context, to our own and others’ aims, goals, and values, to the aspects of the situation that match one or another prototype, and so forth in deciding which cultural model to apply to any given situation. People with different goals or perspectives or vulnerabilities, or … might turn to different cultural models. Presumably the closer the situation approaches that of a given prototype the less wriggle room there would be (at least in regard to social – hence collective – things). Carol Mukhopadhyay’s Ph. D. dissertation (1980) contains wonderful examples of variability in the extension of “cultural precedents” (what we are speaking of here as aspects of “cultural models”). A different kind of perspective issue concerns whether players or observers are looking at a private (in their minds) unrealized instantiation, at a more public discussion of an instantiation in the context of its potential or

208 Chapter 11 – Problems – messages vs. codes hypothesized realization, or at a concrete situation for which a cultural model which might serve as a basis for interpretation is being sought. Cultural model prototypes, as in the semantics of language, are – I think76 – inferred from commonalities across ones experience of more or less similar events. But, unlike with language, the learner does not have the anchor provided by the shared Saussurean signifier (“sound image” of the word in question). Similarity is in part an intuitive thing based on general experience in the culture. But, also it would seem to be dependent on contexts of use, what is accomplished, and so forth. The prototype’s delineation would first of all be a function of the salience of the alternative forms (depending on the frequency with which they are encountered and the relative clarity of their delineation from extraneous noise), but then would be significantly influenced at least by the function to which the cultural model is directed, the smoothness of the fit of alternative forms with the function, the range of contexts in which the alternative forms are used, and the guidance provided by what seems marked vs. unmarked. 3.2. Alternative or conflicting cultural models Conflicting cultural models (or rules) are surely present within any given culture; even any simple study of aphorisms shows that our cultural armory is replete with weapons for any and all purposes. But in regard to the implications of such conflicts for our anthropological interpretation of cultural models – or for our native conceptualization of them – I would want, really strongly want, to separate cultural models and their content from meta-rules for the instantiation of these cultural models. That is, the presence of contradictory cultural models causes us no distress (except when some dilemma foregrounds the contrast – and then it is the dilemma that distresses us, not the inconsistencies among the cultural models). What distresses us is the problem of deciding which cultural model to apply to a given situation and how to apply it. As abstractions, the cultural models themselves contain no such guidelines; among our wider cultural knowledge (a “Mishnah” to the 76

The “I think” hedge is by way of saying that what follows are considerations that either follow from the theory or are based on an analogy with the prototypes of my semantic theory. Absent better formal specification of the definitions of actual empirical examples of cultural models, such characterizations are necessarily speculative. But I think that such informed and motivated speculations are useful for the development of empirical measures, and so I am offering them.

Aspects and attributes of cultural models

209

cultural model’s “Torah”, if you will) are traditional expectations of who uses or doesn’t use which cultural models for what reasons, and also among our wider cultural knowledge is our meta-knowledge of changing trends in these expectations (a “Gemara”, if you will). And then our personal preferences – what we like, wherein we see our individual selves – exist (are constructed, elaborated, and interpreted) within that frame. It still remains the case, though, that one important research problem, once actual cultural models are successfully identified, described, and framed, is to figure out how they interact. Clearly, under some (but, equally clearly, not all) conditions, strong cognitive dissonance concerns rise to the fore, and force us to confront such contradictions – and, I think, in some sense in such situations, more general “rules” trump more local ones. As an example, consider desegregation in the 1960s. Going into the process the abstract belief systems of many white Americans apparently included “all men are created equal” and “colored folks are inferior”. External conditions – such as a good economy, a great and painful overseas war just recently fought over similar issues, rising and militant Black consciousness joined with a willingness to fight for equal rights, and so forth – clearly played some important role in raising the given contradiction to consciousness and making it a source of pain. Once the contradiction was foregrounded, enough Americans – especially youth and, even (especially ?) those in the South – got caught up sufficiently in the dilemma to lead to a public demolition of the second proposition (even if, to be sure, it was not so effectively demolished in all private hearts and minds). 3.3. Degrees and aspects of specificity of cultural models Cultural models – even those that seem to be rule-like definers of some activity – are not necessarily to be seen as rigid rules for the execution of that activity. They can – as in the “rules” of jazz (thanks to Gene Anderson for the example) – specify “spontaneous improvisation”, even while they also provide a framework within which the improvisation is to take place. The apparent “rule” of some mid-to-late 20th century art forms to the effect of “Thou shalt break every possible rule” was less random than that conceptualization implies, and more tied to known and structured patterns which were being rejected or undermined. This is to say that there do exist guidelines (grammar-like “rules”, if you will) for improvisation. And such “rules” did/do change over time. And, maybe, indeed, it is the emergence of a “rule” which thereby highlights (raises to consciousness) a new potential innova-

210 Chapter 11 – Problems – messages vs. codes tion (i.e., a regularity to be broken). We can’t break “rules” which don’t exist; can’t play off of regularities that are un-felt. I think this has a lot to do with all the creative arts – even including creative work in science, technology, baseball, and the art of daily life. Furthermore, while the rate may vary, it is clear that all cultures are always changing (if nothing else, consider the analog of linguistic “drift”). If cultural models are to exist, then it follows that they too must be capable of changing. We know that language – grammar, word meanings, registers, etc. – is always changing, and that change does not lessen the temptation we have to treat it as a formal, rule-bound phenomenon! 3.4. Where we have good vs. weak vs. no cultural models We do not have effective cultural models for all areas of our lives, and the cultural models we do have are always subject to change. Since our knowledge of cultural models – and, behind that, the very existence of particular cultural models – depends upon experience, areas of new experience within a culture may well not (yet) exhibit well-worked out cultural models, or new experiencers may well not yet have learned the ones that exist. Since cultural models are collective representations, and thus depend upon social interaction, it is also possible that, in areas (even of common experience) that we don’t share with others and that we don’t much talk about, there might exist no good cultural models. 3.5. Rules for breaking rules The case of “rules for breaking rules”, such as what to do when the traffic light gets stuck on “red”, and of understandings regarding “rules” which are not always followed, do not represent ambiguities or contradictions in the cultural models themselves, but instead are best seen as issues of instantiation and application or invocation. We have a cultural model and we have knowledge of when to follow it, when to think about maybe not following it, and what the conditions (involving, maybe, amount of traffic, presence or absence of cops, our own private self-conception, our haste and reasons for that haste, etc.) are that speak to the decision. That is, this is not a problem with cultural models themselves, but rather has to do with how we interpret them and what we expect of them.

Aspects and attributes of cultural models

211

3.6. Variability in cultural models A problem perhaps seems posed by our varying individual human natures (as personality differences, skill and ability differences, differences in taste, etc.). Partially different people avail themselves of different cultural models as models for their own behavior in one or another specific area. Such pressures may well be one of the forces (along with differences in context, stake, knowledge, etc.) that drive the creation of alternative cultural models. And we each use different cultural models to interpret the varying behavior of our different acquaintances (in any given area of culture). And, here, it is time to remind ourselves that cultural models are not simply or only models for (our own) action; they also exist as models for explaining, inter alia, the behavior of others. And we do have our Talmudlike commentaries concerning, inter alia, who is likely to follow which model in which circumstances, and what might be the reasons for such behavior.

3.7. Instantiation issues in apparent variability or vagueness of cultural models A problem is raised by the apparent vagueness of some cultural specifications (thanks, again, to Gene Anderson for the example) – e.g., how hard is “too hard” when pressing a friend for some decision, how much is “too much” when drinking or eating, etc.. I tend to see this is an instantiation issue. I don’t see cultural models so much as “vague” as “abstract” or “generic”. Instantiation, then, can sometimes be simple and straightforward (the red light example) and other times be complex, variable, situationally specific, etc. Some cultural models exist only to define abstract standards; some to explain bad behavior (e.g., in sex, the “player”), etc. Some, like the no sex before marriage, seem, today, to continue to exist only as a standard to press sexually mature but socially and economically immature children to adhere to – and the reason for its continuing absolute phrasing (going back to when it was a more absolute rule) has to do with avoiding discussions of the “how old is ‘old enough’” type. And behind all that is a continuing culturally shared realization that sex is heavy stuff, carrying much potential for pain and disruption if not handled with care.

212 Chapter 11 – Problems – messages vs. codes 3.8. Variability in cultural models vs. in individuals’ knowledge of them As many have noted, we learn from interaction and from generalizing from our interactions – and, indeed, seem to have some sort of inborn disposition to seek generalizations across the detail of variant versions of any given situation. And, as many have noted, our generalizing process at any given moment is not equally far along across all the areas of our interaction; in some areas we have “it” really figured out while in others we have only the roughest of approximations. It has been sometimes suggested that such tentativeness and provisionality (or that such a range of certainties) should be built into our conceptualization of cultural models. I disagree, and instead would phrase this slightly differently. We presume the existence out there of a (potential) rule (or regularity) that is clear, sharp, etc., and then see our own individual understanding of that as sometimes only approximate or otherwise inadequate. Different cultural models (or other kinds of collective representations too, for that matter) are known varyingly well by different people. One implication is that there may well exist cultural models (or other collective representations) that people are aware of and consider to be “out there” but that, in fact, no one knows well! The idea behind this is what I think is a real tendency we have to presume structure and to ascribe it to some social entity. So the pattern that we actually experience in some given case might not clear enough for us to figure out more than the vaguest sense of a cognitive structure, but we still, in that case, presume that such a structure must be there, and make some guess about the relevant “there”77. This situation is 77

Eugene Anderson, a colleague of mine, reading a draft of this manuscript, offered the following comment regarding the messiness of cultural models and the levels at which they are held or imputed. The illustration includes some illustration of how we make sense of apparently anomalous content – in this case the romantic feelings of a country swain and by what signs we read him as a rustic: A different question is how messy “love” is – there being no similarity in actual mood or brain waves between passionate erotic love, love of one’s kids, and love of boiled cabbage. Hence the humor in a Renaissance Spanish song by Baltasar del Alcazar about a man who loved only three things: Ines, ham, and eggplants with cheese (no one ever forgets the eggplants with cheese). So when love CM’s are messy, it’s partly because we can never figure out what love is, in the first place, let alone apply any simple models to it. “Captured by Love,” by Baltasar del Alcazar, 16th century (and I assure you the Spanish doggerel is every bit as deliberately awful as mine).

Aspects and attributes of cultural models

213

a little like the well-discussed example of the shaman who has to fake finding the illness-causing object in his patient, but who believes in the system (by which such foreign objects are introduced and cause illness) and is certain that there exist others (such as, perhaps, his teacher) who really can (and do) find the illness-causing object – in a situation in which we, as external observers, know, of course (!), that no one can do it. Three things have captured my heart’s love, if you please: beautiful Ines, ham, and eggplants with cheese. It was Ines, first, who got into the place, so that I now am dragged by all that’s not Ines. A year I’ve been out of my head since she fed me for lunch some Aracena ham and eggplants with cheese to munch. Ines has got first prize, but I don’t know, I swear, which of them has got of my soul a bigger share. In taste and seriousness the difference’s not worth a damn between Ines, eggplants with cheese, and Aracena ham. Ines in her beauty; ham from Aracena; eggplants and cheese, that dish of ancient noble Spain. And I faithful in love judge them of equal degrees – all are one: Ines, ham, and eggplants with cheese. Of course Baltasar is adopting the persona of the simple rustic swain who hasn’t quite figured out the difference between his heart and his stomach but loves all the more sincerely for it – a stock theme in Renaissance Spanish poetry. (Baltasar’s simple poetry with self-consciously awful prosody “mistakes” removes doubt.) So we have a cultural model here.

214 Chapter 11 – Problems – messages vs. codes I would also add that our inborn predisposition is also to separate out rules for message construction from statements contained in such rules. That is, on the one hand, we keep trying to abstract out grammar-like regularities, prototypic referents or instantiations, and so forth, while, on the other hand, we use these grammar-like regularities as frames for isolating out and pointing attention to the relevant content – that is, whatever is the particular feature of some unique situation that we wish to call attention to, address, operate on, etc. In other words, in our actions I do not see us as natives at all confounding the patterns – the collective representations – with the actual actions of actual individuals in the manner that anthropological analysts seem to keep doing. 3.9. Conformity Then there is the question of conformity. It has been offered that “If everybody had a different set of grammar and usage rules, nobody could talk. Conversely, the more a group of people communicate (directly or indirectly), the more they need to have the same underlying grammar rules”, and that a similar pressure exists re cultural models. Does intense interaction among a group of people pressure them to bring their cultural models into conformity with one another’s. “Conformity” seems a reasonable inference, but that conformity need in no sense be equivalent to “identity”. We can live – and do so frequently – in social entities which contain individuals possessing divergent internal sets of rules. What such situations do force us to do is to come up with conceptualizations that allow us to understand and deal with such divergence. And since our major way of understanding the behavior of others seems to be by reference to what would lead to or enable us, ourselves, to act in such a manner, that attempt at extended intersubjectivity becomes one push for the creation of divergent cultural models (or rule sets). Even within language itself we each use common grammatical and usage rules to say sometimes very different things. And, even within language (though much less than with culture), different ones of us preferentially exhibit different favored grammatical constructions. 4. Collective representations reprise Having talked a lot about cultural models, as if they simply exist, it is appropriate to remind ourselves that, like language, cultural models have no objective existence outside of the minds of those who use them. As many

Collective representations reprise

215

of us have all along been saying, this is the curious and interesting problem with any and all “collective representations” – whether langue, society’s values, or cultural models. They only exist as individual representations of them, and it is only the shaping constraints of active interaction and of shared experience that effectively make them “collective”, and then only to the degree that these constraints actually manage to come into play. That is, only to the degree that “individual representations” of “collective representations” exist and are similar across the individuals within some community can we speak of the “collective representations” of that community. At the level of individual representations of X, it is “turtles all the way down” – where X can be other individuals, social entities, animals, pseudo animate objects (such as robots or cartoon characters), ad hoc communities, abstract categories of people, etc. etc. “Collective Representations” exist when some social grouping constrains some set of individual representations; the social grouping can be as small and ad hoc as “you and me” or something as big as “America”; the objects of the collective representations can be as local and concrete and dumb as a “number 2 pencil” or as local and individual as “John Doe’s behavior patterns”, or as large and sublime as “the stock market” (“in its infinite wisdom” ?).

5. Looking back My sense is that cultural models are fairly brief (“lean mean machines”, as some old football team was once described)78. That is, they are more like generic templates than totally filled in plans; they specify key parameters via the specification of default values for them and add in enough other detail to make the scene concrete and plausible (even if still general and generic). These models are then, in use, instantiated by users as they fill in the given situation’s actual parameter values and add in other needed parameters with their values. It is these filled-out plans (instantiated cultural models) that people then use to understand or think about particular situations79 (again, whether interpretative or action-planning). These instantiations include the specific particular attributes of the actual people and objects, the nuances of particular points of view and interests, etc.. These instantiations are individual, and part of individual cognition (individual “schemas”), but can be discussed and thus coordinated across a set of interacting individuals. 78 79

A suggestion made by Victor De Munck in a personal communication. Giovanni Bennardo’s observation in a personal communication.

216 Chapter 11 – Problems – messages vs. codes It is also possible for people in a population to share knowledge of commonly occurring situations and thus of major aspects of the instantiations relevant to these situations – as, for example, considering Schank and Abelson’s (1977) restaurant simulations, the kinds of restaurants people in a given group typically go to and their typical reasons for going to a restaurant. Thus, for commonly encountered situations, we (as members of whatever group or groups are relevant) tend to develop more or less standardized instantiations which members of the group are all pretty much aware of and make mental use of. We then adapt these standardized instantiations to immediate local detail, including, potentially, how we want the business to come out, how we feel about those we are dealing with, etc. It is seems that in situations, where we are torn, we are capable of instantiating alternative cultural models (or, if relevant, filling out alternative standard instantiations) – that is, mentally applying the alternatives to the given situation – and comparing how they come out. This comparative application is nicely illustrated in Carol Mukhopadhyay’s study of the allocation of household tasks (sexual division of labor) in an American population in which the wives were all working (Mukhopadhyay 1980, and see Mukhopadhyay 1984). The next level is the application of the instantiated mental model to an actual behavioral situation – involving what I am calling, after Trubetzkoy, “realization”. This is more of an application than an instantiation because a) it involves a person integrating the chosen cultural model(s) into that person’s own individual schemas, and b) it involves the person’s actual actions or interpretation of the actual actions of others, and thus involves a process of “negotiation” – whether directly and overtly via talking or indirectly and implicitly via tentative actions constantly adjusted for the (re)actions of others in the situation. As the action situation on the ground flows, the players can change the cultural models that they are using to conceptualize and plan it – and can do so even fairly frequently if enough is at issue and if the situation is fluid enough. In youthful male/female interactions, for example, such recasting seems maybe common. What is “at issue” concerns likely action outcomes, but also the status of the players’ various egos, their standing amongst their friends and strangers, their sense of their own status as moral beings, and so forth.

Whither now

217

6. Whither now I do offer – though I am aware that there are those who disagree – that cultural models thus characterized can be studied empirically, can be rigorously modeled, and can have outcomes derived from them which can be tested empirically. To do so requires a means of formalizing them – or, rather, proposed models of them. Relevant future research needs arise when one considers the empirical validation of proposed cultural models. Centrally important are the formal and empirical issues delineated in Section 2.1 of Chapter 9. The formal concerns included the clarity with which cultural models need to be defined and their boundaries delineated. On the empirical side we need to see characterizations of cultural models across a much wider range of domains and situations; we need many more examples of how they are used in routine situations and adapted to more novel circumstances. The relationship of cultural models to individual schemas requires further research. Similarly in need of further work is the relationship of cultural models to the language used by members of the given culture to describe or characterize them. Given what I have seen of different kinds of models of proposed cognitive structures, algebraic models – at least, at the moment – seem unlikely candidates for modeling cultural models; either cultural models lack the tight logical structure required or we do not yet understand them well enough. For the moment something more like computer programs seem the best device for modeling the kind of basic story line or picture with conditional contingencies that seems to characterize proffered models of cultural models. Computer simulations represent a major potential tool for exploring potential regularities of process and structure, while being explicit about special local or more ad hoc constraints or conditions – for, thus, giving us the information which will be needed down the line for any more elegant representations (see Kronenfeld 1976 for a discussion of these issues). Another important tool in developing our knowledge of cultural models – their shapes, boundaries, modes of operation, mixture of variability and sharedness, etc. – will be controlled experiments. Ethnographic descriptions, however insightful, seem insufficient to the task. It is too hard to control for variant representations, special personal circumstances, and so forth. The experiments need not be of actual action – though some such in sufficiently benign circumstances might prove quite useful – but can be at the level of verbal reactions to scenarios, attributes, and the like, especially where the aim is to try to sort cultural models from their mental instantiations.

218 Chapter 11 – Problems – messages vs. codes For the moment I have neither any computer simulation of cognitive content nor any careful experiments to offer. But, I have so far offered some suggestions regarding formal detail, and I do in Chapters 10 and 13 offer some substantial evidence that something like cultural models exist – even if I cannot yet provide the kinds of sharp delineation and precise demonstrations that I would like.

Chapter 12 Other theoretical issues and relationships

1. Cultural models – truth status A colleague has asked whether I think cultural models are real entities within native cognition or are just anthropologically created descriptive devices. This is, in part, the old “God’s Truth vs. Hocus Pocus” argument from Bloomfieldian linguistics re grammars and other linguistic analytic concepts. Is an analysis considered to be the truth about what is in speakers’ heads or simply some kind of a summary of the regularities seen in their behavior? That has always seemed to me to be a false dichotomy. Any description or model (grammar or whatever) represents an assertion about the truth of what it specifically addresses and entails no similar assertion about other stuff (even stuff it makes formal use of); no model (theory, description, etc.) is (or is intended to be) true about every aspect of the modeled universe. Descriptions of cultural models as a general form or descriptions of specific cultural models – even if empirically validated and otherwise successful – will never be simply equivalent to the cultural models they purport to describe. It’s worth periodically reminding ourselves of that, but I see no big issue here. A second part of this question involves what kind of status one ascribes to the native actors’ cultural models being described. What kind of reality are they asserted to have. My view of this issue goes back to Saussure’s assertion that langue existed and was the proper object of study for linguistics, but was not a property of any individual (being instead a property of a speech community – a social group) and could not be studied directly (but only through its expression in individuals’ acts of parole). Saussure’s characterization of langue was equivalent to important aspects of Durkheim’s understanding of “collective representations”, and hence relevant to much more of culture than grammar alone. Since human conceptions – at least the non-written or recorded ones that make up the vast part of culture and language – do not exist in any metaphysical space outside of individual human beings, there is a clear literal sense in which cultural models or langue do not really exist. But what allows us to study langue and, hopefully, cultural models as if they exist is the fact that speakers of the relevant languages – and participants in the relevant cultural groups – themselves

220 Chapter 12 – Other theoretical issues and relationships act as if they really exist. That is, people – we – don’t just learn how to speak or act, but learn, too, the rules or regularities of the communities to which we belong, and use these regularities to construct our own individual acts. We don’t learn these community regularities as statistical norms or distributions but learn them as (as if) objectively real patterns (sometimes expressible as rules). We are each aware of the ways in which our own predilections deviate from or match the collective pattern. In sum, we ascribe an objective reality to the collective representations of the groups we belong to and otherwise experience. I like the phrase “collective representations” because it gets out the “rule” vs. “regularity” bind and gets out of the idealistic (vs. descriptive) issue raised by “norms”. The collective representations (my sense of them – whatever Durkheim meant or did not mean by the phrase) are ascribed to the group by individuals, and are then represented in individual minds as individual representations of the given collective representations (i.e, as my native speaker’s sense of the dialect of a community to which I belong or with which I interact). The collective representations are inferred from regularities of group performance as productively and coherently organized patterns. The illusion of the objective existence of the collective representations is fostered by a convergence of individual representations of them; the convergence itself is a combination of a) shared and parallel learning within the given community and b) the homogenizing effect of interaction in shared collective actions and in communication. Cultural models, then, become Platonic ideals of the same sort as circles, squares, and other geometric figures – which we also understand and represent in our minds without ever directly experiencing any actual instances (the instances of, say, circles that we actually encounter are all imperfect and all include irrelevant features such as specific size and color [see Kronenfeld 2006b]). And, finally, there is the question of the degree to which native actors are self-conscious of cultural models, or overtly and explicitly self-aware of their structure and content. On the model of langue/language, I would presume that we normally are not conscious of them. Their patterns (shape and/or content) can, as some reason or occasion arises, be raised to consciousness. We, as natives, never routinely have any formal (or formally well defined) analytic knowledge of them – as Sapir and others noted in framing the modern theoretical understanding of the phoneme. But, also, the issue of our awareness adds another whole set of complications. Often, in my experience, what we actually see is a combination of a bunch of explicit and self-conscious fragments linked together subliminally by a felt sense of pattern, often with a lot of slippage and often with ad hoc simplifications or conflations.

Theoretical models and empirical tests

2.

221

Theoretical models and empirical tests

In our consideration of the issues I am raising, we need to keep three kinds of theoretical models distinct from one another. First, there is the system, process, and role being hypothesized for culture relative to all the systematic things that we do communally without being explicitly taught (or ordered). This book’s characterizations of culture, and of related thought, are general, largely informally described, and largely inexplicit; the aim of these characterizations is to convey what one thinks is going on and what one thinks a relevant genuine theory should be aiming at. Second, we have more explicit – and necessarily simplified – models of some specific parts or aspects of that hypothesized culture – such as the Kronenfeld and Kaus (1993) “starling” simulation. The models aim at evaluating the plausibility (or potential sufficiency) of the actually proposed mechanisms for their hypothesized role in the broader theory. Third, we have models aimed at analytically exploring and/or capturing regularities produced by the behavioral systems targeted by the broader theory. Game theoretic models, maximization models, satisficing models, etc. seem to fall in this category. 2.1. Culture as a mixed structure: empirical implications The overall model that is being proposed for culture in this volume is a mixed structure. Parts of it are highly structured in ways that parallel the mathematically complex but precise structures that linguists have found for human language; other parts are more loosely connected in way that are perhaps more reasonably modeled as “neural net” models. The difference is that, even though neural net models have been proposed for language, these neural net models miss the crucial insight that our understanding and production of language is not predicated simply on the pattern of recurrent arrangement or connections that we experience, but is predicated additionally (and importantly) on the structural models that we leap to (as opposed to carefully reason out) regarding significant structural elements and appropriately patterned relations among these. That is, we seem to have some sort of a drive to construct logically consistent (and relatively mechanical, as opposed to statistical) systems (or patterned relations among abstract classes) from which the consistencies we experience can be deduced (or produced) rather than settling for some more direct representations of the consistencies themselves. In part this is a way of saying that we are bad an dealing with probabilistic or statistical processes or statistical data (very bad, as Kahneman and Tversky [see Kahneman, Slovic, and Tversky 1982],

222 Chapter 12 – Other theoretical issues and relationships and Herbert A. Simon, among others, have shown), but instead tend to try to find a story line (often a kind of “just so” story) that captures the regularities we think we see (as illustrated for some important ecological examples in Anderson 1996). We see a problem, guess at an explanation (or hypothesis), and then act on that insight; if it works, fine, if not we try to guess at a better one, and then try that one. I see this claim as relating to findings regarding the episodic nature of much memory. Norms, goals, available means, etc. are all generated by and within this general cultural model. None of them has any privileged position from which the whole enterprise might be envisioned as being driven. Norms, for instance, represent one kind of abstraction that people (whether “natives” or social science analysts) can make from the behavioral patterns they experience. People often can find a variety of norms that apply to any given situation, and often find they have to reason with themselves (or with others) concerning which are actually relevant, which best apply, which lead to the right outcomes, which are plausible to relevant others, which are affected how by situational factors, and so forth. Norms by themselves do not usually account for actual behavior (in any direct sense, at least). Any behavioral predictions based simply and directly on any claimed norms will, I predict, always fail. The same applies to goals, perceived means (especially in many social areas), and so forth. More realistic, fuller models of what knowledge guides our behavior – including goals, norms, expectations, etc. and how these fit together – and of what affective force attaches to the various cognitive components are needed to account for our behavior, and it is to this end that the present conceptualization of culture is aimed. 2.2. Behavioral models and analytic regularities Work by colleagues, under the head of “decision theory” (see Chapter 2), has explored the relationship between micro-economic satisficing models (as well as maximization models and various “portfolio management” models) of some economic activity (e.g., fish selling in the traditional West African marketing system) – our third type of model – and has produced models of the flow of fish-sellers’ actual behavioral decisions – our second type. This work has shown that a much clearer picture of the economic considerations driving the fish marketing decisions emerges from an economic consideration of the very highly predictive behavioral model (ca. 90% accurate predictions of real-life marketing decisions) than is reachable through a direct economic analysis of the pattern or marketing decisions. The difference has

Theoretical models and empirical tests

223

to do with the effects on the decision process of information processing constraints of the sort described by Kahneman and Tversky, and by Simon, among others and of culturally standardized mechanisms that people use to get around such constraints. The fish-selling study and analysis was carried out by Christina Gladwin and Hugh Gladwin, and is described in a series of papers (C. Gladwin 1975, 1980; C. Gladwin and H. Gladwin 1971) similar studies have been conducted by them and by others in a variety of areas including, inter alia, fertilizer choice in Puebla (C. Gladwin 1976, 1980), small farming decisions in Florida (Gladwin and Butler 1984), and choice among medical treatment alternatives in a variety of places (Young 1981). A variety of alternative normative economic models were considered, and evaluated against the simplest descriptive model that best captured the actual specific decisions of actual fish-sellers, including the inputs on which the decisions were based, with the most accuracy. The satisficing model emerged as the clear winner in this case for these people operating under their actual real-world economic, social, and computational constraints; that is, the satisficing model best represented the goals of these actual human actors. The suggestion has been made that game-theoretic approaches in social science relate to actual behavior in ways parallel to the ways in which the economic models studied by the previously described students of decision theory. Alternative game theoretic models offer ways of normatively evaluating behavioral outcomes of players under different combinations of player’s own goals and player’s abstract expectations about goals and behavior of other players. But these models do not represent any kind of actual, direct modeling of the actual thought or behavior of the players (whether they be individuals or collectivities). Other kinds of models – equivalent to the Gladwins’ descriptive ethnographic model of fish selling – have to be constructed for this purpose 2.3. Empirical tests I am much interested in empirical tests of proposed theories, but I want to note that direct empirical tests of this book’s culture hypothesis – even if one assumes that it be absolutely true – will be hard to come by. There are too many different kinds of inputs that people deal with and abstractions of patterns that they infer to make any test easy or straightforward. Even with regard to the more unified and straightforward realm of language, such direct tests of any whole, overall view are exceedingly hard. Instead, one generates deductions from one’s theory that differ from the deductions one

224 Chapter 12 – Other theoretical issues and relationships sees being made from other theories, and looks for empirical data that will decide among the given alternatives. But the test always gets complicated by the fact that different theories always speak, to greater or lesser degrees, to different goals and different conceptions of what a particular kind of theory (e.g., a grammar) should specifically do – and thus never are easily comparable. Culture, as proposed here, is even messier than language, and so even harder to empirically evaluate – even were the proposed view to be absolutely accurate. Instead, useful empirical tests are always of only pieces of the overall theory, often only of models based on or derived from parts of it. Ideally, one picks propositions on which the theory is based, preferable relatively basic ones, and finds ways (situations in which) to empirically evaluate these. But such tests of course are always only partial, and their relevance to the overall theory always depends on some non-trivial reasoning process. The relationship between a test and the theory, then, never is an equivalence relationship, but always is an implicational one – concerning what the test implies about the relationship among the evaluated variables and the relationship of that relationship to the relationship among those variables which is asserted by the theory. In terms of the typology set up above, the third kind of model gives one targets to evaluate, while the second gives one mechanisms. The first serves, in a sense, as a general conception from which the actually found versions of the second and third naturally fall out.

3. Some further methodological thoughts The basic methodological trick, of course, is to sort out what is individual (certainly including actually held emotions, actually known facts, actual decisions about what plan to act on and what attitude to express), what is collective (the kit bag of alternative action plans with conditions of and implications of each, basic knowledge structures in terms of which events, acts, etc. are categorized and thought, and so forth), what is tied to the local instantiation of some collective structure (e.g., what about Baltasar del Alcazar’s Ines herself makes “me” love her, what she actually knows about cooking), what is part of some collective action or activity (– a game, an expression of a class attitude, or …) and so forth. One strand of relevant research concentrates on trying to describe actual cultural models within specific domains such as what various of us have begun to do within the domains of ethnobiology, views of love and marriage, and kinship. To what degree, and by dint of what evidence, do such models seem actually to exist in folk conceptions (vs. merely serving as useful ana-

Some further methodological thoughts

225

lytic constructs)? Within individuals (across situations) – and across members of a population – how clearly defined and bounded do these models seem to be; what social, evaluative, instrumental, communicative, coordinative etc. ends do they seem to serve? How are they actually utilized? In exploring these questions one should be looking to develop both a formal characterization of each particular cultural model and a formal characterization of the properties of cultural models in general. A useful aim, then, would be, as findings and events allow, to create computer simulations of such models and of the processes by which they get instantiated in individual behavior (reasoning, interpretation, action, etc.). 4. What seem pressing empirical issues 1) The interaction of sharing and differentiation or distribution. That is, how to relate what is shared (cf. consensus) to what is differentiated among different subgroups or different individuals, how to picture it all as a system. How to tell differentiation based on overarching sharing from differentiation based on independent worlds. 2) The problems of units and forms. a) How empirically to distinguish the content of a cultural model from the content of its instantiation. How to distinguish a general instantiation from a subtype of the basic cultural model. How to tell exactly what is in a cultural model – whether items, relationships, roles, or … b) How to tell well-formed (models or descriptions of) cultural models from badly formed ones. How to tell cultural models that belong to a given culture from ones that are well-formed but still not part of the given culture – or how to tell familiar own-culture ones from familiar alien-culture ones from unfamiliar alien-culture ones. 3) Individual representations of collective knowledge. How cultural model knowledge is held in individuals. How it is learned and stored – and tuned and adapted. 4) How abstract cultural model knowledge gets applied to actual situations. 5. Neural networks This book’s conceptualization of cultural models within a Parallel Distributed Processing framework is not unrelated to the kinds of neural networks models of knowledge systems that are current, but there are some differences.

226 Chapter 12 – Other theoretical issues and relationships I do think that our inference of the patterns that constitute cultural models is from density patterns of what we experience. “Density patterns” covers both the frequency with which we encounter instances and the significance or effect that we associate with the experienced instances. But also involved are recognition and structuration. “Recognition” refers to our ability to recognize two experiences or encounters as instances of the same thing – in a situation in which there always exist innumerable (if often trivial) differences between the any two experiences. Here the Piagetian insight that we can only see a little more than we know to look for becomes important. The things have to similar enough to be recognized and different enough for their similarity to be noteworthy or informative. In the socially constituted world of cultural concepts there will exist significant effects of learning on perceived similarity – and thus on which network connections get strengthened (dare I say “reinforced”). The actual concept formation process seems to be what Bruner, Goodnow, and Austin (1956) described where we leap to possible generalizations (vs. carefully inducing them), try them out, and – if/when they don’t work – amend them and try again. “Structuration” refers to our apparent drive to impute structure to the patterns we experience and to use that imputed structure as the basis for our attempts at productive generalization and as the basis for our long term memory of what we have experienced. The basic structures of relevance to cultural models seem to be narrative structures – story lines, if you will. But the process of abstraction in which a story line comes to be seen as a linking of slots into which alternative fillers can fall implies the existence of something like form classes. Processes of cross-referencing different story lines, of linking story lines to one another, of assessing the logical consequences of actions (cf. Hutchins’ Culture and Inference [1980] and Schank and Abelson’s Scripts, Plans, Goals and Understanding [1977]), and so forth all together provoke the more complex kinds of structure that we see in cultural knowledge systems (e.g., kinship), in our models of the physical and social worlds, and so forth. That is, we learn (abstract out) patterns, but the forms and linkages of these inferred patterns are shaped by various logical, cognitive, communicative, interactive, and productivity constraints, as well as by the nature of the cognized (piece of the) world.

Chapter 13 Illustrative examples: cultural models

1. Cultural models: empirical examples My suggestion is that cultural knowledge is made up of conceptual entities grouped into domains of action. The relevant structural relations among the entities are not the semantic ones of contrast and inclusion (even if that semantic knowledge of “what is subsumed in what” and “what represents an alternative to what” is always pertinent to our thought and conversation), but are prototypic representations of the relations in the pragmatic world which make the conceptual entities themselves meaningful and useful. Thus, in a ranching context, grass relates to cattle as potential food (and secondarily, perhaps, as being dependent on nutrients, some of which can be provided through cow by-products). There exist other plants which cattle will not eat but which sheep or goats will happily consume. I am offering cultural models as the collective cultural conceptual units via which various kinds of cultural knowledge – including classifications, knowledge of processes and skills, goals, values, identities, etc. – are linked to action – and thus as the units via which conceptual entities are linked together with action into systems that make up “culture” as it is traditionally understood. This knowledge would be what individuals rely on in deciding what to do and how to do it. Individuals do not automatically implement (“instantiate” and then “realize”) any particular cultural model, but instead exercise “agency” both in selecting which out of an array of potentially applicable models to chose to apply and then in deciding how actually to implement it. Empirical questions arise concerning how do we actually know that cultural models exist, how do we know their properties, how do we know their content, and so forth. I cannot yet will not answer these questions, but I do want to describe various pieces of research that strongly suggest that something like this book’s cultural models really exists and that begin to specify some of their content.

228 Chapter 13 – Illustrative examples: cultural models 2. Image studies One piece of research involved asking students in a University of California, Riverside class to provide me with a “sketch of a high school” (using an abstract version of what Kevin Lynch did in his studies for The Image of the City [1960] – see especially his Appendix B). The students were a mix of undergraduate and graduate students and came from a variety of regions (though mostly in southern California). Following Lynch, I analyzed the data by putting together a composite drawing in which the criterion for an element’s inclusion was that a substantial number of the respondents included the element in their drawings, and where the darkness/heaviness of the relevant parts of the drawing represented the number of respondents mentioning the element in question. Adjacencies (what went next to what) in the drawing were similarly determined and their “weight” shown by a fraction in which the numerator was the number of students placing the two next to each other and the denominator was the number who included both elements somewhere in their sketch. The labels on elements were determined in a similar way. The resulting composite drawing is shown in Figure 1. The total number of respondents was small (n =8), but the results still seem sufficiently pronounced so as to be significant. That is, even a small sample produced an informative and meaningful picture – even if a larger sample would have fleshed it out more. The analytic method is moderately “objective” and automatic. The amount of agreement that came out of the task, and the meaningfulness of the resulting picture (given the almost total lack of structure built into the experimental task – i.e., no constraint was placed on the content beyond the need for it to be something that could be drawn) are the evidence that I want to adduce in favor of some sort of shared structure. The respondents were people who had not discussed their high school experiences with each other, and they all went to different high schools in communities that varied significantly in size and degree of urbanization. To be sure there were considerable commonalities across their various high schools (southern California does have a fairly characteristic kind of dispersed high school campus), but that does not explain the particular commonalities in the picture. There exist a large number of similarities across the high schools that did not appear in the collective picture (nor, often, in any of the individual sketches). That is, the respondents were quite selective in a way that cannot be explained simply by referring to similar, parallel individual experiences. There had to be some implicit consensus about the salient elements of a California (or American) high school. But the drawing task was too unusual for this

Image studies

229

“High School”

N.B.:

See article text for presentation conventions.

No library (only 1 mention) No bathrooms (1) No lounges (1) No trees (1) No study hall (0)

Figure 1. Image of southern California high school – Students from anthropology class University of California at Riverside / Mixed undergraduate and graduate students from southern California, early 1980s (N= 8)

230 Chapter 13 – Illustrative examples: cultural models consensus to have emerged directly in terms of anything specific to such a visual representation. That is, there had to exist some more general consensual construct on which the responses to the drawing task were based (“construct” because it did not/could not represent any simple function of direct experience). Such a consensus presumably came out of an interdigitated mix of direct experience of high school, interaction with others based on similar high school properties, and conversation – where the interactive and the conversational components forced some convergence of the individual understandings on which they were based, but where the convergence cannot have been on isolated elements but instead had to be on a representation from which elements could be estimated or computed. Such a representation would seem to have many of the features that I have described as characterizing cultural models. Some sense of how people home in on specific reference communities can be seen in a kind of conversation I have heard (and heard about) involving a new member of a group. The conversation takes off from a question such as “How do you guys X” or “How do folks around here X?”, where X might be something like “decide who’s on which team” or “decide who picks the topic for the next week’s discussion?” or “decide who gets stuck cleaning up after the reception?”, etc. That is, they speak in terms of a group (it can easily be an ad hoc one) and they presume some group understanding that pertains to something that the group is relevant to, and they presume some pattern specific to the group (rather than referring, for instance, to the separate individual dispositions or experiences of the group’s members). The high school sketches cannot themselves have been the cultural models, but rather these have to be specified, narrowed, and adapted mappings that were derived from the cultural models. Note that the form (i.e., a drawing) in which the knowledge is expressed is forced by the research design, and so it does not provide any direct evidence concerning the form or format of the underlying individual or cultural representations. Figure 2 provides some of the evidence for treating the sketch products as a real empirical finding vs. some kind of tautological exercise. Figure 2 shows the consensus pictures that came out of the same task when it was given to a class of Master of Arts students in Merida, Yucatan in Mexico, where the stimulus was “escuela secondaria”. The picture is clearly different from the California one, and different in ways that make sense – but ways that I, at least, would not have predicted ahead of the fact. The comparison of the Merida picture with the Riverside one suggests two types of observation. First, we have content differences that presumably reflect factual differences

Image studies

231

Number of Students listing the given feature (N=8) “Escuela” “Secondaria” “Federal” “No 1”

3 4 3 2

show building soccer field city block fence

8 5 4 4

people shown – on balcony – in classroom – in soccer field

4 2 2 1

regarding the building: – balcony 6 – traditional windows 6 – single building 6

Figure 2. Image of “Escuela Secondaria,” Mexico – Students at INIREB M.A. Program in Merida, Yucatan in 1980s / Students from all over Mexico

in secondary schools and how they are understood in the two places. Second, we have the difference in levels of agreement (or density of the resulting pictures), which – given the comparable number of respondents (again, n=8) – suggests more homogeneity in the Riverside group than in the Merida one. While I cannot prove this claim statistically, I suggest that there is a real difference in homogeneity which reflects not differences in U.S. vs. Mexican culture but, instead, reflects the fact that the Riverside group was drawn regionally from southern California while the Merida group was drawn nationally from all over Mexico. Figure 3 is the image produced by a similar (“quickly sketch a …”) study of “restaurants”. Figure 3 is particularly interesting for its illustration of what happens when the population has two different images – or, at least, perspectives – one a floor-plan and one an external view. The two views came out of a single set of responses from a single sample of 20 students. It is unclear whether or not the two views represent different subsets of the students – or just alternative views held by all of the students.

232 Chapter 13 – Illustrative examples: cultural models

Number of Students listing the given feature (N=20) ENTRANCE shown Location – in corner – on long side – on short side

9 3 3 3

KITCHEN shown Location – as indicated

9 5

TABLES shown – Square tables – Round tables

8 6 3

RECEPTION AREA shown – Near entrance – Around corner – On side

7 7 half half

COUNTER shown Location – by entrance – by kitchen – by both

5 1 2 3

CASHIER shown Location – on counter – Separate – by entrance

5 5 5 5

FLOOR PLAN 9 Overall shape – Clean rectangle 5 – Rectangle with projections or indentations 3 – Irregular shape 1

ROOF shown Shape – pitched as shown – pitched with bevel ends – pitched with flat top – flat

10 5 2 2 2

CHIMNEY shown

10

DOORS shown Location – on end – on side – (both locations – (loc. without actual door

10 7 5 1) 1)

WINDOWS shown Location – by door – Other wall Type – separate – forming wall

10 7 6 7 2

SIGN WITH NAME shown Location – over door – separate sign from bldg. – on roof

8 5 2 2

TABLES shown in windows

4

SHRUBBERY shown

4

PERSPECTIVE PLAN breakdown – show 2 sides – show 1 side – (did both – (gave interior views

11 6 5 1) 3)

Figure 3. Restaurant Images – Students in mixed undergraduate and graduate anthropology class at the University of California at Riverside, Winter 1988

Image studies

233

See article text for presentation conventions.

Figure 4. Image of “Anthropology” – Students from cognitive anthropology class University of California at Riverside / Mixed undergraduate and graduate students from southern California, Winter 1981 (N=8)

For further illustration of the method – drawn from a comparable Riverside class in cognitive anthropology – Figure 4 provides an equivalent “verbal” “picture” (“list 4 or 5 words that come to mind when you hear…”) of “anthropology”. Adjacencies are based on the number of times the given pair of words appeared next to each other on the same answer sheet. Parts of anthropology are central, with neighboring disciplines around the periphery. These findings relate to a very different ethnographic task that Volney Stefflre worked out several decades ago (see Stefflre 1972). He was a psychologist/anthropologist who pioneered an approach to marketing that has since become a major component in business schools’ curricula. He would ask around a hundred or so people a series of questions about actual and ideal products in some category (light whiskeys, instant coffee, politicians, etc.). When a new product was constructed on the basis of his findings (typically a ca. 5 year process), he turned out to have successfully predicted (from those initial responses) both the subsequent market percentage for the new product and how much of the market it took from each already existing product. Stefflre had respondents make statements about each of the products in the set and then he pulled out a subset of questions that seemed most common across respondents yet were diverse and contained some kind of content. By “contained content” we mean a statement that was more informative than simply “X is good”; e.g., “X is rich and full bodied”. The content did not have to be objective. He would then use the statements from the subset

234 Chapter 13 – Illustrative examples: cultural models list as questions (e.g., “Is Y rich and full bodied?”) and ask these questions about all the other products in the set. Each pair of products was given a similarity score based on the number of questions for which the products received the same answer. The set of similarity scores were run through a multidimensional scaling program to produce a spatial model of the given product domain in which the juxtaposition of products in the space represented their similarity to one another. Then, based on the same questions, each person’s ideal version of the product was located in this space. This produced something like a density function of demand relative to the products. At this point one could deduce from the geometric configuration those people who would switch to a new product located at any particular point in the space. The challenge was then to develop a product that elicited the desired evaluations. The process took a while because many of the qualities were not immediately tangible or objective. For example, in one case (as a part of a much larger task) he had to figure out exactly what shade of green betokened a “happy” pickle! Stefflre’s innovation was to ask people about the similarities among products rather than surveying them about their specific preferences (even though these did eventually figure in). What made Stefflre’s procedure work was the fact that people had a shared understanding of what mattered about the various kinds of products, and thus about the relative similarities and differences among products in a certain domain, even though their individual preferences were quite variable. The stability of answers to questions such as the shade of green for a “happy pickle” meant that respondents were answering based on shared cultural understandings. Their individual experiences alone could not have led to their similar responses. In one presentation of his work he commented that there was “no question so idiotic” that you would fail to get a stable set of responses to it from a hundred people in a given culture. By this he meant that the distribution for the next 100 persons from the same population would be pretty much the same as for the first 100. This meant that his initial assessments could be (and indeed, were) run on very small samples. And, since his respondents never (or at least only rarely) had any occasion to have with one another anything like their conversations with Stefflre, they could not have rehearsed or trained themselves on the answers. They were constructing their answers from general knowledge. Indirect evidence, including the specificity of their answers and the sensitivity of the work to context, suggests that they were not answering from some broad set of general knowledge, but rather they had relatively domain-specific understandings. His work does not force acceptance of cultural models, but it is consistent with the idea of cultural models.

Image studies

235

My high school and restaurant examples show uniformity in part because the analytic method homes in on uniformities within a given population. Differences that could represent, a division of labor (or regional or class sub-cultural differences) can be found by examining subpopulations of a population to see if contrasting subpopulations show contrasting uniformities. With the high school and restaurant studies and within the context of the sample population (students in a UC Riverside class), there was no systematic basis for any subdivision. The students’ experiences (both direct and socially mediated indirect) were homogeneous. While there was notable individual variation, there were no contrasting subgroups whose relevant experiences differed substantially. The discussion of the differences between the Riverside and Merida school pictures did suggest that the different density of detail probably resulted from differences in the homogeneity of the two sample populations. Implicit in that suggestion was the claim that a regional Mexican sample (e.g., students drawn from the state of Yucatan) would have produced more shared elements and that, on the other hand, a sample drawn from across the US (still of comparable size) would have shown fewer shared elements. And, as suggested before, the differences in pictures between those produced by a national and a regional sample would result from a combination of factual variability in the objects and situations experienced and communicative variability in how the situations were experienced and talked about. To further illustrate subpopulation differences we now turn to a different topic, namely the make-up of the University. For this study I did not ask for drawings (or physical maps) but rather for verbal information. I collected several kinds of data. One consisted of asking for terms that the stimulus (e.g., “university” or “UC Riverside”) brought to mind. Another consisted of more focused questions, as “What kinds of people are at UCR?” or “What kinds of students are at UCR?” A third consisted of asking questions such as “What is a sophomore?”, “What is a TA?”, etc. Given my primarily pedagogical goals in the classroom, much of the data collection was informal (students shouted out answers that they had previously written down, with a show of hands for who agreed with what, etc.), so I have neither pictures nor numbers for these experiments, but I do want to discuss what I found. My results are both simple and unsurprising. But first, we need some definitions. The set of primary roles in a university represents a sequence or progression: undergraduate to graduate to faculty. The undergraduate role may be divided into: freshman, sophomore, junior, and senior. The graduate role may be divided into something like: first year, second year, finished

236 Chapter 13 – Illustrative examples: cultural models with course work, past candidacy, and ABD (all but degree). Faculty roles may be divided into the professorial series: Assistant Professor, Associate Professor and Professor. Teaching Assistants are an advanced student apprentice faculty blend. Lecturers are outside the normal progression and instructors used to be part of the progression, but are no longer. The basic, generic cover term for those who teach is “professor” (used for all ranks and statuses), but “instructor” and “teacher” can also be used. The findings were that students knew the most about the position they were in – both its subdivisions and its place in the sequence – and the ones they had recently left. Knowledge further back was there, but could become a bit blurry. Knowledge of the immediate future – the normal next step or two – was fairly well known, while steps further ahead were – in the collective case – a blur. That is, freshmen and sophomores knew about juniors and seniors, but often had only a vague sense of some amorphous category beyond that which included graduate students and faculty; juniors and seniors were aware that graduate students were a kind of student (not faculty), but were not very clear about TAs and did not distinguish among different kinds of faculty; graduate students knew about lecturers and the like but were not very clear on the professorial ranks. This pattern is similar to what Hutchins [1995] found regarding the knowledge of the component roles involved in navigating a large naval ship. The point is that there exists a basic framework of shared knowledge that, taxonomically speaking, is at a fairly high level and that more specific knowledge represents expansion of selected nodes of that general knowledge. In the university case I just described the expansion is seriatim and cumulative. Questions such as “What kinds of students are there in the university?”, in the context described above, also produce another kind of answer (besides levels in the progression), namely a list of various “majors”. These are alternative areas (mostly academic disciplines) of primary concentration such as anthropology, economics, biology, English, history, dance, and so forth. These, unlike the levels, are non-cumulative, but as with the levels students know more about the subdivisions of the major they are in than the ones they are not in, and they know more about “neighboring” ones than more distant ones. And this kind of specialization can be repeated for special concentrations within a major. Again, the common, fairly (taxonomically) high level sharing allows communication across the specialties with their attendant individual expertise. This kind of combination of shared general with allocated specific knowledge seems to be what is at issue with folk taxonomies in the first place as well as with our scientific, Linneaean spinoff from European folk taxonomies.

An extended example

237

Folk taxonomies are, of course, chunks of language and thus perhaps not tightly relevant to non-linguistic culture. But the knowledge of parts of the university that we have been considering – even if involving named chunks – involves extensive, non-linguistic knowledge 3.

An extended example: ranches, rangeland, and environmental cultural models

For our final, extended, example, let us turn to some particular aspects of ranching in California as an extended example of how we see cultural models as operating and of how we see them as functioning. Both the ethnographic and survey research were carried out by Dr. Kimberly Hedrick, with some advice and input from me (and from her adviser, Dr. Eugene Anderson). She is the principle author of this discussion.

3.1. Introduction In conducting fieldwork with ranchers, it rapidly becomes apparent that they have a complex of models via which they understand their environment and make decisions concerning ranch management. Ranchers must frequently work with other groups of people in the management of their grazing lands, particularly those that are public lands (under the jurisdiction of the Bureau of Land Management, Forest Service, or National Grasslands). As ranchers repeatedly described their problems of miscommunication with non-ranchers and the false assumptions non-ranchers had about the local environment, it became imperative for us to find a way to study these models of the environment and compare those held by ranchers with those of non-ranchers. Participant observation and depth interviewing were (and are) very useful for producing detailed information about ranchers’ models of the environment, but it remained necessary to find an analytic device for bringing this detail together into a coherent, interpretable picture that captured the uniformities in rancher responses while making sense of the variability that was found. Since the ranchers seemed to understand each other well enough, in spite of variability in the specific situation of each, some sort of dynamic and flexible cultural model seemed implied. The problem of understanding and then modeling the miscommunication of ranchers with members of other groups (including the general – non-expert – public as well as other “expert” groups such as academic

238 Chapter 13 – Illustrative examples: cultural models ecologists and government agents and regulators) suggested the need for similar models of the understandings of members of these other groups. But, as extensive participant observation among multiple groups was impractical, a more expedient way of understanding and describing their cultural models was needed. In an effort to investigate, in particular, the general public’s cultural models of environmental categories – in order to understand the conflict between ranchers and the public, as well as to better understand the form and content of cultural models themselves – we have been exploring the administration of variants of the kind of loose, open-ended free-listing survey task described earlier (for schools and disciplines) to samples of nonranchers. Two different surveys were administered, each among non-rancher university students at UC Riverside. Students were of various majors and ages. The surveys are described following the results.

3.2. Empirical testing of cultural models We began with a survey that was based on verbal labels (words !) that were frequently discussed among ranchers and non-ranchers and that seemed to cause conflict. These labels included environmental categories such as “wetland” and “forest,” land use categories such as “urban” and “ranch,” and certain overarching problematic terms such as “nature” and “wilderness.” This initial survey was conducted with 71 participants drawn from an upper division cross-listed anthropology/business course at UC Riverside; 27 of these were selected in a stratified random sample based on age, gender, and major/educational background. Their responses were entered into a database by term, and were classified into broader categories in order to understand the broader salient features that defined people’s models (for example, “dry” would be classified as “climate-precipitation”). Such a database is a useful heuristic device, as it allows the researcher an efficient way of processing large quantities of data that may then be manipulated, stored, and statistically analyzed with ease. Based on the results of the first survey, it was evident that non-ranchers used primarily visual cues to define their models. It was hypothesized that ecological zone cultural models were in the form of relatively static pictures among non-ranchers. This was then tested by a second survey in which 42 non-ranchers were shown pictures of various ecological zones and asked to write down what type of environment they were viewing and how they knew that it was that type. They were also instructed to indicate if the picture showed a healthy environment or not, and why. The congruence of re-

An extended example

239

sults between the first and second surveys, combined with the new data on non-ranchers’ models of environmental health, showed that it was a reasonable hypothesis that non-ranchers’ models of ecological zones, as well as environmental health, are primarily passive cultural models that take the form of static ideal pictures. Terms that produced confusion (such as “rangeland”) in the first survey matched pictures in the second survey that produced similar confusion as to their appropriate “label.” Terms that produced well-defined cultural models in the first survey (such as “desert” and “forest”), likewise produced clearly defined prototype-extension models in the second survey. 3.3. What is a ranch? Not surprisingly, ranchers’ cultural models of their environments are more complex and deep than those of non-ranchers, and there are some significantly different understandings of certain common labels such as “wilderness” and even “ranch.” First of all, there is the problem of what farms and ranches are. Ranchers have a clearer model of what a ranch is than the general public, but one that is also broader. Ranchers’ model of a “ranch” is an enterprise that uses land extensively to produce animals and/or hay. The prototypic ranch among ranchers may be cattle or sheep, depending on location, but the term “ranch” extends to include the production of ostriches, game, goats, alpacas and llamas, horses, and hay. It typically does not extend to include any crops other than hay. The public is aware there are different types of ranches, but our explorations indicated they do not have any consistent model of a ranch that sufficiently distinguishes it from a farm. They have a general idea that ranches deal more with animals, but they cannot define how this is really different from a farm (see chart below). The public thus often thinks that crop-growing farms are also ranches: “apple ranch,” “peach ranch,” and so forth. The first environmental cultural model survey did show that the public has a prototypic understanding of ranch that has it in livestock production, specifically cattle and/or horses. The fact that respondents only used comparative responses when responding to the term “ranch” indicates that for the public, a ranch is a type of farm. “Farm”, in its prototypic term, seems to denote something like the old “family farm” (which, by now of course, has become practically nonexistent), as indicated by the relatively equal weight given to vegetative and livestock production, as well as the broader spectrum of domestic fauna listed. “Ranch” seems to be, for the general public, a specific type of “farm,”

240 Chapter 13 – Illustrative examples: cultural models Chart: Survey #1 (Responses Generated by Verbal Label Cue) N=42. Population is Non-ranchers. Number of responses in brackets. RANCH

FARM

Fauna, all (42) Specific Fauna (38) Cow (17) Horse (12)

Purpose (39) Crop production (14) General food production (12) Livestock production (11)

Purpose (20) Livestock production (13) Crop production (4) Comparative to Farm (10) Ranch is a type of farm (3) Ranch is the same as farm (3) Ranch is a large farm (2)

Fauna, all (26) Specific Fauna (19) Cow (8) Horse (4) Pig (2) Sheep (2) Chickens (2)

but how it exactly compares is a matter of disagreement. Perhaps it is a large farm, or a type of farm where they predominantly raise animals. It is interesting to note that, however vague their definitions, the public certainly has a prototypic type of ranch – as a cattle ranch or alternatively a horse ranch. Other types of ranches (ostrich, game, bison, etc.) may or may not be under the umbrella of “ranch” in the public’s classification system. It is worth noting that horse breeding places can be referred to by professionals as either ranches or farms. The designation is not about extensive/ intensive land use as much as it is a matter of riding discipline and breed. English disciplines and breeds (Thoroughbreds, Warmbloods, Andalusians, etc.) typically are associated with horse farms. Western disciplines and breeds (Quarter horses, Paints, Appaloosas, etc.) are typically associated with horse ranches. Both horse farms and ranches may be extensive (as in the Thoroughbred farms of Kentucky or the Quarter Horse ranches in the West) or intensive (as with many horse breeders in suburban locations). This model of ranch as a type of farm can be contrasted with ranchers’ views, which hold ranches and farms to be quite separate types of organizations – because of very different economic needs and land uses. Ranchers and farmers, for much of history, saw themselves as antagonistic groups, each fighting to maintain a very different use of the land (particularly during the open range period). While family farmers and ranchers now cooperate due to a dramatically dwindling population and power, they still hold these two categories as quite separate.

An extended example

241

3.4. Environmental categories Secondly, there is the problem of models of various environmental categories, such as “desert” or “grassland.” Ranchers tend to have very detailed, locally oriented understandings of environmental labels. A “desert,” for example, would be modeled after local environments for a rancher, if s/he lived in a desert. Through association with other ranchers throughout the United States and even internationally, ranchers are also keenly aware of ecological differences between different places (especially in different regions, but also even in different local micro-regions) and the necessity of experience with the local environment if one is to know how to manage a given area. Thus, though there may be overarching categories such as “desert,” these are typically informed by local environmental attributes and ranchers resist generalizing to other locations without sufficient data. Their models of various ecosystems tend to be locally based and specific.

Figure 5. Sahara Desert

242 Chapter 13 – Illustrative examples: cultural models Non-ranching non-experts, as indicated by the surveys, respond to certain ecosystem labels with what seem to be widely held generic cultural models but to other labels with much more nebulously varied results. Models tend to be superficial and are sometimes related more to what one would see in National Geographic or on the Discovery Channel than local environments, even when the label is applicable to local environments. “Desert” produced one of the strongest cultural models. Participants described “desert” overwhelmingly in terms of climate, particularly as dry and hot. Other salient features were a lack of vegetation and a prevalence of sand. When confronted with pictures of deserts, the prototypic desert was the Sahara (figure 5), which 97% of respondents labeled as a desert and 78% picked as the best example of a desert. The reasons given included the prevalence of sand (88%), lack of vegetation (27%), and dryness (24%). A picture of a local (Eastern California) desert (figure 6) was a distant second, chosen as the best example of a desert by only 17% of respondents.

Figure 6. California Desert (without cattle showing)

The most interesting finding concerning non-ranchers’ models of deserts was that they were predominantly considered barren and were incongruent with the model of a ranch. Respondents had, among their choices of possible prototypic “deserts,” a picture of an Eastern California desert near the White

An extended example

243

mountains without cattle (figure 6) – and another one with cattle (figure 7). The plants and soil were nearly identical. While 17% of respondents picked the picture without cattle as the prototypic desert, none picked the one with cattle.

Figure 7. California Desert with Cattle (Rangeland)

The picture of the Eastern California desert with cattle (figure 7) was labeled by respondents primarily as a ranch (17%) or farm (14%), or as prairie or plains (26%), and only as a desert by 14% of the participants. Interestingly, the primary reasons given for labeling the Eastern California desert picture without cows (figure 6) as a desert was due to its dryness and its type of plants. The reason for labeling the picture that showed the same ecosystem with cows as “prairie” or “plains” was also claimed to be due to the type of plants. Essentially, the presence of cows in the picture altered people’s perceptions of the surrounding environment, as they could not hold in their minds the models of “ranch” or “farm” and “desert” simultaneously. This is cause for conflict between ranchers and non-ranchers, as much of the Southwestern desert country includes cattle or sheep ranches. Since ranchers know that grazing is prevalent throughout the West, in deserts and forests as well as grasslands, a ranch may be in a desert. Non-ranchers, having little real experience with the broad spectrum that is ranching, indicated in surveys that ranches were an extension of farms, and were associated primarily with grasslands and green vegetative production. Confronted with a picture that

244 Chapter 13 – Illustrative examples: cultural models obviously contained salient elements of both models (cows and dry shrublike vegetation), many respondents shifted their perception of the vegetation to accommodate their model of ranch, since it seemed extremely difficult for people to ignore the cows. 3.5. A use category Finally, there is the problem of the term “rangeland.” Ranchers use this term, and it shows up in popular literature and ecological debates, but it seems to cause a great deal of conflict. Ranchers understand this to be any area where ranchers’ animals graze. It includes multiple ecological zones (mountain meadows, shrublands, deserts, grasslands, forests) and various types of forage. The “range” is the unpopulated areas of the West that are in livestock production. It is where ranches are located. Non-ranchers’ surveys indicated a significant amount of confusion about what “rangeland” meant. Responses overall were much lower than other land use labels (such as “farm” or “urban”), and the most commonly held response was confusion (indicated by responses such as “I don’t know” or “?”). Rangeland was associated somewhat with animals, but not with any particular animal. When respondents indicated a purpose for rangeland, it was frequently livestock production, but few respondents indicated any purpose. It is notable that the only ecological zone that was mentioned in response to the prompt “rangeland” was grassland. When confronted with pictures of various ecological zones and asked to choose the best example of “rangeland,” most respondents picked the only picture with cattle in it, though it was not a grassland. The distant second was a grassland. Other ecological zones were not chosen in sufficient numbers to even indicate an extension of “rangeland” to include them. These included riparian and forested areas. The lack of overlap of “rangeland” and therefore “ranch” with “forest” and “river” bear out what is frequently discussed by ranchers: that non-ranchers do not understand the diversity of ecological zones that are held by ranches. Nonranchers, if they do not hold “forest” and “ranch” as overlapping models, may think that ranches that utilize Forest Service land, for example, are anomalies and should not be given such permits.

An extended example

245

3.6. Environmental health and conflict The most frequently discussed conflict between ranchers and non-ranchers is evaluating rangeland condition. Non-ranchers, both those relatively uneducated in ecology and professional environmental scientists, have too frequently designated well maintained ranch lands as being in poor health and thus well-run sustainable ranches as destructive to the natural environment. While there are certainly a plethora of environmental issues facing grazed lands, particularly coming out of the heritage of the open range period, some of the conflict between ranchers and non-ranchers is due to miscommunication. Ranchers have models of environmental health that are sometimes quite different from those of non-ranchers, particularly those who are not extensively educated in rangeland ecology. In response to pictures of various ecosystems, non-ranchers indicated whether each was healthy or not, and why. The following chart illustrates the models of environmental health that non-ranchers hold. Chart: Survey #2 (Responses Generated by Picture Cue) N=42. Non-ranchers. In order from most prevalent response to least prevalent. HEALTHY

NOT HEALTHY

Presence of plants (all kinds) Green Presence of water Presence of animals Presence of trees Clean air Presence of people

Dirty water Lack of water/dry Not green Lack of vegetation Lack of people

Thus, non-ranchers’ models of environmental health are primarily based upon very superficial and crude attributes: plants, a green color, and water. Pictures of environments that did not contain these characteristics were considered “unhealthy” by the majority of the respondents, even if the ecological zone was such that it naturally did not have high enough precipitation to produce an abundance of vegetation. While ecologists and other trained environmental scientists certainly have a much more sophisticated and complex system for evaluating environmental health, it is worth noting that they may still be influenced by the same underlying cultural models.

246 Chapter 13 – Illustrative examples: cultural models Ethnographic data show that ranchers in severely arid areas are often accused by government ecologists and biologists of overgrazing on lands that they do not use at all! Some areas, due to extreme aridity, do not grow grass or lack plant litter between shrubs, both of which in less arid areas are indications of overgrazing. In this case, the conflict between ranchers and environmental scientists is primarily a problem of locally based cultural models versus more generalized ones. In the case of ranchers’ conflicts with the non-ranching general public, it is primarily due to non-ranchers’ visions of a healthy environment as being lush and green, with abundant water resources, even though much of the West is arid and unable to support such growth. Ranchers, having direct and constant interactions with the environment, learn their knowledge experientially, producing a range landscape that is interactive (that is, humans and the environment are part of one system), practically oriented, and dynamic. Urban/suburban environmentalists and other folks, on the other hand, learn about the environment primarily through formal education and experience “Nature” as something that is a separate category from human landscapes. They escape from their humancreated landscape, the city, by entering the Wilderness, a landscape that is relatively untouched by humankind. Their range landscape is thus radically simplified (from the reality represented by an interface of humans and nature – marked by a careful utilization of natural resources), as well as aesthetic and static. Ranchers’ cultural models for range evaluation have at least three components: natural resources available, temporal knowledge, and knowledge of the land’s social history. When a rancher evaluates rangeland as good, bad, or somewhere in between, s/he is referring to the combined analysis of these three factors. 1. Natural resources available: this is the natural suitability of the land for raising livestock, including rainfall and other water resources, plant communities, topography, soils, and ease of managing livestock at that location (access to land by roads/trails, access of land by the public, etc.) 2. Temporal knowledge: this is knowledge of the history of environmental events and their current manifestations on the land. The current state of the land is evaluated in light of recent natural or human-induced environmental change, positive and negative. 3. Social history: knowledge of the land’s ownership and management. This aids the rancher in evaluating the potential of the land for further use. An example: as we were riding through mountain meadows gathering cattle, the rancher I (KH) was with pointed out a large meadow that was not part

An extended example

247

of his allotment, but was adjacent. The grass was cropped quite short, and bare spots of ground a couple feet in diameter could be seen here and there throughout the meadow. The small stream that ran through the center of the meadow was dry; the banks were eroded quite a bit and trampled so that rather than a stream, the area was deep mud with grass growing over some of it. The water gates along the stream that could regulate water flow were in disrepair. He commented that it was a “shame” that such a “naturally good meadow” was in such “condition.” He described the owner, a “hobby rancher” whose primary career was medicine, and said he just did not know or care enough to manage his herd properly. The meadow was declared to be in a bad state, but was still considered good range if it had better management. The natural resources available, very similar to his own meadows, made the meadow good range, IF it was managed well. It had abundant water sources, good native grasses, and although it would necessitate a 20-mile cattle drive in spring and fall on horseback, it was an ideal summer range: wet, cool, and productive. In its current condition, it was evaluated to be in a bad state, but had the potential to improve under better management. Therefore, it was not “good range,” but even in its current state it was not “bad” either. The abundant natural resources were enough to mitigate the poor management, leaving the land desirable but needing a lot of work. The rancher discussed the time depth, explaining the relatively recent mismanagement and the strong potential for restoration under proper management. The land was thus evaluated with a full knowledge of its recent social history: who leased the allotment, how they (mis)managed it, how many cattle were run on it each year. This social history allowed the rancher to evaluate the land’s future potential. It is interesting to note that the same land would be evaluated quite differently by an many environmentalists. Without the dynamic components of time depth and social history, the land would simply be declared unhealthy. Some would probably seek to put the allotment into non-use. This can be contrasted with the ranchers’ evaluation, and the knowledge of management strategies that could restore the land while keeping it productive.

3.7. Forms of cultural models and action We are ultimately concerned in this example with dynamic (perhaps scenario-like) cultural models of action or behavior, but such models depend on more static picture-like models of what the results of one or another kind

248 Chapter 13 – Illustrative examples: cultural models of action might be. For various reasons many of our examples are of the static sort, but in this ranching discussion at least these are always linked to actions. The static models, especially, are often fairly tightly tied to the classification systems foregrounded in earlier cognitive anthropology; the difference here is that we are focused on the combinations of cues, reference structures (including pertinent classification systems), experience, stake, etc. that drive people’s actual assignment of situations to categories. From this preliminary research it seems that our informants hold ranch relevant knowledge in at least three forms: prototype-extension, classification by distinctive feature, and place on a continuum. Ecological zones, for example, typically have prototype-extension forms among non-ranchers, and, to a lesser extent, among ranchers as well. Non-ranchers, for example, have a prototypic “forest” that is hallmarked by tall conifer trees and lush, green vegetation. Extensions include deciduous forests, tropical rainforests, and dry forests with chaparral (such as those in Southern California). Other cultural categories seem to be defined not by category prototypes but by some salient distinctive feature. These categories may, in turn, be made up of more specific categories that do have prototypic exemplars. These distinctive feature categories may well be analytic or esoteric ones that do not come up in normal discussion or action – and hence lack both good folk labels and good, consistent experiential exemplars. Non-ranchers’ responses to the pictures of riparian areas indicated that there is a general overarching recognition of a category of “lands next to significant water sources,” but without any particular label coming to mind. The labels that were given to relevant instances in our stimulus pictures were all more specific – such as “river,” “stream,” or even “marsh” or “swamp”. The general public does not seem to have strong enough or clear enough experience of the general case to produce any consistent or developed prototype, but people are consistent about a classification system that makes water sources a salient feature that unifies a variety of different places or situations that have water under one overarching category. Finally, there are continua. While non-ranchers seem to have a prototype-extension model of healthy environments, ranchers have a continuum model of environmental health that is based more on sustainability and the combination of natural resources, management history, and potential that either make for or signal sustainability than on any simple classificatory features (such as a green color). Thus, while lands may be evaluated as “either/or” by non-ranchers (either healthy or not), they are evaluated in a much more complex and detailed way by ranchers and placed on a continuum of sustainability and suitability for ranching. (What prototypes

An extended example

249

ranchers do have in this regard seem much more locally and ecologically specific.) This is due in part, it seems, to a combination of two basic types of models that may both be used in making decisions to determine courses of action. Passive cultural models, such as non-ranchers’ model of environmental health, are abstract ideals. In this case, non-ranchers have static pictures in their minds of what a healthy environment is. Active cultural models, such as ranchers’ models of environmental health, are dynamic models of processes and thus are more flexible and situational. They are models of human-directed and natural processes that inform one’s understanding of the data given by the situation. Thus, while non-ranchers may look at rangeland that is desert and judge it as “unhealthy” based on an ideal standard of green vegetation and abundance of water, ranchers will be informed by models of rangeland management, local ecosystemic processes, grazing behaviors of various types of livestock, and so forth, in order to arrive at an evaluation that is not “either/or” but rather is a judgment of potential based on management.

Chapter 14 Gregory Bateson: pulling it all together

1. Bateson’s Naven My original excitement with Gregory Bateson’s Naven was that Bateson had been able to bridge two dichotomies that others seemed either to find overly daunting – or to not find at all: cognitive content vs. affect and individual behavior vs. collective patterns. Bateson deals with both sides of a number of traditionally important oppositions – such as affect vs. cognition (as ethos vs. eidos), behavior patterns vs. structure (as ethos and eidos vs. social structure, etc.), the individual vs. the group, individual behavior vs. cultural norms (including individual variability re goals and actions vs. cultural norms, and personality vs. culture), culture vs. society (as ethos and eidos vs. social structure), and continuity vs. change (as schizmogenesis) – and effectively interrelates the two sides. He does so by way of showing how the various analytic lenses and angles of vision developed in anthropology and related disciplines can be used together to provide a reasonably holistic understanding of a cultural event. His example contrasts with the normal practice of reifying one or the other of these into “the” explanation. As a model for such a careful, integrated, empirically careful analysis, Naven is still unsurpassed – and not even close to being matched.80 Also important, though less salient in the present post-cybernetic world, was the idea of schismogenesis. What stands out today about schismogenesis is not the idea of feedback per se, which we are well used to, but his exploration of how the feedback process could exist not only between indi80

Naven is surprisingly modern (i.e., “advanced”) in some other regards as well. First, Bateson does a better job of telling the reader exactly how he knows each piece of information he presents (direct observation [of how many examples], direct informant report of examples, informant statements about general or normative practice, inference, or etc.) than do most modern ethnographers. This allows the reader some basis for an independent evaluation of how believable or how convincing are Bateson’s generalizations or explanations. And, second, Bateson is very good about presenting women’s as well as men’s perspectives on events, values, goals, etc. That is, he is the least sexist ethnographer (man or woman) that I can easily think of.

Bateson’s system

251

viduals but also between more abstract socio-cognitive constructions such as roles. And, more particularly, what is important is his taking those role relations not as reified abstractions but as patterns in the behavior of actual individuals, and considering how the individuals’ learning and goals were fit into the larger cultural pattern. For our current purposes, then, schismogenesis’s importance refers back to the relationship between the individual and culture and society.81 2. Bateson’s system The question of the relationship between the individual and the collectivity, at least in the cognitive realm (I have no worked out sense of the affective one), is complicated because of the essentially different shape and nature of collective vs. individual structures. This is a difference that Bateson clearly was aware of (see e.g., p. 112, 175 f.) but one whose full extent he perhaps did not fully appreciate – or, at the least, one that he did not underscore. That is, the individual “schema” (to use a post-Batesonian term for the individual’s own cognitive structure that actually produces behavior) is a constantly changing construction of understanding that relates goals, values, knowledge, and affect to action; its application is always contextually sensitive and, in the final analysis, somewhat ad hoc. The collective “cultural model” is an abstract knowledge structure – whether a scenario, a scene, a gestalt, or a set of propositions – that has implications for action, but where the actual application of the model (what we have spoken of as conceptual “instantiation” and “realization” in actual behavior) and its attendant actions depend on how its conditions are interpreted in the concrete world, on how its goals relate to the goals that structure other potentially relevant models and to the goals of the individual using the model, on how its conditions and variables relate to the operative concrete world, and so forth. Cultural models are commonly thought of as only “logical”, but they clearly can include also expectations about appropriate affect, about proper affective triggers and effects, about their affective effect on viewers, and so forth. 81

Historically, of course, schismogenesis was important as a very early application of feedback loops to human cognition and behavior. Indeed, Bateson – by being one of the participants (along with Margaret Mead) in the Macy Foundation meetings that led to the U.S. avatar of the cybernetics revolution – helped produce our current understanding of self-regulating systems (other such meetings were happening elsewhere in the world; Piaget participated in one in France). Much of modern systems analysis and artificial intelligence work traces back to those beginnings.

252 Chapter 14 – Gregory Bateson: pulling it all together Crucial to Naven was a conception – caught in the process of emerging – of how a bunch of very different (and, apparently, from a traditional point of view, contradictory) analytic generalizations could apply to the same sets of behavior by the same individuals. Bateson saw analytic generalizations as patterns found in behavior – truly, demonstrably there – when the behavior was examined from a particular point of view (i.e., for a given set of people in terms of some given set of analytic questions). Analytic categories were creations of the analyst, but the fact of such creation did not mean that all such creations were created equal; better ones were ones which more systematically accounted for observed patterns in the behavior of the groups of people being considered. Individual behavior was never to be understood as any simple, direct reflex of any analytic generalization, but rather was to be seen as the intersection (and interaction) of the variety of different patterns (some captured by the analyst, some perhaps not yet captured, and some outside the given analytic purview) that were relevant in the behaving actor’s situation (part of my reason for distinguishing a cultural model’s “realization” from its “instantiation”). Some of these patterns were cultural – perceived by the actor in the behavior of recognized classes of people around her or him and then utilized and adapted by him or her; some of these patterns were individual – characteristics of that individual which had developed over that actor’s lifespan. Some of the patterns had to do with the logical or conceptual structure of knowledge and its relationship to goals and values, while other patterns had to do with feelings and their appropriateness to various kinds of situations and various attributes of situations. Bateson’s cultural integration was thus not of Radcliffe-Brownian institutions nor of Malinowski’s ways of meeting basic needs, but was, as befits one of the creators of the information age, an integration of systematic conceptual, affective, social, and individual patterns of knowledge, action, and interaction. The individual – according to this view of pattern – is like a water molecule in the ocean, which bounces up and down (and a little bit back and forth in various sideways directions) as part of the various swell patterns, island interference patterns, wind-blown surface patterns, wakes of ships, waves produced by earthquakes and explosions, and so forth. The molecule belongs to none of these patterns, but moves in response to them all82 – and 82

In the case of the water molecule, almost all of its responses are to pressures from its immediate neighbors. For people re cultural patterns there exists, additionally, the possibility of reacting to perceptions of the larger pattern. But, still, even for our humans, many interesting patterns do seem to depend only (as with the water molecules) on immediate neighbors – as seen in Bateson’s explanation of schismogenesis.

Bateson’s system

253

yet the given molecule is no different than all the rest of the molecules in any of the waves or patterns. The various waves are analytic constructs that we create to capture patterns or regularities that we recognize in nature; none of them has any specific, dedicated substance. I want to suggest that this water molecule is not a bad model for understanding how Bateson thinks of individuals in a sea of other individuals; the patterns he describes are like the different kinds of waves; each individual is always moving in response to each of the patterns, and the motion that the observer sees is always an amalgam of the total mix of analytically separable motions. The major difference is the person can (though need not) notice the patterns and react to them via that awareness – which can be thought of, perhaps, as just some added wave patterns! As an example of what I mean by different, independent patterns coming together in each bit of behavior by each individual – as well as indicating how I think individual agency enters into the picture – let us consider an example from Bateson, that concerning Iatmul transvestitism. For reasons relating to social and cognitive structural issues, the Naven ceremonial involves transvestitism, but those social and cognitive concerns say nothing about the tenor or tone of the cross-dressing – the fact that men, when cross dressing, play it as low comedy burlesque, while women play it for all the serious pomp and self-aggrandizement they can. The reason for the tenor has to do with the relative status of Iatmul men and women – and, thus, of the preferred behaviors of men and women. Men are supposed to be proud, aggressive, and individualistic, while women are supposed to be more cooperative, socially supportive, and retiring. Men (Iatmul men, of course), when dressing as women, seem very anxious that they not be mistaken for people who actually aspire to be or behave like women – hence the burlesque (a tendency for Iatmul men to be a bit histrionic seems to reinforce this approach to cross dressing). Women, on the other hand, gain status by crossdressing and cross-behaving, and so seem to try to make the most of their moments in the spotlight. This business of how the sex or gender roles are played seems to have no intrinsic connection with the reasons (in Iatmul culture) for the cross-dressing and cross-behaving, but flows from other Iatmul values and attitudes – ones that, as Bateson notes, are not unrelated to the way that cross-dressing among run-of-the-mill folks works in our own society. The patterns – of how a Naven ceremonial works, of how Iatmul men and women play at cross-dressing – are bigger than single individuals, and not so subject to individual agency; within these patterns people seem a little like the water molecules described above. But active decisions are involved, including whether or not to have a Naven ceremonial (they are, apparently,

254 Chapter 14 – Gregory Bateson: pulling it all together rare), how fancy a one to put on, whether or not/how deeply any given individual chooses to participate (it is extended relatives, vs. focal ones, who perform the ceremonial), and so forth – even if there are social and traditional pressures that may be relevant. The patterns of behavior themselves are not so much a matter of agency because they constitute the public code by which the ceremonial is recognized, its relations played out, and its effects thus effectuated – and because the patterns are the ways in which the central Iatmul-ness of the players is recognized. The function and functioning of culture requires the continuity of such cultural patterns – and it is in this sense that the agency of the players is not so much an issue – even if there still does exist some room for individual tuning and adjusting. I want to note that this way of thinking about the relationship of individuals to cultural patterns (leaving aside for the moment individual patterns), by having cultural patterns emerge from (and continue via) ongoing dynamic interactions with other people, suggests that these collective patterns are not deeply internalized within the individual psyches, but instead represent something closer to choices that an individual makes (under more or less social pressure) about the relevant context and purposes of behavior. One’s range of choices, of course, is not unlimited, but is constrained by a variety of cognitive, affective, and social considerations, including the range of roles (or cognitive or behavior patterns) one knows, how well these roles (or patterns) are known (to oneself and to relevant others), how convincingly one fits the social profile of the role (or pattern), what is understood to be appropriate to the given situation and context, what kinds of stakes various players have in who does what, and so forth. In a sense, then, one’s choice of roles and patterns can always be seen as a matter for negotiation, depending on the circumstances. I am speaking of “patterns”83 – a word Bateson did use, but also one that I have preferred from my own experience – rather than of “rules” or “norms” or such because I think the term better captures the kinds of regularities that native actors (each of us in our native cultures) learn, respond to, and act in terms of. In some cases – as most extremely in the regularities of grammar, but notably also in areas such as kinship – the patterns are systematic enough to be describable by formal rules. Where rules can be created that capture the regularities, they provide a wonderful analytic tool for exploring the 83

Bateson indicates that he got some of his ideas about pattern from Ruth Benedict. I just want to note that in the present discussion the word is being used more in reference to perceived patterns of behavior than to abstracted cultural patterns.

Bateson’s system

255

shape, characteristics, causes, and effects of the regularities; but the leap which many linguists and anthropologists then make to an assumption that the rules exist in that form in native minds and are used as rules to generate native behavior is I think deeply misguided. (I have shown in Kronenfeld 2006b how formal analytic rules can accurately capture regularities without being directly involved in the production of those regularities.) And I think my misgivings are of the sort that Bateson similarly felt regarding the versions he encountered. At least as worked out by Piaget, formal understandings of regularities in the world (including the effects of behavior) presuppose (and are built out of) a more concrete understanding of a range of specific instances of those regularities, and such concrete understanding, in turn, relies upon a range of relevant sensory-motor experience. This progression seems something of what Bateson had in mind with the idea of “meta-” or “deutero-learning”. What this means for the regularities in behavior that we first encounter in others and then take on as our own is that we first must experience the behaviors before we can pull out the regularities that run across them, and we have to capture a range of specific regularities before we can form more abstract conceptions of what is going on. What we are always doing is reacting to apparent patterns in the behavior of those around us, and utilizing those patterns in our own behavior – with guesses about what are the operative regularities, to whom they apply, for what ends they are used, and so forth. We then use feedback from those around us to fine-tune the ones we got right, to recast ones that we really screwed up, and to decide to dump those that no one seems to recognize. This mode of transmission means that our learning of cultural or linguistic patterns is never perfect (in the sense of precisely reproducing the patterns of those from whom we are learning). What we learn of things cultural or linguistic is, mostly, not so much constrained by the evolved formal shape of the conceptual (or affective) thing being learned as by the incidental (sometimes accidental and sometimes the result of exogenous factors) facts of our experience – which versions we encounter in which contexts, and which ones we encounter frequently enough for any given level of learning and generalization. As we encounter things in the world, including the behavior of others (whether people, animate beings, or other objects), we keep looking for possible regularities and attempting to generalize from ones we think we find; and we keep striving for more and more systematic and productive representations of the patterns that seem to be useful. And in this process we try to comprehend the details that preclude what otherwise might seem relevantly contradictory generalizations.

256 Chapter 14 – Gregory Bateson: pulling it all together As noted earlier, the patterns that we systematically develop in interaction with those around us are the stuff of culture and language. Regularities extracted from different groupings of people represent different cultures or sub-cultures. Seen this way, it seems reasonable to conclude that cultural groupings represent a variety of overlapping hierarchies of inclusion (the more systematic nature of language makes the overlap part there trickier). And, it seems to follow, in turn, that we all belong to a great many subcultural (and maybe cultural) groups. This line of thought gets us off into the important questions – about distributed systems of knowledge, about the relationship of shared knowledge to society, and about the lack of any essential or pure instantiation of any cultural system – with which this book is concerned. 3. Cultural models If we see human behavior in terms of the kind of intersecting patterns that I have described, then cultural models can be understood as patterns in which people participate whether they are aware or not. Using our Batesonian perspective we can imagine that these participants are not themselves simply following or adhering to any single cultural model, but are working off of – following but modulating – patterns of behavior that seem relevant to their individual perceptions of their situations, goals, and abilities, and that cohere with their dispositions and senses of themselves. In the above supposition I am conflating two kinds of patterns. One is that viewed by an external observer – as the pattern of observed behavior – and the other is that set of regularities (whether cognitive or affective) followed by any given producer of the behavior. But the two are more tightly tied together than perhaps one might think by the fact that individuals in the culture only themselves experience the former – observed patterns of action – in their interactions with others and so whatever shared patterning that there is (vs. individual variation) in the latter – the process of understanding and constructing that produces and interprets the actions – is the result of similar inferences made from similar stimuli by similar minds in similar contexts. It is the latter kind of pattern and not the former which constitutes a cultural model. This view entails a supposition that in responding to patterns we perceive in the actions of those around us we don’t simply sense and respond to the perceived behavioral pattern itself, but link it to a class of actors, a class of situations, a set of goals, a feeling state and so forth – that is, into a coherent, logical (as Bateson used the term) package that

Some lessons from Bateson

257

makes sense of the sensed behavioral pattern and that enables us ourselves to appropriately (by our understanding) behave in the indicated way should the occasion arise. People need not be self-consciously aware of a pattern in order to follow it. People’s self-awareness then becomes here, as elsewhere in things psychological, an issue for further research – concerning when, why, and under what conditions it emerges. A lack of self-conscious awareness, of course, does not preclude a sense of when behavior (whether real or described or fictional) is consistent – or inconsistent – with how some class of people supposedly behave in some situation. The response to consistent patterns is what generates the cultural models in the first place. Those responses – as with responses to pronunciation in the context of a study of phonology – give us an empirical handle on the status of hypothesized cultural models and enable their experimental study.

4.

Some lessons from Bateson

4.1. How it all goes together By and large it seems that psychological anthropologists have concentrated on either the cognitive side or the affective side; certainly that has been true of me. To some degree this seems to represent a reasonable division of labor – it certainly is hard to do everything well. But this division of labor can only work if there exists some good theoretical and empirical framework for bringing the separate tasks (back) together to produce an integrated and unified understanding of the particular cultures we study as well as of culture as a general construct. I am not in a position to speak to the adequacy or contemporary usefulness of Bateson’s methodology and angle of attack for describing the ethos of Iatmul culture and/or for relating the affective life of individuals to that ethos. It certainly struck me as better than that of his contemporaries at the time, and better than approaches such as “modal personality” and “national character” which were in vogue back when I was a student (but well after Bateson’s work) – but which neglected his insights into the great difference between individual psychology and the psychological patterns encouraged or rewarded by one or another culture. This book’s present concern in regard to affect is not with any theory of affect per se (as important as I do believe such issues to be), but with the problem of integrating and unifying the different approaches. Bateson gives us a model for how to think about the relationship of individual personality

258 Chapter 14 – Gregory Bateson: pulling it all together to the affective and behavioral patterns rewarded, selected for, and maybe shaped by a given culture, and a model for how to relate these affective patterns to knowledge structures. I mean both that the instance he provides (regarding the Iatmul Naven ceremonial) offers an exemplar for how we might pull such relationships together and that his theoretical or analytic discussion of how these separate perspectives (on the behavior of individuals) go together provides us, if not yet with a real theory, at least with a good model for how such a theory might be approached.

4.2. Eidos and cultural models Bateson’s cultural structure and eidos represent the cognitive side of thought and behavior – that is, the logic of what goes with what, what entails what, what skills earn one respect, and so forth. In terms of today’s cognitive anthropology the role of Bateson’s cultural structures is filled by various kinds of cognitive structures (paradigms, taxonomies, prototypic scenes and scenarios, and so forth – shading into “cultural models”) and the role of his eidos by individual “cultural models” (in their sense as models for behavior, interpretation of behavior, etc.). But cultural models differ in some interesting ways from Bateson’s eidos. They are more instrumental, aimed at achieving some goal, than is eidos, which leans more in the direction of presenting modes of cognitive functioning. Since Naven was written well before the advent of cognitive anthropology (or the disciplines of cognitive sciences and artificial intelligence on which it sometimes draws), it is not surprising that Bateson did not there address these more recent kinds of cognitive structures. But I do think we have perhaps missed out on something by not attending more to the kinds of patterns that Bateson did address. The Batesonian kinds of considerations are not totally absent today. I am aware, for instance, of such approaches sometimes emerging in discussions of particular sociolinguistic/cultural settings that have a sharp, specific flavor (such as Samoan Fonos – see Duranti 1981), but the Batesonian considerations seem not common to ethnographic practice in general and, more importantly, not integrated systematically into mainline cognitive studies in anthropology.

Some concluding reactions to Naven

259

5. Some concluding reactions to Naven I would like to wrap up this discussion of Naven by discussing a few relationships that Bateson poses or queries, but does not really address – relationships that we maybe today are in a little bit better position to think about than was he. 5.1. Sociology and culture Bateson queries, but does not really offer insight into, the relationship between sociology and culture. In Naven he particularly explains the contrast between Radcliffe-Brownian sociological structure – the structure formed by groups of people and the relations among these – and his own notion of cultural structure – the structure formed by relations among conceptual units. But these are treated as two (or three – depending on whether one takes collective cognitive structure and collective affective structure as part of a more general psychological structure or not) separate ways of looking at patterns of behavior that run across individuals; no attempt is made to interrelate the two. In this book I suggest (with an explanation of how) that the two are in fact mutually constitutive of each other. That is, I suggest that cultural structures, whether cognitive or affective, are properties of groups of people (vs. separate individuals), and are learned via observation of and interaction with members of relevant groups – and thus are dependent on the groupings produced by the social life of the people involved. Conversely, membership in social groupings is not produced by any disembodied divine fiat, but rather by the rules or means of affiliation defined by a given culture, and fellow members are recognized by means of attributes of their behavior (including language, dress, special knowledge, carriage, and so forth), most of which are learned – as cultural patterns of the given group or subgroup (whether nation, ethnic group, profession, civic club, sports group, or whatever). I don’t mean that affiliation is strictly a matter of cultural learning; we do come equipped with some general social dispositions (as noticed, for instance, long ago by George Homans in The Human Group [1950]) and some communicative and linguistic dispositions (as noted, among many others, by people as diverse as Noam Chomsky and Joseph Greenberg). But I do mean that culture represents the shared game plan that enables human groups to function as systems of distributed cognition. This view goes considerably beyond anything in Naven, but the kind of systems considerations that underlie it do go back in part to Bateson’s concerns with feedback sys-

260 Chapter 14 – Gregory Bateson: pulling it all together tems (including the balanced feedback relations among different components of a cultural system) and with the ways in which patterns of behavior form and replicate culture and address people’s sense of who they are and whom they ally with or oppose. 5.2. Levels of analysis – a demurrer Bateson himself (e.g., p. 298) emphasizes as a major take-home lesson of his enterprise the necessity of keeping levels of analysis clear, distinct, and on the table. By levels he means, for example, statements about particular behaviors vs. statements about regularities in behaviors vs. statements about changes in regularities, and so forth. Those comments of his, while eminently reasonable, seem too abstract to be useful, and not nearly as helpful as his more mundane and concrete discussion and explication of his understanding of Naven and Iatmul culture as seen in it. There are perhaps a couple of reasons – beyond whatever part of it one may simply charge up to differences in taste between him and me. In part, I think that some of the framing notions, from Whitehead, Russell, and others were new and exciting enough at that time, but have, since then, become much more familiar to us. But, also, I think that that kind of a focus on abstract qualities of our anthropological statements – that is, some kind of formal meta-rules – rather than on the actual specifications that we need in order to accurately and productively describe some system of behavior always gets in the way of the kinds of insightful generalizations that science (which Bateson clearly saw himself as doing – e.g., p. 30) depends on. Such meta-statements do come after we have understood some regularity, and are important to the explanations which philosophers and historians of science use to account for what has happened, but I do not think that they ever much contribute to the generation of the actual insights. We anthropologists and social scientists ourselves are still only “natives” in our modes of cognitive and social functioning, and so, as described above, normally and naturally experience and respond only to patterns of experienced (concrete) regularities, not to any kind of abstract or analytic levels.84 I should make clear here that my complaint is not with rough and ready stabs at potential generalizations – such as Bateson’s thoughts regarding Iatmul (and British) transvestism. Naven has many of these and they almost 84

“Levels” and the like are things, rather, that we later introduce as analytic and conceptual tools.

Some concluding reactions to Naven

261

always are stimulating and provocative. These are concrete, if not formally phrased, statements about putative regularities – whether in behavior, attributes, or function, and whether in how they are or how they appear to be changing. These statements are not phrased formally because not enough is known about their referents to allow any precise characterization. The virtue of offering them, beyond some contribution to our ethnographic understanding of Iatmul culture, is that such tentative generalizations of suspected insights guide subsequent structuring of relevant data definition and collection, and thus form part of the feedback process by which better and more formal theoretical formulations are eventually worked out. Many of Bateson’s insights (in Naven and elsewhere [Bateson 1972] – about ethos and eidos, about schismogenesis, etc.) have thus contributed to significant refinement of our theoretical and ethnographic apparatus, even if for most of these insights the process of refinement is not yet finished. My reservation is about any attempt to insist on meta-statements concerning the logical level of putative claims or insights as part of that feedback process.85 My view is based in part at least, perhaps I should confess, on a couple of examples of abstract vs. concrete characterizations which I have seen in my attempts to understand the regularities of kinship systems – those of Tax and Romney, and on my attempts to simulate social systems embodying emergent properties. I think it possible that the post-cybernetic advent of computer programs as ways of formalizing and making explicit relatively 85

We don’t, I think, know how to analyze the logical level of a claim until we have a clear understanding of the actual claim – and thus I think the meta-analysis can only come after the analysis itself. Otherwise our meta-analysis is very likely to be significantly skewed by attributes of the fuzziness (aka looseness and errors) of our understanding of what’s being analyzed, as opposed to representing directly properties intrinsic to our putative insight. This risk is particularly relevant because the kind of levels analysis that Bateson speaks of concerns statements about posited reality rather than that posited reality itself. As an example, Bateson’s early attempts to relate the tone and tenor of Iatmul transvestitism to attributes of the Naven ceremonial might have led to complicated discussions of the pertinent levels of analysis (had he followed his own concluding advice – which he pretty much did not do), while, as he eventually realized, the relationship is purely epiphenomenal, following as we have seen from the intersection of two simple but logically distinct patterns. Once we have understood the interrelationship, a proper statement of it is indeed a meta-statement that addresses the lower level statements which describe the logic of Naven and the logic of cross-dressing, but trying to frame the problem (before its solution) in such meta- terms would not have helped produce the final insight.

262 Chapter 14 – Gregory Bateson: pulling it all together ad hoc guesses and insights and objectively evaluating the systemic effects of such guesses and insights removes some of the kinds of logical problems that the levels concern seemed to offer a way out of.

References

Anderson, Eugene N. 1996 Ecologies of the Heart. New York /Oxford: Oxford University Press. Barth, Fredrik 1992 Towards greater naturalism in conceptualizing societies. In Conceptualizing Society, Adam Kuper (ed.). London: Routledge. Bateson, Gregory 1958 Naven. Palo Alto: Stanford University Press [Original in 1938; the later edition has added material.] 1972 Steps to an Ecology of Mind: Collected Essays in Anthropology, Psychiatry, Evolution, and Epistemology. San Francisco: Chandler Publishing Company Berlin, O. Brent 1972 Speculations on the growth of ethnobotanical nomenclature. Language in Society 1: 51–86. 1992 Ethnobiological Classification. Princeton: Princeton University Press. Boster, James 1985 Requiem for the omniscient informant: There’s life in the old girl yet. In Directions in Cognitive Anthropology, J. Dougherty (ed.). Urbana: University of Illinois Press. 1988 Natural sources of internal category structure: Typicality, familiarity, and similarity of birds. Memory and Cognition 16: 258–270. Bourdieu, Pierre 1977 Outline of a Theory of Practice, translated by R. Nice. Cambridge: Cambridge University Press. Brown, Roger 1965 Social Psychology. First Edition. New York: Free Press. Bruner, Jerome S., Jacqueline J. Goodnow, and George A. Austin 1956 A Study of Thinking (with an appendix on language by Roger W. Brown). New York: John Wiley. Carnap, R. 1938 Foundations of logic and mathematics. In International Encyclopedia of Unified Science, Vol. 1, O. Neurath, R. Carnap and C. W. Morris (eds.), 139–214. Chicago: University of Chicago Press. Casson, Ronald W. 1983 Schemata in cognitive anthropology. Annual Review of Anthropology 12: 429–462. Chomsky, Noam 1957 Syntactic Structures. The Hague: Mouton. 1972 Language and Mind. New York: Harcourt Brace Jovanovich.

264 References Colby, Benjamin N. 1973 A partial grammar of Eskimo folktales. American Anthropologist 75: 645–662. 1975 Culture Grammars. Science 187: 913–919. Colby, Benjamin N. and Lore M. Colby 1981 The Daykeeper: The Life and Discourse of an Ixil Diviner. Cambridge, MA: Harvard University Press. Colby, Benjamin N., James Fernandez, and David B. Kronenfeld 1981 Toward a convergence of cognitive and symbolic anthropology. American Ethnologist 8: 422–450. D’Andrade, Roy G. 1985 Character terms and cultural models. In Directions in Cognitive Anthropology, Janet W. D. Dougherty (ed.), 321–344. Urbana: University of Illinois Press. 1987 A Folk Model of the Mind. In Holland and Quinn 1987, 112–148. 1992 Schemas and Motivation. In D’Andrade and Strauss 1992, 23–44. 1995 The Development of Cognitive Anthropology. Cambridge: Cambridge University Press. D’Andrade, Roy and Claudia Strauss (eds.) 1992 Human Motives and Cultural Models. Cambridge: Cambridge University Press. De Munck, Victor 1996 Love and marriage in a Sri Lankan Muslim community: Towards a re-evaluation of Dravidian marriage practices. American Ethnologist 23: 698–716. De Munck, Victor and Andrey Korotayev 2000 Cultural units in Cross-Cultural Research. In Special Issue: Comparative Research and Cultural Units, guest editor, Victor De Munck. Ethnology 34: 335–348. Dougherty, Janet W. E. and Charles M. Keller 1985 Taskonomy: A practical approach to knowledge structures. In Directions in Cognitive Anthropology, Janet W. D. Dougherty (ed.), 161– 174. Urbana: University of Illinois Press. [orig. 1982] Duranti, Alessandro 1981 The Samoan Fono: A Sociolinguistic Study. (The Australian National University, Pacific Linguistics Monograph B80). 1997 Linguistic Anthropology. Cambridge: Cambridge University Press. 2003 Language as culture in U.S. anthropology. Current Anthropology 44 (3): 323–347. Durkheim, Emile 1938 The Rules of Sociological Method. 8th Edition, translated by Sarah A. Solovay and John H. Mueller, edited by George E. G. Catlin. New York: The Free Press. [French original: 1894]

References 265 1947

The Elementary Forms Of The Religious Life: A Study In Religious Sociology. Translated by Joseph A. Swain. Glencoe, IL: Free Press. [French original: 1912] 1951 Suicide, A Study In Sociology. Translated by John A. Spaulding and George Simpson. New York: Free Press. (French original, 1897) 1984 The Division Of Labor In Society. Translated by W. D. Halls, with an introduction by Lewis A. Coser. New York: Free Press. [French original: 1902] Evans-Pritchard, E. E. 1940 The Nuer. Oxford: Oxford University Press. 1951 Kinship and Marriage Among the Nuer. Oxford: Oxford University Press. Flavell, John H. 1963 The Developmental Psychology of Jean Piaget. New York: D. Van Nostrand Company. Fortes, Meyer 1945 The Dynamics of Clanship among the Talensi. Oxford: Oxford University Press. 1949 The Web of Kinship among the Talensi. Oxford: Oxford University Press. Frake, Charles O. 1961 The diagnosis of disease among the Subunun of Mindanao. American Anthropologist 93: 113–132. 1962 The ethnographic study of cognitive systems. In Anthropology and Human Behavior, T. Gladwin and W. C. Sturtevant (eds.), 72–85. Washington: Anthropological Society of Washington. 1964a Notes on queries in ethnography. American Anthropologist 66: 132– 145. 1964b A structural description of Subanun “Religious Behavior”. In Explorations in Cultural Anthropology, Ward H. Goodenough (ed.), 111– 129. New York: McGraw-Hill. Gatewood, John 2000 Distributional instability and the units of culture. In special issue: Comparative Research and Cultural Units, guest editor, Victor De Munck. Ethnology 34: 293–303. 2001 Reflections on the Nature of Cultural Distributions and the Units of Culture Problem. Cross-Cultural Research, Vol. 35 No. 2, May 2001: 227–241 in press Socially Distributed Cognition. In Cambridge Encyclopedia of Language Sciences, Patrick Hogan (ed.). Gatewood, John B. and John W. G. Lowe n.d. The blind men and the credit union: Adventures with a socially distributed cultural model. Expanded version of paper (“Cultural models

266 References and multiple perspectives: Employees’ understandings of credit unions”) presented at First General Scholarly Meetings of the Society for Anthropological Sciences, February 23–27, 2005, Santa Fe, New Mexico. Gladwin, Christina H. 1975 A model of the supply of smoked fish from Cape Coast to Kumasi. In Formal Methods in Economic Anthropology, S. Plattner (ed.). A special publication of the American Anthropological Association, No. 4: 77–127. 1976 A view of the Plan Puebla: An application of hierarchical decision models. American Journal of Agricultural Economics 58: 881–887. 1977 A model of farmers’ decisions to adopt the recommendations of Plan Puebla. Ph.D. dissertation, Stanford University, Food Research Institute. 1980 A theory of real-life choices: Applications to agricultural decisions. In Agricultural Decision Making, P. Barlett (ed.), 45–85. New York: Academic Press. Gladwin, C. H, and J. Butler 1984 Is gardening an adaptive strategy for Florida family farmers? Human Organization 43 (3): 208–216. Gladwin, Hugh and Christina H. Gladwin 1971 Estimating market conditions and profit expectations of fish sellers at Cape Coast, Ghana. In Studies in Economic Anthropology, G. Dalton (ed.), 122–142. Anthropological Studies 7. Washington, DC: American Anthropological Association. Goffman, Erving 1959 The Presentation of Self in Everyday Life. Garden City, NY: Doubleday. 1963 Behavior in public places: notes on the social organization of gatherings. New York: Free Press 1974 Frame Analysis: an Essay on the Organization of Experience. New York: Harper and Row. 1981 Forms of talk. Philadelphia: University of Pennsylvania Press Gombrich, E. H. 1960 Art and Illusion. Princeton: Princeton University Press. Greenberg, Joseph H. 1968 Anthropological Linguistics. New York: Random House. 1966 Language Universals with Special Reference to Feature Hierarchies. The Hague: Mouton. Grice, H. P. 1957 Meaning. Philosophical Review 67. Gulliver, P. H. 1998 The Family Herds: A Study of Two Pastoral Tribes in East Africa, the Jie and Turkana. London: Routledge and Kegan Paul.

References 267 Hall, Edward 1959 The Silent Language. New York: Doubleday and Co, Inc. 1966 The Hidden Dimension. Anchor Books ed. Garden City, NY: Anchor Books. Hofstadter, Douglas 1999 Gödel, Escher, Bach: an Eternal Golden Braid. New York: Basic Books. Holland, Dorothy 1992 The woman who climbed up the house: Some limitations of Schema Theory. In New Directions in Psychological Anthropology. M. White, T. Schwartz, and C. A. Lutz, (eds.), 68–79. Cambridge: Cambridge University Press. Holland, Dorothy and Naomi Quinn (eds.) 1987 Cultural Models in Language and Thought. Cambridge: Cambridge University Press. Homans, George C. 1950 The Human Group. New York: Harcourt, Brace. Hovland, Arild n.d. How imperative the Sami Ethnos? Sami youth at the Crossroads. 1995 Literal identity? On the textual reconstruction and manifestation of ethnic identity among young Sami. Paper presented at a Symposium on “Ethnic identity: psychological and anthropological perspectives”, Karasjok, Norway, June 12–14, 1995. Hunn, Eugene 1976 Toward a perceptual model of folk biological classification. American Ethnologist 3: 508–524. 1977 Tzeltal Folk Zoology: The Classification of Discontinuities in Nature. New York Academic Press. 1985 The utilitarian factor in folk biological classification. In Directions in Cognitive Anthropology, Janet W. D. Dougherty (ed.), 117–140. Urbana: University of Illinois Press. 1987 Science and common sense: a reply to Atran. American Anthropologist 89: 146–149 Hutchins, Edwin 1980 Culture and Inference: A Trobriand Case Study. Cambridge, MA: Harvard University Press. 1995 Cognition in the Wild. Cambridge, MA: MIT Press. Kahneman, Daniel, Paul Slovic, and Amos Tversky (eds.) 1982 Judgment Under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press. Keller, Charles M. and Janet Dixon Keller 1996 Cognition and Tool Use: The Blacksmith at Work. Cambridge: Cambridge University Press.

268 References Keller, Rudi 1994 On Language Change. The Invisible Hand in Language. London and New York: Routledge. Köhler, Wolfgang 1927 The Mentality of Apes. London: Routledge & Kegan Paul. 1947 Gestalt Psychology, An Introduction To New Concepts In Modern Psychology. New York: Liveright Publishing Corporation Kroeber, A. L. and C. Kluckhohn 1952 Culture: A Critical Review of Concepts and Definitions. New York: Vintage Books. Kronenfeld, David B. 1973 Fanti kinship: The structure of terminology and behavior. American Anthropologist 75: 1577–1595. 1975 Kroeber vs. Radcliffe-Brown on kinship behavior: the Fanti test case. Man 10: 257–284. 1976 Computer analysis of skewed kinship terminologies. Language 52: 4: 891–917. 1980a A formal analysis of Fanti kinship terminology. Anthropos 75: 586– 608. 1980b Particularistic or universalistic analysis of Fanti kinterminology: the alternative goals of terminological analysis. Man 15 (1): 151–169. 1985 Numerical taxonomy: Old techniques and new assumptions. Current Anthropology 26 (1): 21–41 1989 Morgan vs. Dorsey on the Omaha Cross-Parallel Contrast: Theoretical implications. L’Homme 19: 78–101. 1991 Fanti kinship: Language, inheritance, and kingroups. Anthropos 86: 19–31 1996 Plastic Glasses and Church Fathers: Semantic Extension from the Ethnoscience Tradition. (Oxford Studies in Anthropological Linguistics.) New York: Oxford University Press. 1998 Stranger communities and “Sweetheart Dances”. Anthropos 93: 77–88. 2001a Morgan, Trautmann and Barnes, and the Iroquois-type cross/parallel distinction. Anthropos 96: 423–432. 2001b Using Sydney H. Gould’s formalization of kin terminologies: Social information, skewing, and structural types. Anthropological Theory 1: 173–196. (Special Issue: Kinship, guest edited by David B. Kronenfeld) 2001c Introduction: The uses of formal analysis re cognitive and social issues. Anthropological Theory 1: 147–172. (Special Issue: Kinship, guest edited by David B. Kronenfeld) 2004 Definitions of cross vs. parallel: Implications for a new typology; (an appreciation of A. Kimball Romney). Cross-Cultural Research The Journal of Comparative Social Science 38: 249–269.

References 269 2006a

Issues in the classification of kinship terminologies: Toward a new typology. Anthropos 101: 203–219. 2006b Formal rules, cognitive representations, and learning in language and other cultural systems. Language Sciences 28: 424–435. 2008 Fanti Kinship and the Analysis of Kinship Terminologies. Champaign, IL: University of Illinois Press. Kronenfeld, David B. and Henry W. Decker 1970 Structuralism. Annual Review of Anthropology 8: 503–541. Kronenfeld, David B. and Andrea Kaus 1993 Starlings and other critters: Simulating society. Journal of Quantitative Anthropology 4: 143–174. Kronenfeld, David B. and Jerrold E. Kronenfeld 2006 CritSim2: A program for simulating society. In Cybernetics and Systems 2006, Vol. 1, R. Trappl (ed.), 301–303. Vienna: Austrian Society for Cybernetic Studies. n.d. CritSIM: An agent-Based model of Durkheimian Emergent Social Phenomena. Paper delivered June 9th, 2007 in the Human Science and Complexity Series. Kronenfeld, David B. and Gabriella Rundblad 2003 The semantic structure of lexical fields: Variation and change. In Words in Time, Diachronic Semantics from Different Points of View, Regine Eckardt, Klaus von Heusinger, and Christoph Schwarze (eds.), 67–114. Berlin /New York: Mouton de Gruyter. Kronenfeld, David B. and Lynn L. Thomas 1983 Revised lexicostatistical classification of Salishan languages. American Anthropologist. 85: 372–378. Kronenfeld, David B. and Halvard Vike 2002 Collective representations and social praxis: Local politics in the Norwegian welfare state. Journal of the Royal Anthropological Institute 8: 621–643. Lashley, Karl 1951 The problem of serial order in psychology. In Cerebral Mechanisms in Behavior, Lloyd A. Jeffress (ed.), 112–136. New York: Wiley. Lave, Jean 1988 Cognition in Practice: Mind, Mathematics, and Culture in Everyday Life. Cambridge /New York: Cambridge University Press. Lave, Jean and Etienne Wenger 1991 Situated Learning: Legitimate Peripheral Participation. Cambridge, New York: Cambridge University Press. Lehman, F. K. (F. K. L. Chit Hlaing) 1985 Cognition and computation. In Directions in Cognitive Anthropology, Janet W. D. Dougherty (ed.), 19–48. Urbana /Chicago: University of Illinois Press.

270 References 1993

The relationship between genealogical and terminological structure in kinship terminologies. Journal of Quantitative Anthropology 4: 95– 122. 2001 Aspects of a Formalist Theory of kinship: the functional basis of its genealogical roots and some extensions in Generalized Alliance Theory. Anthropological Theory 1: 212–238. Lévi-Strauss, Claude 1960 La Geste d’Asdiwal. Translated, reprinted in The Structural Study of Myth and Totemism, E. Leach (ed.), 1–48. London: Tavistock / New York: Praeger. 1963 Structural Anthropology. Translated by C. Jacobson, R. G. Schoepf. New York: Basic Books. 1964 Tristes Tropiques: An Anthropological Study of Primitive Societies in Brazil. Translated by J. Russell. New York: Atheneum. Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press Longus 1890 Daphnis and Chloe. The Elizabethan version from Amyot’s translation by Angel Day. Edited by Joseph Jacogs. London. [Greek original 2nd Century AD; Amyot’s French Translation 1559; Day’s version 1587] 1968 Daphnis and Chloe. Translated by Paul Turner. (The Penguin Classics) Harmondsworth, Middlesex, England. [Greek original 2nd Century AD] Lyons, John 1977 Semantics, Volumes 1 and 2. Cambridge: Cambridge University Press. Kevin Lynch 1960 Image of the City. Cambridge, MA: MIT Press. Metzger, Duane and Gerald Williams 1963 A formal ethnographic analysis of Tenejapa Ladino weddings. American Anthropologist 65: 1076 –1101. 1966 Some procedures and results in the study of native categories: Tzeltal “Firewood”. American Anthropologist 68: 389–407. Miller, George A. 1956 The Magical number seven, plus or minus two: Some limits of our capacity for processing information. Psychological Review 63: 81–97. Montague, R. 1968 Pragmatics. In Contemporary Philosophy, R. Klibansky (ed.), 102– 121. Florence: La Nuova Italia Editrice. [Reprinted in R. Montague, 1974: Formal Philosophy: Selected Papers, R. H. Thomason (ed.), 119–147. New Haven: Yale University Press] Moore, Robert L. 1981 Modernization and Westernization in Hong Kong: Patterns of culture change in an urban setting, Ph.D. thesis, University of California, Riverside. Ann Arbor: University Microfilms.

References 271 1998

Love and limerence with Chinese characteristics: Student romance in the PRC. In Romantic Love and Sexual Behavior: Perspectives from the Social Sciences, Victor C. De Munck (ed.), 251–284. Westport, CT/ London: Praeger. Morris, Charles W. 1938 Foundations of the theory of signs. In International Encyclopedia of Unified Science, O. Neurath, R. Carnap, and C. Morris (eds.), 77–138. Chicago: University of Chicago Press. Mukhopadhyay, Carol C. 1980 The Sexual Division of Labor in the Family: a Decision-Making Analysis. Ann Arbor: University Microfilms. 1984 Testing a decision process model of the sexual division of labor in the family. Human Organization 43: 227–242. Padgett, John F. and Christopher K. Ansell 1993 Robust action and the rise of the Medici, 1400–1434. America Journal of Sociology 98: 1259–1319. Perry, Mark 2003 Distributed Cognition. In HCI Models, Theories, and Frameworks: Toward a Multidisciplinary Science, John M. Carroll (ed.), 195–223. San Francisco, CA: Morgan Kaurman. Pike, Kenneth L. 1967 Language in Relation to a Unified Theory of the Structure of Human Behavior. The Hague: Mouton and Co. (original 1954, 1955, 1961) Plattner, Stuart 1984 Economic Decision Making of Marketplace Merchants: An Ethnographic Model. Human Organization 43: 252–264. Putnam, Hilary 2003 Meaning and Reference. In Meaning, Mark Richard (ed.), 70–81. Malden, MA: Blackwell. [Republication of Hilary Putnam, 1973, “Meaning and Reference”. Journal of Philosophy 70: 699–711] Quinn, Naomi 1975 Decision-making models of social structure: a critical review. American Ethnologist 2: 19–45. 1987 Convergent evidence for a cultural model of American marriage. In Holland and Quinn 1987: 173–192. 1992 The motivational force of self-understanding: Evidence from wives’ inner conflicts. In D’Andrade and Strauss 1992, 90–126. Quinn, Naomi and Dorothy Holland 1987 Culture and Cognition. In Holland and Quinn 1987: 3–40. Randall, R. A. 1977 Change and variation in Samal fishing: Making plans to “Make a Living” in the Southern Philippines. Ph.D. Dissertation, University of California, Berkeley, Department of Anthropology.

272 References Read, Dwight 1984 An algebraic account of the American kinship terminology. Current Anthropology 25: 417–440. Romney, A. Kimball, William Batchelder, and Susan Weller 1987 Recent applications of cultural consensus theory. American Behavioral Scientist 31: 163–177. Romney, A. K, S. C. Weller, and W. H. Batchelder 1986 Culture as consensus: A theory of culture and informant accuracy. American Anthropologist 88: 313–338. Rumelhart, David E. 1975 Notes on a Schema for Stories. In Representation and Understanding Studies in Cognitive Science. New York: Academic Press. 1980 Schemata: The building blocks of cognition. In Theoretical Issues in Reading Comprehension: Perspectives from Cognitive Psychology, Linguistics, Artificial Intelligence, and Education, Rand J. Spiro, Bertram C. Bruce, and William F. Brewer (eds.), 33–58. Hillsdale, NJ: Lawrence Erlbaum. Rumelhart, David E., James L. McClelland, and the PDP Research Group 1986 Parallel Distributed Processing: Explorations in the Microstructure of Cognition, Vol. I: Foundations and Vol. II: Psychological and Biological Models. Cambridge, MA: MIT Press. Rundblad, Gabriella and David B. Kronenfeld 2000 Folk-Etymology: Haphazard Perversion or Shrewd Analogy? In Lexicology, Semantics and Lexicography: Selected Papers from the 4th G. L. Brook Symposium, Manchester, August 1998, Julie Coleman and Christian J. Kay (eds.), 19–34. Amsterdam /Philadelphia: Benjamins. 2003 The Inevitability of Folk Etymology: A Case of Collective Reality and Invisible Hands. Journal of Pragmatics 35: 119–138. Sapir, Edward 1963 Selected Writings. David G. Mandelbaum (ed.). Berkeley: University of California Press. [Original: 1949. Selected Writing of Edward Sapir in Language, Culture and Personality] 1925 Sound patterns in language. Language 1: 37–51. [Reprinted in Selected Writings of Edward Sapir, D. Mandelbaum (ed.), 33–45] 1933 La réalité psycholique des phonèmes, Journal de Psychologie Normale et Pathologique 30: 247–265. [Reprinted in Selected Writings of Edward Sapir, D. Mandelbaum (ed.), 46–60 as “The Psychological Reality of Phonemes”] Saussure, Ferdinand de 1959 Course in General Linguistics. Edited by Charles Bally and Albert Sechehaye in collaboration with Albert Reidlinger. Translated by Wade Baskin. New York: Philosophic Library. [French original 1916]. (Highly relevant and important is the critical edition prepared by

References 273 Tullio de Mauro, 1973: Cours de Linguistique Générale. Publie par Charles Bally et Albert Sechehaye, avec la collaboration de Albert Riedlinger. Édition critique préparée par Tullio de Mauro. Paris: Payot.) Schallert, Diane 1982 The significance of knowledge. A synthesis of research related to Schema Theory. In Reading Expository Materials, Wayne Otto and Sandra White (eds.), 13–48. New York: Academic Press. Schank, Roger and R. P. Abelson 1977 Scripts, Plans, Goals and Understanding: An Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum. Searle, J. R, F. Kiefer, and M. Bierwisch (eds) 1980 Speech Act Theory and Pragmatics. Synthese Language Library, Vol. 10. Dordrecht: Reidel Shore, Bradd 1996 Culture in Mind: Cognition, Culture, and the Problem of Meaning. New York: Oxford University Press. Sperber, Dan 1996 Explaining Culture: A Naturalistic Approach. Malden, MA: Blackwell. Stefflre, Volney 1972 Some applications of multidimensional scaling to social science problems. In Multidimensional Scaling, Vol. II, A. K. Romney, R. Shepard, and S. B. Nerlove (eds.), 211–243. New York: Seminar Press. Strauss, Claudia and Naomi Quinn 1997 A Cognitive Theory of Cultural Meaning (Publications of the Society for Psychological Anthropology). Cambridge: Cambridge University Press. Trubetzkoy, N. S. 1969 Fundamentals of Phonology. Berkeley and Los Angeles: University of California Press. Translated by Christine A. M. Baltaxe. [Originally published in German as Grundzüge der Phonologie, 1939] Tyler, Stephen A. 1978 The Said and the Unsaid: Mind, Meaning, and Culture. New York: Academic Press. Wallace, Anthony F. C. 1970 Culture and Personality. New York: Random House. Wilmoth, Stan 1987 The Development of Blackfeet Politics and Multiethnic Categories: 1934–84. Ph.D. dissertation, University of California, Riverside. Ann Arbor: University Microfilms. Young, James C. 1981 Medical Choice in a Mexican Town. New Brunswick, NJ: Rutgers University Press.

Index

biological (biology), 1, 6, 8, 16, 60, 78, 86, 100, 102, 139, 141–143, 146–147, 156, 198, 236 classification, 4, 8, 12, 16, 26, 28–30, 147, 162, 163, 172–173, 179, 184, 190–191, 194, 227, 240, 248 code (see also message), 4, 6, 14, 32–33, 74, 76, 78, 90, 98, 117, 125, 131, 134, 149, 177, 180, 198, 254 cognition, 1–2, 17–19, 41, 53, 56, 72, 77, 88, 94, 122, 138, 159, 215, 219, 250– 251 distributed ~, 3, 5, 37, 40, 83, 93–94, 101, 103, 107, 149, 159, 176, 259 cognitive anthropology, 1, 5, 27–28, 88, 164, 200, 233, 248, 258 complexity, 17, 49, 178, 184 computer model, 56, 95–96 computer simulation, 121–122, 219, 225 contrast, 9, 16, 19, 23, 28, 31–32, 51, 54, 65–66, 75, 81, 92, 97, 100, 110–112, 120, 127, 136, 139–142, 144, 147, 152–153, 179–182, 208, 227, 259 convention, 11, 58–59, 65, 67, 92 natural ~, 65, 68 cultural ~ application, 5, 13, 64, 124, 164, 172–174, 182, 184–186, 188, 191, 210, 251 ~ conceptual systems, 2, 5, 31, 162– 163, 166, 173, 184 ~ instantiation, 5, 76, 85, 87, 97, 125, 167, 169–170, 174–176, 185–188, 194–197, 206–208, 211, 224, 252, 256 ~ invocation, 188, 197, 210 ~ model, 2, 4–5, 27–28, 30–31, 33, 35–36, 51, 59, 66, 69, 75, 85, 88,

107, 125, 132–133, 163–188, 189– 218, 219–227, 230, 234, 237–239, 242, 245–249, 251–252, 257–258 ~ modes of thought, 5, 99, 163 ~ precedents, 30, 50–51, 89, 194, 204– 205, 207 ~ realization, 170, 183, 185, 187–189, 196, 208, 211, 216, 251 culture, 2–6, 8–9, 11, 15–27, 31–34, 37– 39, 44, 51, 64, 69, 74, 76–79, 81–105, 107, 114, 117–118, 121, 124–125, 130–135, 142–143, 147, 151, 153– 154, 156, 162–166, 169, 176–178, 191, 196, 198–199, 201, 203, 208, 210–211, 214, 218–219, 221–223, 225, 227, 231, 234, 237, 250–251, 253–254, 256–257, 259–261 consensus, 6, 12, 203, 225, 228, 230 cultural communities, 11–12, 14, 16, 32–34, 38, 76, 79, 84, 87–88, 98, 104, 114–117, 125, 130, 134, 139, 149, 154, 157–158, 160, 166, 169, 176, 178–180, 202, 215, 218, 220, 228, 230, 246 variability, 15, 23, 30, 86, 104–105, 179, 184, 207, 211, 218, 235, 237, 250 decentralization, 83, 93–96, 101, 107, 114 empirical, 8, 21, 25–26, 31, 76, 124, 164– 165, 185, 190, 198, 207–208, 217, 221, 223, 225, 227, 230, 257 ethnicity, 42, 74, 80, 114–115, 120–121, 129, 134–136, 138–141, 145–161, 179, 181 ethnoscience, 28, 47, 56, 77 examples, 1, 4, 31, 36, 38, 54, 63, 66, 68, 70–71, 89, 95, 107, 110, 114, 117, 140,

276 Index 162, 164, 191–196, 200, 205, 207– 208, 222, 227, 235, 248, 250, 261 ethnic groups, 80, 121, 129, 138, 148, 155, 157 Blackfeet, 136, 138–148, 150 Jew, 12, 43, 114–115, 134, 153, 158–161, 177 Sami, 43, 136, 149–161 kinship, 8, 21, 24, 31, 44, 79, 91, 112– 114, 121–124, 172–173, 184, 189– 196, 207, 224, 226, 254, 261 English, 8, 14, 16, 36, 107, 110– 113, 139–144, 159–160, 162, 182–183, 190, 194–195, 236, 240 Fanti, 12, 139, 189–193 Jew, 12, 43, 114–115, 134, 153, 158–161, 177 love, 1, 133, 162, 164, 167, 170–171, 175, 188, 195–196, 212–213, 224 ranching, 164, 227, 237, 243, 248 extensionist semantics, 22, 206 core referent, 139–140, 144 extension, 5, 10–17, 75, 109, 111–112, 115, 139–143, 147, 190–191, 195, 198, 207, 243–244 focal referent, 74, 108 semantic extension, 16, 75, 111 focal member, 128, 163 formal, 36, 38, 47–49, 57, 61–62, 64, 70, 72, 78–80, 86–87, 91, 107, 112, 118, 124, 160, 163, 165–167, 173, 183, 190, 193, 196–197, 202, 208, 210, 217– 220, 225, 246, 254–255, 260–261 generative, 19, 23, 45, 182, 199 hierarchy, 14, 30, 52, 71–72, 78, 80, 130, 178, 193 inclusion, 9, 13, 28, 62, 75, 80, 96, 176, 178, 180–182, 227–228, 256 hierarchy of ~, 71, 78, 176 internalization, 31–33, 88, 167 language, 1–5, 7–10, 14–28, 32–33, 36– 38, 40, 42, 45, 55–56, 73–74, 76–77,

79, 80, 84–85, 88–92, 99, 100, 104, 106–114, 117, 122–123, 125–132, 136, 139, 142–143, 146–149, 151– 152, 156, 158, 166, 169, 173–174, 177, 182–183, 187, 193, 195, 198– 199, 208, 210, 214, 218–221, 223, 237, 256, 259 learned (see also taught), 1, 5, 34, 37, 48, 52, 75, 78–79, 84, 86–89, 107, 124, 130, 145, 147, 161, 166, 173, 178, 187, 190, 201, 210, 218, 225, 255, 259 linguistics, 2, 13, 26–27, 42, 54, 90–91, 170, 172, 219 marking, 30, 52–53, 68, 75, 145, 147– 148, 178, 183, 186 ~ hierarchy, 52, 178, 183, 186 marked/unmarked, 36, 51–52, 75, 105, 113, 141, 144, 147, 167, 170, 181– 183, 206, 208, 246 message (see also code), 26, 38, 58, 69, 110–111, 192, 197, 214 natural decision making, 28, 35, 47–53 decision theory, 30, 54–56, 222–223 paradigm, 44, 60, 64, 190 parallel distributed processing, 2–3, 77, 93, 103 pattern, 5, 14, 16, 19, 34, 37, 45, 50, 53, 59, 65, 77–79, 87, 89, 92, 96, 106, 109–113, 117, 120, 125, 127, 131, 133, 145, 168, 170, 173, 176, 183, 189–193, 199, 201, 209, 212–215, 220–223, 226, 230, 236, 250–261 pdp, 94–95 pragmatics, 1–2, 7–9, 15, 18, 23, 38, 68–69 presupposition, 4, 9, 32, 68, 75, 77, 87, 116, 123, 131, 170, 175, 192, 201 productivity, 5, 19, 23, 35–36, 78–79, 86, 108–109, 118, 131, 172–173, 182, 195, 197, 207, 226, 247, 255 prototype, 109–111, 139–140, 146, 205, 207–208, 248

Index 277 representation, 18, 20–21, 26, 30, 32–33, 57–58, 65–66, 68, 71, 74, 83–86, 90, 99, 126–128, 132–133, 155, 174, 176, 201, 205, 230 collective ~, 2, 9, 19, 23, 32–33, 40– 46, 64, 69, 74–75, 83–85, 90–91, 122, 125–128, 130–133, 148, 155, 168–169, 173, 177, 200, 202, 210, 212–215, 219–220 individual ~, 2, 19, 23, 32–33, 74–75, 78–79, 83–84, 99, 102, 104–105, 118, 126–127, 169, 173, 177, 195, 200, 215, 220 schema, 26, 57–59, 61–62, 64–65, 75, 121, 125–126, 185, 200–201, 251, 263, 282 semantics, 2–3, 7–8, 15, 22, 27, 31–32, 69, 107, 127, 184, 201, 208 signifier (signified), 7, 28, 75, 208 society, 2–3, 6, 16–17, 20, 23–24, 40– 45, 59, 64, 74–81, 90–92, 95–104, 106, 117, 120–131, 134, 136, 138, 142, 148–149, 152–154, 156–157, 170, 200, 215, 250–253, 256

~ social groups, 2, 34, 43, 80, 89, 104, 114, 116, 120–121, 129, 135, 138, 148, 155, 163, 169, 215, 219, 259 stereotype, 35, 118 structure, 4–5, 10, 14, 18, 26, 30–33, 36, 48, 52, 54, 61–62, 66, 74, 77, 79, 81, 90, 94, 100, 103, 106–107, 115, 120, 123–125, 127, 130, 148, 152, 160–161, 163–167, 172–173, 176, 178, 183, 186–187, 196, 212, 217–221, 224, 226–228, 250–253, 258–259 system, 1–6, 9, 12, 16–19, 22–25, 27– 30, 32–33, 37–38, 40, 42–45, 49, 56– 57, 60, 62, 64–65, 74–81, 84, 86, 89108, 110, 113–114, 117–128, 131– 136, 138, 141, 162–164, 166, 168, 172–173, 176, 178, 183–184, 189– 191, 198–199, 206–207, 209, 213, 221–222, 225–227, 235, 240, 245– 246, 248, 251–261 taught (see also learned), 1, 37, 78, 86, 89, 90, 108, 109, 177, 200, 221 taxonomy, 4, 8, 12, 28–29, 147, 178– 179, 193