Handbook of Behaviorism

HANDBOO K OF BEHAVIORISM This Page Intentionally Left Blank HANDBOO K OF BEHAVIORIS M EDITED B Y WILLIAM O ' DONOH...

Author: William O' Donohue | Richard F. Kitchener (eds.)

187 downloads 1764 Views 24MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

HANDBOO K OF BEHAVIORISM

This Page Intentionally Left Blank

HANDBOO K OF BEHAVIORIS M

EDITED B Y WILLIAM O ' DONOHU E University of Nevada Reno, Nevada RICHARD KITCHENE R Colorado State Universit y Ft . Collins, Colorad o

ACADEMIC PRES S San Diego London Boston New York Sydney Tokyo Toronto

This book is printed on acid-free paper.

e

Copyright © 1999 by ACADEMIC PRES S All Rights Reserved. No part of this publication may be reproduced or transmitted in any form or by an y means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher . Academic Press a division of Harcourt Brace & Company

525 B Street, Suite 1900, San Diego, California 92101-4495, US A http ://www.apnet .com Academic Pres s 24-28 Oval Road, London NWI 7DX, U K http ://www.hbuk.co .uk/ap/ Library of Congress Catalog Card Number : 98-85620 International Standard Book Number : 0-12-524190-9 PRINTED IN THE UNITED STATES OF AMERIC A 98 99 00 01 02 03 MM 9 8 7 6 5 4 3 2 1

This book is dedicated to the loving memory of William Thomas O'Donohue, Sr.


CONTENT S

CONTRIBUTORS XVI I PREFACE XIX

1 INTRODUCTION : THE BEHAVIORISM S WILLIAM O ' DONOHUE AND RICHARD F. KITCHENE R

I . The Behaviorisms 1 A. Exegetical Problems 8 B. Emotional Reactions 9 C. Philosophy of Science, Scientific Theory, Methodology, Empirica l Findings, and Clinical Implications 1 0 D. Behaviorism and the Effectiveness of Practical Problem Solving 1 0 E. American Character 1 1 F. Relationship Between Philosophical and Psychological Behaviorisms 1 2 G. Quo Vadis? 1 2 References 13

2 WATSONIAN BEHAVIORIS M EDWARD K . MORRIS AND JAMES T. TOD D

I. Introduction 1 6 II. Intellectual, Cultural, and Individual Context 1 8 VII

VIII

CONTENT S

A. Intellectual Context: Philosophy, Science, and Psychology 1 8 B. Cultural and Individual Context 2 2 III . Before Watsonian Behaviorism 25 A. Watson 's Research Programs 2 6 B. Watson and Psychology in 1913 29 IV. Watsonian Behaviorism 30 A. Watson : 1913 -1920 3 0 B. The Behaviorist Manifesto 3 2 C. "Psychology as the Behaviorist Views It" 3 2 D. A Science of Behavior 45 . After Watsonian Behaviorism 4 9 V A. Watson : 1920—1936 49 B. Heredity and Environment 50 C. Applications to Child Rearing and Advertising 53 D. Watson: 1936—1958 56 VI . Conclusion 5 7 References 59

3 INTERBEHAVIORISM AND INTERBEHAVIORAL PSYCHOLOG Y LINDA J . HAYES AND DEBRA W . FREDERICKS

I . Historical Background 7 3 A. Biography 73 B. The Evolution of Organismic Psychology 7 6 II . Ziggurat of Science and Civilization 77 A. Cultural Matrix 7 7 B. Protopostulates of Scientific Philosophy 78 C. Metasystem of the Sciences 80 D. Postulates of the Individual Sciences 81 III . Postulational System of Interbehavioral Psychology 8 1 A. Interbehavioral Fields as Subject Matter 81 B. Factors Participating in Interbehavioral Fields 8 1 IV. Psychological Events 8 3 A. General Classes of Interbehavior 8 3 B. Varieties of Psychological Events 85 V. Interbehaviorism and Radical Behaviorism 8 6 A. Similarities 86 B. Differences 8 7 VI . Relation to Empirical Evidence 89

IX

CONTENTS

A. Kantor's Contributions to Empirical Research 9 0 B. Others 'Assessment of Kantor's Contributions 9 1 C. Sphere of Influence 9 1 D. Domains of Influence 91 VII. Current Status and Future Directions 9 2 VIII. Epilogue 9 3 References 93

4 EDWARD C . TOLMAN ' S PURPOSIVE BEHAVIORIS M NANCY K . INNI S

I . Purposive Behaviorism 9 9 A . A New Formula for Behaviorism 99 B Purposive Behavior in Amimals and Men 10 1 II . Operational Behaviorism 10 3 III . Performance Vectors 106 A. The Schematic Sowbug 10 6 B. Failure of the Schematic Sowbug 10 9 IV. From Rats and Sowbugs to Man 10 9 V. A Field Theory of Learning 11 0 A. Cognitive Maps in Rats and Men 11 1 B. Tolman's Map Analogy 11 2 VI . The Psychology of Social Learning 11 3 VII. Edward Tolman : Purposive Behaviorist 11 5 Acknowledgments 11 5 References 116

5 CLARK L . HULL ' S BEHAVIORIS M MICHAEL E . RASHOTTE AND ABRAM AMSE L

Introduction 12 0 Brief Biographical Sketch 12 0 I . Theoretical Style 122 A. Influence of Machines 122

X

CONTENT S

B. Plan for Developing a Behavior Theory 12 3 C. Perspective on Hull's Attempt to Make a Formal Behavior Theory 12 6 D. Physiology and Hull—Spence Theory 12 7 II. S —R Analysis of Cognitive Processes 129 A . Knowledge, Foresight, Directing Ideas, and Purpose 129 B . Gestalt Phenomena 136 C. Habit—Family Hierarchies (Networks of Associations) 136 D. Kenneth Spence's Analysis of Transposition 13 8 E. Application to Psychotherapy and Symbolic Processes 14 1 III. Conceptualizations of Reinforcement and Reward : Evolution of Hull' s 1943 Theory 142 A . "Reinforcement" in Hull's 1943 Theory 143 B. Difficulties Recognized by Hull in 1943 144 C. The Elliott— Crespi —Zeaman Experiment 145 D. Spence's Contributions to Reward—Incentive Theory 14 7 E. Amsel's Frustration Theory : A Conditioning Model of Effects of Rewards and Nonreward 15 0 Concluding Comments 15 3 Acknowledgments 154 References 154

6 RADICAL BEHAVIORISM : B .F. SKINNER ' S PHILOSOPH Y OF SCIENC E JON RINGE N

I . Radical Behaviorism as Philosophy of Science 16 0 II. Radical Behaviorism : The Basic Issues 16 1 III. Radical Behaviorism and Mentalism 16 2 A. Privacy 162 B. Intentionality 164 IV. Radical Behaviorism and Knowledge 170 A. "Thinking" "Impossible" "Thoughts"? 171 B. Naturalism and "Knowledge?" 175 V. Summary 176 References 17 7

CONTENTS

7

XI

EMPIRICAL BEHAVIORIS M SIDNEY W. BIJO U

I. II. III. IV. V. VI.

Freud and Psychoanalysis 18 0 Watson and Behaviorism 18 0 Lewin and Field Theory 18 1 Hull—Spence and Learning Theory 18 2 Skinner and ther Experimental Analysis of Behavior 18 4 Empirical Behaviorism : Radical Behaviorism an d Interbehaviorism 18 7 VII. The Influence of Empirical Behaviorism 18 9 VIII. Summary and Conclusion 19 0 References 190

8 TELEOLOGICAL BEHAVIORIS M HOWARD RACHLI N

I. Final Causes 196 A. The Relation of Final Causes to Their Effects 19 7 B. Sensation and Imagination in De Anima 198 II. Causation in Physics 200 III . Causation in Classical and Modern Psychology 20 1 IV. Causation in Skinnerian Psychology 20 3 V. Teleological Behaviorism 20 8 A. Teleological Behaviorism and Modern Philosophy of Psychology 21 0 B. Utility Functions as Final Causes 21 2 Acknowledgments 21 3 References 213

XII

9

CONTENT S

THEORETICAL BEHAVIORIS M

J. E.

R . STADDO N

I . Cognitive Psychology 21 9 A. Philosophical Objections to Cognitive Psychology 22 1 B. AI Objections to Cognitivism 223 II . Theoretical Behaviorism 224 A. The Model is the Behavior 22 4 B. Consciousness 22 9 III . Conclusion 23 8 Acknowledgments 23 9 References 239

10 BIOLOGICAL BEHAVIORIS M WILLIAM TIMBERLAK E

I. Introduction 24 4 A. The Waning Hegonomy of Traditional Behaviorism 24 4 B. Biological Behaviorism : An Inclusive Approach 24 5 C. Chapter Preview 24 5 II . A Brief History of the Study of Behavior 24 6 A. Observation and Anthropomorphism 24 6 B. Training and Implicit Causal Models 24 7 C. Experimental Tests of Causation 248 III . Strengths and Limitations of Manipulation-Centered Behaviorism 24 9 A. Strengths of the Manipulation-Centered Approach 25 0 B. Limitations of the Manipulation-Centered Approach 25 1 IV. An Animal-Centered Biological Behaviorism 25 3 A. Animal-Centered Versus Anthropomorphic Observation 254 B. Animal-Centered Manipulations 257 C. An Animal-Centered Approach to Traditional Experimentation 25 8 D. Constructing an Animal-Centered Causal System Model 26 1 V. The Feeding Behavior System in Rats : A Causal System Model 26 3 A. Structure of the Feeding System 26 4 B. Regulation of a Behavior System 26 7 C. Characteristics of Animal-Centered Causal System Models of Behavior 268

XII I

CONTENTS

VI . Biological Behaviorism and Nonperipheral Causation 268 A. Evolution and Development 269 B. Physiology 270 C. Intervening Concepts 271 D. Mental Life 273 VII . Summary 27 6 References 277

1 1 FUNCTIONAL CONTEXTUALISM : A PRAGMATIC PHILOSOPH Y FOR BEHAVIORAL SCIENC E ELIZABETH V . GIFFORD AND STEVEN C . HAYE S

I . Mechanism and Pragmatism : Basic Definitions 28 7 A. Mechanism 28 7 B. Historical Traditions in Behaviorism 28 7 C. The Core of Pragmatism 288 D. A Pragmatic Approach to Philosophy of Science 290 II. Philosophical Inconsistency in Skinner's Writing 29 1 A. Dewey's Functional Definition of the Stimulus and Response 292 B. The Evolution of the Concept of the Stimulus 29 3 III. Contextualistic Behaviorism 29 4 A. An Interactive, Holistic Unit of Analysis 29 4 B. Flexible Units 29 5 C. Dynamic Change versus Static Structure 29 6 D. Causality and Utility 297 E. Functional Epistemology 29 9 F Pragmatic Goals of Analysis 30 0 G. The Limits of James's Pragmatism 30 1 H. Summary 302 IV. Goals and Contextual Schools 30 3 A. Descriptive Contextualism 30 4 B. Strengths and Weaknesses 305 V. Functional Contextualism 30 6 A. Strengths 30 7 B. Scientific Theory 30 7 C. Behavior Analytic Rules for Action 30 8 D. Integrating Basic and Applied Research 309 E. Weakness 31 0 F Contextualism and Relativism 310

XIV

CONTENTS

G. Verbal and Nonverbal Justification 31 1 H. Summary : Descriptive and Functional Contextualism 31 3 VI. Implications of Functional Contextualism : Expanding the Analysis of Private Events 31 3 A. Public/Private versus Physical/Mental 31 4 B. The Scientific Analysis of Private Events 31 4 VII. Criticisms of Functional Contextualism 31 6 A. Contextualism 's Fidelity to its Assumptions and Promiscuity a t Other Levels 31 7 B. Recent Criticism within Behavior Analysis 31 8 C. Functional Contextualism or Pragmatism? 32 1 VIII . Conclusion 32 1 References 322

12 WITTGENSTEIN ' S BEHAVIORIS M DAVID BLOO R

I. II. III. IV. V. VI. VII. VIII. IX. X. XI. XII. XIII.

Introduction 32 9 Behaviorism as a Family-Resemblance Concept 33 0 The Case for Behaviorism 332 Sensations 33 5 Diaries and Manometers 33 9 The Behaviorist Reading 34 1 A Problem 343 The Institution of Use 344 The Institution of Pain 34 8 The Private Diary as an Institution 35 0 Assimilating Wittgenstein 35 3 Figure and Ground 35 6 Conclusion 35 8 References 359

13 RYLE ' S BEHAVIORIS M ULLIN T. PLAC E

I. Gilbert Ryle 36 2 II. OUR-Behaviorism versus OR-Behaviorism 362

XV

CONTENTS

III . Ryle's Debt to Wittgenstein 36 4 A. Philosophy as Linguistic Clarification 36 5 B. Logical Grammar 365 C. Ordinary Language 366 D. Dispositions 366 E. Aspect and Ontology 36 7 F Reasons and Causes 36 8 IV. What is in Wittgenstein, But Not in Ryle 36 9 A. Language Games 369 B. Linguistic Rules 369 V. Ryle's Innovations : The Failures 37 0 A. "The Ghost in the Machine" 370 B. Categories, Category Differences, and Category Mistakes 37 1 C. "Knowing How" and "Knowing That" 37 3 VI . Ryle's Innovations : The Achievements 374 A . Restricting Conceptual Analysis to On-Duty Sentences 374 B . The Method of Verification as a Clue to Meaning 376 C. The Hypothetical Analysis of Dispositional Statements 378 D . A Comprehensive Survey of Commonsense Psychology 379 E. Most, But Not All, Ordinary Psychological Concepts are Dispositional 379 VII . Ryle's Critics 380 A. Place (1954) 38 1 B. Geach (1957) 383 C. Medlin (1967) 386 D. Armstrong (1968) 391 E. Martin (1994) 393 VIII. Conclusion 395 References 397

14 LOGICAL BEHAVIORIS M RICHARD F. KITCHENE R

I. II. III. IV. V. VI. VII.

Introduction 39 9 The Varieties of Philosophical Behaviorism 40 0 Logical Behaviorism and Psychological Behaviorism 40 2 Logical Behaviorism and Semantic Behaviorism : Physicalism 40 3 Is Logical Behaviorism a Behaviorism at All? 40 5 The Semantics of Logical Behaviorism 40 7 Carnaps's "Naturalistic" Semantics 40 9

XV I

CONTENTS

VIII . Conclusion 41 5 References 416

i5 QUINE ' S BEHAVIORIS M ROGER E . GIBSO N

I . Introduction 41 9 II . Stage Setting 42 0 III . Quine's Behaviorism/Empiricism 42 2 IV. Behaviorism and Philosophy of Language 42 4 A. Meaning, Synonymy, and Analyticity 424 B. Language-Learning 42 7 C. Translation and Indeterminacy 42 9 V. Philosophy of Mind 43 2 VI . Conclusion 434 References 43 5 SUBJECT INDEX

437

CONTRIBUTOR S

Numbers in parentheses indicate the pages on which the authors ' contributions begin.

Abram Amsel (119) University of Texas at Austin, Austin, Texas 7871 2 Sidney W. Bijou (179) University of Nevada-Reno, Reno, Nevada 8955 9 David Bloor (329) University of Edinburgh, Edinburgh, United Kingdo m EH8 9JT William O'Donohue (1) University of Nevada, Reno, Nevada 8955 9 Debra W. Fredericks (71) University of Nevada, Reno, Nevada 89557-006 2 Roger E. Gibson (419) Washington University, St. Louis, Missouri 63130-4899 Elizabeth V. Gifford (285) University of Nevada, Reno, Nevada 89557-000 1 Linda J . Hayes (71) University of Nevada, Reno, Nevada 89557-006 2 Steven C . Hayes (285) University of Nevada, Reno, Nevada 89557-000 1 Nancy K . Innis (97) Department of Psychology, University of Western Ontario , London, Canada N6A 5C 2 Richard F. Kitchener (1, 399) Colorado State University, Ft . Collins, Colorado 8052 3 Edward K. Morris (15) Department of Human Development and Family Life, Dole Human Development Center, University of Kansas, Lawrence, Kansa s 66045-213 3 Ullin T. Place (361) Willowtree Cottage, Boltby, Thirsk, North Yorkshire , England Y07 2DY Howard Rachlin (195) SUNY at Stony Brook, Stony Brook, NY 11794-000 1 XV I I

XVIII

CONTRIBUTOR S

Michael E . Rashotte (119) Florida State University, Tallahassee, Florida 32306 273 0 Jon Ringen (159) University of Iowa, Iowa City, Iowa 5224 0 J. E . R. Staddon (217) Duke University, Durham, NC 27706-770 6 William Timberlake (243) Indiana University, Bloomington, Indiana 4740 5 James T. Todd (15) Department of Psychology, Eastern Michigan University , Ypsilanti, Michigan 48197

PREFAC E

Why publish a book on behaviorism? Behaviorism is dead, killed by the cognitive revolution. So why resurrect the corpse? These are good questions and reflect the current status of behaviorism as an approach to the study of the mind . However, we believe that, to paraphrase Mark Twain, "reports of the demise o f behaviorism are greatly exaggerated." But even if this is not true and behavior ism is dead, we believe that as the present century draws to a close, it is important to consider the status of a movement that, arguably, was the dominant approach to the mind in the 20th century . Finally, as our Introduction points out, w e believe it is important to realize that there are a wide variety of behaviorisms an d not a single, monolithic approach . On the received view, psychological behaviorism was killed by the cognitiv e revolution, which occurred in the '60's . The centerpiece here, no doubt, wa s Chomsky's critique of Skinner's Verbal Behavior. But there were other threads in the tapestry of cognitivism that collectively resulted in behaviorism's loss o f hegemony including not only new approaches to the study of the mind (e .g ., arti ficial intelligence) but also philosophical critiques, and degenerating empirical research programs . But what were the ingredients common to behaviorism tha t resulted in its demise? To which version of behaviorism were these critiques directed? And were these targets adequately characterized by their critics or wer e they just straw men ? Rarely have these questions been probed . In particular, rarely have individuals set out the commonalties of all the versions of behaviorism, for in fact there is not just one behaviorism but several behaviorisms and (to borrow anothe r metaphor) there are no common, essential features but rather only a family re semblance between the various members of the behaviorist family . Consider, fo r XIX

xx

PREFAC E

example the case of Edward Chase Tolman, who saw himself as a staunch behaviorist, but who many cognitivists today cite as a ligitimate member of the cognitivist family . Where should we put Tolman? Arguably, he was as much of a behaviorist as a cognitivist . But if so, then how do we distinguish behaviorism from cognitivism? It won't do to say (as one cognitivist famously did), "cognitivists are interested in what is going on inside the skin of the organism and behaviorists aren't" since this is also true of paradigm-cases of behaviorism, e .g. , Hull and Skinner. In short, it is our view that no one has yet set out the feature s distinguishing behaviorism from cognitivism . But before that can be done, it i s necessary to survey the various types and varieties of behaviorism and this w e have attemped to do by including a cross section of the most important varietie s of the behaviorisms . In reading these slections, the reader (hopefully) will have a sense that behaviorism is not dead but instead continues to foster a variety of (what Lakato s called) progressive research programs . They were progressive both in the sense of being empirically progressive research programs generating new and interesting empirical data and conceptually progressive research programs—modifying their theoretical views in the face of criticisms in such a way that one ca n call these modifications "logically and conceptually progressive" . That the death of behaviorism has been greatly exaggerated can only be seen in another way and this is the continuing one is tempted to say perennia l appeal of behaviorism as an approach to the study of the mind. One has, for ex ample, not only the increasing influence of a Wittgensteinian kind of behaviorism in a variety of contexts (e .g., prototype theory, social constructivism), on e also has signs that the newer cognitivists connectionists, roboticists, and dynamic systems theorists are incorporating elements of behaviorism into their respective accounts of cognition . Indeed, there are reasons to think that as th e classical symbolic approach to cognition wanes, there has been a corresponding importation of several features of behaviorism . Although the case of robotics i s the clearest example of this, one can also find this present in the other non-classical approaches of connectionism and DST . (Time constraints prevented us from including a manuscript on this .) Clearly, this raises this question (onc e again) of what features of behaviorism can be combined with other approaches to the study of the mind . As the year 2000 approaches, we owe it to ourselves to look back at the 20 th century and to survey its accomplishments and failures, its strengths and weaknesses . What seems characteristic of this century's predominant approach to th e study of the mind, both in philosophy and in psychology, was a sympathy to wards behaviorism. Anyone who has studied the history of 20th century psychology would immediately run head on into this fact that the hegemony of behaviorism lasted from the '20s until the '60's (and we would argue much longer) . What were the strengths of this psychological approach and what were its weaknesses? What elements can be retained as having lasting value and what elements should be rejected as being unfruitful? Although the history of 20 th cen-

PREFACE

XX I

tury philosophy of mind does not reveal a corresponding hegemony of philosophical behaviorism, as least to the same degree, nevertheless such a position deserves primacy of place . In canvassing the different philosophical approache s to the mind, behaviorism deserves a special importance not only because of its historical importance, say, between 1920 to 1960, but also because, unlike psychology, philosophical approaches tend to less constrained by the factors o f time . There is a sense in which philosophical behaviorism is as viable an optio n today as it was in 1949 (contrary to what most current textbooks claim) . But again, even if philosophical behaviorism is only of historical interest, it may b e of value to winnow out those strands that are of more positive value from thos e that are not . Once again, this requires a renewed critical sensitivity to the actual views of behaviorists and to historical details— something that has been lackin g in many accounts . These, then, were some of the considerations leading the editors to put together a selection of readings on behaviorism . We have not, of course, covere d every variety of behaviorisms, but we do believe that we have covered most o f them . Once again, we wish to thank our families for their endless patience and understanding, our many contributors, and to our publisher for having the re solve to publish such a book .


INTRODUCTION : TH E BEHAVIORISM S

WILLIAM O ' DONOHU E University of Nevada, Reno

RICHARD F. KITCHENE R Colorado State University

I . The Behaviorisms A. Exegetical Problems B. Emotional Reaction s C. Philosophy of Science, Scientific Theory, Methodology, Empiri cal Findings, and Clinical Implication s D. Behaviorism and the Effectiveness of Practical Problem Solvin g E. American Characte r F Relationship between Philosophical and Psychological Behav iorisms G. Quo Vadis? I . THE BEHAVIORISM S It is an interesting fact that behaviorism is usually referred to in the singular . It is commonplace to see statements such as "Behaviorism is . . . " as if ther e were a single, unified conception called Behaviorism and all the various behav iorists agreed about the conceptual foundations of behaviorism . This construal o f Handbook of Behaviorism

'

I

Copyright © 1999 by Academic Press . All rights of reproduction in any form reserved .

2

O ' DONOHUE AND KITCHENE R

a single (or unified) behaviorism is perplexing, as two facts should be readily apparent to all but the most casual reader of this literature : 1) Historically there have been a number of influential behaviorists each with his own characteristi c school; and 2) There have been (and still are) substantial disagreements amon g these behaviorists concerning a variety of questions . In psychology, for example , Watson, Kantor, Skinner, Hull, Spence, and Tolman had significant disagreements on a number of different kinds of issues ; and in philosophy, the same can be said for Carnap, Quine, Wittgenstein, and Ryle . The divergence of opinions become even more apparent when one compares a philosophical behavioris t such as Quine with a psychological behaviorist such as Tolman ; here one sees very different kinds of claims . In an important sense, disagreement implies taking different positions on th e same issue . But in the usual case behavorial philosophers and behavorial psychologists disagree because they are addressing different issues, or are addressing the same issue at a different level of abstraction (e .g., philosophers use a meta-language to talk about issues psychologists address in their object language) . Why should we, therefore, assume that all the important claims of thes e different scholars are the same or sufficiently similar to be felicitously subsume d under the single label behaviorism? The chapters that follow depict the nature o f these disagreements . A reading of this book should make it abundantly clear that there are man y varieties of behaviorism at least fifteen and hence we should refer to thes e accounts as behaviorisms rather than behaviorism . This gives rise to the following questions : a) What do these forms of behaviorism have in common, if any thing? Does the use of the singular term behaviorism refer to some set of genera l attributes all these different subtypes share? ; b) What common properties make each one of these behaviorisms a legitimate form of behaviorism as opposed to something else? ; c) What are their differentia what differentiates the behaviorism of Hull, for example, from the behaviorism of Skinner ? There are several interesting approaches one can take in addressing thes e questions . First, one can employ the classic Greek definitional strategy of genus et differentia. This involves finding a property or properties all of these share and then finding a property or properties that distinguish behaviorism from all th e other paradigms . One then can go even further and find the differentia distinguishing each of the individual species of behaviorism from other forms . How ever, it is not clear what the best candidate for the genus might be ; Zuriff (1985 ) in his excellent Behaviorism: A Conceptual Reconstruction has asserted tha t there are no necessary and sufficient definitional properties for behaviorism . A second possibility is to use prototype theory (Rosch, 1978) . For example , one could take Watson's behaviorism perhaps because of its historical priorit y as well as the familiarity of some of its features as the prototype and clai m that the other kinds of behaviorism fit this prototype (to varying degrees) because of their similarity on key features . Skinner's behaviorism then is similar t o this prototype because of his emphasis on learning and the environment, his distrust of the mental, his connection with evolutionary theory, and so on . This still

1 INTRODUCTION : THE BEHAVIORISMS

3

leaves open the question of what is the prototype and what are the dimension s upon which any similar judgements are made . Should we take, for example, Watsonian behaviorism as a paradigm-case of behaviorism ? Another related approach would be to invoke Wittgenstein's (1958) concep t of family resemblance and argue that although these share no necessary and sufficient properties, they do share a "complicated network of similarities overlap ping and criss-crossing : sometimes overall similarities, sometimes similarities o f detail" . Wittgenstein (1958) states : Consider for example the proceedings that we call `games .' I mean board-games, cardgames, ball-games, Olympic games, and so on . What is common to them all?—Don 't say: "There must be something common, or they would not be called `games ' "—but look and see whether there is anything common to all. —For if you look at them you wil l not see something that is common to all, but similarities, relationships, and a whole serie s of them at that . . . . Are they all "amusing"? Compare chess with noughts and crosses . Or is there always winning and losing, or competition between players? Think of patience . In ball games there is winning and losing ; but when a child throws his ball at the wall and catches it again, this feature has disappeared . Look at the parts played by skil l and luck; and at the difference between skill in chess and skill in tennis. Think now of games like ring-aring-a-roses ; here is the element of amusement, but how many othe r characteristic features have disappeared! And we can go through the many, man y other groups of games in the same way ; can see how similarities crop up and disappear . And the result of this examination is : we see a complicated network of similarities over lapping and criss-crossing ; sometimes overall similarities, sometimes similarities of detai l (p . 31-32) .

On this approach, the task would be to explicate this "complicated network o f similarities" . Zuriff has suggested some interesting candidates in the web o f behaviorism : 1. Psychology is a branch of natural science . Thus, there are continuities in the kind of research methods, theories, questions, and results between science s such as biology, physics, and psychology . This would seem to eliminate, how ever, other versions of behaviorism that do not emulate the natural sciences, fo r example, the various versions of Action Theory and contemporary social constructivism that claim that psychology should not be construed as a natural science but rather as a cultural or hermeneutic science . 2. Psychological evidence should be objective evidence, obtained from observation and experiments conducted on public objects . Various arguments are given to delimit what exactly constitutes objective evidence, for example, it must satisfy the criterion of being intersubjectively verifiable and replicable by others . 3. Introspection of the private phenomenal world is problematic and shoul d be avoided . 4. Psychology's data should concern molar behavior, not physiological processes . Thus, there is an autonomous science of behavior independent o f physiology . 5. Mental terms such as volition, intention, and purpose are problematic an d either proscribed or translated into a nonmentalistic language .

4


6. Theoretical concepts ought to be properly tied to behavioral data . Althoug h there are disagreements among behaviorists concerning what constitutes legitimate theoretical entities, in general, behaviorists are minimalists and are reluctant to admit constructs such as central states, intervening variables, hypothetica l constructs, and so on, if they are more than summary terms for empirical observations . In the terminology of MacCorquodale and Meehl (1948) the only theoretical terms allowed would be intervening variables, not hypothetical constructs . 7. Behaviorism is associated with learning research, which attempts to relat e environmental events to responses . 8. The adaptation of the organism to its environment is often a central concern of behaviorists . Thus, evolutionary theory is used to address long-term "ultimate" questions, while learning research is used to address more proximat e questions . The interrelationships of evolution and learning are thus an importan t issue . 9. Notions of thinking and cognitive notions in general divert attention awa y from behavior and the environment and their relationships . Hence, the intensionality of cognitive constructs is not scientifically acceptable and must be avoided . 10. Behaviorism rejects putative internal causes particularly human agenc y and instead focuses on external, environmental relations with behavior . All behavior is assumed to be lawfully related to the environment . This externalism heightens the objectivity of the science of human behavior by placing causes i n the observable environment. 11. It is possible and useful to translate mentalistic sentences into sentence s that appeal only to behavior and other observable events . 12. Questions about language and in particular about meaning can be profitably researched by studying verbal behavior, particularly how humans lear n language . 13. Philosophical questions, such as the nature of knowledge, should be naturalized . Epistemological questions can be answered by indigenous psychologica l accounts of learning . The scientific approach should be thoroughgoing and applied to epistemological questions regarding science itself . Not every form of behaviorism would agree with each of these claims, wit h the divergence between psychological approaches and philosophical approache s being the most noticeable . Most psychological behaviorisms would endors e many of these claims, in particular, 1, 2, 3, 4, 6, 7, and 10 ; hence, there would be key similarities between their approaches . But philosophical approaches to behaviorism would accept fewer of these principles, for example, 5, 6, 8, 9, 11, 12 , 13 perhaps would be controversial . Furthermore, within the psychological behaviorist camp there would be disagreements over many of these principles, fo r example, Tolman would not be sympathetic to 9 whereas Skinner would ; Hul l would not be as sympathetic to 10 but Tolman would be, and so on . How does one evaluate the various planks of the behaviorist program? Man y of these claims do not appear to be empirical, factual claims, but rather prescrip-


5

tions regarding how behavior ought to be studied by psychologists to order t o make optimal progress . To the extent that these claims have empirical implications they are certainly underdetermined by current evidence . Thus, these claims constitute a bet, a bet regarding the best decisions to make at a number of choic e points for producing the most successful explanation/understanding/prediction / control of human behavior . Since the empirical results are not yet completely in , any bet is to be vindicated by its current status and future success . At present , one can mainly give conceptual–logical arguments and reasons for believing i n the behaviorist program as well as appealing to whatever empirical evidence i s available . Because (modern) behaviorism is approximately a century old, one can begin to evaluate the fruits of this behavioristic bet . This is a complicated undertakin g and necessitates resolving three other questions : 1) What are the proper outcom e criteria to evaluate the fruitfulness of this bet? 2) Since bets are essentially comparative—there must be alternatives that are seen as less fruitful these claim s imply that making these moves will prove better than making alternative moves . Hence, another question is : What are the other candidates? The final questio n raised is : 3) What is the actual value of these outcome variables for all these candidates? We will discuss later one relevant consideration in evaluating the out comes of the bet on the behaviorisms when we describe the relationship betwee n these behaviorisms and the resolution of certain practical problems . One implication of this "family resemblance" approach to the different type s of behaviorism is that at times there will be some family resemblances betwee n behaviorism and nonbehavioral approaches : Freud believed that all behavio r was determined by past events and the current situation; cognitive psychologists emphasize experimental studies and rely on overt behavior as their de pendent variable ; physiological psychologists want their data to be "objective " and emphasize the prediction and control of behavior ; and sociobiologists emphasize the evolutionary influences on behavior (note the overlap with Skinner' s construct of the contingencies of survival) . Thus, this network of similaritie s among the behaviorisms connects them to other approaches toward the study o f behavior. Finally, another possibility for characterizing behaviorism arises from within certain forms of behaviorism . One can take a behavioral account of verbal behavior and view this as a question about human behavior that can be analyze d experimentally to determine if and what relationships exist between the occurrence of the response behaviorism and certain environmental events . Skinner stated, "What happens when a man speaks or responds to speech is clearly a question about human behavior and hence a question to be answered with th e concepts and techniques of psychology as an experimental science of behavior " (Skinner, 1957, p . 5) . Thus, in Skinner's account, behaviorism can be construe d as a tact (Skinner, 1957) and the variables of which this verbal response is a function can be experimentally explored . (One must remember that there ar e other behavioral accounts of language, and therefore another choice point concerns which of these one will use .)

6


Whichever of these strategies one uses, it is certainly the case that these different forms of behaviorism complicate definitional matters considerably . But it is not only these definitional issues that are complicated . The following questions also become more complex : What were the historical influences in the rise of these behaviorisms ? Historically, how did these forms of behaviorism relate to one another ? Conceptually, how do the behaviorisms relate to one another? What is the current status of these behaviorisms are they dead, dying o r thriving ? 5. How does one appraise the value (and future prospects) of these behaviorisms ? 1. 2. 3. 4.

These questions become exponentially complex given the many varieties of behaviorism, but they can hardly be ignored . Behaviorism dominated much of twentieth century American psychology, either as the predominate point of view, or, in the later part of the century, as the view that was being reacted against, for example, in the cognitive revolutio n (Baars, 1986) . Watson published Animal Education in 1903, "Psychology as th e Behaviorist Views It" in 1913, and the Psychological Care of the Infant in 1928 . Guthrie's Psychology of Learning was published in 1935 . Hull's Principles of Behavior was published in 1943 . Skinner published The Behavior of Organisms in 1938, Walden Two in 1948, Verbal Behavior in 1957, Contingencies of Reinforcement in 1969, and About Behaviorism in 1974 . Students and followers o f these behaviorists as well as contemporary behaviorists such as Bijou, Hayes , Rachlin, Staddon, Timberlake, and Quine published and continued to publis h during the last three decades of the century. But what about behavioristic approaches prior to Watson? Should Hobbes and the French philosophes (Cabanis , de la Mettrie) be characterized as behaviorists? What about those philosopher s who stressed the crucial role of practical reason (as opposed to theoretical reason) and hence laid primacy on praxis, for example, Aristotle, Hegel, Marx , Dewey, James, Mead, and so on? What about those philosophers of Action Theory, popular in the 1960s (Hamlyn, 1958 ; Melden, 1961 ; Peters, 1958), most of whom have denied they were behaviorists? Certainly objections can be made t o terming these philosophical views behaviorism but is this because individual s tend to associate behaviorism with certain narrow interpretations, for example , that of Watson? If one interprets behaviorism widely enough, these individuals could be included . Once again, there appears to be a wide variety of behaviorisms . Clearly, if there are several types of behaviorisms (behavioristic psychologies), there are also several types of philosophical behaviorisms . For example, i n the twentieth century, we have seen very different versions of behaviorism in th e work of the logical positivists (Carnap, Hempel, Feigl, and Bergmann), Gilber t Ryle (1949), Ludwig Wittgenstein (1958), and W. V. 0 Quine (1960) . To date, there has been no definitive history of the behaviorisms and if such a


7

history were ever to be written, one would first have to decide what criteria t o employ in selecting the various candidates . The family resemblance (or prototype) model suggests one will have to include many more behaviorisms than w e have selected and hence push the history of behaviorists back further than th e twentieth century. Clearly, the historical questions about behaviorism cannot be separated fro m conceptual questions of their relationship . But what about the current standing of the behaviorisms? Is not behaviorism dead? Was it not killed by the cognitive revolution ? It is a myth that the cognitive revolution killed behaviorism (O' Donohue & Naugle, 1997) . Behaviorism is still very much alive (although its hegemony ha s been eclipsed by the cognitive revolution) . This book contains chapters by Sidney Bijou on his empirical behaviorism, Howard Rachlin on his teleological behaviorism, John Staddon on his theoretical behaviorism, William Timberlake o n his biological behaviorism, and Steven Hayes on his contextual behaviorism . Quine's philosophical behaviorism is still influential in philosophy, as i s Wittgenstein's aberrant kind of behaviorism . Ryle's behaviorism together with Wittgenstein ' s have provided some of the material that critics of the representational theory of mind (e .g ., Coulter, 1983) have used against certain forms o f contemporary cognitive psychology . In fact, there are indications of a new rapprochement between behaviorism and contemporary cognitive psychology sinc e the emergence of connectionism, robotics, and dynamic systems theory as viabl e alternatives to the classical symbolic approach to the mind . Skinnerian behaviorism has been the most influential form of psychologica l behaviorism in this century at least in the second half of this century and i t continues to be influential even after the cognitive revolution . There is a scholarly association of people working in his tradition (Association for Behavio r Analysis) currently with 2,600 members ; there are approximately two dozen scholarly journals that either publish basic operant research (e.g ., Journal of the Experimental Analysis of Behavior), or publish clinically relevant research base d on Skinner's work (e .g., Journal of Applied Behavior Analysis) or publish conceptual papers relating to radical behaviorism (Behavior Analyst and Behavio r and Philosophy) . These journals continue to be read, cited, and most importantly, continue to publish new findings . Wyatt, Hawkins, and Davis (1986 ) found that these journals publish over 500 articles annually and have ove r 20,000 subscribers . These are but a few examples because there are other organizations and other ways that Skinner's work is continuing to have an influence . Thus, the legacy of B .F. Skinner is currently alive and well . (We could also cite the 1984 issue of the influential journal, Brain and Behavioral Sciences, devoted to the work of Skinner .) These rumors and false attributions may be hurting it s future . Behaviorism is still alive and developing (as the behaviorisms of Hayes , Timberlake, Rachlin, Staddon, and Quine attest) . Some forms of psychological behaviorism have been alleged to be dead be cause of their purported dependence on outmoded philosophical systems . One of

8

'

O DONOHUE AND KITCHENE R

the most common arguments of this type runs as follows : Logical positivism i s dead. Behaviorism was the extrapolation of logical positivism to questions abou t human behavior. Therefore, behaviorism is dead . Smith (1986) has shown that this view of behaviorism's philosophical relations comprises three distinct theses : 1) the importation thesis, that major behaviorists imported their philosophical and methodological beliefs from logical positivism ; 2) the subordination thesis, that the behaviorists subordinated their psychological views to these prio r philosophical commitments ; and 3) the thesis of linked fates, according to whic h the fate of behaviorism is thus linked to the fate of logical positivism . Smith examined the validity of these theses using the major neobehaviorists Tolman , Hull, and Skinner and found each to be false . If this is correct, then the demise of logical positivism did not carry with it the demise of behaviorism . Thus far, we have sought to argue three points : 1) There are several behaviorisms, both philosophical and psychological, not a single one ; 2) This plurality makes already difficult questions concerning the history and conceptual character of behaviorism even more difficult to answer ; and 3) Any claims regarding the death of behaviorism are seen as premature . We will now turn our attentio n to several other matters : 1) the exegetical problems that have plagued the behaviorisms ; 2) the strong emotional reactions to the behaviorisms ; 3) the multi-level systematicity of the behaviorisms ; 4) the association of the behaviorisms with effective responses to practical problems ; and 5) the American character of th e behaviorisms . We realize that these are somewhat unconventional topics, but w e think they need to be highlighted to properly understand the role of the behaviorisms in twentieth century psychology . A . EXEGETICAL PROBLEMS

The Oxford English Dictionary defines exegesis as "to interpret . . . exposition (of a sentence, word, etc .)" . Any scholar's writings or remarks can be accurately or inaccurately characterized . Watson's remark that he could take any infant and condition them to become a member of any occupation, is often cited a s an example of the excessive environmentalism of behaviorism, or worse, the extreme ideological nature of behaviorism . However, unfortunately what is often missing is that Watson in the next sentence admits that this is an exaggeration . Careful interpretation would necessitate including this important sentence . Todd and Morris (1992) have done a nice job of identifying many misreading of Skinner that have occurred in psychology textbooks and the popular media. MacCorquodale (1970) has done an excellent job of depicting the exegetical errors in Chomsky ' s (1959) review of Skinner's Verbal Behavior. Moreover, behaviorists are even personally attacked ; many rumors circulated about Skinner' s treatment of his children, for example (O'Donohue, 1998) . Similar points could be made about the standard interpretation of the logical (or analytical) behaviorism of the logical positivists . There appears to have been several different interpretations of what philosophical behaviorism meant to thinkers such as Carnap,

I INTRODUCTION : THE BEHAVIORISMS

9

Neurath, Feigl, and Hempel . Standard interpretation and textbook accounts and criticisms gloss over many of these important differences, assimilating them al l to a single model, which none of them actually held, and then showing such a view to be untenable . Thus, there have been many caricatures of the behaviorisms . Much ink has been spilt constructing straw men and then showing the problematical nature o f them . This can be a fairly effective rhetorical strategy. It's easy to dismiss some thing that is patently ridiculous, for example, by claiming that behaviorists say that people don't think . However, little intellectual progress is made under these circumstances . We would encourage the reader to pay careful attention to characterizations of the behaviorisms, particularly those found in secondary sources . Good scholarship demands a careful reading and a faithful depiction of claims . We hope that faithful characterizations of behaviorism increase in the future . B. EMOTIONAL REACTIONS

Another striking feature of the behaviorisms is that they tend to invoke strong emotional reactions, usually negative ones . Part of this comes from the sensitiv e nature of the subject matter human nature and human behavior. However, compared to other approaches to the study of human behavior and its analysis , behaviorism still elicits stronge reactions . We conjecture that some of this is due to the earlier point that behaviorism has often been mischaracterized in such a way to deliberately invoke negative re actions . However, behaviorists also assume part of the responsibility here . A number of prominent behaviorists were superlative rhetoricians . Here, one is re minded of the logical positivists' rhetoric about the meaningless of metaphysics . Ryle's "ghost in the machine" was a wonderful polemical phrase . Watson's manifesto contains persuasive, strident, almost scientistic prose . Skinner's Conceptual Nervous System ridicules certain theoretical moves . But above all, Skinner's book Beyond Freedom and Dignity has been a red flag . This book boldly states the implications of determinism and therefore challenges fundamental notions that are cherished by many in Western Society. Similarly, Skinner's Walden Two presents a radical notion of what he sees as a more reasonable society one that is very different from the way society is currently structured. Behaviorists have often addressed sensitive topics in iconoclastic and emotionally charged ways . But even when behaviorism deals with more mundane topics, it is still fairly radical . O'Donohue, Callahan & Ruchstuhl (1998) have suggested that the behaviorisms approach radically conflicts with folk psychology the informal notions of human behavior and its proper study held by the common person . Behaviorism has not only positive doctrines, but also negative ones critiques o f many received views . Some negative reactions to behaviorism are probably du e to its criticisms of these popular positions . Thus, behaviorism is seen as odd and disturbing .

10


Moreover, the effectiveness of some of the applied aspects of behaviorism i s frightening to some . When attempts to control or influence the behavior of others appear ineffective or weak, the underlying paradigm is seen as less threatening . However, more powerful technologies of behavior that actually influenc e people and change behavior raise an even larger specter because it elicits questions of whether these means will be used for good ends or not and whether the y will be used properly, for example, when individuals are properly informed an d can exert reasonable countercontrol . Finally, some have embraced behaviorism as a way of life or (some migh t say) even as a religion . The fervor and zealotry of some proponents of the behaviorisms have been of concern to others . Behaviorism at times has been see n by some as a finished product that provides all sorts of answers to all sorts o f questions . Some behaviorists have asserted their position with a smugness an d condescension others find off-putting . Thus, at times the dogmatic way som e have embraced behaviorism has aroused strong reactions in others . C. PHILOSOPHY OF SCIENCE, SCIENTIFIC THEORY, METHODOLOGY, EMPIRICAL FINDINGS, AND CLINICAL IMPLICATION S

Behaviorism is most properly considered a philosophy of the science of human behavior. It is not the science itself, but rather the meta-position in whic h basic questions about what is the proper subject matter of psychology and ho w this subject matter should be properly studied are raised (Skinner, 1974 ; Zuriff, 1985) . But this philosophy of behaviorism, in turn, leads to certain lower-level theo-

ries, analyses, and conceptual models . Learning theory and evolutionary theory , for example, have played an important part in behaviorism . If one were to strip these away, one would have something not easily recognized as behaviorism . Certain empirical regularities and psychological phenomena also form part o f behaviorism ; for example, habituation, classical conditioning, and operant conditioning . Next, certain characteristic research methodologies are part of the underlying conceptual network : use of animal preparations, certain characteristic dependent variables (learning curves), and so on . Finally, there is an applied aspect to certain kinds of behaviorism : behavior modification, child raising techniques, educational technologies, and even societal restructuring have all bee n proposed as implications of the more basic behavioral research model . The behaviorisms have often been multi-layered and more encompassing than alternative approaches to the study of human behavior. D. BEHAVIORISM AND THE EFFECTIVENESS OF PRACTICA L PROBLEM SOLVING

One branch on the behaviorisms' trunk involves behavioral interventions for a variety of applied problems . These include : behavior modification, behavio r therapy, applied behavior analysis, behavioral education, and behavioral medi-

I INTRODUCTION : THE BEHAVIORISMS

1 1

cine . Behaviorisms ' applied efforts have been very influential in the developmen t of effective strategies for a number of applied problems . For example, a recen t and very influential report written by a theoretically eclectic committee organized by the American Psychological Association identified which intervention s had good evidence that they worked . Of the twenty-seven techniques, over twenty behavior therapy techniques were listed as "empirically validated" (Task Force on Promotion and Dissemination of Psychological Procedures, 1995) . Many of these therapies are heavily influenced by Skinner's work . As a case in point, the application of Skinner's work to the behavioral deficits of mentally retarded children and adults has resulted in such a significant increase in skills for these individuals that many who were once institutionalized are now living independently (see for example Rush, 1995) . Certainly other psychologists such a s Freud also have been influential in developing therapy techniques . But, the difference between Freud and Skinner is that the work of Skinner and his student s has resulted in interventions that have been clearly shown to actually help peopl e regarding a significant number of their problems . Some behaviorists would see in these positive outcome data a pragmatic vindication of their meta-scientific views . This is some of the "cash value" of thei r more abstract commitments . Thus, an unique aspect of twentieth-century behaviorism is that their associated applied efforts have arguably been more successfu l than their competitors . E . AMERICAN CHARACTER

All the psychological behaviorists in this book are American . Among the philosophical behaviorists, Quine is American, Ryle was English, Carnap an d Wittgenstein were German-Austrian, with distinguished academic careers i n America and Great Britain . Psychological behaviorism (and, to a lesser extent , philosophical behaviorism) has a distinct American character . Psychological behaviorism was most influential in American psychology, had moderate influence in other English speaking countries, was much less influence on the continent and in Asia (There are some exceptions to this generalization in Central an d South America) . Likewise, philosophical behaviorisms were most influential i n Anglo-Saxon countries . (Although logical positivism was briefly influential i n Austria, Germany, and Poland) . Why is this the case? First, the United States and Great Britain had the positive economic conditions that would allow citizens to be able to pursue psychology and philosophy . A country needs relative affluence to be able to support re searchers, teachers, and students in pursuing this avenue of questions . However, beyond these economic prerequisites, certain American values and historica l contingencies have contributed to behaviorisms' American character and American residence . These include : the American view of the plasticity of the individual; its optimism and utopian world view ; its rural, agrarian character, which involves the necessity of understanding, handling, and even training animals ; its

12


value placed on mechanical skills (necessary in conducting animal research) ; its capitalism ; its pragmatism (emphasizing the cash value of ideas) ; its anti-intellectualism (as opposed to more scholastic paradigms such as psychoanalysis) ; its emphasis upon simple, direct communication . However, this is not to say that all aspects of behaviorism were agreeable to the American values . Certainly behaviorism's determinism is an exception. However, some of these components of th e American culture and some of the corresponding characteristics of the behaviorisms make for a more comfortable relationship . F. RELATIONSHIP BETWEEN PHILOSOPHICA L AND PSYCHOLOGICAL BEHAVIORISM S

The historical and conceptual relationships between psychological and philosophical behaviorisms are complex . Smith (1986) has shown that one should no t assume that because there are certain family resemblances between some o f these views, that one actually influenced the other . To cite just one example, naturalized epistemology is an area that has gained increasing attention in recent years and shows the fruitful convergence of philosophical and psychological behaviorisms . Quine and Skinner show a remarkable agreement on this point, al though it is not clear how much either influenced the other, and (quite surprisingly) so do many of the views of Carnap (whose influence on Quine and vic e versa is well known) . According to one interpretation of naturalistic epistemology, questions concerning knowledge are most fruitfully studied in the psychologist' s learning laboratory (although Skinner's or Watson's account of language bears little resemblance to Carnap's and Quine's) . Other philosophical behaviorists do not share this belief in naturalistic epistemology. For example, Ryle would clearly reject it and would hold on to a more traditional, philosophical interpretation of epistemology. In fact, Ryle's view o f the value and promise of psychology (especially behavioristic psychology) wa s notoriously rather bleak . Wittgenstein's philosophical views concerning forms o f life and one's enculturation in these collective modes lends itself to a naturalisti c interpretation —a social–psychological interpretation but his widely cite d comments about the poverty of psychology does not speak well for the possibility of a collaboration between philosophy and psychology . G . QUO VADIS ?

As the twentieth century comes to a close, it is a propos to reflect upon and critically examine one of the most influential philosophical and psychological schools of the twentieth century behaviorism . We have had over eight decade s to see the empirical and conceptual fruits of such an approach . We have learne d that certain aspects of some of these behaviorisms do not have the resources t o handle certain conceptual issues ; we have learned that several empirical components of these approaches are in error and/or in need of revision . But other as-

13

i INTRODUCTION : THE BEHAVIORISMS

pects have proved their mettle, showing themselves to be of considerable valu e in studying and understanding humans and their actions . We are tempted to say that, in Lakatos' (1978) term, behaviorism is a progressive research program, not a degenerative one . That, of course, is a historical value judgment with whic h others would find themselves in disagreement . Nevertheless, we have argued tha t the demise of behaviorism has been greatly exaggerated and based upon misreading and misunderstanding . What behaviorism in the twenty-first century will look like is uncertain, bu t based on its past track record, we believe it will continue to be a viable way o f understanding our world . At the same time, however, we are confident that it will evolve and develop into a set of behaviorisms quite different from those discussed in this volume .

REFERENCE S Baars, B . J . (1986) . The cognitive revolution in psychology. New York: Guilford. Chomsky, N . (1959) . Review of Skinner' s "Verbal Behavior" . Language, 35, 26-58 . Coulter, J . (1983) . Rethinking cognitive theory. New York: St . Martin's Press . Hamlyn, D . (1958) . Behaviour. Philosophy, 28, 19-40. Lakatos, I. (1978) . Falsification and the methodology of scientific research programmes . In J . Worall & G. Currie (Eds .), Imre Lakatos . Philosophical Papers (Vol. 1, pp . 8-101) . Cambridge, England : Cambridge University Press . MacCorquodale, K. (1970) . On Chomsky ' s review of Skinner's Verbal Behavior. Journal of th e Experimental Analysis of Behavior, 13, 83-99 . MacCorquodale, K ., & Meehl, P. H . (1948) . On a distinction between intervening variables and hypothetical constructs . Psychological Review, 42, 111-132 . Melden, A . I. (1961) . Free action. London: Routledge & Kegan Paul . O ' Donohue, W. (in preparation) . Skinner's psychology. London : Sage. O'Donohue, W., Callahan, G ., & Ruchstuhl, L.E. (In press). Epistemological barriers to radical behaviorism. The Behavior Analyst. O'Donohue, W., & Naugle, A . (in preparation) . The anomalous cognitive revolution . Peters, R . S . (1958) . The concept of motivation. London: Routledge & Kegan Paul . Quine, W. V. O . (1960) . Word and object. Cambridge, MA : M .I .T. Press . Rosch, E . (1978) . Principles of categorization . In E . Rosch & B . Lloyd (Eds .), Cognition and categorization. New York : Wiley. Ryle, G . (1949) . The concept of mind. New York: Barnes & Noble . Skinner, B . F. (1957) . Verbal behavior. Englewood Cliffs : Prentice Hall. Smith, L . D . (1986) . Behaviorism and logical positivism: A reassessment of the alliance. Stanford , Ca : Stanford University Press . Todd, J . T., & Morris, E. K . (1992) . Case histories in the great power of steady misrepresentation . American Psychologist, 47, 1441-1453 . Wittgenstein, L. (1958) . Philosophical investigations . (G .E .M . Anscombe Trans.) . New York : Macmillan . Wyatt, W. J ., Hawkins, R. P., & Davis, P. (1986) . Behaviorism: Are reports of its death exaggerated? The Behavior Analyst, 9, 101-105 . Zuriff, G . (1985). Behaviorism: A conceptual reconstruction . New York : Columbia University Press .


G

WATSONIAN BEHAVIORIS M

EDWARD K . MORRIS

JAMES T . TOD D

University of Kansas

Eastern Michigan University

I. Introductio n II. Intellectual, Cultural, and Individual Contex t A. Intellectual Context: Philosophy, Science, and Psychology B. Cultural and Individual Contex t III. Before Watsonian Behavioris m A. Watson's Research Program s B. Watson and Psychology in 191 3 IV. Watsonian Behaviorism A. Watson : 1913—1920 B. The Behaviorist Manifesto C. "Psychology as the Behaviorist Views It " D. A Science of Behavio r V. After Watsonian Behaviorism A. Watson: 1920—1936 B. Heredity and Environmen t C. Applications to Child Rearing and Advertisin g D. Watson : 1936—1958 VI. Conclusion References

Handbook of Behaviorism

15

Copyright © 1999 by Academic Press. All rights of reproduction in any form reserved .

16

MORRIS AND TODD

I . INTRODUCTIO N

When John B . Watson (1878–1958) was born on January 9, 1878, in Travelers Rest, South Carolina, Wilhelm Wundt (1832–1920) was in the process o f founding "experimental psychology" in Leipzig, Germany (Wundt, 1873) . As a science, Wundt's psychology was German : intellectual, idealistic, and academic . As a psychology, it was mentalistic : Its subject matter was the contents of consciousness ; its method was introspection ; it was a psychology of the adult human mind. This was the psychology taught to the first generation of American s seeking doctoral degrees in Europe (Boring, 1950, pp . 505 -549) . This was the psychology Edward B . Titchener (1867–1927) transformed in America into th e first school of psychology structuralism (see Titchener, 1897) . Thirty-five years later, in 1913, Watson founded another school of psycholog y in two lectures delivered at Columbia University in New York City, the first o n February 24 and the second on April 3 . The lectures were published as "Psychology as the Behaviorist Views It" (Watson, 1913b) and "Image and Affection i n Behavior" (Watson, 1913a) . Watson's psychology was American : pragmatic , matter-of-fact, and professional . It was behavioral in its methodology and it s metaphysics : Its subject matter was behavior; its methods were objective. It was a psychology of nonhuman and human behavior, the behavior of infants, children, and adults . This was the school of psychology that Watson himself s o named behaviorism (Watson, 1913b) . In 1920 Wundt died, and Watson resigned from the academic position he hel d in psychology. Behaviorism was then becoming the psychology taught to Americans seeking doctoral degrees in the United States . By 1938, 25 years after the Columbia lectures, the transformation was complete: Experimental psychologists studied behavior, not consciousness or mind (see Woodworth, 1938) . Methodologically, psychology is little changed today : Psychologists still study behavior, but no longer as a subject matter in its own right, as it was for Watson . Instead, behavior is a basis for objective inferences about brain, mind, and cognition and theories thereof (see Hulse, Fowler, and Honig, 1978 ; Lachman , Lachman, and Butterfield, 1979) . Watson's metaphysical behaviorism, in contrast, never held a prominent position in psychology (see Bergmann, 1956), al though other behaviorisms have been as self-statedly naturalistic notably, J . R . Kantor's (1888–1984) interbehavioral psychology (Kantor, 1924, 1926 ; see L . J . Hayes, this volume) and B . F. Skinner's (1904–1990) radical behavioris m (Skinner, 1938, 1953 ; see Ringen, this volume) . Coincidentally, the former' s main systematic outline (Kantor, 1958) and the latter's first journal (Journal of the Experimental Analysis of Behavior) were published the year Watson die d 1958 . This synopsis of the history and system of Watsonian behaviorism is generally the received view, yet it underdetermines Watson's conceptual and empirical programs both those that were seeming and those that were real . That is, al-

2 WATSONIAN BEHAVIORISM

17

though Watson's behaviorism can be succinctly characterized, his views varie d within and across the corpus of his work (Logue, 1985b) . A scholar's work is expected, of course, to evolve over time, but in Watson's case this evolution has (a ) made his behaviorism difficult to describe as a self-consistent system and (b) al lowed variations within it to support contrasting accounts, both among themselves and with what Watson wrote (e .g ., Buckley, 1989 ; Harzem, 1993 ; see Herrnstein, 1969 ; Kimble, 1997 ; Samelson, 1980) . Thus, whereas Watsonian behaviorism was clearly an objective psychology, its details varied and were sometimes undefining of his "-ism." And Watson's last word was not necessarily hi s best word . In this chapter, then, we do not depict the one "true" Watsonian behaviorism, but rather attempt to describe Watson's system more broadly, notin g variations within it both supposed and actual . We begin by presenting the intellectual and cultural context out of which behaviorism emerged, including Watson in his own time . We then turn to the behaviorism he founded, organizing it into three sections . First, we recount Watson's program of research between 1900, when he entered graduate school at th e University of Chicago, and 1913, when he published his Columbia lectures . Second, we describe the behaviorism he founded in the first of these publication s (Watson 1913b) and as it matured throughout the remainder of the decade, culminating in Psychology from the Standpoint of a Behaviorist (Watson, 1919 , 1924b, 1929) . In this context, we address three of its main features : its goal o f prediction and control ; its methodological and metaphysical stances ; and its approach to thinking, language, and implicit response . Third, we describe Watson' s contributions to psychology between 1920, when he resigned from Johns Hopkins University, soon to begin a career in advertising, and 1930, when he published his last book, the revised edition of Behaviorism (Watson, 1924a, 1930) . Here, we address a fourth feature of his behaviorism his stance on heredit y and environment and review his contributions to advertising and child rearing . These three sections correspond, respectively, to Watson's career as a researcher , a systematist, and a popularizer of behaviorism . He of course played all of thes e roles in each phase of his career, but their relative emphases changed with time, occasion, and opportunity. Our account is drawn from Watson's written work (see Todd, Dewsbury , Logue, & Dryden, 1994) and related primary and secondary sources from tha t and subsequent eras (e .g ., Bergmann, 1956 ; Lashley, 1923 ; see Wozniak, 1993a, 1993b, 1994) . It is informed as well by an increasing body of Watson scholar ship (e .g ., Buckley, 1989 ; Logue, 1985a ; Samelson, 1981 ; see Todd & Morris , 1994) and authoritative texts (e .g ., Boakes, 1984 ; Leahey, 1997 ; Malone, 1990, pp . 93-118) . We do not, however, cover popular treatments of thi s material (e .g ., Cohen, 1979) or texts that draw from them because, althoug h interesting, their "facts" and interpretations are sometimes mistaken (Harris , 1981 ; Larson, 1981) . Our account is selective, especially where we interrelat e intellectual, cultural, and individual history, which may suggest a presentist

18

MORRIS AND TOD D

perspective on Watson's contributions in their own time (see Samelson, 1974 ; cf. Stocking, 1965), but this is not our intent . That is, we do not mean to justify Watsonian behaviorism as the "correct" outcome of some inevitable historica l progress, but rather relate the contexts and contingencies within which it evolved . We begin by describing an intellectual tradition in which philosophy wa s tending toward psychology, psychology toward science, and scientific psychology toward the study of behavior. We then explain how the progressive , pragmatic culture of the United States fostered a behavioral psychology . We end with Watson himself, who was born into a poor, rural, southern family, but who achieved success in the city through hard work, education, and a profession . II . INTELLECTUAL, CULTURAL, AN D INDIVIDUAL CONTEX T A . INTELLECTUAL CONTEXT : PHILOSOPHY, SCIENCE, AND PSYCHOLOG Y

The Scientific Revolution (ca . 1600–1750) changed how nature was known and what was known about it, as well as assumptions about the nature of natur e (Lindberg, 1992) . Religion, superstition, and scholastic traditions were superseded by reason, observation, and experiment, as urged by Francis Bacon (1561–1626) . Nicolaus Copernicus (1473–1543) discerned that the earth wa s not the center of the universe, but positioned in a heliocentric orbit, so tha t Galileo Galilei (1564–1642) and Johannes Kepler (1571–1630) could more accurately predict its orbit and offer a more coherent account of the solar system. The physics of objects in motion became better understood through Isaac New ton's (1642–1727) contributions ; it was then explained mathematically, obviating any recourse to essences and animism . A material, mechanical worldview was replacing spiritual and occult traditions (Westfall, 1971) . The revolution, though, had disquieting effects . Enlightenment philosophy and natural science met with religious and romantic resistance . Even scientist s themselves were reluctant to relinquish mystical, hermetic traditions (e .g . , alchemy ; see Bonelli & Shea, 1975 ; Shapin, 1996) . The revolution took time t o complete . A similar pattern occurred later with the ascent of evolutionary biology and psychology. In the latter, the structuralists opposed bringing the study o f behavior into psychology proper ; in their view, only the introspective analysis of consciousness was proper psychology (Titchener, 1897, 1910) . Later, these op posed programs were loosely conjoined in a common purpose to study function—the function of consciousness and behavior in species and individuals . This was the second school of psychology functionalism (Angell, 1907) . A s psychology evolved in America, however, with practical ends in a practical culture, this compromise failed . The objective study of behavior superseded both


19

the structural and the functional analysis of consciousness, leaving behavioris m "the only consistent and logical functionalism" (Watson, 1913b, p . 166) . 1. Experimental Psycholog y With the Scientific Revolution came modern philosophy, in part through Rene Descartes's (1596–1650) attempt to reconcile mind and body . Descartes posite d that human and nonhuman physiology, and thus behavior, operated according t o mechanical principles (e .g ., the reflex), while humans alone possessed a sou l the seat of consciousness independent of a materialist account . With a distinc t concept of consciousness, associationistic and empiricist philosophers such a s David Hartley (1705–1757), David Hume (1711–1776), John Lock e (1632–1704), James Mill (1773–1836), and John Stuart Mill (1806–1873) undertook the rational analysis of mind (Brown, 1985) . Hume, Hartley, and Locke most influenced Watson (1936) . With a materialist account of body and behavior , physicists and physiologists, such as Franz Josef Gall (1758–1828) on brai n function, Charles Bell (1774–1842) and Francois Magendie (1783–1855) o n the central nervous system, Hermann von Helmholtz (1821–1894) on sensor y physiology, and Gustav Fechner (1801–1887) and E. H . Weber (1795–1878) on sensory perception (see Boring, 1942), undertook the empirical analysis of the body. While at Chicago and Johns Hopkins, Watson would be influenced by the next generation of biologists and neurologists . Wundt (1873) brought together these philosophical, physiological, and psychophysical methods and in 1879 founded psychology's first laboratory, thereb y establishing a new discipline . Its methods involved presentations of carefull y measured stimuli under controlled conditions ; its procedures entailed the rigorous introspection of these stimuli as perceived in consciousness ; its data were self-reports of this conscious content (e .g ., sensations, feelings, and images) . Although variously transformed in the succeeding years (Leahey, 1981), thi s was "experimental psychology" in turn-of-the-century America . In this context , Watson's early research in comparative psychology and animal behavior was peripheral ; indeed, it was often considered physiology, not psychology. Thi s disciplinary hegemony eventually became a source of contention (see Wozniak , 1993b) and the impetus for Watson's behaviorism . As Watson (1913b) wrote: "[e]ither psychology must change its viewpoint so as to take in the facts of behavior, whether or not they have bearings upon the problems of 'consciousness' ; or else behavior must stand alone as a wholly separate and independent science " (p . 159) . 2. Toward Behaviorism Not all philosophy and psychology sought to harmonize mind and body, how ever . Other scholars and scientists attempted to naturalize them both and science as well . Some of these efforts were precursors to Watson's behaviorism ; others were parallel programs from which it would draw .

20

MORRIS AND TOD D

a. Philosophy of Science Among the precursors of behaviorism was Bacon's ([1620] 1960) descriptiv e positivism and his insistence that science, then often abstract and contemplative, was intimately allied with technology and craft (see Smith, 1992, 1995) . Man y scientific advances emerged in the context of solving technological problems , and many technological advances emerged from solving scientific problem s (e .g., in navigation, astronomy, and optics ; weaponry, ballistics, and physics ; steam engines and thermodynamics), making the goals of science both intellectual and practical . This practical knowledge was derived from working directl y with the subject matter itself, where the "truth" of this knowledge was synonymous with prediction and control . Epistemology was to be empirical (see PerezRamos, 1988) . Bacon ([1624] 1942) also had a reformist, even utopian, perspective : Scientific and technical knowledge could and should be applied to solvin g social problems and building a better world . Watson's behaviorism would reflect these ideals, too . Knowledge was bes t gleaned from observation and experiment, the results of which were applied t o problems in human behavior and social organization, for instance, in child development, education, and business . Indeed, for Watson (1913b), applied scienc e was science itself: Experimental pedagogy, the psychology of drugs, the psychology of tests, and psychopathology are all vigorous growths . These are sometimes wrongly called "practical " or "applied" psychology . Surely there was never a worse misnomer. In the future there may grow up vocational bureaus which really apply psychology . At present these fields are truly scientific and are in search of broad generalizations which will lead to the control of human behavior. (p . 169)

Many of Watson's subsequent publications were devoted to applied psycholog y (e.g ., Watson, 1916a, 1927a ; Watson & Lashley, 1920) and to recommendation s for social engineering that is, for changing social and cultural practices (e .g. , Watson & Watson, 1928) . b. Evolutionary Biology and Neuroscienc e From 19th century science came two other programs evolutionary biolog y and Russian neuroscience (see Boakes, 1984) . Charles Darwin (1809—1882) offered an evolutionary account of the continuity of species (Darwin, 1859) an d proposed natural selection as a mechanism through which it proceeded (Darwin , 1871) . Darwin's theory was, of course, revolutionary in its own right, with it s own disquieting effects (Cosslett, 1984) . It challenged assumptions about the status of humans in the natural order and, with that, the meaning of life . Neither was now outside the purview of science . Later, Watson (1913b) would challenge related assumptions : "[T]he behavior of man and the behavior of animals mus t be considered on the same plane ; as being equally essential to a general under standing of behavior" (p . 176) . This challenge was founded on two consequences of Darwin's (1872) work :


21

comparative psychology and the psychology of adaptation (see Boakes, 1984) . Comparative psychology inquired after the evolutionary basis of the mental an d behavioral traits and the capacities of humans and nonhumans (George John Romanes, 1848—1894 ; e .g ., Romanes, 1882, 1884) . In the beginning, its method s were anecdotal, and its inferences about the minds of nonhumans were subjective . In making comparative psychology scientific, C . Lloyd Morgan (1852—1936) dismissed subjective inferences (Morgan, 1894), but his admonitions were also extended to objective inferences about animal mind (Costall , 1993), which Watson (1913b) later applied to humans : This attempt to reason by analogy from human consciousness process to the consciou s processes in animals, and vice versa: to make consciousness, as the human being know s it, the center of all behavior, forces us into a situation similar to that which existed in biology in Darwin ' s time . (p . 161 )

As for the psychology of adaptation, once the continuity of species was established, continuity in mental and behavioral adaptation for instance, in menta l and behavioral processes (e .g ., association) could also be assumed to hold across humans and nonhumans (Spencer, 1897) . If so, then adaptation and association could be isolated and studied in nonhuman species (e .g ., cats) in nonnative environments (e .g ., puzzle boxes ; see Thorndike, 1898) . The psychology of adaptation was also a catalyst for functionalism, founde d by John Dewey (1859—1952), William James (1842—1910), and James Rowlan d Angell (1869—1949) . Functionalism was concerned with the purpose and use o f mind, consciousness, and behavior in nonhumans and humans alike (see Dewey , 1886 ; James 1890, 1892) . Watson would enroll at the University of Chicago to study philosophy with Dewey, and Angell became a major influence, but Watso n (1903) dismissed mind and consciousness, and he began his career studying th e behavioral adaption of the white rat in systematic laboratory preparations (e .g . , mazes), where adaptation became learning (Watson, 1913b) : The time seems to have come when psychology must discard all reference to consciousness ; when it need no longer delude itself into thinking that it is making mental states th e object of observation. We have become so enmeshed in speculative questions concernin g the elements of mind, the nature of conscious content . . . that I, as an experimental student, feel that something is wrong with our premises and the types of problems that develop from them. (p . 163 )

The second program from science to behaviorism was Russian neuroscience , but it was a parallel development, not a precursor. Adopting Descartes's account of bodily action, Russian psychology materialized mind into brain, which it the n explained in terms of reflexes . This approach was the tradition of Iva n Sechenov's (1829—1905) objectivism, Vladimir Behkterev's (1867—1927) reflexology, and Ivan Pavlov's (1849—1936) experimental analysis of the refle x and its conditioning . Watson's 1915 presidential address to the American Psychological Association (APA) (Watson, 1916b) drew on Behkterev (1906, 1913)

22

MORRIS AND TODD

and Pavlov (1906 ; Yerkes & Morgulis, 1909) offering the reflex as an objectiv e means for analyzing sensations . Later, Watson used it to interpret emotional behavior (Watson & Morgan, 1917) and then as a basis for synthesizing conditioned emotional reactions (Watson & Rayner, 1920) . c. Philosophy of Psychology Returning to philosophy, Ernst Mach (1838–1936) reformulated Bacon' s views and August Comte's (1798–1857) positivism in the context of Darwinia n theory and offered a new philosophy of science (Mach, [1883] 1960, [1886] 1959 ; see Blackmore, 1972) . He advanced a descriptive positivism in whic h metaphysical entities, and theories thereof, were rejected . He analyzed scientific concepts historically, so as to reveal any metaphysical bases and biases . And he saw science itself as an economical, efficient adaptation to the world, wher e knowledge was not a search for truth, but for functional relations among event s (see Abbagnano, 1967 ; Smith, 1995 ; cf. Loeb, 1916) . The search for functional relations made prediction and control the goals o f science . Indeed, the same year the APA was founded, William James (1892) observed that the aim of science was "practical prediction and control" (p . 148 ) and claimed that for psychology to become a science, independent of philosophy, this must be its goal, too and it became so (Watson, 1913b, p . 158) . A distinctly American philosophy—pragmatism—founded by Charles S . Peirce (1839–1914) and James at Harvard, and George Herbert Mead (1863–1931) and Dewey at Chicago (see Scheffler, 1974 ; cf. Pepper, 1942 on contextualism ; see Gifford S . C . Hayes, this volume) arose in this context . In this view, nature, mind, and behavior were acts in context, historically dependent and so too was science . As science advanced, new findings revealed the limitations of prior knowledge and prior assumptions, such that "effective action" be came a better criterion for truth than coherence or correspondence (Peirce, 1905 ; see Ayer, 1968) . Watson (1913b) likewise challenged prior knowledge and assumptions and became thoroughly pragmatic, although not obviously a philosophical pragmatist . This was the intellectual tradition out of which Watson's behaviorism emerged . It was also a context in which an objective psychology was becomin g almost inevitable . Such a psychology was emblematic of cultural values in th e United States and well suited to individuals such as Watson . This is material to which we now briefly turn. B. CULTURAL AND INDIVIDUAL CONTEX T

1 . The Culture : Social and Political The 1800s found the United States a new nation, but not without a history . Enlightenment philosophy and evangelical religion emigrated from Europe, th e latter proselytizing for conversion and change in both soul and action . Watson (1913b, 1919) would proselytize, too, even lecturing to crowds on street corner s in New York City (Baum, 1994) . As a nation unto itself, the business of America


23

became business, where practical, technical knowledge was as highly valued a s pure science (Burnham, 1989 ; Smith & Marx, 1994) . America was changing in other ways, too . Individual freedom was giving way to institutional control (Wiebe, 1967) . A rural, agricultural society was being displaced by a growing urban industrial complex (Bakan, 1966) . A form of radical environmentalism arose in which action not heredity, family, or tradition—dictated achievement . Intellectually, what emerged in response to questions about human morals and comportment was a faith in science and technology (see Burnham, 1989) . Socially, what emerged in response to the stresses an d strains in American society was Progressivism . Science, technology, and Progressivism became the bases for a promise of progress in building a new nation, as well as individual self-improvement . Progressivism (1890–1920) was a social and political movement designed t o bring order out of confusion that is, out of weakening religious certainties an d traditions and out of the complex and changing demands of new social and economic institutions (see Buenker, Burnham, & Crunden, 1977) . The nation sought order through reform, from which a cult of efficiency emerged, along with rational, disinterested professionalization in business, industry, and governmen t (Wiebe, 1967) . For themselves, individuals sought order through self-control an d self-improvement (Bledstein, 1976), the latter based on education, especially ir f professions that could contribute to Progressivism's social and political agend a (O'Donnell, 1985) . Progress was predicated on the assumption that it could b e achieved through science and technology . Prediction and control were thus mor e than the goals of American science : They were part of the fabric of the culture as well (see Ross, 1991), to an extent that is difficult to fathom today (see L . D . Smith, 1992) . 2. The Individual : John Watson This was the world into which John Broadus Watson was born . His family was impoverished, southern, and largely uneducated (see Brewer, 1991 ; Buck ley, 1989 ; Watson, 1936) . His father, though, gave him the skills and sensibilitie s necessary for surviving in rural America that is, for managing a farm and its equipment, livestock, and animals . His mother gave him a sense of duty to th e Baptist church and a faith in success through education . Watson's father abandoned the family, however, and Watson's world was suddenly less certain . When the family then moved into Greenville, South Carolina (pop . 20,000), Watson found his rural skills and sensibilities poorly suited to success in the city . He be came rebellious and insubordinate, a disinterested student, a boy growing up uncertain about himself in uncertain times . Watson eventually realized that one means of escaping his circumstances was through an education and a profession, for which university-based specializations were becoming a means (Buckley, 1989, pp . 17–32) . At the age of 16 , Watson convinced himself and others that he should be accepted at Furman University, a small Baptist college in Greenville . There, he developed a strong interest in philosophy, which eventually led him to psychology . He graduated 5 years

24

MORRIS AND TOD D

later with a masters degree, but still poor and with few close friends . When hi s mother died a year later, Watson was freed from his family and his Baptist obligations, but not their enduring influence (Creelan, 1974), to pursue a career . Among the professions, he would find that psychology as both a science and a technology could generate the practical knowledge necessary for successfu l working in a self-consciously progressive culture . For his graduate work, Watso n chose the Department of Philosophy and Psychology at the modernist Universit y of Chicago (f. 1890 ; see Buckley, 1989, pp . 33-58) . a . Persona We delve no further into Watson biography at this point, but touch on it agai n later. However, we would be remiss in not mentioning that Watson's behavioris m was a function not only of the intellectual and cultural contexts just described but also of him as an individual . We do not mean by this to bring zeitgeist an d great person history into conflict, but rather to point out that, although scientist s are products of broad intellectual and cultural forces (R. I. Watson, 1971), they are also individuals who interact uniquely with the events of their day and thu s contribute uniquely to the events that follow (Boring, 1955) . For Watson, two points are relevant . First, Watson's contributions to behaviorism had more the imprimatur of his character than those of the later behaviorists because their contributions wer e conditional on the preceding history of behaviorism, which dampened the impact of their individual character. Watson's contributions, however, occurred be fore a history of behaviorism existed, allowing his persona to have more of a n impact . Second, until other distinct and well-recognized behaviorisms developed, society, science, and psychology could not place Watson's behaviorism i n the context of behaviorism more generally because, again, behaviorism had n o history. As a result, his personal views were sometimes mistakenly seen as defining features of behaviorism rather than one man's perspective . For instance, his admonition against allowing children to develop strong emotional ties to thei r parents (Watson & Watson, 1928, pp . 69-87), about which he later demurre d (Watson, 1936), was seen as a peculiarly behavioral perspective (e .g ., Buckley, 1989, pp . 168-169) . The admonition, though, was unrelated to Watson ' s behaviorism as an "-ism," albeit perhaps conditional on his childhood . Beyond these comments, we are reluctant to offer much more of an accoun t of Watson as an individual for three reasons . First, we never knew Watson personally. Second, Watson was rarely interviewed (Burnham, 1994 ; cf. Hannush, 1987), his autobiography is not especially revealing (see Watson, 1936), and later in life he burned all his papers (Burnham, 1994) . Third, descriptions of Watson drawn from the existing archival records (e .g ., his correspondence with colleagues) seem sometimes influenced by predispositions for or against behaviorism and thus may not be entirely objective (see, e .g ., Buckley, 1989 ; Harzem, 1993) .


25

Nonetheless, we think we can fairly state that Watson had a disciplined , straightforward, and self-assured style (Brewer, 1991)—a style that was a positive force in the evolution, promotion, and dissemination of behaviorism (Lovie , 1984) . In addition, he could count among his close colleagues and friends bot h Titchener and Angell, and his second marriage was a noted success . Watson' s style, however, was also judged to be arrogant, rigid, and unreflective and thu s detrimental to some of his academic, familial, and personal relations (Brewer, 1991 ; Buckley, 1989, pp . 112–133 ; Buckley, 1994 ; Horowitz, 1992) . These relations, though, were multiply determined : Watson challenged deeply seated, long standing traditions, which was bound to evoke criticism about his behavioris m and his promotion of it . Any more than this, though, we are reluctant to say ; Watson's persona awaits fuller treatment . At this point, we turn to his career and his contribution s

III . BEFORE WATSONIAN BEHAVIORIS M When Watson arrived at the University of Chicago in the fall of 1900, h e found an academic environment oriented toward Progressive reform and th e "new professionalism" (Bledstein, 1976 ; Buckley, 1989, pp . 33–58) . It was a context in which Watson could improve and advance himself through a career i n psychology . He began by studying philosophy under Dewey, but his interes t soon faded . As he said later : "I never knew what he was saying then, and unfortunately for me, I still don't know" (Watson, 1936, p . 274) . In place of philosophy, Watson took courses in functional psychology from Angell, who became his intellectual mentor, encouraging him not only to de scribe behavior but to analyze it as well. Watson was thus led to courses in neurology from Henry Donaldson (1857–1938) and physiology from Jacques Loe b (1859–1924) . Loeb was the champion of a self-consciously pragmatic approac h to the life sciences, with prediction and control as its goal (Loeb, 1900, 1916 ; see Pauly, 1987) . He encouraged Watson to study the physiology of the dog' s brain under his guidance, but Watson's future advisers Angell and Donaldson—did not feel that "Loeb was a very `safe' man for a green Ph .D . candidate" (Watson, 1936, p . 273) . 1 They suggested instead that Watson "make a study o f the psychical development of the white rat in correlation with the growth of it s nervous system" (Watson, 1903, p . 5) . This would be the subject of his dissertation . Comparative psychology was more Watson's metier than philosophy anyway. Indeed, given his background, he was arguably an "animal man" before he kne w 'What Angell and Donaldson meant by not "a very `safe' man" has never been fully explained . It presumably implied something about Loeb's science (e .g ., its unrelenting objectivism) or his persona, both of which were thought extreme (see Pauly, 1987) . To Richard Herrnstein, it may also have meant that Loeb was Jewish and that having a Jewish adviser was not a safe start to an academic career (R. J. Herrnstein, personal communication, 1988 ; cf. Winston, 1996) .

26

MORRIS AND TOD D

it (see Bakan, 1966) . Later, his behaviorism would be more a product of his re search in comparative psychology than of formal philosophy. Indeed, Watson' s behaviorism was taken seriously because of the stature he had achieved betwee n 1900 and 1913 through the extent and quality of his research (Bergmann, 1956 ; Boakes, 1984 ; Buckley, 1989, pp . 85–87) . The 1900–1913 period was Watson' s most productive as a scientist and spanned more than half of his academic career. We organize it here in terms of his research programs during this perio d programs in comparative psychology, ethology, and psychophysics (see Todd & Morris, 1986) . A . WATSON'S RESEARCH PROGRAM S

1 . Comparative Psychology Watson's dissertation, Animal Education, was subtitled An Experimental Study of the Psychical Development of the White Rat, Correlated with th e Growth of Its Nervous System (Watson, 1903) . It addressed the following questions : (1) How far is it possible (dealing with the psychological side of the problem) to give a systematic account of the gradual unfolding of the associative processes in the rat? (2) Is it possible . . . to find out whether or not medullated [myelinated] nerve fibers . . . are a conditio sine qua non of a rat' s forming and retaining definite associations? (3) Is there any demonstrable connection between the increasing complexity of the psychical life . . . and the number of the medullated fibers in the cortex? (p . 5 )

These were important issues in turn-of-the-century comparative psychology, i n which assertions about behavior were often based on theories about the nervou s system (e.g ., the brain structures requisite for learning), not on experimental evidence. Using research preparations involving problem boxes and mazes in the tradition of Morgan (1894), Thorndike (1898), and Small (1900), Watson found tha t rats could not find food hidden in sawdust before 12 days of age, but that by 2 3 days they performed as quickly as adult rats, albeit not as efficiently. Through surgical procedures, he also established that rats were not born with myelinated nerve fibers in their brains, but rather that myelinization increased over th e course of the first 24 days, although even then it was not complete . Myelinization was thus not the conditio sine qua non it was theorized to be ; the observational and experimental evidence showed otherwise . Shortly thereafter, Watson (1904) underlined the importance of experimentation : We do plead for long and careful studies in more restricted lines than that represented b y simply taking an animal and watching its general behavior . It is time to put the animal i n such situations that some mental act may be exhibited to the exclusion of others . (p. 363) .

Watson's dissertation "classic in developmental psychobiology" (Dews bury, 1994, p. 143) was completed in 1903, when he received Chicago's firs t Ph .D . in psychology. He graduated magna cum laude and remained with the department as a laboratory supervisor and instructor and an assistant to Angell for


27

another 5 years, until 1908, when he moved to Johns Hopkins University . Of hi s experience at Chicago, Watson (1936) later wrote : I learned my James under Angell, patience and exactness in research from Donaldson , and from Loeb the fact that research need not be uninteresting . I learned also under Angell how to lecture and express my thinking in words simply chosen . (p . 275)

At Johns Hopkins, he would find students, colleagues, and mentors of the same caliber James Mark Baldwin (1861–1934), H . S . Jennings (1869–1947) , Arthur O . Lovejoy (1873-1962), Adolf Meyer (1866–1943), and Curt Richte r (1894–1988) some of them already inclined toward behaviorism, among the m Knight Dunlop (1875–1949), Karl S . Lashley (1890-1958), and Robert Yerke s (1876-1956) . a. Kinaesthetic and Organic Sensations Between the receipt of his doctorate and his 1913 Columbia lectures, Watso n published more than a dozen additional articles related to his dissertation . Some of them addressed the relative importance of the sensory systems for learning and were conducted with Harvey Can (1873-1954), then a student of Angell' s (see Can & Watson, 1908) . The best known of these articles was "Kinaestheti c and Organic Sensations : Their Role in the Reactions of the White Rat to th e Maze" (Watson, 1907a) . In preparation, Watson trained and tested rats' maze performance under different conditions for instance, at various lighting levels and with visual and tactual cues but to no effect . Drawing on surgical skills acquired at Johns Hopkins in 1905, he created three groups of rats that could not, respectively, see , hear, or smell. Their performance after surgery, though, was no different than be fore on a variety of measures (e .g., at different starting points) . Finally, Watson made one rat blind, deaf, and anosmic and removed all its whiskers . Although this rat took more trials to learn the maze, its final running speed matched that o f the controls . Watson's (1907a) conclusion was that : "The kinaesthetic impressions coupled with certain other intra-organic impressions are the only necessar y sensory factors used in the formation of the maze association" (pp . 84–85) . That is, the stimuli necessary for learning were private events . 2 . Ethological Investigation s Although Watson (1904) emphasized the importance of laboratory experimentation, he also valued careful observational and field research . On invitation from the Carnegie Institution, he made field trips in 1907, 1908, 1910, and 191 3 to study the Noddy and Sooty Tern Colony on Key Bird of the Tortugas Islan d Group off southern Florida (Watson, 1907b, 1908a, 1910a, 1910c, 1912b) . There, he made careful records of the birds' distribution on the island ; their feeding, nesting, and mating habits ; their care for their young ; and the young birds ' development . These observations yielded what would later be called a n ethogram, a comprehensive account of an organism's naturally occurring activi-

28

MORRIS AND TOD D

ties . In addition, Watson (1908a) raised a number of birds on his own and note d that they followed him about (i.e ., to imprint ; Gray, 1963 ; see Lorenz, 1935) . 2 In his field studies, Watson (1908a) investigated the extent and basis of mate , egg, and nest recognition and the birds' performance in mazes . In the latter, h e found that they were more variable than rats in their maze-running times because they instinctively stopped and "idled" that is, stood motionless for periods o f time (cf. the Breland effect ; see Breland & Breland, 1961) . In other studies, Watson (1910a) analyzed the birds' homing abilities and found that they could retur n to the islands from unfamiliar locations hundreds of miles away. Taken as a whole, Watson's field observations and studies were among the earliest and mos t careful work in ethology (Tinbergen, 1951, pp . 52, 146) . 3. Psychophysical Studie s After Watson moved to Johns Hopkins, he and his colleague at Harvard University, Robert Yerkes, began a project commissioned by the APA : They developed standard procedures for assessing the visual capacities of nonhumans . Watson was responsible for designing the equipment and developing the assessmen t methods, which were reported in considerable detail (Yerkes & Watson, 1911) . In addition, he conducted several psychophysical studies . For instance, using a discriminated choice procedure, Watson (1909b) investigated the differentia l sensitivity of monkeys to colors, observed their tendency to respond to position , and noted their preference for food rewards that were earned as opposed to thos e made freely available (i .e., contrafreeloading ; see Osborne, 1977) . The psychophysical equipment and assessment methods Watson and Yerkes designe d and developed, and their findings, were significant steps forward in the quantitative assessment of nonhuman perception (Boakes, 1984) . 4. Conclusio n We can see in Watson's research at Chicago the hallmarks of his research an d theory to come . He constructed, standardized, and maintained most of his apparatus which was consistent with his hands-on approach to understanding his subject matter (Boakes, 1984; Todd & Morris, 1986) . He used and advocated "genetic" or developmental methods (e .g ., longitudinal designs) for answerin g 2As an example of how Watson and his work would later be viewed, Konrad Z . Lorenz (1903—1986), who was awarded the 1973 Nobel Prize in Physiology and Medicine along with Nik o Tinbergin (1907—1988) and Karl von Frisch (1886—1982), once commented, "If J . B . Watson ha d only once reared a young bird in isolation, he would have never asserted that all complicated behavior patterns were conditioned " (Lorenz, 1950, p . 233) . But later Lorenz responded to Dewsbur y (1994) : I begin with the confession that I am quite aware of having done some injustice to the behaviorists in general and to John Watson in particular. What I know about him, is onl y what Karl Buhler made me read of his works and this was calculated, I think, quite consciously to irritate me and raise my objections. (p . 144 ) But Lorenz reflected that "I feel that having done some really ethological work on terns he i s more reprehensible for having held his later views" (p . 144, emphasis mine) .


29

questions about behavior, which he would later adapt for studying infant an d child behavior (e .g ., Watson, 1919) . His work was descriptive and inductive an d avoided subjective and objective inferences about consciousness . For example , although he used terms such as psychical and association, they were usually made in reference to observable relations such as habit (Watson, 1903, 1913b) . He was also critical of conclusions that lacked empirical support, especiall y those that were physiologically reductionistic (Yerkes, 1905) . Finally, he extrapolated his findings to practical matters, noting in Animal Education, for instance , that if young rats could learn without having a fully developed neurological system, then so too should human infants (Watson, 1903, pp . 121–122) . B . WATSON AND PSYCHOLOGY IN 191 3

By 1913, Watson was established. As for his personal life, he married Mar y Amelia Ickes (1885–1973) in 1903 ; she was from a prominent family and wa s once his student at Chicago . Between them, they had two children Mary (Polly) (b . 1905) and John (b . 1907) . Professionally, Watson had attained considerable stature as a psychologist . His research was known for being carefully de signed and conducted, and his field observations for being lengthy and detailed—and they were published in prestigious journals (e .g ., Science, Psychological Review) . His reputation as a scientist would never be doubte d (Buckley, 1989, pp . 112–133) . He was also becoming a systematist . He had written reviews of the literature (e .g ., Watson, 1908b, 1911a) and reviewed now classic texts in comparative and human psychology (e .g ., Jennings, 1906 ; Thorndike, 1911 ; Titchener, 1911 ; Washburn, 1908 ; see Watson, 1907c, 1908c , 1911b, 1911c) . And he had begun describing the importance of research on animal behavior in the popular press (e .g ., Harper's Monthly Magazine ; see Watson, 1909a, 1910b, 1912a) . In all, Watson amassed more than 65 publication s during this 10-year period and attained commensurate professional status . H e was the director of the Johns Hopkins psychology program, cofounded an d edited the Journal of Animal Behavior, ran and edited the Psychological Review, and coedited the Psychological Bulletin . In 1914 he was selected by his peers to be the 24th president of the APA and in 1915 founded the Journal of Experimental Psychology. Although well respected in his discipline, when Watson moved to Johns Hopkins in 1908, he was not yet sure what psychology uniquely constituted . He wa s unwilling to make it over into physiology or leave it to the structuralists . The former lent scientific credence, but at the cost of independence ; the latter had n o credence at all in the other sciences . If psychology was to become a science, i t had to produce results at the level of behavior useful results . If psycholog y was to garner university and foundation support, it had to generate public inter est. These became major themes in psychology and in Watson's career . The former led him to redefine and systematize psychology as behaviorism ; the latter made him the first "pop" psychologist .

30

MORRIS AND TODD

1 . Psychology in 191 3 The importance of Watson's empirical and conceptual contributions notwithstanding, they were not outside other advances in psychology at the time (see Boakes, 1984 ; Dewsbury, 1984 ; Leahey, 1992 ; Wozniak, 1993a, 1993b) . In comparative psychology, for instance, both nonhuman and human behavior had bee n carefully observed as early as the 1870s (e .g ., Darwin, 1877 ; Spaulding, 1873) . Psychologists were concentrating more on measurement, research design, experimental control, and quantitative analysis (e .g ., Baldwin, 1895 ; Jennings, 1906 ; Lubbock, 1882 ; Yerkes, 1905, 1907) . Nonhuman behavior was eventually take n into the laboratory, where research could be conducted with greater care, usin g puzzle boxes and mazes (see Small, 1900 ; Thorndike, 1898) . And research results were being interpreted more objectively than before (see, e .g ., Jennings , 1907) . The structuralist program was also being challenged, both internally and externally (Coon, 1993) . Introspective methods were criticized for being illusory , inadequate, and logically flawed (Dodge, 1912 ; Dunlap, 1912) . The concept of consciousness was questioned, with arguments marshaled for dismissing it for being unworkable (Perry, 1904), refining its definition (Dewey, 1906), recastin g it as function rather than as substance (James, 1904), and analyzing its functio n in evolutionary and developmental terms (e .g ., adaptation ; Judd, 1910) . Along with these challenges, the nature and status of psychology as a science were also being called into question (Yerkes, 1910) . The alternative was functionalism (Angell, 1907, 1913), which was more eclectic in orientation (e .g . , Calkins, 1906) and more practical in application (e .g ., to individual differences , the effects of practice ; see Cattell, 1906) . In becoming functional and practical , psychology was becoming a science of behavior (Meyer, 1911), but not quickl y or decisively enough for Watson (1907c, 1910b) . IV. WATSONIAN BEHAVIORIS M

Watson was, of course, aware of the objectivist and functionalist trends in psychology, because his own work was integral to them (Watson, 1903, 1909b ) and his colleagues, including Angell, Baldwin, Dunlap, Jennings, Lashley, an d Yerkes, were advancing these positions . He was cognizant, too, of the criticisms of the structuralist program, which he tentatively applied to the concept of consciousness within functionalism itself (Watson, 1910b, 1913b) but not at first in a strong public stance . Before we turn to the position Watson took, we offer a n overview of his career between 1913 and 1920 . A. WATSON : 1913—1920

Although Watson would offer psychology a new school—behaviorism—h e did not at first alter his research program. He did not need to . His research in


31

comparative psychology, ethology, and psychophysics was already behavioral . With his wife, for instance, he assessed the psychophysical reactions of rats and rabbits to monochromatic light (Watson & Watson, 1913) . He also tested a theory that birds homed on the basis of their sensitivity to infrared light by measuring their spectral sensitivity, but found no evidence for the theory (Watson , 1915) . With Lashley, he continued his research on homing in terns (Watson & Lashley, 1913b, 1915a, 1915b), observed the behavioral development of a monkey they had brought to Bird Key (Lashley & Watson, 1913), and studied the effects of practice on human behavior (Watson & Lashley, 1913a) . His research with nonhumans, though, largely culminated with the publication of Behavior : An Introduction to Comparative Psychology (Watson, 1914) . Afterward, his interests turned toward the analysis of human behavior, with an emphasis on infan t and child development, which he conducted in Adolph Meyer's (1866–1943 ) Phipps Clinic at Johns Hopkins between 1916 and 1920 . He also develope d stronger systematic interests in attempting to reformulate psychology as a natural science . When Watson was selected to become president of the APA, he planned to speak at the convention on thinking as subvocal speech (see Watson, 1913b, pp . 172–174), but he and Lashley could not overcome some technical difficulties involved in measuring interoceptive muscle movements supposedly correlate d with thinking . So instead he read a paper titled "The Place of the Conditione d Reflex in Psychology" (Watson, 1916b), deriving his views from Behkterev and Pavlov (see Behkterev, 1906, 1913 ; Pavlov, 1906) . The conditioned reflex there after became an important, but not exclusive, principle in his analysis of behavior, albeit central to his theory of emotions (Watson & Morgan, 1917) . For Watson, World War I was an opportunity, not a loss . He became among a handful of scientists who would guide science ' s role in government and industr y in the years thereafter (see Buckley, 1989, pp . 99–111) . In the war, he served a s a major in the U .S . Army Signal Corps and conducted research with the Militar y Intelligence Division, applying scientific methods to personnel selection an d training . The war also allowed him to broaden the scope of his research ; for instance, he studied the effects of oxygen deprivation on human performanc e (Watson, 1918) and studied anti-venereal-disease films and their relation to education about sexually transmitted diseases (Lashley & Watson, 1920 ; Watson & Lashley, 1920) . Although he was now studying applied problems in human behavior, Watso n did not shift his publications toward popular venues, but his style of writing di d change . In his first systematic text on behaviorism Psychology from the Stand point of a Behaviorist (Watson, 1919, 1924b, 1929) he described physiolog y and behavior in detail, but supported his overall analysis with examples draw n from everyday human behavior . These examples were designed to convince a lay audience of behaviorism's potential to deal scientifically with human problem s (Keller, 1994) . After 1920, he moved increasingly in this direction . Watson's final research project at Johns Hopkins culminated in the well-

32

MORRIS AND TOD D

known Watson and Rayner (1920) study "Conditioned Emotional Reactions," i n which they classically conditioned a child's fear of a white rat by pairing the ra t with a startling noise . Although the research was methodologically flawed an d ethically questionable (see Harris, 1979), it was a conceptually compellin g demonstration of Watsonian behaviorism and later took on near-mythologica l status (Samelson, 1980) . 3 We turn now from Watson's biography to the behaviorism he founded and elaborated on during the remainder of the decade . B. THE BEHAVIORIST MANIFEST O The most significant event in the development of Watson's behaviorism was , of course, the publication of "Psychology as the Behaviorist Views It" (Watson , 1913b), later referred to as the behaviorist manifesto for its forceful, crusading tone (Woodward, 1948, p . 69 ; see Samelson, 1994) . It was also judged the single most important publication in the first 50 years of the Psychological Review (Langfield, 1943) . We begin by quoting and briefly analyzing its well-know n opening paragraph, after which we expand on three features of Watson's system , as introduced in the manifesto and later extended . These features are the goal o f prediction and control ; the distinction between methodological and metaphysica l behaviorism ; and the nature of consciousness, thinking, and implicit responding . We conclude with a review of Watson's science of behavior, specifically its purpose and purview, methods, and organization of the subject matter. C . "PSYCHOLOGY AS THE BEHAVIORIST VIEWS IT" The first paragraph of Watson's (1913b) "Psychology as the Behavioris t Views It" offers perhaps the best and most concise statement of Watsonian behaviorism, albeit general enough to permit varied interpretations : Psychology as the behaviorist views it is a purely objective experimental branch of natural science . Its theoretical goal is the prediction and control of behavior . Introspection forms no essential part of its methods, nor is the scientific value of its data dependen t upon the readiness with which they lend themselves to interpretation in terms of consciousness . The behaviorist, in his efforts to get a unitary scheme of animal response, recognizes no dividing line between man and brute. The behavior of man, with all of its refinement and complexity, forms only a part of the behaviorist ' s total scheme of investigation . (p . 158 )

Although this statement was consistent with then-current developments i n comparative, human, and philosophical psychology, Watson took a bolder, more far-reaching position than his colleagues and peers (see Burnham, 1968 ; Harrell & Harrison, 1938) . In just 99 words, he had (a) introduced and defined the term 'Watson made a moving picture of the "Little Albert" experiment—the first use of cinematography by a psychologist in the United States . In this and his other use of film, Watson was a pioneer (Harris, 1982) .


33

behaviorist4 (b) declared a new subject matter for psychology, (c) aligned psychology within the objectivist tradition, (d) combined the truth criteria of pragmatism with the technological ideal of prediction and control, (e) rejected introspective methods as the exclusive methods of psychology, (f) Opposed th e requirement that psychology be explained in terms of consciousness, (g) pro claimed that the scope of psychology was the behavior of all organisms, not jus t humans, (h) asserted the continuity of species and the possibility of broadly applicable principles of behavior, and (i) reaffirmed the conceptual unity of huma n and nonhuman psychology. This new school of psychology—behaviorism—was subject to much philosophical, psychological, and public discussion and debate, especially as th e decade progressed (Samelson, 1981), in part because Watson challenged established traditions, was forceful in his style, and was not always unambiguous . The last characteristic was due in part to how Watson presented the main features of his behaviorism, to which we now turn, beginning with prediction an d control . 1 . Prediction and Contro l Beyond emphasizing the importance of objective observation and experimentation, Watson wrote little about the science of psychology per se before 1913 . In his manifesto and afterward, however, he was definitive and consistent abou t its goal : "Its theoretical goal is the prediction and control of behavior" (Watson, 1913b, p . 158) . By theoretical, Watson presumably meant in principle, becaus e prediction and control were not always achievable in practice. What he meant by prediction and control, however, depended on whether Watson was speaking about psychology as a science, a discipline, a profession, or a basis for social engineering and reform (Morris, Todd, & Midgley, 1993) . For him, prediction-andcontrol 5 was pertinent to these areas as, respectively, a means for understanding behavior, promoting behaviorism, professionalizing psychology, and changing cultural practices (see Samelson, 1981) . Thus, as Watson's interests and audiences shifted, so too did his meaning of prediction-and-control . a. The Analysis of Behavio r Watson's research in comparative psychology, ethology, and psychophysic s entailed careful observational analyses of behavior, predicated on the assumption that behavior was lawful and hence predictable . In studying the develop mental psychobiology of rats, the nesting habits of terns, and the spectral sensitivity of birds, he was observing patterns of change, the organization of behavior , and orderly behavioral relations . Describing these patterns, organizations, an d relations was the goal of his science, in which coherence, correspondence, an d prediction were bases for understanding behavior . 4Behaviorism and behavioristic followed the next year (Watson, 1914, p. 1 ; see Schneider & Morris, 1987) . 5We use prediction and control in reference and preference to control, because control implies prediction (e .g ., about change in behavior and the direction of that change) .

34

MORRIS AND TOD D

Most of Watson's research, though, was experimental . He placed great value on demonstrating not just observing the lawfulness of behavior in con trolled laboratory and field situations (Watson, 1904) . In studying problem solving in rats, maze learning in birds, and discrimination learning in monkeys, he was analyzing behavior as a function of physiological and environmental variables, where the prediction of behavior was a function of experimental control , that is (i .e ., where prediction and control was his goal) . Prediction and control was also the pragmatic criterion of "truth" or understanding of these functional relations . That is, prediction-and-control was not only a means for discerning the variables of which behavior was a function but also a means for verifying those functional relations once they were discerned . In Watson's (1924b) words "Behavioristic psychology attempts to formulate, through systematic observatio n and experimentation, the generalizations and principles which underlie man' s behavior" (p . 5) . That is, only through prediction-and-control did Watson thin k behavior could be fully understood. b. Behavioral Synthesis Watson's interests soon broadened and changed, however. He conducted observational and experimental analyses and became interested in behavioral synthesis as well that is, in "the building up of habits from simple reflexes fo r practical and theoretical (and ethical) purposes" (Watson, 1914, pp . 53-54) . Synthesis was also achieved through prediction-and-control : Functional relation s were no longer just observed or demonstrated and verified but now created . In this perspective, Watson (1924b) sounded like Loeb (1900, 1916), making science in part engineering : Every scientist feels that he makes progress in his field to the extent to which he can gai n control over the material with which he works. . . . The psychologist likewise, having chosen human behavior as his material, feels that he makes progress only as he can manipulate or control it. (p . 7)

Although Watson was more interested here in control than analysis, predictionand-control was never arbitrary (i .e ., control for control's sake) . Rather, prediction-and-control was the basis for synthesizing new functional relations . c. Applied Psychology After 1920, Watson increasingly promoted behaviorism, within both psychology and the culture at large . In these contexts, prediction-and-control still served theory through synthesis, but synthesis was now extended toward the solution of practical problems . In this context, Watson was making progress in understanding behavior by extending prediction-and-control outside the laboratory, where i t became a basis for the promise of behaviorism in applied psychology and socia l engineering . Watson's famous 1924 "dozen healthy infants" statement, for instance, appealed to prediction-and-control not only for understanding behavior but also fo r application :


35

Give me a dozen healthy infants, well-formed, and my own specified world to bring the m up in and I'll guarantee to take any one at random and train him to become any type o f specialist I might select—doctor, lawyer, artist, merchant-chief, and yes even beggar-ma n and thief, regardless of his talents, penchants, tendencies, abilities, vocations, and race o f his ancestors . (Watson, 1924a, p . 82 ; 1930, p. 104)

In stating that he could take normal, healthy children and teach them to becom e any kind of professional his audience might select, Watson was calling for critical science . He was criticizing conclusions made about heredity and environment based largely on rational argument and correlational analyses . The former lacked empirical evidence ; the latter lacked experimental control . In this statement, Watson was also making a case for the role of the environment in human development, as well as for the practical implications of behaviorism . Indeed, throughout this period, he took an increasing interest in socia l problems, such as education (Watson, 1917) and the prevention of venereal disease (Watson & Lashley, 1920), in which prediction-and-control was both a conceptual tool for analyzing these problems and a practical means for solvin g them . d. Conclusion Watson's meaning of prediction-and-control depended on context . In the be ginning, he used it in analyzing behavior as a subject matter unto itself ; later, for synthesizing it ; and toward the end, for promoting and applying psychology i n the social arena . These goals were consistent with 19th-century natural science (James, 1892) and did not overly concern Watson's peers (Samelson, 1985) , many of whom were eager for psychology to become a science independent o f philosophy. In addition, these goals were psychology's rationale for seeking in creased support within universities and from private foundations (Buckley, 1989 ) and were inherent in Progressive social reform as well (O'Donnell, 1985) . No one meaning of prediction-and-control sufficed for all the contexts in whic h Watson used it. 2. Methodological or Metaphysical Behaviorism The most distinctive feature of Watson's behaviorism in 1913 was his rejection of consciousness, mind, and image as the subject matter of psycholog y (Bergmann, 1956 ; Heidbreder, 1933, pp . 234—286 ; Woodworth, 1921, pp . 8—10) . Its other features for instance, its objectivism could be found throughout most of the rest of psychology (see Wozniak, 1993a), but its rejectio n of consciousness was what made Watson's behaviorism more than a variant o f objective psychology. The rejection of consciousness, though, did not determine whether Watson was a methodological or metaphysical behaviorist (see Day, 1983 ; Moore , 1981) . As a methodological behaviorist, he could have rejected consciousnes s because, although it might be real, it was not intersubjectively verifiable . Thus , he could study behavior objectively, but had to put consciousness aside ; it wa s not something to which the methods of science could be applied . As a metaphys-

36

MORRIS AND TOD D

ical behaviorist, Watson could have rejected consciousness as an unreal, conceptual construct . On this view, nothing was set aside ; consciousness was a quality of behavior, not a structure or essence . Of course, Watson could simply have refused to address the concept of consciousness and move ahead with his behaviorist program, which he sometimes did (Calkins, 1930 ; Heidbreder, 1933 ; Lashley, 1923) . In the final analysis, Watson was probably a metaphysica l behaviorist, but his position is difficult to discern because most of his argument s were methodological, not metaphysical (Bergmann, 1956 ; Heidbreder, 1933) . a. The Manifesto Even in his manifesto, Watson (1913b) equivocated over whether consciousness was real, but irrelevant, or a conceptual fiction . In the first paragraph, for instance, he suggested that introspection was not "essential," thereby making th e methodological concession that consciousness existed . This stance was then reinforced by the statement : "It is agreed that introspection is the method par excellence by means of which mental states may be manipulated for the purpose o f psychology" (p . 158) . Later in the manifesto, though, Watson offered metaphysical reasons for why consciousness was an unnecessary and obstructionistic construct, pointing out that it led to "the absurd position of attempting to construc t the consciousness content of the [nonhuman] animal whose behavior we hav e been studying" (p . 159) . This process of constructing consciousness was based more on metaphysical suppositions than on "direct observation under experimental conditions" (p . 161) . Watson was concerned that psychology not adop t Romanes's (1882, 1884) position without at least considering its deep methodological difficulties, which even Titchener recognized (see Jaynes, 1976) . Watson's equivocation continued throughout the manifesto . For instance, al though he offered metaphysical points against the concept of consciousnes s arguing that it arose only when simpler processes did not produce satisfactor y explanations he also wrote that, in his own case, "closer examination [of mental imagery] leads me to deny in my own case the presence of imagery in the Galtonian sense" (p . 173) . 6 If Watson's "closer examination" were accomplishe d through introspection, then his seeming acceptance of consciousness mad e sense, because he could not reject introspection if he himself had used it t o 6 In this context, Watson may also have inadvertently established the idea that he was incapable of visual imagery, which Skinner (1959) cited as a possible reason for Watson's rejection of consciousness . Watson (1913b), who was most likely speaking of introspective exercises he conducted whil e teaching from Titchener's textbooks, stated : Until a few years ago I thought that centrally aroused visual sensations were as clear a s those peripherally aroused . . . . However, closer examination leads me to deny in my own case the presence of imagery in the Galtonian sense . (p. 173 ) Watson was, of course, referring to the outcome of exercises in formal introspection . According to Dunlap (1932, p . 45), though, Watson himself reported no problems with informal visual imager y and employed it effectively when designing his apparatus .


37

gather data damaging to the structuralist position (see also Watson, 1913a , p . 423) . Watson's equivocation, though, may not have represented any actual wavering between methodological and metaphysical behaviorism but more a rhetorica l tactic . In 1913, he was interested in establishing and expanding psychology as a distinct discipline . As a systematist, however, Watson an "animal man " could not afford to alienate the structuralists and functionalists, who dominate d organized psychology. That is, he could not, at the time, remake psycholog y completely in his own image because the result would have been an extension o f his animal psychology to human psychology, which would have been unacceptable to his colleagues . By provisionally allowing consciousness, his behavioris m could be accepted as a type of objective psychology by psychologists whose systematic and programmatic interests partially coincided with his own (see Wood worth, 1921, p . 2) . b. The Behaviorist's Standpoin t By the time Watson (1919) published Psychology from the Standpoint of a Behaviorist, his methodological objections to consciousness were accepted throughout psychology (Bergmann, 1956 ; Woodworth, 1921) . The study of animal behavior could easily be objective, avoiding inferences about animal consciousness (see Jennings, 1907 ; Loeb, 1900 ; Sherrington, 1906 ; Thorndike , 1911 ; but see also Kohler, [1917] 1925 ; Washburn, 1908) . In human psychology , the reality of consciousness was still accepted, but its relative inscrutabilit y prompted objective studies of behavior in which it (or cognition, as it was beginning to be called ; see Tolman, 1922 ; see also Leeper, 1951 ; T. V. Moore, 1939) might or might not be inferred as a causal construct . Given the general acceptance of methodological behaviorism, this woul d have seemed the right time for Watson to reassert his metaphysical arguments , but he did not, perhaps because his growing popular audience would not hav e appreciated or understood them . He continued to emphasize the methodological problems, but then adopted a new strategy : He simply declared that behavioris m had banished consciousness and moved on to other issues . As a result, although Watson was praised for his commitment to objectivity, his incompletely developed rejection of consciousness was evidence for the necessity of retaining it . He seemed to have underestimated how deeply pervasive the concept of consciousness was in psychology. He could not banish it without appearing to rejec t a large portion of what was thought essential to psychology (Kohler, 1947 ; Woodworth, 1921) or of what it was to be a human being (Glenn, 1936) . Watson (1928) later offered a variety of conceptual and historical reasons to explain the persistence of the concept of consciousness (e .g ., as an extension of religious beliefs), but by then psychology had largely abandoned structuralism . Abandoning structuralism, however, did not mean that psychology accepte d Watsonian behaviorism . Instead, psychology adopted an approach to the min d that corresponded more closely to the popular, commonsense homuncular con-

38

MORRIS AND TODD

ception of mental functioning (Moore, 1939 ; Tolman, 1922) . This approach di d not make difficult technical distinctions (e .g ., between perception and apperception ; see Wundt, 1912) or require arcane mentalistic arguments (e .g ., about imageless thought ; see Titchener, 1909) . Instead, the mind operated as people generally believed it did, which probably broadened the general appeal of mentalistic psychology. In philosophical psychology, this approach would late r adopt a stance similar to that of the logical positivist movement in the philosophy of science (e .g ., Blumberg & Feigl, 1931) to systematize the transformatio n of vernacular concepts into seemingly scientific ones (see Moore, 1995 ; Smith , 1986) . In the learning theories to come, a qualified acceptance of the fundamental validity and usefulness of vernacular concepts came to guide much theorizin g about intervening variables (e .g ., Tolman, 1932) and hypothetical construct s (e.g ., Hull, 1943) . Watson (1913a, 1913b, 1919, 1924a) had probably hoped that his theory o f thinking as subvocal speech and his analysis of the physical nature of private experience, which we address later in this chapter, would serve as a useful substitute for speculations about consciousness and mind . He might have been better served, however, by a frontal attack on mentalism, but his one attempt to explai n private experience not as conscious content but as perceiving in the absence o f the thing perceived (i .e ., conditioned seeing) was confined to a footnote (Watson, 1924b) : "From the standpoint of theory, this distinction between a visual re action and a "visual image" is important. There is inherent in it the differenc e between a thoroughgoing monism and a thoroughgoing mentalism" . This was Watson's most metaphysically sophisticated statement on the concept of consciousness, but its implications went beyond the popular exposition Watson wa s developing, and he pursued it no further (p. 352) . c. Behavioris m

By the time Watson (1930) published the second edition of Behaviorism, h e probably did not regard consciousness as metaphysically real in any sense . H e not only attempted to undermine the concept of consciousness by citing its origins in prescientific thinking (Watson, 1928, p . 79 ; 1930, p . 2) but also noted that the concept was being invoked in an illogical fashion by affirming the consequent : Psychologists "do not tell us what consciousness is, but merely begin t o put things into it by assumption ; and then when they come to analyze consciousness, naturally they find in it just what they put into it" (Watson, 1930, p . 4) . Even here, though, Watson continued to undermine his metaphysical arguments . First, he continued to deny the utility of consciousness on methodologica l grounds, noting that behavior could be publically observed while consciousnes s could not (Watson, 1930, pp . 5-6) . Second, he apparently dismissed concepts such as mental images solely on the basis of their having been associated wit h structuralism (Dunlap, 1932 ; Heidbreder, 1933) . And third, he dwelled on th e physical attributes of thinking (e .g ., subvocal speech) as his primary avenue of attack on the issue of privacy . The point that thinking could have physical cone-


39

lates was credible in an anatomical sense, but Watson's failure to address suc h topics as the nature of sensation (e .g ., hearing oneself speak when thinking) suggested an extreme denial of a perfectly real phenomenal experience (Heidbreder, 1933) . d. Conclusio n Although Watson generally took the metaphysical position that consciousnes s did not exist as an entity, he never developed an effective case against it . H e seemingly rejected, rather than explained, phenomenal experience and concentrated on arguments about the methodological difficulties inherent in introspection. As a result, confusion arose over whether he was a methodological or meta physical behaviorist . As this volume attests, methodological and metaphysical behaviorism exist in greater variety and subtlety than suggested in Watson's work alone (see Zuriff , 1985) . For instance, methodological behaviorism may (a) accept the phenomena l facts of consciousness and mind but place them outside the purview of scientifi c psychology because they are intersubjectively unverifiable (e .g ., operationism ; see Stevens, 1939) ; (b) reject consciousness as directly verifiable but accept objective inferences of consciousness and mind based on behavioral criteria (e .g . , purposive behaviorism ; see Tolman, 1932) ; or (c) accept the facts of consciousness and mind as behavior, albeit as "private events," which makes this positio n also a metaphysical one (e .g ., radical behaviorism; see Skinner, 1945) . Metaphysical behaviorism, in turn, may range from (a) denials that consciousness a s an entity exists as anything but an epiphenomenon (e .g ., early Watsonian behaviorism) to (b) accepting the facts of consciousness as real, albeit private, events (Skinner, 1945), to (c) ordinary-language analyses of consciousness and mind as descriptive categories of dispositions to behave (e.g ., Ryle, 1949 ; Wittgenstein , 1953 ; see Bloor, this volume ; Place, this volume) . Taking Watson at his word, h e adopted several of these methodological and metaphysical stances at the sam e and different times, all the while leaning toward the latter. To clarify his position more fully, we address a third feature of his behaviorism the relation amon g thinking, language, and implicit responses . 3. Thinking, Language, and Implicit Responses Although Watson rejected consciousness and mind as the subject matter o f psychology, he still had to address the activities historically associated with them for instance, thinking, feeling, and imagining in a behaviorally consistent manner. The received view is that Watson developed a peripheral accoun t that localized thinking in the larynx, emotions in the glands, and imagery in the movements of the eye (see, e .g ., Calkins, 1930, p . 42 ; Heidbreder, 1933) . Watson certainly invited this conclusion by using peripheralist examples that identified generalized activity (e .g ., sexual arousal) with the responses of specific organs (e .g ., genital reactions ; Watson, 1913a), but this was not exactly hi s position .

40

MORRIS AND TOD D

a . The Manifesto When Watson (1913b) published his manifesto, he was a comparative psychologist. Although animal consciousness was still debated, comparative psychologists had little trouble publishing objective studies that avoided any discussion of consciousness (Boakes, 1984) . In the manifesto, then, where Watso n employed examples drawn from research on nonhuman behavior, consciousnes s was not a troublesome issue . In using these examples, though, Watson (1913b , pp . 172, 175) seemingly slighted such topics as thinking and language . When he did address them, he apparently did so as an afterthought, because his treatmen t appeared entirely in a series of footnotes likely added after the main text wa s composed (Samelson, 1994) . The tentative nature of this early theory of thinking and language can be see n in Watson's equivocating over the continuity of research methodology from non human to human behavior. For instance, he asserted early in his manifesto tha t human and nonhuman behavior were of essentially the same character and coul d be studied with the same methods but then later conceded that objective studie s of human perception might also effectively employ "language methods" uniqu e to humans (see Fechner, [1860] 1966) . As Watson (1913b) described it : I should prefer to look upon this abbreviated method, where the human subject is told i n words . . . to equate two stimuli ; or to state in words whether a given stimulus is pre sent or absent, etc ., as the language method . . . . The method becomes possible merely by virtue of the fact that in the particular case the experimenter and his animal have systems of abbreviations or shorthand behavior signs (language), any one of which ma y stand for a habit belonging to the repertoire of the experimenter and his subject . (p . 172)

Here Watson formally adopted the position that words served as substitutes fo r the objects and events to which they referred . Although Watson (1913b) had clarified the role of human verbal responses in the study of perception, and wrote approvingly of the use of nonsense syllable s in objective research on memory (e .g ., Ebbinghaus, [1885] 1913), he admitted in the main text of the manifesto that he was unable to deal effectively with "th e more complex forms of behavior, such as imagination, judgment, reasoning, and conception" (p . 173) . After complaining that his critics were no more effective themselves, he digressed into a lengthy footnote in which he formally introduce d the so-called motor theory of thinking : The hypothesis that all of the so-called "higher thought" process go on in terms of fain t restatements of the original muscular act (including speech here) and that these are integrated into systems which respond in serial order (associated mechanisms) is, I believe a tenable one. (p. 174)

Watson was not the first to make such a proposal . Several versions of the motor theory had already been proposed and would have been familiar to psychologists at the time (see Samelson, 1994, pp . 14-16) . For instance, an article by Pinter (1913), "Inner Speech during Silent Reading," was under review at th e Psychological Review at the same time as Watson's manifesto and may have


41

prompted Watson to include the footnote . Pinter (1913) himself cited an earlie r article "Automatic Movements of the Larynx" (Curtis, [1899] 1900) that re ported on attempts to measure the laryngeal correlates of thinking . In any event, according to Watson (1913b), the motor theory "makes reflective processes as mechanical as habit" (p . 174) and the processes responsible fo r explicit habits also responsible for implicit ones : "The scheme of habit which James long ago described where each return of afferent current releases th e next appropriate motor discharge is as true for thought processes as for overt muscular acts" (p . 174) . Aside from the prima facie appeal of the motor theory to an objective psychologist, adopting an existing theory also allowed Watson to avoid being completely unorthodox . By citing James (1890), he could give additional support t o his unconventional view of thinking by associating it with a then-conventional and widely accepted view of the reflex (see Dewey, 1896) . This tactic would mute critics who might question the upstart Watson but not the well-respected James . Also, when critiques of the motor theory did arise (e .g ., Woodworth , 1931, p . 72), they would not be about the basic tenets of behaviorism (e .g ., th e rejection of consciousness) . Rather, they would be critiques of a theory that Watson could claim was not yet completely worked out . Once his motor theory was introduced, Watson (1913b) used it to dismiss th e principle of ideomotor action, which traditional associationists regarded as th e primary means by which mental events came to initiate behavioral ones (James , 1890) . As an alternative, he argued that sensorimotor action actually accounted for all behavior. Referring to an article by Thorndike (1913), published in th e same issue of Psychological Review as his manifesto, Watson (1913b) noted : Thorndike . . . casts suspicions upon ideo-motor action . . . . I heartily agree with him. I should throw out imagery altogether and attempt to show that practically all natura l thought goes on in terms of sensori-motor process in the larynx . (p . 174)

This was Watson's first specific identification of the larynx as the locus of thinking, an idea he would later retract. In the closing paragraph of this footnote, Watson (1913b) addressed an explicitly behavioral implication of his theory . He suggested that increasing effective thinking could be viewed as the adaptation of private laryngeal responses t o increasingly demanding environmental stimulation : According to my view, thought processes are really motor habits in the larynx, improvements, short cuts, changes, etc ., in these habits are brought about in the same way tha t such changes are produced in other motor habits . This view carries with it the implicatio n that there are no reflective processes (centrally initiated processes) : The individual is al ways examining objects in the one case objects in the now accepted sense, in the othe r their substitutes . . . the movements in the speech musculature . (p . 174 )

The conceptual importance of this statement should not be underestimated . Watson clearly expressed the position that learning, even when it involved behavio r unobservable to others, consisted of environmental events . For Watson, there

42

MORRIS AND TOD D

was no centrally located mind operating independently of the world. Instead , there was an organism interacting with an environment, albeit sometimes a private, but nevertheless real, environment . b. Image and Affection Some reprintings of "Psychology as the Behaviorist Views It" integrate th e aforementioned footnote into the main text or omit it entirely (Samelson, 1994) , which has obscured the hurried and haphazard manner in which Watson introduced his theory of thinking . Because his initial account was cursory and incomplete (Watson, 1913a, p . 421), he published a companion piece later the sam e year "Image and Affection in Behavior" (1913a) in which he attempted t o clarify some of the details . Four points bear mentioning . First, Watson's (1913a) rejection of the image, as conceived by the structuralists, was virtually complete . He had expressed doubts about it in his manifesto but had neither entirely ruled it out nor completely denied the usefulness of introspection . In contrast, in "Image and Affection," Watson (1913a) described th e concept of the image or "centrally aroused sensation" as "the most serious obstacle in the way of a free passage from structuralism to behaviorism" (p . 421) . Fo r a determinist grounded in physiology, Watson could not accept the existence o f hidden central processes operating according to mental principles derived primarily from philosophy. He was also critical of introspective studies that , through their collective unreliability, were unable to provide useful answers t o questions about behavior. All Watson (1913a) offered adherents of the imag e doctrine was a footnote granting images the status of epiphenomena : I may have to grant a few sporadic cases of imagery to him who will not be otherwis e convinced, but I will insist that the images of such a one are sporadic, and as unnecessary to his well-being and well-thinking as a few hairs more or less on his head . (p . 423)

This acceptance of images as epiphenomena, though, was inconsistent with Watson's metaphysical stance concerning consciousness . In his reappraisal of Watson's contributions, the philosopher Gustav Bergmann (1906—1987 ; Bergmann , 1956) correctly noted : [Watson] confused "centrally aroused" with "uncaused ." Images, he concluded, are children of the devil, or, more precisely, of an interacting mind. Thus, he decreed that ther e are none, except perhaps a sporadic few which do not matter . Strangely, he did not see that even a single one would suffice to destroy his case . (p . 274)

This criticism echoes those of prominent psychologists of Watson's time an d later, who took the existence of mind as a given (Calkins, 1913, 1930 ; Jastrow, 1930 ; Kohler, 1947 ; Titchener, 1914 ; Tolman, 1932 ; Woodworth, 1931) and wh o regarded Watson's rejection of it as "silly" at best (Bergmann, 1956, p . 269) or a denial of the fundamental essence of humanity at worst (Glenn, 1936 ; King, 1930) . Second, Watson (1913a) introduced the term implicit behavior to describe subvocal speech and other covert activities (pp . 423 -424) while he used explicit


43

to describe behavior with observable dimensions . This implicit–explicit dimension based on physical scale was designed primarily to deal with the problem o f privacy as defined by the structuralists, but Watson's description of it in physica l terms trapped him conceptually, because he was unable to fully explore the functional implications of the concept of implicit events . Later, more sophisticated analyses would substitute Watson's implicit–explicit distinction with a view i n which psychological privacy was defined by the relative accessibility of an even t by specified observers (Skinner, 1945) . Third, Watson (1913a) introduced the possibility that implicit response s might intervene in explicit responses . He wrote, for instance, that "[w]here explicit behavior is delayed (i .e ., where deliberation ensues), the intervening time between stimulus and response is given over the implicit behavior (to `thought processes')" (pp . 423-424) . Watson (1913a) did not expand on this point furthe r but would reiterate it in subsequent publications (e .g ., Watson, 1928 , pp . 82–92) . Fourth, Watson (1913a) offered probably his most direct criticism of the obstructionistic effect of mentalistic psychology . Traditional psychologists, he argued, exhibited a tendency to invest the brain with unobservable powers an d processes that corresponded, not uncoincidentally, with what needed to be explained : "The tendency to make the brain itself something more than a mechanism of coordinating incoming and outgoing impulses has been very stron g among psychologists, and even among psychologically inclined neurologists " (p. 424) . Watson's concern was not simply with obstructionism, however . The structuralists had claimed that the study of thinking and other mental processes required special techniques, such as introspection, primarily because thos e processes occurred entirely in the brain and possessed unique, nonphysical at tributes (e .g ., clearness and extensity; see Titchener, 1929) . Watson (1913a) argued that mentalistic explanations of this sort were due to the unconscious retention of prescientific preconceptions about the mind and behavior : When the psychologist threw away the soul he compromised with the conscience by setting up a "mind" which was to remain always hidden and difficult to access . The transfe r from the periphery to cortex has been the inventive for driving psychology into vain an d fruitless searches of the unknown and unknowable . (p . 424 ; see also Watson, 1928 , pp . 79—80)

c . Thinking and Language Late r Watson's position on thinking and language did not change significantly afte r its formulation in "Image and Affection in Behavior ." He expanded his treatment into chapter-length presentations, however, and shifted his emphasis away fro m technical arguments to everyday examples . In Behavior : An Introduction to Comparative Psychology (Watson, 1914), for instance, he included reworke d versions of his 1913 articles (see Lovie, 1984), as well as a chapter in which h e identified the ability to speak as the primary difference between humans and

44

MORRIS AND TOD D

nonhumans . In Psychology from the Standpoint of a Behaviorist, Watson (1924b, pp . 331–367) devoted a chapter to the development and retention of explicit and implicit language habits, but his use of detailed medical diagrams and technica l descriptions of the vocal apparatus, along with the addition of the conditione d reflex as a behavioral mechanism, suggested firmer empirical grounding for hi s theory than was the case . Watson did however, distance himself from his earlier identification of thinking as "laryngeal habits" by addressing material on nonlanguage forms of thinking such as manual sign language and on the effects of larynx removal, which did not, of course, impair thinking . In Behaviorism, Watson (1924a, 1930) included two chapters on talking , thinking, and thinking without words, in which he sought to clarify his positio n and then move on . His point of clarification concerned laryngeal movements : I wish here expressly to affirm that in developing this view I never believed that the laryngeal movement . . . . as such played the predominating role in thought . I admit in my former presentations I have, in order to gain pedagogical simplicity, expressed myself i n ways which can be so interpreted . We have all had the proofs before us time and agai n that the larynx can be removed without completely destroying the person 's ability to think . (1924a, p . 191 ; 1930, p. 238 )

In moving on, Watson described two implications of his theory of thinkin g and language, specifically with respect to having a repertoire of "word substitutes ." The first dealt with the efficiency of action : "Think of what it means in the economy of time and ability to call out cooperation from groups to hav e word substitutes for objects common to all members" (1930, p . 234) that is, to be able to refer to events verbally . The second implication concerned cover t problem solving : "Many of our discoveries come largely through this ability to manipulate a world of objects not actually present to our senses" (1930, p . 234) for example, when one "thinks through" a problem . However, Watson' s suggestion that in speaking each word elicits each subsequent word created conceptual difficulties in accounting for the number and variety of specific stimulus response relations that must be conditioned for ordinary speech . As Lashley (1951) pointed out later, chaining theory was untenable for physiologica l reasons . d. Conclusion Even though the structuralists' conception of consciousness became untenable over the course of Watson's career, the existence of consciousness as a t least "self-awareness" was too obviously true to abandon (Heidbreder, 1933) . Thus, Watson could not avoid dealing with consciousness ; however, his theory of implicit behavior was not convincing to most of his contemporaries (se e Calkins, 1930 ; Jastrow, 1930 ; Lashley, 1923 ; Woodworth, 1931) . Although these critics praised Watson's advocation of objectivity, they found his rejection o f consciousness excessive (e .g ., Dunlap, 1932) . Watson also seemed to have an in tolerance for anything associated with introspective psychology . Heidbreder (1933) described this intolerance as Watson's unwillingness to grant organisms

45


the ability to respond to internal events as they did to external ones, even thoug h he clearly allowed attribution (see also Calkins, 1930; Jastrow, 1930) . Watson had gone too far ; his behaviorism was too radical . But perhaps Watson did not go far enough in describing the implications o f his theory . Although he was trying to deal with "self-awareness" as people' s identification of their implicit behavior, his preoccupation with its structural features left him little opportunity to explore how self-awareness might be conceptualized as responses to one's own behavior, for instance (e .g ., hearing what seems to be one's voice during thinking) . As it was, Watson's (1919) most sophisticated attempt to account for private experience as behavior was relegate d to a footnote on the image in Psychology from the Standpoint of a Behaviorist, most of which involved reductionistic speculations about the neurophysiologica l origins of stimuli identified as images . Only in the footnote's final sentences did Watson (1919) offer the concept of seeing as a conditionable behavior, distinguishing the "visual reaction" a behavior from the "visual image" —a construct . Apparently, no one noticed this distinction, and unfortunately Watson neve r developed it in a manner that might have prevented the understandable conclusion that he was merely substituting muscular movements and glandular secretions for mental activity (Bergmann, 1956) . In the end, though, Watson advanced a number of important points about thinking, language, and implicit behavior, among them the concept that words serve stimulus functions in behavior rather than conveying ideas and the view that covert problem solving is physical activity . These were points that the behaviorisms to follow would develop more fully. D. A SCIENCE OF BEHAVIO R

Up to this point, we have described the historical, intellectual, and cultura l contexts from which Watsonian behaviorism evolved, some of its general characteristics, and three of its main features . We have not, however, described Watson's science per se. Although his science was largely underdetermined by behaviorism as a system of psychology, it reveals still additional details taken to b e distinctive of it, albeit less central than those presented thus far. For this reason , we review his science more briefly, organizing it in terms of its purpose an d purview, its methods, and its systematic organization of behavior as a subjec t matter for a natural science . Throughout this material, we draw on example s taken from Watson's (1919, 1930) own analyses . 1 . Purpose and Purview As for the purpose and purview of psychology as the behaviorist views it , Watson (1919) wrote : Psychology is that division of natural science which takes human activity and conduct a s its subject matter. It attempts to formulate through systematic observation and experimen -

46

MORRIS AND TOD D

tation the laws and principles which underlie man ' s reactions . Every one agrees that man's acts are determined by something, and that, whether he acts orderly or not, ther e are sufficient grounds for his acting as he does act, if only these grounds can be discovered. (p . 1 )

Here, again, we see that prediction-and-control was not, in itself, the goal o f Watson's science . For Watson (1919) : "As a science psychology puts before her self the task of unraveling the complex factors involved in the development o f human behavior from infancy to old age, and with finding the laws for the regulation of behavior" (p . 8) . Prediction-and-control was the means for findin g these laws and principles, after which psychology could apply them to individua l behavior and social engineering . a . Responses and Stimuli The laws for the regulation of behavior were the laws of "adjustment, " which Watson (1919, pp . 8 -14 ; 1930, pp . 11—16) conceptualized as relations among responses and stimuli . As for whether Watson's responses and stimuli, and thei r relations, were molar or molecular, he has been interpreted both ways (see, e .g . , Herrnstein, 1967 ; Kimble, 1997) . On the one hand, Watson (1919) stated that th e goal of psychology was the "ascertaining of such data and laws that, given th e stimulus, psychology can predict what the response will be ; or, on the othe r hand, given the response, it can specify the nature of effective stimulus" (p . 10 , emphasis in original) . On the other hand, although he had borrowed the terms response and stimulus from physiology, he did not restrict their meaning to physiology alone . A response encompassed acts defined in terms of natural-language categorie s of action (e .g., talking, swimming) or action to some end (e.g ., writing a letter, building a house) . As we have seen, responses could also be explicit (e .g ., playing tennis) or implicit (e .g ., thinking), as well as hereditary (i .e ., unconditioned, such as sneezing) or habitual (i.e ., conditioned, such as staying on good term s with members of your gender) . Watson's (1930) occasional classification and examples of end-organ activity notwithstanding, he believed that "the behavioris t is primarily interested in the behavior of the whole man" (p . 15) . As for stimuli, they generally encompassed situations, also defined in terms of natural-languag e categories (e .g ., a lecture hall) or their effects (e .g ., injury) . Stimuli, too, could be either explicit (e .g ., street noises) or implicit (e .g ., muscle movement) . To this analysis Watson added that adjustments were complex . Acts depen d on an individual's history hereditary, organic, and behavioral such that an y one stimulus can "call out" more than one response (i .e ., have multiple effects), both across and within individuals . As for situations, they comprise multipl e stimuli, some of them conditional ; as such, causation is multiple and context dependent . Although Watson sometimes wrote that adjustments could be reduce d to simpler units (e .g ., reflexes), he also contended that lawfulness and order could be discerned at more complex levels of adjustment . By many accounts, then, Watson's analysis was not reductionistic (e .g ., Kitchener, 1977 ; Malone,


47

1990 ; Wozniak, 1994), although it was sometimes unsophisticated and inconsistent . 2. Methods Watson (1919, pp . 24—47 ; 1930, pp . 20—47) described a wide range of objective methods and procedures for psychological research, organizing them into five categories . The first was observation, either unaided or enhanced through instrumentation, but in either case systematic, such that the results might be subjected to analysis through descriptive statistics . Instrumentation referred to apparatuses that improved observation, such as measuring responses (e .g ., stop watches), presenting stimuli (e .g ., chronoscopes), and describing the setting (e .g ., laboratory illumination, hours of food deprivation) . The second class of methods was the conditioned reflex method, as employed by Pavlov (1906) in the analysis of conditioned glandular reflexes and by Behkterev (1913) in the analysis of conditioned motor reflexes . The third class was the verbal report. Watson did not deny, for instance, that we could observe both explicit and implicit responses and stimuli and offer reports o f them . These responses were, in part, records of an individual's adjustmen t to the environment, no matter what the dimension, size, or location of th e adjustment . The fourth class of methods was testing, which included (a) test s for determining general levels of behavior (e .g ., the Binet— Simon test o f intelligence) and (b) tests to determine special abilities (e .g ., arithmetic, telegraphy), especially in personnel selection . The final class was social experimentation, some of it experiments in nature, including the passage of law s (e .g ., Prohibition) and governmental reform (e .g ., the fall of the Russian monarchy) . 3. The Organization of Behavio r Watson organized behavior in a variety of ways . Some of them involved formal categories, such as the origins and loci of behavior, while others were functional classifications concerned with the laws and principles of behavior . Still others classified behavior according to its domain or content, described in ordinary language terms . a. Formal Organization Watson's (1919, pp . 14—15, 194—202, 231—134, 269—275, 1930, pp . 16—18) formal classification of behavior included the distinction between heredity responses (i .e ., unlearned responses, such as unconditioned emotional responses, and reflexes) and habits (i.e., learned responses, such as conditione d emotional responses and thinking) . Heredity itself could be either instinctive (e .g ., the grasping reflex) or emotional (e .g ., love, fear, and rage) . Other forma l distinctions involved (a) whether behavior was explicit (e .g ., typewriting) or implicit (e.g ., subvocal "mental" arithmetic) and (b) whether it was manual (e.g .,

48

MORRIS AND TOD D

eye–hand coordination), laryngeal (e .g ., talking aloud, thinking to oneself), o r visceral (e .g., emotional reactions) . b. Functional Organization Watson (1919, pp . 298–300) also classified behavior according to function that is, according to the laws and principles of behavioral regulation . Because of the multiple effects, multiple causation, and conditionality of an individual's behavioral adjustments, behavior was exceedingly complex but not undetermined. The determiners, though, were simpler and more orderly, of which Watson liste d six, encompassing stimulus, the current context, and behavioral history . These determiners were (a) the previous, most recent response to stimulu s (recency) ; (b) the most frequent response to a stimulus (frequency) ; (c) the response most "closely connected" to the overall situation or general context (e .g . , responses appropriate to Sundays but not Saturdays) ; (d) the emotional tension or arousal imparted by a situation, affecting what response will be called out b y a stimulus (e .g ., polishing versus brandishing a gun) ; (e) temporary intraorgani c factors (e .g ., headaches) ; and (f) the individual's life history, which included th e history of all of the foregoing events and their effects . Although Watson cited and appealed to classical conditioning in the traditio n of Pavlov (1906 ; Yerkes & Morgulis, 1909) and Behkterev (1906, 1913), his ac count and analysis of behavior in terms of the conditioned reflex was often loos e and analogical (see Malone, 1990, pp . 98-101) . He more commonly referred to the first three determiners listed earlier recency, frequency, and stimulu s change in accounting for behavior. These three determiners also accounted fo r the relation between responses and their consequences, but he otherwise rejecte d Thorndike's (1898, 1911) law of effect for being fanciful and teleological (Watson, 1930, p . 206) . c. Domain and Content Finally, Watson classified behavior in terms of natural-language categories, among the most often discussed being emotion (Watson, 1919 , pp . 194–230 ; 1930, pp . 140–195) ; manual, motor, or bodily habits (Watson , 1919, pp . 269–309, 1930, pp . 196–223), language (Watson, 1919 , pp. 310–347 ; 1930, pp . 224–251) ; thinking (Watson, 1930, pp . 224–268) ; memory (Watson, 1919, pp . 301– 308) ; work, both academic and vocational (Watson, 1919, pp. 348–391) ; and personality (Watson, 1919, pp . 392–420 ; 1930, pp. 269–304) . 4 . Conclusion In developing a system of psychology, Watson addressed both metatheoretica l and empirical issues, and his behaviorism encompassed both . As for his science , it was not restricted to molecular as opposed to molar analyses, habits as op posed to heredity, explicit as opposed to implicit behavior, or motor as oppose d to mental activity. His metatheory was consistent with these details but would

49


not have been inconsistent were they otherwise, as long as they were objectivel y and empirically grounded .

V. AFTER WATSONIAN BEHAVIORIS M A . WATSON : 1920–193 6

During Watson ' s work on "Conditioned Emotional Reactions " with his student Rosalie Rayner (b . 1899), the two became romantically involved . Their relationship caused a scandal within the Johns Hopkins community, and in October 1920 the university forced Watson to choose between his academic position and Ms . Rayner. He chose Ms . Rayner and resigned. Watson and his first wife, Mary Ickes, were then divorced, but the proceedings caused a public scandal, i n part because both women came from prominent families (Buckley, 1994) . Afterward, Watson and Rayner were married and became active, social New Yorkers , with Watson even profiled in The New Yorker. He also became a father again, thi s time of William Rayner (Billy) (1921–1954) and James Broadus (Jimmy ) (b . 1924) . Watson's resignation from Johns Hopkins was a resignation from academi c psychology . With his public notoriety, and without sympathetic support fro m colleagues and friends, academic positions were unavailable to him . So he turned to the world of business specifically, to advertising and in 1921 took a permanent position with the J . Walter Thompson advertising agency in Ne w York, rising quickly to a position as vice president . He was a success in his second career (Buckley, 1989, pp . 134–147) . Although embittered about leaving the university and frustrated by his inability to conduct further research, Watson still wrote about psychology . Over the next 10 years, he published numerous articles for popular magazines and industry publications, as well as three new books Behaviorism (1924a, 1930), The Ways of Behaviorism (1928), and Psychological Care of Infant and Child (Watson & Watson, 1928) . In these books, he continued to promote his views on consciousness and mind ; he refined his position on the contributions of heredit y and environment to behavior ; and he began advocating for broad-scale social engineering (Buckley, 1989, pp . 148–176) . He also became the first "radical behaviorist" and his behaviorism the first "radical behaviorism" (Schneider & Morris, 1987) . Mary Calkins (1863–1930) coined these terms to describe wha t she saw as Watson's extreme views (Calkins, 1921) . ' In this section, we turn to a fourth feature of Watson's behaviorism his position on heredity and environment because the received view of his position 'Today, radical behaviorism refers to Skinner's behaviorism (Skinner, 1945, p . 294) . Here, behaviorism is the philosophy of his science (Skinner, 1974, p . 3) and radical denotes "thoroughgoing"

or "root" in the sense that Skinner did not deny events taking place within the skin of the organis m but conceptualized them as more behavioral (Zuriff, 1985) .

50

MORRIS AND TOD D

is generally drawn from his writings during this period . We then recount Watson's writings on child rearing and his work in advertising . We conclude by de scribing the final years of Watson's career and his life. B . HEREDITY AND ENVIRONMENT Watson's position on heredity and environment is often described as radically environmentalistic, a position in which most behavior is learned and genetic s play a negligible role (see Todd, 1994 ; Todd & Morris, 1992) . This view is generally derived from what Watson (1924a, 1930) wrote in Behaviorism . When h e introduced behaviorism in 1913, however, he was not committed to an environmentalist stance and accepted both heredity and environment as determiners o f behavior. Some aspects of this position changed over the course of his career , but others did not. Indeed, in some cases, the importance Watson attached to genetically based species differences actually increased . This assertion may seem paradoxical, given Watson's (1924a, p . 82) "dozen healthy infants" statemen t quoted earlier and his eventual rejection of the concept of instinct (Watson , 1930, pp . 93-139), but for Watson the lack of genetically based instincts in humans was itself a product of genetics (see Skinner, 1959) . 1 . Before Behaviorism (1924) Although Watson (1913b, 1914) did not reject genetic or instinctive contributions to human behavior in the early part of his career, he did give experience th e primary role (Boakes, 1984, 1994 ; Dewsbury, 1994 ; Todd & Morris, 1986) . Hi s doctoral dissertation, for instance, demonstrated that learning could occur i n newborn rats, a finding he then generalized to newborn humans (Watson, 1903) . Heredity, however, played a role. Instincts, he wrote, "determine in large measure our choice of companions, occupations, and our pleasures" (Watson, 1912a, p . 381) . And, in his manifesto, Watson embraced the continuity assumption of evolutionary biology, stating also that "[t]he psychology which I should attempt to build up would take as its starting point, first , the observable fact that organisms, man and animal alike, do adjust themselves to the environment by means of hereditary and habit equipments" (Watson, 1913b, p . 167).

A year later Watson (1914) defined instinct as "a combination of congenital responses unfolding serially under appropriate stimulation" (p . 106) and defende d it as being "short, useful, and convenient" (p . 106), despite its "abuse" by others as a catch-all explanation for human behavior (see, e .g ., James, 1890 ; Thorndike , 1904) . Watson eventually abandoned this justification for the term instinct and began citing the lack of hard data to support the hereditarian viewpoint as his reaso n for assigning a larger role to the environment . The hereditarians had conducte d no systematic research to support their position directly, whereas much of Watson's research on infants and children was designed, in part, to determine the rel-


51

ative contributions of heredity and environment (Watson, 1930, pp . 93–139) . Watson's practice of conducting longitudinal studies probably derived from th e importance he attached to such techniques for studying instinctual behavior i n animals (Watson, 1908a ; see Boakes, 1984, 1994 ; Dewsbury, 1994 ; Todd & Morris, 1986) . According to Watson (1914), a detailed developmental analysi s was a prerequisite for understanding the relative contribution of heredity to behavior : It is because of the intimate connection between instinct and habit that one most desire s to get a clearer knowledge of the animal ' s repertoire of perfected instinctive responses , his partial and incomplete adjustments, and have his tendencies towards adjustment be fore beginning controlled work upon habit formation . (p . 44 )

By 1916, Watson was conducting research on the behavior of human infant s (e .g ., Watson & Morgan, 1917 ; see Samelson, 1994) and, perhaps as a result o f his observations, was beginning to doubt the role of instincts in human behavior . At the time, the nature–nurture debate was in full swing, in both psychology and the culture at large, with important implications for immigration laws, rac e relations, and women's rights (Gould, 1981) . Thus, Watson could hardly write about human behavior without addressing the issue of heredity, especially be cause his developmental research with human infants was making him a recognized authority in these matters . The position Watson adopted at this point was that humans might not exhibit complex "pattern instincts" (e .g ., nest building and mating rituals) as do othe r animals (Watson & Morgan, 1917 ; see Watson, 1914) . This position, though , was not dictated by his social agenda . He had done the requisite research and also knew that one of the important conceptual bases for the belief in human instincts recapitulation theory (i .e ., that embryological development retraces the adult stages of the organism ' s evolutionary ancestors) was empirically untenable (Watson, 1930, p . 287 ; see Gould, 1977), despite its wide adoption by developmentalists (e .g ., Freud, 1930 ; Hall, 1904) . Thus, although Watson (1919) continued to use the term instinct to refer to all unlearned behavior, he claimed that humans did not possess pattern instincts . In addition, he took the position that genetic differences among humans had littl e relevance to their behavior : It lies well within the bounds of probability that if we were able to obtain a new-bor n baby belonging to the dynasty of the Pharaohs, and were to bring him up along with th e other lads in Boston, he would develop into the same kind of college youth that we fin d among the other Harvard students . (p . 12)

2. After Behaviorism (1924 ) In Behaviorism, Watson (1924a) continued to assert that most or all comple x human behavior is learned and that all normal healthy human infants have virtu ally identical behavioral potentials, barring inherited structural differences (Watson, 1924a, pp . 74-107) . He also began to argue that the term instinct had outlived its usefulness .

52

MORRIS AND TOD D

a. Empirical Concerns As for his views on the behavioral potential of humans, Watson's (1924a , p . 82 ; 1930, p . 104) rhetorically forceful "dozen healthy infants" statemen t seems to have been mistaken for a manifesto-like summary of his overall position. But Watson was not denying that genetics played a role in behavior : To th e contrary, he was arguing that inheritance was responsible for the prevalence o f instincts in nonhumans, as well as the relative lack of them in humans . O f course, he may have been wrong, but that was an empirical issue, not a conceptual feature of his behaviorism. In any event, Watson had come to reject the continuity assumption he had asserted in his manifesto (Logue, 1978) and instea d adopted the view that species were likely to be uniquely adapted to their particular environments . For instance, in discussing the differences between the result s of his doctoral dissertation using rats and a replication using guinea pigs (Allen , 1904), Watson (1924a) wrote that "[s]uch a wide divergence of birth equipmen t in the two animal species so closely related (both rodents) proves how unsafe i t is to generalize, on the basis of infra-human studies, as to what the unlearne d equipment of man is" (p . 87) . In a later article, Watson (1927b) reiterated thi s point : "Just because [man] has an evolutionary history is no proof that he mus t have instincts like the stock from which he sprang" (pp . 228-229) . In addition, although the "dozen healthy infants" statement was intended to express Watson's view of the power of the environment and the practical possibilities of behaviorism, it was not central to his analysis of the contributions o f heredity to behavior. Watson's contemporaries made relatively little of this statement, and it did not become a standard feature of textbook presentations of Watson until the 1950s (Todd, 1994) . Indeed, Bergmann (1956) labeled Watson's environmentalism a "pseudo methodological tenet" of his behaviorism . Instead, the statement allowed Watson to reiterate the point that the hereditarians lacked data to support their views and served as a transition to his call for a research-based approach to replace the culturally derived assumptions tha t guided the hereditarian viewpoint . In material just preceding the "dozen healthy infants" statement, Watson (1924a) wrote : Our conclusion then, is that we have no real evidence of the inheritance of traits. . . . But let one adopted child who had a bad ancestry go wrong and it is used as incontestabl e evidence for the inheritance of moral turpitude and criminal tendencies . As a matter of fact, there has not been a double handful of cases in the whole of civilization of which records have been carefully enough kept for us to draw any such conclusions—menta l testers, Lombroso, and all other students of criminality to the contrary notwithstanding . (p . 82 ; Watson, 1930, pp . 103—104 )

Immediately following the statement is Watson's (1924a) overlooked cavea t concerning the hereditarians' lack of data : "I am going beyond the facts and I admit it, but so have the advocates of the contrary and they have been doing it fo r many thousands of years" (p . 82 ; Watson, 1930, p . 104) .


53

b. Conceptual Concern s Watson also had conceptual concerns about instincts as determiners of huma n behavior. In Watson's time instincts were invoked to explain a wide variety o f human activities and emotions, including climbing, imitation, emulation, rivalry , pugnacity, anger, resentment, sympathy, hunting, acquisitiveness, kleptomania , cleanliness, shame, and love (James, 1890) . As hidden internal motivators, these and other instincts could be invoked at will to explain any behavior without otherwise analyzing it . In addition, this was a concern with the teleological implications of instincts, where instincts were conceptualized as strivings for goals an d purposes and hence were mentalistic . As with Watson's other views, this was no t a unique perspective in psychology, but was consistent with already-availabl e critiques of the instinct doctrine (e .g ., Bernard, 1921 ; Dunlap, 1919 ; Kuo, 1922) . In Watson's (1924a) view, innate behavior occurred not because some mysterious internal force impelled it but because of the interaction of an organism' s biological structure with its environment . He explained this interaction with an analogy involving an extension of Loeb's (1907, 1918) theory of tropisms i n which behavior was accounted for as the actions of a physiochemical mechanism stimulated into action by environmental events : [I]f we need no mysterious way of accounting for the motion of the boomerang, if th e laws of physics will account for its motions—cannot psychology see in this a muc h needed lesson in simplicity? can it not dispense with instincts? can we not say that man i s built of certain materials put together in certain complex ways, and as a corollary of the way he is put together and of the material out of which he is made—he must act (unti l learning has reshaped him) as he does act? (p . 86 ; emphasis in original )

3 . Conclusion In contrast to standard accounts, radical environmentalism was not a fundamental feature of Watson's behaviorism . In fact, Watson explicitly accepted th e role of genetic factors in behavior, but rejected instincts as conceptual causes . Rather, an organism inherited a biological structure that interacted with its environment to produce behavior. Changing either the structure or environment o f the organism through learning, injury, growth, and other factors changed behavior. This view rested on Watson's own research and the views of other scientists of his day . C . APPLICATIONS TO CHILD REARING AND ADVERTISIN G After Watson left Johns Hopkins, he conducted no further research in comparative psychology and child development, except for supervising the importan t studies by Mary Cover Jones (1896–1987) on the elimination of children's fear s (Jones, 1924a, 1924b) . These studies were sequels to "Conditioned Emotiona l Reactions" (Watson & Rayner, 1920) and are now considered landmarks in the

54

MORRIS AND TOD D

history of behavior therapy (Krasner, 1988) . Watson continued to write about behaviorism, of course, but he turned more to applied psychology, mainly chil d rearing and advertising . Although expert in the latter (Coon, 1994), he would later admit that he did not know enough about the former child rearing t o write what he wanted to (Watson, 1936) . Watson's interest in applied psychology was apparent early in his caree r where it was nurtured at the University of Chicago by leading figures in progressive, pragmatic philosophy and science (e .g., Dewey, Loeb) . In his dissertation , for instance, Watson concluded by generalizing his findings on early learning t o human infants . In his manifesto, he wrote approvingly of the applied branches o f psychology, and continued to advocate for them throughout the decade (Watson , 1917) . World War I offered him opportunities to conduct applied research (e .g . , Lashley & Watson, 1920 ; Watson, 1918) . After he left university settings, Watson would become an applied psychologist . 1 . Child Rearing Watson did not conduct any research with children until his work with huma n infants in Meyer's laboratory in 1916 (Buckley, 1989 ; Samelson, 1994) . Initially, he was interested in their unlearned reactions to various stimuli, but he eventually focused on their emotional development (Watson, 1919 ; Watson & Morgan , 1917) . Watson published prolifically in these areas and consistently included data from his infant studies in his books (Watson, 1919, 1924a, 1924b, 1930) . Although his theory of emotional development was oversimplified (see, e .g . , Sherman, 1929), it was reasonably well received by his colleagues, and Watso n was soon regarded as an expert in child development (Buckley, 1989, pp . 148–176) . This research also opened up new avenues for Watson's popular articles, be ginning with "Practical and Theoretical Problems in Instinct and Habits" (Watson, 1917), published in Suggestions of Modern Science Concerning Educatio n (Jennings, Watson, Meyer, & Thomas, 1917) . Here Watson summarized his work on unconditioned behavioral reactions, handedness, and emotional development ; complained about the uselessness of IQ tests for education (e .g ., saying that they classified but did not teach) ; and called for greater respect for the environment' s impact on human behavior from before birth into old age : "I sit and watch with amazement the middle aged and aged given no mean exhibition of the fact th e human race is never too old to learn" (pp . 94–95) . Although Watson had rejected Thorndike's (1898) law of effect, many of hi s analyses of child rearing appealed to the significant role of rewards and punishments (Thorndike, 1913 ; see Watson, 1917, 1924a, pp . 144–147 ; 1930, pp. 180-186) . Moreover, although he took a nonmediational approach in his earlie r research, in his child rearing work Watson suggested that rewards and punishments might produce emotional responses that lead in turn to changes in behavior (Watson, 1917, pp . 96–97 ; cf. Amsel, this volume) . Watson would write more than a dozen articles on child rearing and related


55

topics for magazines such as Cosmopolitan, Harper's, and McCall's, eventually reworking them with Rayner into their book, Psychological Care of Infant and Child (Watson & Watson, 1928) . But, unlike the more practically oriented parenting books of the day (e .g ., Gesell, 1930 ; Thorn, 1927), relatively little of advice was systematic or directly applicable . They were less interested in teachin g parents about diapering and feeding than in the emotional adjustment of childre n as they grew into adulthood . Like Freud (1856–1939), the Watsons suggested that many psychologica l problems had their roots in childhood (Mead, 1976) . According to them, for instance, "invalidism" the tendency of adults to make vague physical complaints—was caused by too much "coddling" in childhood (Watson & Watson, 1928, pp. 69–87) and could be prevented by treating the child as an adult . Kissing, hugging, and other open displays of affection should therefore be minimize d to reduce unnecessary conditioned emotional reactions . As for temper tantrums , they could be traced to children's having associated everyday events with overl y restrictive clothing and environment . Tantrums, then, could be prevented by providing looser clothing and teaching the child greater self-sufficiency (Watson & Watson, 1928, pp . 88–112) . Watson and Watson (1928, pp . 152–183) also advocated an honest, open approach to sex education, although an approach still tied to the times . Like most academically based authors, they advised that masturbation was a normal reaction to stimulation of the genitalia and caused no physiological harm. They warned, however, that although relatively harmless, it should be minimized be cause it interfered with normal sexual socialization . Children should be encouraged to interact with peers of both genders . As misguided or quaint as some of the Watsons' advice may be today, it wa s not unique or radical in its time . For instance, although the Watsons warne d against hugs and kisses, they were also early opponents of corporal punishmen t (e .g ., spanking ; see Watson, 1924a, pp . 144–147 ; 1930, pp . 180–186 ; Watson & Watson, 1928, p . 62) . Some of their advice was also well researched, specificall y their recommendations on the importance of information-based sex education a s an effective means of preparing teenagers to deal appropriately with sexual situations (see Watson & Lashley, 1920) . Finally, most of what the Watsons wrot e about could be found in books written from other perspectives (Mead, 1976) , some of them offering harsher advice than anything the Watsons suggested (e .g . , pinning boys' sleeves to the mattress at night to prevent masturbation ; Fenton, 1925, pp . 284–285 ; Thorn, 1927, p . 280) . Some of the Watsons' other advic e was congruent with more modern viewpoints (e .g ., avoidance of corporal punishment ; Spock, 1988) . Although Psychological Care of Infant and Child sold well, and Watson wrote for widely distributed magazines (e .g ., Watson, 1922b, 1927a), the impac t of his advice is difficult to assess . Psychological Care of Infant and Child wa s but one of many books on child care, parenting, and "hygiene" available at th e time (e .g ., Fenton, 1925 ; Gesell, 1930 ; Thorn, 1927 ; see Chicago Association for

56

MORRIS AND TODD

Child Student and Parent Education, 1926, 1928) . Also, by the 1940s, Arnold Gesell (1880–1961) and his colleagues (e .g., Gesell & Ilg, 1943) had proposed a normative-maturational theory of child development that would be much mor e popular than the Watsons' approach . The Watsons' book remains in print only a s a historical curiosity. In any event, little of its advice was derived from behavior ism. 2 . Advertising At J . Walter Thompson, Watson was instrumental in developing several important advertising campaigns for personal care items in the 1920s, includin g Johnson & Johnson's baby powder, Pebeco toothpaste, and Pond's cold cream (Coon, 1994) . He also wrote articles for the firm's in-house publications and fo r The Coca Cola Company, some of them describing findings from market re search, for instance, "What Cigarette Are You Smoking and Why?" (Watson, 1922c ; see Watson, 1922a, 1927c) . Later, he contributed articles that showe d how behaviorism might be applied to advertising and business, among the m "Our Fears : How they Develop" (Watson, 1934) for Red Barrel, Coca Cola' s journal. This would be Watson's last known published article . When Watson entered the industry, advertising was reshaping itself, particularly through the introduction of testimonial advertising and the "fear-sex emulation" formula, which appealed to emotions rather than reason (Rorty, [1934 ] 1976, p . 242) . Watson had no fundamental impact in this regard, however, be cause testimonial advertising and appeals to emotion appeared before he entere d the industry (Coon, 1994 ; contra Buckley, 1982) . Also, whereas the testimonial is sometimes said to reflect a Pavlovian process the association of beautifu l people with a product (Gorn, 1982) before the 1930s it was as easily associated with the structuralist concept of ideomotor action (James, 1890) . In this view, any ideas created by exposure to stimuli would lead to action if they be came sufficiently prominent in consciousness (Scott, 1903, 1913/1978) . Watson's only original contribution to advertising was seemingly to have used high society figures and European royalty, rather than famous actresses, in hi s advertisements (Coon, 1994) . As in child rearing, Watson' s impact on advertising has been overestimate d (Ewen, 1976 ; Gold, 1987 ; Pease, [1958] 1976 ; Rorty, [1934] 1976) . In fact, two years before his death, Watson himself summarized his impact on the industry by noting that psychology, per se, was unimportant . Advertising simply followed its own procedures (Burnham, 1994) . This was hardly the status he had earlier assigned the psychology of advertising when he lauded it as an exemplar of applied psychology (Watson, 1913b) . D. WATSON : 1936–1958

As a psychologist, Watson's career as a researcher and systematist in university settings spanned only 20 years (1900-1920) . Afterward, he was a systema-

57


tist and a popularizer of psychology for another 16 years (1920–1936), but the n as an advertising executive. He published his last book in 1930 and very little after that . Other than his brief autobiographical chapter (Watson, 1936), his onl y known publication after 1936 was a letter that appeared in the May 1950 issue of the Journal of Psychology Club at Furman University. Watson's life changed dramatically in 1935, however . He resigned from J . Walter Thompson to join William Esty and Company, and Rayner died unexpectedly of pneumonia . After her death, Watson began withdrawing from his social and intellectual life, focusing mainly on business and reportedly preferrin g the farm animals at his Westport, Connecticut estate to his colleagues (Buckley, 1989, pp . 177–184; Burnham, 1994) . When he retired from Esty in 1945, Watson became even more reclusive . He sold his estate and moved into a small farmhouse with his secretary-companion. In 1957 the APA honored Watson with an award for his lifetime work . Although he traveled to New York to receive it , he feared making an emotional display and had his son James accept the hono r in his place . The award's citation read : To Dr. John B . Watson, whose work has been one of the vital determinants of the for m and substance of modem psychology. He initiated a revolution in psychological thought, and his writings have been the point of departure for continuing lines of fruitful research .

Watson died a year later, on September 25, 1958, in New York City . VI . CONCLUSIO N

In this chapter, we have offered an account of Watsonian behaviorism structured along three lines of inquiry : We examined the intellectual, cultural, and individual context from which behaviorism and Watson, himself, emerged ; we ordered Watson's professional activities in terms of the successive phases of his career ; and, within this framework, we analyzed the main features of his contributions . First, as for the context, we found that philosophy, psychology, and societ y were in many ways prepared for Watsonian behaviorism that is, for an objective, empirical, functional, and pragmatic psychology . We also found a youn g man whose persona could be described in similar terms, who was seeking a profession that would make a difference, and who lived in a culture that supporte d progressive ideas . Neither Watson nor his behaviorism were, however, necessar y for psychology to become objective, empirical, functional, and pragmatic . With or without his metaphysical behaviorism, for instance, methodological behavior ism would have emerged, and nothing in Watson's individual context or person a was distinctive enough that he must have become the first behaviorist. What was critical was the unique confluence of culture, psychology, and Watson himself, a point to which we return in closing . Second, in ordering Watson's activities according to the phases of his career , we see that the latter corresponded to his contributions to psychology as a scien-

58

MORRIS AND TOD D

tist, a systematist, and a popularizer. From 1900 through 1912, Watson wa s mainly a researcher in animal behavior and comparative psychology, in both th e laboratory and the field, conducting studies that made him well known and respected. Between 1913 and 1920, Watson continued to conduct research, increasingly with humans, but he became more of a systematizer, publishing extensively on psychology from the standpoint of a behaviorist . When he left academia in 1920, his empirical contributions ceased and his systematic wor k was little further developed, but he greatly popularized a science of psycholog y in the culture at large. The received view of Watsonian behaviorism is generally drawn from this last phase of his career, from selected aspects of the second phase, and from almos t none of the first phase. As a result, the importance of his early empirical an d conceptual contributions and his place in the larger intellectual community fo r formulating, establishing, and situating behaviorism often go unappreciated . The same is true for the influence of Watson's early work on his subsequent career and contributions . We have tried to make these early influences clearer an d more explicit . Third, throughout the foregoing, we interspersed descriptions and analyses o f the main features of Watsonian behaviorism, several subsidiary features, and th e nature of Watson's science . These features included (a) the goal of predictionand-control, (b) Watson's methodological and metaphysical stances on the mind body problem, (c) his analysis of thinking, language, and implicit response, (d ) the relation between heredity and environment, (e) implications of Watson's behaviorism for advertising and child rearing, and (f) Watson's science of behavior . Here we found that prediction-and-control had different meanings dependin g on context, no one of them essential, except where prediction-and-control mean t natural science itself. Methodological and metaphysical behaviorism wer e clearly distinctive positions, yet they were just as clearly both "behavioral ." Watson's analysis of thinking, language, and implicit response was one of severa l possible interpretations, and not all of it was unique . The heredity-environmen t issue was one on which Watson took different stances, but for him the issue wa s largely empirical, not philosophical : He could just as well have emphasized heredity and still been the first behaviorist . Watson's contributions to advertisin g and child rearing were neither novel nor necessarily predicated on his behavior ism : His advertizing campaigns were consistent with industry standards, and hi s child rearing advice was a variant of extant practices, although perhaps influenced by idiosyncracies of his own childhood . As for Watson's science of behavior, few aspects of its purpose and purview , methods, or systematic organization distinguished it from the other behaviorism s of his day or from those to come . Most behaviorisms, for instance, attempt naturalistic stances and seek principles of behavior, often in the language of stimuli and responses . Likewise, direct observation and conditioning are distinctive features of psychology in the learning-theory tradition, while verbal reports ar e commonplace in research on human learning and cognition. Although psycho logical tests for the assessment of intelligence, abilities, and personality gener-

59


ally belong to psychologies that are not nominally behavioral, experimental psychology is today behavioral in its methods and for Watson these were method s for measuring behavior . Finally, Watson's organization of behavior by form and content has often been mirrored throughout psychology, but his analysis of function has been superseded on the basis of empirical research . The latter should b e expected, however, in a science of psychology . Excepting Watson's naturalistic stance, little of his context, career, and contributions seems uniquely determining of, or determined by, his behaviorism . Th e latter's distinctiveness was due to a unique confluence of variables . In othe r words, although behaviorism would not have been Watsonian without Watson , an objective, empirical, functional, and pragmatic psychology would hav e emerged and been accepted in the culture of his day . Taken together, however, Watson, psychology, and the culture provided a unique configuration of a persona, a discipline, and a time and place from which evolved a distinctive and singularly important step toward psychology's becoming a science . Watsonian behaviorism remains today a benchmark against which other behaviorisms an d psychological systems are measured. What might Watson have thought about psychology at present, as it approaches the millennium, nearly a century after he began graduate school? Hi s vision was prescient: Experimental psychology has not been defined in structuralist image for more than 75 years . It is today methodologically behavioral . Watson's metaphysical behaviorism, however, never took hold to any appreciable degree, except in some of the other behaviorisms . A completely naturalisti c and independent science of behavior thus remains an elusive goal for psychology and even for some varieties of behaviorism . As a consequence, Watson's observation about psychology in 1936 might well be his observation today : I still believe as firmly as ever in the general behavioristic position I took overtly in 1912 . I think it has influenced psychology. Strangely enough, I think it has temporally slowe d down psychology because the older instructors would not accept it wholeheartedly, an d consequently they failed to present it convincingly to their classes . The youngsters did not get a fair presentation, hence they are not embarking wholeheartedly upon a behavioristi c career, and yet they will no longer accept the teachings of James, Titchener, and Angell . I honestly think psychology has been sterile for several years . We need younger instructors who will teach objective psychology with no reference to the mythology most of us present-day psychologists have been brought up upon . When the day comes, psychology wil l have a renaissance greater than that which occurred in the Middle Ages . I believe a s firmly as ever in the future of behaviorism . (Watson, 1936, p . 281 )

REFERENCE S Abbagnano, N . (1967) . Positivism . In P. Edwards (Ed .), The encyclopedia of philosophy (pp . 414-419) . New York : Macmillan . Allen, J. (1904) . The associative processes of the guinea pig . A study of the psychical development of an animal with a nervous system well-medullated at birth. Journal of Comparative Neurology and Psychology, 14, 293-259. Angell, J . R. (1907) . The province of functional psychology . Psychological Review, 14, 61-91 .

60

MORRIS AND TOD D

Angell, J. R . (1913) . Behavior as a category of psychology . Psychological Review, 20, 255-270 . Ayer, A . J. (1968) . The origins of pragmatism : Studies in the philosophy of Charles Saunders Peirc e and William James. London : Macmillan. Bacon, F. (1942) . New Atlantis. in G . S . Haight (Ed.), Essays and new Atlantis (pp. 243-302 . New York: Walter J. Black . (Original work published 1624) Bacon, E (1960) . The new organon. (E H. Anderson, Ed .) . Indianapolis, IN: Bobbs Merrill. (Original work published 1620 ) Bakan, D . (1966) . Behaviorism and American urbanization . Journal of the History of the Behaviora l Sciences, 2, 5-28 . Baldwin, (1895) . Mental development in the child and the race . New York : Macmillan . Baum, W. M . (1994) . John B . Watson and behavior analysis : Past, present, and future. In J. T. Todd & E. K. Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 133-140) . Westport, CT: Greenwood. Behkterev, V. (1906) . La psychologie objective. Revue Scientifique, 6, 353-357 . Behkterev, V. (1913) . La psychologie objective. Paris : Librairie Felix Alcan . Bergmann, G. (1956) . The contribution of John B . Watson . Psychological Review, 63, 265-276 . Bernard, L . L. (1921) . The misuse of instinct in the social sciences . Psychological Review, 28, 96-119. Blackmore, J . T. (1972) . Ernst Mach. Berkeley, CA : University of California Press. Bledstein, B . (1976). The culture of professionalism: The middle class and the development of higher education in America . New York : Norton . Blumberg, A . E ., & Feigl, H . (1931) . Logical positivism . Journal of Philosophy, 28, 281-296 . Boakes, R . A. (1984) . From Darwin to behaviourism : Psychology and the minds of animals. Cambridge, England : Cambridge University Press . Boakes, R. A. (1994). John B . Watson ' s early scientific career : 1903-1913 . In J. T. Todd & E . K. Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 144-150). Westport, CT: Greenwood. Bonelli, M. L . R ., & Shea, W. R . (Eds .) . (1975) . Reason, experiment, and mysticism in the scientifi c revolution . New York : Science History Publications. Boring, E . G . (1942) . Sensation and perception in the history of experimental psychology. New York: Appleton-Century. Boring, E. G. (1950) . A history of experimental psychology (2nd ed .) . New York : Appleton-Century Crofts . Boring, E . G . (1955) . Dual role of the zeitgeist in scientific creativity. Scientific Monthly, 80, 101-106 . Breland, K ., & Breland, M . (1961) . The misbehavior of organisms . American Psychologist, 16, 661-664 . Brewer, C. L. (1991) . Perspectives on John B . Watson. In G. A. Kimble, M . Wertheimer, & C . White (Eds.), Portraits of pioneers in psychology (pp. 171-186) . Hillsdale, NJ : Erlbaum. Brown, S . C . (Ed .). (1985) . Philosophers of the Enlightenment . Sussex, England : Harvester. Buckley, K. E. (1982) . The selling of a psychologist : John Broadus Watson and the application o f behavioral techniques to advertising. Journal of the History of the Behavioral Sciences, 18, 207-221 . Buckley, K. W. (1989) . Mechanical man: John Broadus Watson and the beginnings of behaviorism. New York : Guilford. Buckley, K . W. (1994) . Misbehaviorism: The case of John B . Watson' s dismissal from Johns Hopkins University. In J . T. Todd & E . K . Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 19-36) . Westport, CT: Greenwood. Buenker, J. D ., Burnham, J. C., & Crunden, R . M . (1977). Progressivism . Cambridge, MA : Schenkman. Burnham, J. C . (1968) . On the origins of behaviorism. Journal of the History of the Behavioral Sciences, 4, 143-151 . Burnham, J . C. (1989) . How superstition won and science lost. New Brunswick, NJ : Rutgers University Press.


61

Burnham, J. C . (1994) . John B . Watson : Interviewee, professional figure, symbol . In J. T. Todd & E. K . Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 65-73) . Westport, CT: Greenwood . Calkins, M . W. (1906) . A reconciliation between structural and functional psychology . Psychologica l Review, 13, 61-81 . Calkins, M . W. (1913) . Psychology and the behaviorist. Psychological Bulletin, 10, 288-291 . Calkins, M . W. (1921) . The truly psychological behaviorism . Psychological Review, 28, 1-18 . Calkins, M. W. (1930) . Mary Whiton Calkins . In C . Murchison (Ed.), A history of psychology in autobiography (Vol . 1, pp . 31-62) . Worcester, MA : Clark University Press . Carr, H., & Watson, J. B . (1908) . Orientation in the white rat. Journal of Comparative Neurology an d Psychology, 18, 27-44 . Cattell, J . M . (1906) . Conceptions of methods of psychology. In H. J . Rogers (Ed.), Congress of arts and science, Universal Exposition, St. Louis, 1904 : Vol. 5. Biology, anthropology, psychology, sociology. Boston : Houghton, Mifflin. Chicago Association for Child Student and Parent Education . (1926) . Intelligent parenthood: Proceedings of the Mid-West Conference on Parent Education, March 4, 5, and 6, 1926 . Chicago : University of Chicago Press. Chicago Association for Child Student and Parent Education. (1928). Building character: Proceedings of the Mid-West Conference on Parent Education, February, 1928. Chicago : University of Chicago Press. Cohen, D. (1979). John B. Watson: The founder of behaviorism . London : Routledge & Kegan Paul . Coon, D . J . (1993) . Standardizing the subject : Experimental psychologists, introspection, and th e quest for a technological ideal . Technology and Culture, 34, 757-783 . Coon, D . J . (1994). "Not a creature of reason" : The alleged impact of Watsonian behaviorism on advertising in the 1920s. In J . T. Todd & E . K . Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 37-64) . Westport, CT: Greenwood. Cosslett, T . (Ed.) . (1984) . Science and history in the nineteenth century . Cambridge, England : Cambridge University Press . Costall, A. (1993) . How Lloyd Morgan's canon backfired. Journal of the History of the Behavioral Sciences, 29, 113-122 . Creelan, P. G . (1974) . Watsonian behaviorism and the Calvinist conscience . Journal of the History of the Behavioral Sciences, 0, 95-118 . Curtis, H . S . (1899/1900) . Automatic movements of the larynx . American Journal of Psychology, 11 , 237-239 . Darwin, C . R . (1859). The origin of species. London : Murray. Darwin, C . R . (1871) . The descent of man. London : Murray. Darwin, C . R . (1877) . A biographical sketch of an infant. Mind, 2, 285-294 . Day, W. F. (1983). On the difference between radical and methodological behaviorism. Behaviorism , 11, 89-102 . Dewey, J . (1886) . Psychology. New York : Harper. Dewey, J . (1896). The reflex arc concept in psychology . Psychological Review, 3, 357-370 . Dewey, J . (1906) . The terms "conscious " and "consciousness ." Journal of Philosophy, Psychology, and Experimental Methods, 3, 39-41 . Dewsbury, D . A . (1984) . Comparative psychology in the twentieth century. Stroudsburg, PA : Hutchinson . Dewsbury, D . A . (1994) . John B . Watson : Profile of a comparative psychologist and proto-ethologist . In J. T. Todd & E . K. Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp. 141-144) . Westport, CT : Greenwood. Dodge, R. (1912) . The theory and limitations of introspection . American Journal of Psychology, 23, 214-229 . Dunlap, K. (1912) . The case against introspection . Psychological Review, 19, 404-413 . Dunlap, K. (1919) . Are there any instincts? Journal of Abnormal Psychology, 14, 307-311 . Dunlap, K. (1932) . Knight Dunlap . In C. Murchison (Ed .), History of psychology in autobiography (Vol 2, pp . 35-61) . Worcester, MA : Clark University Press .

62

MORRIS AND TOD D

Ebbinghaus, H . (1913) . Memory : A contribution to experimental psychology (H. A. Ruger & C . E. Bussenius, Trans .) . New York: Teacher's College. (Original work published 1885 ) Ewen, S . (1976) . Captains of consciousness : Advertising and the social roots of the consumer culture. New York: McGraw-Hill . Fechner, G. T. (1966). Elements of psychophysics . New York: Holt, Rinehart & Winston. (Original work published 1860) Fenton, J . C . (1925). The practical psychology of babyhood . Mental development and mental hygien e in the first two years of life . Boston : Houghton Mifflin . Freud, S . (1930) . Civilization and its discontents. New York : Norton. Gesell, A . (1930). The guidance ofmental growth in infant and child . New York : Macmillan. Gesell, A ., & Ilg, F. L . (1943) . The infant and child in the culture of today. New York : Harper & Brothers . Glenn, P. J . (1936) . Psychology: A class manual in the philosophy of organic and rational life . St. Louis, MO: Herder. Gold, P. (1987) . Advertising, politics, and American culture : From salesmanship to therapy. New York : Paragon House . Gorn, G . (1982) . The effects of music in advertising on choice behavior : A classical conditioning approach . Journal of Marketing, 46, 94-101 . Gould, S . J . (1977) . Ontogeny and phylogeny. Cambridge, MA : Harvard University Press. Gould, S . J . (1981) . The mismeasure of man . New York : Norton . Gray, P. H. (1963) . The descriptive study of imprinting in birds from 1863 to 1953 . Journal of General Psychology, 63, 333-346 . Hall, G . S . (1904) . Adolescence: Its psychology and its relations to physiology, anthropology , sociology, sex, crime, religion, and education (Vols . 1–2) . New York: Appleton . Hannush, M . J. (1987) . John B . Watson remembered : An interview with James B . Watson. Journal of the History of the Behavioral Sciences, 23, 137-152 . Harrell, W., & Harrison, R . (1938). The rise and fall of behaviorism. Journal of General Psychology, 18, 367–421 . Harris, B . (1979) . What ever happened to little Albert? American Psychologist, 34, 151–160. Harris, B . (1981) . A non-historical biography of John B . Watson . Contemporary Psychology, 26, 66–67 . Harris, B . (1982) . The role of film in John Watson 's developmental research program: Intellectual, disciplinary, and social influences . In G . Eckardt, W. G . Bringmann, & L. Sprung (Eds .), Contributions to a history of developmental psychology : International William T. Preyer Symposium (pp . 359–366) . Berlin : Mouton . Harzem, P. (1993) . The discrediting of John Broadus Watson . Mexican Journal of Behavior Analysis , 19, 39–66. Heidbreder, E . (1933). Seven psychologies. Englewood Cliffs, NJ : Prentice-Hall . Herrnstein, R . J . (1967). Introduction . In J . B . Watson, Behavior: An introduction to comparativ e psychology (pp . xi–xxxi) . New York : Holt, Rinehart & Winston. Herrnstein, R. J. (1969) . Behaviorism. In D . A Krantz (Ed .), Schools of psychology : A symposium (pp . 51-68) . New York : Appleton-Century-Crofts . Horowitz, F. D . (1992) . John B . Watson's legacy : Learning and environment . Developmental Psychology, 28, 360-367 . Hull, C . L . (1943). Principles of behavior. New York : Appleton-Century-Crofts . Hulse, S . H., Fowler, H., & Honig, W. K . (Eds .) . (1978) . Cognitive processes in animal behavior. Hillsdale, NJ: Erlbaum. James, W. (1890) . Principles ofpsychology (Vols . 1–2) . New York : Holt. James, W. (1892) . A plea for psychology as a natural science . Philosophical Review, 1, 146–153 . James, W. (1904). Does "consciousness " exist? Journal of Philosophy, Psychology, and Scientifi c Methods, 1, 477-491 . Jastrow, J . (1930) . Joseph Jastrow. In C . Murchison (Ed .), History of psychology in autobiography (Vol 1, pp. 135-162) . Worcester, MA : Clark University Press .


63

Jaynes, J . (1976) . The origin of consciousness in the breakdown of the bicameral mind . Boston : Houghton Mifflin . Jennings, H . S . (1906) . Behavior of the lower organisms . New York : Macmillan . Jennings, H . S . (1907) . The interpretation of the behavior of the lower organisms. Science, 27, 698-710 . Jennings, H . S ., Watson, J . B ., Meyer, A., & Thomas, W. I . (Eds .). (1917) . Suggestions of modern science concerning education. New York: Macmillan. Jones, M . C. (1924a) . The elimination of children ' s fears . Journal of Experimental Psychology, 7, 383-390 . Jones, M . C. (1924b) . A laboratory study of fear: The case of Peter. Pedagogical Seminary, 31 , 308-315 . Judd, C . H. (1910) . Evolution and consciousness. Psychological Review, 17, 77-97 . Kantor, J . R . (1924) . Principles of psychology (Vol . 1). Chicago : Principia Press . Kantor, J. R . (1926) . Principles of psychology (Vol . 2) . Chicago: Principia Press. Kantor, J . R . (1958) . Interbehavioral psychology. Chicago: Principia Press. Keller, F. S . (1994) . A debt acknowledged . In J . T. Todd & E . K. Morris (Eds .), Modern perspective s on John B. Watson and classical behaviorism (pp . 125-130) . Westport, CT: Greenwood . Kimble, G . A. (1997) . John B . Watson ' s Behaviorism: A retrospective review. Contemporary Psychology, 42, 23-28. King, W. P. (Ed .) . (1930) . Behaviorism: A battle line. Nashville, TN : Cokesbury Press . Kitchener, R . F. (1977) . Behavior and behaviorism. Behaviorism, 5, 11-71 . Kohler, W. (1925) . The mentality of apes (E . Winter, Trans .) . New York : Harcourt, Brace, & World . (Original work published 1917 ) Kohler, W. (1947) . Gestalt psychology : An introduction to new concepts in modern psychology. New York : Liveright. Krasner, L. (1988). In memoriam : Mary Cover Jones . Behavior Analyst, 11, 91-92 . Kuo, Z . Y. (1922) . How are our instincts acquired? Psychological Review, 29, 34-365 . Lachman, R., Lachman, L . J ., & Butterfield, E . C . (1979). Cognitive psychology and informatio n processing: An introduction. Hillsdale, NJ : Erlbaum . Langfield, H. S . (1943) . Fifty years of the Psychological Review Psychological Review, 50, 143-155 . Larson, C . A . (1981) . Epilogue to the review of John B . Watson's biography. Contemporary Psychology, 26, 62. Lashley, K. S . (1923). The behavioristic interpretation of consciousness . Psychological Review, 30, 237-272, 329-353 . Lashley, K . S . (1951) . The problem of serial order in behavior. In L. A . Jeffress (Ed.), Cerebral mechanisms in behavior: The Hixon Symposium (pp. 112-146) . New York : Wiley . Lashley, K. S ., & Watson, J . B . (1913). Notes on the development of a young monkey . Journal of Animal Behavior, 3, 114-139 . Lashley, K . S ., & Watson, J . B . (1920) . A psychological study of motion pictures in relation t o venereal disease campaigns . Social Hygiene, 7, 181-219 . Leahey, T. H . (1981) . The mistaken mirror : On Wundt ' s and Titchener's psychologies . Journal of the History of the Behavioral Sciences, 17, 273-282. Leahey, T. H. (1992) . Mythical revolutions in the history of American psychology . American Psychologist, 47, 308-318 . Leahey, T. H . (1997). History of psychology : Main currents in psychological thought. Englewood Cliffs, NJ : Prentice-Hall . Leeper, R. (1951) . Cognitive processes . In S . S . Stevens (Ed .), Handbook of experimental psychology (pp. 730-757) . New York: Wiley. Lindberg, D . (1992) . The beginnings of western science . Chicago : University of Chicago Press . Loeb, J. (1900) . Comparative physiology of the brain and comparative psychology . New York : Putnam. Loeb, J . (1907) . Concerning the theory of tropisms . Journal of Experimental Zoology, 4, 145-156 .

64

MORRIS AND TOD D

Loeb, J . (1916) . The organism as a whole . New York : Putnam. Loeb, J . (1918) . Forced movements, tropisms, and animal conduct . Philadelphia: Lippincott . Logue, A. W. (1978) . Behaviorist John B . Watson and the continuity of species . Behaviorism, 6, 71 7 9 Logue, A. W. (1985a). The origins of behaviorism: Antecedents and proclamation . In C . Buxto n (Ed .), Points of view in the modern history of psychology (pp . 141-167) . New York: Academic Press . Logue, A . W. (1985b). The growth of behaviorism : Controversy and diversity. In C . Buxton (Ed .) , Points of view in the modern history of psychology (pp . 169-196) . New York : Academic Press. Lorenz, K. Z. (1935) . Der Kumpan in der Unwelt des Vogels . Journal of Ornithology, 83, 137-213 , 289-413 . Lorenz, K. Z. (1950) . The comparative method in studying innate behavior patterns . In J. F. Danielli & R . Brown (Eds.), Physiological mechanisms in animal behavior (pp . 221-268) . New York: Academic Press . Lovie, A. D . (1984) . Ethnographic discourse analysis and J. B . Watson : The behaviorist as propagandist . In W. J . Bako, M . E. Hyland, H . Van Rappard, & A. W. Staats (Eds.), Current issues in theoretical psychology (pp . 151-164) . New York : Elsevier. Lubbock, J . (1882) . Ants, bees, and wasps . A record of observation on the habits of the socia l Hymenoptera . London : Kegan Paul, Trench . Mach, E . (1959) . The analysis of sensations. New York : Dover. (Original work published 1886) Mach, E . (1960). The science of mechanics (T . J . McCormack, Trans .) . LaSalle, IL : Open Court . (Original work published 1883 ) Malone, J. C . (1990) . Theories of learning: A historical approach. Belmont, CA : Wadsworth. Mead, D . E . (1976) . Six approaches to parenting. Provo, UT : Brigham Young University Press . Meyer, M . (1911) . The fundamental laws of human behavior. Boston: Badger. Moore, J . (1981) . On mentalism, methodological behaviorism, and radical behaviorism . Behaviorism, 9, 55-77 . Moore, J . (1995) . Some historical and conceptual relations among logical positivism, behaviorism , and cognitive psychology. In J. T. Todd & E . K . Morris (Eds.), Modern perspectives on B . F Skinner and contemporary behaviorism (pp . 51-74) . Westport, CT: Greenwood. Moore, T. V. (1939) . Cognitive psychology. Philadelphia: Lippincott . Morgan, C . L. (1894) . An introduction to comparative psychology . London : Walter Scott. Morris, E . K., Todd, J . T., & Midgley, B . D . (1993) . The prediction and control of behavior : Watson , Skinner, and beyond . Mexican Journal of Behavior Analysis, 19, 103-131 . O'Donnell, J . M . (1985) . The origins of behaviorism : American psychology, 1870-1920. New York : New York University Press . Osborne, S . R. (1977) . The free-food (contrafreeloading) phenomena : A review and analysis . Animal Learning and Behavior, 5, 221-235 . Pauly, P. J . (1987) . Controlling life : Jacques Loeb and the engineering ideal in biology. New York : Oxford University Press . Pavlov, I . P. (1906) . The scientific investigation of the psychical faculties or processes in the highe r animals . Science, 24, 613-616. Pavlov, I. P. (1927) . Conditioned reflexes. New York: Oxford University Press . Pease, O . (1976) . The responsibilities of American advertising : Private control and public influence, 1920-1940. New York: Arno Reprint. (Original work published 1958 ) Peirce, C . S . (1905). What pragmatism is . Monist, 15, 161-181 . Pepper, S . C . (1942) . World hypotheses : A study in evidence. Berkeley CA: University of Californi a Press. Perez-Ramos, A. (1988) . Francis Bacon's idea of science and the maker's knowledge tradition . Oxford, England: Oxford University Press . Perry, R . B . (1904) . Conceptions and misconceptions of consciousness . Psychological Review, 11 , 282-296.


65

Pinter, R . (1913). Inner speech during silent reading . Psychological Review, 20, 129–153 . Romanes, G. J . (1882) . Animal intelligence. London : Kegan Paul, Trench . Romanes, G . J . (1884) . Mental evolution in animals . New York: Appleton . Rorty, J . (1976) . Our master's voice : Advertising. New York: Arno Reprint . (Original work publishe d 1934 ) Ross, D. (1991) . The origins of American social science. New York : Cambridge University Press . Ryle, G . (1949) . The concept of mind. New York: Barnes & Noble . Samelson, F. (1974) . History, origin myth, and ideology: "Discovery" of social psychology . Journal for the Theory of Social Behavior, 4, 217–231 . Samelson, F (1980) . John B . Watson's little Albert, Cyril Burt's twins, and the need for a critica l science . American Psychologist, 35, 619-625 . Samelson, F (1981) . Struggle for scientific authority: The reception of Watson's behaviorism . Journal of the History of the Behavioral Sciences, 17, 399–425 . Samelson, F (1985) . Organizing the kingdom of behavior : Academic battles and the organizational policies in the twenties . Journal of the History of the Behavioral Sciences, 21, 33–47 . Samelson, F (1994) . John B . Watson in 1913 : Rhetoric and practice. In J . T. Todd & E . K. Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp . 3–18) . Westport , CT: Greenwood . Scheffler, I . (1974). Four pragmatists: A critical introduction to Peirce, James, Mead, and Dewey. London : Routledge & Kegan Paul . Schneider, S . M., & Morris, E . K. (1987) . A history of the term radical behaviorism: From Watson to Skinner. Behavior Analyst, 10, 27-39 . Scott, W. D . (1903) . The theory of advertising : A simple exposition of the principles ofpsychology in their relation to successful advertising. Boston : Small, Maynard . Scott, W. D . (1978) . Psychology of advertising (5th ed.) . New York: Arno Reprint. (Original work published 1913 ) Shapin, S . (1996) . The scientific revolution . Chicago : University of Chicago Press . Sherman, M. (1929) . The differentiation of emotional responses in infants : I . Judgments o f emotional responses from motion picture views and actual observation . Journal of Comparative Psychology, 7, 265–284. Sherrington, C . S . (1906) . The integrative action of the nervous system. New Haven, CT: Yal e University Press. Skinner, B . F (1938) . The behavior of organisms : An experimental analysis. New York : Appleton Century. Skinner, B . F (1945). The operational analysis of psychological terms . Psychological Review, 52, 270 – 277, 291– 294 . Skinner, B . F (1953) . Science and human behavior. New York: Macmillan . Skinner, B . F (1959) . John Broadus Watson, behaviorist . Science, 129, 197–198 . Skinner, B . F (1974) . About behaviorism . New York : Knopf. Small, W. S . (1900) . Experimental study on the mental processes of the rat . American Journal of Psychology, 11, 133–165 . Smith, L . D . (1986) . Behaviorism and logical positivism : A reassessment of the alliance. Stanford , CA : Stanford University Press . Smith, L. D . (1992) . On prediction and control : B . F Skinner and the technical ideal of science . American Psychologist, 47, 216–223 . Smith, L . D . (1995) . Inquiry nearer the source . In J. T. Todd & E . K . Morris (Eds .), Modern perspectives on B. F. Skinner and contemporary behaviorism (pp . 39–50). Westport, CT: Greenwood. Smith, M. R., & Marx, L . (1994) . Does technology drive history? The dilemma of technological determinism. Cambridge, MA : MIT Press . Spaulding, D . A . (1873) . Instinct . With observations on young animals. Macmillan's Magazine, 27, 282–293 .

66

MORRIS AND TOD D

Spencer, H. (1897) . The principles ofpsychology (3rd ed .). New York: Appleton . Spock, B . (1988) . Dr. Spock on parenting: Sensible advice from America's most trusted child-care expert. New York : Simon & Schuster. Stevens, S . S . (1939) . Psychology and the science of science . Psychological Bulletin, 36, 221-263 . Stocking, G . W. (1965). On the limits of "presentism" and "historicism " in the historiography of the behavioral sciences . Journal of the History of the Behavioral Sciences, 1, 211-217 . Thorn, D . A . (1927). Everyday problems of the everyday child. New York: D. Appleton . Thorndike, E . L. (1898) . Animal intelligence : An experimental study of the associative processes i n animals . Psychological Review Monographs, 2 (Suppl. 8) . Thorndike, E. L . (1904) . Introduction to the theory of mental and social measurements. New York: Science Press . Thorndike, E . L. (1911) . Animal intelligence : Experimental studies . New York : Hafner. Thorndike, E . L. (1913) . Educational psychology: Vol. 2 The psychology of learning . New York: Teachers College . Tinbergen, N . (1951) . The study of instinct. Oxford, England : Oxford University Press . Titchener, E. B . (1897) . An outline of psychology. New York : Macmillan. Titchener, E . B . (1909) . Lectures on the experimental psychology on the thought processes . Ne w York: Macmillan . Titchener, E. B . (1910) . A textbook ofpsychology. New York: Macmillan. Titchener, E. B . (1911) . A note on the consciousness of self. American Journal of Psychology, 22, 540-552. Titchener, E . B . (1914) . On psychology as the behaviorist views it . Proceedings of the American Philosophical Society, 53, 1-17 . Titchener, E . B . (1929) . Systematic psychology : Prolegomena. Ithaca, NY: Cornell University Press . Todd, J. T. (1994). What psychology has to say about John B . Watson: Classical behaviorism in textbooks, 1920-1989 . In J . T. Todd & E. K. Morris (Eds .), Modern perspectives on John B. Watson and classical behaviorism (pp. 75-108) . Westport, CT: Greenwood . Todd, J. T., Dewsbury, D . A., Logue, A. W., & Dryden, N . W. (1994). John B . Watson : A bibliography of published works . In J. T. Todd & E . K. Morris (Eds .), Modern perspectives on John B . Watson and classical behaviorism (pp . 169-178) . Westport, CT: Greenwood. Todd, J . T., & Morris, E . K. (1986) . The early research of John B . Watson : Before the behavioral revolution. Behavior Analyst, 9, 71-88 . Todd, J . T., & Morris, E . K. (1992) . Case histories in the great power of steady misrepresentation . American Psychologist, 47, 1441-1453 . Todd, J . T., & Morris, E. K . (Eds .) . (1994). Modern perspectives on John B. Watson and classical behaviorism. Westport, CT : Greenwood. Tolman, E . C. (1922) . A new formula for behaviorism. Psychological Review, 29, 44-53 . Tolman, E. C . (1932) . Purposive behavior in animals and men . New York: Appleton-Century-Crofts . Washburn, M . E (1908) . The animal mind. New York: Macmillan . Watson, J. B . (1903) . Animal education : An experimental study of the psychical development of the white rat, correlated with the growth of its nervous system . Chicago : University of Chicag o Press . Watson, J. B . (1904) . Some unemphasized aspects of comparative psychology. Journal of Comparative Psychology and Neurology, 14, 360-363 . Watson, J . B . (1907a) . Kinaesthetic and organic sensations : Their role in the reactions of the white rat to the maze. Psychological Review Monograph Supplement, 8 (33), 1-100. Watson, J. B . (1907b) . Report of John B. Watson on the condition of the noddy and sooty ter n colony, Bird Key, Tortugas, Florida . Bird Lore, 9, 307-316. Watson, J . B . (1907c) . [Review of Behavior of lower organisms] . Psychological Bulletin, 4 , 288-291 . Watson, J. B . (1908a) . The behavior of noddy and sooty terns . In Papers from the Tortuga s Laboratory of the Carnegie Institution of Washington, 2, 187-255 . (Carnegie Institutio n Publication #103)


67

Watson, J. B . (1908b) . Imitation in monkeys . Psychological Bulletin, 5, 169–178 . Watson, J . B . (1908c) . [Review of The dancing mouse] . Journal of Philosophy, Psychology, and Scientific Methods, 5, 184-189 . Watson, J . B . (1909a) . How animals find their way home . Harper's Monthly Magazine, 119, 685–689. Watson, J . B . (1909b). Some experiments bearing on color vision in monkeys . Journal of Comparative Psychology and Neurology, 19, 1–28 . Watson, J . B . (1910a) . Further data on the homing sense of noddy and sooty terns. Science, 32, 470–473 . Watson, J. B . (1910b) . The new science of animal behavior. Harper's Monthly Magazine, 120, 346-353 . Watson, J . B . (1910c). Report of Prof. John B . Watson on the work on Bird Key. Carnegie Institution of Washington Yearbook, 9, 144–146 . Watson, J . B . (1911a) . Literature for 1910 on the behavior of vertebrates . Journal of Animal Behavior, 1, 430–447. Watson, J . B . (1911b) . [Review of Animal Intelligence] . Science, 34, 317 . Watson, J . B . (1911c) . [Review of A textbook of psychology] . American Journal of Psychology, 22, 313-316 . Watson, J . B . (1912a) . Instinctive activity in animals : Some recent experiments and observations . Harper's Monthly Magazine, 124, 376–382. Watson, J . B . (1912b) . Report on the limits of the spectrum for birds . Carnegie Institution of Washington Yearbook, 11, 162–163 . Watson, J . B . (1913a) . Image and affection in behavior . Journal of Philosophy, Psychology, an d Scientific Methods, 10, 421–428 . Watson, J . B . (1913b) . Psychology as the behaviorist views it . Psychological Review, 20, 158–177 . Watson, J. B . (1914) . Behavior: An introduction to comparative psychology. New York : Holt. Watson, J. B . (1915) . Studies on the spectral sensitivity of birds . Carnegie Institute Publications , 211, 85–114. Watson, J . B . (1916a) . Behavior and the concept of mental disease . Journal of Philosophy, Psychology, and Scientific Methods, 13, 589–597 . Watson, J . B . (1916b) . The place of the conditioned-reflex in psychology . Psychological Review, 23 , 89–116. Watson, J . B . (1917) . Practical and theoretical problems in instinct and habits . In H . S . Jennings, J. B . Watson, A. Meyer, & W. I. Thomas (Eds .), Suggestions of modern science concerning education (pp . 51–99). New York: Macmillan. Watson, J . B . (1918) . Preliminary report of the effect of oxygen hunger upon handwriting function . Washington, DC : Military Aeronautics, Medical Research Laboratory Division, pp. 182–185 . Watson, J. B . (1919) . Psychology from the standpoint of a behaviorist . Philadelphia: Lippincott . Watson, J . B . (1922a, September) . Are booklets read and kept? J. Walter Thompson Compan y Bulletin . Watson, J. B . (1922b, January) . Can science determine your baby's career before it can talk? New York Sunday Magazine, p. 4. Watson, J . B . (1922c, July) . What cigarette are you smoking and why? J. Walter Thompson Company News Bulletin, pp 1–1 5 Watson, J . B . (1924a) . Behaviorism. New York : People's Institute. Watson, J. B . (1924b) . Psychology from the standpoint of a behaviorist (2nd ed.) . Philadelphia : Lippincott. Watson, J . B . (1927a, November) . Are you giving your child a chance?—This behaviorist speaks . McCall's, pp. 64, 74 . Watson, J . B . (1927b) . The behaviorist looks at instincts . Harper's Monthly Magazine, 155, 228–235 . Watson, J . B . (1927c, April) . Can psychology help in the selection of personnel? J. Walter Thompso n Company News Bulletin, pp . 9–10, 13 .

68

MORRIS AND TODD

Watson, J. B. (1928) . The ways of behaviorism. New York: Harper & Brothers . Watson, J . B . (1929) . Psychology from the standpoint of a behaviorist (3rd ed.) . Philadelphia : Lippincott . Watson, J. B . (1930) . Behaviorism (rev. ed .) . Chicago : University of Chicago Press . Watson, J. B . (1934, July) . Our fears : How they develop. Red Barrel. Watson, J . B . (1936). John Broadus Watson. In C. Murchison (Ed.), A history of psychology i n autobiography (Vol . 3, pp . 271-281). Worcester, MA : Clark University Press . Watson, J. B ., & Lashley, K . S . (1913a) . The effect of the amount of practice in learning archery . Carnegie Institution of Washington Yearbook, 12, 180-181 . Watson, J. B ., & Lashley, K . S . (1913b). A study of homing instinct in the noddy and sooty ter n which nest upon Bird Key, Tortugas . Carnegie Institution of Washington Yearbook, 12, 179-180. Watson, J . B ., & Lashley, K . S . (1915a) . An historical and experimental study of homing. Carnegie Institute Publications, 211, 7-60 . Watson, J . B ., & Lashley, K. S . (1915b) . Homing and related activities of birds . Bird Lore, 17, 460-461 . Watson, J. B ., & Lashley, K . S . (1920) . A consensus of medical opinion upon questions relating to sex education and venereal disease campaigns . Mental Hygiene, 4, 769-847 . Watson, J . B ., & Morgan, J . J . B . (1917) . Emotional reactions and psychological experimentation . American Journal of Psychology, 28, 163-174 . Watson, J . B ., & Rayner, R. (1920) . Conditioned emotional reactions . Journal of Experimental Psychology, 3, 1-14 . Watson, J. B ., & Watson, M . I. (1913) . A study of the responses of rodents to monochromatic light . Journal ofAnimal Behavior, 3, 1-14 . Watson, J . B ., & Watson, R . R . (1928) . Psychological care of infant and child. New York: Norton . Watson, R . I . (1971). Prescriptions as operative in the history of psychology . Journal of the History of the Behavioral Sciences, 7, 311-322 . Westfall, R . S . (1971) . The construction of modern science : Mechanisms and mechanics. Cambridge , England : Cambridge University Press . Wiebe, R. (1967). The search for order, 1877-1920 . New York : Hill & Wang .

Winston, A . S . (1996) . "As his name indicates" : R . S . Woodworth's letters of reference an d employment for Jewish psychologists in the 1930s . Journal of the History of the Behavioral Sciences, 32, 30-43 . Wittgenstein, L . (1953) . Philosophical investigations . New York: Macmillan . Woodworth, R. S . (1921) . Psychology. New York : Holt. Woodworth, R. S . (1931) . Contemporary schools ofpsychology. New York: Ronald Press . Woodworth, R. S . (1938) . Experimental psychology. New York : Holt. Woodworth, R . S . (1948) . Contemporary schools of psychology (rev. ed.) . New York : Ronald Press . Wozniak, R. H. (Ed .) . (1993a) . Experimental and comparative roots of early behaviorism : Studies of animal and infant behavior. London : Routledge/Thoemmes Press . Wozniak, R. H . (Ed.) . (1993b) . Theoretical roots of early behaviorism : Functionalism, the critique of introspection, and the nature and evolution of consciousness . London : Routledge/Thoemme s Press. Wozniak, R . H . (Ed .) . (1994). Reflex, habit, and implicit response : The early elaboration of theoretical and methodological behaviourism . London : Routledge/Thoemmes Press . Wundt, W. M . (1873) . Grundzug der physiologischen psychologie . Leipzig, Germany : Englemann. Wundt, W. M . (1912) . An introduction to psychology (2nd . ed.) . New York: Macmillan. Yerkes, R . M. (1905) . Animal psychology and criteria of the psychic . Journal of Philosophy, Psychology, and Scientific Methods, 2, 141-149 . Yerkes, R . M . (1907) . The dancing mouse . New York: Macmillan . Yerkes, R . M . (1910) . Psychology and its relations to biology. Journal of Philosophy, Psychology, and Scientific Methods, 7, 113-124 . Yerkes, R . M ., & Morgulis, S . (1909) . The method of Pawlow in animal psychology. Psychological Bulletin, 6, 257-273 .


69

Yerkes, R. M ., & Watson, J . B . (1911) . Methods of studying vision in animals . Behavio r Monographs, 1 (2), iv-90. Zuriff, G. E. (1985) . Behaviorism : A conceptual reconstruction . New York : Columbia University Press .


3 INTERBEHAVIORISM AN D INTERBEHAVIORA L PSYCHOLOG Y

LINDA J . HAYES AND DEBRA W . FREDERICK S University of Nevada

I . Historical Backgroun d A. Biography B. The Evolution of Organismic Psycholog y II . Ziggurat of Science and Civilizatio n A. Cultural Matrix B. Protopostulates of Scientific Philosophy C. Metasystem of the Sciences D. Postulates of the Individual Science s III . Postulational System of Interbehavioral Psycholog y A. Interbehavioral Fields as Subject Matte r B. Factors Participating in Interbehavioral Field s IV. Psychological Event s A. General Classes of Interbehavio r B. Varieties of Psychological Event s V. Interbehaviorism and Radical Behavioris m A. Similaritie s B. Differences VI . Relation to Empirical Evidenc e A. Kantor's Contributions to Empirical Research B. Others' Assessment of Kantor's Contribution s C. Sphere of Influence D. Domains of Influence VII . Current Status and Future Directions VIII . Epilogue References Handbook of Behaviorism

71


72

HAYES AND FREDERICKS

Jacob Robert Kantor (1888–1984) was a prolific contributor to the psycho logical and philosophical literatures for over six decades, publishing more tha n 100 articles and books on interbehaviorism, in addition to numerous book reviews and commentaries (Smith, 1976) . Paradoxically, interbehaviorism is one of the most overlooked systems in psychology ' s history : It is rare to see his work mentioned in a psychology textbook. Even books on the history of psycholog y do not mention his contribution (although see Lundin, 1979, for an exception) . His work is virtually unknown in the philosophical community as well . Speculation as to the reasons for this obscurity has focused on a number of is sues, including his ponderous writing style, his harsh and unwavering criticism of dualism, the incompatibility of his postulates and those of the dominant scientific collectivity, the absence of an experimental methodology and research tradition even his Jewish heritage (Morris, Higgins, & Bickel, 1983 ; Parrott , 1983a; Smith, 1996a ; Verplanck, 1983, 1995) . Kantor's mentoring style ma y also have played a part in this outcome . He collaborated very little (although se e Kantor and Smith, 1975, for an exception), and held his students up to what has been reported to be a higher standard than some were willing to endure (Ru e Cromwell, personal communication to D . W. Fredericks, November, 1994) . Finally, there were few second generation interbehaviorists trained by Kantor him self, and their opportunity to produce a third generation was limited. That Kantor's work has been overlooked is unfortunate if a rapid pace o f change in intellectual perspective is of value . His position, having been articulated outside of traditional stimulus–response mechanisms or familiar mentalistic approaches, might have simply been ahead of his time (Smith, 1996a) . In recent years, the essence of Kantor's position has become more mainstream (see Delprato, in press, for a discussion of similarities to Kantor's perspective in con temporary psychology) . This evolution of thinking of which Kantor's wor k was an example might have proceeded more rapidly particularly in psychology, had his work been better known . Whatever the reasons for this course of events, the fact remains that interbehaviorism is little known and little appreciated by contemporary psychologist s and philosophers . Our purpose in writing this chapter is thereby twofold . First , we wish to provide an introduction to Kantor's position on its own terms and i n doing so we will show that it is an example of a valuable formulation of contemporary relevance . Second, we wish to examine some of the commonalities between his position and those of other behavior scientists of this century . Although we will focus specifically on interbehavioral psychology, aspects of interbehavioral philosophy that are appropriate to our purposes will also be ad dressed . )

1Kantor's contribution to philosophy (e .g ., Kantor, 1945, 1950 and 1981) is broad in scope . Both its own specifications and its critique of philosophical traditions are more extensive than can be adequately addressed in this chapter.

73

3 INTERBEHAVIORISM AND INTERBEHAVIORAL PSYCHOLOGY

I . HISTORICAL BACKGROUN D A. BIOGRAPHY

Jacob Robert Kantor's family immigrated to the United States from a Russia n speaking area of Eastern Europe sometime during the mid to late 1800s . Kantor was born in Harrisburg, Pennsylvania in 1888, and lived there until his famil y moved to Chicago around 1900, where he remained until finishing his doctora l studies at the University of Chicago (P. T. Mountjoy, personal communication to D . W. Fredericks, November 17, 1996) . During this period, the University of Chicago was the home of the Functiona l School of Psychology, also known as the psychology of adaptation (Leahey , 1992, p . 278) . William James had been a member of the faculty, and one of his followers, Roland Angell, was the chairman of the Department during a portio n of Kantor's time there . Pavlov's influence on the department was also being felt , as John B . Watson's work (e .g ., Watson, 1914) had come into the forefront o f thinking in behavioral psychology during this time . During his studies at the University of Chicago, Kantor began formulating a naturalistic alternative to prevailing perspectives in psychology and philosophy . In his view, these perspectives obstructed scientific progress because of thei r commitment to dualism . He could not align with either the functionalists or the behaviorists of the day because, in his view, neither group had freed themselve s from dualistic thinking (J. R. Kantor, personal communication to L . J . Hayes, May 20, 1981) . Kantor completed his dissertation, entitled "The functional nature of philosophical categories", in 1917 .2 Prior to finishing his degree, he took an appointment at the University of Minnesota, in 1915 . Kantor (1976) described his time in Minnesota as one of the most significant periods of the development of inter behavioral psychology. It was there, through the experience of articulating hi s views to students, that he systematized his perspective into what came to be de scribed as "Organismic Psychology" (Kantor, 1976, p . 122) . The name "Organismic Psychology" was intended to implicate the whole organism as the relevan t actor in a psychological event, and was promoted as an antidote to the notio n that parts of the organism real or imagined had causal responsibility for th e rest of the organism's action . It was a proclamation against dualism . Upon his graduation from the University of Chicago in 1920, Kantor took a post at Indiana University, where he stayed until the time of his retirement in 1957 . He chaired seven Masters theses and eleven Doctoral dissertations whil e at Indiana (Mountjoy, 1994) . It was during this period that Organismic Psychology was renamed Interbehavioral Psychology to correct misconceptions abou t its relation to biology (Kantor, 1976, 1980a ; Smith, 1996a) . While at Indiana, Kantor wrote his two volume principia, Principles of Psychology (1924, 1926) ; completed his seminal works on social psychology (1923 , 2For a detailed description of Kantor's dissertation, see Mountjoy & Cone (1995)

.

74

HAYES AND FREDERICK S

1929), grammar (1936), and scientific logic (1953) ; explicated the relationship between logic and psychology (1945, 1950) ; and developed his critical analysi s of physiology (1947) . All were published by The Principia Press, a publishin g house established for this purpose by Kantor and his brother in 1932 (J . R . Kantor, personal communication to L . J . Hayes, May 20, 1981 3 ) . Additional aspects of these works were published as articles in numerous journals along the way . He was fluent in French and German, and his work was translated into Polis h and Spanish, among other languages (Smith, 1976) . Kantor was active in the early development of psychology as a professiona l discipline . He was among those responsible for the establishment of The Psychological Record in 1937, and served as its first editor (Mountjoy, 1997) . With the exception of a brief interruption attributable to circumstances surroundin g the Second World War, this journal has been in continuous circulation since . Many of his commentaries, appearing in The Psychological Record and elsewhere, were published under pseudonyms as a means of emphasizing an arena of ideas rather than people in the domain of philosophical criticism . His work i s published under the names "Observer" and "A Mitsorg", the latter meaning "o f concern" in German . Kantor was instrumental in bringing B . F. Skinner to Indiana in 1945 (Skinner, 1979, p. 284) . Skinner credits Kantor with certain significant changes in hi s views, among them the elimination of the concept of drive (Skinner, 1974 , p . 243), and an appreciation of the impossibility of defining a functional stimulus without reference to a functional response and vice versa (Skinner, 1938 , p. 35) . While Kantor's views were not changed by his association with Skinne r in a noticeable way, he held Skinner's work in high regard . Verplanck (1983) , who was a member of the faculty at Indiana at the time, writes that Kantor commented more than once : "Aristotle was the first Interbehaviorist, himself the second, and Skinner the third" (pp . xiv) . A more formal tribute to Skinner's caree r was delivered by Kantor somewhat later in an invited address to Division 25 o f the American Psychological Association, entitled "An Analysis of the Experimental Analysis of Behavior." This paper, in which Kantor praised Skinner' s work as outstanding in the great scientific revolution of the period, was subsequently published in The Journal of the Experimental Analysis of Behavio r (Kantor, 1970) . Kantor left Indiana University in 1957 at the age of 69, shortly after the deat h of his wife, and returned with his daughter, Helene, to assume a position of Professor Emeritus at the University of Chicago, which he held for almost thirt y years . Upon returning to Chicago, he set out "to correct the lack of factual treatment of the career of psychology as a scientific discipline, and undertook t o trace the evolution of psychology as a science" (Kantor, 1976) . These efforts eventuated in a two-volume history entitled The Scientific Evolution of Psychology, completed in 1963 and 1969 . (See Parrott, 1984, for further discussion . ) 3 Mountjoy (1997) presents an alternative explanation, stating that the Principia Press "was in tended to serve as a cooperative publisher for competent scholars in any field of human intellectua l endeavor."


75

As an Emeritus Professor, Kantor completed his systematic works in bot h psychology (Interbehavioral Psychology, 1958) and philosophy (Interbehavioral Philosophy, 1981) ; and extended his analyses of numerous philosophical categories, psychological concepts, and scientific domains . He worked on the problem of observation (Kantor, 1980b), of confusing constructs with events (Kantor , 1980a, 1981), and the analysis of psychological linguistics (Kantor, 1977a) , among many others . He worked toward harmonious and collaborative relation s with other disciplines, including mathematics (Kantor, 1958), history (Kantor , 1971a ; Observer, 1979, 1983), biology (Kantor, 1958, 1971a), chemistry (Kan tor, 1958), physics (Kantor, 1958, 1971a), anthropology (Kantor, 1958, 1971a, 1982a), and sociology (1971a), among others . He regarded his revision of An Outline of Social Psychology (1929), entitled Cultural Psychology (1982a), as his last major contribution to psychology ; and Tragedy and the Event Continuum (1983) 4 his last contribution to philosophy. With these, he said, his intellectual program was complete (personal communication to L. J . Hayes, June 1, 1983) . He remained a productive scholar until hi s death in 1984 . 5 The archive of his life's work is held in the Archives of Psychology at the University of Akron, where the remaining copies of his works published by The Principia Press may be obtained . A newsjournal, The Interbehaviorist, dedicated to interbehavioral psycholog y and philosophy, was founded in 1969 by Noel Smith, and continues to be published out of the University of Nevada today . Kantor was given honorary degree s at Dennison University and the University of Akron (P . T. Mountjoy, personal communication to Debra Fredericks, November 17, 1996) and was honored b y numerous other societies and groups ; among them, Cheiron, the Internationa l Society for the History of the Behavioral Sciences . Kantor was also an early contributor to the program of the Midwestern Association for Behavior Analysis (now, ABA), although he did not follow the convention out of his hometown of Chicago, describing the organization as "flamboyant" (personal communication to L. J . Hayes, May 15, 1980) . ABA is still the home of a Special Interest Group of Interbehaviorists, and his influence continues to be felt at this meeting through the work of a number of his follower s and their students . He made numerous trips to Mexico (well into his nineties) where he had a de voted following and was a frequent contributor to Revista Mexicana de Analisis de la Conducta (RMAC), edited at the time by his friend and pupil, Emilio Ribes . Through this connection, Kantor's impact has been felt in many parts of Latin America and Europe . At least two formal conferences on Interbehavioral Psychology have been held in Europe in recent years, one in the United Kingdom and one in Madrid . As was the case for many of the leading behavioral psychologists of his day, Kantor stood alone in his views . References in his published works to compatible thinkers were few in number ; his criticisms (at least in psychology) were no t 'See Parrott (1983b) for a review. 5 See Smith (1976) for a bibliography .

76


typically directed at the works of particular persons, but rather at contrary formulations . He did not write an autobiography ; and although he published an autobiographical piece in the RMAC (Kantor, 1976), it shed very little light on th e influences and changing circumstances of his life, since most of the piece de scribes stages in the development of interbehaviorism. 6 For all of these reason s his intellectual heritage, as it relates to other thinkers, is difficult to trace . In the following section, therefore, we describe the evolution of his perspective as w e understand it . B . THE EVOLUTION OF ORGANISMIC PSYCHOLOG Y Kantor (1976) writes that his experiences as a young child greatly influence d his later perspectives . Having grown up in a highly diversified ethnic community, Kantor was exposed to continual disharmony between cultural groups . "Traditional beliefs, cultural rituals, diverse social relations, and economic opportunism", according to Kantor (1976, p . 121), led to conventional understandings within groups and discord between groups . He reasoned that conflicts between groups were a result of incongruities between pristine events, that is t o say, confrontable things and events subject to bare contact, and beliefs abou t those events, namely descriptive and explanatory constructions (Kantor, 1976) . Simply stated, different individuals described the same events in different way s and those descriptions were determined not only by contacts with events but als o in accord with conventional beliefs . The most salient observation to Kantor along these lines was the impositio n of a nonnatural element in the description of natural happenings . Inherent to most cultural belief systems, in another words, was the premise of a duality o f existences, conventionally understood as spirit and matter (Kantor, 1980a) . In the psychological domain, this dichotomy had been perpetrated as that of min d and body. Two of Kantor's earliest papers, "Conscious behavior and the abnormal" (1918) and "Human personality and its pathology" (1919), cited by him a s preliminary manifestations of the interbehavioral doctrine (Kantor, 1976) , clearly articulate a rejection of dualistic dogma . More important than his dismissal of dualistic thinking was the obligation i t provided to formulate a philosophical system void of it . His sensitivity to the unknowing imposition of cultural beliefs on scientific understandings became a cornerstone of his system building efforts . While he did not believe that scientists could ever completely disengage from the cultural matrix, he did believe that an awareness of influence from this source on scientific system buildin g was essential to the development of naturalism (Kantor, 1953, pp . 9–13 ; 1958 , pp . 34–36) . Kantor (1976) describes four stages in the development of naturalistic thinking in science, and it is reasonable to assume that these stages also describe hi s 6 See Mountjoy & Cone (in press) for a biography .


77

own development as a natural scientist . The first and second of these, as alread y mentioned, were a rejection of animism and a control of descriptive abstractions , respectively. The third stage entailed the development of field constructions ; and the forth, an explicit treatment of behavior fields . In the sections to follow, we discuss Kantor's progression and contributions through each of these stages . II . ZIGGURAT ' OF SCIENC E AND CIVILIZATIO N The first two of these stages, pertaining to fundamental assumptions and their logic, are illustrated in Kantor and Smith's (1975, p . 410) "Ziggurat of Scienc e and Civilization." Each tier of the ziggurat is built upon the tier below it . The bottom tier is cultural presupposition, upon which rests philosophical protopostulation, that is general, broad-based assumptions ; then scientific metapostulation, that is, more specific assumptions about science ; ending in the top tier with the postulates of the individual sciences (Kantor & Smith, 1975) . To put it another way, the disciplinary work of any individual scientist is rooted in a meta system of scientific assumptions, which is rooted in a matrix of philosophical assumptions, which, in turn, is formulated in keeping with the cultural persuasion s of the time (Kantor, 1958) . Thus, from Kantor's perspective, scientific work al ways begins with an examination of the culture within which the scientist work s (Kantor, 1953, pp . 46–48 ; 1958, pp . 4–6) . A . CULTURAL MATRI X In formulating psychology as a natural science, Kantor (1938, p . 8) contended that one must be concerned with "the ratio of traditional assertion to description based upon actual contact with phenomena" When this ratio was to o high, misdirection occurred in science . Evidence of scientific misdirection, perpetrated by an overproportional participation of biases from cultural sources , was examined in detail in his two-volume history of psychology (Kantor, 1963 , 1969a) . In this work, Kantor explored the development of western scientifi c views in a context of economic, political, religious, military, and other civilizational circumstances . At times, cultural conditions might be conducive to the development of naturalism in science ; at other times, not . Particularly disserviceable to this development, he argued, were conditions of extreme hardship for large groups of people and the sense of powerlessness that accompanied it . "Spiritistic" thinking, Kantor (1958, pp . 6–8 ; 1963, p . 158) argued, arises under such conditions as a means of adaptation ; it was sustained by the power ove r people it affords to smaller groups advantaged under such conditions as a resul t of other sets of cultural circumstances (1963, p . 158) . 7"Ziggurat" is defined in Webster's University Dictionary as a "terraced pyramid of successivel y receding stories ."

78


While civilizational circumstances, and the beliefs they engender, change d over time, particularly resistant to change was the assumption of a duality of existences . The resistance to change of supernatural or nonnatural constructions was partly due to the manner in which such entities become enmeshed in ordinary expression, whereby they were sustained in language itself . In addition to this facet of the problem, Kantor (1982a, pp . 199—201) outlined a number o f other conditions favoring the durability and dispersion of dualism as well as other constructions disserviceable to the evolution of naturalism. (Also see Hayes, Adams & Rydeen, 1994, for a discussion of these factors) . B . PROTOPOSTULATES OF SCIENTIFIC PHILOSOPH Y

The second level of Kantor and Smith's (1975) ziggurat is philosophy . At this level of scientific system building, protopostulates, or assumptions at the level o f the philosophy of science, are outlined . In addition to articulating these protopostulates, Kantor (1981, p . 85) provided a number of specifications for what h e called "valid thinking " at this level, through which the influence of cultural institutions might be minimized . These specifications are summarized below . 1 . Specifications for Valid Thinking A number of these specifications for valid thinking in philosophy centere d around the nature and use of language . Specifically, Kantor warned against conceptualizing language as a means of communicating ideas having their source s in internal, nonnatural entities . Instead, he argued, language was a prominen t form of human adaptation to the environment, which occurred by way of its referential function (Kantor, 1977b) . Its powerful role in this regard was subject t o distortion, however. Because linguistic forms ordinarily referred to aspects of th e nonverbal world, the existence of linguistic forms was assumed to implicate th e existence of their referents, despite an absence of contact with the latter . Kanto r (1969b) referred to such practices as autistic . By contrast, a proper orientation t o language events, articulated as one in which references were made only to wha t is there to be confronted, prevented this form of reification and its implication s for philosophy . Kantor further warned that once nonnatural referents were constructed, the y became the subject of pseudoproblems or verbal puzzles into which scientist s became entangled . As an example of this sort of entanglement, Kantor (1981 ) cites the question: "Can we be certain that there is an external world beyond ones own consciousness?" (p . 97) . The antidote for this problem, he claimed , was to be alert to the imposition of transcendences in scientific work . Proper postulation for the sciences was held to be assured, as well, if philosophers of science were able to escape from certain forms of intellectual bondage. Specifically, Kantor (1981, pp. 95—98) advocated a number of freedoms, including freedom from metaphysical institutions, freedom from universalism, an d freedom from absolutes . In his view, the development of naturalistic views in


79

science could not occur in the midst of struggles to find answers to metaphysical questions, among which he included how the existence of god might be prove d or disproved. No science, in Kantor's view, was concerned with "existences o r processes which transcend the boundaries of scientific enterprises . . . No scientific problem is concerned with a `Reality' beyond events and their investigation" (1958, p . 64) . One implication of this assertion was that questions concerning such entities had no scientific significance . Freedom from universalism was advocated on the grounds that scientifi c work bears the stamp of particular scientists and in as much as no scientist wa s in contact with the totality of all things, valid thinking concerning the univers e as an entity could not be accomplished . The "universe", in other words, is not a scientific concept . (Also see Zimmerman, 1979, on this point) . Freedom from absolutes was urged in appreciation of the fact that scientifi c knowledge was cumulative and corrigible . Valid thinking, thereby, did not ad dress itself to issues of absolute certainty, absolute existence, or absolute truth . Finally, Kantor (1981) claimed that the development of a naturalistic philosophy of science was intricately linked with naturalistic assumptions regarding th e nature of psychological events, particularly complex forms of these events . Fo r example, when thinking is taken to be evidence of the soul's psychic power, scientific system building becomes disconnected from actual observations and ac quires a nonscientific propositional structure . Philosophizing is itself psychological activity, and the adequacy of each depends on that of the other . 2. Protopostulates Proper The protopostulates of a scientific system define that system, as well as clarify the distinction between the formal assumptions underlying scientific thinking and the relatively ill-formed suppositions characteristic of the cultural matrix ou t of which they emerge . A number of these protopostulates are reflected in th e above guide to valid thinking . Remaining are those pertaining to the nature , practices, products, and scope of scientific work, as described below . With regard to the nature of science, Kantor (1958) claimed that science wa s a process of orientation to things and events, and the products that resulted fro m this orientation . In his words : "Science is the enterprise of interbehaving [i .e . interacting] with specific things and events which leads to a definite and precis e orientation with respect to those things and events" (Protopostulate 1, p . 64) . Scientific orientation, in turn, "concerns (a) the existence and identity of thing s and events or their components, and (b) the relationship between either the components of things and events, or between the various things and events them selves" (1958, Protopostulate 2, p . 64) . With regard to the practices of sciences, Kantor observed that different subject matters, as isolated by different sciences, require different means and methods of orientation. As he (1958) put it : "Scientific orientation requires specialized instruments and methods depending upon (a) the specific characteristics o f the events interacted with and (b) the specific problems formulated about them"

80


(Protopostulate 4, p . 64) . These various operations and practices "eventuate i n protocols (records), hypotheses, theories, and laws" (Kantor, 1958, Protopostulate 5, p . 64) . Finally, with regard to the scope of scientific work, Kantor suggested that scientific findings (i .e ., records concerning events and their investigation) and investigative results (i .e ., laws and theories) were sometimes applicable onl y within specific scientific enterprises, and sometimes within the larger culture ou t of which they had emerged . In either case, such applications, he claimed, constituted the authentic basis for scientific prediction and control . (Kantor, 1958 , pp . 65-66) . C. METASYSTEM OF THE SCIENCE S

The third level of the ziggurat of science and civilization, as articulated b y Kantor and Smith (1975), is made up of metapostulates concerning distinctions , commonalities, and interrelations among the various sciences, as well as thei r relative adequacies . The sciences were held to be distinguished by their subject matters by th e particular things and events isolated out of the larger matrix as their specific focus . Given the implications of the concept of the matrix, and an interpretation o f isolation as categorical construction, no grounds existed for suggesting that on e set of events was any more basic or fundamental than any other. While events o f one science often participate in the events of another for example, a psycho logical event necessarily involves a biological organism as a participant this copresence of factors did not suggest that one set of events was based on anothe r (Kantor, 1958) . It followed that the events of no science were reducible to thos e of any other. With regard to their interrelations among the various sciences, and in a general context of interdependence, Kantor (1938, p . 15 ; 1953, p . 5) claims that all sciences were relatively independent. From his perspective, dependencies exist only where subject matters overlapped (e .g., biochemistry, social psychology, astrophysics) . At the same time, all sciences are homogenous with all other sciences (Kan tor, 1938, p . 15 ; 1953, p . 66) . This metapostulate implied that, in the case of in congruence, differences could be traced to inadequacies of orientation to th e events isolated by the sciences in question . The issue here, as at the tier above , was commerce with dualism . In further regard to the relative adequacy of the various sciences, Kantor pointed to the comprehensiveness of their coverage of events, operations , and theoretical constructions making up their domain ; and to their relative awareness of systemological problems, referring again to the adequacy o f their orientation to underlying protopostulation . Finally, all sciences were assumed to be subject to continual corrective reformulation (Kantor, 1958, pp . 66-67) .


81

D . POSTULATES OF THE INDIVIDUAL SCIENCE S At the top tier of the ziggurat, are the assumptions of narrowest scope, pertaining to particular subject matter, means of study, and conceptual formulations . These propositions included descriptions of specific events, definitions of re search operations, and descriptions of constructional practices in the formulatio n of hypotheses, theories and laws (Kantor, 1958, Kantor & Smith, 1975 , pp . 514-417) . We turn now to an examination of the postulational system of the science of psychology, as formulated by Kantor . Naturalism in science i s achieved at this third stage of its development in the articulation of field constructions . III . POSTULATIONAL SYSTEM O F INTERBEHAVIORAL PSYCHOLOG Y A . INTERBEHAVIORAL FIELDS AS SUBJECT MATTE R With regard to the subject matter of psychology, Kantor (1958) took the position that psychology studied interbehavioral fields, comprised of : organisms an d stimulus objects, media of stimulation, and setting factors . Following from the metapostulate concerning the isolation of subject matters by different sciences mentioned above, Kantor further postulated that psychological interbehavior involved actions of the whole organism, not its parts taken separately or independently . The events making up such fields, he continues (1958, pp . 81-82), consiste d of adjustments of organisms to environing things, evolved from bioecological interbehaviors . These adjustments were further held to develop over the lifetime s of individual organisms .' The concept of the interactional field and the premise of continuous evolution implied, in addition, that psychological events occurre d without any internal or external determiners . B . FACTORS PARTICIPATING IN INTERBEHAVIORAL FIELD S While for Kantor the interbehavioral field was all of one piece as a crude dat a event, he held it useful for analytical purposes to describe the factors participating in such fields (Kantor, 1958, pp . 87-90) . Among them were: organismic factors, including organisms, responses, and response functions ; stimulational factors, including objects, stimulus objects, and stimulus functions ; interbehavioral history, subdivided into the organismic and stimulational categories of reactional biography and stimulus evolution, respectively ; media of contact; and settin g factors . $Instincts, inherited behaviors, inborn predispositions, and susceptibilities of all sorts are foreig n to Kantor' s position . See Hayes, Adams, & Dixon (1996) for further discussion .

82


1. Organismic Factors Kantor held that the organism, as a biological entity, participates in a psychological field as an interrelated factor, not as an aspect of the psychological fiel d itself. It constituted merely a locus of response functions (1969a, p . 375) . Responses were conceptualized, from the standpoint of an observer, as potential response functions understood in this manner by virtue of their having historicall y participated in relation to stimulation . 9 Current responding with respect to stimulation, or response function, constituted the essential organismic factor in an interbehavioral field of the psychological variety. 2. Stimulational Factors An object, as a physical entity, likewise participated in a psychological fiel d as an interrelated factor, not as an aspect of the psychological field itself . An object constituted merely a locus of stimulus functions (Kantor, 1969, p . 375) . Stimulus objects were conceptualized, from the standpoint of an observer, as potential stimulus functions understood in this manner by virtue of their havin g historically participated in relation with responding . Currently occurring stimulation with respect to responding, or stimulus function, constituted the essentia l stimulational factor in an interbehavioral field . Neither response functions nor stimulus functions are held to exist separatel y or independently as crude data events . Rather, each implied the other : They wer e symmetrical and reciprocal . Taken together, they constituted a unitary, contemporary, psychological phenomenon . 3. Interbehavioral Histor y As postulated, psychological events (e .g., stimulus–response functions) developed continuously over the lifetimes of organisms . This development was captured, from the organismic side, in the concept of the reactional biography ; and from the stimulational side, as stimulus evolution . Each implied the other : They were symmetrical and reciprocal in their relation (Kantor, 1958, p . 89) . Taken together they constituted interbehavioral history, which consisted of th e evolution of interbehavioral fields . 4. Media of Contact Kantor (1958, p . 89) contended that psychological events included specifi c contact media in addition to stimulus and response functions, whose role it wa s to enable functions to occur at particular times . For example, in order to interbe9The definition of responses in Kantor's system is not entirely clear . The parallel between organ isms, responses, and response functions on one hand, and objects, stimulus objects, and stimulu s functions on the other, is not fully articulated . The definition of responses included herein is an at tempt to clarify this ambiguity by likening responses to the parallel concept of stimulus objects, fo r which a more precise definition is provided by Kantor (1958, p . 88) . It is also possible that Kanto r intended responses to be distinguished from response functions in that the former pertained to th e forms or morphological characteristics of responses in current interaction with stimulation (Kantor , 1958, p . 88) .


83

have visually with a colored object, light is an indispensable enabling factor . (For a discussion of the significance of this concept, see Kantor, 1922) . 5 . Setting Factors Setting factors (Kantor, 1958, p . 89) were conceptualized by Kantor as general surrounding circumstances . They might be localized in stimulational factors , organismic factors, or both . Among the factors constituting the organismic setting are such events as deprivation, fatigue, illness, intoxication, and so on . The role of setting factors was to influence the occurrence or nonoccurrence of particular stimulus—response functions, or to facilitate the occurrence of th e functions in question in varying degrees . Essentially, setting factors were held t o select or actualize particular functions at particular moments .

IV. PSYCHOLOGICAL EVENT S

The fourth stage in the development of naturalism in science, as suggested b y Kantor, occurred with the explicit treatment of different sorts of interbehaviora l fields . To illustrate Kantor's treatment in this regard, we present his four genera l classes of interbehavior . A . GENERAL CLASSES OF INTERBEHAVIO R

Kantor (1924) categorized psychological events into five general classe s based on the circumstances of their origin in the repertoires of individuals an d their distribution among members of the species . Among them are universal actions, basic and suprabasic actions, idiosyncratic actions, and cultural actions, a s summarized below . 1. Universal Behavio r Kantor reasons that some of the actions of individuals are shared among al l members of the species as an outcome of their common organismic characteristics and the natural properties of confronted stimulus objects . All humans salivate when lemon juice is squirted into their mouths, for example . This commonality of action was owed to the natural properties of lemon juice (e .g ., its acidity), and the characteristics of the biological organisms who reacted to thi s substance in this way . Actions having this provenance were called "universal" i n Kantor's (1924, p . 49) system as a reference to the universality of their distribution across the species . The same sort of activity is called "unconditioned" o r "unlearned" in other systems of behavioral psychology . 2. Idiosyncratic Behavior Some of what individuals do is acquired over the course of their lives, an d some of this activity had its origins in the distinctly idiosyncratic circumstances

84


of those individuals . Such behavior is not shared with other members of th e species, except when other members' life circumstances are accidentally simila r (Kantor and Smith, 1975, pp . 41–42) . Inasmuch as no two persons may be in exactly the same place at exactly the same time, no two persons have the sam e history . This is to say, no two persons' idiosyncratic repertoires are the same . This type of activity was called "idiosyncratic" as a reference to the specificit y of its occurrence (Kantor & Smith, 1975, pp . 126–127 ; see also Kantor, 1924 , p . 49) . 3. Cultural Behavio r Cultural actions were said to have some of the characteristics of both universal and idiosyncratic actions . Like universal actions, cultural acts are shared , though not with all members of the species . They are not in this sense "universal," but rather have their origins in ontogenic circumstances that differ across individuals . Like idiosyncratic actions, cultural acts are acquired . More specifically, cultural responses are held to be acquired with respect to common or conventiona l stimulus functions . Such functions operate independently of the natural proper ties of the stimulus objects in which they inhere, which is what is meant by "cultural" in this context (Kantor, 1924, p . 50) . The commonness of the functions of cultural stimulus objects, moreover, is not an accidental result of two or mor e persons reacting in the same idiosyncratic way to them . Rather cultural stimulu s objects have common functions because people had endowed them with specia l properties . Unlike universal actions, which occur as an unavoidable outcome o f interactions involving certain natural properties of stimuli and the biologica l constituencies of organisms, cultural responses are acquired through a culturalization process operating under specific group circumstances (Kantor, 1982a) . For example, a voodoo doll is an object that stimulates cultural reactions o n the parts of a particular group of people . The object itself is not harmful in any way, and fear therefore does not arise as a reaction to the natural properties of a voodoo doll. Rather, fear reactions arise because a functional property of injuriousness has been attributed to this object over the course of a particular group' s cultural evolution (Hayes, Adams and Rydeen, 1994) . To reiterate, cultural behavior is shared action with respect to conventional stimulus function arising under specific group auspices . 4. Basic Behavio r Basic reactions constituted fundamental forms of action, according to Kanto r (1924, pp. 159-181), because they were acquired very early in an individual' s history and thereby served to influence all of the individual's later conduct . While basic reactions "have more to do with things and their crude manipulation, and less with complex understanding and intelligent transformation of suc h objects" (Kantor, 1924, p . 161), the term "basic" was not used to suggest a relative simplicity of action . It referred, instead, to the fact that such behaviors were


85

acquired very early. Moreover, their very early acquisition implicated the participation of other persons in their occurrence . Basic behaviors might remain mor e or less permanently an aspect of an individual's repertoire ; or they might becom e integrated into more and more complex forms of supra-basic behavior . B . VARIETIES OF PSYCHOLOGICAL EVENTS

Kantor's two-volume Principles of Psychology (1924, 1926) undertakes to define and distinguish psychological events of every variety on the basis of th e factors participating in them and the organization of those factors, among othe r criteria . The details of these analyses exceed the scope of this exposition . His coverage is comprehensive, addressing an enormous variety of specific types o f psychological events . Some of these topics were of particular interest to him , eventuating in longer treatises, including : psychological language events (1936 , 1976) and social psychology/cultural behavior (1924, 1982a) . See Parrott (1984 ) for a more extended discussion of his thematic interests . As examples of his work in this regard, we mention two topical areas i n which Kantor's contribution has been particularly astute and is deserving of further consideration by the interested reader. 1. Perceptio n Perceptual activity has been a perennial subject of inquiry throughout the history of science due to the significance of this concept to the issue of scientifi c observation and its epistemological implications . Kantor's (1920a, 1924, 1980b ) analysis of perception is particularly provocative . Perceptual activity, like all other psychological events in interbehavioral perspective, is understood as a n endpoint in a history of functional relations . As such, the cultural and idiosyncratic history of the observer is implicated in a current act of perception . In short, not only do we perceive what is there to be seen, but also what has bee n there over the course of our histories with particular stimuli . Perceptual activity is described as semi-implicit for this reason (Kantor, 1924, p . 249) . Drawn into question by this analysis is the dichotomy of objectivity and subjectivity in science . (See Hayes (1997) for a discussion of the implications of this analysis fo r science . ) 2. Linguistic Acts Kantor's (1936, 1977a) treatment of psychological linguistics has a numbe r of features to recommend it . Among them are a distinction between events o f linguistic function from those of mere linguistic form. In addition, his analysi s makes sense of the concept of reference from a naturalistic standpoint, throug h which the linguistic activity of the listener is elucidated . Linguistic adjustments are held to differ from other psychological events i n that, unlike other events which are comprised of a single response function coordinated with a single stimulus function, linguistic adjustments are constituted of

86


a single response function coordinated with two stimulus functions simultaneously (1936, pp . 73—74) . The response function of the activity of a speaker i s coordinated with stimulus functions having as their sources, on one hand, a listener, and, on the other hand, the thing or event spoken of . This bistimulational analysis made for a referential interpretation of linguistic activity : a speaker (i .e . , referrer) was said to refer someone (i .e ., a referee) to something (i .e ., a referent) . For an analysis of the implications of Kantor's perspective on this subject matter , and a comparison of his formulation with that of other behavior scientists see Schoenfeld (1969) and Parrott (1984, 1986) .

V. INTERBEHAVIORISM AN D RADICAL BEHAVIORIS M The principle commonality between interbehaviorism and other forms of behaviorism is its explicit rejection of dualism and its corresponding adoption o f the behavior of the whole organism in relation to a stimulating environment to be its subject matter. How this subject matter is conceptualized differs across behaviorisms, however ; and when examined in detail, Kantor's position bears littl e similarity to most other forms . Skinner's radical behaviorism shares some features with interbehaviorism, at least on the surface (See Parrott, 1983, for a mor e complete discussion of similarities and differences between interbehavioris m and radical behaviorism) . Hence, we restrict our discussion to this comparison , assuming that differences of greater magnitude, as may exist in more distantl y related behaviorisms, may be extrapolated from this comparison .

A . SIMILARITIE S

1. Philosophical Underpinnings At a relatively gross level, radical behaviorism and interbehaviorism are alike in their commitments to realism and pragmatism : Both Skinner and Kantor assumed the existence of the natural world ; and both evaluated accomplishment s in terms of their utility. Moreover, as behaviorists, both eschewed dualism . 2. Psychological Subject Matter Both interbehaviorism (Kantor, 1938, p . 33) and radical behaviorism (Skinner, 1938, p . 6) took their subject matter to be the behavior of organisms in relation to their stimulating environments, and both took behavior to imply the action of the whole organism . Behavior was described by Kantor (Kantor an d Smith, 1975, pp. 5—6) as "adjustmental" in character. Skinner made a similar point, describing behavior as "adaptive" (1953, p . 90) .


87

B . DIFFERENCES

Apart from these similarities, the two positions are rather different (althoug h see Morris, 1992, for a different perspective) . Some of these differences ma y be seen as mere differences in emphasis, while others reflect irreconcilable premises . We consider the central concepts of response, stimulus, functional relation and psychological history in the following section. 1. Concept of Behavior Skinner's conception of behavior as "action of the organism on the outsid e world" (1938, p . 6) was regarded as incomplete by Kantor. For Kantor, behavior was not the action of an organism but the interaction of an organism and an environment. Behavior was mutually corresponsive, not emissive (Kantor, 1970) . 2. Concept of Stimulus The concept of the stimulus in the two positions is also somewhat different. Kantor (1970) argued that the experimental analysis of behavior "stops short at the surface notion of a stimulus as simply an object or condition that determine s a response" (p . 106) . Objects, stimulus objects and stimulus functions are more differentiated in interbehavioral theory, as previously discussed . Prior to the establishment of an interaction between the organism and environing things, those things are simpl y objects without psychological significance . As an interaction is established, objects take on stimulus functions corresponding to specific response functions . Only at this point do objects become stimulus objects . The matching function s between stimuli and responses, thus established, constitute psychological adjustments (Kantor, 1970) . From Kantor's perspective, what was missing in the radical behavioral formulation is the interaction of stimulation with responding (Kantor, 1969a, p . 377) . Interaction implies action on the parts of both organisms and stimulus objects . That is to say, responding is something that an organism and stimulus object d o with respect to each other. For this reason, stimulating is described as the "mutual and corresponding behavior of an object in an interactional field, along wit h the action performed by the organism" (Kantor & Smith, 1975, p . 32) . 3. Concept of Function The concept of functional relation is central to both radical behaviorism an d interbehaviorism ; however, the term "function" is interpreted quite differently . Function is understood in radical behavioral thinking as something an objec t does to responding in a causal sense (Skinner, 1953, p . 23 ; Parrott, 1983a) . More specifically, stimuli are said to strengthen operants (i .e ., reinforcing stimuli), se t the occasion for responses (i .e ., discriminative stimuli), and to elicit responses

88


eliciting stimuli of the respondent framework) (Skinner, 1953, pp. 107 – 128) . This use of the term function is at odds with Kantor's use of this term . For Kantor (1969a, p . 379), stimulus functions are only the participation of stimulu s objects in psychological events, abstracted from observed correlations of stimulating and responding . As such, stimuli have no special status in these correlations : "They are a phase or factor in the event, an essential part of it and not a preceding or antedating cause" (Kantor, 1938, p . 37) . It might be argued that Skinner does not emphasize the interdependency o f the coordinants in functional relations as does Kantor, but nonetheless recognizes this aspect of their character . The following passage (among others) might be cited as evidence of this recognition: (i.e.,

The relations between the controller and the controlled [are] reciprocal. The scientist in the laboratory, studying the behavior of a pigeon, designs contingencies and observe s their effect. His apparatus exerts conspicuous control on the pigeon, but we must not for get the control exerted by the pigeon . The behavior of the pigeon had determined the design of the apparatus, and the procedure which is used . Some such reciprocal control is characteristic of all science . (Skinner, 1971, p. 161) .

This passage does not speak to the issue, however, What Skinner is suggesting here is that when a pigeon and a scientist interact, sometimes the stimulu s product of the scientist's behavior is the independent variable with respect to th e pigeon's behavior; and sometimes the stimulus product of the pigeon's behavio r is the independent variable with respect to the behavior of the scientist . He is not suggesting that the actions of the scientist and pigeon are interdependent . Interdependent events don't "exert" anything . From the standpoint of interbehavioral thinking, Skinner's misstep is a product first of focusing on only one coordinant of a functional relation at a time ; and second, assuming an identity between functional and causal relations . For Kan tor, the relation between a stimulus and a response depends as much upon the response as it does on the stimulus . Neither coordinant may be abstracted from th e functional relation to account for the other factor not to mention accountin g for itself as well (See Parrott, 1983a, for further discussion) . Skinner's failure to fully appreciate the interdependency of functional relations, reflected in the dichotomy of dependent and independent variables, and i n the confusion of functional with causal variables produced by this dichotomy, i s the source of another difference between radical behaviorism and interbehaviorism . Specifically, the concept of history, central to both positions, is interpreted differently. 4. Concept of History What evolves throughout the course of ontogenic history 10 in Skinner's formulation is an organism, not functional relations, as held in interbehaviorism. 10For a critique of Skinner's view of phylogenic history along interbehavioral lines, see Parrott , 1983a; and Hayes, Adams, & Dixon, 1996) .


89

Skinner (1974, p . 215) contends that an organism is changed by its exposure t o contingencies of reinforcement ; and these changes are what enable it to reac t differently to stimulus events from one moment in this history to the next . Kantor's problem with this type of theorizing is twofold . First, it is not based on confrontations with events : biological evidence to substantiate the hypothesi s of permanent changes in an organism's physiology as a result of exposure t o contingencies of reinforcement does not exist . Second, Skinner's suggestion tha t physiological evidence will explain "why an organism then behaves in a different way possibly at a much later date" (Skinner 1974, p . 215) is organocentric, reductionistic, and causal, none of which are palatable to interbehaviorists (Kan tor, 1969a, pp . 364-368) . For Kantor, the product of interbehavioral history is not a changed organism, but rather changed functional relations . What evolves over the course of an organism's history is its responding with respect to stimulating . 5. Concept of Setting Interbehaviorism and radical behaviorism also differ as to their conceptualizations of the setting or context in which functional relations are taking place . The setting is differentiated from stimulus events in Kantor's perspective, as discussed above . This distinction is not addressed by Skinner. He appeals to a number of variables which do not fit his stimulus category, including deprivation, aversive stimulation and audience variables (Skinner, 1957) . They have not been collectivel y nominated by him since the concept of "drive" was abandoned sometime afte r 1953, however, and have not achieved full paradigmatic significance in hi s system . Parrott (1983a) has argued that it is not setting factors that are missing fro m the radical behavioral account . Discriminative and reinforcing stimuli have th e character of the setting factors of interbehavioral formulation . In her view, it i s stimuli that are missing from the radical behavioral account, and this is due to the underlying causal interpretation of functional relations .

VI . RELATION TO EMPIRICAL EVIDENC E The science of behavior, like most other sciences, has both an investigativ e and an applied branch . Its applied branch is relatively new, having been fostere d in response to problems of adjustment surfacing after the second world war . Its investigative branch is of somewhat longer duration, beginning in the early par t of the century. Longer than either of these is its tradition as a philosophical enterprise : Philosophical positions akin to behaviorism date back at least as far a s Aristotle, if Kantor's (1969, p . 357 ; Smith, 1973) interpretation of Aristotle' s position has credence (Mountjoy, 1966, p . 8) . It is probably safe to say that human behavior has always been a subject of serious contemplation and

90


speculation . Each of these three branches of the science of behavior has contributed to our understanding of behavior each in its own way, with respect t o its own aims . Many behavior scientists of the contemporary period are inclined to find value only in empirical contributions, however . Their own success in producin g contributions of this sort no doubt contributes to this bias, along with ignoranc e as to the role and significance of scientific logic and scientific philosophy . Regardless of the reasons for it, behavior scientists view the investigative branch o f their science to be its most important feature . Hence we must examine Kantor' s contributions to this aspect of scientific work . A . KANTOR'S CONTRIBUTIONS TO EMPIRICAL RESEARC H Kantor did not conduct empirical research . Neither was he a practitioner . Thi s was not seen by him as a personal weakness . It was not an accident (Verplanck, 1983) not a happenstance of opportunity . He made no apology for it . Neithe r did he see it as an inadequacy of his position . These were simply not his roles . J. R. Kantor was a philosopher of the science of behavior. Kantor did not pursue his role as a philosopher in disregard or disrespect o f research and practice, however. On the contrary, from his perspective, philosophizing was intimately connected with these domains . In his view, valid practices and outcomes of science were fundamentally dependent on valid scientifi c philosophizing (Observer, 1968, 1969, 1980) . Included among these aspects o f science were: distinguishing between the authentic existence of events and autistic constructions presumed to be descriptive and explanatory of such existenc e (Kantor, 1981, p . 11) occurring in observational phases ; the systematic organization of investigations and findings (Kantor, 1981, pp . 109) ; and the developmen t of generalizations or laws concerning the events observed (Kantor, 1977b, pp. 9) . In short, his contribution to research and practice, was made explicitly an d deliberately by way of his philosophical work . Kantor's most concentrated and comprehensive treatment of scientifi c methodology and logic is presented in his The Logic of Modern Science (1953) . The value he placed on experimentation is evident in this work . Kantor (1953 , 1978) viewed experimentation as encompassing a wide scope of activities, however ; and warned against ritualism in this domain through which methods an d instruments became the defining characteristics of science . He states "laboratory techniques are certainly not the exclusive methods of investigating events, an d when they become prescribed forms of operations they tend to function as research-hampering traditions" (1953, pp . 104) . In this regard, Kantor (1970) entreated the readership of the Journal of the Experimental Analysis of Behavior to avoid characterizing all psychological events in terms of a limited number of measures, namely rate, ratio, and latency; and to expand its scope "to include the free investigation of human organisms . . . including perceiving, remembering, thinking, and feeling behavior . . . . emotional behavior, volitional and voluntary actions, the creative processes of


91

imagination as well as the behavior called inventing, thinking, problem solving , and reasoning" (pp . 103, 105) . Kantor also addressed himself to numerous areas of application, among them : education (1975), intelligence testing (1920b), psychopathology (1919), and mental retardation (1982b) . In all of these areas his aim was the same : To overcome problems of interpretation and their disserviceable implications for practice occurring as a result of improper postulation and systemic inconsistencies . I t is at least plausible that greater awareness of cultural presuppositions and scientific postulation may lead to greater sophistication and effectiveness in applying scientific knowledge to issues and problems of societal significance. B. OTHERS ' ASSESSMENT OF KANTOR'S CONTRIBUTION S Kantor's assessment of his contribution to the investigative domain is no t shared by other behavior scientists, however. In a survey of then-current and pas t editorial board members of the Journal of the Experimental Analysis of Behavior, Journal of Applied Behavior Analysis, and Behaviorism (Morris, Higgins , and Bickel, 1983), the "lack of a research tradition, or even relevance to research, either basic or applied" was cited as one of the most significant weaknesses of interbehavioral psychology . C. SPHERE OF INFLUENCE Still, Kantor's influence has been felt to some degree . In a citation study conducted by Smith and Ray (1981), covering the years 1917 1976, the author s found 36 references to interbehaviorism in experimental and case studies . The authors concluded that although Kantor's perspective is infrequently cited, it i s viewed positively when adopted, and has therefore largely gone unnoticed rather than examined and found wanting (Smith & Ray, 1981) . In a more recent review by Smith (1996b), over ninety research article s claimed an intellectual debt to interbehaviorism explicitly . These papers were published in a wide variety of journals from mainstream behavior analyti c journals such as the Journal of Applied Behavior Analysis and the Journal of the Experimental Analysis of Behavior, to more remotely connected outlets, including Applied Systems and Cybernetics and the Journal of Emotional and Behavioral Disorders (Smith, 1996b) . Additionally, an increasing trend is evident in citations ; from 8 between 1937—1952 to 56 between 1980—1995 . Smith (1996b ) concludes that contrary to received opinion, interbehavioral research is occur ring, increasing, and demonstrates "the potential for dramatic new directions i n psychology" (p . 11) . D. DOMAINS OF INFLUENC E Interbehavioral thinking is evident in scholarly work in a number of domains among which may be included : human development (Bijou, 1976, 1992, 1993 , 1996 ; Bijou & Ribes, 1996 ; Bijou & Baer, 1978), gerontology (Fredericks &

92


Fisher, 1997), psychological linguistics (Bijou, Umbreit, Ghezzi, & Chao, 1986 , 1988 ; Bijou, Chao, & Ghezzi, 1988 ; Ghezzi, Bijou & Chao, 1991 ; Chaisson & Hayes, 1993), psychopathology (Delprato, 1995 ; Pronko, 1963), learnin g (Schoenfeld, 1969), physics (Smith & Smith, 1996), and in the study of subjectivity (Stephenson, 1953) . This last area warrants further comment . In The Study of Behavior, Stephen son (1953) outlines a method for the study of subjectivity based on the premis e that "what is subjective, such as thinking, and what is observable to others, suc h as playing golf, are in no way distinguishable for scientific purposes" (p . 4) . Stephenson cites Kantor's formulation as a basis for his Q-methodology ; a method by which statements are sorted with regard to the degree of an individual's agreement or disagreement then factor analyzed . Q-method has been used to study a wide range of phenomena not traditionally included in the domain o f the behavior scientist . For example, it has been used to study phenomenolog y (Brown, 1994 ; Delprato & Knapp, 1994 ; Smith, 1994), political subjectivit y (Brown, 1980), nursing (Dennis, 1986), and humor (Kinsey, 1993), to name a few. For a more complete discussion of empirical research from a field theoretica l perspective similar to Kantor's, the reader is referred to Ray and Delprat o (1989) .

VII . CURRENT STATUS AN D FUTURE DIRECTION S The death of a scholar always marks the beginning of the end of his or her influence on the scientific community, at least in terms of its explicit formulation . While devotees may sustain its explicit influence for some period, there is al ways a drift in thinking over time until the connection with the leader become s obscure, at best . When the devotees are few in number, as in the case of interbehavioral psychology and interbehaviorism, the magnitude and duration of specific influence is necessarily limited . The impact of particular scholarly contributions also depends on the nature o f those contributions themselves, specifically their integration in much broader evolutions of cultural understanding . When they reflect, and especially whe n they spearhead these movements, they may sustain a presence beyond the life times of their leaders and disciples in their role as examples of articulation o f these new and unfamiliar understandings (Smith, 1991, Morris & Midgley , 1990) . Kantor's position is assured of this future . His contribution to psychology has been to provide a formulation of this subject matter in the field theoretical perspective of contemporary science more generally —a formulation as yet to be fully adopted in psychology and which may find illustration in Kantor's work (Smith, 1991 ; Morris & Midgley, 1990) .


93

VIII . EPILOGU E Kantor spent his life trying to overthrow what he called "spookology", refer ring to animistic positions in psychology and philosophy . It is endearing to know that at the time of his death, in his 96th year, a handwritten note on his des k read: "No spirits, wraiths, hobgoblins, spooks, noumena, superstitions, transcendentals, mystics, invisible hands, supreme creator, angels, demons, . . . (Delprato, 1996 ; Mountjoy and Hansor, 1986) .

REFERENCE S Bijou, S . W. (1976) . Child development : The basic stage of early childhood . Englewood Cliffs, NJ : Prentice-Hall . Bijou, S . W. (1992) . Behavior analysis . In R. Vasta (Ed .), Six theories of child development (pp . 61-132) . London and Philadelphia : Jessica Kingsley Publishers . Bijou, S . W. (1993). Behavior analysis of child development (2nd rev.) Reno, NV: Context Press . Bijou, S . W. (1996). Behaviorist approaches to human development. In E . De Corte & F. Weinert (Eds .), Encyclopedia of developmental and instructional psychology . Oxford, England : Elsevier Science Limited. Bijou, S . W., & Baer, D . M . (1978). Behavior analysis of child development (rev ed.) . Englewoo d Cliffs, NJ : Prentice-Hall. Bijou, S . W., Chao, C ., & Ghezzi, P. M. (1988) . Manual of instructions for identifying and analyzin g referential interactions, II . The Psychological Record, 38, 401-414 . Bijou, S . W., & Ribes, E . (Eds .) . (1996) . New directions in behavior development . Reno, NV: Context Press. Bijou, S. W., Umbreit, J., Ghezzi, P. M., & Chao, C . (1986) . Psychological linguistics : A natural science approach to the study of language interactions . The Analysis of Verbal Behavior, 4, 23-29 . Bijou, S . W., Umbreit, J ., Ghezzi, P. M., & Chao, C . (1988) . Manual of instructions for identifyin g and analyzing referential interactions . The Psychological Record, 36, 491-518. Brown, S . R. (1980). Political Subjectivity : Application of Q methodology in political science. New Haven, CT : Yale University Press . Brown, S . R . (1994) . Q methodology and interbehavioral phenomenology . The Interbehaviorist, 22(3), 24-26. Chaisson, C ., & Hayes, L . J . (1993). Referential interactions among persons of different status . The Psychological Record, 43, 13-24 . Delprato, D. J . (in press) . Converging Trends with Interbehaviorism . In B . D . Midgley & E . K. Morris (Eds .), Modern Perspectives on J. R. Kantor and Interbehaviorism . Greenwood Press . Delprato, D. J . (1995) . Interbehavioral psychology : Critical, systematic, and integrative approach to clinical services. In W. O ' Donohue & L . Krasner (Eds .), Theories of behavior therapy : Exploring behavior change. Washington, DC : American Psychological Association. Delprato, D . J . (1996) . Jacob Robert Kantor, radical humanist. Presented at a workshop J. R . Kantor's Interbehavioral Psychology : Beyond Mechanism and Mentalism, Cheiron Society, Earlham College, Richmond, Indiana, June 27-30, 1996. Delprato, D . J ., & Knapp, T. (1994) . Q methodology and interbehavioral descriptions . The Interbehaviorist, 22(3), 14-23 . Dennis, K. E. (1986) . Q methodology : relevance and application to nursing research . Advances i n Nursing Science, 8(3), 6-17 .

94


Fredericks, D . W., & Fisher, J. E . (1997) . A behavior analytic alternative to cognitive developmenta l approach to dementia. Paper presented at the twenty third annual convention for the Association for Behavior Analysis, Chicago, IL . Ghezzi, P. M ., Bijou, S . W., & Chao, C . (1991) . Manual for training raters to identify and analyz e referential interactions. The Psychological Record, 41, 473—486 . Hayes, L . J . (1997) . Scientific knowing in psychological perspective . In L. J . Hayes & P. M. Ghezz i (Eds .), Investigations in behavioral epistemology . Reno, NV : Context Press . Hayes, L . J ., Adams, M . A ., & Dixon, M. (1996) . Causal constructs and conceptual confusions . The Psychological Record, 47, 97—112 . Hayes, L . J ., Adams, M . A., & Rydeen, K. (1994). Ethics, choice and value . In L . J . Hayes, G. J. Hayes, S . C. Moore, & P. M . Ghezzi (Eds.), Ethical issues in developmental disabilities (pp. 11—39) . Reno, NV: Context Press . Kantor, J. R. (1918) . Conscious behavior and the abnormal . Journal of Abnormal Psychology, 13, 158-168 . Kantor, J . R . (1919) . Human personality and its pathology . Journal of Philosophy, Psychology, an d Scientific Methods, 16, 236—246 . Kantor, J . R. (1920a) . Suggestions toward a scientific interpretation of perception . Psychologica l Review, 27, 191—216 . Kantor, J. R . (1920b) . Intelligence and mental tests. Journal of Philosophy, 17, 260—268 . Kantor, J . R. (1922) . Can the Psychological experiment reconcile the introspectionists an d objectivists? American Journal of Psychology, 32, 481—510. Kantor, J . R. (1923) . What are the data and problems of social psychology? The Journal of Philosophy, 20(17), 449—457 . Kantor, J. R . (1924) . Principles of psychology, vol . I. Chicago : The Principia Press . Kantor, J. R . (1926) . Principles of psychology, vol II. Chicago: The Principia Press . Kantor, J. R . (1929) . An outline of social psychology. Chicago : Follett. Kantor, J. R . (1936) . An objective psychology of grammar. Chicago: The Principia Press . Kantor, J. R . (1938) . The nature of psychology as a natural science . Acta Psychological, 4,1—61 . Kantor, J. R . (1945) . Psychology and Logic, Volume I. Chicago : The Principia Press . Kantor, J . R . (1947) . Problems of physiological psychology. Granville, OH : The Principia Press . Kantor, I. R . (1950) . Psychology and Logic, Volume II. Chicago: The Principia Press. Kantor, J . R . (1953) . The logic of modern science. Chicago : The Principia Press . Kantor, J . R . (1958) . Interbehavioral psychology. Chicago: The Principia Press . Kantor, J . R . (1963) . The scientific evolution ofpsychology, vol I. Chicago : The Principia Press . Kantor, J . R . (1969a) . The scientific evolution ofpsychology, vol II. Chicago : The Principia Press. Kantor, J . R. (1969b) . Scientific psychology and specious philosophy. The Psychological Record, 19, 15—27. Kantor, J. R. (1970) . An analysis of the experimental analysis of behavior (TEAB) . Journal of th e Experimental Analysis of Behavior, 13, 101—108 . Kantor, J . R. (1971a) . The aim and progress of psychology and other sciences . Chicago : Th e Principia Press. Kantor, J. R. (1975). Education in psychological perspective . The Psychological Record, 25, 315-323 . Kantor, J. R. (1976) . The origin and evolution of interbehavioral psychology . Revista Mexicana de Analisis de la Conducta, 2(2), 120-136 . Kantor, J . R . (1977a). Evolution and revolution in the philosophy of science, Revista Mexicana de Analisis de la Conducta, 3(1), 7—16 . Kantor, J . R . (1977b). Psychological Linguistics. Chicago: Principia Press . Kantor, J . R . (1978) . Experimentation : The acme of science. Revista Mexicana de Analisis de la Conducta, 4(1), 5—15 . Kantor, J. R. (1980a) . Manifesto of interbehavioral psychology . Revista Mexicana de Analisis de la Conducta, 6(2), 117—128 .


95

Kantor, J . R . (1980b) . Perceiving as science and as traditional dogma . Revista Mexicana de Analisis de la Conducta, 6, 3-16 . Kantor, J . R. (1981). Interbehavioral philosophy. Chicago: The Principia Press . Kantor, J . R. (1982a) . Cultural psychology. Chicago : The Principia Press . Kantor, J . R . (1982b) . Psychological retardation and interbehavioral adjustments . The Psychological Record, 32, 305-313 . Kantor, J . R. (1983) . Tragedy and the event continuum. Chicago : The Principia Press . Kantor, J . R., & Smith, N . W. (1975) The science of psychology : An interbehavioral survey. Chicago : Principia Press . Kinsey, D . F. (1993) . Humor communicability. Operant Subjectivity, 17(1/2), 49–61 . Leahey, T. H. (1992) . A history of psychology : Main currents in psychological thought . New Jersey : Prentice Hall . Lundin, R . W. (1979) . Systems and theories of psychology, (second edition) . Lexington, MA: D . C . Heath and Company. Morris, E . K. (1992). The aim, progress, and evolution of behavior analysis . The Behavior Analyst , 15, 3–29 . Morris, E . K ., Higgins, S . T., & Bickel, W. K . (1983) . Contributions of J . R . Kantor to contemporary behaviorism. In N . W. Smith, P. T. Mountjoy, & D . H. Rubens (Eds.), Reassessment in psychology: The interbehavioral alternative . Washington, D. C . : University Press o f America. Morris, E . K., & Midgley, B . D. (1990) . Historical and conceptual foundations of ecobehavioral analysis . In S . R . Schroeder (Ed .), Ecobehavioral analysis and developmental disabilities : Th e twenty-first century (pp . 1-32) . New York: Springer-Verlag. Mountjoy, P. T. (1966) . New dimensions in the history of psychology : A review. Journal of The Scientific Laboratories, Denison University, 47, 5-12 . Mountjoy, P. T. (1994) . Paul Mountjoy on Kantor Graduates. The Interbehaviorist, 22(3), 4 . Mountjoy, P. T. (1997) . Another new journal? The Psychological Record: Volumes I–V; 1937-1945 . The Psychological Record, 47, 3-20 . Mountjoy, P. T., & Cone, D . M. (1995) . The foundational nature of philosophical categories : Jacob Robert Kantor's doctoral dissertation. The Interbehaviorist, 23(2), 5-8. Mountjoy, P. T., & Cone, D . M . (in press) . A biography of J. R . Kantor. In B . D . Midgley & E . K . Morris (Eds .), Modern Perspectives on J. R . Kantor and Interbehaviorism . Greenwood Press . Mountjoy, P. T., & Hansor, J . D . (1986). Jacob Robert Kantor (1888-1984) . American Psychologist, 41(11), 1296-1297 . Observer (1968) . The ambivalence of psychology toward philosophy . The Psychological Record, 18, 641-643 . Observer (1969) . Comments and queries : The basis fallacy in psychology . The Psychological Record, 19, 645-648 . Observer (1979) . Comments and queries : Observations on the history of psychology. The Psychological Record, 29, 256-571 . Observer (1980) . Comments and queries : Theological psychology vs scientific psychology . Th e Psychological Record, 30, 131-133 . Observer (1983) . Comments and queries : Reflections concerning psychological historiography . The Psychological Record, 33, 131-138 . Parrott, L . J . (1983a). Similarities and differences between Skinner ' s radical behaviorism and Kantor ' s interbehaviorism. Revista Mexicana de Analisis de la Conducta, 9(2), 95-115 . Parrott, L . J . (1983b). Valuable constructs : A review of Tragedy and the event continuum by J. R . Kantor. (Chicago : The Principia Press.) . The Interbehaviorist, 12, 8-10 . Parrott, L . J. (1984) . J . R . Kantor ' s Contributions to psychology and philosophy : A guide to further study. The Behavior Analyst, 7, 169–181 . Parrott, L . J. (1986) . On the difference between verbal and social behaviors . In P. N . Chase & L . J .

96


Parrott (Eds .), Psychological aspects of language (pp . 91–117). Springfield, IL : Charle s Thomas . Pronko, N . H . (1963) . Abnormal psychology. Baltimore: Williams and Wilkins. Ray, R . D., & Delprato, D . J . (1989) . Behavioral systems analysis : Methodological strategiew an d tactics . Behavioral Science, 34, 81–127 . Schoenfeld, W. N. (1969) . J. R . Kantor's Objective psychology of grammar and Psychology an d logic: A retrospective appreciation . The Journal of the Experimental Analysis of Behavior, 12 , 329–347 . Skinner, B . F (1938). The behavior of organisms (7th ed.) . New York : Appleton-Century-Crofts . Skinner, B . F (1953). Science and human behavior. New York : The Free Press . Skinner, B . F (1971). Beyond Freedom and Dignity. New York : Bantam. Skinner, B . F (1974). About behaviorism . New York : Knopf. Skinner, B . F (1979). The shaping of a behaviorist. New York : Alfred A. Knopf. Smith, N. W. (1973) . Interbehavioral psychology : Roots and branches. The Psychological Record, 23, 153–167 . Smith, N . W. (1976) . The works of J . R . Kantor: Pioneer in scientific psychology . Revista Mexicana de Analisis de la Conducta, 2(2), 137–148 . Smith, N . W. (1981). Manuscript #2198 . JSAS Catalogue of Selected Documents in Psychology, 11 , 13 . Smith, N. W. (1994) . Phenomenological psychology. The Interbehaviorist, 22(3), 5–13 . Smith, N. W. (1996a) . The evolution of J. R . Kantor's interbehavioral field psychology and som e reasons for its neglect. Presented at a workshop J. R. Kantor's Interbehavioral Psychology: Beyond Mechanism and Mentalism, Cheiron Society, Earlham College, Richmond, Indiana, June 27–30, 1996. Smith, N. W. (1996b) . Interbehavioral research . Unpublished manuscript . Smith, N . W., & Ray, C . E . (1981) . A citation study of the interbehavioral field psychology of J . R. Kantor. Revista Mexicana de Analisis de la Conducta, 7, 117–134 . Smith, N . W., & Smith, L. L . (1996) . Field theory in science : Its role as a necessary and sufficient condition in psychology. The Psychological Record, 46(1), 3–21 . Stephenson, W. (1953) . The study of behavior, Q-technique and its methodology . Chicago : The University of Chicago Press . Verplanck, W. S . (1983). Preface in N. W. Smith, P. T. Mountjoy, & D. H . Ruben (Eds .), Reassessment in psychology: The interbehavioral alternative, (pp . xi –xxv) . Washington D . C . : University Press of America, Inc . Verplanck, W. S . (1995) . Some reflections on Kantor, Kantorians, and Kantor's career. The Interbehaviorist, 23(1), 6–12 . Watson, J . B . (1914) . Behavior. An introduction to comparative psychology . New York : Henry Holt & Co . Zimmerman, D . W. (1979) . Quantum theory and interbehavioral theory . The Psychological Record, 29, 473-485 .

EDWARD C . TOLMAN ' S PURPOSIVE BEHAVIORIS M

NANCY K . INNI S University of Western Ontari o

I . Purposive Behavioris m A. A New Formula for Behavioris m B. Purposive Behavior in Animals and Men II . Operational Behavioris m III . Performance Vectors A. The Schematic Sowbu g B. Failure of the Schematic Sowbu g IV. From Rats and Sowbugs to Ma n V. A Field Theory of Learning A. "Cognitive Maps in Rats and Men " B. Tolman's Map Analogy VI. The Psychology of Social Learnin g VII. Edward Tolman : Purposive Behaviorist Reference s Edward Chace Tolman introduced his purposive behaviorism in the early 1920s and, although he continued to refine his concepts over the next thre e decades, the essential features of his theory remained the same . Tolman wa s born in West Newton, Massachusetts, on April 14, 1886 . He was the second son of James P. Tolman, a successful cordage manufacturer, and the former Mar y


97


98

INNI S

Chace Cheney, who came from a long line of Quaker reformers . His brother , Richard C . Tolman (1881-1948), both rival and role model for Edward, becam e an eminent physical chemist at the California Institute of Technology . Following family tradition, both Tolman boys attended the Massachusetts Institute of Technology . Edward received his B .S . in electrochemistry from MIT in 1911 . Wanting to follow his brother in pursuing a career in science, but also having stron g humanitarian interests, Edward was attracted to the new field of psychology . In the summer of 1911, he took introductory courses in philosophy and psychology at Harvard, and confirmed that psychology was, indeed, his field . He entered the graduate program at Harvard that fall, completing his M .A. in 1912 and hi s Ph .D . in 1915 (Tolman, 1952) . His doctoral dissertation, "Studies in Memory, " was supervised by Hugo Munsterberg, director of the Harvard Psychologica l Laboratory. In 1915, Edward Tolman married Kathleen Drew ; they had thre e children. His first appointment was as an instructor at Northwestern University (1915-1918) . He then moved to the University of California, where he remained on the faculty until he was fired in 1950 for refusing to sign an anticommunist loyalty oath . A group of nonsigners led by Tolman, who opposed the oat h because it was an infringement on academic freedom, took the university regent s to court, and the nonsigners were eventually reinstated (Innis, 1992b) . Edward Tolman died in Berkeley, California, on November 17, 1959 . Tolman was introduced to behaviorism, as it was then being promoted b y John B . Watson (1914), in a course he audited with Robert M . Yerkes, one of hi s Harvard professors . 2 Tolman liked the fact that this system was objective, an d thus scientific . However, he could never accept Watsonian behaviorism wholeheartedly. Tolman wanted a nonphysiological approach that considered behavior as more than simply reflex reactions to stimuli . Other teachers at Harvard als o provided ideas that contributed to the approach that Tolman would eventuall y develop . The most important of these teachers were Edwin B . Holt and Ralph Barton Perry, both of whom were part of a philosophical movement known a s New Realism (Holt et al ., 1912) . "Around this time, Perry and Holt began extending the neorealist program to psychology by formulating a behavioris m which gave the mentalistic concepts of purpose and cognition an objective statu s in the natural world" (Smith, 1982, p . 154) . For example, Holt (1914, 1915) "needed only behavior to define mind but asserted that this behavior could onl y be described in the language of will and purpose" (Kuklick, 1977, p . 418) . Boring (1950) pointed out the obvious similarities between Holt's ideas and those o f Tolman, as have others (e .g ., see Kitchener, 1977 ; Kuklick, 1977 ; Smith, 1982, 1986, for Tolman's relationship with the neorealists), but there is no direct evidence that Tolman ever regarded Holt as a mentor . 3 Tolman did, however, indi 2 E . C. Tolman to R . M . Yerkes, March 16, 1932, Yerkes Papers, Yale University Library ; see als o Tolman (1952) . 3 Boring and Boring's (1948) representation of Tolman with two mentors at Harvard—Munsterberg and Holt—may have been Boring's own idea. Kuklick's (1977) discussion of Tolman ' s ideas i s based primarily on Boring's (1950) account of Tolman's psychology.

'

4 EDWARD C . TOLMAN S PURPOSIVE BEHAVIORISM

99

cate that Perry's ideas (e .g ., Perry, 1918, 1921) "tremendously influenced " him (Tolman, 1959, p . 94) and "laid the basis for [his] later interest in motivatio n and, indeed, gave [him] the main concepts" for his theoretical system (Tolman , 1952, p . 325) . Initially, in Tolman's purposive behaviorism, behavior implied a performance , the achievement of an altered relationship between the organism and its environment ; behavior was functional and pragmatic ; behavior involved motivation an d cognition; behavior revealed purpose . But for Tolman, in contrast to theorist s such as William McDougall (1912, 1926), who also advanced a purposive behaviorism, purpose was not a mentalistic concept it was part and parcel of th e behavior being observed, an obvious legacy from the new realists (Smith, 1982) . As Tolman's own theoretical ideas evolved, his definition of purpose change d somewhat, and it became more of an inferred construct reflecting underlyin g motivational and cognitive processes . Always, however, Tolman viewed purpos e as an objectively defined construct with no teleological implications . By the mid-1930s, he had renamed his system operational behaviorism, and more explicitly denoted how he defined his terms . In this chapter, I trace the development of Tolman's theoretical system, examining the aspects of his theory tha t have continued to remain viable and describing some of his ideas that did not .

I . PURPOSIVE BEHAVIORIS M In 1922, Edward Tolman introduced a "new formula for behaviorism," proposing a nonphysiological behaviorism that could treat the problems of motive, purpose, and determining tendency scientifically . Over the next 10 years , he published a number of Psychological Review articles in which he attempte d to define objectively these and other psychological constructs that behaviorist s such as John Watson had repudiated as subjective and mentalistic . In 1932, th e ideas developed in these articles were presented as a system of psychology in th e book Purposive Behavior in Animals and Men, which has continued to be th e primary source for Tolman's ideas . A. A NEW FORMULA FOR BEHAVIORIS M

Shortly after he arrived at the University of California, Tolman introduced a laboratory course in comparative psychology and, with his students, began t o study the behavior of rats in mazes . The fact that he was working with animal s facilitated Tolman's acceptance of behaviorism as his theoretical framework . Here was an approach that could be applied to rats as well as to human beings . However, Tolman could not accept the idea that "a mere `muscle-twitchism ' such as Watson's" was a "really useful behaviorism" (Tolman, 1952, p . 330) . For Tolman behavior was molar, involving an adjustment of the organism with respect to its environment, not a molecular, physiological response .

100

INNI S

In an early paper, "A New Formula for Behaviorism," Tolman (1922) introduced four concepts : stimulating agency, behavior-cue, behavior-object, and behavior-act. The first thing to note here is Tolman's use of hyphenated terms t o represent his concepts .' In his later writings, and particularly in Purposive Behavior, these hyphenated neologisms would proliferate, and there is some evidence to suggest that they were irritating and perhaps discouraged people fro m reading, and thus coming to accept, the ideas that they represented .' Behavior, o f course, is a function of the stimulus environment, and in these terms we hav e Tolman's attempt to deal with this stimulus–response (S–R) relationship . By attaching behavior to cue and object he hoped to emphasize the interdependenc y of stimulus and response. The term behavior-cue was meant to objectify the old notion of sense-qualities, or sensations produced by stimuli (stimulatin g agency), the domain of the early psychophysicists . It is only by observing differential behavior toward a series of stimuli that we can say something about thei r effects on the organism we can never directly know what any other organis m subjectively experiences, even if that other organism is another human being . With the behavior-object concept, Tolman's aim was to objectively encompas s the idea of how stimuli are perceived that is, the meaning they have for the organism. This meaning was defined by the behavior-act, the final observed response . In other words, in the presence of particular stimuli, behavior is directe d toward (or away from) the behavior-object (goal) . This "relation of behavior-ac t to behavior-object" was "the extremely important problem of motive" in that i t involved the causal factors controlling behavior, such as "desire, emotion, instinct, habit, determining set ." And this was "the problem which the older analytical formulation [Structuralism] tended to obscure and make almost impossible " (Tolman, 1922, pp . 51–52, emphasis in original) . That Tolman, at this early stage in his career, was determined to develop a very general and all-encompassing theory of psychology can be seen from th e concluding statement in this article . Here he claimed that the value of the new formation [sic] will be in part theoretical, in that it will bring under a single rubric all the apparently different and contradictory methods of actual psychology; but in part, also, practical, in that it will allow for a more ready and accurate treatment o f the problems of motive, purpose, determining tendency, and the like . (1922, p . 53) .

Tolman's task, then, was to show how these motivational and cognitive concepts could be dealt with objectively and thus be part of a behavioral, rather tha n a mentalistic, system . In the Psychological Review articles, he provided objective definitions for instinct (Tolman, 1920), emotions (Tolman, 1923), ideas (Tolman, 1926), and consciousness (Tolman, 1927), as well as for purpose and cognition (Tolman, 1925, 1928) . Then, in Purposive Behavior, Tolman presented data, mainly from studies of rats in mazes carried out in his own laboratory, t o support the theoretical ideas initially developed in these articles . 4 Note that in reprinting this article (Tolman, 1951) these hyphens have been removed. 5 Young (1933), R: M . Elliott to E. G . Boring, February 2, 1931 ; E . G . Boring to R . M. Elliott, January 12, 1932, Boring Papers, Harvard University Archives, Pusey Library, Harvard University .

4 EDWARD C . TOLMAN ' S PURPOSIVE BEHAVIORISM

10 1

B. PURPOSIVE BEHAVIOR IN ANIMALS AND MEN

Tolman began his book by carefully establishing the need for a psychology i n which "mental processes are most usefully to be conceived as but dynamic aspects, or determinants, of behavior . . . functional variables which intermediate in the causal equation between environmental stimuli and initiating physiological states or excitements, on the one side, and final overt behavior, on th e other" (1932, p . 2) . This causal equation, of course, is what Tolman (1932 ) sought to explain. At the core of his system, then, were the immanent purposiv e and cognitive determinants of behavior that had to be objectively and functionally defined . Any behavior-act has "the character of getting-to or getting-from a specific goal-object or goal-situation" (p . 10), and it persists until the object o r situation is achieved . This "persistence until" character of the action, which ca n be directly observed, defines it as purposive . The selection of a particular route , or means, of reaching (or getting away from) the goal can also be directly observed, as can the disruption of this behavior if the situation is changed . In thes e observations, then, we have an objective measure of the animal's cognitions, th e meaning that the behavior-object has for it . Cognitions are expectations concerning the relation between signs, environmental support stimuli (the earlier behavior-cues), and significates, the goals (behavior-objects) that they indicate . Two types of expectations were identified : those that were long term, depending on genetics or past experience (memory) , called means-end-readinesses, and those specific to the current, concrete situation, sign-gestalt-expectations . The distinction, now popular with comparativ e cognition researchers, between what is called reference and working memor y (e.g ., Honig, 1978) seems to embody the difference that Tolman was trying t o identify. However, at the time, this distinction was generally ignored by those interpreting the theory . Tolman (1952) also felt that the sign-Gestalt-expectatio n concept, itself, had been misinterpreted, particularly in Spence's (1942) accoun t of the theory. For Tolman (1952), "a sign-gestalt-expectation [was] not to b e conceived as just an S-S association" the relationship was meant to impl y much more . It was a Gestalt, in which "the smallest unit of experience is no t just a mere sensory-perceptual pattern but such a pattern suffused with instrumental meaning" (Tolman, 1952, p . 332) . Cues in the animal's environment pro vide information about expected goals and, consequently, the means of achieving them . To complete his system, Tolman introduced two other behavior determinants , capacities and behavior-adjustments . Tolman listed a number of capacities tha t animals possess, either as a result of heredity or past experience, includin g manipulanda capacity, which determines the animal's ability to recognize th e properties of objects that make motor manipulations possible (p . 448) ; discriminanda capacity, which refers to the ability of the animal to use the properties o f objects (the behavior-cues of his earlier formulation) to make sensory differentiations (p . 441) ; means-end-capacities, which involve the ability of the anima l to deal with or expect means-end-relations (p. 449) ; retentivity (memory) ;

102

INNI S

consciousness-ability; ideation-ability; and creativity (p . 439) . Differences i n these capacities, of course, denote individual and species differences . In fact , Tolman had an early interest in innate individual differences, and his first publication involving animal research concerned the question of whether the ability t o learn in rats is inherited (Tolman, 1924) . This project was soon taken over by hi s student, and later colleague, Robert Tryon, who became a leader in the field o f behavior genetics (Innis, 1992c) . Tryon persisted in encouraging Tolman not to ignore individual differences, and Tolman continued to include these variables i n his system (e .g ., Tolman, 1932, chap . 24 ; 1938) . To account for all the aspects of psychology dealt with by the introspectionists, Tolman had to objectively define ideation and consciousness. He did s o with the notion of behavior-adjustment, the last of his behavior-determinants . Behavior-adjustments were "surrogates" for "sampling" or actual "runningback-and-forth" responses (Tolman, 1932, p . 438), either overt (conscious awareness ; that is, attention) or covert (ideas), which served to enhance particular aspects of the environment . In an earlier paper, Tolman gave this graphic ac count of a behavior-adjustment, using as an example the behavior of a rat hesitating before choosing an arm of a T maze . "The rat stops and wiggles his nos e from side to side, and then finally chooses . And . . . he usually chooses th e correct one, more often than when he does not hesitate" (Tolman, 1926, p. 367) . Later, these responses were labeled vicarious-trial-and-error behavior (Muenzinger, 1938) and became an important feature of Tolman's only attempt t o quantify his theory . Much of Purposive Behavior is devoted to presenting empirical support fo r Tolman's position by showing, through innumerable studies, many of which ar e now textbook classics, that even an animal as lowly as the rat can display purposive action with respect to a goal . For example, experiments carried out by M . H . Elliott, one of his graduate students, demonstrated that rats learn about—expect—specific properties of goal objects (e .g ., bran versus sunflower seeds) . This expectation affects both rate of learning and level of performance onc e learning has occurred. Equally motivated rats learned more quickly with preferred foods, and their behavior was disrupted when the goal-object wa s changed that is, when their expectation was disconfirmed (M . H . Elliott, 1928 ; Tolman, 1932) . One of the most important findings to support Tolman's cognitive learning theory involved what he called latent learning . Studies by his students (Blodgett, 1929 ; Tolman & Honzik, 1930) showed that hungry rats learned about the spatial properties of a maze even when no food was received in th e goal box . If food was presented after several days of training under these no food conditions, the rats performed as well as rats that had been receiving foo d from the outset of the study. This finding that reward (reinforcement) acts o n performance, not learning, sparked a heated controversy with the Hullian learning theorists who, like Thorndike before them, emphasized the necessity of reinforcement in the learning process . Throughout the 1930s and 1940s, a healthy rivalry existed between researchers associated with the positions held by Hull


10 3

(e .g ., 1937, 1943 ; see Rashotte and Amsel, Chapter 5) and Tolman, stimulatin g research and debate on a number of other issues, as well as latent learning . Tolman (1932) concluded Purposive Behavior by summarizing the important constructs of his system . There are four causes of behavior : stimuli, heredity, training, and physiological state. Intervening between these causal factors an d the observed behavior are behavior-determinants, which include capacities de pending on genetics and past experience ; capacities relating to particular situations or motivational states ; the "immanent purposive and cognitive determinants, which lie in the very warp and woof of the S ---> R sequence" ; and the "back-and-forth-behavior-adjustments " (p . 412) . Tolman applied his system to all of psychology. But although he believed that he had presented a useful an d valid approach, Tolman included a final caveat, pointing to the ephemeral qualit y of a system : "each system is so obviously bound to be wrong ." He went on to caution, "may neither you nor we ever seek to hold up these propositions, sav e in a somewhat amused, a somewhat skeptical, and a wholly adventure-seekin g and pragmatic behavior-attitude" (p . 394) . Tolman's approach always remaine d pragmatic, and he continued to maintain that "doing science" should involv e having fun (Tolman, 1959) . Although Purposive Behavior was not widely reviewed, in general, it was received positively . Although deploring the neologisms, P. T. Young (1933), who reviewed it for the American Journal of Psychology, indicated that it wa s "delightfully informal and on the level . . . distinctly a contribution to psychology and . . . well worth the careful study of serious students" (p . 178) . There was also a brief summary in England's New Statesman and Nation, remarking on Tolman's "Pickwickian" use of language .' Colleagues of Tolman, includin g E . G. Boring, were very positive with respect to the scholarship of the text, whil e maintaining that it was difficult . ' In an introduction to Purposive Behavior in Animals and Men, R . M . (Mike) Elliott, editor of the Century Psychology Series in which it was published, provided an astute assessment of Tolman's system :

Professor Tolman's argument may be ignored in some quarters; it will certainly be amplified by himself and others as new and crucial research data come to light ; it will not, I think, be radically revised, that is "disproved" ; and it will never be discredited, that is , shown to be fictitious or unnecessary. Behaviorism of this sort has come of age . (1932 , p . viii-ix .)

II . OPERATIONAL BEHAVIORIS M Edward Tolman was soon amplifying his arguments . Tolman's system now offered a method for objectively identifying the determining tendencies underlying an organism's behavior the delineation of operationally defined interven 6 New Statesman and Nation,

Saturday, May 14, 1932. 7 E. G . Boring to R . M . Elliott, January 12, 1932.

104

INNI S

ing variables . In Purposive Behavior he had considered his purposive and cognitive determining tendencies as immanent determinants that is, as intrinsic t o the function (fi) relating independent (environmental stimuli and physiologica l state) and dependent (behavior) variables . However, because of the complexit y of all but the simplest reflexive reactions, Tolman realized that for psychologist s to adequately explain behavior that is, to identify the determining tendencie s producing behavior it was necessary to break down the independent–dependent variable relationship into subordinate functions . These constructs, then , must be functionally related to both the independent (f2) and the dependent (A) variables (Tolman, 1935, 1936) . Intervening variables, as Tolman labeled these constructs, represent the motivational and cognitive processes outlined in Purposive Behavior, "stated and tested in terms of concrete repeatable operations " (Tolman, 1936, p . 89) . Tolman (1936) called his revised approach "operationa l behaviorism" for two reasons : (1) it took the same "positivistic attitude," labele d "operational" by Bridgman (1927 ; see also Stevens, 1935), that physicists an d philosophers of the time were adopting to determine the operations needed t o define scientific constructs ; and (2) the behavior the researcher was observing involved an organism acting on operating on its environment . In 1937, as president of the American Psychological Association, Tolman took the occasio n of his presidential address to present to his colleagues the most recent version o f his system of psychology (Innis, 1992a) . Tolman (1938) used a series of diagrams, similar to the one shown in Figure 4 .1, to illustrate his system of psychology and show how his constructs wer e conceived to function in determining behavior . In the example shown in the figure, the dependent variable, choice of the left arm (Bleft) of a T maze (depicted on the right), was for Tolman, as for all behaviorists, a function, fi, of independent variables (depicted on the left) . For example, psychologists, beginning with Thorndike, had been interested in the way behavior changes across trials, the traditional learning curve . Tolman showed that the effect of choices and their out comes on previous trials, represented by the 1(OBO) variable in Figure 4 .1, i s but one of a number of independent variables acting in concert to control the choice observed on a given trial . Tolman identified two types of independent variables, individual difference and environmental. The individual difference variables are indicated by th e acronym H .A .T.E ., which refers to Heredity, Age, previous Training, and Endocrine, drug, or vitamin conditions (Tolman liked to joke that he hated to stud y individual differences) . These variables have a controlling effect on each of th e environmental variables . Along with 1(OBO), Tolman included variables representing the maintenance schedule (i .e ., conditions of deprivation) ; the relevance of the goal object, given the conditions of deprivation ; the kinds of stimuli avail able to the rat in the maze ; the type of motor response required of the animal ; and the configuration, or pattern, of units in the maze . To identify the causal factors controlling behavior, Tolman emphasized that simply considering the relationship between the independent and dependent

10 5


INDEPENDENT VARIABLES (W) IV

f

DEPENDENT VARIABLES (DV)

f,

f

INTERVENING VARIABLES

DV

H .A.T .E . Maintenance Schedule

DEMAND

N

H .A.T.E. Goal Object

APPETITE

H.A.T.E. Stimuli

DIP I. ERENTIATION

H.A.T .E. Motor Response

Bleft // //

MOTOR SKILL

H.A.T.E . E(OBO)

/ HYPOTHESES / /

H .A.T.E . Pattern of Maze Units

/ BIASES /

FIGURE 4 . 1 Diagram to illustrate the functional relationships among the Independent, Intervening, and Dependent variables in Tolman ' s system of psychology, modelled on the diagrams that Tolman presented in his 1937 presidential address to the American Psychological Association (se e Tolman, 1938) .

variables was not enough . Rather, this function must be broken down and considered in terms of the two functions relating independent variables (f2) and dependent variables (A) to intervening variables . For Tolman a theory was no mor e than a set of intervening variables (Tolman, 1938, p . 9) . The task, then, was t o operationally define these intervening variables by means of standard experiments in which one independent variable, correlated with the construct bein g studied, is systematically varied while all others are held constant . The diagra m in Figure 4 .1 lists six intervening variables, each related to a particular independent variable . The intervening variables listed by Tolman and he was quick t o point out that this was an oversimplification of what is really going on wer e demand, appetite, differentiation, motor skill, hypotheses, and biases . Definin g experiments for many values of these variables had already been reported in the literature, with a number of the studies having been conducted by Tolman an d his students . For example, defining experiments with respect to demand an d

106

INNI S

appetite included M . H. Elliott's (1928, 1929) studies with different goal-object s and levels of deprivation . Tolman wanted to identify the rules that would permit prediction of the behavioral outcome for all independent variables, and so understanding the f3 functional relationship was of most interest . What the nature of this function was , Tolman did not know ; but he was quite confident that the "ways of combinatio n of the intervening variables [did] not seem those of scalar addition" (Tolman , 1938, p . 21) . Tolman saw a more promising approach to finding the "f3 rules or equations" (p . 24) in Kurt Lewin's topological psychology (1935, 1936), and h e soon introduced a model that incorporated these ideas . III . PERFORMANCE VECTOR S The diagram in Figure 4 .1 shows the functional relations that must be define d in order to explain choice-point performance, which Tolman referred to a s achievement behavior. He also identified another set of responses, called catalyzing behavior, observed just before the actual choice was made . This behavior, he suggested, functioned to "enhance the values of one or more of the independen t variables" (1938, p . 33) . Tolman adopted Muenzinger's (1938) term vicariou s trial and error (VTE) to refer to this hypothetical "running-back-and-forth" o r "sampling response" that he had identified many years earlier (Tolman, 1926) . In a grant proposal to the University of California Research Board in 1937, Tolma n suggested that this catalyzing VTE behavior might be an "objective correlate o f conscious awareness" (attention) . 8 For the next 5 or 6 years, Tolman and his students devoted most of their time to research on vicarious trial and error behavior . Much of this work was also aimed at providing support for a model with whic h Tolman (1939) hoped to quantify his theory . This model represented the f3 function relating the intervening variables to the catalyzing and achievement behaviors (see Fig . 4 .1) . Along with a graduate student, Clark Crannell, Tolman devised the representation of the model shown in Figure 4 .2 . A . THE SCHEMATIC SOWBUG The model seems to have been introduced, as were many of Tolman's theoretical ideas, in lectures to his undergraduate motivation class . In March 1938, h e presented the class with "a new idea . . . . I don't know whether it works or not ; probably it doesn't. But if it does it would be a ringer." 9 The idea came while h e was trying to show the students how Kurt Lewin's field theory of motivation , 8 Psychology Research Grant 220 (1937—38), University of California Archives, Bancroft Library, University of California—Berkeley . 9 Mimeographed transcript of Tolman's lecture notes, March 28, 1938, p. 145 . Unpublished manuscript, University of California—Berkeley . Tolman Papers, Archives of the History of American Psychology, Akron, Ohio .


10 7

FIGURE 4 .2 Tolman's schematic sowbug . Drawn from a photograph of a representation of the sowbug made by Berkeley graduate student, Clarke W . Crannell . Photograph courtesy of the Department of Psychology, University of California, Berkeley .

which influenced much of his own thinking, could be applied to an animal' s choice behavior . Lewin (1935, 1936) used an oval to represent an individual' s psychological "Life-Space," in which a field of forces, represented by vectors , produces valenced action (a directed response) . Tolman decided to animate Lewin's representation by adding motor appendages indicated by orientatio n vectors, on each side, and progression vectors, behind the oval (see Figure 4 .2) . The organism had a receptor organ, a sensory surface that, in one (unpublished ) account of the model, Tolman suggested could be wide or narrow ; if wide the organism was "smarter," that is, could deal with more stimuli . This sensory surfac e seems to be a very early version of the cognitive map, perhaps the most well known of Tolman's constructs . Cognitive maps could be broad or narrow, with broad maps providing the animal with more flexibility in responding (Tolman , 1948) . By adding legs and a sensory organ to the oval representation, Tolma n produced what looked to one student like a common sowbug, hence the rathe r uncomplimentary label, "schematic sowbug," that became attached to the model . The quantitative rules of action represented by the schematic sowbug were part of a molar, field theory of learning . As in Lewin's theory, a topological representation and vector mathematics were used to show and predict the

108

INNI S

strength of orientation (VTEs) and progression (choice of and movement toward a stimulus) . The sowbug was bilaterally symmetrical, with "the underlyin g mechanism . . . analogous to the mechanism assumed by Loeb [1918] in hi s doctrine of `forced movements' in lower organisms ." (Tolman, 1941, p . 369) . For the model, Tolman (1939) assumed that, "figuratively speaking," "all highe r organisms are in a certain underlying sense tropistic" (p . 323, emphasis is i n original) . The long vectors, on the side opposite to the single stimulus source shown in Figure 4 .2, indicate that the sowbug is pushed in the direction of tha t source . The lengths of the vectors designating orientation and progression wer e defined by the area under the orientation and progression distributions shown within the organism. The progression distribution determines the animal's movement ; the orientation distribution, its attention . The shape of the orientation distribution represents the relative intensity of stimuli (and these could be from any dimension) impinging on the sowbug, while its height is determined by orientation need and tensions (psychological energy), concepts directly related to thos e in Lewin's theory of motivation . The height of the progression distribution is determined by a "specific hypothesis" as to the value of a particular stimulus source that is established as a result of past experience . Thus the schematic sow bug was a mechanistic model, "a physico-mathematical formula . . . by means of which the effects of . . . independent variables . . . are (in a given discrimination set-up) to be theoretically combined so as to predict the actual resultant dependent variables" (Tolman, 1941, p . 369, emphasis in original). For Tolman, the model represented the f3 function in his theory, and as such was an attempt to show how external stimuli (through their associated intervening variables) combine to produce performance . Of major interest, of course, was the situation in which two stimuli were pre sent simultaneously. Could the model predict the rat's choice-point performance? An additional assumption was introduced stating that during the perio d of direct stimulation by one of the stimuli "the orientation need" for that stimulus became "temporarily weaker" and the other recovered (Tolman, 1939, p . 329) . The behavior of the model then predicted orientation—attention—to firs t one and then the other stimulus, or VTEing . Are there more VTEs when the discrimination is easy or difficult? Intuitively one would think that more VTE s would occur with a difficult discrimination . However, using the sowbug model , Tolman (1939) accurately predicted (before seeing the data, as he was quick to point out) Muenzinger's (1938) finding of more VTEs with easy discriminations . Other studies (especially with humans), however, showed more VTEs when th e discrimination was difficult . Tolman (1941) was able to show that this differenc e depended on the type of problem facing the animal (and the model) . When the animal was learning what to do (in modern terminology, establishing a referenc e memory), more VTEs occurred on easy discriminations . However when the task required was known (e .g ., choosing on the basis of brightness), then VTEing occurred more often when the discrimination was difficult (black versus dar k gray) . This model, with its separate orientation and progression vectors, also has

'


10 9

some similarity to the theory of attention in discrimination learning later pro posed by Sutherland and Mackintosh (1971) in which animals first learn abou t stimulus properties (selective attention) and then acquire the appropriate response . B . FAILURE OF THE SCHEMATIC SOWBU G

Tolman ' s purposive behaviorism was, and still is, criticized for being vagu e and nonquantitative . The schematic sowbug was an attempt to devise an explici t set of rules —a mechanistic model for a field theory of animal learning . I t never caught on . Why was this model ignored? It was not because Tolman eventually rejected it ; he included the first sowbug paper among the articles that "t o him meant steps in the development of his theoretical system" (Tolman, 1951a , p . v) . Moreover, in the final account of his system, Tolman (1959) took severa l pages to discuss ideas related to those first presented in the sowbug model, usin g somewhat different terminology. He introduced the idea of a perceptual discrimination space, to refer to what on the sowbug he called the sensory surface, an d identification and pragmatic vectors to replace the orientation and progressio n vectors of the sowbug . The use of vector mathematics to predict performanc e was one of the most important features of the sowbug model, and Tolman continued to believe that "some such vector interpretation, even though it may hav e to be considerably modified, is intriguing and has empirical suggestiveness " (1959, p . 141) . The sowbug model may simply have been ahead of its time . People in the early 1940s were not ready to accept the type of mechanical model — a cybernetic model represented by the sowbug, and so rejected or ignored it . In 1979, Donald Campbell wrote that "many of us still expect out of cybernetics , servosystem engineering, automaton theory, and artificial intelligence to com e the next comprehensive behavior theory . . . . Tolman's theory, both purposive and behavioristic, is the theory of the 1930s closest to such a development" (p . 187) .

IV . FROM RATS AND SOWBUGS TO MA N From the early 1940s on there is a trend in Tolman's writing away from a strictly theoretical psychology to one that emphasizes the role of the psychologist in meliorating the human condition . As a pacifist, Tolman was particularl y disheartened by World War II, and he was determined to use psychology to pre vent another war. To this end, he published a slim volume, Drives toward War, which was an odd combination of motivation textbook and prescription fo r peace . The first words of the text encompass his topic : "War is horror, cruelty, death ; but it is also selflessness, devotion, exaltation" (Tolman, 1942, p . vii) . The inherent conflict in human nature is at once identified . This was a book abou t motivation about the dynamics of human behavior.

1'1O

INNI S

In the text Tolman (1942) examined the inherent biological drives (the appetites and aversions) and the social drives of animals and men . He was particularly concerned with the role of the psychological dynamisms, essentially th e Freudian dynamisms fixation, repression, displacement, projection . These dynamisms (and the resulting behavior) tend to be blindly persistent, in contrast t o learned beliefs or expectations about reaching a goal, which can be modified if the environment changes . Tolman stressed that a proper channeling of psycho logical (that is, motivational) energy is essential to human welfare, as well a s personal happiness . He developed a "drive-conversion scheme" that woul d "combine the emphasis upon biological drives with a moderate amount of self assertion and a large amount of collectivity" (p . 100) . Once Tolman had organized his ideas on the role of the dynamic mechanisms in social behavior, hi s long-standing commitment to social action meant that he must draw attention t o the implications for the future of mankind . Thus, while Drives toward War is a textbook on motivation, it is also a book with one major underlying theme how to change society so that war between nations is no longer inevitable ; s o that man, even if he cannot control, can at least harness his aggressive tendencies . It was a message of hope ; there is, even with our given needs and dynamisms, the possibility for lasting world peace . Tolman's prescription for peace involved creating a world where psychological energy was constructively channeled . His utopia would be somewhat socialistic so that no one experienced too great frustrations of the biological drives ; it would be a society in which children easily identify with parents ; and above all it would involve loyalty to a "supernational" group, a worldwide political unio n with which all men could identify. This "World-Federation" would have all the trappings of a state (flag, anthem, etc .) to facilitate loyalty. Since aggression is a basic drive, this state must have a "common enemy" as a target for the aggression of the group . Perhaps nature itself would be this target, as William Jame s (1910) had suggested more than a generation before. Although he abhorred war, Tolman also despised the atrocities that were occurring in Nazi Germany, and he offered his services to his government, spending several months during the war in Washington with the Assessment Program of the Office of Strategic Services (OSS) helping to select individuals who had the potential to be successful agents for the American government. As a result , Tolman's interest in the area of personality and social psychology was stimulated (Tolman, 1952) . This change in interest is evident in several papers published in the late 1940s and early 1950s, in which he applied his theoretical idea s directly to human behavior.

V. A FIELD THEORY OF LEARNIN G

In 1947, Edward Tolman delivered the 34th Annual Faculty Research Lectur e at the University of California. The occasion, celebrating his long and distin-


11 1

guished career in the psychology department at Berkeley, provided Tolma n (1948) with an opportunity to summarize the research that he and his student s had carried out over more than three decades, and to differentiate his own system, that he now called a field theory of learning, from the generally more popular S –R approach . "Cognitive Maps in Rats and Men," the title of Tolman's talk, reflects the key elements of his psychology at this relatively late stage of his career : a spatial metaphor for memory representation ; the rat as a prototypical re search subject; and mankind as the true and proper subject of his concern . Tolman concluded his faculty lecture by incorporating some of the ideas on th e dynamics of human behavior, first developed in Drives toward War, into the cognitive map model that was now explicitly central to his general learning theory . A . COGNITIVE MAPS IN RATS AND ME N

Tolman (1948) began his lecture by pointing out how his theory of animal learning differed from the popular S –R theories of what he called the "telephone-switchboard school ." For him, a different metaphor was more descriptive of what was really going on . He suggested that "in the course of learning some thing like a field map of the environment gets established in the rat's brain," an d that the rat's "nervous system is surprisingly selective as to which . . . stimuli it will let in at any given time" . The analogy, then, is of a "central office" that i s "far more like a map control room than it is like an old-fashioned telephone ex change" (p . 192) . Besides introducing a compelling analogy, Tolman also for th e first time gives a physiological referent for one of his constructs . Perhaps thi s was due to the influence of David Krech (Krechevsky), who had recently joined the Berkeley psychology department and was an active participant in the livel y discussions in Tolman's research seminar . Tolman (1948) then summarized a number of the seminal experiments that h e and his students had conducted with rats in mazes, the results of which sup ported his theoretical claims . He outlined studies of latent learning initiated in the late 1920s (e .g ., Tolman & Honzik, 1930), the work on "hypotheses" in rat s carried out in the early 1930s by Krechevsky (e .g ., Krechevsky, 1932a, 1932b) , and the research on the role of vicarious trial and error in visual discrimination , discussed previously, that had been the focus of efforts in his laboratory in th e early 1940s . He also described an unpublished study by a recent Ph .D . student, Bradford Hudson, in which rats that had received shock near a stimulus paired with food responded by avoiding this stimulus, sometimes even by piling sawdust over it, when they were placed in the same situation again . The important factor seemed to be that the stimulus was present after the shock was received. This study, in particular, showed the selective nature of the rat's behavior in establishing a cognitive map . Finally, Tolman reviewed findings from a series o f spatial orientation studies that had just been completed by Benbow Ritchie and Donald Kalish . These experiments, now still frequently cited in the literature , showed that the rats learned something very general about the location of impor-

112

INNI S

tant stimuli in their environment rather than simply a series of responses . For ex ample, rats were trained on an elevated maze to take a long path to food an d then, on a test trial, were given a choice among many possible paths . The largest proportion of animals took the path that provided the shortest route to the foo d box, rather than the original path on which they were trained (Tolman, Ritchie , & Kalish, 1946) . This and similar studies provided strong support for the ide a that the animals had a general representation of the situation ; they had constructed a cognitive map . The cognitive map model has recently become popular with researchers in a new subdiscipline of psychology, comparative cognition . An ever-growing number of animal-learning researchers have adopted an information-processing approach and now consider themselves to be cognitivists, and, with this change i n focus, a new appreciation of Tolman's ideas has emerged . Citations of "Cognitive Maps in Rats and Men" and Purposive Behavior are more frequent now than they were when Tolman was alive . Tolman also extended the cognitive map idea by suggesting that two types o f maps (1) broad and comprehensive or (2) narrowly fixed could be established . Results from experiments with animals suggested that conditions such as brain damage, overtraining, reduced stimulus availability, or highly motivatin g or frustrating conditions could result in narrow strip maps . In a maze, then, under such conditions, a rat would not respond on the basis of a wide array of environmental stimuli, but rather fixate on a single, particular path . It was at this point that Tolman (1948) turned his attention to human behavio r and "some . . . of the so-called `psychological mechanisms' which the clinica l psychologists and the other students of personality have uncovered as the devil s underlying many of our individual and social maladjustments ." These he believed could be "interpreted as narrowings of our cognitive maps due to to o strong motivations or to too intense frustration" (p . 207) . Later, in an addres s made as the recipient of the Kurt Lewin Memorial Award presented by the Society for the Psychological Study of Social Issues (SPSSI), Tolman again pointe d out the relevance of his approach not only for rats in mazes, but for two types of human social learning that he alluded to briefly at the end of his faculty lecture , namely "(a) the social learning of the clinical patient in the therapeutic session , and (b) the almost equally complicated social learning involved in the mechanism of identification with an ingroup" (Tolman, 1949, p . 18) . B . TOLMAN'S MAP ANALOG Y

There were two important themes in Tolman's "Cognitive Maps" paper. The first, and most enduring, is the map model, developed over years of researc h with rats in mazes . The second is the extension of this model to human socia l learning . We will see how this second theme was developed by Tolman in th e next section . With respect to the cognitive map metaphor, Tolman had alway s felt comfortable with spatial analogies, and they are evident in his writing from

'


11 3

the 1920s on . Smith (1990) presents an eloquent discussion of their importanc e to the theoretical development of Tolman's ideas : "For Tolman, the extended metaphor of mazes and maps was more than a manner of expression it was the very language of his deepest thoughts and beliefs" (p . 249) .

VI . THE PSYCHOLOG Y OF SOCIAL LEARNIN G Tolman used his Lewin Memorial Award lecture to expand on his field theory , now applied equally to humans and to rats, by introducing some new construct s and elaborating on old ones . In this talk honoring Kurt Lewin, Tolman was eager to show his appreciation for the influence Lewin had on his ideas, and he did so by adding "certain new features to the Lewin model" (Tolman, 1949, p . 18) . Tolman saw his Cognitive Map (or Sign-Gestalt-Field) construct to be equivalent to Lewin's (1935) idea of Life Space, but he differed somewhat in the way he rep resented Lewin's Psychological Person, with its inner and outer cores . Tolman depicted an Inner Life Space, called the Personality Structure, and an Outer Life Space, the Cognitive Map. The personality structure, comprising drives, beliefs , values, and perceptual readinesses, corresponded to Lewin's inner core ; the outer core Tolman placed inside the cognitive map and called it the Behaving Self The behaving self in this model was essentially Tolman's "old `schematic sowbug ' dusted off" (Tolman, 1949, p . 9) . In experiencing the world, "that which is learned or acquired by either rat or man is first a cognitive map and, secondly, new or modified drives, beliefs, values and perceptual readinesses" (Tolman, 1949, p . 10) . Let us look first at the personality structure . Drives were represented by compartments denoting specific drives as well as a compartment for a general physiological (motivational ) energy, libido . And a new metaphor was introduced . Each compartment was assigned both positive and negative charges (indicated by plus and minus signs) , "an electromagnetic elaboration" that Tolman found useful "as a way of summarizing the facts of wanting or of positively valuing, on the one hand, and of no t wanting or negatively valuing, on the other" (Tolman, 1949, p . 7) . Positive or negative charges were also associated with goal-objects and represented perceptual readinesses that arose as the result of the value of the object . Drive compartments were connected to goal-objects by channels that represented instrumental beliefs such as " ìf hungry, food is good ' and ìf food is good, food-cans are good' " (Tolman, 1949, p . 8) . The practice of using plus and minus signs to indicate positive and negative goal objects has continued to remain popular in man y learning theories . Its introduction by Tolman was clearly influenced by his ow n background in engineering and the way in which Lewin had represented the valences associated with various regions of the Life Space in his theory . As a result of experience the organism establishes a set of "concrete expectations . . . as to the distances, directions," and other aspects of the stimulus envi-

1 14

INNI S

ronment (Tolman, 1949, p . 8) . "The acquisition of the particular [cognitive] map depends on the actual stimulus configuration presented by the environment, o n the frequency and recency with which this stimulus configuration has been explored, on how emphatic its various features are, and finally on the individual' s perceptual readinesses for such features" (Tolman, 1949, pp . 10-11) . Thus, as a result of experience, certain aspects of the environment have either a positive o r a negative value determining the direction that behavior will take . The behaving self, within the cognitive map, contained "need-pushes . . . derived from the negative charges of the drive tensions" (Tolman, 1949, p . 9) for the various drives (e .g ., for food, sex, and pain avoidance) in the personality structure, and were represented by vectors . Tolman presented his model, and described som e specific applications, using a series of diagrams in which topological force an d electric charges were prominent features . This model was further refined by Tolman (1951b) in four chapters in Talcott Parsons' and Edward Shils's book Toward a General Theory of Social Action . In this presentation, the main change to the model was in the way the intervenin g variables were designated. Tolman now talked about a Need System, a BeliefValue Matrix, and a Behavior Space. In contrast to drive (the physiological condition of the organism), need was used to indicate "a postulated resultant, intervening behavioral process to be defined in the last analysis as a readiness to ge t to and to manipulate in a consummatory fashion (or to get from) certain othe r types of object" (Tolman, 1951b, p . 288), a long-familiar aspect of Tolman' s theory . In line with his more recent conceptualizations, the need system contained many compartments (reflecting the different needs), each with positiv e and negative "electromagnetic charges" indicating a readiness to approach (o r avoid) certain objects in the environment . The familiar concepts of means-end-readiness and sign-gestalt-expectation , introduced in Purposive Behavior, were now embodied in the belief-value matrix and the behavior space constructs . These new terms emphasized the idea tha t "each single sign-gestalt-expectation is always part of a larger field of expectations and that any simple means-end-readiness—belief-value unit is also part of a larger field or matrix of such units" (Tolman, 1952, p. 332 ; emphasis in original) . The belief-value matrix was "constituted by the cognitive categorizations, beliefs, and values" of the individual relative to, for example, in food seeking behavior, "hunger deprivation, hunger gratification, foods, and restaurants" (Tolman, 1951b, p . 290) . A belief-value matrix contained "universals," whereas the behavior space contained "particulars ." The behavior space was defined as "a particularized complex of perceptions (memories and inferences) a s to objects and relations and the `behaving self,' evoked by the given environmental stimulus situation and by a controlling and activated belief-value matrix (or perhaps several such matrices)" (Tolman, 1951b, p . 296) . In the early 1950s, after much persuasion, Tolman consented to contribute a chapter to Sigmund Koch's series Psychology : A Study of a Science. It would turn out to be his last publication . In his chapter, Tolman (1959) reexamined his


11 5

theory and made very explicit the ways in which he had modified the theoretica l constructs over the previous decade . Perhaps because of the complexity of the new way in which the system was presented (using a series of complex symbols rather than diagrams), it is not often cited . It must be emphasized, however, that the basic ideas of Tolman's original theoretical position had endured . The chap ter, titled "Principles of Purposive Behavior," presented essentially the same behaviorism that Tolman had introduced more than 40 years earlier.

VII . EDWARD TOLMAN : PURPOSIVE BEHAVIORIS T To a "true" behaviorist, "purposive behaviorism" no doubt was (and perhap s to some still is) an oxymoron. Indeed, Tolman's behaviorism was unlike an y other. Tolman believed that behavior could not need not be explained in a way that excludes mentalistic terms . Rather than get rid of them, he wanted t o give them objective, operational definitions . In place of the sterile mathematic s and empty organisms of his competitors, Tolman proposed a rich theoretical structure in which purpose and cognition played well-defined parts as potentiall y measurable intervening variables . For him, actions were infused with meaning ; behavior was goal directed that is, motivated and purposive. However, adopting this view did not mean that it was impossible to develop mechanistic rules t o account for the behavior observed . Tolman's sowbug model was an attempt to d o just that . The internal structure of the sowbug, with its needs and hypothese s (motives and cognitions), was not available to the observer . It could not be directly known, and Tolman did not provide rules for directly measuring thes e variables . But the stimuli impinging on its sensory surface were available, a s were the responses made to them, and these reflected the state of the system , which could then be inferred . Indeed, the sowbug bears some resemblance t o Braitenberg's (1984) vehicles, interesting mechanical models of emotional an d cognitive states . Today, such models are not particularly alien ; in Tolman's day they were alien, and so they were rejected or ignored . When Tolman used the terms purpose and cognition, it was impossible for many to believe that he was a behaviorist, despite his protestations . The schematic sowbug, in its various incarnations, which Tolman was still advocating in the final version of his system , shows that he was .

ACKNOWLEDGMENT S I thank the Social Sciences and Humanities Research Council of Canada for support of my wor k on Tolman . Thanks are also due to Jennifer Higa and John Staddon who made helpful comments o n a previous version of this chapter.

116

INNI S

REFERENCE S Blodgett, H . C . (1929). The effect of the introduction of reward upon the maze performance of rats . University of California Publications in Psychology, 4, 113-134 . Boring, E . G. (1950) . A history of experimental psychology (2d ed.) . New York: Appleton-CenturyCrofts. Boring, E . G ., & Boring, M . D. (1948) . Master and pupils among the American psychologists . American Journal of Psychology, 6, 527–534. Braitenberg, V. (1984) . Vehicles: Experiments in synthetic psychology . Cambridge, MA: MIT Press. Bridgman, P . W. (1927) . The logic of modern physics . New York : Macmillan. Campbell, D . T. (1979) . A tribal model of the social system vehicle carrying scientific knowledge . Knowledge: Creation, Diffusion, Utilization, 1, 181–201 . Elliott, M . H . (1928). The effect of change of reward on the maze performance of rats . University of California Publications in Psychology, 4, 19–30 . Elliott, M . H. (1929) . The effect of appropriateness of reward and complex incentives on maze performance . University of California Publications in Psychology, 4, 91–98 . Elliott, R . M . (1932) . Editor's introduction . In E. C . Tolman, Purposive Behavior in Animals an d Men (pp. vii–ix). New York: Century Co. Holt, E. B . (1914) . The concept of consciousness. New York: Henry Holt Holt, E. B . (1915) . The Freudian wish and its place in ethics . New York: Henry Holt . Holt, E . B ., Marvin, W. T., Montague, W. P., Perry, R . B ., Pitkin, W. B ., & Spalding, E . G. (1912) . The new realism : Cooperative studies in philosophy. New York : Macmillan. Honig, W. K. (1978) . Studies of working memory in the pigeon . In S. H . Hulse, H . Fowler, & W. K. Honig (Eds .), Cognitive processes in animal behavior. Hillsdale, NJ : Lawrence Erlbau m Associates . Hull, C . L . (1937) . Mind, mechanism, and adaptive behavior . Psychological Review, 44, 1–32 . Hull, C . L . (1943) . Principles of behavior. New York : Appleton-Century-Crofts . Innis, N . K . (1992a) . Animal psychology in America as revealed in APA presidential addresses . Journal of Experimental Psychology : Animal Behavior Processes, 18, 3–11 . Innis, N. K . (1992b) . Lessons from the controversy over the loyalty oath at the University o f California . Minerva, 30, 337–365 . Innis, N. K. (1992c) . Tolman and Tryon : Early research on the inheritance of the ability to learn . American Psychologist, 47, 190–197 . James, W. (1910). The moral equivalent of war. New York : American Association for International Conciliation . Kitchener, R . F. (1977) . Behavior and behaviorism. Behaviorism, 5, 11–71 . Krechevsky, I. (1932a). `Hypotheses' in rats . Psychological Review, 6, 516–532. Krechevsky. I . (1932b) . The genesis of "hypotheses" in rats . University of California Publications i n Psychology, 6, 45-64 . Kuklick, B . (1977) . The rise of American philosophy. New Haven, CT: Yale University Press . Lewin, K. (1935) . A dynamic theory of personality. New York : McGraw-Hill . Lewin, K. (1936) . Principles of topological psychology. New York: McGraw-Hill . Loeb, J . (1918). Forced movements, tropisms, and animal conduct . Philadelphia: Lippincott . McDougall, W. (1912) . Psychology: The study of behavior. London: Oxford University Press . McDougall, W. (1926) . Men or robots . In C . Murchison (Ed .), Psychologies of 1925 . Worcester, MA: Clark University Press . Muenzinger, K. F. (1938). Vicarious trial and error at a point of choice : I . A general survey of its relation to learning efficiency. Journal of Genetic Psychology, 53, 75--86. Perry, R . B . (1918) . Docility and purposiveness . Psychological Review, 25, 1–20. Perry, R . B . (1921) . A behavioristic view of purpose . Journal of Philosophy, 18, 85–105 . Smith, L . D. (1982) . Purpose and cognition : The limits of the neorealist influence on Tolman' s psychology. Behaviorism, 10, 151–163 .


11 7

Smith, L . D. (1986). Behaviorism and logical positivism: A reassessment of the alliance. Stanford , CA: Stanford University Press . Smith, L. D. (1990) . Metaphors of knowledge and behavior in the behaviorist tradition . In D . E . Leary (Ed.), Metaphors in the history of psychology. (pp 239-266) . New York: Cambridge University Press . Spence, K . (1942). Theoretical interpretations of learning . In F. A . Moss (Ed .), Comparative psychology. New York: Prentice-Hall. Stevens, S . S . (1935) . The operational definition of constructs. Psychological Review, 42, 517-527 . Sutherland, N . S ., & Mackintosh, N . J. (1971) . Mechanisms of Animal Discrimination Learning. New York: Academic Press . Tolman, E. C . (1920) . Instinct and purpose . Psychological Review, 27, 217-233 . Tolman, E . C . (1922) . A new formula for behaviorism . Psychological Review, 29, 44-53 . Tolman, E . C. (1923) . A behavioristic account of the emotions . Psychological Review, 30, 217-227 . Tolman, E . C . (1924) . The inheritance of maze-learning ability in rats . Journal of Comparative Psychology, 4, 1-18 . Tolman, E . C . (1925) . Purpose and cognition : The determiners of animal learning. Psychological Review, 32, 285-297 . Tolman, E. C . (1926) . A behavioristic theory of ideas . Psychological Review, 33, 352-369 . Tolman, E . C . (1927) . A behaviorist's definition of consciousness. Psychological Review, 34, 433-439 . Tolman, E . C . (1928) . Purposive behavior. Psychological Review, 35, 524-530 . Tolman, E. C . (1932) . Purposive behavior in animals and men . New York : Century Co. Tolman, E . C . (1935) . Psychology vs . immediate experience . Philosophical Science, 2, 356-380 . Tolman, E . C . (1936) . Operational behaviorism and current trends in psychology. Proceedings of th e 25th Anniversary Celebration of the Inauguration of Graduate Studies, University of Southern California (pp. 89-103) . Los Angeles : University of Southern California . Tolman, E . C . (1938) . The determiners of behavior at a choice point . Psychological Review, 45, 1-41 . Tolman, E . C . (1939) . Prediction of vicarious trial and error by means of the schematic sowbug . Psychological Review, 46, 318-336 . Tolman, E . C. (1941) . Discrimination vs . learning and the schematic sowbug . Psychological Review, 48, 367-382. Tolman, E. C . (1942) . Drives toward war. New York: Appleton-Century. Tolman, E. C . (1948) . Cognitive maps in rats and men . Psychological Review, 55, 189-210 . Tolman, E. C . (1949) . The psychology of social learning . Journal of Social Issues, Supplement 3 , 5 18 . Tolman, E . C . (1951a) . Behavior and psychological man . Berkeley and Los Angeles : University of California Press . Tolman, E . C . (1951b) . A psychological model . In T. Parsons & E . A. Shils (Eds.), Toward a general theory of action (pp . 279-361) . Cambridge, MA : Harvard University Press . Tolman, E . C . (1952) . Edward Chace Tolman. In E .G . Boring, H . S . Langfeld, H . Werner, & R. M . Yerkes (Eds .), A history of psychology in autobiography (Vol . 4, pp . 323-339) . Worcester, MA : Clark University Press. Tolman, E . C . (1959) . Principles of purposive behavior. In S . Koch (Ed .), Psychology : A study of a science (Vol. 2 , pp. 92-157) . New York: McGraw-Hill . Tolman, E . C ., & Honzik, C . H . (1930). Introduction and removal of reward and maze performance in rats . University of California Publications in Psychology, 4, 257-275 . Tolman, E . C., Ritchie, B . F., & Kalish, D . (1946). Studies in spatial learning : I . Orientation and the short-cut . Journal of Experimental Psychology, 36, 13-25 . Watson, J. B . (1914) . Behavior : An introduction to comparative psychology . New York : Henry Holt . Young, P. T. (1933) . Review of Purposive behavior in animals and men. American Journal of Psychology, 45, 177-178.


5 CLARK L . HULL' S BEHAVIORIS M

MICHAEL E . RASHOTTE

ABRAM AMSE L

Florida State University

University of Texas at Austi n

Introduction Brief Biographical Sketc h I . Theoretical Styl e A. Influence of Machine s B. Plan for Developing a Behavior Theory C. Perspective on Hull's Attempt to Make a Forma l Behavior Theory D. Physiology and Hull—Spence Theory II . S —R Analysis of Cognitive Processe s A. Knowledge, Foresight, Directing Ideas, and Purpose B. Gestalt Phenomena C. Habit—Family Hierarchies (Networks of Associations ) D. Kenneth Spence's Analysis of Transposition E. Application to Psychotherapy and Symbolic Processe s III . Conceptualizations of Reinforcement and Reward : Evolution of Hull's 1943 Theory A. "Reinforcement" in Hull's 1943 Theory B. Difficulties Recognized by Hull in 1943


119


120

RASHOTTE AND AMSE L

C. The Elliott– Crespi–Zeaman Experiment D. Spence's Contributions to Reward–Incentive Theory E. Amsel's Frustration Theory : A Conditioning Model of Effects of Reward and Nonreward Concluding Comments

INTRODUCTIO N For a period of 23 years, beginning about 1929, Clark Leonard Hul l (1884–1952) devoted his academic efforts toward developing a sophisticate d version of behaviorism that was fully committed to the language and orientations of stimulus–response (S –R) psychology . The success and influence of hi s work are a matter of record (e.g ., Amsel & Rashotte, 1984 ; Logan, 1959) . As behaviorism enters a new second century, and as cognitive science and neuroscience have come to be dominant influences in psychology, it is timely to reflec t on the forces that motivated Hull's attempt, the breadth and character of his effort, and some points of its success and failure . It is our opinion that Hull woul d have warmly embraced many of the present-day developments, and that a reading of his work, and the critical responses it attracted, can provide useful historical perspective on some modem-day efforts in behavioral, cognitive, and neural science . In the three main sections of this chapter, we illustrate some important aspects of Hull's theoretical work and its influence . We begin with an overview of the style of his theoretical approach . Next, we review selected theoretical analyses developed by Hull and his associates to conceptualize cognitive phenomena within the S –R framework analyses that greatly expanded the scope of S – R behaviorism and have had a lasting effect in twentieth-century psychology. Finally, we illustrate how Hull's theoretical approach responded to the pressures o f empirical findings in one focused area the conceptualization of reinforcemen t and reward that have been, and remain, enormously influential in behavior theory. We begin with a biographical sketch of Clark Hull .

BRIEF BIOGRAPHICAL SKETC H Clark Leonard Hull was born on a farm near Akron, New York, on May 24, 1884, and died in New Haven, Connecticut, on May 10, 1952, where he wa s Sterling Professor of Psychology at Yale University . When he was 3 or 4 years of age, his family moved to an unimproved farm in West Saginaw, Michigan, where he spent his early years under "pioneer condi-

5 CLARK L. HULL ' S BEHAVIORISM

12 1

tions" (Hull, 1952b, p . 143) . He was educated in local elementary and high schools, at Alma College and the University of Michigan (bachelor's degree , 1913), and at the University of Wisconsin at Madison, where he did his graduate work in experimental psychology (master's degree, 1915 ; doctoral degree , 1918) . Hull was 34 years old when he earned his doctorate, and his career as a scientist continued for an additional 34 years, right up to the time of his death . His scientific standing was recognized by election to membership in a number o f societies, including the American Academy of Arts and Sciences (1935) and th e National Academy of Sciences (1936) . He was awarded the Warren Medal of th e Society of Experimental Psychologists (1945), and he served as president of th e American Psychological Association in 1935-1936 . Hull's accomplishments were made against a background of generally poor health that began at the tim e of his graduation from high school when a bout of typhoid fever left him with a poor memory. A few years later poliomyelitis left him with a permanently paralyzed leg . In his later years, he was plagued with a heart condition . His first academic appointment was at the University of Wisconsin, where h e stayed for about a decade after receiving the Ph .D . and rose through the ranks t o professor in 1925 . During those years he worked on a variety of projects including the effects of tobacco smoking on mental and motor efficiency, the improvement of vocational guidance through improved aptitude testing, and the scientific basis of hypnosis (see Hull, 1952b) . In 1929, at the age of 45 years, Hull accepted the position of research professor of psychology at the Institute of Psychology (later the Institute of Human Relations) at Yale University, where he remained for the rest of his career. He had no formal teaching assignment at Yale , but engaged in graduate instruction through a weekly seminar that attracted graduate students and personnel of the institute for discussion of a variety of is sues in behavior theory. Hull is, perhaps, best known for Principles of Behavior: An Introduction t o Behavior Theory (1943b), the first volume of a planned three-volume effort. In Principles, he described a set of formal axioms constituting a logical syste m from which hypotheses about mammalian behavior could be deduced . This wa s to be the first of a planned three-volume effort . The second volume, A Behavio r System : An Introduction to Behavior Theory Concerning the Individual Organism (Hull, 1952a), appeared shortly after his death . It was intended to illustrate the ability of the system to generate fruitful hypotheses and deductions for th e behavior of animals in nonsocial settings . The third volume, which was never written, was to have applied his system to some elementary phenomena of mammalian social behavior. In addition to these three volumes, Hull's publishe d works include a large number of experimental papers and 21 Psychological Review articles that exerted great influence in psychology at the time of their publication and continue to be influential even today (Amsel & Rashotte, 1984 ; Spence, 1952a) . Some of the key influences arising from Hull's contributions t o behavior theory are discussed in the remainder of this chapter .

I22

RASHOTTE AND AMSE L

I . THEORETICAL STYL E A . INFLUENCE OF MACHINES

Among the forces that motivated Hull's work in behavior theory, his endurin g interest in machine technology should be especially noted . Since Descartes , much theorizing in psychology has been influenced by machine metaphors . The hydraulic system of Sigmund Freud, the telephone switchboard of the early connectionists, and the computer-based models of memory and of present-day connectionist approaches come to mind as some examples . In the early years of thi s century, Hull was influenced by the emerging technology of machines that coul d carry out some rudimentary functions of human cognition (e .g., calculation of algebraic outcomes) . Indeed, entries made in his private notebooks (Hull, 1962) , his construction of a machine to aid in the calculation of correlation coefficient s (Hull, 1925a, 1925b), early publications that described electromechanical de vices that could "learn" (Baernstein & Hull, 1931 ; Hull & Baernstein, 1929 ; Krueger & Hull, 1931), and comments in a later paper about having constructe d a machine made of "sliding disk segments of sheet metal" designed to "exhibi t the conclusions logically flowing from all of the known syllogisms" (Hull , 1935a, p . 219) illustrate his deep interest in machine technology and its possibl e relevance for understanding psychological processes . An example of his thinking is provided in the following quotation from a 1931 paper titled "An Electro-Chemical Parallel to the Conditioned Reflex ." Thi s passage advocates the value of designing and constructing machines (robots ) that simulate adaptive behavior as part of the behavioristic program to banish mentalistic explanations from behavior theory. It is believed that the construction and study of models of the type described above [i .e. , physical devices capable of learning] will aid in freeing the science of complex adaptiv e mammalian behavior from the mysticism which ever haunts it . The belief is very wide spread and persistent that certain complex forms of adaptation cannot take place by an y imaginable concatenation of materials without the mediation of some nous, entelechy, soul, spirit, ego, mind, consciousness, or Einsicht. There is, on the other hand, the op posed belief that the above explanatory concepts are but the names of disembodied functions (ghosts) which, insofar as they have any objective existence, are themselves in need of explanation and may conceivably be duplicated by adroit concatenation of materials . The conflict between these points of view is probably at a permanent impasse unless th e second party can bring forth tangible evidence of synthetic achievement in this direction . That a considerable degree of success in this practically virgin field will reward a vigorous effort can scarcely be doubted . As progress is made by the second group it may b e anticipated that the first will gradually retreat to the more and more inaccessible parts o f the psychological terrain . How successful the attempts to construct "psychic " machines will prove, time alone will tell . It will be too early to venture a final judgment until a s much labor and ingenuity have been expended in this attempt as have been devoted to th e development of the steam engine, the electric motor, or the printing processes . Unless some practical use for such adaptive mechanisms should develop, this will naturally be a very long time . It is not inconceivable, however, that in the demands for higher and highe r degree of automaticity in machines constantly being made by modern industry, the ultra

5 CLARK L . HULL ' S BEHAVIORISM

12 3

automaticity of the type of machine here considered, may have an important place . In that event the exploration of the potentialities lying in this radically new order of automaticity would be comparatively rapid . (Krueger & Hull, 1931, 267–268 ; emphasis in original )

Many modern-day theorists in psychology and cognitive science will fin d their own thinking reflected in Hull's proposal that important advances can b e made by adopting the logical and quantitative discipline necessary to construc t machine simulations of behavioral phenomena . They will also empathize with Hull's discussion of the difficulties inherent in such an undertaking . For example, designing a machine that behaves very differently depending on smal l changes in the pattern of stimulation to which it is subjected was seen by Hull a s a major problem (e .g ., Hull, 1929 ; Krueger & Hull, 1931) . Subsequent developments in the study of "compound stimulus learning" in conditioning experiment s (including stimulus configuration) and "pattern recognition" in cognitive studie s are illustrative . Of course, the materials available for constructing actual "psychic machines" were extremely limited in Hull's day, and he devoted much of the rest of his career to constructing "virtual machines" (theoretical systems ) that simulate (predict) behavioral phenomena of interest . Just as the design of an actual machine calls for adherence to strict logical and quantitative rules in combining a set of physical materials, Hull's theoretical systems attempted to use the same approach in combining a set of conceptual terms . The success of behavioral predictions from his systems were the means by which the adequacy of hi s theoretical "machines" was to be judged . B . PLAN FOR DEVELOPING A BEHAVIOR THEOR Y

About the time he moved from the University of Wisconsin to Yale University, in 1929, Hull began to put into practice a "master plan" for developing a behavior theory. At Yale, he joined the Institute of Human Relations, where psychologists, sociologists, economists, and cultural anthropologists pursued th e ideal of a unified approach to social science, and it was in this atmosphere tha t Hull began his major theoretical work. He proposed to publish three key volumes that would cover "in an elementary manner the range of ordinary mammalian behavior" (Hull, 1952a, p . vii) . His orientation is well expressed by the following passage from his autobiography, in which, having acknowledged tha t his approach was probably unfluenced by his early training in the physical sciences, he writes : I came to the definite conclusion around 1930 that psychology is a true natural science ; that its primary laws are expressible quantitatively by means of a moderate number of ordinary equations; that all the complex behavior of single individuals will ultimately b e derivable as secondary laws from (1) these primary laws together with (2) the condition s under which behavior occurs ; and that all the behavior of groups as a whole, i .e ., strictly social behavior as such, may similarly be derived as quantitative laws from the same primary equations . With these and similar views as a background, the task of psychologist s obviously is that of laying bare these laws as quickly and accurately as possible, particu -

124

RASHOTTE AND AMSE L

larly the primary laws . This belief was deepened by the influence of my seminar students , notably Kenneth W. Spence and Neal E. Miller. It has determined the most of my scientific activities ever since, and the longer I live the more convinced I am of its general soundness. (Hull, 1952b, p . 155 )

To undertake this task, Hull decided to write a series of theoretical papers that would address various issues on which he wished to attract comment . The ideas in such papers would then be refined and presented in a unified volume, alon g with experimental evidence, to stand as the principles from which he would predict behavior. Hull described this work as really amounting to a case of "mechanical design" (Hull, 1962, p . 833), and the first of his papers appeared in the Psychological Review in 1929 (Hull, 1929) . In all, he wrote 21 articles and tw o brief notes on learning theory in that journal through 1950 . These works include theoretical accounts of such topics as complex cognitive processes (e .g ., knowledge, purpose, foresight, directing ideas), Gestalt phenomena (e.g ., the appearance of novel responses in problem solving, the behavior of young children i n spatial situations), species and individual differences, and behavior in social situations . We think that these Psychological Review articles best represent his orienting attitudes and the true scope of his effort to develop a behaviorism . All of these articles have been reprinted in a single volume, with commentary added b y us (Amsel & Rashotte, 1984) . Readers of the present chapter might consult tha t earlier volume for more biographical details concerning Hull, and for our comments on the ways ideas in his papers relate to previous and subsequent developments in psychology, as we saw it in the early 1980s . Briefly, the intellectual roots of Hull's planned behavioristic theory lay in a diverse set of important ideas concerning behavior and the nature of theory . A central focus of the plan was to blend the conceptual and experimental advance s in Edward L . Thorndike's (1898, 1911) S –R psychology and Ivan P. Pavlov' s (1927) approach to behavioral plasticity into a unified account. Both of these approaches were mechanistic, of course, as suited Hull's taste . He also included a strong focus on behavior as an adaptive process, and on the role of motivation i n energizing and directing behavior, reflecting the influences of Charles Darwin and Sigmund Freud, respectively . Finally, Hull emulated the theoretical form o f Isaac Newton's Principia, which had such a powerful effect on physics . In many of his writings, Hull attempted to outline his theoretical arguments as formal an d quantitative statements within a hypothetico-deductive methodology . He mad e heavy use of the intervening variable, a model Edward C . Tolman had introduced into psychology (Tolman, 1938), and he endorsed Bridgman's operationalism as a requirement in the definition of these variables . Later, he and others in the Hullian group wrote extensively on the nature and advantages of thi s theoretical form over vaguely stated theories or atheoretical approaches . Arriving on the scene after John B . Watson had made behaviorism well known and controversial in American psychology, Hull's plan for an S –R behaviorism was very ambitious . It aspired to predict the behavior of individuals in isolation, as well as in group settings . It aspired to conceptualize the bases for

'

5 CLARK L . HULL S BEHAVIORISM

12 5

adaptive behavior in a broad sense, including certain cognitive processes and the performance differences between species and individuals . It aspired to be logically rigorous and mathematical as a way of ensuring that its assumptions an d predictions were clear and available for comparison with competing theories . The first volume of Hull's planned three-volume work was intended to be a primer of behavior theory that would simply set out the principles of the forma l system to be used in approaching specific topics in behavior theory. That volume, Principles of Behavior: An Introduction to Behavior Theory, published i n 1943 when Hull was 58 years old, is often cited (incorrectly, we think) as th e core of his theoretical work. It made little, if any, attempt to elaborate the complex behavioral implications of the system . Indeed, the book was devoted to an analysis of classical and instrumental conditioning to the exclusion, even, of the more complex analyses of purpose, foresight, and so on characteristic of Hull' s own earlier work in the 12 Psychological Review articles he had published up t o that time . He had used the formal postulate style in an earlier monograph, Mathematico-Deductive Theory of Rote Learning (Hull et al ., 1940), in which he attempted to explore the logical implications of conceptualizing rote learning i n terms of Pavlovian processes . He considered that work a "dress rehearsal" fo r Principles of Behavior, and both works illustrate one way that behavioris m might utilize a highly formalized theoretical approach . Immediately after publication of Principles, Hull began working on the second volume in the series and in 4 years had completed a draft of 12 chapters . This second book was intended to demonstrate the derivation of more complex phenomena of behavior from the already-published postulates of the system i n his "primer." However, difficulties in quantifying some of the theoretical variables, and frail health, delayed publication, and the book, A Behavior System : A n Introduction to Behavior Theory Concerning the Individual Organism, which was finally completed in his 68th year, appeared in 1952, shortly after his death . A sample of its chapter titles conveys the range of phenomena to which the postulates were applied : "Simple Trial-and-Error Learning," "Discrimination Learning," "Learning within the Individual Behavior Link," "Behavior in Relation t o Objects in Space," "Multidirectional Maze Learning," "The Problem-Solvin g Assembly of Behavior Segments," "Value, Valuation, and Behavior Theory." Some of these chapters were updated versions of papers Hull had published earlier, some had circulated in the psychological community for several years in the form of his seminar proceedings, and the chapter on discrimination learning borrowed heavily from classic papers by his close associate Kenneth Spence (1936 , 1937, 1938, 1942) . The third volume Hull had planned was never written . Of it , Hull said in his autobiography, "for many years I have felt that by far the mos t important portion of the system for civilization in general would be found in th e third volume, which would concern the strictly social relationships among subjects" (1952b, p . 162) . As noted previously, he believed that the principles underlying social behavior were not fundamentally different from those underlyin g individual behavior and was optimistic that the elementary phenomena of mam-

126

RASHOTTE AND AMSE L

malian social behavior could be derived from the primary "laws" stated in Principles . Obviously, this unfinished part of Hull's work was intended as an inter face between psychology and the disciplines of sociology, economics, and cultural anthropology, all of which were represented in the faculty of the Institute o f Human Relations at Yale . Kenneth W. Spence played an intimate role in the development of Hull's theorizing (Amsel & Rashotte, 1984), and after Hull's death Spence assumed th e central position in the Hullian movement . In the 10 years or so preceding Spence's death, in 1967, the more formal parts of the theory became known a s the Hull–Spence approach (e .g ., Logan, 1959) . Well before Hull's death the major experimental work testing the theory was done at the University of Iowa by Spence and his students . Partly as a result of the work at Iowa, Spence favored revisions in certain portions of the theory, and these were reflected in his ow n most unified theoretical statement made in his Silliman Lectures at Yale, published as Behavior Theory and Conditioning (Spence, 1956) . C . PERSPECTIVE ON HULL'S ATTEMPT TO MAKE A FORMAL BEHAVIOR THEORY Hull's endorsement of formal theorizing stands out in the initial chapters of Principles of Behavior, where we find a discussion of the deductive nature of successful scientific theories in the natural sciences and a commentary on the application of this approach in the behavioral sciences : Following the natural-sciences model, the behavioral scientist elaborates a set of postulates, or first principles, and uses them as premises in deducing, by rigorous logic, inferences o r theorems about behavioral phenomena . The postulates are abstract principles derived from available empirical evidence and influenced by shrewd guesses abou t the process underlying behavior. These postulates often involve hypothetical entities ("intervening variables"), invented by the theorist to organize his thinkin g about the relationships among experimental manipulations and measurement s (independent and dependent variables) related to behavioral phenomena of inter est . The theory can then be evaluated by translating the deductions from the theory into experimental operations and seeing how it fares in the laboratory . This theoretical approach advocated in psychology by Tolman and then b y Hull had great appeal in the abstract . In practice, however, the difficulties prove d overwhelming . First of all, the systematic gathering of empirical evidence abou t learning (not to say behavioral phenomena in general) began only late in th e nineteenth century, and the experimental methods were often crude and overl y complex in the early work . Accordingly, when Hull was writing Principles, about 60 years ago, the empirical base with which he started was very weak an d he relied heavily on shrewd conjectures in the formulation of his postulates . (In a later section of this chapter, concerned with reinforcement and reward in Hullia n theory, we illustrate how such deficiencies came to light as a result of experimental work and how the theory changed as a result.) Second, the absence of a

'


12 7

sound empirical data base for the known phenomena of learning made it difficul t for Hull to be as meaningfully quantitative as he would have liked to be . Nevertheless, in Principles, he went through the motions of quantification in the cas e of some constructs (e .g ., excitatory potential SER), apparently to point the direction for future endeavors . Finally, Hull's overall theoretical objective (Hull , 1952a, p . vii) to state the principles that allow deductions of "the range of ordinary mammalian behavior" (even if only "in an elementary manner") was extremely ambitious for the 1940s ; indeed, the objective seems unattainable for th e 1990s . Hull realized, and stated early in his second book, A Behavior Syste m (1952a), that, in practice, rigorous deductions about even moderately complex learning phenomena could not be made from the original postulate set because it was incomplete and not sufficiently quantitative. In fact, he characterized the deductions as "informal," and commented that it was probably too early to attemp t rigorous deductions about even a limited range of behaviors . The formal styl e and the scope of Hull's theorizing in Principles was clearly premature, at least in the sense that it was based on too little empirical work. Of those theorists who came earliest under his influence, he was emulated i n style but not in scope by K. W. Spence, and in neither style nor scope by N. E . Miller and O . H . Mowrer, although all three worked with many of the basic concepts Hull developed. The later work of Spence and his students harked back i n some respects, and particularly in its emphasis on motive–incentive factors in behavior, to Hull's papers of the thirties on purpose and foresight . From the present perspective, Hull's own final assessment of the formal theoretical styl e seems correct : It was (and still seems to be) premature to undertake formal theorizing covering a wide range of areas or topics even in an area about which muc h is known, the psychology of learning . The present inclination is to make theory "in smaller pieces," to limit theoretical integration to behavioral phenomena of lesser scope, and to investigate the neural mechanisms underlying such phenomena . But, as we shall also see, important spin-offs from Hull's theoretical effort s were still occurring, to go along with the tradition of appreciation for the natur e of theory and its relation to experimentation (Bergman & Spence, 1944 ; Hull , 1943a, 1943b, 1943c ; Miller, 1959 ; Spence, 1948, 1952b, 1952c, 1957) . Despite severe criticism (Koch, 1954) there is little doubt that Hull ranks as one of th e most influential theorists in modern psychology . D . PHYSIOLOGY AND HULL – SPENCE THEORY

In view of the great interest and progress being made today in the neurobiology of learning, memory, and performance, it is of interest to note Hull's attitudes toward including physiological mechanisms in behavior theory, and Kenneth Spence's long-running disagreement with Hull on this matter. As the twentieth century draws to a close, some research programs based on evolve d versions of Hull–Spence theory include a focus on neural mechanisms underlying behavior theory. We draw attention to one of these programs later in this

128

RASHOTTE AND AMSE L

chapter in consideration of Amsel's recent work in developmental neurobiolog y of learning and memory. Here, we present excerpts from private correspondenc e between Hull and Spence to characterize the tensions within the Hull–Spenc e system concerning this issue . Hull was in favor of using physiological terms, partly to attract students whom he assumed would find speculations about the physiological nuts an d bolts of the system to be more easily grasped than the abstract mathematical formulations he sought to develop . (This is, perhaps, a carryover from his own dee p interest in the structure of machines that carry out psychological functions .) Spence, on the other hand, argued that such physiological speculations detrac t from behavior theory, much the same type of argument that is often associate d with Skinner's behaviorism (see Ringen, this volume) . From the present-day perspective, neural network approaches to learning would seem to be close to the spirit of Hull's thinking with, of course, the benefit of modern computational machines and major advances in neurobiology. Spence's position is illustrated in a letter to Hull in 1943 . Spence had received an autographed copy of the newly published Principles and speculate s about the probable response by reviewers : I shall await with interest the reaction of the reviewers . As I have indicated several times , I feel quite sure that most of the people who read the book will completely miss the point and will go off into picayunish criticism of some of the neurophysiological "notions " that you have elaborated. This, of course, represents our old difference in viewpoint . I have always been very unhappy about the fact that you have been inclined to throw in hypotheses as to the mediational mechanisms underlying the abstract mathematical concepts . I can just see the critics bemoaning the extent to which you hypostatize little things insid e the brain . In other words I fear that you will have the non-physiologically minded peopl e jumping on your theorizing as being nothing but neurophysiological speculation and the physiologically minded people for having made what from their point of view would be naive neurophysiological speculation . To me, of course, the meat of the discussion is th e framework of mathematical theory (constructs and postulate relations) and I would have been much happier if the whole discussion had been kept on a very abstract level . (Kenneth W. Spence to Clark L . Hull, September 8, 1943, Kenneth Spence Papers, Archives o f the History of American Psychology, University of Akron . )

Hull responded promptly, noting that he was "plugging along" on the manuscript for his second book (ultimately, A Behavior System, which appeared almost a decade later), before commenting on Spence's earlier letter : Thanks much for your remarks about the new book . I think that I am following out you r preference substantially in the writing of the second volume, as I think you will find little or nothing of the objectionable physiological nature in the Chapter II. The same, I think , will be true in all of the others . I simply take these equations and go on from there, paying no attention whatever to the physiological suggestions which I sprinkled more or les s through the first volume . I am, of course, not very much concerned with what the reviewers say; what I am interested in is what the graduate students say . Those are the boys who will decide the matter. I would be very much interested to have relayed to me the reactions of various types of graduate students, both the Lewinian type and your own students . As you know, my motive in introducing the sub-molar physiological suggestions

'

5 CLARK L. HULL S BEHAVIORISM

12 9

was the hope that it would make it somewhat easier for relatively naive graduate student s to secure a sense of reality from the theory, which many people have difficulty in doin g from the mere inspection of equations . (Clark L . Hull to Kenneth W. Spence, September 21, 1943, Kenneth Spence Papers, Archives of the History of American Psychology, University of Akron .)

II . S—R ANALYSIS O F COGNITIVE PROCESSE S A . KNOWLEDGE, FORESIGHT, DIRECTING IDEAS, AND PURPOS E The adequacy of an S –R analysis of behavior was severely challenged by behavioral phenomena that implied that the behavior of animals is at least some times guided by future events rather than by the prevailing stimulus conditions . A commonplace example is that a laboratory rat performs a response that pre vents the occurrence of a future event or chooses a new pathway that allows it t o reach a distant goal in the shortest time when its established pathway to the goa l is blocked. In common language, behaviors such as these indicate "foresight, " "knowledge," "purpose." It was not obvious that S –R psychology, with its insistence on behavioral control by prevailing stimulus conditions, could conceptualize adequately such instances of behavior that appeared to be controlled by th e not-here and the not-now. Hull provided an S –R account of foresight, knowledge, and purpose in a series of articles published in the Psychological Review in the early 1930s . These papers introduced several concepts that remained influential in S – R psychology, even though Hull, himself, neglected much of the content of thes e papers in his later work, particularly in the statement of his theory in Principle s of Behavior in 1943 . In these early theoretical papers Hull exhibited a notable tendency to employ certain principles in a loosely structured argument to deduce from stimulus-response language the behavioral phenomena attributed t o foresight and purpose . In this respect, these papers anticipated Hull's later attempt to build a formal theory with rigorous deductive logic . These early papers also were characterized by an emphasis on the biologically adaptive significance of processes, such as foresight and purpose, and provide a good example of the Darwinian orientation that was particularly prominent in Hull's earl y writings . It will become clear in the following discussion that, while Hull appealed to a variety of principles in these papers, his analysis of purposive behavior prove d particularly significant for future theoretical developments because it introduce d the fractional anticipatory goal response, rG -SG . The proposal was that the stimuli present while an organism is striving to reach a goal will, themselves, com e to evoke some components of the goal response in anticipatory form (r G) through Pavlovian conditioning ; and, that such responses generate propriocep-

130

RASHOTTE AND AMSE L

tive feedback stimuli (5G) that are then available to enter into associations and direct behavior. With this concept, Hull formalized his view that Pavlovian conditioned reflexes play an important, if not a dominant, role in guiding behavior i n Thorndikeian learning situations . In 1929, Hull published the first of his 21 Psychological Review articles . [All of these have been collected in their original form and pagination, and republished on the centennial of Hull's birth, with extensive commentary (Amsel & Rashotte, 1984) .] It concerned the functional significance of Pavlov's conditioned reflex and was distinctively Darwinian in tone, emphasizing the significance of Pavlovian excitation and inhibition in achieving the fine behavioral adjustments necessary for survival . For example, it included consideration o f signaled avoidance learning in which the experimental animal takes protectiv e action in advance of an impending aversive event, as if it anticipates the harmfu l event . In important papers published in the two succeeding years, Hull (1930, 1931 ) emphasized that animals with "knowledge" and "foresight" enjoyed additional adaptive advantages . These papers served to illustrate a way in which S —R theory could conceptualize these cognitive phenomena in mechanistic terms, an d they were perhaps the first conditioning-based accounts of "purposive behavior . " Hull's analysis was made with reference to a basic case in which a sequence o f behaviors (R 1 —R2 —R3 —R4 —R5 ) occurs because the stimulus properties of th e world are arranged in a way that evokes those responses in that sequence (i .e . , S1 - S2 - S3 - S4 - S5 . This situation is shown in the following diagram taken from Hull (1930) : S3

The World :

The Organism :

R~

S4

R3

-~ S S

R4

R5

If this sequence of S —R events is repeated several times, certain principles come into play that illustrate how an organism gains a rudimentary knowledge of it s world, and shows foresight and purpose . One principle, response-produced proprioceptive feedback, well supported b y physiological studies even in 1930, is that each response produces a distinctive , internal feedback stimulus that has the potential to operate as a cue like all other stimuli . The resulting state of affairs, reproduced in the following diagram, i s that when the response sequence is activated by the external stimulus sequence, associations form between the response-produced, internal feedback stimuli an d the responses that follow them . These associations are symbolized as dotted arrows in the following diagram :

'

13 1


The World :

The

Organism :

S1

S2

R 1 -0-s 1

.

.-S3 •

S 4

R22 -0—r') 2- - -+ K 3 —n0-0 33

S

-0-0

44

5 -- -

The second principle, redintegration, states that any and all stimuli present at a given time acquire the capacity to evoke the responses that follow them. Thi s further principle leads to the deduction that once these feedback stimuli occur i n conjunction with certain responses, they can evoke the responses in the sequence, even in the absence of external stimuli . In the event, for example, the external stimulus sequence were interrupted at S 1, the entire response sequenc e would run off anyway, as shown in the following : The World : S 1

The Organism :

R1--*-s 1-♦`R2--0--s2

S4_ . R 5 - -

The third principle holds that the response sequence can run off at a faste r tempo when cued by response-produced stimuli than by external stimuli . Thi s principle of "foreshortening" makes it possible for R 2 , say, to appear in apparent anticipation of the external stimulus S 2 . But the preceding diagram shows tha t this anticipation is more apparent than real because R2 is in fact cued by a pre ceding stimulus, the feedback stimulus (s I ) from R 1 . Anticipatory responses o r " intrusions " such as these are often taken as evidence of foresight, and followin g is a simple S —R account of how they develop. Hull was impressed with an economic imperative that energy expenditure be reduced to a minimum in the context of adaptive biological functioning — a fourth major principle evident in his early work . One instance of this principle of behavioral parsimony is the tendency for responses to diminish in magnitude t o a level just adequate for the production of some biological adaptive consequence . Consider the operation of such a principle in the case of the response sequence described earlier. Suppose that it is adaptive to perform R3 prior to the occurrence of S 3 , and that a successful outcome of the response sequence depends only on R3 . The third principle provides the basis for how R3 comes to antedate S 3 : the response sequence runs off at a faster tempo than the external stimulus sequence because the responses in the sequence are links, chained togethe r by response-produced feedback stimuli . The fourth principle holds that the intensity of R 1 and R2 may diminish greatly because the sole function of these responses is to produce feedback stim -

132

RASHOTTE AND AMSEL

uli that will allow the chain of responses to run off. Indeed, R 1 and R 2 may diminish even to a covert level, so long as their intensity at that level is sufficien t to produce feedback stimuli for the next response in the sequence . Pure stimulu s act was the term employed by Hull for responses whose principal function is t o produce stimuli that control other responses . He viewed these acts as being of the greatest biological significance, because the stimulation they provided for the control of behavior was ultimately independent of the stimulation provided b y the external environment . The fact that behaviors might seem to appear spontaneously or to be unrelated to the prevailing stimulus conditions was not, i n Hull's view, necessarily an embarrassment for S –R theory. In fact, such behaviors fit rather easily into the S –R schema, when analyzed in terms of the set o f principles he invoked. Pure stimulus acts, particularly those that have receded t o a covert level, were viewed as rudimentary forms of directing ideas or thoughts , and were particularly obvious analogues of internal mechanisms that control behavior. Relying on this analysis, Hull argued that "knowledge" could be conceptualized as the running off of a sequence of behavior (perhaps at a covert level) with out stimulus support from the world that normally evokes it, and that, when suc h internalization occurs, it indicates "that a functional parallel of this action segment of the physical world has become a part of the organism . Henceforth, the organism will carry about continuously a kind of replica of this world segment . In this very intimate and biologically significant sense the organism may be sai d to know the world" (Hull, 1930, p . 514) . In this same paper, he proposed that "foresight" be viewed as the running off of a response sequence at a faster temp o than is possible when that sequence depends solely on stimuli from the externa l world . Hull's analysis of "purpose" in the paper "Goal Attraction and Directing Ideas as Habit Phenomena" (Hull, 1931) was based on yet another mechanism that was a brilliant invocation of the idea that Pavlovian conditioned response s can be significant factors in the control of instrumental behavior . In accepting this idea, Hull joined other analytical thinkers of the time who recognized that Thorndikeian response–reinforcer procedures can always be said to include a stimulus–reinforcer relation of the Pavlovian sort . Responses that produce the reinforcer in Thorndikeian conditioning are necessarily performed in some con text of stimulation that requires that a Pavlovian stimulus–reinforcer be embedded within that Thorndikeian procedure . Hull gave much careful consideration to the manner in which Pavlovian conditioning might develop within and influence—performance in Thorndikeian learning . For this purpose, he introduced the concept of the fractional anticipatory goal response (r G sG), which was a Pavlovian mechanism designed to account for anticipation or purpose in S – R learning theory. Behavior is said to be purposive when responses leading to a goal are directed or guided in some way by that goal . The difficulty to be overcome in an S –R theory is that, in these cases, the goal occurs only after the response has been made .

13 3


In his analysis, Hull first considered the features of behavior that might lea d an observer to describe it as purposive . In the case of a rat learning to ru n through a multiple T -unit maze for food, responding is variable in the early trial s but later is said to take on a "purposeful" quality because the rat (a) speeds up a s it approaches the goal ; (b) begins turning movements, in anticipation of reachin g the final choice point, often resulting in incorrect choice of maze arms in earl y trials ; and (c) begins to make masticatory movements before the goal is reached . While the acceleration in responding as the rat approaches the goal was easil y accounted for by the goal-gradient hypothesis, which holds that responses wil l be more firmly connected to stimuli the closer the S –R event is to the reinforcer (i.e., the goal), the remaining two characteristics of "purposive" behavior d o seem to reflect the anticipation of a future event. As the following discussio n shows, it was these anticipatory characteristics of behavior that Hull explaine d by appealing to Pavlovian conditioning . Briefly, Hull's analysis portrayed the Thorndikeian response as a sequence of S –R connections including a final one in which the goal stimulus (S 4 represents the food-pellet goal stimulus in the following diagram ; designated elsewhere as S G) evokes the goal response, RG (e .g . , grasping, chewing, swallowing) . The top two rows of the following diagra m show this aspect of Hull's analysis : ~~-

N

1 The.

-1

r--

r--

-I

I

I

\

OrgaIlit ;P1

-1

_I

_

R

Id / `,

I)

s'

j/

/

J

C7

i‘ -

t_

,

I)

/

-i –t c4''G 4 _1

\,

--

G

The portion of the diagram labeled S D represents the stimulus condition tha t arises from the animal's motivational state . This is the constant portion of the stimulus complex that persists throughout the sequence of S –R events, and i s present when the goal is reached . Through Pavlovian conditioning, S D comes t o evoke a fractional part of the final goal response (rG –s G ), and because S D is constant throughout the response sequence, it should elicit this rG –sG throughout the sequence, although most strongly in the portions nearest the goal . In this manner, Hull provided a straightforward account, in Pavlovian terms, of how th e goal, or a fractional representation of it, can be available to the organism before the goal is physically present .

134

RASHOTTE AND AMSE L

The manner in which rG —sG comes to guide behavior is also reasonably straightforward . The SG portion of rG —sG represents the proprioceptive feedback stimulation associated with occurrence of the conditioned goal response . Because sG is present while the S —R segments of the Thorndikeian response are occurring, it becomes associated with those responses in each S —R segmen t through a simple contiguity principle (dashed arrows in the diagram) . Once associated with the Thorndikeian responses, SG would direct their occurrence in the same way as other stimuli do . The rG —s G mechanism can, in principle, accoun t for masticatory movements as the hungry rat approaches food at the goal . Such movements can be regarded as fractional Pavlovian conditioned portions of the consummatory response at the goal, just as salivation is a measurable fraction of the response to the food-in-mouth unconditioned stimulus (US), or to its anticipation, in a Pavlovian experiment . By simple extrapolation, this principle ac counts for anticipatory turning movements in a multiple-unit maze before the ra t reaches a choice-point in the maze ; it accounts, particularly, for the fact that in a multiple-unit maze, the anticipatory turns are likely to be in the same directio n as the final goal turn, especially in the latter units of the maze . In short, rG —sG seems to provide a plausible account of significant aspects of purposive behavio r entirely within the framework of simple conditioning principles . Hull's analysis of purpose was a brilliant coup . It allowed S —R theory to invade the territory of cognitive learning theory and even to compete successfull y with cognitive theory in predicting experimental results . Consider one case in which Hull's S —R analysis triumphed over Tolman's cognitive position . Edward C . Tolman (see Innis, this volume) conceptualized learning as the formation of cognitions about how one event leads to another, and viewe d learned behavior as reflecting the operation of cognitions about the consequences of behavior in a given environment (Tolman, 1932) . In one test of hi s position (Tolman, 1933), Tolman trained hungry rats to choose the white-curtained entrance in a black—white visual discrimination to gain access to a foo d box . An incorrect choice (black) resulted in confinement in a box without food where shock was given . After the rats were well trained in choosing the white — > food box alternative, they were placed directly in the food box without runnin g through the alley and received an electric shock . The point of this manipulation was to change the significance of the goal box so that, when, on a subsequen t trial, the rat was placed at the beginning of the runway, its cognitions its "insightful" behavior, in Tolman's language should now lead it to expect shock a t the end of the runway . Consequently, the rat should refuse to run . In the test, to Tolman's considerable surprise, and in his picturesque language, "[the rats] immediately dashed off gaily and just as usual through the whole discriminationapparatus and bang whack into the very food compartment in which [they had] just been shocked" (Tolman, 1933, p . 250) . Although Tolman suggested some possible reasons for the failure of the rats' cognitions in this situation, he conceded that the data might be a victory for S —R theory . He had reasoned that S —R theory should predict that the rats would fail to run only when the reaction


13 5

of running to the goal was followed by shock . (In fact, Tolman later trained rat s under the latter conditions, and found that they did fail to run in the test . ) In an example of how Hull's S –R analysis could make powerful prediction s of a "cognitive" sort, Neal Miller (1935) used the logic of Hull's fractional anticipatory goal response to set up conditions in which rats would indeed be expected to manifest "foresight ." Shocked at the goal, even without having run to it, they would show evidence of "cognitive" functioning that they anticipate d the shock that would occur at the end of the runway . Miller reasoned that shoc k at the goal would disrupt running on a subsequent test trial only if stimuli in th e goal box evoked a response that afforded a characteristic pattern of interoceptiv e stimulation . In that case the anticipatory goal response (rG) in the runway woul d have a distinctive feedback stimulus (SG) similar to that present when the anima l was in the goal box . Consequently, the goal–shock experience should conditio n reactions evoked by shock to the distinctive interoceptive stimuli and SG shoul d transfer the reactions to shock from the goal box to the alley and inhibit runnin g in the alley. Miller noted that in Tolman's experiment the goal response may no t have been sufficiently distinctive to provide a characteristic feedback stimulus i n the goal box . This lack of distinctiveness would account for Tolman's failure t o demonstrate foresight in his rats . Miller conducted an experiment in which the animals were required to mak e a highly distinctive goal-approach response at the end of the runway, such as climbing up and making a sharp turn to the right to enter a narrow tunnel wher e food was available. Under these conditions, the results were as Tolman had expected on the basis of a cognitive theory, but had been unable to demonstrate i n his experiment : Rats shocked at the goal while performing a distinctive goal response showed less vigorous running to the goal for food when subsequently re leased at the start of the runway, even though the runway training for food an d the shock training in the goal were given independently . Miller reported several other manipulations in this elegant paper that strengthened the foothold of S – R theory in the domain of the cognitive . These experiments by Tolman and Mille r are early examples of what is now called the "reinforcer devaluation" experiment, which has figured in some recent theoretical efforts to understand the nature of reinforcement (e .g ., Colwill & Rescorla, 1986) . A recollection by B . F. Skinner provides an interesting historical note concerning Hull, the rG –SG mechanism, and one of Skinner's own efforts to deflate assertions about the special nature of "cognitive" processes . Here are Skinner's words : In 1937 someone had said that the use of tokens was a kind of symbolic behavior beyon d the reach of rats, and I countered by teaching a rat . . . to pull a chain to release a marble from a rack, pick the marble up in its forepaws, carry it across the cage, and drop i t down a slot . . . . When Hull visited my laboratory . . . in September, 1937, he saw [the rat] at work and noticed that the rat licked the marble as it carried it across the cage . "Anticipatory goal behavior," he called it. I had another explanation : We never washed th e marbles and they became rather tacky and no doubt had some flavor if not nutritiona l value . (Skinner, 1977, p . 1007 ; emphasis in original)

136

RASHOTTE AND AMSE L

Skinner also pointed out that the heralded later discovery that goal-directed instrumental response sequences are sometimes disrupted by intrusion of "inappropriate" anticipatory goal responses (Breland & Breland, 1961) may have been anticipated by Hull decades earlier during this visit . Certainly, the eventual realization that such "misbehavior," as the Brelands termed it (Breland & Breland, 1961), originates in the Pavlovian contingencies inherent in Thorndikeia n training situations is in line with this assertion . Later in this chapter, we see that Hull's rG –s G mechanism subsequently be came a central organizing idea in Hullian accounts of reward and nonrewar d learning, in ways that were not anticipated in the original paper on "purpose . " B. GESTALT PHENOMEN A The Gestalt psychologists Wolfgang Kohler and Max Wertheimer, the American "sign-Gestalt" learning theorist Edward C . Tolman, and the field theoris t Kurt Lewin were among the most vocal critics of Hull's version of behavioris m as it was being developed . Hull's responses to these criticisms provided the texture of much of his work in the 1930s and beyond . Some of his Psychological Review papers reflect attempts to deal with behavioral phenomena that wer e commonly interpreted as clear indicators of "Gestalt" processes, which, for many, fell outside the explanatory scope of S –R behaviorism . Papers dealing with the appearance of sudden "insight" in the solving of problems and demonstrations of "reasoning" by rats (Hull, 1935a, 1935b), and with the way "detour " problems are solved (Hull, 1938), are illustrative . More generally, Hull struggled to work out a logical way to decide which of several theoretical approaches t o these and other phenomena was most fruitful (e .g ., Hull, 1935a, 1935b) . A more complete discussion of these issues from our perspective can be found in Amse l and Rashotte (1984) . We comment here on two aspects of the Hullian approac h to Gestalt phenomena . One concerns Hull's proposal to utilize "habit-famil y hierarchies" as the basis for some theoretical accounts . This idea provided S – R theory with a theoretical mechanism underlying flexibility in learned respons e outcomes . From today's perspective, these hierarchies bear some resemblance to the associative networks found in "neural network" approaches to learning an d performance . (In the fifth chapter of Principles of Behavior Hull provides an interesting discussion of hierarchies of unlearned response tendencies and their importance in a mechanistic behavior theory .) The other aspect of Hull's approach we will discuss is the famous theory proposed by Hull's student an d close associate, Kenneth Spence (1936, 1937), to account for the Gestalt phenomenon of "transposition." C . HABIT—FAMILY HIERARCHIES (NETWORKS OF ASSOCIATIONS ) To account for complex behavior in which there is response variability, and in which a successful response suddenly appears without it having been trained in

137


that situation, Hull introduced an ingenious associative mechanism that combine d two types of stimulus—response "compound habit mechanisms" (Hull, 1934a, 1934b) . Figure 5 .1 shows the separate "divergent" and "convergent" habit mechanisms, and their combination into the hybrid "habit—family hierarchy ." For Hull, a compound habit mechanism involved multiple associative tendencies . The divergent mechanism (shown as a) is characterized by a set of independent excitatory response tendencies that can be visualized as radiating out in a Part a

■■

A — •ft

au.

R

01.,

ft

...

‘

R

Rz

3

4

Part b

UM. .11. WM.

•WO OM. Oft. 40.

VR6

:

Part c

5

4

=r .w.

M r.. s r

.VIP ■I■W

r r ,s

•a. !VW

%

4'

Or

Sn

'''

r

'

R

6

1

50 /

N

'I,

,

SP

R.

.b

%

r 1".

RP

sm. or

FIGURE 5 . 1 Parts a, b, and c represent the diagrams shown as Figures 1, 2, and 8, respectively, in Hull's (1934a) "The Concept of the Habit-Family Hierarchy and Maze Learning : Part I" (Paper 6 in Amsel & Rashotte, 1984) .

138

RASHOTTE AND AMSE L

fan-shaped pattern from a single stimulus . These response tendencies are in competition in the sense that they cannot be executed simultaneously, and they pro vide a basis for flexibility in responding in a familiar stimulus situation. The convergent mechanism (shown as b) provides for functional equivalence amon g several stimuli because they all evoke the same response (R G) . Hull proposed that once this mechanism is established, it would be understandable that a new response established under one of these stimulus conditions (e .g ., sm) could appear with little or no delay when another of the stimulus conditions occurs (e .g ., sp ) . The habit–family hierarchy (shown as c) is a combination of the two individual mechanisms, joined by "a number of distinct behavior sequences which are supposed to differ greatly from one another both as to the qualitative nature o f their activity and as to the length of sequence ." (Hull, 1934a, p38) The length o f the sequences in the diagram represents inversely the excitatory strength of eac h at point S A : The shortest response to reach the goal is the top one (initiated b y R 1 ) ; the least direct sequence is the bottom one (initiated by R 4) . Obviously, i n this network of excitatory tendencies Hull has provided an ingenious way withi n the framework of S –R theory for considerable flexibility in behavior. Shoul d one of the pathways to the goal be blocked, for example, other alternatives ar e available in an orderly fashion . Hull exploited this idea in many settings an d noted that it serves as "the dominant physical mechanism which . . . provides the basis for a purely physical theory of knowledge" (Hull, 1934a, pp . 40–41) . Tolman, in his good-natured way, commented about the attractiveness of Hull' s theoretical diagrams : You are all familiar with such diagrams . They are very clever and can be invented, as I know to my cost, to explain practically any type of behavior, however far distant from a n instance of conditioning such a behavior might at first sight appear. I have, therefore, the greatest respect for them . And, even though I argue against them, I find myself continually being intrigued and almost ready to change my mind and accept them and Hull after all. (Tolman, 1938, pp. 13 -14)

Indeed, Hull was at his most creative in these proposals by which he extende d the S –R approach to some important behavioral phenomena studied by Gestal t psychologists, such as the demonstrations by Wolfgang Kohler and by Norma n R. F. Maier that novel behavioral solutions can appear suddenly when animal s are striving to achieve a goal (Hull, 1935a, 1935b), and Kurt Lewin's finding s concerning how young children solve the problem of reaching a goal when th e direct path is blocked by a detour (Hull, 1938) . More generally, Hull' s habit–family hierarchies provided S –R behaviorism with a significant new theoretical tool for conceptualizing complex behavior. D. KENNETH SPENCE'S ANALYSI S OF TRANSPOSITION

One of the major challenges to the stimulus–response analysis of learnin g arose out of learning experiments that appeared to show that animals do not

5 CLARK L . HULL' S BEHAVIORISM

13 9

learn specific responses to specific stimuli in simultaneous discriminations but to relations between the stimuli . In this challenge, as well as in some others for which discrimination learning was the vehicle, Gestalt psychologists and Tolmanian behaviorists found themselves on the same side . For example, in an early study reported by Wolfgang Kohler (1939), chimpanzees were reinforced for responding to a light-gray but not to a darker-gray stimulus card . In a subsequent test the originally positive (light-gray) card was paired with a new lighter-gra y card, and the chimpanzees responded to the new card and not to the previousl y positive stimulus . Choice of the new card in this test suggested that during original training the animals had learned to respond to a relationship ("lighter than") , and not to the specific shades of gray. This phenomenon was termed transposition because it seemed analogous to what happens when music is transpose d from one key to another: All of the elements (notes) are changed but the overall relationship is preserved, and the music is still recognizable as the same . In the mid-1930s, Kenneth W. Spence published a series of classic papers o n discrimination learning . In one of these papers, an S —R analysis of relationa l learning was elaborated (Spence, 1937) . Spence pointed out that transpositio n does not always occur, and that "relational" theories of transposition, which emphasized the role of abstract relative judgments by animals, were not satisfactor y because they failed to specify when transposition would fail . Furthermore, simply saying that discrimination learning is relational does not provide a basis fo r the systematic experimental analysis of transposition . Spence (1936, 1937) pro posed an extension of his earlier theory, a kind of Pavlovian model in which discrimination learning was seen as reflecting the divergence of excitatory associative strength to the reinforced stimulus (S +) and inhibitory associative strengt h to the nonreinforced stimulus (S — ) . The great advantage of this "absolute" ove r a "relational" theory was that it made rather specific predictions about the outcomes of relatively complex discrimination learning experiments . In so doing it illustrated the power of the S —R approach in specifying variables for effectiv e experimental analysis . The details of Spence's theory are illustrated in terms of hypothetical discrimination tests among shades of gray . Spence's assumption was that gradients o f excitation and inhibition generalize along a continuum of gray stimuli as i s the hypothetical case after a discrimination has been formed between gray shades 8 (S +) and 11 (S —) on the basis of reinforcement of responses to S + and nonreinforcement of responses to S — . From the point of view of "relational " theory, the animals should learn the abstract rule that the "lighter than" of a pai r of gray stimuli on this dimension is the correct (positive) stimulus . From th e point of view of Spence's S —R analysis, however, the reinforcement of responses to S + yields increments in excitatory strength to that stimulus, while th e nonreinforcement of responses to S — yields increments in inhibitory strength t o that stimulus . The asymptotic levels of excitatory and inhibitory strengths to S + and S — are given by the heights of the appropriate curves directly above S + and S — , re-

140

RASHOTTE AND AMSE L

HIGH Strength of Association

♦

♦

LOW

1

(

I

\ 1

I

1

1

1

1

1

1

I

1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 1 8

Light

4 4 S+ S SHADE OF GREY

► Dark

FIGURE 5 .2 Hypothetical gradients of excitation (solid line) and inhibition (dotted line) afte r discriminative training with stimuli of two shades of gray . Responses to gray-shade 8 (S +) were reinforced : responses to gray-shade 11 were nonreinforced. The net excitatory strength at each shade of gray is the algebraic summation of the excitatory and inhibitory association strength at that shade . The net strength at selected stimuli is shown as a vertical arrow (after Spence, 1937) .

spectively. As Figure 5 .2 indicates, generalization gradients of excitation and inhibition to surrounding shades of gray were assumed by the theory, so that mos t stimuli on this continuum of brightness have both excitatory and inhibitor y strength . Spence's theory made additional assumptions : (1) that the net associative strength at any point on the continuum is the algebraic summation of the excitatory and inhibitory associative strengths (length of double-headed arrow) , and (2) that when confronted with any pair of stimuli the animal's response wil l be to the stimulus with the absolutely greater net excitatory strength . As indicated, the net excitatory strength to the positive training stimulus S + (shade 8) i s obviously greater than to S — (shade 11), and the animal will respond to S+ . If, in a transposition test, stimulus 5 is pitted against the former S + (stimulus 8) , the theory predicts responding in a relational manner : the net excitatory strength is greater at S 5 than at S 8 . The theory also predicts, however, that transposition will fail when, for example, S3 is tested against S 8 or S 5 . In these cases, responding should be to the stimulus closer or identical to the original positive training stimulus . At the time Spence's S —R account of transposition was proposed there wa s little, if any, experimental evidence about the nature of gradients of excitatio n and inhibition in discrimination learning, and there were no tests along the kind s of extended dimensions and hypothesized interactions of such gradients in discrimination learning, as they related to the phenomenon of transposition . The theory was attractive because it provided a set of mechanisms for the interpretation of transposition that were strictly deduced from a small number of assumptions ; however, the major prediction that transposition would break down i f one went far enough out on a dimension seemed very much against common sense . While Spence and his students performed several experiments that foun d evidence for the reversal of transposition, there was, of course, still considerabl e


14 1

debate about how best to interpret the phenomenon of transposition (e .g ., Ehrenfreund, 1952 ; Mackintosh, 1974) . In subsequent years, the ability to achiev e sharp, symmetrical generalization gradients in discrimination experiments with pigeons revived interest in the quantitative features of Spence's theorizing abou t gradients of excitation and inhibition and the assumption of algebraic summation (e .g ., Hearst, 1968) . In the 1930s the principal significance of Spence's theory was that it provided a further instance of how a variety of behavioral phenomena could be deduce d from a small number of assumptions compatible with S –R theorizing in the tradition of Thorndike and Pavlov. Spence's theoretical and experimental work on discrimination learning (e .g., Spence, 1938, 1940, 1942, 1952c, 1960a) was, an d has remained, the most effective counterbalance to cognitive-oriented interpretations of discrimination learning . More generally, Spence's brilliant theoretica l writings on this subject and others had an important influence on Hull's thinking, particularly on the more formal statements of Hullian theory in 1943 an d later (Amsel & Rashotte, 1984) . E . APPLICATION TO PSYCHOTHERAP Y AND SYMBOLIC PROCESSE S As we have noted, Hull's appointment to the Institute of Human Relations a t Yale University, in 1929, provided the opportunity for interactions with sociologists, cultural anthropologists, and other behavioral scientists . One result of these interactions took the form of attempts to employ principles of simple learning in the analysis of complex behavior. For example, Neal E . Miller and John Dollard published books titled Social Learning and Imitation (Miller & Dollard, 1941) and Personality and Psychotherapy (Dollard & Miller, 1950) , both of which drew heavily on the basic Hullian approach . In many respect s these books stand as earlier versions of Hull's intended third volume, which wa s of course never written . The opening paragraph of Personality and Psychotherapy conveys the spirit of this kind of undertaking : This book is an attempt to aid in the creation of a psychological base for a general scienc e of human behavior. Three great traditions, heretofore followed separately, are brought together. One of these is psychoanalysis, initiated by the genius of Freud and carried on by his many able students in the art of psychotherapy . Another stems from the work o f Pavlov, Thorndike, Hull, and a host of other experimentalists. They have applied the exactness of natural-science method to the study of the principles of learning . Finally, modem social science is crucial because it describes the social conditions under which huma n beings learn . The ultimate goal is to combine the vitality of psychoanalysis, the rigor o f the natural-science laboratory, and the facts of culture . We believe that a psychology o f this kind should occupy a fundamental position in the social sciences and humanities — making it unnecessary for each of them to invent its own special assumptions about human nature and personality . (Dollard & Miller, 1950, p . 3)

An important consequence of these efforts to apply Hullian learning principles to complex phenomena was that many of the basic concepts of Hullian the-

142

RASHOTTE AND AMSE L

ory had to be "liberalized" (Miller, 1959) . For example, Hull's (1943b) view tha t need reduction is the source of reinforcement for learned S –R association s proved overly restrictive since many instances of reinforcement (e .g ., a small food pellet) were not seen to be related to obvious reduction in bodily needs . Accordingly, for example, Miller and Dollard liberalized Hull's position and pro posed that the basis of reinforcement is the reduction of any strong stimulus (e .g ., the response of pulling down a window blind is reinforced by the elimination of glare) . Reinforcement can result from the termination of "arbitrary" stimuli that, through classical conditioning, come to play the same functional role a s do primary drives such as hunger and thirst the principle of "secondary reinforcement" around which considerable research had already been initiated . Hull's other tightly formulated principles were similarly strained in learning based analyses of such phenomena as imitation, copying, reasoning, sympto m learning, Freudian repression, and techniques of psychotherapy. Furthermore , the analysis of neurotic behavior patterns and psychotherapy was strongly influenced by Miller's brilliant analyses, from animal learning experiments, of the establishment, resolution, and displacement of conflicts (e .g ., Miller, 1944, 1959) . It is interesting that although Miller and Dollard's analysis of neurosis an d psychotherapy was influential in alerting practitioners to the importance o f learning processes, it never attracted a fervent band of adherents, dedicated t o revolutionizing the practice of psychotherapy. The person with a Hullian orientation who was, perhaps, most identified with this movement was Joseph Wolpe , with his early writings on reciprocal inhibition and the resultant development o f systematic desensitization, a conditioning therapy (e .g ., Wolpe, 1958) . It is not possible to elaborate here on the many early contributions that other s made to an analysis of complex behavior within the general tradition of Hullia n thought. We must, at least, call attention to the many influential contributions o f O . H . Mowrer to S –R analyses of neurotic behavior, psychotherapy, language , and symbolic processes (e .g ., Mowrer, 1939, 1940, 1953, 1960a, 1960b) ; to D . E . Berlyne's Hullian-oriented theoretical analyses of exploratory behavior , thinking, and aesthetic behavior (Berlyne, 1960, 1965, 1971) ; to Charles Os good's S –R analyses of language and symbolic processes (Osgood, 1953) ; and to J . G . Taylor's thoroughgoing account of perception from an S –R perspectiv e (Taylor, 1962) .

III . CONCEPTUALIZATIONS O F REINFORCEMENT AND REWARD : EVOLUTION OF HULL'S 1943 THEOR Y The term reward appears only once in the index of subjects in Hull's Principles. (The terms punishment and fear do not appear in the index at all .) The concept of reinforcement, of course, plays the major role . It is defined as the


14 3

diminution of a need (whether appetitive or aversive) that is necessary for the formation and strengthening of habit . Since the term reward was not used by Hull in 1943, it may seem odd that w e choose to trace its meaning from its neglect by Hull to its development by others who worked in Hull's tradition . The point is that the discussion that follows ad dresses itself to the concept of reward and not necessarily to reinforcement, an d this conceptualization of reward turns out to be very close to the meaning it ha d for Hull in 1943 "incentive ." A. "REINFORCEMENT" IN HULL'S 1943 THEOR Y

For the reader unfamiliar with the formal statements of Hull, we quote fro m Principles the major features of Postulate 4, to which we have already alluded : Whenever an effector activity (r —~ R) and a receptor activity (S — s) occur in close temporal contiguity ( :Cr), and this sCr is closely associated with the diminution of a need (G ) or with a stimulus which has been closely and consistently associated with the diminutio n of a need (G), there will result an increment to a tendency (OsHR) for that afferent impulse on later occasions to evoke that reaction. The increments from successive reinforcement s summate in a manner which yields a combined habit strength (sHR ) which is a simple positive growth function of the number of reinforcements (N) . (Hull, 1943b, p . 178 )

For Hull, behavior was not related directly to habit strength but to effectiv e excitatory potential, S E R (excitatory potential minus total inhibitory strength) . This feature of the theory is expressed in Postulate 7, which specifies that al l "primary drives" contribute to the total drive strength (D), which multiplies the current value of sHR to determine the strength of SER , and, consequently, the strength of responding in a given situation : generally, SER = f( SHR ) X f(D) : Any effective habit strength (sHR ) is sensitized into reaction potentiality (sER) by all primary drives active within an organism at a given time, the magnitude of this potentialit y being a product obtained by multiplying an increasing function of sHR by an increasin g function of D. (Hull, 1943b, p. 253 )

Hull was led to this conceptualization by a number of experimental findings , but particularly by the data of Perin (1942), which are shown in Figure 5 .3 as Hull portrayed them (p . 228) in Principles. Since neither magnitude nor delay of reinforcement was varied in Perin's experiment, the group differences at the end of acquisition were taken to reflect th e strength of sHR as a function of number of reinforcements . It is clear that Hull takes the relation between number of responses in extinction and number of reinforcements in acquisition to be a simple positive growth function, as woul d follow if sHR increased in the manner required by Postulate 4 . Further, the figur e shows that when drive is varied in extinction (reduced from the acquisition leve l to 3 hours for one group and left unchanged for the other), the curves reach different asymptotes as would be expected if SER was the product of D and SHR (Postulate 7) .

RASHOTTE AND AMSE L

14 4 60

22 HOURS HUNGE R

c

Z

50

H V z

840 w o O

a 30 -

O z

o F=

20 -

.

3 HOURS HUNGE R •

oC ce

m 10 •

D z 0

10

20

30 40 50 60 70 NUMBER OF REINFORCEMENT REPETITIONS (N)

80

90

FIGURE 5 .3 Number of unreinforced bar presses required by differently motivated groups o f rats to reach a criterion of extinction, plotted as a function of the number of times the bar press wa s reinforced with food during acquisition . (Hull assumed that, under certain circumstances, this response measure would reflect the strength of SER. )

B. DIFFICULTIES RECOGNIZED BY HULL IN 194 3

Hull obviously knew that the 1943 treatment of the Perin data was an over simplification or, at the very most, that this conceptualization would be useful only under certain restricted (ideal) experimental conditions . For example, Perin used reactions to extinction (n) as the dependent variable, but there were alread y in existence important data by L . G . Humphreys (1939a, 1939b), of which Hul l was aware (1943b), that restricted the use of resistance to extinction as a measure of habit strength to cases in which reinforcement is continuous : Humphrey s had already shown in a series of experiments that resistance to extinction i s greater following a relatively small number of reinforcements if those have occurred in the context of "partial reinforcement," the quasi-random intermixture of reinforcements and nonreinforcements . [Hull's own theoretical writings neve r dealt adequately with the so-called partial reinforcement extinction effec t (PREE), and it fell to others, such as Amsel (e .g ., 1958, 1962) and Capaldi (e .g . , 1966), to provide influential theoretical treatments of partial reinforcement, an d in Amsel's case, also discrimination learning, in the context of modified Hullian theory . ] A second limitation on the applicability of the theory was in Hull's treatment of response latency (or, inversely, speed) as the dependent variable . The 1943


14 5

theory specified that response latency declines as a negatively accelerated function of number of reinforcements (N), assuming a constant drive level (D) . Hul l recognized that this relationship can hold only when the promptness of reinforcement depends on the promptness of responding, as it does ordinarily in th e case where, for example, a rat runs from a start box through a runway to obtai n food at the goal . There are cases, however, in which reinforcement and the rapidity of the response are negatively correlated, such as when the rat must learn t o delay leaving the start box to obtain reward, or is reinforced for some "self-imposed" ritual of slow responding (actually "time-taking") . In cases of this sort, frequency of reinforcement increases as latency (time to respond) increases . Hul l recognized this as a further limitation of the theory's predictiveness . In later years, Frank A. Logan (1960), who worked with Hull on his last book, developed what he called a "micromolar" approach in which the concept of negativel y correlated reinforcement figured prominently . This approach further increase d the flexibility of theory in the Hullian tradition, particularly on the dependent variable side . The oversimplicity of Hull's habit-strength function, which asserted that habi t strength grew as a function of the number of reinforcements, S HR = f(N), was further questioned in our work (Amsel & Rashotte, 1969 ; Rashotte & Amsel, 1968) ; we showed that not only did rats learn idiosyncratic rituals to "take time " under one of Logan's negatively correlated reward conditions, but that these rituals emerged in extinction, even after prolonged intervening training on a simpl e continuous reinforcement schedule (CRF) in which the rituals were eliminated. C. THE ELLIOTT – CRESPI – ZEAMAN EXPERIMEN T Another difficulty for Hull's SH R = f(N) arose just before Principles of Behavior appeared in print . L . P. Crespi (1942), in a result partly anticipated b y Elliott (1928) and soon replicated by Zeaman (1949), reported two effects that were troublesome for Hull's concept of reinforcement. First, rats trained to ru n to a large reward abruptly slowed down to a level below that of a small-rewar d control group when the reward size was reduced (Crespi's "depression effect") ; and rats trained to a small reward abruptly increased their speed when rewar d size was increased ("elation effect") . The Elliott–Crespi–Zeaman findings posed three problems for Hull's 194 3 theory. First, because habit strength is a function of reward magnitude, larger reward is expected to yield a higher level of S HR to multiply D, and therefore a high level of performance (e .g ., running speed) . This much was found in the firs t phase of the Crespi –Zeaman experiments . However, because S HR is assumed t o represent a relatively permanent change in associative strength, decreases in reward size in the second phase of the experiments should have little immediat e effect on response strength (running speed) . Instead, Crespi and Zeaman reported pronounced decreases in speed after a single trial at the larger rewar d value, and the depressed running speed recovered to a new asymptotic level ap-

146

RASHOTTE AND AMSE L

propriate to the smaller reward only after four trials . The shift from small to large reward also produced immediate shifts toward higher asymptote than did a large reward from the outset . (This last finding has been harder to replicate.) The clear implication of these findings, in any case, was that magnitude-of-reinforcement (reward– size) manipulations did affect behavior, but not solely throug h SHR, as the 1943 theory required . Pressured by these findings and by critiques of the theory's handling of reinforcement, Hull revised his conceptualization of reinforcement in later (1951 , 1952a) versions of the postulates . The revised conceptualizations took magnitude and delay of reinforcement out of the determination of habit strength an d made S HR a function only of the number of reinforcements . The action of magnitude and delay of reinforcement were now expressed through two new theoretical factors, representing incentive motivation (symbolically, K and J) : K and J were now incentive–motivational factors in the sense that, added to D, they multiplied habit SHR to determine excitatory potential SE R. The split-off of associative–reinforcing function from motivational–incentive function was now complete . The "law" of habit formation, S HR = f(N), ex pressed habit strength to be a function only of the number of reinforcements (N) . Reaction potential (S ER) was still determined by S HR along with drive strength (D), but now added to D, as undifferentiated incentive motivational factors, wer e K and J, and new computational formulas were provided for determining the values of K and J and SER (Hull, 1951) . Hull's new conceptualization of reaction potential took care of most aspect s of the Crespi – Zeaman experiment . For example, his new equations for K (Postulate 7) made it possible for a change in reward size (either upward or down ward, even if it occurred only once) to cause the value of K to change substantially toward the value required by the new reward size . And because th e changed values of K and J entered multiplicatively into the equation for S E R, a changed reward value (K) or delay in reward (J) could cause behavior such as response speed in a runway to move to a new asymptotic level appropriate to tha t reward in very few trials . The revised theory did not handle the overshooting (elation) and undershooting (depression) effects Crespi identified in shifts from small to large reward an d large to small reward, respectively . Crespi held that these effects were emotiona l and should disappear with continued training . The work on successive negative contrast (SNC) shows clearly that this is the case in the shift from large to small reward size (see Amsel, 1992, for a review) . Hull accepted this analysis but di d not feature it in his theorizing because he was uncertain about the generality o f the phenomena in question . Hull's own discussion of reward-shift in A Behavio r System (1952a) centered on the latent-learning experiment . The classic latent learning experiment is one in which animals are run through a maze for severa l trials without reward, and are then shifted to a series of large rewards (e .g ., Blodgett, 1929) . In some of these experiments (Tolman & Honzik, 1930) a downward shift in reward is also included . In these experiments, as in those of Crespi and


14 7

Zeaman, upward shifts are followed by abrupt improvement in performance (reduction in errors in the maze) and downward shifts result in increased errors . However, as Hull observed, "overshooting" is found in neither case . The entire history of reward-shift experiments and the theoretical treatment they receive d by Hull and by many other theorists has recently been reviewed in great detai l (Flaherty, 1996) . D. SPENCE'S CONTRIBUTIONS TO REWARD–INCENTIVE THEORY An important contributor to Hull's formal theorizing from the beginning, Kenneth W. Spence continued the development of what came to be calle d Hull–Spence theory in important ways . A significant part of the empirical an d theoretical work of Spence and his students at Iowa was directed at motive–incentive factors in behavior, and this emphasis was reflected in Spence's Sillima n Lectures at Yale University, published as the book Behavior Theory and Conditioning (1956) . In Chapter 5 of this book, Spence's treatment of "Reinforcemen t in Instrumental Learning" is primarily a discussion of how the theory must be adapted to handle new data on magnitude and delay of reward. It was characteristic of Spence to let vigorous experimental analysis guide th e development of his theorizing, and the problem of magnitude of reward in instrumental learning is a good example of that style . In 1956 Spence reported several experiments, conducted in his laboratory, which appear to replicate all aspects of the Crespi–Zeaman experiments except "overshooting" of th e asymptote for large reward following a shift from small to large reward (Spence , 1956) . The problem for Spence was that the number of trials before the shif t from small to large magnitude of reward in the Crespi–Zeaman experiments wa s so limited that running speed had not reached asymptote in the preshift phase . Consequently, the apparent "elation effect" following the shift seemed attributable to the uninteresting fact that speed was simply approaching its true asymptote . A period of intensive experimental work on this problem left Spence with a firmer empirical base for the theoretical treatment of reward magnitude . There remained three facts to be accounted for in a theory of the action of rewards : (a) rate and asymptote of responding are higher for large reward than for small re ward ; (b) performance level shifts quickly following increases and decreases i n reward size ; and (c) the Crespi–Zeaman "depression effect" occurs response speed falls below the low-reward asymptote following a shift from high to lo w reward. At this point, there is an obvious return to the spirit of Hull's writings in th e 1930s where rG –sG played such a central theoretical role . Indeed Spence had prepared the reader for his view of reward–incentive effects in Chapter 2 of Behavior Theory and Conditioning (see Figure 5 .4) : Spence's position, like Hull's earlier one, was that in every case of instrumental learning there is an inherent classical conditioning component . Further, ac cording to the Pavlovian principle of stimulus substitution, different reinforcers,

B. SA

SA

rg• $g ___ ..

rg-Sg - --* Rwc

SG

RG

F 1 G U R E 5 .4 Diagrammatic representation of the conditioning processes involved in instrumental learning. A. Early in training. This diagram shows how classical conditioning of the consummatory response (R G), represented by arrows marked II, is an integral part of instrumental learning . The arrows marked I represent the instrumental conditioning of the locomoting response. B. Later in training. The stimulus cues of the alley (S A , S A ') are assumed to evoke an anticipatory, fractiona l portion of the consummatory response, denoted rg . The interoceptive cue (s g) produced by rg , in turn, is assumed to acquire a habit loading for the instrumental locomoting response .

such as food or water, should produce qualitatively different R G S, and therefore different conditioned r gs . As in Hull's classic conceptualization, r g is assumed to produce distinct feedback stimulation, symbolized s g , and the mechanism is designated rg s g . What is new in this portion of Spence's theorizing is that h e gives rg motivational properties, something Hull had not done earlier, while retaining in sg the major associative mechanism of expectation that Hull had conceptualized 25 years earlier. More formally, Spence (1956) proposes that incentive motivation, K, represents, quantitatively, the strength of rg —sg . In this way, rg — sg is tied directly to one of the theoretical variables that determine the valu e of SER (henceforth E, as Spence omits the subscripts S and R) . Further, Spence' s account restricts the effects of reward in instrumental appetitive learning to K and to the associative properties of s g. The growth of H (sHR ) is determined simply by contiguity, the number of S —R conjunctions . We can recognize this position as the ultimate form of an evolution of thinking in which positive reinforce-

'


14 9

ment (or reward) plays an ever-decreasing role as an associative variable, as i t did under Hull's strong reinforcement principle of 1943 . In recognition of th e essential similarity of D and K as general motivational factors, and as a result o f other experiments, Spence held D and K to be additive [E = H X (D + K)], not multiplicative, in determining excitatory potential . In the revised postulate s (1952a) Hull's formula had been SER = SH R X D X K. It remained for Spence to account for the depression effect, that respons e strength falls below the low-reward asymptote following a shift from large t o small reward . He elected, as Crespi had before, to attribute this "undershooting " of the asymptote to an emotional reaction, "frustration," and he regarded frustration as inhibitory . There was no specific mechanism for the relation of frustration to inhibition until some years later when Spence (1960b) endorsed the theoretical view of his student Amsel (1958), who proposed such a mechanism in what is now known as Frustration Theory. Spence's treatment of the role of reward in instrumental learning stimulated a number of experiments, because it identified several experimental operation s that should influence performance through K . In particular, the theoretical strategy of specifying r g –s g3 as the basis of K encouraged tests of the theory in whic h explicit classical conditioning manipulations were made on the baseline of instrumental learning experiments . For example, Spence (1956) reported experiments that attempted to vary strength of conditioning of r g to stimuli of the S A variety by manipulating the time the rat was allowed to eat in the goal box . Although experiments such as these suggest that the concept of r g –s g as the basis for K-value stimulated research, it was often impossible to employ th e logic of the approach as the basis for unequivocal predictions about instrumenta l responding . The problem lay in the lack of precise understanding of the basis for conditioning of r g –sg : the runway was conceptualized as a sequence of exteroceptive stimuli preceding the goal-box stimulus and food ; the theory assumed that the goal-box stimuli were conditioned to r g because of their close temporal proximity to the food US, and that stimuli earlier in the runway evoked r g through generalization. Accordingly, the strength of r g –sg at any point in the runway should be determined by the similarity between runway cues at that poin t and the goal-box stimuli . In Spence's view, these assumptions were tentative an d needed refinement. For one thing, as Spence pointed out, proprioceptive as wel l as exteroceptive cues would surely also affect the quantitative details of generalization of rG –S G across the runway . Another difficulty, which became more an d more apparent in the sixties and later, arose from the growing evidence fro m studies of successive-compound conditioning that processes far more comple x than generalization influence the strength of conditioning to stimuli in a sequence preceding a US (e .g ., Baker, 1968 ; Egger & Miller, 1962 ; Holland, 1992 ; Razran, 1971 ; Rudenko, 1974 ; Wickens, Born, & Wickens, 1963) . Additiona l Subscripts are sometimes lowercase (g or f) and sometimes uppercase (G or F) . This follows the author's usage in each instance .

150

RASHOTTE AND AMSE L

difficulties with the r g –s g analysis had also been presented by Logan, in his boo k Incentive (1960) . E. AMSEL'S FRUSTRATION THEORY: A CONDITIONING MODEL O F EFFECTS OF REWARD AND NONREWAR D As we have noted, both Hull and Spence accepted the view that "frustration " (or some other emotional factor) accounted for the depression of running spee d below a small-reward asymptote when magnitude of reward was shifted from large to small, but neither theorist provided a detailed account of how frustratio n might enter into the general structure of the theory . Other researchers had als o reported signs of "emotional" upset in animals at the beginning of extinction , and some had suggested that emotion should be included in making prediction s about the strength of responding in extinction (e .g ., Skinner, 1950) . As early as 1951, at a meeting of the Southern Society for Philosophy and Psychology, Amsel proposed a conceptualization of the role of anticipatory frustration, in the context of the rG –s G mechanism, as a third factor to be added t o Hull's two-factor theory of inhibition . In this paper Amsel was extending Hull' s two-factor theory of inhibition and proposing a third factor, anticipatory frustration (rF –s F ), as important in understanding a variety of nonreinforcement-relate d inhibitory effects . He addressed himself specifically to the Crespi–Zeaman effect, which we have already discussed, but included a discussion of other "paradoxical" effects that occur when rewards and nonrewards are intermixed i n learning . In the same year, J . S . Brown and I . E. Farber (1951) advanced a theor y of frustration, also in a Hullian framework . Their elegant, but complex, theory related frustration to conflicting response tendencies, as well as to the operation of nonreinforcement and thwarting . The theory was appended to a more genera l conceptualization of emotions as intervening variables, and it is, perhaps, for thi s reason that it did not have the impact on experimental work it might otherwis e have had. Amsel subsequently elaborated his views in a series of empirical and theoretical papers (Amsel, 1958, 1962, 1967) that included Amsel and Roussel's (1952 ) demonstration of the "Frustration Effect" (FE), an account of the primary or energizing (D-like) effects of frustration . The whole approach came to be known a s Frustration Theory. It accounted not only for the primary (activating) effects of frustration resulting when an expected reward does not materialize, but also fo r the partial reinforcement extinction effect (PREE) and, ultimately, for a family of other paradoxical effects of reward schedules in both discriminative an d nondiscriminative instrumental learning (see Amsel, 1986, 1992) . In informal terms, Amsel's theory assumed that when nonreward, reduced reward, or delayed reward occurs in place of an expected reward, the animal experiences a temporary aversive motivational state, primary frustration . With the conditions that produce primary frustration specified, the theory outlined thre e properties of frustration . The first, primary frustration (R F ), is a hypothetical un-

'


15 1

conditioned reaction to the frustrating event . The theory specifies that RF will exert a transient motivational (energizing) effect on responses with which it coincides, increasing, particularly, the vigor with which these responses are per formed . The behavioral manifestation of such a consequence of R F (the FE) is an effect that has been documented by many (perhaps hundreds of) empirical studies, most of them employing some variant of the double-runway apparatus firs t used by Amsel and Roussel (see Amsel, 1992, for a more extended account) . The second property of frustration is the frustration drive stimulus (S F), a feedback stimulus from primary frustration that acts to cue, guide, and direct behavior (i .e., it acts associatively ; Amsel & Prouty, 1959 ; Amsel & Ward, 1954) . The third property of frustration, and the one that was actually proposed first, an d that bears perhaps the heaviest theoretical burden, is conditioned (anticipatory) frustration (rF –s F ) . This factor refers to the manner in which frustration influences responses that precede the frustrating event, and here the theory relies o n the logic derived from its Pavlovian antecedents : With repeated occurrences, stimuli (conditioned stimuli, or CSs) accompanying primary frustration (R F) come to evoke a classically conditioned form of R F, designated rF –s F . As with rG –s G , of which it is an aversive counterpart, rF - SF is evoked first by stimuli in the region of the goal event, and later moves forward in time into the instrumental response . This conditioned response (rF –sF) is assumed to increase in strength as a function of trials, reaching an asymptotic strength appropriate to th e strength of RF . The role of r F –sF in instrumental responding was specified most clearly in later versions of the theory that emphasize that its function varies with its strength (Amsel, 1967) . While rF contributes to incentive motivation, K, at al l values of its strength, it provides an aversive guiding stimulus (5 F ) only at value s above some threshold . At weak intensities, therefore, the role of rF is only an invigorating one . As rF becomes stronger, however, SF becomes an aversive stimulus and evokes unconditioned and conditioned responses that are antagonistic t o the referent instrumental response . Obviously, such competing responses evoke d by rF –s F provide a plausible account Amsel provided this account in his 195 1 paper of depression of running speed below the small-reward asymptote following a shift from large to small reward in the Crespi – Zeaman experiment . Spence might have used such a mechanism in 1956 as the basis for his inhibitio n (I), which, it will be recalled, represented the effects of competing responses an d subtracted from E to weaken performance. He did later adopt rF –s F as the basi s for I (Spence, 1960b) . Perhaps the most referenced portion of Amsel's theory concerns the influence of rF –s F when the instrumental response continues to be performed in its presence. There are two cases, depicted in Figure 5 .5 . The first is the partial reinforcement experiment in which reward for an instrumental response is restricte d to a random percentage (usually half) of trials . The four-stage hypothesis (Amsel, 1958) outlines a sequence of events leading to the development of persistence . The second case concerns the presence of differential external stimuli (S +

152

RASHOTTE AND AMSE L

PARTIAL REWARD REWARD TRIALS SA

Z Cr

Q

S

sr' R App

O

(I)

SA

SR

`

3

SNon

RA Pp

RR

R

R

.

SNo, elf ``~

• rR st RApp

F F AV SA Conflict rR - sR-p

r - s --_♦R

(1)

Q

A

2

J

SA

SR

RAN,

DISCRIMINATION LEARNIN G

NONREWARD TRIALS REWARD TRIALS

RA p

R

SR

RNon R

RApp

R

7

RF

NONREWARD TRIALS

, s rR

S Non R

RNon R

RAOO

RR

SNon

sl App R

rR

r

F; 'F

R

may

sbRApp RF

R AV

Conflict r

R - sR

°

Ap

R p

F

aRAV

R s R

4 RAp

rFw s

r

p

FIGURE 5 .5 Diagrammed sequence of hypotheses relating frustrative nonreward to stages o f partial reward and discrimination learning . (adapted from Amsel, 1958) .

and S —), and the role of r F — sF as a mediating mechanism in the formation of si multaneous or successive-trial discrimination learning . The assumption that r F —sF becomes conditioned to the instrumental respons e has provided theoretical leverage in topic areas beyond the scope of Hull's original theory, and indeed beyond the problem areas to which frustration theory wa s initially applied (see Amsel, 1992) . The conditioning model theory of Amsel ha s integrated the explanations of a number of phenomena besides the invigorating effect of primary frustration (the FE), the Crespi — Zeaman effect, and the partia l reinforcement extinction effect (PREE) . The theory was extended in 1962 to ac count for a variety of phenomena (see Appendix, Amsel, 1992, pp . 233ff) : emotional—motivational factors in discrimination learning ; the retention and durabil ity of persistence ; the overlearning —extinction effect ; the overlearning —reversal effect in discrimination learning ; the phenomenon of subzero extinction (a finding analogous to the depression effect) ; and the paradoxical Haggard—Goodric h partial reinforcement acquisition effect (PRAE), in which partially rewarded ani mals respond more vigorously in early stages of the instrumental response-chai n in acquisition than do continuously rewarded animals . Also accounted for are th e action of certain drugs such as alcohol and sodium amobarbital to attenuate th e PRAE and the PREE ; the phenomenon of behavioral contrast ; the appearance o f certain "adjunctive" behaviors including schedule-induced polydipsia in Skinnerian—operant experiments ; the role of the limbic system, particularly the role


15 3

of the amygdala and the hippocampus, in short-term memory–based learning i n certain "paradoxical" phenomena such as the PREE ; transfer of persistenc e across situations and motivational–reinforcement conditions, including the effects of early experience on later behavior ; and the phenomenon of "regression" : return to earlier successful modes of behavior. Amsel (1972, 1992) extended the theory to provide a more general account of persistence, viewing partial reinforcement as only one instance of a more general case . This extended theory relies heavily on the proposal that stimuli that initially disrupt instrumental responding (e .g ., stimuli to which subjects "habituate") become counterconditioned to the evocation of the instrumental respondin g they initially disrupt, with the consequence that instrumental responding will, in the future, be relatively unaffected by these and other frustrative or distracting stimuli . This general theory of persistence stems from the work of Amsel and hi s students and others (e .g ., Terris, German, & Enzie, 1969 ; Wagner, 1966) on transfer of persistence . In the last 20 years, this and other work has led, quite naturally, to an increasing interest by Amsel and his students in the study not only of the PREE, but also persistence and a number of other behavioral effect s based on a variety of reward schedules, all of this from a strong ontogenetic –developmental perspective, but also with a behavioral neuroscience emphasis, particularly on hippocampal function (see Amsel, 1986, 1992, 1994) . The material reviewed in the preceding subsections provides one illustration of how behavior theory of the Hullian variety initiated and responded to empirical findings . In this case, where some important changes in the conceptualizatio n of reinforcement, reward, and nonreward in Hullian theory are reviewed fro m 1943 onward, we can see that Hullian behaviorism has been strongly disposed t o refine its initial formulations and to develop new ones within the S –R context a s the empirical evidence requires . CONCLUDING COMMENT S Hull's S –R behaviorism was a major addition to the intellectual achievements of psychology during the twentieth century . Our present review of selected aspects of his efforts characterizes some important features of his approach . However, a firsthand reading of his theoretical papers in th e Psychological Review (Amsel & Rashotte, 1984) and of his classic books such as Principles of Behavior (Hull, 1943b) and A Behavior System (Hull, 1952a) remains the best way to grasp his thinking and the possibilities it raises for behavior theory even today. What is the enduring value of Clark L . Hull's work? We addressed this issue several years ago in a chapter, titled "Postscript : The Decline and Lasting Influence of Clark Hull," which followed our extended commentary on his 21 Psychological Review papers (Amsel & Rashotte, 1984) . Our comments at that time remain apt. Therefore, we conclude with an excerpt from our earlier postscript:

154

RASHOTTE AND AMSE L

What is most important in Hull's surviving influence is perhaps the scientific style an d tradition : the willingness to provide explicit and falsifiable predictions from a set of hypotheses derived from the existing literature ; the enormous body of experimental results , certainly not always confirmatory, these hypotheses have generated ; the economy of a family of interrelated constructs, empirically derived ; the recognition in the context o f general lawfulness that if explanations are as complex (have as many degrees of freedom ) as the phenomena they seek to explain, they offer no explanatory advantage ; the emphasis not only on the purely associative but also the motivational processes (the latter almost completely neglected since the return of cognitive structuralism in the 1960s) ; the hope for and attempt at quantification in learning theory, still to this day a science of ordina l differences and interactions . If not the content—and much of this certainly survives — then the approach, the strategy, and the example of dedication are identifiable aspects o f Hull's legacy. (Amsel & Rashotte, 1984, p . 507)

ACKNOWLEDGMENT S We express our deep thanks to Anna Tapsak, at the University of Texas, Austin, who provided u s with excellent and timely technical assistance in preparing this chapter. We are indebted to Ruth Hull Low and Janet Taylor Spence for permission to quote from private correspondence written by Clark Hull and Kenneth Spence, respectively . We thank Kindler Verlag for permission to incorporate adapted portions of an earlier chapter w e wrote that appeared in a German language version : Amsel, A, & Rashotte, M . E . (1977). A perspective on S –R learning theory in America with particular reference to Clark L . Hull, hi s precursors, and his followers . In Zeier, H . (Ed .). The psychology of the 20th century: Vol. 4 . Pavlov and his followers : From Classical conditioning to behavioral therapy. Zurich : Kindler Verlag (i n German) .

REFERENCE S Amsel, A . (1958) . The role of frustrative nonreward in noncontinuous reward situations . Psychological Bulletin, 55, 102–119 . Amsel, A. (1962) . Frustrative nonreward in partial reinforcement and discrimination learning: Some recent history and a theoretical extension . Psychological Review, 69, 306–328 . Amsel, A . (1967) . Partial reinforcement effects on vigor and persistence. In K . W. Spence & J. T. Spence (Eds .), The psychology of learning and motivation (Vol . 1, pp. 1–65) . New York: Academic Press . Amsel, A. (1972) . Behavioral habituation, counterconditioning, and a general theory of persistence. In A. H . Black & W. F. Prokasy (Eds .), Classical conditioning II: Current research and theory (pp. 409–426) New York : Appleton-Century-Crofts . Amsel, A. (1986) . Developmental psychobiology and behaviour theory : Reciprocating influences (Daniel E . Berlyne memorial lecture) . Canadian Journal of Psychology, 40, 311–342. Amsel, A . (1992). Frustration theory: An analysis of dispositional learning and memory . Cambridge, England : Cambridge University Press. Amsel, A . (1994) . Precis of Frustration theory : An analysis of dispositional learning and memory, Cambridge University Press, 1992 . Psychonomic Bulletin and Review, 1, 280–296 . Amsel, A., & Prouty, D. (1959) . Frustrative factors in selective learning with reward and nonreward as discriminanda . Journal of Experimental Psychology, 57, 224-230 . Amsel, A ., & Rashotte, M . E. (1969) . Transfer of experimenter-imposed slow-response patterns to


15 5

the extinction of a continuously rewarded response . Journal of Comparative and Physiological Psychology, 69, 185-189 . Amsel, A ., & Rashotte, M . E . (1984) . Mechanisms of adaptive behavior: Clark L. Hull's theoretical papers, with commentary. New York: Columbia University Press . Amsel, A ., & Roussel, J . (1952). Motivational properties of frustration : I . Effect on a running response of the addition of frustration to the motivational complex . Journal of Experimental Psychology, 43, 363-368 . Amsel, A ., & Ward, J . S . (1954) . Motivational properties of frustration : II. Frustration drive stimulus and frustration reduction in selective learning . Journal of Experimental Psychology, 48, 37-47 . Baernstein, H. D ., & Hull, C . L . (1931) . A mechanical model of the conditioned reflex . Journal of General Psychology, 5, 99-106 . Baker, T. W. (1968) Properties of compound conditioned stimuli and their components . Psychological Bulletin, 70, 611-625 . Bergman, G ., & Spence, K . W. (1944). Operationism and theory in psychology . Psychological Review, 51, 47-68 . Berlyne, D . E . (1960) . Conflict arousal and curiosity. New York: McGraw-Hill . Berlyne, D . E . (1965) . Structure and direction in thinking. New York : John Wiley & Sons . Berlyne, D . E . (1971) . Aesthetics and psychobiology. New York: Appleton-Century-Crofts . Blodgett, H. C . (1929) . The effect of the introduction of reward upon the maze performance of rats . University of California Publications in Psychology, 4, 113-134 . Breland, K ., & Breland, M. (1961). The misbehavior of organisms. American Psychologist, 16, 681-684 . Brown, J . S ., & Farber, I . E . (1951) . Emotions conceptualized as intervening variables—with suggestions toward a theory of frustration . Psychological Bulletin, 48, 465-495 . Capaldi, E . J. (1966) . Partial reinforcement: A hypothesis of sequential effects . Psychological Review, 73, 459-477 . Colwill, R . M . & Rescorla, R. A . (1986) . Associative structures in instrumental learning . In G . H . Bower (Ed.), The psychology of learning and motivation (Vol . 20, pp . 55-104). New York : Academic Press . Crespi, L . P. (1942) . Quantitative variation of incentive and performance in the white rat . American Journal of Psychology, 55, 467-517 . Dollard, J., & Miller, N . E. (1950) . Personality and psychotherapy. New York : McGraw-Hill . Egger, M. D ., & Miller, N . E . (1962) . Secondary reinforcement in rats as a function of informatio n value and reliability of the stimulus . Journal of Experimental Psychology, 64, 97-104. Ehrenfreund, D . (1952) . A study of the transposition gradient . Journal of Experimental Psychology, 43, 81-87 . Elliott, M. H. (1928) . The effect of change of reward on the maze performance of rats . University of California Publications in Psychology, 4, 19-30 . Flaherty, C . E (1996) . Incentive relativity. Cambridge, England : Cambridge University Press . Hearst, E . (1968) . Discrimination learning as the summation of excitation and inhibition . Science, 162, 1303-1306 . Holland, P. C . (1992). Occasion setting in Pavlovian conditioning . In D . Medin (Ed.), The psychology of learning and motivation (Vol . 28, pp . 69-125) . San Diego, CA : Academic Press . Hull, C . L. (1925a) . An automatic correlation calculating machine. Journal of the American Statistical Association, 20, 522-531 . Hull, C . L. (1925b). An automatic machine for making multiple aptitude forecasts . Journal of Educational Psychology, 26, 593-598 . Hull, C . L . (1929) . A functional interpretation of the conditioned reflex . Psychological Review, 36, 498-511 . Hull, C . L. (1930) . Knowledge and purpose as habit mechanisms . Psychological Review, 37, 511-525 . Hull, C . L . (1931). Goal attraction and directing ideas conceived as habit phenomena . Psychological Review, 38, 487-506.

156

RASHOTTE AND AMSE L

Hull, C . L. (1934a) . The concept of the habit-family hierarchy and maze learning : Part 1 . Psychological Review, 41, 33–54. Hull, C . L. (1934b). The concept of the habit-family hierarchy and maze learning : Part 2 . Psychological Review, 41, 134–152. Hull, C . L . (1935a) . The conflicting psychologies of learning—a way out. Psychological Review, 42, 491–516 . Hull, C . L. (1935b) . The mechanism of the assembly of behavior segments in novel combination s suitable for problem solution . Psychological Review, 42, 219–245 . Hull, C . L. (1938) . The goal gradient hypothesis applied to some "field force" problems in youn g children. Psychological Review, 45, 271–299 . Hull, C . L . (1943a) . A postscript concerning intervening variables. Psychological Review, 50, 540 . Hull, C. L . (1943b) . Principles of behavior: An introduction to behavior theory. New York : Appleton-Century-Crofts . Hull, C . L. (1943c) . The problem in intervening variables in molar behavior theory . Psychologica l Review, 50, 273–291 . Hull, C . L . (1951) . Essentials of behavior. New Haven, CT : Yale University Press. Hull, C . L . (1952a) . A behavior system: An introduction to behavior theory concerning the individua l organism. New Haven, CT: Yale University Press . Hull, C . L . (1952b) . Autobiography. In E . G. Boring, H . S . Langfeld, H . Werner, & R. M . Yerke s (Eds .), A history of psychology in autobiography (Vol . 4, pp . 143–162). Worcester, MA: Clark University Press . Hull, C . L. (1962) . Psychology of the scientist : IV. Passages from the Ìdea Books' of Clark L . Hull. Perceptual and Motor Skills, 15, 807 – 882 (monograph supplement) . Hull, C . L ., & Baernstein, H . D. (1929) . A mechanical parallel to the conditioned reflex. Science, 70, 14–25 . Hull, C . L., Hovland, C . I ., Ross, R . T., Hall, M., Perkins, D. T., & Fitch, F. B . (1940) . Mathematicodeductive theory of rote learning. New Haven, CT: Yale University Press. Humphreys, L . G . (1939a) . Acquisition and extinction of verbal expectation in a situation analogou s to conditioning. Journal of Experimental Psychology, 25, 294–301 . Humphreys, L . G . (1939b) . The effect of random alternation of reinforcement on the acquisition and extinction of conditioned eyelid reactions . Journal of Experimental Psychology, 25, 141–158 . Koch, S . (1954) . C. L . Hull . In : W. K. Estes et al. (Eds .), Modern learning theory (pp. 1–176) . New York: Appleton-Century-Crofts . Kohler, W. (1939). Simple structural functions in the chimpanzee and in the chicken . In E . D . Elli s (Ed.), A source book of Gestalt psychology (pp . 217–227). New York : Harcourt, Brace. (Translation ; originally published in German in 1918 : Aus der Anthropoidenstation auf Teneriffa . IV. Nachweis einfacher Strukturenfunktionen beim Schimpansen and beim Haushuhn : Ober eine neue Methode zur Untersuchung des bunten Farbensystems . Abh . Preuss . Akad . Wiss., Berlin, pp . 1–101 .) Krueger, R . G., & Hull, C . L . (1931) . An electro-chemical parallel to the conditioned reflex . Journal of General Psychology, 5, 262–269 . Logan, F. A . (1959) . The Hull-Spence approach . In S . Koch (Ed .), Psychology: A study of a science (Vol. 2, pp. 293–358) . New York : McGraw-Hill. Logan, F. A. (1960) . Incentive : How the conditions of reinforcement affect the performance of rats. New Haven, CT: Yale University Press . Mackintosh, N. J. (1974) . The psychology of animal learning. New York : Academic Press. Miller, N . E. (1935) . A reply to "Sign-gestalt or conditioned reflex? " Psychological Review, 42, 280–292 . Miller, N . E. (1944) . Experimental studies of conflict . In J . M . Hunt (Ed.), Personality and th e behavior disorders (pp . 431–465) New York: Ronald Press . Miller, N. E . (1959) . Liberalization of basic S-R concepts : Extensions to conflict behavior, motivation, and social learning. In S . Koch (Ed .), Psychology: A study of a science (Vol. 2, pp . 196–292) . New York: McGraw-Hill .


157

Miller, N . E., & Dollard, J . (1941) . Social learning and imitation . New Haven, CT: Yale University Press . Mowrer, O . H. (1939) . A stimulus-response analysis of anxiety and its role as a reinforcing agent . Psychological Review, 46, 553-566. Mowrer, O. H . (1940). Experimental analogue to "regression" with incidental observation on "reaction formation ." Journal of Abnormal and Social Psychology, 35, 56-87 . Mowrer, O . H . (1953) . Motivation and neurosis . Current Theory and Research in Motivation (pp . 162-185) . Lincoln: University of Nebraska Press . Mowrer, O . H . (1960a) . Learning theory and behavior. New York: John Wiley & Sons . Mowrer, O . H . (1960b). Learning theory and the symbolic processes . New York: John Wiley & Sons . Osgood, C. E . (1953) . Method and theory in experimental psychology . New York: Oxford University Press . Pavlov, I . P. (1927) . Conditioned reflexes : An investigation of the physiological activity of th e cerebral cortex. (G . V. Anrep, Trans .) . Oxford, England : Oxford University Press. Perin, C . T. (1942) . Behavior potentiality as a joint function of the amount of training and the degre e of hunger at the time of extinction. Journal of Experimental Psychology, 30, 93-113 . Rashotte, M . E ., & Amsel, A . (1968). Transfer of slow-response rituals to the extinction of a continuously rewarded response . Journal of Comparative and Physiological Psychology, 66, 432-443 . Razran, G . (1971) . Mind in evolution : An east-west synthesis of learned behavior and cognition . Boston : Houghton-Mifflin. Rudenko, L. P. (1974). On the functional structure of conditioned reflexes to serial stimuli in dogs . Acta Neurobiologica Experimentalis, 34, 69-79 . Skinner, B . F. (1950) . Are theories of learning necessary? Psychological Review, 57, 193-216. Skinner, B . F. (1977) . Herrnstein and the evolution of behaviorism . American Psychologist, 32, 1006-1012 . Spence, K . W. (1936) . The nature of discrimination learning in animals . Psychological Review, 43, 427 -449 . Spence, K. W. (1937) . The differential response in animals to stimuli varying within a singl e dimension. Psychologial Review, 44, 430-444. Spence, K . W. (1938) . Gradual versus sudden solution of discrimination problems by chimpanzees . Journal of Comparative Psychology, 25, 213-224 . Spence, K. W. (1940). Continuous versus non-continuous interpretations of discrimination learning . Psychological Review, 47, 271-288 . Spence, K . W. (1942) . The basis of solution by chimpanzees of the intermediate size problem . Journal of Experimental Psychology, 31, 257-271 . Spence, K. W. (1948) . The postulates and methods of behaviorism. Psychological Review, 55, 67-78 . Spence, K . W. (1952a) . Clark Leonard Hull, 1884-1952 . American Journal of Psychology, 65, 639-646 . Spence, K . W. (1952b) . Mathematical formulations of learning phenomena. Psychological Review, 59, 152-160 . Spence, K. W. (1952c) . The nature of response in discrimination learning. Psychological Review, 59, 89-93 . Spence, K . W. (1956) . Behavior theory and conditioning . New Haven, CT : Yale University Press . Spence, K. W. (1957) . The empirical basis and theoretical structure of psychology, Philosophy of Science, 24, 97-108 . Spence, K . W. (1960a) . Conceptual models of spatial and non-spatial selective learning . In : K. W. Spence, Behavior theory and learning : Selected papers (pp . 366-392). Englewood Cliffs, NJ : Prentice-Hall . Spence, K. W. (1960b) . The roles of reinforcement and non-reinforcement in simple learning . In K . W. Spence, Behavior theory and learning : Selected papers (pp . 91-112). Englewood Cliffs, NJ : Prentice-Hall .

158

RASHOTTE AND AMSE L

Taylor, J . G. (1962) . The behavioral basis of perception . New Haven, CT: Yale University Press . Terris, W., German, D ., & Enzie, R . (1969) . Transitional resistance to the effects of aversive stimulation. Journal of Comparative and Physiological Psychology, 67, 264-268 . Thorndike, E . L. (1898) . Animal intelligence : An experimental study of the associative processes i n animals . Psychological Monographs, 2, no . 4 . Thorndike, E. L. (1911) . Animal intelligence : Experimental studies. New York : Macmillan. Tolman, E . C . (1932) . Purposive behavior in animals and men. Berkeley and Los Angeles : University of California Press . Tolman, E. C . (1933) . Sign-gestalt or conditioned reflex? Psychological Review, 40, 246-255 . Tolman, E . C . (1938). The determiners of behavior at a choice point . Psychological Review, 45, 1-41 . Tolman, E . C ., & Honzik, C . H. (1930) . Introduction and removal of reward and maze performanc e in rats . University of California Publications in Psychology, 4, 257-275 . Wagner, A . R. (1966) . Frustration and punishment. In R. N . Haber (Ed.), Current research in motivation (pp. 229-239) . New York : Holt, Rinehart & Winston. Wickens, D. D., Born, D . G., & Wickens, C . D . (1963) . Response strength to a compound conditioned stimulus as a function of the element interstimulus interval . Journal of Comparative and Physiological Psychology, 56, 727-731 . Wolpe, J . (1958) . Psychotherapy by reciprocal inhibition . Stanford, CA: Stanford University Press. Zeaman, D . (1949). Response latency as a function of the amount of reinforcement . Journal of Experimental Psychology, 39, 446-483 .

6 RADICAL BEHAVIORISM : B . F. SKINNER ' S PHILOSOPHY OF SCIENC E JON RINGE N

University of Iowa

I . Radical Behaviorism as Philosophy of Science II . Radical Behaviorism : The Basic Issues III . Radical Behaviorism and Mentalism A. Privacy B. Intentionality IV. Radical Behaviorism and Knowledg e

A. "Thinking" "Impossible" "Thoughts" ? B. Naturalism and "Knowledge? " V. Summary Reference s

The cause and root of nearly all evils in the sciences is this, that while we falsely admire and extol the powers of the huma n mind we neglect to seek for its true helps. —Francis Bacon


1 59


160

RINGE N

I . RADICAL BEHAVIORISM AS PHILOSOPH Y OF SCIENC E In "To Know the Future," Skinner (1990) asserts : "Bertrand Russell once sai d that one of the two great aims of his life was to discover what could be known . I could say that one of mine has been to discover what it means to be a knower " (p . 103) . In this way, Skinner identifies what he repeatedly (e .g., Skinner, 1983 , p . 395) described as the first, most enduring, and most central concern of his career in psychology, the theory of knowledge . Surprisingly little discussion of the nature of this Skinnerian concern exists in the literature on Skinner's life an d work . Yet, it is as a theory of knowledge that the nature and significance of radical behaviorism is most clearly exhibited . In Skinner's view, scientific activities are exemplary epistemic activities . Skinner (1971) asks : "What, after all, have we to show for non-scientific or prescientific good judgement, or common sense, or the insights gained through personal experience? It is science or nothing" (p . 160) . Hence, Skinner's concern with knowing naturally focuses on science . Radical behaviorism constitutes the philosophy of science Skinner offers as his contribution to the theory of knowledge . Understanding radical behaviorism requires consideration of its relations to other aspects of Skinner's work . Skinner describes the basic relations in quite traditional ways . Thus, he distinguishes his science, the technology constitute d by applications of this science, and the philosophy of this science and technology. His science consists of the experimental analysis of behavior and the distinctive instruments, experimental procedures, and concepts it deploys (e .g ., the experimental conditioning chamber, operant and respondent behavior and conditioning, schedules of reinforcement, behavioral synthesis) . Behavior technolog y or applied behavior analysis is the field concerned with applications of the science (e.g ., applications to the design of better social practices in areas as divers e as behavior therapy, learner-paced and machine-assisted instruction, child rearing, and the practice of science itself) . Radical behaviorism is the philosophy of this science and technology. What Skinner says about radical behaviorism a s philosophy presents a challenge to traditional (and many contemporary) view s about what the theory of knowledge could possibly be . Skinner's radical behaviorism embodies the thesis that understanding science , in general, and the experimental analysis of behavior, in particular, is best achieved through the scientific analysis and interpretation of scientific behavior . In short, Skinner argues that understanding knowing requires scientific analysis of the activities of scientists . This is clearly a "bootstrap" operation that Skinne r takes to exemplify the Baconian view that the more we learn the better we ca n learn . Skinner attempts to demonstrate that from a radical behaviorist analysis , better characterizations of science and better epistemic practices emerge . Such an interpretation illuminates Skinner's consistent opposition to mental ism, clearly displayed in three exemplary essays : "The Operational Analysis of Psychological Terms" (1945), "Are Theories of Learning Necessary?" (1950),

6 RADICAL BEHAVIORISM : B . F. SKINNER ' S PHILOSOPHY OF SCIENCE

16 1

and "Selection by Consequences" (1981) . Reflection on themes in these essay s and their development in Skinner's other work provides a basis for assessing Skinner's frequent complaint that he was misunderstood as well as for highlighting the fundamental challenge radical behaviorism presents to received theorie s of knowledge. II . RADICAL BEHAVIORISM : TH E BASIC ISSUE S According to Skinner (1969, p . 267) ; the basic issue raised by radical behaviorism concerns "the usefulness of mentalistic concepts ." Skinner's antimentalism is a confluence of several themes that have been obscured by the mistake n presumption that radical behaviorism is a form of logical positivism (see als o Smith, 1986) . On Skinner's (Baconian) view, a main aim or function of scientific activity i s to develop descriptions and understandings of phenomena that enable predictio n and control . It is this function that grounds the Baconian view that knowledge i s power : Science enables more successful human action by forging better (descriptive and explanatory) tools for prediction and control of natural phenomena . Science provides a "fuel for success" by providing a means for more effectivel y realizing or pursuing human ends or goals . Skinner clearly holds that modern science exemplifies a successful practice o f this sort, and at its core is experimental investigation . In proper Baconian spirit , Skinner holds that science is the search for causes, that the identification o f causes enables prediction and control, and that experimental investigation, properly conducted, enables the identification of causes . This much of Skinner's radical behaviorism is a rather traditional and unremarkable view of science . How ever, Skinner's elaboration of this traditional view reveals what is most radical about Skinner's radical behaviorist philosophy of science . The experimental analysis of behavior is, for the most part, a straightforwar d extension of Skinner's traditional Baconian view to the study of human and non human behavior. Thus, Skinner proposes and pursues the experimental study o f the causal–functional relations between the behavior of whole organisms an d the environment in which the behavior develops and occurs . What is unique , challenging, and largely misunderstood about Skinner's radical behaviorism i s Skinner's argument that this view of science provides a basis for skepticis m about mentalism, in particular, and about various influential approaches to th e development of theories of learning and intelligent action, in general . Discussion of this part of Skinner's work has been extremely misleading, ye t it is central to his radical behaviorism . I aim to clarify how these various views , properly understood, constitute a coherent radical behaviorist philosophy of science and hence a serious contribution to the theory of knowledge and epistemi c practice . I will first attempt a clear presentation of those themes and then indi-

162

RINGE N

cate how they are misrepresented by being treated as exemplifications of logica l positivist doctrine, and underappreciated as a unique and fundamental challeng e to the causal theories of the mind that underwrite most of contemporary (naturalistic and nonnaturalistic) epistemology . The discussion will illustrate how Skinner's radical behaviorism both instantiates and is in tension with the traditiona l view that The philosophy of science can usefully be divided into two broad areas . The epistemology of science deals with the justification of claims to scientific knowledge . The metaphysics of science investigates philosophically puzzling features of the world described by science . In effect, the epistemology of science asks whether scientific theories are true , whereas, the metaphysics of science considers what it would tell us about the world i f they were true . (Papineau, 1996, 1 )

III . RADICAL BEHAVIORISM AND MENTALIS M Skinner's radical behaviorism highlights and criticizes two basic features o f mentalism : privacy and intentionality. The first is the focus of Skinner's critiqu e of introspection construed as a method for observing private mental states . Skinner's "The Operational Analysis of Psychological Terms" exemplifies his argument here. Skinner's approach to intentionality emerges in discussion of mentalistic explanation construed as causal explanation by states that (succeed or fail i n varying degrees to) represent or copy (stand for or be about) aspects of the worl d in which their bearers act . Here "Selection by Consequences" is exemplary . Th e centerpiece of both these critiques is what Skinner rightly views as his fundamental achievement : the characterization and exploration of differences betwee n operant and respondent behavior and conditioning . A key feature of this discussion is the emergence of a clear appreciation of the distinctive features of selection processes, exemplified in natural selection and operant conditioning, respectively. Skinner clearly recognizes that these distinctive features identify a causa l mode that is different from both the mechanistic causation exemplified in respondent conditioning, S —R behaviorism, and various cognitive theories, an d the "purposive causation" constitutive of the Aristotelian doctrine of final causation and embraced by various forms of vitalism (Ringen, 1993a) . The significance of these points has been obscured by characterizations of radical behavior ism as an extremely constrained and antitheoretical form of logical positivism . On this interpretation, radical behaviorism is a form of analytical behaviorism . Criticisms of this misleading view will be discussed in relation to Skinner's "Ar e Theories of Learning Necessary?" A. PRIVACY An early argument against mentalism and for an experimental analysis o f epistemic activity appears as Skinner's 1945 "The Operational Analysis of Psy-


16 3

chological Terms ." This critical assessment of introspective psychology exemplifies Skinner's view that an experimental analysis of behavior can both illuminat e epistemic practice and serve as an occasion for its improvement . Thus, Skinner notes that psychological terms are traditionally described as terms that refer t o mental states, and (at least some) mental states are traditionally described as entities that the bearer (and only the bearer) can directly apprehend, perceive, o r observe . It is this characterization of mental terms that encourages the practice of experimental introspection . Skinner argues that an experimental analysis o f behavior indicates that this practice is epistemically problematic . The problem s are revealed by analysis of variables controlling discrimative (verbal) responses . Skinner asserts : A verbal response to a private stimulus may be maintained in strength through appropriate reinforcement based upon public accompaniments of consequences, . . . or throug h appropriate reinforcement accorded the response when it is made to public stimuli, th e private case occurring by induction when the stimuli are only partly similar. . . . None of the conditions . . . permits the sharpening of reference which is achieved, in the cas e of public stimuli, by a precise contingency of reinforcement. . . . It is, therefore, impossible to establish a rigorous scientific vocabulary for public use, nor can the speake r clearly "know himself' in the sense in which knowing is identified with behaving discriminatively. (1945, p . 274 )

Skinner maintains that "a similar analysis could be made of all terms descriptiv e of motivation, emotion, and action in general, including (of special interest here ) the acts of seeing, hearing, and so on" (1945, p . 275) . Here Skinner exhibits ho w a behavior analytic "interpretation" of introspection and observation can illuminate and provide a basis for improving epistemic practice . In traditional terms , one might be inclined to say that Skinner here illustrates how the metaphysics o f science can illuminate and shape its epistemology . Skinner makes two basic points : Differential reinforcement of publicly accessible responses to publicly accessible entities (e .g ., stimuli and responses) can b e more precise than differential reinforcement of responses occasioned by privat e entities . Given this perspective, Skinner argues that in the normal course o f things, no verbal responses to private entities can be established that will serv e to precisely discriminate one private state from another. Ergo, Skinner concludes, reports of introspection are problematic as a basis for investigating th e private entities traditionally identified as constituents of conscious mental life . This point enables and provides justification for adopting improved epistemi c practice : namely, experimental methods that, unlike introspection, facilitate the development of verbal repertoires that precisely and reliably discriminate entities in the various domains of inquiry. The main point of such verbal repertoire s is what Skinner takes to be a main aim of science, namely, to facilitate more effective action by facilitating the prediction and control of phenomena in variou s domains of interest and inquiry . In this sense, the verbal practices of scientists are tools for the prediction and control of natural phenomena . The relative merits of various verbal practices are to be assessed in terms of their relative success in

1 64

RINGE N

(or promise for) enhancing prediction and control . Skinner notes that an experimental analysis of behavior, unlike traditional approaches to epistemology , makes clear the point of such improved epistemic practices . Thus, from the perspective of an experimental analysis of behavior : We may understand why terms referring to private events have never formed a stable and acceptable vocabulary of reasonably uniform usage . This historical fact is puzzling to adherents of the "correspondence school " of meaning . Why is it not possible to assign names to the diverse elements of private experience and then to proceed with consistent and effective discourse? The answer lies in the process by which " terms are assigned to private events," a process which we have just analyzed in a rough way in terms of the reinforcement of verbal responses . (1945, p. 274)

Hence, "The Operational Analysis of Psychological Terms" provides an ex ample of radical behaviorism in practice . The experimental analysis (or, mor e precisely, the behavior analytic interpretation, Skinner, 1957, pp . 11—12) o f epistemic practices in psychology, specifically the practice of introspection an d the use of psychological terms in introspective reports of conscious mental life , illuminates and provides a principled basis for criticizing that practice as well a s for selecting better practices . Experimental analysis of behavior undermines th e description of introspection as observation of private events and processes . Thi s is a first step in Skinner's critical analysis of mentalism . The analysis here provides an early example of the significance of "the three-term contingency" fo r understanding laboratory practice . In this case, Skinner draws conclusions about the character of verbal responses to private states from the assumption that suc h responses are discriminative responses established through the process of operant conditioning . Skinner's argument here exemplifies discussion of that part of philosophy of science constituting the metaphysics of science, an account o f what must be the case if a given scientific description of the world is true . B . INTENTIONALITY 1 . Positivism, Mentalism, and Theories Skinner's conclusions about the inappropriateness of introspective methods i n science are now widely shared (Lyons, 1986) but Skinner's style of argument fo r that conclusion is not so widely appreciated . In spite of what Skinner says in hi s discussion of introspection, radical behaviorism has been treated as a form of logical positivism (Fodor, 1968 ; Scriven, 1956 ; compare Ringen, 1993b ; Smith, 1986) . Skinner cites "The Operational Analysis of Psychological Terms" as an occasion on which he explicitly rejects the logical positivist philosophy of science . Skinner is quite correct in so citing this article . Lest there be any doubt, conside r the following : The main purpose of this discussion has been to define a definition by considering an ex ample. To be consistent the psychologist must deal with his own verbal practices by de -


16 5

veloping an empirical science of verbal behavior . He cannot, unfortunately, join the logician in defining a definition, for example, as a "rule for the use of a term" (Feigl) ; he must turn instead to the contingencies of reinforcement which account for the functional relation between a term, as a verbal response, and a given stimulus . This is the " operationa l basis" for his use of term ; and it is not logic but science . The philosopher will call this circular. He will argue that we must adopt the rules of logic in order to make and interpret the experiments required in an empirical science o f verbal behavior. But talking about talking is no more circular than thinking about thinking or knowing about knowing . Whether or not we are lifting ourselves by our own boot straps, the simple fact is that we can make progress in a scientific analysis of verbal behavior. Eventually we shall be able to include, and perhaps to understand, our own verba l behavior as scientists. If it turns out that our final view of verbal behavior invalidates ou r scientific structure from the point of view of logic and truth-value, then so much th e worse for logic, which will also have been embraced by our analysis . (1945, p. 277)

The logical positivists emphasized "the logic of science ." Many positivists took a Kantian view of the logic of science as describing the a priori conditions for the possibility of empirical science . In this sense, that logic of science was treated as foundational for science . The view that the methodology of scienc e could itself be an empirical science was rejected (as viciously circular) . (Se e Bon Jour, 1994 ) Radical behaviorism is orthogonal to these positivist doctrines . It should , thus, come as no surprise that attempts to portray radical behaviorism as a for m of logical positivism occasion some fundamental misunderstandings . Perhap s the most widespread of these misunderstandings concerns the relations between Skinner's antimentalism and his general skepticism about psychological theories . In "Are Theories of Learning Necessary?" Skinner forcefully attacked th e emphasis on theory construction in psychology . He gave expression to this skepticism in a number of other well-known publications . One constant in these discussions is the description of the kind of theory that Skinner is rejecting . Thi s constant description is accompanied by a wide variety of criticisms of psychologists' concern with such theories . Skinner (1950) argues agains t theory . . . [which is] any explanation of an observed fact which appeals to events taking place somewhere else, at some other level of observation, described in different terms , and measured, if at all, in different dimensions . (p 69 )

Skinner identifies three types of theory in the psychology of learning that fi t this definition : those providing explanations in terms of (1) physiological or electrochemical changes, (2) mental phenomena, or (3) "the Conceptual Nervou s System" (which Skinner, 1950, p . 70, describes as "not . . . an actual structure undergoing physiological or biochemical changes, but only . . . a dynamic system with a certain output") . On Skinner's (1950) view, these are all "theories in the sense that they are no t expressed in the same terms and could not be confirmed with the same method s of observation as the facts for which they are said to account" (pp . 69-70) .

166

RINGE N

There is a large literature assessing Skinner's basis for skepticism about suc h theories (see Kitchener, 1996) . Much of it begins with the mistaken characterization of Skinner as a logical positivist and concludes that Skinner holds an unwarrantedly restrictive view of what is required to ensure the empirical significanc e of technical terms in experimental science . Such an interpretation of Skinne r misleads more than illuminates . Three points are germane here. The first is that Skinner's description of theory delimits by exclusion a class of terms appropriate for discriminating variables that can be experimentally manipulated and measured . This exclusion highlights his view that the technical terms in experimental science should be effective tools for prediction and control of phenomena being investigated in th e laboratory. Skinner finds this problem particularly acute in psychology, becaus e everyday language for describing the behavior of humans and nonhumans alik e is mentalistic through and through and, as a consequence, is laden with implications that Skinner finds to be obstacles to the precise description of behavior, o f antecedent causes or controlling variables that might be experimentally manipulated, and of relations between behavior and its causes . The links to private events provide one example of obstacles of this sort. Skinner claims that there are others . Some of these other obstacles are the subject of discussion in "Selection by Consequences," which is the focus of the next section . This article is use fully introduced by the second and third points that need to be made by way o f illuminating the significance of the first . The second point is that Skinner's conception of theory relativizes theory t o domains of experimental inquiry . Thus, the terms used to describe independen t and dependent variables in an experimental investigation of spiking potentials o f neural cell assemblies are not, in the relevant sense, theoretical with respect to the domain under investigation, but they are, in the relevant sense, theoretica l with respect to experimental investigations of the environmental variables con trolling the behavior of whole organisms . Skinner maintains that the experimental analysis of behavior is better served by a technical vocabulary shaped an d maintained by the contingencies of experimental research than one importe d from everyday life or another experimental or theoretical domain . In Skinner' s view, the relations between verbal behavior in these different domains remains a matter for scientific investigation and interpretation . Finally, it needs to be noted that there is a type of theory construction that Skinner explicitly pursues and hence at least implicitly accepts . In "The Generic Nature of Stimulus and Response," Skinner notes that to be scientifically useful , terms for stimulus and response should discriminate features of behavior and it s antecedents that are lawfully related . Thus, Skinner (1935) says : In a reflex preparation the observed correlation is never between all properties of bot h stimulus and response . Some properties are irrelevant . The relevant properties are accordingly taken to define classes and the reflex is regarded as a correlation of generic terms . (pp . 476—477)


16 7

A reflex is accordingly defined as a correlation of a stimulus and a response at a leve l of restriction marked by the orderliness of changes in the correlation . (p. 477 ) What is wanted is the "necessary and sufficient" correlation of a stimulus and a response . The procedure recommended by the present analysis is to [experimentally] discover the defining properties of a stimulus and a response and to express the correlatio n in terms of classes . (p . 472 )

The conception of theory exemplified here is well expressed in the title of Wilfred Sellars's "Concepts as Involving Laws and Inconceivable withou t Them" (1948) . In this sense of theory, virtually all the terms employed in the experimental analysis of behavior are theoretical terms . Skinner's skepticism abou t theories is not an expression of any radical empiricist or positivist view that science is grounded in what is given in presuppositionless experience . Rather it is an expression of the view that technical scientific terms should usefully discriminate the features of the world that scientific investigation finds to be causally related . Skinner's objection to mentalistic and related terms in psychology is tha t they present systematic obstacles to that task . I turn now to develop this view . I highlight an aspect of Skinner's skepticism about theories that is largel y overlooked in discussions of mentalism and its relation to the relative merits o f radical behaviorism and cognitivism . What I say about this issue should be sufficient to suggest the ways in which the characterization of Skinner as a positivis t obscure links between his skepticism about theories and his rejection of mental ism . Specifically, I present reason to conclude that the conventional view o f Skinner's skepticism about mentalism has matters almost exactly backward . On the received view, Skinner's antimentalism is a consequence of his skepticism about theories, and his skepticism about theories is grounded in a radicall y constrained form of logical behaviorism, an application of a restrictive (radical empiricist) form of logical positivism to psychology (see Kitchener, this volume) . I present reason for concluding that, correctly understood, Skinner's skepticism about theories in psychology is grounded in his skepticism about mental ism, and his skepticism about mentalism is, in turn, grounded in his view that th e aim of science is the experimental determination of causal relations that enable the prediction and control of behavior . Skinner's basic objection to mentalism i s that mentalistic concepts are an obstacle to achieving that aim . Mentalistic concepts fail to meet the test embodied in Skinner's operationism : The ultimate criterion of the goodness of a concept is not whether two people are brough t into agreement but whether the scientist who uses the concept can operate sucessfull y upon his material—all by himself if need be. . . . What matters . . . is . . . whether he is getting anywhere with his control over nature . (Skinner, 1945, p. 383 )

2. Mentalism and Selection by Consequences "The Operational Analysis of Psychological Terms" provides one of Skinner's case studies of how a scientific analysis of the verbal behavior of psycholo-

168

RINGE N

gists can illuminate scientific methodology and provide a basis for improving it . Another is provided by "Selection by Consequences" and the earlier work, Contingencies of Reinforcement, in which similar points were developed . Here Skinner explores the significance of his (1974) claim that "operant behavior i s the . . . field of purpose and intention" (p . 55) . His main thesis is that the causal processes producing the behavior traditionally called purposive and intentional are instances of selection by consequences, a causal mode exhibited in the analogous processes of operant conditioning (the contingencies of reinforcement) and natural selection (the contingencies of survival) . Skinner argues tha t operant conditioning stands in the same relation to the adaptedness of the behavior of organisms traditionally explained by purpose and creative intelligence a s natural selection stands to the biological adaptations natural theologians attributed to the purpose and creative intelligence of divinity. He suggests that just a s we learned that design could be produced without a designer we are learnin g that intelligence (and purpose) can be produced without mind . He also suggest s that just as we learned that mysteries rather than productive problem solving ar e the end result of biological research driven by the model of creative intelligen t design, so are we learning a similar point about psychological explanations driven by a similar model. Here Skinner's remarks make contact with recent an d exciting work in naturalistic philosophy and psychology. Skinner's point here is essentially that made in philosophers' (e .g ., Darden & Cain, 1989) discussions of interfield relations : that phenomena in one field that exhibit characteristics similar to those in another may turn out to be the result of similar processes . In this instance, the similarities between the variation and subsequent adaptedness of behavior and the variation and adaptedness of biologica l structures suggest that the causal patterns that explain the first set of phenomena may be similar to those earlier found to explain the second . Skinner argues that they are . Both processes are selection processes, the common features of whic h are exhibited in operant conditioning and natural selection . Skinner calls thi s causal mode "selection by consequences ." He argues that this causal mode should displace mentalistic explanation for similar reasons and in the same wa y that natural selection displaced appeals to divine creation (and various forms of vitalism) in biology (Kitcher, 1993) . This is the basis of Skinner's antimentalism. 3. Mentalism, Teleology, and Causatio n Skinner argues that mentalism constitutes a pattern of explanation that historically has proven to be a scientific dead end . Conceptually, Skinner compares mentalism in psychology with animism in physics and vitalism in biology . He argues that mentalistic explanations mislead . Like animistic and vitalistic explanations, mentalistic explanations suggest that a causal order has been identifie d when, in fact, none has been found : Instead, teleological or goal-directed propensities are named . The controlling (i.e ., causal) variables that produce such


16 9

propensities remain unspecified, and it is their specification that is the aim of science. In this way, Skinner maintains that mentalism in psychology, like animis m and vitalism, is a scientific dead end . Mentalism does not enable progress in the search for the causal variables that facilitate the prediction and control of behavior. In contrast, Skinner argues that (as a matter of empirical fact) operant behavior is the field of purpose and intention, and in this manner proposes that stud y of the causal relations exhibited in the processes of operant behavior and conditioning represent a scientifically more promising approach to understanding behavior than mentalistic analysis can provide . Conceptually, Skinner suggests a n analogy between the study of evolution by natural selection and the study of behavior change by operant conditioning : Both involve a causal mode Skinner dubs "selection by consequences ." Skinner suggests that historically, natural selection stands in relation to vitalism and intelligent design in biology as operan t (instrumental) conditioning stands to mentalism and cognitive theories of rational choice in psychology . These arguments are mainly heuristic . They suggest that from the perspective of natural science, mentalism appears reactionary and radical behaviorism a progressive extension of the Darwinian achievement establishing how variatio n and adaptedness of behavior fit an explanatory pattern of the same type (selection by consequences) as that which explains variation and adaptedness in biological characteristics . Skinner here holds out the prospect of the kind of explanatory unification that philosophers (e .g ., Kitcher, 1993) identify as a central feature marking Darwin's achievement as an instance of scientific progress . The preceding arguments, of course, are not conclusive . Proof must include empirical evidence that the ranges of behavior for which mentalistic explanations are characteristically invoked can be explained (predicted, and controlled ) by processes of respondent and operant conditioning . Such evidence would sup port skepticism concerning cognitivist claims that mentalistic explanations are the most promising (and perhaps essential) route to whatever degree of predictability intelligent action exhibits (see, e .g ., Dennett, 1978, 1987 ; Fodor, 1987) . Skinner presents the achievements of radical behaviorist program for the experimental analysis of behavior as evidence sufficient to warrant continued (an d more) vigourous pursuit of that program . Cognitivists often contest radical behaviorist interpretations of such evidence . Surveying these debates is beyond the scope of this chapter, and, of course, any such survey would be inconclusive . (For a useful beginning, see Smith, 1994 .) For my purposes, however, it suffice s to have exhibited the nature of Skinner's criticism of mentalism . (Compar e Rachlin, 1992 ; and Stout, 1996 on the prospects for a "teleological behavior ism .") The main point could be summarized as follows : Mentalism and radical behaviorism embed the behavior of organisms in different kinds of processes , namely teleological, and, exclusively, nonteleological (e .g ., mechanistic and se-

170

RINGE N

lectionistic) processes, respectively. The significance of this contrast emerge s sharply in relation to traditional approaches to the epistemology and philosoph y of science. IV. RADICAL BEHAVIORIS M AND KNOWLEDG E

According to conventional wisdom: Science is a way to teach how something gets to be known, what is not known, to wha t extent things are known (for nothing is known absolutely), how to handle doubt and uncertainty, what the rules of evidence are, how to think about things so that judgments can be made, how to distinguish truth from fraud, and from show. (Gleick, 1992, p . 285 )

Such a view of science fits comfortably with a traditional view that knowledge i s justified true belief. (Knowers are those who either have knowledge or have the capacity to acquire it, e.g ., through learning either from experience or from instruction .) Many have argued, however, that this view of knowledge (and knowers) does not fit comfortably with the radical behaviorist conception of psychology as a natural science . The main argument for this conclusion finds tension s between the aims of radical behaviorism and each of the three constituents o f knowledge . The arguments suggest that "radical behaviorist epistemology" is an oxymoron . Belief (like doubt, certainty, and judgment) is a paradigmatic mental state . Radical behaviorism is antimentalistic . Justification is traditionally viewed as a relation between (the contents of propositional attitudes such as) belief and evidence, reasons, and rules of evaluation . Arguably, these relations are not causa l relations and rules of evaluation are not lawlike regularities governing the (psychological or sociological) acceptability of beliefs (or propositions) given certain evidence or reasons . As we have seen, radical behaviorism aims to identify th e causes and controlling relations that permit the prediction and control of behavior . Finally, truth is traditionally viewed in terms of either coherence (a relation among sentences, propositions, or the contents of belief) or correspondence ( a relation between the world and sentences, propositions, or the contents of be lief) . Both coherence and correspondence are traditionally construed as semanti c (or logical/conceptual) relations, and these are arguably not causal or lawlike relations . Skinner proposes a radical behaviorist (i .e ., a nonmentalistic, causal) accoun t of scientific behavior. It is hard to see how such a project could even begin to address the central concerns of a traditional approach to the theory of scientifi c knowledge . Critics (e .g ., Putnam, 1988) argue that radical behaviorism could not possibl y give an account of the justification of claims to true scientific belief . At best, i t could suggest how to change the subject. At worst, like full-scale skepticism, i t would be self-defeating : It would undermine any possible warrant for its own ac-


17 1

ceptance . It should be clear that considerable difficulty attends any attempt t o describe how Skinner's radical behaviorism relates to these criticisms . Nevertheless, I will try .

A . "THINKING" "IMPOSSIBLE" "THOUGHTS"?

One unsatisfactory approach is to note that Skinner often distinguishes th e technical language of various sciences from everyday language . Thus : We need a language of feelings and states of mind in our daily lives . It is the language of literature and most of philosophy . Clinical psychologists use it to learn many things abou t the histories of their clients that they could not discover in any other way . There are tw o languages in every field of knowledge, and it would be foolish to insist that the technical version always be used . But it must be used in science and especially in a science of behavior. (Skinner, 1985, p . 300)

It is a mistake to view this acknowledgment of two languages as the suggestion that in Skinner's view radical behaviorists and traditional epistemologists are engaged in nonoverlapping tasks and hence that their respective ways of de scribing phenomena can never conflict . That Skinner holds they can is suggested by various already-quoted passages from "The Operational Analysis of Psycho logical Terms ." Here and elsewhere he clearly asserts that the theory of knowledge should be scientifically assessed, and he introduces radical behaviorism a s one competitor in this field . Furthermore, in many of his essays, Skinner takes clear aim at cognitive approaches to scientific psychology. At the core of much recent work in cognitive science is an assumption share d by traditional epistemologists and classical cognitivists : Intelligent human actio n is guided by contentful cognitive states that to a greater or lesser degree succee d or fail in representing states of affairs in the world in which the actions guided by these states occur. Mental representations cause behavior. Skinner calls thi s view the copy theory. It is one principal target of radical behaviorist criticism . Versions of the copy theory (e .g ., Cummins, 1996) unite classical cognitivist psychology and much contemporary philosophy of science around two compelling theses : Scientific activity is both a form of intelligent action and an exemplary epistemic activity . Successful scientific activity produces better representations ("copies") of the world, and such representations can be objects of belie f and causes of intelligent action . These theses are so much a part of conventiona l wisdom that it is hard to see how things could be otherwise . Skinner's radical behaviorist project for a scientific epistemology might wel l be epitomized by remarks that in fact inspired and shaped it . Bertrand Russel l (1927) remarks : The word "knowledge" , like the word "memory", is avoided by the behaviorist . Nevertheless, there is a phenomenon commonly called "knowledge" which is tested behavioristically in examinations . I want to consider this phenomenon . . . with a view to deciding whether there is anything in it that the behaviorist cannot deal with adequately . (p . 88)

172

RINGE N

Skinner (1979, p. 10) acknowledges reading and being inspired by the book i n which this quotation appears . The quotation can serve as a brief characterizatio n of both the promise and the problems of radical behaviorist epistemology as wel l as the tensions between radical behaviorism and traditional views of knowledg e and knowing . These tensions emerge along several dimensions . One approach to understanding knowing and knowledge is to aspire to a theory of knowledge that is independent of (and, perhaps, presupposed by) any possible science including scientific psychology . Skinner clearly rejects such a non naturalistic view (e .g ., see p . 164) . He must be counted among the philosophical naturalists who hold that "every epistemology needs a psychology" (Kitche r 1993, p . 65) . Radical behaviorism is problematic in part because the psycholog y that traditional epistemology appears to require is in conflict with the antimentalism of radical behaviorism . Indeed, an extension of Skinner's antimentalis m suggests that from the radical behaviorist perspective, traditional mentalisti c epistemology (even if naturalistic, see James, 1891) instantiates a scientificall y problematic form of vitalism. The issues here are profound and excruciatingly hard to keep straight, but on e cannot begin to appreciate Skinner's radical behaviorism without assessing how his project relates to this issue . The challenge is to provide a coherent alternative to patently commonsensical views about the nature of scientific knowledge , specifically to provide a coherent alternative to mentalism . Meeting this challenge requires "thinking" "impossible" "thoughts" in exactly the way that natural theologians' challenge to 19th-century biologists did, namely to explai n how there can be design in nature without invoking a designer . Here, Skinner' s own comparison of natural selection and operant conditioning, his consistent an d telling criticisms of the "copy theory" of the mind, as well as recent developments in cognitive neuroscience provide a framework for articulating a set of alternatives to mentalism, particularly the representational theory of the mind . 1 . The Copy Theory of the Min d Skinner characterizes a variety of cognitivist theories as copy theories . Information-processing theories and symbol-processing theories of cognitive capacit y and function are the prime examples, but various theories of perception are characterized as copy theories as well . And, it should be clear that one of the earlies t of Skinner's sustained criticisms of this form of mentalism appears as a principled criticism of introspection as a method for experimental psychology . The copy theory of the mind and the correspondence theory of truth exhibi t interesting relations to one another that exemplify an important theme in contemporary cognitive science . The common theme is Cartesian : Human cognition, at least, involves a capacity for representing states of affairs in nature mor e or less faithfully, and the representations are constituents of the causal processes that initiate and guide intelligent action . Such a view holds out the prospect o f representations that are completely faithful to the states of affairs they represent . A familiar conception of epistemic activity naturally emerges : The aim of basic

'

6 RADICAL BEHAVIORISM : B . F. SKINNER S PHILOSOPHY OF SCIENCE

17 3

science is knowledge, ideally a theory representing the truth, the whole truth , and nothing but the truth about a given domain of inquiry . On this view, such an aim constitutes an intrinsic value of epistemic activity . Wholly, or even partially, realizing this aim is also acknowledged to be of instrumental value in a way tha t coheres nicely with the representationalist theory of the mind. Assuming that truth is a matter of correspondence, it is argued that accurate representations o f the world provide "fuel for success" in attaining any human aims whos e achievement is contingent on states of affairs that obtain in nature. True beliefs (or propositions) indicate what states of affairs exist to be negotiated and exploited in attaining human desires . Scientific theories, human beliefs (thoughts , perceptions), and the linguistic expressions of each are construed as more or less faithful copies or representations of states of affairs in nature . The task of science and the theory of knowledge as well as the conceptual framework for human psychology is constituted by these commonsense views . Skinner's criticism of the copy theory is the argument that these themes present obstacles to the scientific description and understanding of human behavior including human epistemic activities . In this sense the themes are obstacles to scientific and epistemi c progress . Radical behaviorism represents a challenge to these themes and a pro gram for describing and explaining intelligent action in radically different ways . It is beyond the scope of this chapter (and the present capacities of the author) to assess or even fully develop the details of this challenge, but some of the dimensions can be made clearer. 2. Copy Theory, Intentionality, and Teleology Skinner often (e .g ., 1969, pp . 265–266) remarks on the similarities between descriptions of our environments and descriptions of our mental states . Thus , knives are described as sharp or dull and so are pains ; the etymology of " excruciating" includes "crucifixion" ; actions that "miss their mark" still count as rational, if they were appropriate to circumstances the actor mistakenly believed t o obtain (and the mistaken belief was rationally held) . Cognitivists maintain tha t descriptions of mental states are descriptions of an "intentional environment " and not descriptions of the "geographical (e .g ., physiological, physical, or social) environment" Items in the intentional environment have content or meaning ; beliefs are about states of affairs that may or may not obtain . Faithful representation is a matter of mental entities being copies of or corresponding wit h what is the case independently of any representation . Skinner and his cognitivist critics agree that descriptions of the intentional environment are theoretical in exactly the sense that Skinner takes them to be in relation to an experimenta l analysis of behavior. Radical behaviorism embodies principled skepticism abou t the scientific value of any such descriptions of an intentional environment . Cognitivists argue that they are scientifically indispensable . Here Skinner's radical behaviorist arguments begin with the thesis that experimental investigation o f any domain is facilitated by a technical language devised for that purpose . They are further developed in relation to the analogy between operant conditioning

174

RINGE N

and natural selection . Both connect with recent work in philosophy of mind an d evolutionary epistemology. I turn to develop these connections, briefly. A main connection between the radical behaviorist attack on the copy theory and intentionality and evolutionary epistemology emerges in relation to concern s with teleology. Teleology is intimately related to the kind of intentionality exemplified in the copy theory . Following Charles Taylor (1964), Jonathan Bennet t (1990) provides what remains the most succinct expression of that relation . Bennett gives expression to two neglected themes that have been constitutive of discussions of mind for a least two millenia : The first theme is that "vital" activitie s seem to be patterned in means–ends relations ; specifically, what occurs consistently at earlier times is what is appropriate in the circumstances for bringin g about some specific state of affairs (i .e ., a goal state) later. Furthermore, withi n limits, variations in what reaching the goal state requires are matched by activities appropriate for goal attainment in those circumstances . In this sense, vital phenomena exhibit intelligence . Taylor expresses this idea by saying that certai n phenomena fit teleological generalizations . This first theme is consistent with a purely behavioral concept of intelligence and teleology . The second theme links mentalism and teleology. Taylor (1964) notes : In an "intentional system " . . . the condition of an action occurring is that it be believed adequate to the goal, and not simply that it is, in fact, adequate . And, the two may not go together. The situation as it really is may differ from the situation under its intentional description for the agent, that is, the intentional description may not hold of it . . . . The teleological account holds not of the agent in its "geographical" environment, but of th e agent in his "intentional" environment, the environment as it is for him . (p. 62)

Some behavior appears to violate teleological generalizations . For example, behavior may deviate from what is appropriate for attaining a specified goal . This deviation may occur for a variety of reasons, but occasionally, in the human case, at least, it seems appropriate to explain such deviations as instances o f goal-directed (or intentional) action directed by mistaken beliefs (e .g ., as guided by misrepresentations of the circumstances in which the action occurs, and s o caused by a misperception of what is required for goal attainment) . Here th e teleological generalization is preserved by its extension into the "intentional environment ." The behavior inappropriate for goal attainment in actual circumstances is exhibited as appropriate for the circumstances represented in the min d (or intentional environment) of the behaving system . In this sense, the intelligence of the behaving system is also preserved, and at this point, the intentionality of mentalism, the copy theory of the mind, and teleology are linked . I sugges t (following Bennett, 1983, but contra Bennett, 1993) that this is the framewor k within which Dennett's (1987) intentional stance finds its most natural and mos t intelligible home . I doubt that the traditional notion of mental content (intentionality in the sense of "aboutness") can be found outside this teleological framework . It is clear that this framework is the one to which Skinner takes radical behaviorism to provide a clear and scientifically preferable alternative . There are, however, a considerable number of obstacles to accepting, appreciating, or even fully or coherently describing the implications of a radical behav-


17 5

iorist epistemology . Radical behaviorism raises a skein of problems about truth , reference, meaning, and intentionality whose character is best suggested by th e past absurdity (but current obvious commonsense facticity) of the propositio n that there can be design without a designer. Skinner's radical behaviorism treat s this skein of problems as challenges to be met rather than demonstrations of the absurdity of behaviorism . That solutions to these problems appear to re quire "impossible" "thoughts" for their resolution is an accurate measur e both of what is radical about radical behaviorism and of how truly revolutionary it is . It is hard to imagine what the alternatives to mentalism in epistemolog y might be . This difficulty has occasioned remarks that at the level of the theory o f knowledge (and, perhaps, ethics), radical behaviorism becomes incoherent . Th e challenge to articulate a coherent alternative to mentalism is, of course, perfectl y legitimate, but Skinner's repeated comparisons between the role of natural selection in biology and operant conditioning in psychology suggest a possible way to proceed in responding to the charge ; the incoherence is no more than a failure of imagination, rather like that embodied in the nineteenth-century charge of in coherence leveled at those who suggested there could be design in nature without a designer of nature . Efforts to work out the nature of this view and to meet the challenges it pose s can illuminate both radical behaviorism and the traditional alternative to it . In the interest of encouraging such efforts, I conclude with a brief survey of a variety of views in the literature that exhibit one aspect or another of a set of theme s united in radical behaviorism, and some remarks on the strengths and weaknesses of the specific combination of these themes that radical behaviorism exemplifies . B . NATURALISM AND "KNOWLEDGE? "

Skinner advocates a nonmentalistic causal account of knowing and rejects a copy theory of perception and semantic (coherence or correspondence) theorie s of truth. A variety of recent work in the theory of knowledge embodies aspect s of this view, and exhibits affinities with radical behaviorism . Paul Churchland explicitly entertains the prospects for a "successor concep t for truth" (Putnam, 1988, p . 60) . In A Neurocomputational Perspective he describes a notion of representation as activation vectors in the weight–error spac e of connectionist systems that illustrates how a set of causal relations might constitute a notion intriguingly like that of true belief or true sentence without introducing a copy theory of the mind (Churchland, 1992, 161ff) . Causal (or informational) semantics (e .g ., Dretske, 1981) attempts a causal account of intention (o r content) of the sort semantic notions require . The causal account of content suggested by the biosemantics of Ruth Millikan (1984, 1993) and others treats representation and content as relations produced by selection processes exemplifying the causal mode Skinner terms "selection by consequences ." Panyot Butchvavrov's work (1994) on direct realism articulates a view of perception

176

RINGE N

that accounts for the perception of and appropriate discriminative responding t o diverse properties of the world without appeal to the formation or processing of internal copies of these properties . Karl Popper's (1972) well-known proposa l for epistemology without a knowing subject exemplifies conceptions of knowledge without belief . None of these views is uncontroversial, and none constitute s a complete theory of knowledge, much less a coherent radical behaviorist epistemology that would satisfy Skinner. Nevertheless, they each challenge traditiona l views in ways that a radical behaviorist can appeal to as evidence that variou s surprising features of radical behaviorism are not only conceivable, but are i n fact under discussion . It is beyond the scope of this chapter to attempt any synthesis of these views that might suggest the outlines of a radical behaviorist theory of knowledge and knowers . It suffices to have indicated the dimension s along which such an account is at odds with traditional views .

V . SUMMAR Y B . F. Skinner proposes that psychology be pursued as an experimental natura l science . His radical behaviorism identifies traditional commonsense mentalistic concepts and modes of explanation as a serious obstacle to that pursuit. Skinner's defense of radical behaviorism consists of arguing that mentalistic concept s constitute a systematic diversion from the task of developing concepts that ca n serve as effective tools in the experimental search for the causal factors an d functional relations that will enable prediction, control, and scientific interpretation of human and nonhuman behavior . Mentalistic concepts embody the cop y theory of the mind . As such they encourage the problematic attempt to observ e the features of a private mental (or intentional) environment. They also embody the kinds of teleology that have proved to be scientific dead ends . For these reasons, and not because of antecedent commitment to logical positivist dogma , Skinner takes mentalism to be properly identified as the kind of theory that constitutes a distracting and unnecessary speculative diversion from the essential discipline of experimental research . In the place of mentalism, Skinner recommends a radical behaviorist philosophy of scientific psychology : a scientific interpretation of the epistemic activities of scientific psychologists . This interpretation provides reason for pursuing a kind of experimental program that over the past 60 years has in fact proved productive, and the program avoids the pitfalls of vitalistic and animistic teleolog y from which modern natural science represents a break. In addition, the program promises coherence with modern biological science, most strikingly in providin g the kind of explanatory unification in terms of selection by consequences tha t has been the hallmark of progress in post-Darwinian evolutionary biology . Realizing this promise remains a daunting challenge, but contemporary developments include encouraging signs . Nonmentalistic approaches to artificial intelligence have emerged in force. Selectionist approaches to cognition are being


17 7

pursued in work on brain development and function as well as in contemporary work in philosophy of mind, philosophy of science, and philosophy of language . Skinner might be pleased to see a radical behaviorist prediction being confirmed : Behaviorism, as we know it, will eventually die—not because it is a failure but because i t is a success . As a critical philosophy of science, it will necessarily change as a science o f behavior changes, and the current issues which define behaviorism may be wholly re solved . (Skinner, 1969, p . 267)

REFERENCE S Bennett, J. (1983) . Cognitive ethology : Theory or poetry. Behavioral and Brain Sciences, 11 , 356-358 . Bennett, J . (1990). Linguistic behaviour. Indianapolis, IN: Hackett. Bennett, J . (1993) . Comments from a cautious ally. Behavioral and Brain Sciences, 16, pp . 381–385 . Bon Jour, L . (1994) . Against naturalized epistemology . In P. A. French, T. E . Uehling, Jr., H . K . Wettstein (Eds .) Philosophical Naturalism. Midwest Studies in Philosophy (Vol . XIX, pp . 283–300) Notre Dame, IN : University of Notre Dame Press . Butchvarov, P. (1994) . Direct realism without materialism. In P. A. French, T. E . Uehling, Jr., H . K . Wettstein (Eds .) Philosophical naturalism. Midwest Studies in Philosophy (Vol . XIX, pp . 1–21 ) Notre Dame, IN : University of Notre Dame Press . Churchland, P. (1992) . A neurocomputational perspective . Cambridge, MA : MIT Press . Cummins, R . (1996) . Representations, targets, and attitudes . Cambridge, MA : MIT Press . Darden, L ., & Cain, J . (1989) . Selection type theories. Philosophy of Science, 56, 106-129 . Dennett, D . (1978) . Brainstorms : Philosophical essays on mind and psychology . Cambridge, MA : MIT Press . Dennett, D . (1987) . The intentional stance. Cambridge, MA : MIT Press . Dretske, F. (1981) . Knowledge and the flow of information . Cambridge, MA : MIT Press . Fodor, J . (1968) . Psychological explanation . New York : Random House . Fodor, J . (1987). Psychosemantics: The problem of meaning in philosophy of mind . Cambridge, MA : MIT Press . Gleick, J. (1992) . Genius: The life and science of Richard Feynman . New York : Random House. James, W. (1891) . A plea for psychology as a `natural science .' Philosophical Review, 1, 146–153 . Kitchener, R . F. (1996). Skinner's theory of theories . In W. O'Donohue & R . Kitchener (Eds) ., The philosophy of psychology. London, Sage. Kitcher, P. (1993) . The advancement of science: Science without legend, objectivity without illusions . New York : Oxford. Lyons, W. E. (1986). The disappearance of introspection . Cambridge, MA : MIT Press. Millikan, R. (1984). Language, thought, and other biological categories : New foundations fo r realism . Cambridge, MA: MIT Press . Millikan, R . (1993). White queen psychology and other essays for Alice . Cambridge, MA: MI T Press . Papineau, D . (Ed .) . (1996) . The philosophy of science. New York : Oxford . Popper, K. (1972) . Objective knowledge: An evolutionary approach. New York : Oxford . Putnam, H. (1988) . Representation and reality. Cambridge, MA : MIT Press . Rachlin, H . (1992) . Teleological behaviorism . American Psychologist, 47, 1371–1382 . Ringen, J . (1993a) . Adaptation, teleology, and selection by consequences . Journal of the Experimental Analysis of Behavior, 60, 3–15 . Ringen, J . (1993b). Critical naturalism and the philosophy of psychology. New Ideas in Psychology, 11, 153–177 .

178

RINGE N

Russell, B . (1927) . Philosophy. New York : Norton. Scriven, M . (1956) . A study of radical behaviorism. In H . Feigl & M . Scriven (Eds .), The foundations of science and the concepts of psychology and psychoanalysis . Minnesota Studies in Philosophy of Science (Vol . 1, pp . 88-130) . Minneapolis : University of Minnesota Press. Sellars, W. (1948) . Concepts as involving laws and inconceivable without them . Philosophy of Science, 15, pp. 287-315 . Skinner, B . F. (1935) . The generic nature of the concepts of stimulus and response . Journal of General Psychology, 12, 40-65 . (Reprinted in Skinner, 1972, pp. 458-478 . ) Skinner, B . F (1945) . The operational analysis of psychological terms. Psychological Review, 52, 270-277 . (Reprinted in Skinner, 1972, pp . 370-384 . ) Skinner, B . F. (1950) . Are theories of learning necessary? Psychological Review, 57, 193-216 . (Reprinted in Skinner, 1972, pp . 69-100 . ) Skinner, B . F. (1957) . Verbal behavior. New York: Appleton-Century-Crofts . Skinner, B . F (1969) . Contingencies of reinforcement : A theoretical analysis . New York: AppletonCentury-Crofts . Skinner, B . F. (1971) . Beyond freedom and dignity. New York: Alfred A. Knopf. Skinner, B . F. (1972) . Cumulative record: A selection of papers (3rd ed.) . New York: AppletonCentury-Crofts . Skinner, B . F. (1974) . About behaviorism. New York : Alfred A. Knopf. Skinner, B . F (1979). The shaping of a behaviorist. New York : Alfred A. Knopf. Skinner, B . F (1981) . Selection by consequences . Science, 213, 501-504 . Skinner, B . F (1983) . A matter of consequences. New York : Alfred A . Knopf. Skinner, B . F (1985). Cognitive science and behaviourism. British Journal of Psychology, 76 , 291-301 . Skinner, B . F (1990). To know the future . Behavior Analyst, 13, 103-106 . Smith, L . (1986) . Behaviorism and logical positivism: A reassessment of the alliance. Stanford, CA : Stanford University Press . Smith, T. (1994) . Behavior and its causes: Philosophical foundations of operant psychology. Boston : Kluwer. Stout, R. (1996) . Things that happen because they should: A teleological approach to action. New York: Oxford. Taylor, C . (1964). The explanation of behaviour. New York: Humanities.

EMPIRICAL BEHAVIORIS M

SIDNEY W . BIJO U University of Nevada —Reno

I. Freud and Psychoanalysis II. Watson and Behavioris m III. Lewin and Field Theory IV. Hull—Spence and Learning Theor y V. Skinner and the Experimental Analysis of Behavio r VI. Radical Behaviorism and Interbehaviorism VII. The Influence of Empirical Behaviorism VIII. Summary and Conclusion Reference s Empirical behaviorism is a combination of Skinner's radical behaviorism an d Kantor's interbehaviorism focusing on human development. This variation o f behaviorism is presented as a case history that is, as an account of a lon g search for a philosophy of science for psychology . The quest began in 1935 when I was a graduate student at Columbia University . I was taking Robert S . Woodworth ' s course in a survey of contemporary psychology. His lectures and assigned readings on psychoanalysis, Gestalt, organismic, and personalistic psychology I found bewildering . What I wanted t o know was which view had the greatest promise for a science of psychology . On this issue, Woodworth remarked that psychology was in a state of flux and only a genius could unify the field and move it in the direction of the natural sciences . At this point, I took it upon myself to find that person and to join him or her i n reaching that goal .


17 9


180

BIJO U

I . FREUD AND PSYCHOANALYSI S Freud seemed a possibility ; he was a determinist in that he considered every behavior, thought, and emotion to be caused by some underlying psychologica l process . He claimed that his theory could account not only for the behavior o f individuals but could also provide insight into the areas of arts and literature an d serve as the underpinning of cultural anthropology and history . So to understand psychoanalysis and its claims better, I enrolled in a cours e taught by the psychiatrist Fritz Wittels at the New School of Social Research i n New York City. It became clear that for Freud, the subject matter of psychology was the mental apparatus, and the goal of psychology was to discover how th e mental apparatus is constructed and how the forces within it operate . Unfortunately, many of Freud's concepts and principles were stated in ambiguous terms , thus precluding experimental verification . He also believed that the data of psychology should be collected in therapeutic or quasi-therapeutic situations b y psychoanalysts who preferably themselves had been analyzed . His restrictions on who could collect data and the situations in which they were collected raised questions about the generality of the information obtained . These considerations and others (see Shaffer, 1936) convinced me that Freud was not the genius I wa s looking for.

II . WATSON AND BEHAVIORIS M I turned to behaviorism and reflexology and began to read some of the publications of the spokesman for that view : John B . Watson (1919, 1930) . I was impressed with Watson's simple and direct style, his practical orientation, his frequent references to laboratory research with animals and young children and t o biophysiology, as well as his expansive and positive attitude about the future o f psychology as a natural science . Compared with Freud, he was ultrascientific . In my enthusiasm and naivete, I wrote to Watson, then vice president of the J . Walter Thompson Advertising Company, telling him of my interest in the behaviora l approach to child development and asking him to suggest a topic for my master's degree research . He replied promptly with the suggestion that I study how young children learn "muscle sense ." How, for example, does little Jimmy kno w that his arms are stretched out at shoulder height when his eyes are closed? H e proposed also that for subjects I might contact Patty Hill-Smith, director of th e model kindergarten at the nearby Horace Mann School at Teachers College . Thi s kind of problem was most appealing to me, but I could not find a member of th e psychology department who was willing to serve as my adviser for such a study . I was sorely disappointed, but to make progress in my graduate work, I undertook to complete a project that a member of the faculty, Louise E . Poole, had started . This project involved the measurement of nonverbal intelligence i n young retarded children (Bijou, 1938) . The study was more than a pragmatic

7 EMPIRICAL BEHAVIORISM

18 1

move since I had had a growing interest in the problems of retarded persons, after attending a summer session at Letchworth Village, a large residential institution for mentally retarded persons, not far from New York City. This summer session, sponsored by the New York State Department of Mental Health, left me with the impression that psychologists should not only administer psychologica l tests to retarded persons but should also take advantage of their knowledge i n learning and conditioning and lead the way in developing and carrying out rehabilitation programs . III . LEWIN AND FIELD THEOR Y Among the postmaster's degree courses I took was one on Gestalt and organismic psychology given by George W. Hartmann at Columbia's Teachers College . It was here that I was introduced to the work of Kurt Lewin (1935, 1936) . Lewin's philosophy of science seemed to me not only to hold promise for unifying psychology, but his research with children was original and seemed to hav e the potential for advancing knowledge in developmental psychology. In a discussion with Hartmann, he suggested that I consider completing my graduate work at the University of Iowa, where I could study with this stimulating man who had recently immigrated from Germany . Perhaps Lewin's neoGestalt psychology was the approach I had been looking for . In 1937 I discontinued my further studies at Columbia and accepted an appointment as clinical psychologist at the Delaware State Hospital and Menta l Hygiene Clinic . Two years later, in 1939, I was accepted for graduate study i n Iowa's department of psychology and was given a part-time assistantship . Upon my arrival in Iowa City, I learned to my chagrin that Lewin was not available a s my adviser because his appointment was in Iowa's Child Welfare Research Station and he was not a member of the psychology department. I was told that if I wanted to remain in psychology and keep my assistantship, I would have to select an area of concentration in psychology . I selected learning since it was closest to my interest in child clinical psychology . Kenneth W. Spence was assigne d to me as an adviser . I did, though, take a course from Lewin titled "Theory of Psychology an d Personality Development." Lewin was indeed a stimulating although somewha t disorganized lecturer, and the content of his course was appealing because i t dealt with the behavior of children . However, what bothered me was tha t Lewin's analysis of an individual's behavior was described in terms of response–response relations rather than stimulus–response relations . All the conditions of an interaction situations, responses, setting, and motivation were entered into a field on the basis of an individual's perceptions, hence the relationships among the variables in a field were correlations among response variables . My interest was in stimulus-response functional relationships . Besides, I was beginning to believe, through the learning course with Spence, that the

182

BIJOU

learning theories of Hull, Tolman, and Guthrie had more promise than Lewin' s field theory for the future of psychology .

IV . HULL—SPENCE AND LEARNING THEOR Y

Among the highlights of my training at Iowa were Spence's two-semeste r course in animal learning and conditioning and his informal seminar (The Mon day Night Group), which was devoted entirely to an intensive chapter-by-chapter review of Hull's manuscript for his Principles of Behavior (1943), Gustav Bergmann's course in philosophy of science (mostly logical positivism), E . F. Lindquist's course in statistics (mostly small sample research designs), and Joh n A. McGeoch's course in systematic psychology and his course in human learning (mostly rote memory studies) . My dissertation (Bijou, 1943, 1951), which combined my interest in psychopathology and learning, was a study of experimental neurosis in rats modele d after Pavlov's investigations with dogs in conflict situations . However, I used the instrumental rather than the classical conditioning model . By the time I received my degree I was convinced that it was Hull who wa s the genius destined to bring psychology to the threshold of the natural sciences . I was satisfied that the objective of the science of psychology was to develop a general theory of behavior that would account for events between the stimulu s input and response output of an organism and that research would be concerned with testing hypotheses to tease out the properties of the internal variables an d their relationships. Even though the theory dealt with the behavior of individuals, the research typically involved small-sample, group designs since hypotheses testing was the focus of interest . The fact that, thus far, his approach was based on the behavior of nonverbal organisms apparently was not a matter o f concern for Hull, because he believed that the complex human behavior of verbal organisms would eventually be incorporated into his system . All that was necessary was to redefine Freud's concepts and principles in a way that woul d make them scientifically acceptable . Some of Hull's students had already begun to try to do that . Sears and his colleagues, for example, had been using Freudian concepts in their studies of children, and there was much discussion about th e frustration–aggression hypothesis . In 1941, with a doctoral degree in hand, I sought an academic appointment i n a market that was shrinking rapidly because the country was in the throws o f World War II . University enrollments were down, professors were leaving to joi n the armed forces, and those ineligible for the draft as I was were given shor t shrift by personnel interviewers . In desperation I took an appointment as a re search psychologist at the Wayne County Training School in Michigan, a residential school for high-functioning retarded children and young adults . Becaus e my research activities at the school seemed trivial compared with helping m y country win the war, I took a leave of absence to accept a commission in th e army.


18 3

After my discharge from the Army Air Force in 1946 and an additional yea r at the Wayne County Training School, on the recommendation of Spence I wa s recruited by Skinner, then chairman of the department of psychology at Indian a University, to serve as assistant professor and director of a new clinical program . Skinner wanted a clinical program with an experimental orientation . I taught a course in experimental psychopathology heavily saturated wit h readings by Hull's students, particularly Neal Miller, Robert Sears, and Car l Hovland . In addition I set up a laboratory to continue research on conflict in rats . The course went well, but the research was faltering . A graduate student, Bill Rigby, and I found that the temporal gradients of conflict were not stable, an d the avoidance gradient was not steeper than the approach gradient as predicte d by Miller (1944) . When I expressed my disappointment to Skinner, he smile d and said, "The animal is always right ." During my brief tenure at Indiana, I was stimulated by the frequent discussions among faculty and students about the merits and limitations of the Skinnerian, Hullian, and Kantorian views . Among other things, Skinner and Kanto r held a well-attended joint seminar, and Skinner's students, such as Bill Estes an d Norm Gutman, and Hull's disciples, including Bill Jenkins, Bill Verplanck , Doug Ellson, and I, held informal seminars . It was during this time that I had several discussions with Kantor both on psychology in general and on Hull's behavior theory in particular. Kantor took issue with Hull that psychology shoul d be devoted to studying, almost exclusively, events that presumably go on between stimulus input and response input. Furthermore, he considered Hull's system to be reductionistic because Hull believed that his internal variables woul d eventually be shown to be related to biophysiological structures and functions . And lastly, Kantor was skeptical about Hull's emphasis on the hypothetico-deductive method for theory construction and research, averring that both deductive and inductive methods are proper tools for a science of psychology . During this time at Indiana, there was also considerable research on basic and applied problems . I was particularly interested in a study by Fuller (1949) and one by Greenspoon (1955) . Fuller showed that a profoundly low-functioning institutionalized patient who, according to the hospital staff, was unable to learn , could in fact increase his arm movements when they were followed by warm milk and sugar and decrease them when the contingency was discontinued . And Greenspoon, in an experimental situation parallel to the operant conditioning animal model, demonstrated that the verbal behavior of college students, in th e form of saying words, could be manipulated by the experimenter's delivery o f contingent verbal behavior— "uh-hum" and visual and auditory stimuli . I left Indiana University in 1948 for an appointment as associate professor and director of the Institute of Child Development at the University of Washing ton . The institute at that time consisted only of a child psychological clinic located in the department of psychology . In a short while, I was instrumental in moving the clinic into a building that housed a nursery school . By 1950 the "new" Institute of Child Development consisted of a two-unit nursery school, a child development clinic, and a research laboratory . For the development of a

1 84

BiJo u

research program, I sought the guidance of Robert R. Sears and his colleagues , who were then conducting two kinds of research . One was on the relationships between mothers' child-rearing practices and attitudes on one hand, and thei r children's behavior in preschool on the other. In these studies they used grou p designs and correlations to analyze their data . (e .g ., Sears, Whiting, Nowlis, & Sears, 1953) . The other type of research carried on by the Sears's group deal t with the behavior of individual children using projective doll play technique s (Sears, 1947) . I was acquainted with projective doll play as a therapeutic technique (Bijou, 1954) ; now I had an opportunity to evaluate it as a research tool . A graduate student, Sara Lee Cohen, and I set up a situation with an open, furnished doll house and four dolls representing a father, mother, sister, and brother. The procedure was simple : Children, one at a time, were brought into the laboratory and told that they could play with the toys as much and as long as they liked . An observer, in full view, recorded the verbal and nonverbal behavior o f each child . The resulting protocol was interpreted in Freudian-like terms, such as fantasy aggression, nurturance, and frustration . The entire procedure differed markedly from my previous experience with laboratory research . There was no control over the antecedent and consequen t conditions, and the responses, which varied widely, had to be codified and continually checked for reliability. The method allowed only for a description of what the child did and said, and inferences had to be checked by further re search . The method also foreclosed the possibility of demonstrating functiona l relationships between stimulating conditions and the child's responses . Unwilling to discard my training and experience in research to study the behavior o f children, I continued my search for a point of view that would allow me to study the behavior of individual children in a way that would be compatible with th e science of psychology. V. SKINNER AND TH E EXPERIMENTAL ANALYSI S OF BEHAVIO R My exposure at Indiana to Skinner's research with animals and to studies b y Fuller (1949) and Greenspoon (1955) on operant conditioning in humans led m e to try Skinner's model for a systematic research program with children . To tha t end, I built an apparatus that required a child to drop a ball into a hole in order t o receive a trinket, thus producing a situation with objective, countable responses , and clearly discernible and manipulable antecedent and consequent conditions . With this crude apparatus I carried out a study (Bijou, 1955) on the effect of intermittent reinforcement on the frequency of ball dropping and on extinction o f that response . The design was a compromise between a within-subject and between-subject design . Eighteen 4-year-olds divided into two groups served as subjects . In one group, each child earned six trinkets for six ball-drop responses


18 5

followed by a 3 .5-minute extinction period . In the other group, each child als o earned six trinkets, but the trinkets were distributed over 30 ball-drop response s and a 3 .5- minute extinction period. During extinction, the first group made a n average of 15 responses ; the second group made an average of 22 responses . The difference of seven responses was statistically significant . The study was replicated on 21 4-year-olds with one change in the apparatus . To make the stimulu s accompanying the delivery of the trinkets more distinctive, a buzzer was adde d to the sound of the motor that operated the trinket-delivery device . Finding s from this study verified even more dramatically those obtained in the first study. Despite these positive results, I was dissatisfied because the method had a shortcoming : It did not yield data for a functional analysis of individual behavior. Furthermore, the small number of responses seemed to be more suitable fo r theory testing than for empirical demonstration research . In the second study (Bijou, 1957), I followed operant procedures scrupulously. This time I constructed an apparatus with a lever (an O' Cedar spong e mop handle) for responding because the ball-drop response led to unnecessary complications . For example, sometimes a child's ball would miss the hole , whereupon he or she would chase around the room to retrieve it . I also installed a Gerbrands recorder to record the responses and the delivery of reinforcers, an d a Gerbrands feeder to deliver small objects : trinkets, cookies, and low-calorie candies . With the new setup, I studied the behavior of individual 4-year-olds o n several fixed-interval and fixed-ratio schedules of reinforcement and obtaine d hundreds of responses in orderly relationships to the schedules of reinforcement. The second study the "pure" operant study inspired a number of investigations on basic problems by various members of the institute staff : Baer on es cape and avoidance on two schedules of reinforcement (1960), on the effect o f positive reinforcement on extinguishing responses (1961), on laboratory control of thumb sucking by withdrawing and re-presenting reinforcement (1962b), an d behavior avoiding reinforcement withdrawal (1962a) ; Baer and Sherman (1964) on the control of generalized imitation ; Bijou (1961) on discrimination as a baseline for individual analysis ; and Bijou and Orlando (1960, 1961) on schedules of reinforcement and discrimination in retarded children at the newly established laboratory at the Rainier State School . That second operant study also served to inspire the nursery school staff to engage in a series of studies on problem behavior in the nursery school . The first dealt with the extinction of regressive crawling by a 3 .5-year-old girl (Harris , Johnston, Kelley, & Wolf, 1964), the results of which were so dramatic that th e method (Bijou, Peterson, & Ault, 1968 ; Bijou, Peterson, Harris, Allen, & Johnston, 1969) was adapted for the elimination of crying and whining, isolate play , and excessive passivity (Harris, Wolf, & Baer, 1964) . That second operant study was also the basis for a series of studies in the Child Development Clinic in which the investigators explored procedures fo r training mothers to serve as therapists for their own children (Wahler, Winkel , Peterson, & Morrison, 1965) . A spin-off of this series was one in which a mother

186

BIJO U

was trained in her home to be a therapist for her 4-year-old child with seriou s aggressive behavior (Hawkins, Peterson, Schweid, & Bijou, 1966) . As the work of the institute became more widely known, requests for servic e came from outside the city of Seattle . One such call was from a physician i n charge of the children's ward of the Washington State Hospital requesting assistance to get a 3 .5-year-old "schizophrenic" boy to wear glasses to compensat e for the surgical removal of cataracts . Although we had not treated an autistic child before, I accepted the case because I had observed a laboratory study b y Ferster and DeMyer (1961) in which young autistic children learned simple responses to vending-type machines . Montrose Wolf, research assistant professor , and Todd Risley, a graduate student, undertook to treat the child with the assistance of the hospital psychologist, Hayden Mees (Wolf, Risley, & Mees, 1964) . They not only created programs to reduce problem behaviors (tantrums, self-destructive behavior, and bedtime and eating problems) and getting him to wear hi s glasses, but also trained the attendants, nurses, and parents to carry out the pro grams and record data . After 7 months of intensive treatment, the child was wearing his glasses most of the time and his problem behaviors were under sufficient control to warrant his discharge from the hospital . Six months later, hi s mother reported that he continued to wear his glasses, did not have tantrums o r sleeping problems, and was becoming more verbal . She summarized it all b y saying that for the first time he had become a source of joy for the family. In 1961, I spent a sabbatical year at Harvard with Skinner to study comple x behavior in retarded children . For subjects, I set up a laboratory at the Fernald State School near Cambridge . Using a match-to-sample apparatus developed b y James Holland, I explored the programming of antecedent stimuli to facilitat e the learning of right-left visual discrimination (Bijou, 1968b) . While at Harvard I took the opportunity to informally survey the academi c programs for young retarded children in the Northeast. All seemed to be relatively ineffectual . That impression together with my exposure to programmed instruction in Skinner's Teaching Machine Project convinced me that a teachin g program based on behavioral principles would be superior to anything I ha d seen. So on my return to Seattle, I constructed a classroom at the Rainier Stat e School that would facilitate individualized teaching, hired two behaviorally oriented teachers Cecilia Tague and John Kidder and arranged to have Montrose Wolf and Jay Birnbrauer of the institute staff serve as consultants . Together we developed a motivational system based on a token economy and a curriculu m consisting of programs for reading, writing, arithmetic, and related subjects . Th e average IQ of the children was in the low 60s, and all had serious behavior problems . Although the children came to the class in small groups, each received individualized instruction. All the children made good progress, their behavio r problems were eliminated, and they actually enjoyed coming to class (Bijou , Birnbrauer, Kidder, & Tague, 1966 ; Birnbrauer, Wolf, Kidder, & Tague, 1965) . The reading part of the curriculum was adapted for use with a computer touch


18 7

screen and is now known as the Edmark Reading Program recently acquire d from the publisher by IBM . VI . EMPIRICAL BEHAVIORISM : RADICA L BEHAVIORISM AND INTERBEHAVIORIS M As our research progressed, the teaching members of the institute felt th e need to present child development to our students according to the concepts an d principles used in our research program . So Donald Baer and I took it upon our selves to do just that in a series of paperback books . But in thinking through the project, we realized that it could not be done only on the basis of Skinner's radical behaviorism as gleaned from his Science and Human Behavior (1953) . S o my search for an acceptable view for psychology and child development as a natural science was on again . My conversations with Kantor while at Indiana led me to read some of his books and papers, particularly his Interbehavioral Psychology (1959) . It becam e apparent that his approach was compatible with Skinner's, yet provided man y concepts that were needed to round out a complete theory of child development , as, for example, his conception of developmental stages and his strong emphasi s on setting factors in all human behavior. Our plan, then, was to coordinate radical behaviorism and interbehaviorism into an approach we called empirical behaviorism . This decision terminated my search : I was now convinced that th e work of two geniuses Skinner and Kantor could be integrated into a psychological system that would meet the criteria of a natural science . The first volume (Bijou & Baer, 1961) was devoted to principles . Child development was described as a branch of behavioral psychology that deals exclusively with observable and potentially observable interactions between the behavior (responses) of an organism and environmental events (stimuli) and i s closely related to both biophysiology and cultural anthropology. Human development was defined as progressive changes in interactions between behavior an d stimulating conditions expressed as B = f(S), with S consisting of current stimuli and a history of past stimuli in similar situations . The child was conceptualized as the locus of responses, modifiable by respondent and operant processes , and the environment as specific stimuli and setting events, some originating i n the biophysiology of the child but most by far in the social, physical, and chemical aspects of the environment . For the purposes of research and practical application, the lifelong sequence of interactions that make up the psychological development of a person was divided into three major stages foundational, basic , and societal on the basis of mixed biological and cultural criteria (Kantor & Smith, 1975) . In general it was assumed that developmental interactions proceeded from simple to complex, and emotional behavior and self-control wer e analyzed as examples of complex interactions with intertwined operant and respondent interactions .

188

Buou'

In the second volume (Bijou & Baer, 1965) the principles were applied to the foundational stage, which includes the prenatal, neonatal, and infantile sub stages . Both prenatal and neonatal development were analyzed in naturalisti c terms ; the prominent psychological progressions in infancy (motor, social, emotional, and self-care) were analyzed as progressions in operant interactions ; an d beginning verbal behavior, which marks the end of the foundational stage, wa s analyzed in terms of the baby's ability to utter speech sounds and his or her interactions with the caretaker, whose behavior had acquired conditioned reinforcing properties by the pairing of his or her proximity, attention, and affection wit h the unconditioned stimuli inherent in normal child-rearing activities . In the third volume (Bijou, 1976), the principles were applied to the basi c stage, the early childhood years in which the acquired interactions become th e roots of a child's personality. In this volume, exploratory behavior, curiosity , play, cognition, intelligence, and social behavior were analyzed primarily in operant terms, while child behavior treatment, early childhood education, and parent training were presented as examples of practical application . Volume 1 was revised twice. In the first revision (Bijou & Baer, 1978), the main changes dealt with the relationship between behavior and stimulating conditions reflecting the continuing influence of interbehaviorism . The formula B = f(S) was replaced with Bf H Sf, where Bf and Sf stand for behavior an d stimulus functions and the two-headed arrow as the reciprocal relationship between the two . The point was made that behavior (responses) and stimuli refer to biological interactions, whereas response and stimulus functions refer to psycho logical interactions in that they include the history of past interactions . Further more, the relationship between response and stimulus functions are reciprocal , not linear as had been originally presented ; an interaction changes behavior and at the same time changes the environment : All variables interact mutually resulting in changes in both the behavior of the child an d the functional properties of variables in the environment . Sometimes the changes are subtle ; sometimes they are dramatic . Sometimes they fluctuate between retardation and acceleration ; most of the time they are progressive. (Bijou & Baer, 1978, p . 29 )

Another significant change in the first revision was the expanded treatment o f complex interactions including conflict, decision making, emotional and affective interactions, self-management, biofeedback, problem solving, thinking, an d curiosity. The second revision of the child development book (Bijou, 1993) included four main changes . To take care of a glaring omission from the original and first revision, a chapter on verbal behavior was added based on Skinner's (1957) operant analysis of the speaker and Hayes and Hayes's (1989) analysis of the th e listener. Another change aimed to clarify the meaning of setting factors, which wer e now defined as "the immediate circumstances determining which particular response function in a person's repertoire will operate at the moment ." A definition

18 9


such as this emphasizes that historical interactions greatly shape most current interactions (Kantor, 1946, p . 37) . The other changes were additions . One was a behavior analysis of intelligence as measured by psychological tests (Bijou, 1971), and the other was Skinner's (1973) discourse on the relationship between basic and applied research .

VII . THE INFLUENCE O F EMPIRICAL BEHAVIORIS M An indication of the influence of empirical behaviorism was noted by Kazdi n in his History of Behavior Modification : Experimental Foundations of Contemporary Research (1978) when he stated that "Bijou, Wolf, Baer, and severa l other investigators at the University of Washington began applied operant re search that represented the most extensive efforts of the early 1960s and established the University of Washington as the nucleus of applied operant research " (p . 273) . His reference was to the applied studies on child behavior treatment , early education, special education, and parent training described in the previou s section, and to the development of a research methodology for descriptive an d experimental studies in natural settings (Bijou, Peterson, & Ault, 1968 ; Bijou , Peterson, Harris, Allen, & Johnston, 1969), and to a functional analysis of behavior assessment (Bijou & Peterson, 1971 ; Marholin & Bijou, 1977, 1978) . The influence of empirical behaviorism can also be gauged by the number o f books with this orientation . Outstanding among them are the treatment of handicapped children (Lovaas, 1981), parent training (Latham, 1993), the behaviora l treatment of Down's syndrome children (Dmitriev, 1997), early education (Allen & Goetz, 1982 ; Evans, 1975 ; Goetz & Allen, 1983), remedial speech and language training (Sloane & MacAulay, 1968), education (Jensen, Sloane, & Young, 1988 ; Sulzer-Azaroff & Mayer, 1986), educational psychology (Kratochwill & Bijou, 1987), classroom management (Sloane, 1976), and single-subject research in special education (Tawney & Gast, 1984) . Evidence of the interest in empirical behaviorism abroad is the fact that th e Bijou and Baer series has been translated into Spanish, Italian, Japanese, an d Portuguese, and several psychologists in other countries have published book s that have an empirical behaviorism view. There are, for example, the Perini book (1997) in Italian on educational psychology and the Perini and Bijou boo k (1993) on the education and treatment of retarded children, the Soriano boo k (1996) in Spanish on child clinical psychology, the Azuma book (1975) i n Japanese on behavior modification of retarded persons . Another indication of the impact of this view is its practical application . The Portage Project of Early Childhood Education (Bijou, 1991), a home-based pro gram in which the mother is trained to teach her own handicapped child is an ex ample . The International Portage Project has been adopted in more than 3 0 countries . In Japan, it has become the nationally recognized way of helping par-

190

BIJO U

ents to work constructively with their handicapped children . Another example i s the establishment of a child laboratory in the department of psychology of th e Central University of Venezuela in Caracas, and a third example is the extensiv e program for the education and treatment of handicapped children and their parents at the Ann Sullivan Center in Lima, Peru . In addition, there have been numerous papers and chapters on the theory o f child development . Among them are the Baer paper (1970) on the age-irrelevan t conception of development, his paper (1973) on the control of the developmen t process, and his paper (1976) on the organism as host ; the Roales-Ruiz and Bae r chapter (1996) on the behavior analytic view of human development ; the Bijou paper (1968a) on ages, stages, and the naturalization of human development, hi s paper (1979) on some clarifications of the meaning of a behavior analysis o f child development, his paper (1966) on a functional analysis of retarded development, his paper (1971) on environment and intelligence, his chapter (1996) on the role of setting factors in an analysis of child behavior ; Wahler and Fox (1981) on setting events ; and Morris et al . (1982) on the relationship between behavior analysis and development . VIII . SUMMARY AND CONCLUSIO N My long and persistent search for the genius who might unify psycholog y and have it take its place among the natural sciences prompted me to evaluat e the views of Freud on psychoanalysis, Watson on behaviorism, Lewin on fiel d theory, Hull on behavior theory, Skinner on radical behaviorism, and Kantor o n interbehaviorism . I found that there was not one such person but two Skinne r and Kantor who filled the bill . Although their terminologies differ and eac h emphasized different aspects of psychology, their positions are essentially compatible and could therefore be combined into a system that we call empirical behaviorism. On the basis of this view, I, together with the staff at the Institute of Child Development of the University of Washington, developed a systematic re search program and a conceptual system for normal and deviant child development that stimulated practical programs for the treatment of normally developin g children with behavior problems, and the treatment of handicapped children, a s well as programs in early childhood education, special education, and paren t training in this country and abroad .

REFERENCE S Allen, K . E., & Goetz, E . M . (Eds .) (1982) . Early childhood education: Special problems, specia l solutions . Rockville, MD : Aspen Systems . Azuma, T. (1975) . Behavior modification of the mentally retarded . Tokyo: Meiji Tosko . Baer, D. M. (1960) . Escape and avoidance response of preschool children to two schedules o f reinforcement withdrawal . Journal of the Experimental Analysis of Behavior, 3, 155-160 .


19 1

Baer, D . M . (1961) . Effect of withdrawal of positive reinforcement on an extinguished response i n young children . Child Development, 32, 67-74 . Baer, D . M . (1962a) . A technique of social reinforcement for the study of child behavior : Behavior avoiding reinforcement withdrawal. Child Development, 33, 847-858 . Baer, D . M . (1962b) . Laboratory control of thumb sucking by withdrawal and representation o f reinforcement . Journal of the Experimental Analysis of Behavior, 5, 525-528 . Baer, D . M. (1970) . An age-irrelevant concept of development . Merrill-Palmer Quarterly Journal of Behavior and Development, 16, 238-245 . Baer, D . M . (1973) . The control of the development process : Why wait? In J . R . Nesselroade & H . W. Reese (Eds .), Life-span developmental psychology : Methodological issues (pp . 185-193) . New York : Academic Press . Baer, D . M . (1976) . The organism as host. Human Development, 19, 87-98 . Baer, D . M ., & Sherman, J . A. (1964) . Reinforcement control of generalized imitation in youn g children . Journal of Experimental Child Psychology, 1, 37-49 . Bijou, S . W. (1938) . The performance of normal children on the Randall's Island Performanc e Series . Journal of Applied Psychology, 22, 186-191 . Bijou, S . W. (1943) . A study of experimental neurosis in the rat by the conditioned respons e technique . Journal of Comparative Psychology, 36, 1-20. Bijou, S . W. (1951) . A conditioned response technique to investigate experimental neurosis in the rat . Journal of Comparative and Physiological Psychology, 44, 84-87 . Bijou, S . W. (1954) . Therapeutic techniques with children . In L . A . Pennington & I. A . Berg (Eds .) , An introduction to clinical psychology (2nd ed ., pp . 608-631) . New York : Ronald Press . Bijou, S . W. (1955) . A systematic approach to an experimental analysis of young children . Child Development, 26, 161-168 . Bijou, S . W. (1957) . Methodology for an experimental analysis of child behavior . Psychologica l Reports, 3, 243-250 . Bijou, S . W. (1961) . Discrimination performance as a baseline for individual analysis of youn g children . Child Development, 32, 163-170 . Bijou, S . W. (1966) . A functional analysis of retarded development . In N . R. Ellis (Ed .), International review of research in mental retardation (Vol. 1, pp. 1-19) . New York : Academi c Press . Bijou, S . W. (1968a). Ages, stages, and the naturalization of child development American Psychologist, 33, 419-427 . Bijou, S . W. (1968b) . Experimental analysis of left-right concepts in retarded children. In N. R . Ellis (Ed .), International review of research in mental retardation (Vol 3, pp . 65-96) . New York : Academic Press. Bijou, S . W. (1971) . Environment and intelligence: A behavioral analysis . In R. Cancro (Ed.) , Intelligence : Genetic and environmental influences (pp . 21-239). New York: Grune & Stratton. Bijou, S . W. (1976) . Child development: The basic stage of early childhood . Englewood Cliffs, NJ : Prentice-Hall . Bijou, S . W. (1979) . Some clarification on the meaning of behavior analysis of child development . Psychological Record, 29, 3-13 . Bijou, S . W. (1991). An overview of early childhood programs around the world . In J . Herwig & M . Stine (Eds .), Proceedings from the Third International Conference (pp . 10-15) . Portage, WI : Cooperative Educational Service Agency . Bijou, S . W. (1993) . Behavior analysis of child development (2nd ed.) . Reno, NV: Context Press. Bijou, S . W. (1996) . Setting factors in the analysis of child development . In S . W. Bijou & E . Ribes (Eds .), New directions in behavior development (pp . 147-154) . Reno, NV : Context Press . Bijou, S . W. & Baer D. M . (1961) . Child development: A systematic and empirical theory (Vol. 1). Engelwood Cliffs, NJ : Prentice-Hall . Bijou, S . W., & Baer, D . M . (1965) . Child development: II. Universal stage of infancy. New York: Appleton-Century-Crofts .

192

BIJO U

Bijou ., S . W. & Baer, D . M . (1978). Behavior analysis of child development. Englewood Cliffs, NJ : Prentice-Hall . Bijou, S . W., Birnbrauer, J . S ., Kidder, J. D ., & Tague, C . E. (1966). programmed instruction as a n approach to the teaching of reading, writing, and arithmetic to retarded children. Psychologica l Record, 16, 505-522 . Bijou, S . W., & Orlando, R . (1960) . Single and multiple schedules of reinforcement i n developmentally retarded children . Journal of Experimental Analysis of Behavior, 4, 339-348 . Bijou, S . W., & Orlando, R. (1961) . Rapid development of multiple schedule performances with retarded children . Journal of Experimental Analysis of Behavior, 4, 7-16 . Bijou, S . W., & Peterson, R . F. (1971) . The psychological assessment of children : A functional analysis . In P. McReynolds (Ed .), Advances in psychological assessment (Vol. 2, pp. 63-78) . Palo Alto, CA : Science and Behavior Books. Bijou, S . W., Peterson, R . F, & Ault, M . H. (1968) . A method to integrate descriptive and experimental field studies at the level of data and empirical concepts . Journal of Applie d Behavior Analysis, 1, 175-191 . Bijou, S . W., Peterson, R. F, Harris, F R., Allen, A. K., & Johnston, M . S . (1969) . Methodology for experimental studies of young children in natural settings . Psychological Record, 19, 177-210 . Birnbrauer, J . S ., Wolf, M . M ., Kidder, J. D., & Tague, C . E . (1965) . Classroom behavior of retarde d pupils with token reinforcement . Journal of Experimental Child Psychology, 2, 219-235 . Dmitriev, V. (1997) . Tears and triumphs . Seattle, WA : Peanut Butter Publishing . Evans, E . D. (1975) . Contemporary influences in early children education (Rev . ed ., pp . 147-200) . New York: Holt, Rinehart, and Winston . Ferster, C . B . & DeMyer, M. K . (1961) . The development of performances in autistic children in a n automatically controlled environment. Journal of Chronic Diseases, 13, 312-345 . Fuller, P. R . (1949). Operant conditioning of a vegetative human organism. American Journal of Psychology, 62, 587-590 . Goetz, E . M., & Allen, K. E. (Eds .) . (1983) . Early childhood education: Special environmental, policy, and legal considerations . Rockville, MD : Aspen. Greenspoon, J. (1955) . The reinforcing effect of two spoken sounds on the frequency of two responses. American Journal of Psychology, 68, 409-416. Harris, F. R ., Johnston, M . K ., Kelley, C . S ., & Wolf, M . M. (1964) . Effects of positive social reinforcement on regressive crawling of a nursery school child . Journal of Educationa l Psychology, 55, 35-41 . Harris, F R ., Wolf, M . M ., & Baer, D . M. (1964) . Effects of adult social reinforcement on chil d behavior. Young Children, 20, 8-17 . Hawkins, R . P., Peterson, R. F, Schweid, E ., & Bijou, S . W. (1966) . Behavior therapy in the home : Amelioration of problem parent-child relations with the parent in the therapeutic role . Journal of Experimental Child Psychology, 4, 99-107. Hayes, S . C ., & Hayes, L . J. (1989) . The verbal action of the listener as the basis for rule-governe d behavior. In S . C . Hayes (Ed .), Rule-governed behavior: Cognitive contingencies and instructional control (pp . 153-190) . New York: Plenum . Hull, C . L . (1943). Principles of behavior. New York : Appleton-Century. Jensen, W. R., Sloane, H . N., & Young, K . R . (1988) . Applied behavior analysis in education. Englewood Cliffs, NJ : Prentice-Hall. Kantor, J. R . (1946) . The aim and progress of psychology. American Scientist, 34, 251-263 . Kantor, J. R . (1959) . Interbehavioral psychology (2nd rev. ed .) . Bloomington, IN : Principia Press . Kantor, J . R., & Smith, N . W. (1975) . The science of psychology : An interbehavioral survey . Chicago: Principia Press . Kazdin, A. E . (1978) . History of behavior modification: Experimental foundations of contemporary research . Baltimore, MD : University Park Press . Kratochwill, T. R ., & Bijou, S . W. (1987) . The impact of behaviorism on educational psychology. I n J . A. Glover & R . R. Rounding (Eds.), A history of educational psychology (pp. 131-157) . New York: Plenum .


19 3

Latham, G . I. (1993) . The power of positive parenting. Salt Lake City, UT: Northwest Publishing . Lewin, K. (1935) . A dynamic theory of personality. New York: McGraw-Hill . Lewin, K . (1936) . Principles of topological psychology. New York : McGraw-Hill . Lovaas, O . I . (1981) . Teaching developmentally disabled children . Austin, TX : Pro-ed. Marholin, D., & Bijou, S . W. (1977) . A behavioral approach to the assessment of children's behavio r disorders . Child Welfare, 56, 93–106 . Marholin, D ., & Bijou, S . W. (1978) . Behavioral assessment : Listen when the data speak. In D . Marholin, II (Ed .), Child behavior therapy (pp .13–36) . New York: Gardner Press . Miller, N . E. (1944) . Experimental studies of conflict . In J. M . Hunt (Ed.), Personality and th e behavior disorders (Vol. 1, pp . 431–465) . New York : Ronald. Morris, E . K ., Hursh, D . E ., Winston, A . S ., Gelfand, D . M ., Hartmann, D . P., Reese, H. W., & Baer, D. M. (1982) . Behavior analysis and developmental psychology. Human Development, 25 , 340–364 . Perini, S . (1997) . Psicologia dell'educazione . Bologna, Italy : Societa editrice it Mulino . Perini, S ., & Bijou, S . W. (1993) . Lo sviluppo del bambino ritardato. Milan, Italy : Franco Angeli. Rosales-Ruiz, J., & Baer, D. M . (1996). A behavior analytic view of development . In S . W. Bijou & E. Ribes (Eds .), New directions in behavior development (pp . 155-180) . Reno, NV: Context Press . Sears, R. R. (1947) . Influence of methodological factors on doll play performance . Child Development, 18, 190–197 . Sears, R. R ., Whiting, J . W. M ., Nowlis, V., & Sears, P. S . (1953) . Some child rearing antecedents of aggression and dependency in young children . Genetic Psychology Monographs, 47, 135–234 . Shaffer, L. F. (1936) . The psychology of adjustment: An objective approach to mental hygiene . Boston: Houghton Mifflin . Skinner, B . F. (1953) . Science and human behavior. New York: Macmillan. Skinner, B . F. (1957) . Verbal behavior. New York: Appleton-Century-Crofts . Skinner, B . F. (1973) . Some relations between behavior modification and basic research . In S . W. Bijou & E . Ribes-Inesta (Eds .), Behavior modification : Issues and extensions (pp . 1–6). New York: Academic Press. Sloane, H . W., Jr. (1976). Classroom management : Remediation and prevention. New York : John Wiley & Sons . Sloane, H . W., Jr., & MacAulay, B . D . (1968) . Teaching and the environmental control of verba l behavior. In H . N . Sloane, Jr., & B . D. MacAulay (Eds .), Operant procedures in remedial speec h and language training (pp. 3-18) . Boston : Houghton Mifflin. Soriano, C . L . (Ed .) . (1996). Manual de psicologia clinica infancia y adolescancia . Valencia, Spain : Promolibra . Sulzer-Azaroff, B ., & Mayer, G . R . (1986) . Achieving educational excellence . New York : Holt , Rinehart & Winston. Tawney, J . W., & Gast, D . L . (1984) Single subject research in special education . Columbus, OH : Charles E. Merrill . Wahler, R. G., & Fox, J. J . (1981) . Setting events in applied behavior analysis : Toward a conceptual and methodological expansion . Journal of Applied Behavior Analysis, 14, 327–338 . Wahler, R . G ., Winkel, G . H ., Peterson, R . F., & Morrison, D . C . (1965) . Mothers as behavior therapists for their own children . Behaviour Research and Therapy, 3, 113–124 . Watson, J . B . (1919) . Psychology from the standpoint of a behaviorist . Philadelphia : Lippincott. Watson, J . B . (1930) . Behaviorism (Rev. ed.) . Chicago : University of Chicago Press . Wolf, M . M., Risley, T. R ., & Mees, H. (1964) . Application of operant conditioning procedures to the behavior problems of an autistic child . Behaviour Research and Therapy, 1, 305–312.


8 TELEOLOGICA L BEHAVIORIS M

HOWARD RACHLI N

State University of New York

I . Final Cause s A. The Relation of Final Causes to Their Effects B. Sensation and Imagination in De Anim a II . Causation in Physics III . Causation in Classical and Modem Psychology IV. Causation in Skinnerian Psychology V. Teleological Behavioris m A. Teleological Behaviorism and Modern Philosophy of Psychology B. Utility Functions as Final Causes References The object of this chapter is to present a view of final causes (teleological explanations), as conceived by Aristotle (tr . 1984), to show how a form of mode m behaviorism, which I call teleological behaviorism, makes use of final causes , and to argue that this form of behaviorism provides both a vehicle for predictio n and control of behavior and a potential meaning for mental terms at least as useful as that potentially provided by physiological or cognitive psychology . ' 2 For a detailed exposition of current work in the areas discussed in this chapter and its relation t o language, consciousness, and other psychological issues, see Rachlin (1989, 1991, 1994, 1996 , 1997) . Handbook of Behaviorism

1 95


196

RACHLI N

There is much disagreement among modem philosophers of science abou t what is a final cause and about what Aristotle meant by final cause or teleological explanation. (See Davidson, 1980, Donagan, 1987, Dretske, 1988, Ringen , 1985, Staddon, 1973, Taylor, 1964, and Wright, 1976, for discussions of causation in psychology and modem interpretations of Aristotle.) The details of th e philosophical debate are far beyond the scope of this chapter, therefore the pre sent exposition may be regarded less as an interpretation of Aristotle's actual views than as a reconstruction of those views for the purposes of modem behaviorism . According to at least one contemporary Aristotelian philosopher (Randall , 1960, p. 66) : "For Aristotle life or psyche is the behavior of the organism as a whole in its environment . Aristotle is thus a thoroughgoing behaviorist." An d Kantor (1963, p . 161), considers behaviorism the modem movement closest t o Aristotle's "naturalistic" psychology . Thus there is at least some basis in philosophy for the present approach . I . FINAL CAUSE S Aristotle's concept of cause was much wider than the modem one . A causal explanation of a process, for him, was an answer to a question about th e process whatever might follow the word because in a sentence (Hocutt, 1974; Randall, 1960, p . 124) . Aristotle refers to four types of causes : material, formal, efficient, and final . Material and formal causes explain the nature of "substances" static objects ; efficient and final causes explain the dynamic behavior of objects, inanimate objects, and organisms . Although I am mainly concerned here with dynamic processes, hence efficient and final causes, it is worth noting the relation between the two static causes . For Aristotle a substance consists of matter taking on a certain form. Thus a circle as a substance might be said to consist of a piece of cardboard taking o n the algebraic form a2 + b 2 = r 2 (an example Randall attributes to Spinoza) . We may say, "This piece of cardboard is a circle because it has the for m a2 + b 2 = r 2 ." But no substance ever perfectly fits its form (no piece of cardboard can be a perfect circle) . For Aristotle, forms are really categories int o which various material objects may be classified ; science is the process of classification . Randall (1960, p . 51) quotes John Dewey as follows : "Classification and division are [for Aristotle] counterparts of the intrinsic order of nature ." Now let us consider the two dynamic causes, efficient and final . Final cause s are to efficient causes as formal causes are to material causes . In our modem way of thinking causes precede their effects . Aristotle's efficient causes do precede their effects . For instance, Aristotle says in De Anima (book II, chap . 12 , para. 424a) that our sense organs are affected by forms of objects "the way in which a piece of wax [the organ] takes on the impress of a signet ring [the form

8 TELEOLOGICAL BEHAVIORISM

19 7

of the object] without the iron or gold [the matter] ." When discussing the effect s of objects on sense organs Aristotle does use efficient-cause explanations . Bu t most of De Anima ("On The Soul") is devoted not to this subject but to the relation between objects and whole organisms. Such processes are labeled with the familiar terms sensation, perception, imagination, and thought and are explaine d in terms of final rather than efficient causes . The relation of final causes to their effects parallels the relation of formal causes to their effects —a final cause is a form of movement (or classification o f movement) abstractly conceived ; its effects are the particular movements that comprise the form (or fit into the classification) . A . THE RELATION OF FINAL CAUSES TO THEIR EFFECT S

A final cause is to its effects as a wider concept (e .g ., a dance) is to the particulars (e .g ., the steps) that make it up . Since all classes must have members (fo r Aristotle but not for Plato) all causes must have effects . The difference between a final and a formal cause is that the effects of a final cause are movements (lik e steps in a dance) while the effects of a formal cause are material objects (such a s pieces of cardboard) . A final cause like dancing or playing golf or playing a Mozart sonata may be said to "embrace" its particular effects (steps, shots , movements) . Instrumental acts not embraced by the larger concept (like payin g the fiddler, buying a club, or entering a concert hall) are not inclusive effects o f that final cause but they may be effects of a still wider, still more abstract fina l cause. Analysis of final causes yields ends, ends that consist of abstract patterns of the movements that comprise them, ends that embrace those patterns . Final causes are not simply efficient causes in reverse . An effect of an efficient cause follows its cause but an effect of a final cause does not precede its cause ; it fits into its cause . In some sense a particular movement must occur first in order fo r a pattern of movements to emerge just as the movements of a sonata must be played before the sonata can be said to have been played . In that sense and i n that sense only a final cause follows its effects . Loosely speaking, efficient causes are answers to the question, "How does this or that movement occur?" Analysis of efficient causes ultimately yield s "mechanisms" that may range from simple billiard-ball–like interactions to complex computer circuits to complex neurochemical processes (Staddon , 1973) . Correspondingly, final-cause analyses are attempts to answer the question, "Why does this or that movement occur?" Of what more molar proces s does this particular movement form a part? Answers to the question how, regard less of their completeness, do not automatically answer the question why. Ac cording to Aristotle (Physics, book II, chap . 7, para . 198a) it is the business o f physicists to know all the causes of their object of study . Since psychology, ac cording to Aristotle, is a branch of physics, he would argue that psychologist s ought to know all the causes, final as well as efficient, of the behavior of organ-

198

RACHLI N

isms . (But he also says that formal and final causes are much more importan t than material and efficient causes particularly in psychology.) To illustrate how final causes explain mental terms let us consider Aristotle's accounts of sensation and imagination. B . SENSATION AND IMAGINATION IN DE ANIMA

A particular sensation is conceived by Aristotle as a particular instance of a discriminative pattern . The difference between the pattern of a person's action s (the person's soul) before sensation and afterward is not that the soul contain s something it did not have before but that afterward any complete description o f the person's actions must contain a description of aspects of the object (its sensory qualities) . What has the power of sensation [a whole living animal] is potentially like what the perceived object is actually ; that is, while at the beginning of the process of its being acte d upon the two interacting factors [the behavior of the animal and the qualities of the sensible object] are dissimilar, at the end the one acted upon [the behavior of the animal] is assimilated to the other [the color, smell, sound, etc. of the object] and is identical in qualit y with it. (De Anima II, chap . 5, para . 418a) .

The process Aristotle is describing is behavioral discrimination . Objects may be classified according to their colors (say) . The animal's discrimination enables an observer of the animal to classify its behavior in the same way (by the colors o f objects) . Sensing the color of an object is the act of discriminating between that colo r and others . If a driver generally stops at red lights and goes at green lights, th e driver may be said to discriminate between red and green . Given that this discrimination (and other discriminations) between red and green do generally occur, then, in a particular instance, when the driver actually does stop at a re d light, we may say that the driver sensed the color of the light. We can then classify that bit of behavior under that particular discrimination . Doing so (making the classification) is an act (our act) of matching behavior and color. Stopping at red, saying "red," underlining things in red, and so forth are discriminations o f the same sort . The investigation and analysis of sensation is the investigation an d analysis of this process (of discrimination) . Most misunderstandings of behaviorism by philosophers and psychologist s alike arise at this fundamental level . For example, being deaf, according to a (teleological) behaviorist, consists of failing to discriminate in overt behavio r (verbal and nonverbal) between sounds and nonsounds . Because such discrimination requires a correlation over time of many individual events, it is not possible to tell if someone is deaf (or if you yourself are deaf) at a moment in time . On the other hand, for a mentalist, a cognitivist, a physiologist, hearing a soun d is fundamentally instantaneous ; it is a "sensation" consisting of a single spiritual or physiological event . Mentalists, cognitivists, and physiologists assume tha t behaviorists must identify sensations with individual behaviors (like a rat's lever


19 9

press) . They therefore pose to the behaviorist what they believe to be unanswerable questions such as the following : "What is the difference between two awak e individuals, one of them stone deaf, who are both sitting immobile in a room i n which a record-player is playing a Mozart string quartet?" (Jeffrey Gray, as quoted by Staddon, 1993, p . 123) . The answer is that the correlation betwee n sounds and the behavior of the deaf person is zero (this is what it means to be deaf), while the correlation between sounds and the behavior of the other perso n is nonzero ; Gray has merely pointed to two corresponding momentary events in two entirely different correlations . Hearing a sound is not a momentary event like a rat's lever press but a temporally extended event like a rat's rate of lever pressing . A rat currently pressing a lever 100 times per minute may at a given instant be "immobile," jus t as a hearing person may at a given instant be "immobile" in the presence of a sound. What differentiates this rat from another rat (not pressing the lever at all ) is not behavior at an instant but behavior over time . Similarly, what differentiate s the hearing from the deaf person is behavior over time . Behavioral difference s over time define hearing and deafness . If a person's behavior correlates wit h sounds over time, that person can hear, regardless of what goes on in his or her consciousness or for that matter in his or her brain . This is Aristotle's point an d it is the main point of teleological behaviorism . This theory holds for perceptio n as well as sensation . The crucial difference between sensation and perception, according to Aristotle, lies not in the internal mechanisms mediating the tw o processes, not even in the animal's movements, but in the complexity of th e stimulus that classifies the movements . A pigeon discriminating between the presence and absence of a red light, by pecking faster or slower, is sensing (more precisely, seeing) that light . But a pigeon discriminating between the presenc e and absence of a person in a picture, by pecking faster or slower (Herrnstein , Loveland, & Cable, 1976), is perceiving those people . (For Aristotle, it is only i n creative thought that behavior becomes more complex .) Aristotle's conception of imagination follows from that of sensation and perception . Aristotle says, "imagination must be a movement resulting from an actual exercise of a power of sense" (De Anima, book III, chap . 3, para. 429a) . That is, as far as the actual movements of the animal are concerned, imaginatio n is the same as sensation or perception . The difference is that the discriminate d object is present in the world during sensation or perception, while during imagination the object is not present in the world. Note that Aristotle does not say that the objects present in the world during sensation and perception (which these activities discriminate among) are present inside the animal (as representations, internal images, neural discharges, or anything else) during imagination . What he does say is that the movements the animal makes during imagination with the objects absent are the same as those the animal makes during sensation or perception with the objects present . If you generally behave one way in the presence of, and another way in the absence of, red lights you are discriminating between red lights and other things . But if, on an occasion, you behave in the

200

RACHLI N

absence of a red light as you normally do in its presence you are, on that occasion, imagining a red light . For Aristotle, imagining is acting not dreaming : vividness of imagination is not vividness of interior image but of overt behavior. Suppose two people are asked to imagine a lion present in the room . One closes her eyes and says, "Yes , I see it, a mane and a tail, that's right, it's walking around," and so forth . The other runs screaming for the door. For Aristotle the first person is not imagining a lion but imagining a picture of a lion . The second person is really imaginin g the lion . The location, intensity, orientation, or even existence of an image in tha t person's head would be entirely irrelevant to the process . A good imagination i s not just an aid or a tool in good acting . Rather for Aristotle good acting is goo d imagining . Aristotle's final-cause analysis of perception, memory, and creative thought i n De Anima, and his discussion of freedom, responsibility, and all of ethical behavior in Nichomachian Ethics, follows a similar form —a particular act is identified as a particular perception or thought, as free or unfree, as good or bad, no t on the basis of the particular internal acts (spiritual, cognitive, or physiological) that may efficiently cause it, but rather on the basis of the temporally extende d pattern of overt behavior into which the particular act fits that is, on the basi s of its final cause . As regards introspective knowledge of one's own mental states, one Aristotelian interpreter says : Aristotle has no reason to think that psychic states—perceptions, beliefs, desires—mus t be transparently accessible to the subject, and to him alone . Even if there are such states , this feature of them is not the feature that makes them psychic states . Psychic states, for human souls as for others, are those that are causally relevant to a teleological explanation of the movements of a living organism. (Irwin, 1980, p. 43 )

II . CAUSATION IN PHYSIC S For Aristotle, the motion of all physical objects is explicable in terms of fina l as well as efficient causes . (The other two causes, formal and material, appl y only to static objects .) A stone, according to Aristotle, naturally moves toward th e center of the earth unless hindered. As long as the stone does indeed move toward the center of the earth its movement may be explained in terms of its own ends without reference to other substances . But if the stone should be thrown up in the air by a boy or a volcano, the movement of the stone would have to be explaine d according to ends other than those ascribable to it alone those of the boy or the volcano (in that sense the stone was not "free") . Talking about the ends of a stone's movement sounds odd to us because the success of Renaissance physic s with its rigorous insistence on efficient causes has given us our modern conception of what does and does not constitute proper scientific language .


20 1

According to Rorty (1982, p . 191) : Galileo and his followers discovered, and subsequent centuries have amply confirmed , that you get much better predictions by thinking of things as masses of particles blindl y bumping each other than by thinking of them as Aristotle thought of them animistically , teleologically, and anthropomorphically. They also discovered that you get a better handl e on the universe by thinking of it as infinite and cold and comfortless than by thinking of it as finite, homey, planned, and relevant to human concerns. Finally, they discovered that i f you view planets or missiles or corpuscles as point-masses, you can get nice, simple predictive laws by looking for nice simple mathematical ratios . These discoveries are the basis of modern technological civilization . We can hardly be too grateful for them . But they do not, pace Descartes and Kant, point out any epistemological moral . They do not tell us anything about the nature of science or rationality. In particular, they did not result from the use of, nor do they exemplify, something called "scientific method. "

In this passage Rorty implies that in physics final-cause explanations have les s predictive power than efficient-cause explanations . Whatever their weakness in physics, this chapter will argue that in psychology final-cause explanations may have as much or more predictive power than efficient-cause explanations . Even in modern physics, final causes have a part to play . According to Max Planck, the founder of quantum theory : The cause efficiens, which operates from the present into the future and makes future situations appear as determined by earlier ones, is joined by the cause finalis for which, inversely, the future—namely a definite goal—serves as the premise from which there can be deduced the development of the processes which lead to this goal . (Quoted in Yourgrau and Mandelstam, 1968, p. 165 )

If efficient causes have proven to be inadequate to explain physical phenomen a at the most fundamental level they may be expected to be still less adequate i n psychology . Therefore an attempt at final-cause explanation in psychology i s worthwhile . (See Silverstein, 1988, for a similar argument . )

III . CAUSATION IN CLASSICAL AN D MODERN PSYCHOLOG Y Just as Renaissance physics replaced Aristotelian final causes with efficien t causes, Descartes attempted to replace final causes of human and nonhuman animal behavior with efficient causes . According to Descartes, all nonhuman behavior is efficiently caused by external stimuli . In Descartes's model a stimulus acting through a sense organ is mechanically transmitted through the nerves to th e brain whereupon animal spirits (supposed to be a material cause of life) are re leased to flow back through the nerves to expand the muscles and cause movement. The difference between humans and other animals is that in addition to this completely automatic mechanism, which causes involuntary behavior, humans have another way to direct animal spirits to various nerves their will, which, acting through the pineal gland in the brain, can alter the direction of the

202

RACHLI N

flow of animal spirits . Whether a person's behavior is voluntary or involuntary i t is always, according to Descartes, efficiently caused . Voluntary behavior is efficiently caused by the will while involuntary behavior is efficiently caused directly by external stimulation . Thus Descartes brought Renaissance psycholog y into line with Renaissance physics . Contrary to Aristotle's conception of the nonprivacy of mental states (see previous quote from Irwin, 1980), for Descartes the essential fact about menta l states is their privacy . "I think therefore I am" is a pronouncement about what i s "clear and distinct" in the privacy of the mind . The existence of anything els e one's own body, its behavior, and the bodies and behavior of all other thing s rests, for Descartes, on this internal and private clarity. Since Descartes, psychology has, in various ways, attempted to explain th e action of the will . One method has been to incorporate all behavior, voluntary a s well as involuntary, into more or less complicated reflex systems . Sechenov' s (1863/1965) Reflexes of the Brain and Pavlov's (1927) Conditioned Reflexes rep resent such systems . According to Sechenov, "the initial [efficient] cause of all behavior always lies not in thought, but in external sensory stimulation, without which no thought is possible" (p . 321 emphasis in original) . The American behaviorists Watson (1913), Guthrie (1935), Hull (1952), and Tolman (1949) differed strongly on how to describe the internal efficient cause s of behavior, but their ultimate object, no less than that of Sechenov and Pavlov , was to explain behavior in terms of its internal efficient causes, to get behind behavior itself to discover its underlying mechanisms . Disputes between thes e behaviorists and between behaviorists and cognitive psychologists, as well a s between various schools of cognitive psychology have centered around th e question, what are the internal efficient causes of behavior? At one extreme , Sechenov, Pavlov, and their modern descendants (for example, Rescorla, 1988 ) attempt to trace the "reflexes of the brain ." The connectionists Thorndike (1911 ) and Hull (1952) spoke more abstractly in terms of internal stimulus-respons e (S –R) connections or associations perhaps eventually reducible to physiologica l reflexes . More molar behaviorists such as Tolman (1949) abandoned the possibilit y that mental constructs could be reduced to reflexes but still retained the goal o f describing internal constructs (if not mechanisms) . In Tolman's hands S –R psychology became S-O-R psychology where the 0, standing for organism, was conceived (at least by Tolman's cognitive followers) as a mediating efficient cause between environment and behavior . The cognitive psychologist John R . Anderson (1991, p . 513) says : "I have always felt that something was lost whe n the cognitive revolution abandoned behaviorism ." But he goes on, "In doing this , however, I do not want to lose the cognitive insight that there is a mind betwee n the environment and behavior." (If by between Anderson meant only logical inclusiveness, as in "there are secrets between you and me," he would be identifying cognitivism with teleological behaviorism. But Anderson more likely mean s between to stand for physical separation, as in "there is baloney between the tw o slices of bread .")


20 3

For some modern philosophers of psychology a given mental state may be internally represented by the action of a given computer mechanism consisting o f individual components, none of which may actually represent the mental state . Dennett (1978) calls such mechanisms "intentional systems ." As a very simple example of such a system Dennett (1978, pp . 71–89) cites Thorndike's law of effect. In the operation of the law of effect, reinforcement strengthens a modifiable connection, causing behavior that appears purposive despite the fact that th e organism's purpose itself has no coherent internal representation . Another example of an intentional system would be a chess-playing computer that, say, had a tendency to bring out its queen too soon even though no such tendency was explicitly programmed in the machine . Mental states, according to this sort of cognitive psychology, may be emergent qualities of behavior . Research on "neural networks" (Grossberg, 1982) are perhaps the clearest example of how comple x cognitive and behavioral processes may emerge from the concatenation of muc h simpler efficient causes . Another sort of modern cognitive philosophy (Fodor, 1981) insists that all mental states are internally represented as such and interact with each other i n the mind to cause behavior. According to Fodor (1981, p . 5) "mental causes typically have their overt effects in virtue of their interactions with one another" (emphasis in original) . Psychology would consist of the analysis of such interactions . Philosophers of this latter cognitive school frequently (and not without a degree of justification) accuse those of the former of being behaviorists . The issue within classical behaviorism as well as between classical behavior ism and modern cognitive psychology (of either school) is not whether psychology consists of the analysis of an efficient-cause mediating process between environment and overt behavior ; all follow Descartes on the necessity of a n efficient-cause analysis . The issue is whether mental terms are necessary in suc h an analysis . In general, aside from Skinner, behaviorists and cognitivists alike hav e viewed psychological theory as about efficient causes mediating between environment and behavior . To Skinner we owe the renaissance of the Aristotelian focus on the behavior of whole organisms in their environments . IV. CAUSATION I N SKINNERIAN PSYCHOLOG Y

In the more than half century between his first published paper and his deat h in 1990 Skinner's views naturally varied considerably . It is not an object here t o trace that development . One thread that runs throughout his thought, however, i s his ultimate concern with behavior of the organism as a whole. The concept of the operant as a class of individual movements of whole organisms having a common effect, the causal power attributed to contingencies of reinforcement of operants, the weight assigned to an organism's and a species' history of reinforcement, are strains of Aristotelian thought in Skinner's psychology .

204

RACHLI N

A contingency of reinforcement (a broader term for schedule of reinforcement), for instance, may be conceived as an Aristotelian final cause of the individual acts of which it is comprised . Note that an individual reinforcer considered in isolation could not be a final cause ; a reinforcer strictly follows the act it causes . A reinforcement contingency includes the act it causes thus a contingency may be a final cause . However, while Skinner was never deeply concerned with the logic or philosophy of the concept of causation, there are signs that he was not comfortabl e with teleological explanation as such . Skinner (1938, p . 69) held that an individual behavior-reinforcer contiguity served as the (efficient) cause of subsequent response-rate increases . His concept of "superstition " rests on adventitious contiguities as (efficient) causes of future behavior. Even Skinner's most molar construct, "history of reinforcement," is construed as existing wholly in the past . Skinner saw a person's reinforcement history as the efficient cause of his or he r subsequent behavior rather than the final cause of the behavior comprising th e history . As part of his discussion of nutrition and reproduction Aristotle (De Anima book II, chap . 4, para . 415b) distinguishes between the soul as an efficient caus e and the soul as a final cause . The soul may be the efficient cause of behavior only in the sense that parents, by their reproductive activity, pass their behavioral characteristics to their offspring. That is, the reproductive (the most basic) par t of the parents' souls efficiently causes their offspring's souls . But within a person's lifetime his or her soul is the final not the efficient cause of behavior. An individual's "history of reinforcement" (Skinner's concept closest to Aristotle' s "soul") thus would have been viewed differently by Aristotle and Skinner. Fo r Skinner it was an efficient cause of behavior. Aristotle would have viewed it no t as an efficient cause, preceding any particular act, but as a final cause embracin g the act . Skinner's conception of reinforcement has seemed to some modern commentators (Ringen, 1985 ; Staddon, 1973) to involve efficient causation but wit h temporal gaps . An individual response-reinforcer contiguity may immediately increase response rate, but then what stands between one response and another? Certainly a history of reinforcement, considered as an efficient cause of tempo rally remote behavior, must act through some (presumably internal) mechanism . In referring to this possibility Staddon (1973, p . 55) says : I suggest that the apparent simplicity of the relation between operant behavior and its consequences implies not an absence of mechanism, not that operant behavior is "undifferentiated material," in Skinner's phrase, but rather the existence of more and riche r mechanisms than have hitherto been seriously contemplated. . . . Explanations in terms of "purpose" or "final cause" are always incomplete .

This statement is true as far as it goes . But Staddon does not add that explanations in terms of efficient cause (mechanisms in Staddon's terms) are also incomplete . Good automobile mechanics (if such exist) are not necessarily good dri-


20 5

vers (may not be able to predict and control the movements of the car as a whole) . The price Skinner paid for the rejection of teleology was a simultaneous rejection of mental terms and acceptance of private inner causes (Zuriff, 1979) . Day (1969, p . 318) referred as follows to an exchange between Skinner and th e philosopher Michael Scriven : "Skinner is objecting . . . not to things that are private but to things that are mental ." Here is Skinner's (1953, p . 279) not atypical explanation of the causal status of an idea : "If the individual himself reports Ì have had the idea for some time but have only just recently acted upon it,' h e is describing a covert response which preceded the overt ." For Aristotle, an idea would not be a covert response at all, but a pattern of wholly overt responses including the individual's verbal report as one particular part of the pattern . In relegating mental concepts such as ideas to the interior of the organis m (the same place where the organism's physiology exists), Skinner is in line wit h modern cognitive psychology . Both Skinner, in the preceding passage, and muc h of modern cognitive psychology place mental and physiological events togethe r ("cognitive neurophysiology") inside the organism . The difference is that fo r Skinner such internal events are unimportant simply because they are internal (they are not controlling variables) . Skinner says (1953, p . 35) : The objection to inner states is not that they do not exist, but that they are not relevant in a functional analysis . We cannot account for the behavior of any system while staying wholly inside it; eventually we must turn to forces operating upon the organism fro m without. Unless there is a weak spot in our causal chain so that the second link is not law fully determined by the first, or the third by the second, then the first and third links mus t be lawfully related. If we must always go back beyond the second link for prediction and control, we may avoid many tiresome and exhausting digressions by examining the thir d link as a function of the first. Valid information about the second link may throw ligh t upon this relationship but can in no way alter it.

For modern cognitive psychologists on the other hand internal events are critically important their study is the essence of psychology itself . Consider the following from Dennett (1978, p . 154, footnote) regarding the philosophers Ryl e and Wittgenstein who like Aristotle identified mental terms with molar actions o f whole organisms : Ryle and Wittgenstein are the preeminent modern theorists of the personal [whole organ ism] level . In fact, in their different ways they invent the enterprise, by showing that ther e is work to be done, that there are questions that arise purely at the personal level, and tha t one misconceives the question if one offers sub-personal [i .e ., cognitive] hypotheses or theories as answers . Typically readers who do not understand, or accept, these difficul t claims see them as evading or missing the point, and complain that neither Ryle no r Wittgenstein has any positive psychological theory to offer at all . That is true : the personal level "theory" of persons is not a psychological theory.

Skinner (1938) claimed that there are only two basic kinds of behavior : respondents, classes of behavior (like a person's pupillary dilation or a dog's salivation) elicited by immediate antecedent stimulation, and operants, classes o f behavior correlated with immediate environmental consequences . Examples of

206

RACHLI N

operants are rats' lever presses, pigeons' key pecks, and all animal behavior normally considered voluntary. Aspects of the environment crucial for respondent and operant dynamics are antecedent stimuli of respondents, reinforcers of operants consequences that generally increase an operant's rate of emission (operant conditioning being the study of the relationship of operants and reinforcers) and discriminative stimuli, in the presence of which a give n operant–reinforcer relation obtains . An example of a discriminative stimulus i s the open–closed sign on the door of a shop, signaling a given relation betwee n door pushing (the operant) and door opening (the reinforcer) . Nowhere here are mental terms . Sometimes Skinner (1953) offered "interpretations " in which the use of mentalistic vocabulary in everyday speech is explained in terms of operants, reinforcers, and discriminative stimuli . Self-control, for instance, is held to be nothing but operant avoidance of certai n discriminative stimuli ("get thee behind me Satan"), like crossing the street t o avoid the enticing smell of a bakery. Skinner's nonmentalistic terminology has served very well in the analysis o f discrete operants like pigeons' key pecks, rats' lever presses, and humans' butto n pushes (Honig & Staddon, 1977) . Furthermore, the patterns of behavior discovered in one situation with one species often appear in other situations with othe r species . These patterns change in systematic ways with motivational variable s like reinforcer deprivation and drug dosage. Skinnerian techniques have bee n very successful in areas of behavior therapy ranging from treatment of sever e psychoses to weight control ; their great advantage in these applications is thei r resolute focus on consequences contingencies of reinforcement . For example , many women (and some men) suffer from agoraphobia they are housebound—they have panic attacks in public places and consequently refuse t o leave home . In searching for causes and treatment of such behavior the Skinnerian behavior therapist considers not just its antecedents but also its consequences—avoidance of work, avoidance of sexual temptation, attention fro m relatives and friends, and so forth . Focusing on the actual consequences of dysfunctional behavior has led in many cases to the development of successful treatment—by substitution of less dysfunctional behavior to achieve equivalent ends . Skinnerian techniques have been successfully applied also in business management and in areas normally considered cognitive, like the teaching of readin g and mathematics to children and in college-level courses as diverse as anatom y and foreign languages . However, despite this success, it has not been possible either in the operant laboratory or in the many areas of application of Skinnerian behaviorism to di vide all behavior neatly into specific respondents and specific operants . A respondent must be correlated with an antecedent stimulus and an operant with a consequent reinforcer. What, for instance, reinforces the act of refusing an offered cigarette by a smoker trying to quit? Having to deal with and talk abou t such obviously important acts, behavior therapists have taken two roads, neithe r satisfactory.


20 7

Some, like Homme (1965), have developed an operant psychology of the hid den organism, speaking of inner (covert) respondents, inner operants ("coverants"), and inner discriminative stimuli . According to these psychologists th e person who refuses an offered cigarette can just reinforce the act himself (pa t himself on the back, so to speak) . This conception has both logical and empirical problems . Logically, if a person can reinforce her own actions why should sh e ever withhold reinforcement of any action? What reinforces the giving and withholding of self-reinforcement (Catania, 1975)? Empirically, there is just no evidence that self-reinforcement works and some evidence that it does not work (Castro & Rachlin, 1980) . The other road taken by behavior therapists has led to cognitive behavio r therapy (Mahoney, 1974) . Cognitive behavior therapists retain Skinnerian techniques for acts that are clearly reinforced . But, where environmental reinforcer s are not obvious or immediate, cognitive behavior therapy abandons behavioris m entirely and refers to mental states as inner causes . Thus the person who refuse s the cigarette may do so because he believes it is better for his health and becaus e he wants to be healthy. A therapist might then try to strengthen the person's be lief and desire by logical argument or by asking him to repeat a statement of hi s belief over and over or by reinforcing the statement of the belief . Even this last procedure is cognitive, not behavioral, because it rests on the assumption that th e statement is merely evidence of an internal state and that the reinforcement acts not only on the external statement but also on the inner belief . After all it is th e refusal of cigarettes (what the belief is said to cause) rather than the verbal statement that the therapist is ultimately trying to strengthen . In principle there i s nothing wrong with cognitive behavior therapy. If people do have beliefs as inner states and if beliefs can cause specific actions, then changing the belief wil l change the action . From a viewpoint that sees mental terms as descriptions of final causes, how ever, the cognitive behavior therapists are making what Ryle (1949) calls a "category mistake" If a belief is nothing but a pattern of actions, then a statement o f a belief is merely one of those actions . Altering the statement would affect th e belief only to the extent that it affects one part of the pattern, not the centra l source of all parts . Beethoven's fifth symphony, for instance, is a pattern of note s with four very familiar notes at the beginning . When you hear those notes from an orchestra you can be pretty sure that you are about to hear the rest . To alter those four notes is to seriously alter Beethoven's fifth symphony (perhaps to th e extent that you would want to call it something else) but not necessarily to hav e any effect on the other notes . But then again, there may indeed be a central state, more or less coherentl y represented in the nervous system, more or less innate, that controls all the behavior that an outside observer would call evidence of a person's belief. Altering a belief (like, say, substituting the score of Schubert's ninth for Beethoven' s fifth) would then alter all of its behavioral effects . Cognitive behavior therapist s are thus trying to get at the central antecedents, the efficient cause, the core, the

208

RACHLI N

nub, the nut, the origin, the control room of belief behavior, the very essence of a person's belief. The problem is that they have abandoned the aspect of Skinner' s program that made it so successful its concentration on consequences rather than antecedents . A therapist who focuses on the central efficient causes (th e how) of a person's belief tends to lose sight of the reinforcers (what a person i s getting by behaving in this way) what effects the person's belief has on her relations with her family, friends, environment in general the why of the belief . The failure of behavioral psychology to deal with mentalistic concepts suc h as beliefs (which philosophers call "propositional attitudes" or "intentional states") and pains (which they call "raw feels") has by and large left treatment o f strictly mental disfunction to cognitive and physiological psychology . A comprehensive behavioral analysis of such concepts does not currently exist, but suc h analyses are possible and may be highly useful . For instance, according to Skinner (1945), a pain is an internal private stimulus ; purely behavioral analysis and treatment can never get at pain itself, only "pain behavior." Thus Skinner essentially leaves the field to critics of behavior ism (e .g ., Putnam, 1980) who argue that behaviorism is inadequate to deal wit h pain itself. Nevertheless from a teleological viewpoint pain and pain behavio r may be construed as one thing . Rachlin (1985) argued that pain treatment base d on the equality of pain and pain behavior (widely construed) has been at least a s effective (and if generally accepted might be more effective) than treatment based on pain as an internal state .

V . TELEOLOGICAL BEHAVIORIS M

Guthrie (1935) distinguished between acts and movements. A movement, ac cording to Guthrie, is a particular set of muscular contractions resulting in a particular locus of bodily displacement . (Guthrie's definition of movement is thu s much narrower than Aristotle's .) An act, according to Guthrie, is a coordinate d pattern of movements leading to some definable result . Waving goodbye, for example, would be an act . Ten different instances of a person's waving goodby e might involve ten wholly different sets of movements but constitute a single act , repeated ten times . Guthrie claimed that the apparent learning of acts was essentially an accident of the learning of particular movements . Skinner's (1938) concept of operant conditioning reversed Guthrie's claim . An operant (an act as Guthrie defined it) may be directly reinforced, accordin g to Skinner, without regard to particular movements . Teleological behaviorism expands Skinner's original concept of reinforcement from a single event dependent on a single operant for instance, a singl e food pellet delivery immediately following a single lever press to a pattern o f environmental events perhaps only vaguely contingent on an overlapping patter n of operants . In Herrnstein's (1969) conception of avoidance, for instance, the ef-


20 9

fective contingency is a negative correlation imposed between the rate of aversive events and the rate of operant emission, not any individual consequenc e (hypothetical or real) contingent on any particular operant . Even so abstract a conception as belief is such a pattern of overt observabl e acts and consequences . Take away the observable acts and the belief goes awa y as well ; it is just as if you were to take away the lights and shadows on th e movie screen. Without lights and shadows, the action, the characters, their motives, their beliefs, all go away as well. Teleological inferences and models are inferences and models about respondent and operant contingencies that may not be present at the moment but ma y serve as the context for current actions (Staddon, 1973) . A grocer may give a poverty-stricken customer a free loaf of bread because the grocer is at heart a sympathetic and generous person or as part of a promotional campaign . The grocer's differing motives form differing potential contexts for (and causes of) hi s act. These contexts differ by virtue of differing respondent and operant contingencies operating over periods of time much wider than the present moment . The question of which motive the grocer really has can be answered only by reference to such overt contextual events : the existence or nonexistence of a promotional campaign and the grocer's behavior when no such promotional campaign is being waged . One important feature of the teleological viewpoint bears emphasis : becaus e the temporal context of a brief event extends into the future as well as the past , the grocer's motives may be determined by reference to future as well as to pas t events . To decide what the grocer's motives really were, you could perform a n experiment post hoc : for example, sending poor customers into the store on different occasions . If the grocer should die immediately after his generous act (an d historical records are lost or nonexistent), there is no way, even in principle, t o determine its behavioral context (the grocer ' s motives) . Supposing contrary to fact, that Beethoven's fifth symphony and Schubert's ninth symphony had thre e identical contiguous notes, it would be like trying to determine which symphon y you were listening to from those three notes alone . A mental act such as a motive has meaning only over an extended span o f time . To sample only a piece of that time is to make a guess as to the motive . A s wider and wider samples are taken the grocer's motive becomes more and mor e knowable, but it will never be 100% knowable ; in principle, its context is infinite . This is true even with respect to the grocer's knowledge of himself . Th e grocer is in a privileged position only by virtue of the quantity rather than th e quality of his information . He has a larger sample of his own behavior than any one else but by no means a better sample . Indeed his information as to his own motives may be much worse than that of an outsider because he can observe hi s own behavior (as a whole person) only by reflection (through its various outcomes), whereas outsiders may view his behavior directly . Thus, the teleological behaviorist turns on its head the common mentalist notion that people have privileged access to their own mental states . In this respect, teleological behaviorism

210

RACHLI N

is like Freudian psychology (although, of course, different in many other respects) . A . TELEOLOGICAL BEHAVIORISM AND MODER N PHILOSOPHY OF PSYCHOLOGY At this point it may fairly be asked why, if Aristotle was really a teleological behaviorist, are modern philosophers, particularly those who consider them selves Aristotelians, so dead set against behaviorism? The answer is not tha t these philosophers misunderstand Aristotle, but they do misunderstand behavior ism, especially its molar character; and they completely ignore teleological behaviorism. However, one point of agreement between modern Aristotelian philosophy and behaviorism is a mutual rejection of introspection and the privat e nature of mental events. Donagan (1987, p. 54) says that "A false and deepl y confused doctrine that was philosophically fashionable is still encountered : namely, that taking propositional attitudes [like beliefs] and persisting in them are items in one ' s flow of private consciousness, which are named by private ostensive definition, and which have complex causal relations with one's bodil y states ." Donagan seems to be arguing against introspection as a method in psychology. Aristotelian philosophers in general do indeed deny the privileged status o f introspection . But Donagan goes on : "It does not follow that description of actions in terms of their doers' propositional attitudes can be analyzed withou t residue in terms of patterns of their surface behavior" (1987, p . 54) . Let us discuss that residue . Donagan does not say that if you removed all of the surface behavior you could still have the residue left . Rather, he says, when we look at the surface behavior we perceive it and describe it in such a way that our perceptions and descriptions cannot be analyzed into strictly behavioral terms just a s you could not analyze the personality of the characters in a movie into lights an d shadows . Then, one must ask, how do we perceive beliefs and other propositional attitudes? The answer, according to Donagan and most contemporar y philosophers, Aristotelians and others, is that we perceive them as arising from inside of the person, ourselves or someone else, who has them . Dretske (1988, p . 3) says : "Behavior is endogenously produced movement , movement that has its causal origin within the system whose parts are moving . " Thus, according to Dretske, acts to which no internal cause is attributable are no t even behavior let alone voluntary behavior. For Dretske overt behavior as w e perceive it comes with a sort of (efficient) causal tail. A rat's lever press, if it i s perceived as behavior, is perceived as being caused by something inside the rat . The lever press itself plus the cause comprise the rat's act . Correspondingly, when we perceive a person's actions as arising from that person's belief, we als o perceive behavior with a tail extending inside the person . But this time the tail wags the person, so to speak. The part of our perceptions that reaches inside th e person contains not just a single internal efficient cause as in the case of a rat's


21 1

lever press but an entire representational system . The person's cognitive stat e consists of the overt act and the internal representational system together . (Dretske's analysis of behavior is thus essentially cognitive .) The representational system (the reason for the act) like the causal tail on the rat's lever press i s the "residue" inevitably left over from any conceivable analysis of surface behavior alone . This is how modern philosophy justifies modern cognitive psychology—internal states as either efficient causes of or less-determinate "reason s for" external behavior. But the present discussion of Aristotle's philosophy suggests that the residue after the most penetrating finite behavioral analysis is just more behavior lik e the residue of uncertainty about the probability of a coin being unbiased after X number of tosses . There is inevitably a residue in this case, but it exists becaus e an ideal analysis would take an infinite amount of time not because the residu e exists in another place . There is no question that sometimes, in our society, in our linguistic environment, we attach inner causes to external behavior . Such attachment may harmlessly be called a branch of "folk psychology," which then may be said to have developed into modern cognitive science (Stich, 1983) . But as J . R . Kantor (1963) argued, there is another branch of folk psychology, a sort of folk behaviorism, that is only satisfied when an act is explained in terms of its consequences . Our folk psychology tells us to be more satisfied with an explanation o f a murder, for instance, in terms of consequences such as money or sex than i n terms of inner compulsions . When both sorts of explanation are available w e tend to choose the former. It is only when there is no clear and direct environmental consequence that we are driven to the latter. But since in such cases w e try to invent a plausible internal representational system, as cognitive psychology is set up to do, why is it not just as valid to try to discover a plausible environmental consequence even if unclear and indirect? This it seems is exactl y what Aristotle did do in corresponding circumstances when he looked for fina l causes as behavioral explanations . Kantor (1963) claimed that a behavioristic folk psychology is the true on e the only one on which a scientific psychology could be built . Why, then, it may fairly be asked, would we ever in our everyday mutual interactions or our technical discourse talk as though mental events were internal efficient causes of ou r behavior? Taylor (1964) blames it on "atomism," which he defines (p . 11) as "the notion . . . that the ultimate evidence for any laws we frame about th e world is in the form of discrete units of information, each of which could be as i t is even if all others were different ." On this basis, Taylor criticizes both modern cognitive psychology ("centrism") and classical behavioral psychology ("peripheralism") . Taylor (1964, p . 12) says : "Teleological explanation is, as has often been remarked, connected with some form of holism, or anti-atomistic doctrine ." However, the sort of holism Taylor advocates differs from the behavioristi c holism advocated here . Taylor's (and recently Dretske's) holistic approach embraces an organism's current internal as well as its current overt behavior

212

RACHLI N

(peripheralism plus centrism) . The teleological behaviorist's holistic approach , on the other hand, embraces an organism's past and future overt behavior as wel l as its current overt behavior. The former is a holism of space, the latter, a holis m of time . B . UTILITY FUNCTIONS AS FINAL CAUSE S

The utility functions discoverable by behavioristic methods (Rachlin, Battalio, Kagel, & Green, 1981) are examples of final causes . Utility functions are essentially mathematical summaries of observed behavior. They stand to a particular choice under particular constraints as the equation of a circle stands to particular circles . The empirical procedure for investigation of utility functions is as follows : 1. Observe behavioral patterns under a selected set of environmental constraints . 2. Infer a utility function from the observed patterns assuming that utility i s maximized within the set of constraints observed or imposed in step 1 . 3. Given the utility function and the assumption of maximization, predict be havior under a new (previously unobserved) set of constraints . 4. Revise the utility function on the basis of deviation between predicted an d actual behavior. Behavior under a set of imposed constraints thus "reveals" the subject's preferences (Samuelson, 1973) . The better the utility functions are known, the bette r behavior is predicted and controlled . Utility functions converge naturally onto parts of the intentional idiom the concepts desire, appetite, will, self-control , value . Whether utility functions will ever provide a precise analysis of other intentional idioms such as belief knowledge, and intelligence, terms that seem to refer to verbal as well as nonverbal behavior, is still (as Lacey & Rachlin, 1978 , argue) an open question. If a behavioral analysis ultimately fails, the failure wil l be due to the complexity of the task . It will be like the failure to precisely predic t and control the weather rather than any intrinsic inaccessibility or opacity of it s subject . The present analysis therefore agrees with modern Aristotelian philosopher s that mental terms (propositional attitudes, dispositional states) form at least par t of our fundamental "data language." The issue is what to do with those data . Should they be analyzed into current overt behavior plus internal causes and reasons as some part of our folk psychology seems to dictate? Or should they be analyzed wholly on the level of overt behavior and its consequences over time a s some other part of our folk psychology seems to dictate? Aristotle might hav e argued that both avenues need to be pursued we should know all of the cause s of our object of study ; success in either, far from blocking the other, will illuminate its path .

21 3


Teleological analyses, because of their molar nature, are necessarily less precise than efficient-cause analyses . A given act may be truly understood only some time, perhaps a considerable time, after it occurs because the context (th e final cause) of an act extends into the future as well as into the past . An individual lever press is uncaused in exactly the same sense that an individual event ha s no probability . A probability (conceived as a relative frequency) may be known as accurately as you want it to be known if you are willing to wait . The same goes for the causes of an individual lever press, and, arguably, the same goes for a person's belief (including one's own belief) . Because the air is currently so full of antibehaviorist polemics from psychologists of mentalistic, cognitive, and physiological orientations as well as fro m philosophers of all orientations (but see Stout, 1996, for a defense of teleologica l behaviorism by a philosopher), it may seem that in defending the behavioristic approach I am attacking all of the others . On the contrary, progress in answerin g either question, why or how, awaits progress in answering the other . ACKNOWLEDGMENT S This chapter is modified from an article in American Psychologist, November 1992, pp . 1371 1382, copyright ©1992 by the American Psychological Association, Inc . The modification was prepared with the assistance of grants from The National Institute of Mental Health and The Nationa l Institute on Drug Abuse . Hugh Lacey and Albert Silverstein provided extremely helpful criticism o f an earlier draft in a different form .

REFERENCE S Anderson, J . R . (1991) . More on rational analysis . (Author's response to commentary .) Behavioral and Brain Sciences, 14, 508-513 . Aristotle (tr. 1984) . The complete works of Aristotle (2 volumes) (Jonathan Barnes, Ed .). Princeton , NJ : Princeton University Press . Castro, L., & Rachlin, H. (1980) . Self-reward, self-monitoring, and self-punishment as feedback in weight control . Behavior Therapy, 11, 38-48 . Catania, A . C . (1975) . The myth of self-reinforcement . Behaviorism, 3, 192-199 . Davidson, D. (1980) . Essays on actions and events. Oxford, England : Clarendon Press . Day, W. (1969) . Radical behaviorism in reconciliation with phenomenology. Journal of th e Experimental Analysis of Behavior, 12, 315-328 . Dennett, D . (1978) . Brainstorms : Philosophical essays on mind and psychology. Montgomery, VT : Bradford Books. Donagan, A . (1987) . Choice: The essential element in human action. London : Routledge & Kegan Paul . Dretske, F. (1988). Explaining behavior: Reasons in a world of causes. Cambridge, MA : MIT Press . Fodor, J . A . (1981) . Representations : Philosophical essays on the foundations of cognitive science. Cambridge, MA : MIT Press. Grossberg, S . (1982) . Studies of mind and brain: Neural principles of learning, perception, development, cognition, and motor control. Amsterdam : Reidel Press .

214

RACHLI N

Guthrie, E. R . (1935) . The psychology of learning . New York : Harper. Herrnstein, R. J. (1969) . Method and theory in the study of avoidance . Psychological Review, 76, 49-69 . Herrnstein, R . J ., Loveland, D . H ., & Cable, C . (1976) . Natural concepts in pigeons. Journal of Experimental Psychology : Animal Behavior Processes, 2, 285-303 . Hocutt, M . (1974). Aristotle ' s four becauses . Philosophy, 49, 385-399. Homme, L . E . (1965) . Perspectives in psychology. XXIV : Control of coverants, the operants of th e mind . Psychological Record, 15, 501-511 . Honig, W. K ., & Staddon, J. R . (Eds .) . (1977) . Handbook of operant behavior. New York : PrenticeHall . Hull, C . L. (1952) . A behavior system: An introduction to behavior therapy concerning th e individual organism. New Haven, CT: Yale University Press . Irwin, T. N. (1980). The metaphysical and psychological basis of Aristotle ' s ethics . In A . O . Rorty (Ed .), Essays on Aristotle's ethics (pp . 35-53) . Berkeley and Los Angeles : University of California Press . Kantor, J . R . (1963). The scientific evolution of psychology (Vol . 1) . Chicago : Principia Press . Lacey, H. M ., & Rachlin, H. (1978) . Behavior, cognition, and theories of choice . Behaviorism, 6, 177-202 . Mahoney, M. (1974) . Cognitive behavior modification . Cambridge, MA : Ballinger. Pavlov, I . P. (1927) . Conditioned reflexes (G . V. Anrep, Trans .) . London : Oxford University Press. Putnam, H . (1980) . Brains and behavior . In N. Block (Ed .), Readings in philosophy of psychology (Vol . 1) . Cambridge, MA : Harvard University Press . Rachlin, H . (1985) . Pain and behavior. Behavioral and Brain Sciences, 8, 43-83 . Rachlin, H . (1989) . Judgment, decision, and choice . New York : W. H. Freeman. Rachlin, H . (1991) . Introduction to modern behaviorism (3rd ed.) . New York : W. H . Freeman . Rachlin, H . (1994) . Behavior and mind: The roots of modern psychology. New York: Oxford University Press. Rachlin, H . (1996) . Reply to Lacey. Behavior and Philosophy, 23, 79-84 . Rachlin, H . (1997) . The self and self-control. In G . Snodgrass (Ed .), The self across psychology : Self-recognition, self-awareness, and the self concept . New York Academy of Sciences Annals . Rachlin, H., Battalio, R., Kagel, J ., & Green, L . (1981) . Maximization theory in behaviora l psychology . Behavioral and Brain Sciences, 4, 371-388 . Randall, J. H., Jr. (1960) . Aristotle. New York: Columbia University Press . Rescorla, R . A . (1988) . Pavlovian conditioning : It' s not what you think it is . American Psychologist, 43, 151-160. Ringen, J. (1985) . Operant conditioning and a paradox of teleology. Philosophy of Science, 52 , 565 -577. Rorty, R. (1982) . The consequences of pragmatism . Minneapolis : University of Minnesota Press . Ryle, G. (1949) . The concept of mind. London : Hutchinson House. Samuelson, P. A. (1973). Economics: An introductory analysis (9th ed .) . New York : McGraw-Hill. Sechenov, I . M. (1863/1965) . Reflexes of the brain . In R . J . Herrnstein and E . G. Boring (Eds.), A source book in the history of psychology (pp . 308-320) . Cambridge, MA : Harvard University Press . Silverstein, A. (1988) . An Aristotelian resolution of the ideographic versus nomothetic tension . American Psychologist, 43, 425 -430. Skinner, B . F (1938) . The behavior of organisms. New York: Appleton-Century . Skinner, B . F (1945) . The operational analysis of psychological terms . Psychological Review, 52 , 270-277, 291-294 . Skinner, B . F. (1953) . Science and human behavior . New York: Macmillan . Staddon, J . E. R. (1973). On the notion of cause with applications to behaviorism . Behaviorism, 1, 25-64 . Staddon, J. E. R . (1993) . Behaviorism. London: Duckworth .


21 5

Stitch, S . (1983) . From folk psychology to cognitive science . Cambridge, MA : MIT Press, Bradford Books . Stout, R . (1996) . Things that happen because they should : A teleological approach to action. New York : Oxford University Press . Taylor, C . (1964) . The explanation of behavior. London : Routledge & Kegan Paul . Thorndike, E. L. (1911) . Animal intelligence . New York : Macmillan . Tolman, E. C . (1949) . There is more than one kind of learning . Psychological Review, 56, 144-155 . Watson, J . B . (1913) . Psychology as the behaviorist views it. Psychological Review, 20, 158-177 . Wright, L . (1976) . Teleological explanations. Berkeley and Los Angeles : University of California Press . Yourgrau, W., & Mandelstam, S . (1968) . Variational principles in dynamics and quantum theory . New York : Dover. Zuriff, G . E . (1979) . Ten inner causes . Behaviorism, 7, 1-8 .


9 THEORETICA L BEHAVIORIS M

J . E . R . STADDO N Duke University

I . Cognitive Psychology A. Philosophical Objections to Cognitive Psycholog y B. Al Objections to Cognitivism II . Theoretical Behavioris m A. The Model Is the Behavio r B. Consciousness III . Conclusion Reference s

One of the principal objects of theoretical research in any department of knowledge is to find the point of view from which the subject appears in its greatest simplicity . —J. Willard Gibbs

Behaviorism began as a positivist reaction against introspectionism, which i s one variety of mentalism . It was the dominant movement in American psychology for several decades, but has recently been pressed on two fronts : by cognitive psychology, which is a new form of mentalism, and by neurobiology, wit h which it never had any real quarrel . Like any successful new invention, behaviorism soon morphed into a number of variants (see Box 9 .1) . But in reacting to Handbook of Behaviorism

217


218

STADDO N

Classical : The behaviorism of Watson ; the objective study of behavior; n o mental life, no internal states ; thought is covert speech Methodological : The objective study of third-person behavior ; the data of psychology must be intersubjectively verifiable ; no theoretical prescriptions ; has been absorbed into general experimental and cognitive psycholog y Neo- : Hullian and post-Hullian, theoretical, physiological, group data, no t dynamic, not parsimoniou s Purposive : Tolman and his followers : mentalistic, theoretical, not parsimoniou s Radical : Skinnerian behaviorism ; includes behavioral approach to mental life ; not mechanistic ; internal states not permitted ; antitheory in practic e Teleological : Post-Skinnerian, purposive, close to microeconomic s Theoretical : Post-Skinnerian, accepts internal states ("the skin makes a dif ference") ; dynamic, parsimonious theor y BOX 9 . 1 Varieties of Behaviorism

its critics it lost sight of some of its original purpose, to make a science of behavior. Hullian behaviorism, for example, limited theory to concepts supposed t o reflect the physiological basis for behavior . Since knowledge of brain physiolog y was then vestigial (and is in all probability still minuscule in relation to what re mains to be known), Hullian and neo-Hullian quasi-physiology now appears archaic . Tolman, as Nancy Innis (this volume) points out, was attracted to mental ism and purpose . His view would now be termed cognitive ; it is open to many of the same criticisms as cognitivism (see later section) . B . F. Skinner's radical behaviorism was rhetorically compelling to those desirous of action prediction and control, with the emphasis on control . Although impoverished conceptually, it came with powerful experimental credentials . But despite Skinner's well-known acceptance of theory as "a forma l representation of the data reduced to a minimal number of terms" (Skinner, 1950/1961, p . 69), his opposition to theory in practice crippled radical behavior ism in the long term . Skinner's emphasis on data led at first to creative exploration of a novel set of experimental techniques but then, all too soon, to endles s permutations of reactive experiments ("Experiment A manipulated X, but faile d to measure Y: I shall manipulate X and measure Y") . In his defense of theory, the late Alan Newell (1973) pointed out that "you can't play twenty questions wit h nature and win" meaning that natural processes can rarely be understood b y the exhaustive enumeration of simple cause–effect relations . In physics, for ex ample, no amount of inductive research could have extracted the law of invers e squares from the fact that planets move in elliptical orbits . An infinitude of balls rolling down inclined planes would not suffice to reveal that force equals mas s times acceleration . Experimental analysis by itself can never make sense of be-

9 THEORETICAL BEHAVIORISM

21 9

havior. Theoretical imagination is also required . Because Skinner ruled out internal states (cf . Staddon, 1997), radical-behaviorist "theory" was perforce limite d to laws on the model of Weber and Fechner—as if physics were to be held t o the level of Boyle's law, never advancing to "explanation of an observed fact which appeals to events taking place somewhere else, at some other level of observation, described in different terms, and measured, if at all, in different dimensions" (Skinner, 1950/1961, p . 193) . 2 Behavioristic psychology was a physics without thermodynamics or atomic theory, dominated theoretically fo r 35 years by a Boyle-type empirical generalization, the matching law (Herrnstein , 1961 ; see Staddon, 1982, for the gas law parallel) . Of course, the essence of theoretical physics of scientific theory in any domain—has nothing to do with "levels" or "dimensions ." The essence is as the great Willard Gibbs stated it : to find "the point of view from which the subjec t appears in its greatest simplicity ." That is the central aim of theoretical behavior ism (Staddon, 1993a, 1997) . In this chapter I first comment on some of the problems with cognitivis m the chief cause of behaviorism's eclipse and a point of view that can be confuse d with theoretical behaviorism and then describe theoretical behaviorism and it s approach to a handful of traditional philosophical puzzles . I . COGNITIVE PSYCHOLOG Y If we do not postulate some agent who selects and uses the stored information, we must think of every thought and every response as just the momentar y resultant of an interacting system, governed essentially by laissez-faire economics (Neisser, 1967, p . 293) . Probably few cognitive psychologists now feel a s Neisser did in 1967 that the brain–mind is best understood as an autonomou s agent processing "information ." Even such icons of cognitivism as Marvin Minsky (1985) now argue for a "society of mind," for a view of intelligent behavio r as the outcome of complex interaction among agents that may be autonomou s but whose more salient attributes are stupidity and multiplicity . The mind, mos t cognitivists seem to agree, is a self-organizing system, not a homunculus . Laissez-faire economics is not so far from the current model after all . Yet the decision-making-agent view of mind is still (I would argue) the motivating impulse for much cognitive psychology because cognitivists privileg e introspection. They are at bottom mentalists. "Psychology is the science of mental life," said William James in 1890, and George A . Miller echoed the sentiment in the first salvo of the cognitive counterrevolution, his 1962 book Psychology : The Science of Mental Life. Not only do cognitive psychologists want to study mental life, they want to study the mental life of human beings . This also puts them at odds with behaviorists, who argue from evolutionary biology that sim 2In a strange perversion of positivism, Skinner ruled out theories of this type, which is to say, any real theory at all (cf . a cogent critique by Ben Williams, 1986) .

220

STADDO N

pier, hence more comprehensible, antecedents of the human mind are to b e found in the minds of animals . The need for experiments in which reward and punishment may be freely administered also implies that scientific psycholog y must be centered on animal experiments . But cognitivists were soon dismissing this work as "rat psychology." Psychologists had been chafing against the constraints of biology and behaviorism for two generations when they were liberated by the "cognitive revolution ." The new cognitive psychology legitimated the study of language, mental representation, and consciousness, which (it was argued) had been ruled out or trivialized by the old behaviorism . At least one apostle of the cognitive revolution has placed himself squarely against parsimony : The sole purpose of science is not to frame parsimonious laws [my emphasis] . Notwithstanding Watson 's and Skinner ' s emphasis on the prediction and control of behavior, science aims to explain phenomena, not merely to describe them in laws . And explanations, of course, take the form of theories . Hence, if mental states exist, a complete psychological explanation should give an account of them even if they could be dropped from the laws of behavior. (Johnson-Laird, 1988, pp . 17-18).

This statement is of course a great non sequitur, because it assumes behavior an d mental states are sufficiently independent of one another that it may be possibl e to find adequate laws describing one that will fail to include the other, a curious violation of the materialist view of mind. I would be happy to rephrase Johnson Laird's plaint as "if mental states exist, they will be explained by the laws of behavior" with the unspoken corollary that if mental states are not explained , perhaps they do not exist in a scientific sense . Perhaps mental states are not measurable in the usual third-party way ; perhaps they are not "public knowledge" in John Ziman's (1968) felicitous phrase, and so lie outside science in whic h case the attempt to explain them will fail anyway (more on this in a moment) . But at this point we are engaging not in scientific argument but in speculation . Which stock is more likely to rise : assuming mental states and then trying to explain them, or trying to explain what we really know and can measure, and assuming the mental states will come out in the wash? Take your pick, accordin g to temperament. In any event, philosophical niceties could be shelved in the mid-1950s, because at that time the new mentalism was rendered respectable by the digita l computer. Mentalistic theories could be simulated, programmed made up into a working model. Until then, rigor and quantification had been almost a monopoly of the behaviorists . The vague, verbal theorizing of the Gestalt psychologist s and other protocognitive psychologists could easily be dismissed . But now the ability to simulate mentalistic theories with digital precision conferred prestig e and respectability on cognitive psychology . Cognitive psychologists no longe r need feel embarrassed by hard-headed behaviorists . How has the computer-assisted excitement of the original cognitivists panned out? Pretty well, for a while . But now some philosophers are arguing that cognitivism is failing because the computer metaphor is wrong and because cogni-


22 1

tivism is not mental enough . And some artificial intelligence (AI) researcher s and roboticists are arguing that the emphasis on symbolic thought is largely irrelevant to the development of truly intelligent systems . I believe that the philosophical objections to contemporary cognitive psychology merely underscore the importance of a sophisticated behaviorism and the AI objections partly de fine it. Let us look at the philosophy first. A . PHILOSOPHICAL OBJECTION S TO COGNITIVE PSYCHOLOG Y John Searle (1992) has mounted the most organized attack on mainstrea m cognitive psychology . He objects to its central tenet : that the brain is like a computer and cognitive operations like a program running on that computer . Searle' s main objection is that anything can be conceived of as a computer: a ball rolling down an inclined plane, for example, is an analogue computer that compute s Newton's laws of motion ; a nail driven into a plank computes the distance i t must travel given the force of the hammer and the friction of the wood . The problem, says Searle, is that computational properties are not intrinsic to an y system ; they are interpretations imposed from outside . The scientific problem i s not to understand what is being computed, but how the machine works . The brain-as-computer idea is very widely accepted in cognition . It was firs t summarized in psychology by Feigenbaum and Feldman (1963 ; see also Johnson-Laird, 1988 ; Neisser, 1967 ; Pylyshyn, 1984) . The founding genius for the view was Alan Turing (1950), who proposed as a test for "thought" a comparison between the responses to typed questions of a human being and an appropriately programmed computer. True AI will have been created, argued Turing, if a machine interrogated via e-mail cannot be distinguished from a real human being .' A lot depends, in this very behavioristic definition of human intelligence , on the intelligence of the interrogator . As Joseph Weizenbaum (1976) showe d years ago with his little program Eliza, which simulates a nondirective psychotherapist, a naive questioner is easily fooled into attributing much more t o such a program than its simple structure requires . It may be that cognitive psychologists are also naive in imputing to humans and animals complexities beyond the minimum necessary to explain their behavior. Such, at least, is the suspicion that motivates theoretical behaviorism . Searle's second objection is to accuse cognitivism of the familiar homunculu s fallacy, a charge that Neisser's landmark textbook tends to sustain (and a charge that can also be leveled at radical behaviorism, see later section) . Sophisticate d cognitivists are of course aware of this problem : There is no "little man in the head" looking at some sort of screen on which the products of sensation are displayed . Nevertheless, the ways in which important problems are presented impl y such a little man, says Searle : "Typical homunculus questions in cognitive sci 3There is now a prize, the Loebner Prize, offered for the best Turing test computer program . S o far, the contestants have not been impressive .

222

STADDO N

ence are such as the following : `How does the visual system compute shap e from shading ; how does it compute object distance from size of retinal image?" (1992, p . 214)' . The savvy cognitivist might respond by saying that what is re ally meant is "How is information about object distance represented in th e brain?" and "How is this information derived from the retinal input?" He wil l need to add, of course, "How do we know these things?" In other words, w e need to know something about how states of mind and brain are translated int o things we can measure that is, into behavior . Clearly, the question of representation cannot be separated from questions o f computation and action but it often is, as Searle's questions show . When all three questions are considered together, it is difficult to see how they amount to asking more than "How does the brain work to make behavior?" which is no t necessarily a computational or even a cognitive question . Searle's case against the computer metaphor for the mind is persuasive, an d much of the standard cognitivist boilerplate about "information processing" and computational "levels" is less than compelling on close examination . No information is being processed, perhaps nothing is being "represented" : the external world is changing, neurons are firing, chemicals are circulating, and physical actions are occurring . The job for psychologists and neurobiologists is to find the connections among these events, with the psychologist concentrating on environment–behavior relations and the neurobiologist on relations that involve th e brain . There are other cognitive conceits that Searle might have attacked . One of the most egregious is the curious distinction between competence and performance . Like computational "levels," this is another "trust-me" distinction that depend s for its validity on the reader's acceptance of the proposer's view of the subject . The subject in this case is language, and the proposer is mathematical linguis t Noam Chomsky (e .g ., 1964) . The distinction is less influential now than it use d to be, but the kind of thinking it represents still permeates cognitive psychology . The idea is that language follows certain syntactic "rules," but that actual examples—speech as she is spoken or writing as she is written fall short of perfection because of constraints such as memory limitations . The arbitrariness of this distinction is obvious from a thought experiment . Suppose someone gives you a device of unknown function with a numerical key pad . You try it out by typing in various combinations, and you get the followin g set of outputs : 2 * 2 = 4, 20 * 20 = 400, 25 * 25 = 625, 30 * 30 = 900 . "Great!" you think . "I've figured it out . It's a calculator that multiplies ." But then you try 40 * 40 = 999—whoops! It turns out that the calculator multiplie s fine, so long as the answer is less than 1000 . Anything greater, and it just give s the largest number of which it is capable : 999 . How shall we describe the behavior of this device? The competence–performance people would say, "Well, th e deep rule underlying this behavior is multiplication, but because of memory limitations, answers greater than 999 cannot be produced ." There is an exactly parallel argument in linguistic theory to account for the fact that people are limited


22 3

in their ability to understand and produce deeply embedded, but perfectly grammatical, sentences . "Fine," you may well say. "What's wrong with the competence-performance account of the behavior of this `constrained' calculator?" The answer is "Nothing, so far ." The problem is that there are alternative accounts , but deciding among them demands additional information . For example, one alternative is that the device does not multiply at all, bu t rather just consults a fixed lookup table that comprises all the three-digit (or fewer) products of two-digit numbers . A "brain lesion" (i .e ., damage to the device memory) would support this view if it simply knocked out certain answers but left others involving the two multiplicands unaffected . More general effects might support the competence–performance alternative . Conversely, a purel y behavioral experiment showing that the time taken to complete a given multiplication is independent of the size of the multiplicands would tend to support th e lookup idea, and so on . The point is that the competence–performance idea , which implies an organization involving a rule-following engine limited b y memory and other constraints, represents just one kind of model for what is happening in the linguistic black box . We are under no obligation whatever to accep t it without knowing both the exact nature of the rules that are being enforced an d the precise constraints that limit their output. Without the details, we are simpl y invited to accept as a matter of fact what is nothing more than an arbitrary limitation on theoretical inquiry. ' B . AI OBJECTIONS TO COGNITIVISM

The theme that intelligence is the product of collective action of many unintelligent agents dismissed as "laissez-faire economics" by Neisser—has been carried forward by researchers in robotics, the engineering of intelligent, autonomous artificial creatures . Robotics researcher Rodney Brooks has writte n critically of the current state of cognitive science : No one talks about replicating the full gamut of human intelligence any more . Instead we see a retreat into specialized subproblems, such as ways to represent knowledge, natura l language understanding, vision or even more specialized areas . . . . I . . . believe that human level intelligence is too complex and little understood to be correctly decompose d into the right subpieces at the moment . . . . We will never understand how to decompose human level intelligence until we've had a lot of practice with simpler level intelligence . (1991, p . 140)

So, the latest word from the cognitive-science frontier is that the standard AI approach is being increasingly called into question . We might do better to return 'The competence–performance distinction—the idea that real language is an imperfect readou t of a perfect rule-making engine—has recently encountered difficulties with experimental data o n language learning: "Although there is plenty of evidence that humans use language creatively (sayin g and understanding things that have never been said before) and well (with very low error rates), ther e is very little evidence for the claim that `perfect' knowledge underlies our (occasionally) imperfec t behavior" (Bates & Elman, 1996, p . 1849) .

224

STADDO N

to simpler systems to animals, to "rat psychology ." Brooks provides a plausible biological argument for this position . He notes that language and symboli c communication generally are very recent arrivals on the evolutionary scene . Most of evolutionary time was taken up developing creatures that could do littl e more than find their way about . Distinctively human abilities such as speech an d writing arrived only during the last few ticks of the evolutionary clock . Perhaps we should walk before we try to talk .

II . THEORETICAL BEHAVIORIS M A . THE MODEL IS THE BEHAVIOR Teleological behaviorism (see Rachlin, this volume) is the attempt to understand behavior terms of outcomes . This is also the approach of classical rational choice microeconomics, which tries to explain human behavior as a set of strategies designed to achieve some optimal end, such as maximal profit or some other "good" (cf. Becker, 1987 ; Hirshleifer, 1984) . The teleological approach has been powerful and effective, particularly when the processes that guide behavior ar e complex . But it is not causal . The organism may arrive at some optimal result , but the result is not the (efficient) cause of the behavior . It is better to find efficient causes the antecedent factors that determine behavior and the mechanisms through which they act if we can . Skinner was ambivalent about causal analysis . Operant behavior, his major interest, is of course behavior guided by its consequences . So one Skinnerian theme is teleological . But he also emphasized the organism's history as the determinant of its behavior. Emphasis on antecedents, rather than consequents, is causal. Unfortunately, Skinner's emphasis was entirely on reinforcement history—he had no interest in nonreinforcement effects such as habituation an d motivational change, for example . And because he eschewed real theory, Skinner also provided no way to interpret the effects of complex histories no way for the effects of events in the remote past to combine with the effects of more re cent events, for example . Perhaps for this reason, the experimental emphasis of single-organism re search on operant conditioning has been overwhelmingly on what are known a s reversible effects . For example, a pigeon might be trained for many experimental sessions to peck a key for food reinforcement on some intermittent schedule o f food reinforcement . After 20 or more daily experimental sessions, its respons e rate, the average number of key pecks over an hour period, will stabilize at som e value such as 65/minute . This value will be largely determined by the average rate at which the bird gets food (food deliveries/hour), which in turn is deter mined by the parameters of the schedule (e.g ., the average minimum interfoo d interval for a variable-interval schedule) . The pigeon's rate of responding is largely independent of the bird's history prior to this particular protracted ex-


22 5

perimental condition. If the schedule parameter is changed, from 1 food delivery/minute to 2/minute, say, the bird will settle down at a higher peck rate , 75/minute, say. But if the schedule parameter is changed back again to 1/minute , the rate of pecking will eventually return to 65, its prechange value . Thus, the relation between schedule parameter and response rate is a reversible property o f behavior. The problem is that the same behavior on these two occasions does no t mean that the pigeon is in the same state . That is to say, the pigeon will not necessarily behave in the future in exactly the same way as if it had never had th e intervening experience . Behavior is often reversible, but the state of the organism rarely is . Hence, by restricting their attention to reversible properties of behavior, radical behaviorists avoided dealing with the deeply historical nature o f learned behavior in higher organisms . I will give just one example . Imagine an experimental situation with pigeon s that involves not one but two response keys . Pecks on each key are paid of f probabilistically, like a Las Vegas – style one-armed bandit . Let us look at two different experimental histories . Pigeon A is trained at first to peck on the left response key and paid off on the probabilistic schedule ; pecks on the right have no effect . After 50 or so reinforcements for pecking left, bird A is trained to pec k only on the right; pecks on the left go unrewarded . After 300 reinforcements fo r right pecks (this might take four or five daily experimental sessions), pecks o n neither key are rewarded (this is termed extinction) . Pigeon B is trained through out (350 reinforcements) to peck the right key only for food, then extinguished , like pigeon A . There are two things to notice about this simple experiment : on the last day o f rewarded training, the behavior of the two pigeons will be indistinguishable : both will be pecking exclusively on the right key. But in extinction, their behavior will be very different: Pigeon B, which has never gotten food for peckin g left, will simply peck more and more slowly on the right until pecking ceases al together. But pigeon A, with its history of a few rewards for pecking left, will show some pecks to both keys before quitting entirely. What can this differenc e in behavior in extinction mean? If behavior is all that matters, the two birds were in the same state at the end of right training . So why did they behave differently when conditions changed? There are two ways to deal with this divergence . The solution adopted by teleological behaviorism is to redefine behavior in a temporally extended way. Behavior becomes behavioral history. Since our two pigeons have different histories, the fact that their response to a new condition (extinction) is not the sam e is no longer a puzzle . The problem with this view is that is provides no condensation of the data . It provides no rationale for similarities among histories . Not all histories produce the kind of divergence I have just described . It is possible , in principle and to some degree in practice, to group together histories that are equivalent, in the sense that the future behavior of the animal (in response to new conditions) will be the same after any history in the set. For example, in a version of the experiment I have described we might find that an animal with the

226

STADDO N

history (number of rewards in parenthesis) right (10), left (10) behaves the same in extinction as an animal whose history is left (5), right (15), left (10) . The usual way to describe this equivalence is to say that the two animals are in the same state after these two equivalent histories. The aim of the theoretical behaviorism, therefore, is to understand the complete set of internal states of our animal, where each state is defined by a set of equivalent histories .5 There are onl y two ways to understand internal states : through a deep understanding of the underlying neurophysiology—which may be very complex indeed (cf . Staddon & Zanutto, 1998) or through the construction of dynamic models . Theoretical behaviorism (TB) is a natural descendant of both classical an d Hullian behaviorism. It gets from classical behaviorism the conviction that w e learn about the organism only through its behavior . It rejects, however, the vie w shared by Watson and Skinner that psychology need refer only to stimuli and responses . Contra Skinner, it argues that the skin does make a difference : Events inside the organism (e .g ., the changes wrought by past history) are state variables, not stimuli or responses (see later section) . Contra cognitivism, internal states are not necessarily mental ; contra Hullian behaviorism, internal states are not necessarily physiological (cf . Staddon & Bueno, 1991) . TB sees internal states as purely theoretical constructions based on information from historica l experiments . Nevertheless, it shares with Hullian behaviorism the idea that th e ultimate aim of behavioral study is the derivation of mechanisms. By mechanism I refer not to physiology, but simply to an algorithmic mode l that relates an organism's history the stimuli it has received and the response s it has made to its future behavior. There is some expectation that the ingredients of such a model may eventually prove to have something to do with th e underlying physiology, but the aim of the model is to provide the simplest ex planation for environment–behavior relations, not brain–behavior or environment–brain relations (see Staddon & Bueno, 1991, and Staddon & Zanutto , 1998, for fuller accounts) . Theoretical behaviorism is interested in mechanisms for entirely practica l reasons . The argument runs like this : Classes of equivalent histories are define d by reference to potential experiments . But real animals are very complicated, and historical experiments take time and, ideally, an indefinite supply of identical subjects (replicas) . Consequently, there is no way that the full set of interna l states of a real animal can be enumerated experimentally . Theoretical creativity 'This approach is an application to experimental psychology of logical concepts from automata theory (e .g ., Minsky, 1969) . See, for example, Staddon (1973, 1993a, 1993c, 1997) and Davis, Staddon, Machado, and Palmer (1993) . The term internal often raises behavioristic hackles, but I hav e difficulty understanding why. Suppose we test two identical (replica) organisms at time t after tw o different histories and find their behavior to be identical . Are they in the same state? We do not know until we do some more tests . Suppose these tests reveal a difference at some later time t + k. The n at time t the animals must have been different . Since they were the same externally (i.e . , same behavior) at time t any difference between them at that time must be internal: hence, interna l state .


22 7

is necessary, and theories do not emerge automatically from "orderly arrangement of data" : invention is also required . In practice, therefore, the main way t o specify sets of equivalent histories is through dynamic models that define how moment-by-moment experience changes the state of the organism . These theories can be compared with data, tested (if they do well enough with what is already known), modified or abandoned (all theories are eventually modified o r abandoned), revised, and tested again, in the usual scientific way . I began this section with the assertion that the model is the behavior, whic h might be the slogan for theoretical behaviorism. The basis for this claim is as follows : 1. Behavior, as simple physically defined acts, is literally uninterpretable . The same motor pattern can be a threat, a formal signal, an apparently irrelevan t accompaniment of some other action, or many other things . As Skinner, as wel l as countless critics of behaviorism, have tirelessly pointed out, behavior must be defined in some more general way if distinctions like these are not to be lost . Skinner's suggestion was to group together all acts with a common reinforcin g consequence, which is an improvement over raw physical description, but stil l teleological and still insufficient to explain, for example, why one organism develops one repertoire under a set of reinforcement contingencies and behave s afterward in one way whereas another develops a different set or behaves in a different way afterward . History, and its effects on the organism, must be captured in some theoretical way. 2. Skinner, like many other psychologists, agreed that the findings of psychology provide "what is to be explained" by neurophysiology, just as the law s of chemistry are, ultimately, to be explained by physics . Since behavior, physically defined, is intrinsically ambiguous, it cannot provide "that which is to be explained " by neurophysiology. What neurophysiology must explain is what th e organism is doing—what kind of a machine it is . It follows that the behaviorist' s aim must be to arrive at an accurate dynamic model . It is in this sense that "the model is the behavior" the model is what neurophysiology must explain, not behavior tout court . How does theoretical behaviorism differ from cognitivism? Both are theoretical and both assume internal states . One difference is that theoretical behavior ism is explicitly historical and dynamic (see Fig . 9 .1) . It is not concerned directly with representation, a core issue for cognitivists, but with the way that th e organism is changed by its experience . A second difference is that theoretical behaviorism makes no presumptions about either its subject matter or its theoretical constructs . Cognitive psychology is "the study of mental life" ; TB is the study of the mechanisms of behavior, where mechanism is whatever works to ac count for behavior, and behavior is whatever can be usefully observed or measured, including reports of conscious experience (see later section) . Theoretical behaviorism assumes in advance neither that the categories of consciousness are

228

STADDO N

STIMULUS STATE RESPONS E I I -p.

El -~

BI

T 12 ~—~

El ~

B2

M E

El

13

_Op.

--► B3

IN —■ --■ BN FIGURE 9 . 1 The view of the organism in theoretical behaviorism . If time is divided into discrete time steps, then at each step a stimulus (which may be "no stimulus "), I, produces a response , B (which may be "no response"), and changes the organism's state, S, or both .

inevitable ingredients of any valid theory, nor that they must be immediately explicable by such a theory . "Theories are where you find them," says TB ; exploit whatever works . "If places for mental states can be found in the resulting analysis, fine ; if not, no problem ." Figure 9 .1 provides only a framework . It tells us nothing about how stimul i and responses are to be defined . It does not specify the properties of the states : how many there are, the rules by which they change, and so on. These are th e concern of specific theories, and I refer the reader to original sources for mor e details .' Four types of theoretical structure are illustrated in Figure 9 .2 . A common theme that seems to be emerging from recent work is that many of the mos t striking and reliable dynamic properties of operant conditioning seem to emerg e from relatively simple structures with multiple time scales . Other essential ingredients seem to be competition and nonlinear interactions between variables . Models assembled from these ingredients have been shown to account for th e basic properties of operant learning, including apparent exceptions to the reinforcement principle, such as "instinctive drift" and "superstition, " for the dynamic properties of habituation, and for several properties of complex choice . (See references in note 6 and the legend to Figure 9 .2 for specific applications of this approach .) The next section shows how theoretical behaviorism applies to some questions on the borderline between psychology and philosophy . 6 The theoretical approach to behaviorism is growing, and it is hard to single out particular individuals. For surveys see Staddon (1983) and Commons, Grossberg, and Staddon (1991) and earlie r volumes in the same series . See also many of the contributions to the annual Society for the Quantitative Analysis of Behavior (SQAB) conference .

22 9


R

Assignment of Credit Process integrator memories WTA r A Ei

c_r4

B C

1► ~~

1

es

stimulus (X) -04 V® 1 —

Feedback-typ e habituation unit +

response (Vo)

Vint

Integrator 'do__

p

o n

sse

Rate-sensitive habituation cascad e

response

stimulus

R : Reinforcement input E : "noise" input stimulus

Cascaded integrator ©

output 0

FIGURE 9 .2 Theoretical structures that have been used to account for aspects of operant behavior. Assignment of credit (left) : A parallel structure in which the strength of each response is determined by a reinforcement "memory" (which may be a simple integrator, a rate-sensitive cascade , or some more complex process). There is also a separate "noise" input, s, to each memory . The memory outputs compete according to a nonlinear rule, such as " winner-take-all" (WTA), so that the strongest response is the one that actually occurs . This system shows nonassociative operant conditioning—response emission and selection by reinforcement as well as anomalies such as "instinctive drift" — and is sensitive to delay of reinforcement and contingency effects (Staddon & Zhang, 1991 ; Dragoi, 1996, has incorporated the essential properties of this model into a more comprehensive model for operant conditioning) . Habituation : The top right diagram shows a single feedback–typ e (rather than feedforward) system that shows habituation of a reflex response . Cascaded versions o f this system (right center) show rate sensitivity, the fact that habituation to less frequent stimuli is slower and less complete, but often more persistent, than habituation to closely spaced stimuli (Staddon, 1993b ; Staddon & Higa, 1996) . Integrator cascade: An ingredient of several models allows for response delays : a model of feeding dynamics (Staddon & Zanutto, 1997, 1998b ; Killeen, 1995, has explored the same problem more analytically) and models of time discrimination (Killeen & Fetter man, 1988 ; Machado, 1997). An essential feature of all these models is the existence of multiple tim e scales (i .e ., elementary processes, in serial or parallel arrangements) that occur at different rates .

B . CONSCIOUSNES S

Consciousness is a hot ticket in cognitive psychology today . Respected foundations sponsor conferences on it ; Nobel prize winners aspire to understand it; philosophers and psychologists collaborate to define it (e .g ., Crick & Koch, 1992 ; Dennett, 1991 ; Dennett & Kinsbourne, 1992 ; Flanagan, 1992 ; Gray, 1992) : "There was a large measure of agreement among both scientists an d philosophers at a recent symposium that not only is there a real problem of consciousness but that it is a scientific problem and that the time has come for scientists to tackle it" (Gray, 1992, p . 277) . Roughly 1000 philosophers, psychologists, neuroscientists, computer scientists, physicists, and other cognoscent i

230

STADDO N

participating in "Tucson II : Toward a Science of Consciousness 1996" failed to agree on what consciousness is, but felt the topic important enough to be wort h five days in the desert (Ezzell, 1996) . Consciousness is the topic that most clearly differentiates behaviorists fro m nonbehaviorists . Just what scientific problems does it pose? How would a theoretical behaviorism attempt to answer them, and how would its answer diffe r from the answer offered by cognitive psychologists ? TB takes the Turing test view of consciousness . This view is not accepted b y everyone, however. John Searle (1992), if I understand him correctly, makes th e argument that even if a device were to be found that could pass the Turing test, i t would not be conscious . I have three reactions : First, the assumption that such a device can be created solely from hardware may be false, in which case we nee d say no more . Second is the obvious question : Assuming it can be created, how do you know it is not conscious? The only answer to this question is : because it does not pass the Turing test, which is contrary to the first assumption . In other words, if the only way we know that someone (or something) is conscious is be cause it answers our questions appropriately, then, by definition, a machine tha t can pass the Turing test must be conscious . My third reaction is simply to wait and see . If a machine is ever created tha t passes the Turing test, people will soon enough treat it as one of their own . If we are willing to grant consciousness to a dog, or to someone whose ability to communicate is as impaired as Helen Keller's, are we likely to withhold it from a de vice that speaks and responds indistinguishably from a human being? The debat e will surely go the way of old controversies over apparent paradoxes such as action at a distance or wave-particle duality, neither of which now seem as impossible as they did when first proposed .

FIGURE 9 .3 The phi phenomenon . Spot A is lit briefly, and then after a short delay, Spot B i s lit. After repetition of this sequence, a single lighted spot appears to move from A to B and back .


23 1

1 . Consciousness and Color-Ph i Cognitive confusions about consciousness can be illustrated by recent discussions of a perceptual effect known as the phi phenomenon . The phi phenomeno n is the perceptual fusion into the percept of a single moving image of a succession of similar images flashed at brief intervals . When you watch a movie, your retina is stimulated 24 times each second with 24 static images . An object that takes up adjacent positions in each successive image is perceived as movin g smoothly . The effect can be demonstrated experimentally with a single brigh t spot that is successively presented at one place and then at an adjacent place (se e Figure 9 .3) . There is a related effect in which the two spots are different colors . What is then seen is a single moving spot that changes color at about the mid point of its travel . This color-phi effect was particularly puzzling to philosopher Nelson Goodman . Daniel Dennett and Marcel Kinsbourne (1992) discuss th e problem this way: Goodman wondered : "How are we able . . . to fill in the spot at the intervening placetimes along a path running from the first to the second flash before that flash occurs?" . . . Unless there is precognition, the illusory content cannot be created until after some identification of the second spot occurs in the brain . But if this identification of the second spot is already "in conscious experience" would it not be too late to interpose the illusory color switching-while-moving scene between the conscious experience of spot 1 and the conscious experience of spot 2? . . . [Other experimenters] proposed that the intervening motion is produced retrospectively, built only after the second flash occurs, an d "projected backwards in time ." . . . But what does it mean that this experienced motio n is "projected backwards in time"? (1992, p . 186)

Presented in this way, the color-phi effect might well puzzle anyone, but as w e shall see, this description is highly misleading . Dennett and Kinsbourne maintain the suspense by inventing two supposedly standard cognitive ways of dealing with this effect . One, which they term Orwellian, is that we experience things in one way, but then revise our memories, much as Minitruth in Orwell' s 1984 revised history. The color-phi effect thus becomes a post hoc reinterpretation : two spots are experienced, but a smoothly moving, color-changing spot i s reported. Dennett and Kinsbourne term the other standard approach Stalinesque, by analogy with Stalin's show trials, in which false evidence was reported accurately. In this view, what is reported is what was actually experienced, though what was experienced was not what (objectively) happened . All clear ? Probably not but the reader is now ready for Dennett and Kinsbourne's alternative. They dismiss the Stalinesque and Orwellian accounts in favor of wha t they term a multiple drafts model : Our Multiple Drafts model agrees with Goodman that retrospectively the brain creates th e content (the judgment) that there was intervening motion, and that this content is the n available to govern activity and leave its mark on memory . But our model claims that the brain does not bother "constructing" any representations that go to the trouble of "fillin g in" the blanks . (p . 194)

In the multiple drafts model consciousness becomes a distributed construct, like

232

STADDO N

`the British Empire' (their analogy), which is not uniquely located in time o r space . If you still do not get it, do not worry : The essence of the correct view is Den nett and Kinsbourne's phrase "creates the judgment ." "Multiple drafts" and so forth are irrelevant . What is important is that a given history causes a certai n state of the organism ("judgment," if you prefer) . That is all . This is the view of theoretical behaviorism, which I can explain in the following way. First, note that like all other psychological phenomena, the color-phi effec t involves three conceptually separate domains . a . Domain 1 The first is the domain of felt experience, the phenomenological domain . There is a certain quality (philosophers call this quale) associated with the colorphi experience. This quality is subjective and we can say nothing about it directly . From a scientific point of view, I cannot say whether "green" looks th e same to you as to me ; I can only say whether you make the same judgments about colored objects as I do. I can also know if you say the same things abou t color-phi–type stimuli as I do . This point is a commonplace in philosophy, but apparently it needs to be reiterated from time to time : "That different peopl e classify external stimuli in the `same' way does not mean that individual sens e qualities are the same for different people (which would be a meaningless statement), but that the systems of sense qualities of different people have a common structure (are homeomorphic systems of relations) " (Hayek, 1952/1979, p . 37) . Note that this is a behavioristic position, but it is not the version of behavioris m dismissed by Dennett and Kinsbourne (1992) when they say : "One could, then, `make the problems disappear' by simply refusing to take introspective report s seriously" (p . 187) . As we will see shortly, the question is not whether phenomenological reports should be ignored of course not but what to do with them . It is worth noting that the TB approach to what Skinner calls "private events " is quite different from Skinner's . TB assumes that private events are really private, as far as their quale is concerned. But because stimuli and responses (albei t broadly defined) are all that Skinner permits by way of theory, for him, privat e events must also involve stimuli and responses . Skinner asserts provocativel y that "the skin makes no difference ." Of course, the skin makes a great deal of difference if you contrast the effect of placing a sharp object inside versus out side the chest cavity, you will get the idea . Skinner's approach to private events implies an inner homunculus, eve n though he ridicules the idea . In Skinner's parody, when a man's finger is pricked , electrical impulses illuminate a television screen in the brain . Then, says Skinner, a "little man wakes up, sees the flashing screen [internal stimulus], reache s out, and pulls a lever [internal response] . More flashes of lightning go down th e nerve to the muscles, which then contract" (1963, p . 951) . Skinner's own ac count of private events lacks the neurological imagery, but is otherwise quit e similar : "Each of us is in special contact with a small part of the universe en-


23 3

closed within our own skin . . . . Each is uniquely subject to certain kinds o f proprioceptive and interoceptive stimulation ." Who, exactly is "in . . . contact with a small part of the universe . . . within our own skin"? Who but a homunculus . And what are these (unmeasured) "interoceptive" stimuli? There are no internal stimuli and responses unless we are ready to acknowledge the existence of someone ready to sense the stimuli and make the responses, in whic h case his internal functioning must be explained, which leads to another interna l homunculus, which itself demands explanation, and so on . Skinner's attempt to evade internal states implies the very internal homunculus that he overtly rejects . b. Domain 2 The second domain is physiological, the real-time functioning of the brain . The color-phi experiment says nothing about the brain, but another experiment , which I will discuss in a moment, does include physiological data. c. Domain 3 The third domain is the domain of behavioral data, "intersubjectively verifiable" reports and judgments by experimental subjects . The reports of people in response to appropriate stimuli are the basis for everything objective we can know about color-phi . I believe that much of the muddle in the various cognitive accounts arises firs t of all from confusion among these three domains . For example, Dennett and Kinsbourne (1992) write that "conscious experiences are real events occurring i n the real time and space of the brain, and hence they are clockable and locatabl e within the appropriate limits of precision for real phenomena of their type" (p . 235) . Well, no, not really . What can be clocked and located are reports of conscious experiences and measurements of physiological events . Conscious experiences are Domain 1, which has neither time nor space, but only ineffable qualia. The only evidence we have for these qualia (at least, for someone else's) is Domain 3. And we can try and correlate Domain 3 data with Domain 2 data and infer something about the brain correlates of reported experiences . But that is all . Despite Dennett and Kinsbourne's picturesque prose, their discussion promote s a muddle . All becomes much clearer once we take look more closely at Domain 3: What did the subjects see? That is, what did they say about it, and when did the y say it? The real-time events in the color-phi experiment are illustrated in Figur e 9 .4, which is a version of the general framework of Figure 9 .1 tailored to this experiment . Time goes from top to bottom in discrete steps .' At time 0 the red spo t is lit and goes out ; there is a delay ; then the green spot is lit ; there is another delay ; and then the subject reports what he has seen, namely a continuously moving red spot that changes to green halfway through its travel, which I will denote by "RRRRGGGG ." 'The discreteness is not theoretically necessary, but makes the sequence easier to illustrate .

234

STADDO N

Stimulus Subject Respons e DISTANC E

T

R

I STATE... ESTATE...

M

STATE. . .

E

STATE . . . STATE. . . STATE .. . STATE. . . G

f STATE. . . STATE. . . "RRRRGGGG"

FIGURE 9 .4 Event history in the color-phi experiment . A red spot (R) is briefly lit at time 0 ; a little later a green spot (G) is lit . After an additional short delay, the subject responds that he has seen a moving spot that changes color (RRRRGGGG) . The boxes labeled "state " are simply to indicate that the subject' s state is changing from time step to time step .

The stimuli and responses in the diagram are in Domain 3. (The states are no t in any of the three domains, of course : they are neither qualia nor brain measurements nor behavior. They are inferred constructs, and their properties are a s yet undefined .) The problems with this example seem to arise over how the subject's response is to be interpreted . What does "RRRRGGGG" mean? Is this re ally what the subject sees? Interpreting the response seems to be the heart of the puzzle, but the unknowable quale here is scientifically irrelevant . From a third person (i .e., scientific) point of view, the response just tells us something abou t what other "control" experiments might give the same response. Figure 9 . 5 shows one such control experiment . In this experiment, a single spot really is moving and changing color at the midpoint—RRRRGGGG— and the subject' s report is appropriately "RRRRGGGG ." The similarity between the responses t o the objectively moving stimulus and to the color-phi stimulus is what the statement "the color-phi stimulus looks like a continuously moving spot that change s color" means. The point is that we (i .e ., an external observer) cannot judge the subject's quale, but we can judge if his response is the same or different on tw o occasions . And as for the subject, he can also judge whether one thing looks like another . These same–different judgments are all that is required for a scientifi c account.

23 5


The comparison between these two experiments casts the color-phi experiment in a form that poses an answerable scientific question, namely : "What kinds of process give the same output to the two different histories illustrated i n the two figures?" More generally, what characterizes the class of histories that give the response "RRRRGGGG"? The answer will be some kind of well-defined process, which can be simulated on a computer or built into a microchip . One can even get a glimpse of the appropriate process . It will be one in whic h the representations of temporally adjacent events tend to inhibit one another, s o that "end" events are more salient than events in the middle of a series . Thus, th e sequence RRRRGGGG (small letters indicating weak stimuli) might well hav e much the same effect on an observer as the sequence RRRRGGGG. Perhaps th e effect of the middle stimuli, RRRGGG, is in fact negligible, in which case th e G ( . . . indicating no stimulus at all) would color-phi sequence R also produce the response RRRRGGGG . I have described such a process to illustrate a potential mechanism for the serial-position effect in verbal learning, i n which end stimuli also have a stronger effect than stimuli in the middle of a

Stimulus Subject Respons e DISTANC E

R

T

I STATE. .. R

M E

I STATE... R

I STATE .. . I STATE . . .

R

V

I STATE . . .

G

I STATE. ..

G

I STATE...

G G

STATE . . . STATE . . . "RRRRGGGG "

FIG U R E 9 .5 Event history in a control experiment that produces the same response (subjec t report) as the color-phi experiment . A smoothly moving illuminated spot (indicated by the change i n the position of R or G from time step to time step) changes color midway from red to green . After a delay, the subject reports, veridically (RRRRGGGG) . It is the identity of the responses to the tw o different histories shown in Figures 9 .4 and 9.5 that defines the color-phi effect. The effect is explained once we have rules for the states and state transitions that will yield the same output to thes e two different input sequences .

236

STADDO N

sequence (primacy and recency ; Staddon, 1998) . No doubt something similar could be devised to account for color-phi . There is one remaining problem with this account . We can readily concede the existence of a process that will give the same output to three different inpu t sequences : RRRRGGGG, R G, and RRRRGGGG. The question is , why is the response "RRRRGGGG," rather than "RRRRGGGG" or "R . . . G"? Why do people report the truncated or modified sequences as appearing lik e the unmodified sequence? Why privilege one of the three possible interpretation s over the others? It is here that one must appeal to innate mechanisms . Roge r Shepard (e .g ., 1987) has argued persuasively that our perceptual mechanism s have evolved to interpret a given stimulus sequence in the way that has bee n most probable during our evolutionary history . Thus, by responding "RRRRGGGG," rather than one of the other ways, we may simply be playin g the evolutionary odds . Given that these three sequences all produce the same internal state, the most likely state of the world is RRRRGGGG (rather than one o f the others), so RRRGGGG is what we perceive (report) . It is worth noting that the scientific problem posed by color-phi, as I hav e framed it, parallels exactly the history of research on another perceptual phenomenon : color vision . An early discovery was that people sometimes see "red " (for example) when no spectrally red light is present just as people sometime s see movement when nothing is actually moving . Later research expanded on thi s theme through the study of aftereffects, color-contrast, and "land" effects, 8 eventually showing a wide range of disparities between the color seen and the wavelengths present . The solution to the problem was the discovery of processin g mechanisms that define the necessary and sufficient physical-stimulus condition s for a person to report green, red, or any other color. 2 . Consciousness and the Brain Dennett and Kinsbourne (1992) also discuss conscious effects of electrical brain stimulation that have elicited almost mystical responses from susceptibl e commentators . In a provocative series of experiments, Benjamin Libet (e .g . , 1965, 1985) has studied the relation between recorded and applied brain electrical events and reports of consciousness . For example, in one experiment Libe t studied the tingling sensation that can be produced by a brief electrical pulse t o the hand (the left hand, say) . He found that a similar sensation could be produced by stimulating the subject's cortex (say the left cortex, so that the sensation is referred to the right hand) . Libet has reported instances in which the subject's left cortex (sensation in right hand) was stimulated before his left hand, ye t 'Polaroid Corporation founder Edwin Land became famous in the 1960s for his demonstration s of full-color "Mondrian" pictures from superimposed monochromatic images . For example, given two appropriately constructed black-and-white slides in two projectors, one with white light and on e with spectral red light, full-color images can be produced in the combined image (e .g ., Land & Mc Cann, 1971) . Land showed that the two projectors could have spectrally pure lights close together in the spectrum and still yield a full-color image . These experiments proved definitively that color perception is not linked directly to the spectral composition of the image .


23 7

the subject reported sensations in the reverse order : first left hand, then right. Libet himself views these results as a challenge to materialism, and respecte d knights of science and philosophy Sir John Eccles and Sir Karl Popper seem t o agree . They write in mystification that "this antedating procedure does not see m to be explicable by any neurophysiological process . . . . The antedating sensory experience is attributable to the ability of the self-conscious mind to make slight temporal adjustments, i .e ., to play tricks with time" (Popper & Eccles, 1977, cited in Dennett & Kinsbourne, 1992, p . 187) . It is hard to see why these results amaze . The subject's report in every case comes after the stimulating events have ceased no "time reversal" has occurred . An electrical stimulus to the cortex that is quite different from any naturally generated brain event might well take longer to interpret (i .e ., reach the point of verbal expression) than a stimulus applied to sensory end organs expressly adapted to receive stimuli . Hence, it should not surprise that the brain-induced sensation is reported as occurring after the peripherally produced one . The reversal of temporal order is puzzling only if we have some idea that consciousness is a sort of driving force, rather than a property of our system for self report . In another experiment puzzling to some, Libet (1985) asked subjects to make "spontaneous" decisions to move one hand while looking at a clock and noting the precise time they "formed the intention" to move . Libet recorded the movement, the judged time of the intention, and the time of the preintentional "readiness potentials" on the surface of the subjects' scalps . He found that times subjects reported for the genesis of their intentions lagged behind the readines s potentials by almost half a second . This finding seems to rule out a real "executive" role for consciousness . This result is interesting, but not in any way paradoxical unless one has a rather naive view of consciousness as a "first cause" of action . The control scheme illustrated in Figure 9 .6, for example, is perfectly consistent with Libet' s result . The idea is that the "consciousness" functional subsystem receives input from the outside world and passes on instructions to "everything else" to pro duce the desired behavior according to the required "spontaneous" rule . Libet' s "readiness potentials" come from everything else, not from "consciousness ." In turn, whenever "everything else," following the rule as instructed, initiates an action, it reports its behavior to "consciousness ." Each of these steps takes some time . Thus, the action of "everything else," reflected immediately in its readines s potential, occurs before "consciousness" has time to process the incoming repor t that an action is imminent and to note the real time coming in on the "stimulus " line . No doubt there are many other functional arrangements that can duplicat e the time delays that Libet found . Dissociations of the sort just described can be observed without electroencephalogram (EEG) paraphernalia . If you have ever participated in a quiz under time pressure you know that very often you know that you know the answer to a question quite some time before the answer itself arrives in consciousness .

238

STADDO N

"everything else"

I

"consciousness"

4k_

stimuli

I

inside

t

response s

outside

FIGURE 9 .6 Schematic diagram for a suggested relation between two brain subsystems : the part devoted to " consciousness " and that to "everything else ." Consciousness communicates with th e outside world and also receives input from everything else (i .e ., the unconscious systems of th e brain). Each step—from outside world to consciousness, from consciousness to everything else, an d back, and the response—takes some time . In such a system, brain signals directly from everythin g else may well precede the response of recognition from consciousness .

Sometimes, of course, the delay is very long indeed . In the so-called tip-of-thetongue phenomenon, for example, you may have the "feeling of knowing" day s before the answer pops into consciousness . If something corresponding to Li bet's readiness potential could be measured, perhaps it would occur more or les s contemporaneously with the "feeling of knowing" and thus some time before th e actual answer becomes available. According to the scheme illustrated in Figur e 9 .6, the feeling of knowing is transmitted to consciousness from "everythin g else" in advance of the actual information desired and in response to a "re quest" for the information from consciousness . The muddled cognitive view of consciousness grows out of a conflation of th e subjective (Domain 1) and objective (Domains 2 and 3) levels . This conflation has led to needless puzzles about the issues raised by experiments such as thos e mentioned . Moreover, the naive and dismissive view of behaviorism shared, in a fashionable "cognitive correctness," by much of the psychological communit y has prevented serious consideration of the kinds of arguments I have just offered . (The behaviorists themselves are partly to blame, of course, for reasons I de scribed earlier.) Nevertheless, the lack of a behavioristic perspective on thes e problems has allowed views that often verge on the mystical to gain a respectfu l audience. III . CONCLUSIO N Behaviorism was once the dominant movement in American psychology . It was eclipsed by the "cognitive revolution" in the late 1970s . Two things seem to

23 9


have favored the cognitive movement : (1) the digital computer, which for th e first time allowed mentalistic ideas to be simulated and overcame behavioristi c criticisms that cognitive theories were vague and anecdotal ; and (2) the takeover of behaviorism by Skinnerian radical behaviorism, with its strong bias agains t formal theory and its belief that psychology is nothing but the collection of orderly experimental data . In recent years, cognitivism has come under increasing attack. The computer metaphor has been criticized as fundamentally inadequate, and workers on behavior-based artificial intelligence have suggested that the cognitive-science approach through representation has failed to achieve real intelligence . Lowerlevel, nonsymbolic tasks, they argue, are more fundamental to biological intelligence, evolved earlier, and are more difficult to recreate by artificial methods . At the same time, a new theoretical behaviorism is emerging that shares th e behavior-based–Al emphasis on intelligent behavior as the outcome of interactions among independent, unintelligent agents . The core assumption of TB i s parsimony and an ecumenical attitude toward theoretical constructs . The aim i s "to find the point of view from which the subject appears in its greatest simplicity" without commitment in advance to constructs that are either mentalistic o r neurophysiological . Theoretical behaviorism can deal with mentalistic problems such as "consciousness" without either ignoring them, obscuring the distinctio n between what is inside versus what is outside the organism (like radical behaviorism), or confusing what is felt with what can be measured (like some recent cognitive discussions) . Theoretical behaviorism promises to provide theoretical links between behavior and the brain that rest on real understanding of the behavior, rather than on mentalistic presumptions about how brain–behavior relations "must" be arranged.

ACKNOWLEDGMENT S Research was supported by grants from the National Science Foundation and the Nationa l Institute of Mental Health.

REFERENCE S Bates, E ., & Elman, J. (1996) . Learning rediscovered. Science, 274, 1849-1851 . Becker, G. S . (1987). Economic analysis and human behavior . In L . Green & J . H . Kagel (Eds .) , Advances in behavioral economics, (Vol. 2, pp . 3-17) . Norwood, NJ: Ablex . Brooks, R . A. (1991) . Intelligence without representation . Artificial Intelligence, 47, 139-159 . Chomsky, N . (1964) . Current issues in linguistic theory. The Hague: Mouton . Commons, M . L ., Grossberg, S ., & Staddon, J . E. R . (Eds.) . (1991) . Neural networks of conditioning and action, the XIIth Harvard Symposium. Hillsdale, NJ : Erlbaum. Crick F., & Koch, C . (1992, September) . The problem of consciousness . Scientific American, pp. 152-159.

240

STADDO N

Davis, D . G . S ., Staddon, J. E . R ., Machado, A., & Palmer, R . G. (1993) . The process of recurrent choice . Psychological Review, 100, 320-341 . Dennett, D. C . (1991) . Consciousness explained. Boston : Little Brown . Dennett, D . C., & Kinsbourne, M. (1992) . Time and the observer: The where and when o f consciousness in the brain . Behavioral and Brain Sciences, 15, 183-247 . Dragoi, V. (1996) . A dynamic theory of acquisition and extinction in operant learning . Neural Networks, 9, 1-29. Ezzell, C . (1996, June) . The enigma of consciousness : Still unanswered . Journal of NIH Research, 8, 24-25 . Feigenbaum, E ., & Feldman, J . (Eds .) . (1963) . Computers and thought. New York : McGraw-Hill. Flanagan, O. (1992) . Consciousness reconsidered. Cambridge, MA: MIT Press, Bradford . Gray, J . A. (1992) . Consciousness on the scientific agenda . Nature, 358, 277 . Hayek, F. A. (1952/1979). The counterrevolution of science : studies in the abuse of reason . Indianapolis, IN: Liberty Press . Herrnstein, R . J . (1961) . Relative and absolute strength of response as a function of frequency of reinforcement. Journal of the Experimental Analysis of Behavior, 4, 267-272. Hirshleifer, J . (1984, August) . On the emotions as guarantors of threats and promises. UCLA Economics Department Working Papers, 337, 30 . James, W. (1890) . Principles of psychology. New York: Henry Holt. Johnson-Laird, P. N . (1988) . The computer and the mind. Cambridge, MA : Harvard University Press . Killeen, P. R . (1995) . Economics, ecologics, and mechanics : The dynamics of responding unde r conditions of varying motivation . Journal of the Experimental Analysis of Behavior, 64 , 405 -431 . Killeen, P. R., & Fetterman, G. J . (1988) . A behavioral theory of timing. Psychological Review, 95, 274-295 . Land, E. H., & McCann, J . J. (1971) . Lightness and retinex theory . Journal of the Optical Society of America, 61, 1-11 . Libet, B . (1965) . Cortical activation in conscious and unconscious experience . Perspectives in Biology and Medicine, 9, 77-86 . Libet, B . (1985). Unconscious cerebral initiative and the role of conscious will in voluntary action. Behavioral and Brain Sciences, 8, 529-566 . Machado, A . (1997) . Learning the temporal dynamics of behavior. Psychological Review, 104, 241-265 . Miller, G. A . (1962) . Psychology: the science of mental life. New York: Harper and Row. Minsky, M . (1969) . Computation: Finite and infinite machines Cambridge, MA : MIT Press. Minsky, M . (1985) . The society of mind. New York: Simon & Schuster. Neisser, U. (1967). Cognitive psychology. New York : Appleton-Century-Crofts . Newell, A. (1973). You can't play 20 questions with nature and win : projective comments on th e papers of this symposium . In W. G. Chase (Ed .), Visual information processing (pp. 283-308). New York : Academic Press . Popper, K. R., & Eccles, J . C . (1977) . The self and its brain . Berlin : Springer-Verlag . Pylyshyn, Z . (1984) . Computation and cognition : Toward a foundation for cognitive science . Cambridge, MA: MIT Press, Bradford . Searle, J . (1992). The rediscovery of the mind . Cambridge, MA : MIT Press, Bradford . Shepard, R. N . (1987) . Evolution of a mesh between principles of the mind and regularities of the world. In J . Dupre (Ed.), The latest on the best : Essays on evolution and optimality (pp . 251-275) . Cambridge, MA : MIT Press, Bradford. Skinner, B . F. (1950/1961). Are theories of learning necessary? Reprinted in Skinner, B . E (1961) . Cumulative record (2nd ed .). New York : Appleton-Century-Crofts . Skinner, B . E (1963) . Behaviorism at fifty. Science, 140, 951-958 . Staddon, J . E. R. (1973) . On the notion of cause, with applications to behaviorism . Behaviorism, 1, 25-63 .


24 1

Staddon, J . E. R . (1982) . Behavioral competition, contrast, and matching . In M. L . Commons, R. J . Herrnstein, & H . Rachlin (Eds .), Quantitative analyses of operant behavior : Matching and maximizing accounts : Vol. 2. Quantitative analyses of behavior (pp . 243-261) . Cambridge, MA : Ballinger. Staddon, J. E . R. (1983). Learning and adaptive behavior. Cambridge, England: Cambridge University Press . Staddon, J . (1993a) . Behaviorism : Mind, mechanism, and society . London: Duckworth . Staddon, J . E . R . (1993b). On rate-sensitive habituation . Adaptive Behavior, 1, 421-436 . Staddon, J . E . R . (1993c) . The conventional wisdom of behavior analysis . Journal of the Experimental Analysis of Behavior 60, 439-447 . Staddon, J . E . R . (1997) . Why behaviorism needs internal states . In L . J. Hayes & P. M . Ghezzi (Eds .), Investigations in behavioral epistemology . (Pp . 107-119) . Reno, NV: Context Press . Staddon, J. E . R . (1998) . The dynamics of memory in animal learning . In M . Sabourin, F. Craik & M . Robert (Eds .), the Advances in Psychological Science, Vol . 2 . Proceedings of the XXVI International Congress of Psychology, Montreal, August, 1996 . Hove, UK: Psychology Press . Pp . 259-274 . Staddon, J . E. R ., & Bueno, J . L . O. (1991). On models, behaviorism and the neural basis o f learning . Psychological Science, 2, 3-11 . Staddon, J . E . R., & Higa, J . J . (1996) . Multiple time scales in simple habituation . Psychologica l Review, 103, 720-733 . Staddon, J . E . R ., & Zanutto, B . S . (1997) . Feeding dynamics : Why rats eat in meals and what thi s means for foraging and feeding regulation . In M . E . Bouton & M . S . Fanselow (Eds .), Th e functional behaviorism of Robert C. Bolles : Learning, motivation, and cognition. Washington, DC : American Psychological Association . Pp . 131-162 . Staddon, J . E . R., & Zanutto, B . S . (1998). In praise of parsimony. In C . D . L . Wynne & J . E. R . Staddon (Eds .), Models for action : Mechanisms for adaptive behavior. New York: Erlbaum . Pp . 239-267 . Staddon, J . E . R ., & Zhang, Y. (1991) . On the assignment-of-credit problem in operant learning . In M . L . Commons, S . Grossberg, & J . E . R . Staddon (Eds .), Neural networks of conditioning and action, the XIIth Harvard Symposium (pp . 279-293) . Hillsdale, NJ : Erlbaum. Turing, A . (1950) . Computing machinery and intelligence . Mind, 59, 433-460 . Weizenbaum, J. (1976) . Computer power and human reason : from judgment to calculation . Sa n Francisco : W. H . Freeman .

Williams, B . A . (1986) . On the role of theory in behavior analysis . Behaviorism, 14, 111-124 . Ziman, J . (1968) . Public knowledge : The social dimension of science . Cambridge, England : Cambridge University Press.


10 BIOLOGICA L BEHAVIORIS M

WILLIAM TIMBERLAK E Indiana University

I. Introductio n A. The Waning Hegonomy of Traditional Behavioris m B. Biological Behaviorism : An Inclusive Approac h C. Chapter Preview . A Brief History of the Study of Behavior II A. Observation and Anthropomorphis m B. Training and Implicit Causal Model s C. Experimental Tests of Causatio n III . Manipulation-Centered Behavioris m A. Strengths of the Manipulation-Centered Approach B. Limitations of the Manipulation-Centered Approach IV. An Animal-Centered Biological Behavioris m A. Animal-Centered versus Anthropomorphic Observation B. Animal-Centered Manipulation s C. An Animal-Centered Approach to Traditional Experimentatio n D. Constructing an Animal-Centered Causal Syste m Mode l V. The Feeding Behavior System in Rats : An Example of a Causal System Mode l


243


244

Ti MBERLAK E

A. Structure of the Feeding System B. Regulation of a Behavior Syste m C. Characteristics of Causal Systems of Behavio r VI . Biological Behaviorism and Nonperipheral Causatio n A. Evolution and Developmen t B. Physiology C. Intervening Concepts D. Mental Life VII . Summary I . INTRODUCTIO N With his blunt and tendentious presidential address at the 1913 meetings of the American Psychological Association, John Broadus Watson (1913) is said t o have founded behaviorism . Foreshadowing his future success in advertising , Watson followed his speech with a multipronged campaign of experiments, conceptual journal articles, an argumentative book (Watson, 1919), and a bit of old fashioned arm twisting . His views struck a chord in American culture, and behaviorism quickly gained sufficient influence to set the agenda for much of scientific psychology over the next 50 years . Watson's behaviorism emphasize d the influence of the environment rather than heredity, the importance of peripheral as opposed to central (especially mental) causation, and the importance o f simple learning mechanisms in the construction of complex behavior. The behaviorist approach rejected anecdotes, introspection, and anthropomorphism i n favor of emulating the focus of the physical and medical sciences on careful experimental manipulations and standardized apparatus and procedures . A . THE WANING HEGONOMY OF TRADITIONAL BEHAVIORIS M By the 1960s the dominance of behaviorism had begun to falter . Its center had fractionated on issues of theory and conceptual analysis . Its sphere of influence was invaded by the study of cognition, ethology, development, and neuroscience . Even the strengths of traditional behaviorism, its focus on immediat e peripheral causation and use of standardized apparatus, procedures, and terms , were viewed as limiting (see Johnston, 1981) . Researchers interested in human cognition suggested behaviorism was a lengthy and unfortunate distraction from the true concern of psychology the study of the mind (see Gardner, 1987 ; Wasserman, 1993) . The focus of psychology began to shift toward both the developing techniques of neuroscience and the study of human-related cognitiv e phenomena, such as thinking, information processing, problem solving, concep t formation, memory, language, and consciousness . To complete the atmospher e of siege, biologist E . O . Wilson (1975) argued that the entire behavioral level of

24 5

10 BIOLOGICAL BEHAVIORISM

explanation would soon be assimilated by population biology from above and neurobiology from below. B . BIOLOGICAL BEHAVIORISM : AN INCLUSIVE APPROAC H Despite the frequent current assumption that behaviorism is now in its twilight, I see considerable evidence that a behavior-based analysis remains fundamentally sound and, in fact, is critically important in understanding behavior , evolution, physiology, and mind . Viewed through the lens of history, we can se e continued progress from unsystematic observation and anthropomorphic inference of causation through the development of a firm conceptual and procedura l basis for a modem science (Galef, 1996) . Traditional behaviorism provided technological and conceptual underpinnings for modem research in areas as divers e as cognition (see Staddon, 1993), neuroscience (Kandel, Schwartz, & Jessel , 1991), behavioral biology (Timberlake, 1993), the study of imitation and cultur e (Heyes & Galef, 1996), artificial intelligence (Beer, 1990), and animal welfar e (Dawkins, 1990) . However, none of these current incarnations of behavioris m appears to incorporate the full strengths and diversity of its ancestral strains o r current relatives . This chapter is based on the premise that these diverse elements can be integrated within the approach of biological behaviorism . The biobehavioral approach is based on animal-centered observation and experimentation that attempts to analyze and account for the complex causal structure underlying the fi t between an animal and its environment . In the version of biological behaviorism presented here, this causal structure takes the form of systems of behavior that incorporate perceptual-motor structure and multiple motivational processes . While clarifying and preserving the contribution of traditional behaviorism, biological behaviorism potentially relates better to genetics, ontogeny, and physiology on the one hand, and motivational, ecological, and evolutionary function on the other. C . CHAPTER PREVIEW In the first section of this chapter, I recount a few elements of the histor y of the study of behavior that highlight the progress from anthropomorphis m to experimental manipulations and scientific laws . Following a discussion o f strengths and limitations associated with the manipulation-centered focus o f American behaviorism, I consider the advantages of an animal-centered emphasis more akin to ethology, and how American behaviorism and ethology migh t be combined to produce the causal systems analysis characteristic of biological behaviorism . After presenting an example of a causal systems model in the for m of the feeding behavior system of rats, I touch briefly on how biological behaviorism deals with four areas of nonperipheral causation that have long been con-

246

TIMBERLAK E

cerns of behaviorism : evolution and development, physiology, intervening concepts, and mental events . Finally, a proviso for historical reasons as well a s limitations on both space and my expertise, the majority of examples I consider concern nonhuman animals . However, beginning with the section on manipulation-centered behaviorism the term animal can be read as referring to both human and nonhuman animals .

II . A BRIEF HISTORY O F THE STUDY OF BEHAVIO R The essence of biological behaviorism is the observation, prediction, and explanation of behavior where behavior is defined broadly and inclusively as the dynamics of living forms . In this section I briefly trace some trends in the stud y of behavior from before the Greeks through the twentieth century. Across thi s time frame the explanation of behavior moved from causal inference based primarily on speculations about human feelings and intentions (anthropomorphism ) through models of animals implicit in training procedures to causal inferenc e grounded in experimental analysis . A. OBSERVATION AND ANTHROPOMORPHIS M

One has only to look closely at cave paintings or sift through traditional folk tales to realize that humans have been fascinated by behavior whether or not i t has immediate significance for their survival . The proverbial Martian visitin g earth today would encounter large numbers of humans intently observing behavior in sporting events, soap operas, circling hawks, water striders, children , movies, mimes, pets, joggers, and squirrels . An observant Martian would not e that humans also are compelled to infer that the behavior of others is caused b y motivations and emotions similar to those they attribute to themselves . Youn g children are inveterate observers, delighted to attribute human intentions an d feelings to even an inanimate object so long as it occasionally moves or other wise resembles something living . Teenagers spend long hours on the phone dissecting the emotions and intentions of their classmates and themselves . Eve n adults routinely speculate on the causation of behavior and surprisingly often at tribute causal motivations and feelings to nonliving entities like government organizations and automobiles . In observing other animals, most humans are drawn to human-like qualitie s of appearance and behavior and easily assign complex underlying motivations and emotions . Thus, we believe that ants behave responsibly and with foresigh t in storing food against the coming winter, while grasshoppers irresponsibly fiddle the summer away. Anthropomorphic attribution of motivation and feelings reached a high point in the writings of George Romanes, a disciple of Darwin . Although cautious in his initial statement of evolutionary theory (Darwin, 1859),


24 7

Darwin (1871) strongly encouraged anthropomorphism in his efforts to provid e evidence for evolutionary continuity between other animals and humans . Romanes (1884) attempted to formalize this practice as "ejective inference," the assumption that to know what another creature is doing and why, we must ask what we as humans would feel and do in the same circumstances . Although both the philosophical and practical difficulties of anthropomorphic inference are overwhelming, it appears a very difficult practice to avoi d (Kennedy, 1992) . The involuntary quality of inferring human-like motives for behavior, its early appearance in development, and the absence of a firm boundary between animate and inanimate objects in attributing intentions, all suggest a perceptual bias based in our social evolution . A likely guess is that anthropomorphism is based on a "tool kit" of mechanisms assembled in the course of evolution to promote adaptive social interaction among humans (Beer, 1992 ; Cosmides & Tooby, 1987) . A major problem with this social tool-kit is that it often is not successful i n predicting the behavior of even familiar humans . In the cases of unfamiliar humans from different cultures or nonhuman animals, the success of the kit can b e very spotty. There is no overwhelming reason to assume that all humans shoul d behave alike and for the same reasons, and even less cause to assume that othe r animals act as humans . Human sensory and motor capacities show considerabl e variability among individuals and can differ drastically from those of othe r species . The reliability and measurement of inferred desires and feelings ar e even less well anchored . Without considerably more knowledge of the organis m under observation, an anthropomorphic approach to behavior stands in imminen t danger of misperceiving and misinterpreting the nature of behavior and its determinants . B . TRAINING AND IMPLICIT CAUSAL MODEL S

Historically, significant improvements in predicting behavior arose a s hunters, trainers, and herders spent considerable time in contact with particular animals . Sustained interactions provided a demonstrably better basis for predicting behavior than initial anthropomorphic speculations . Early cave paintings reflect considerable knowledge of how animals could be hunted and killed, as d o the folktales of most cultures and the lore of the few cultures of "primitive" hunters that still exist (Mountjoy, 1980) . A few historical manuscripts provide intriguing insights into animal training . The ancient Sumerians, a powerful culture predating the Greeks, left records around 1400 BC describing how to acclimatize horses slowly to pulling heavy loads . By this shaping procedure the y were able to produce animals that willingly pulled heavy war chariots at hig h speeds for long distances . Perhaps the most impressive combination of observation and manipulatio n was left by Frederick II, Emperor of the Holy Roman Empire, in his treatise o n falconry, The Art of Hunting with Birds (1250/1943) . He reported remarkably

248

TIMBERLAK E

modern procedures for training a bird to hunt cooperatively with a human . For example, he described in detail how to shape a bird to fly to the trainer's glov e by gradually increasing the distance from which it was recalled . Further, he generalized the response to a variety of environments by fading the environment s into each other (Mountjoy, 1980) . In short, the emergence of pursuits like organized hunting, herding, and other purpose training required less dependence on anthropomorphic inference an d more concern with development of a working model of the particular animal . Sophisticated prescriptions for training an animal must be based on clarifying its relation to particular stimuli and how its perceptual-motor and motivational organization can be shaped and controlled by a technology of rewards and punishments . Although anthropomorphism is occasionally invoked in the assumptio n that pleasure and pain explain why behavior occurs, trainers base their work o n models of the organism that facilitate the understanding and control of its behavior. For example, training a falcon to wait before attacking involves at least an implicit understanding of the falcon's sensory-motor capabilities and its typical foraging strategies . C. EXPERIMENTAL TESTS OF CAUSATIO N

From our current vantage, experimental manipulations that test and develo p causal explanations are critical in accounting for behavior, but systematic manipulation and measurement are rare in the historical record. The mini-experiments that occurred in the process of developing a training regime are usually not s o much tests of a particular causal hypothesis as a pragmatic search for a successful technique . Concern with an explicit causal model of behavior did not becom e frequent until after the ground swell of experimental work in the physical and biological sciences that followed the Renaissance . Clear examples of causal research began to appear during the last quarter o f the nineteenth century (Galef, 1996) . Most researchers were concerned with separating the roles of learning and instinct in determining behavior. Douglas Spalding (1872) did a remarkable series of experiments investigating the extent to which movement, directed pecking, and following in chicks depended on experience . Sir John Lubbock (1882) did research on insects showing, among othe r things, that scent controlled the following responses of caterpillars . C . Lloyd Morgan (1896) explored how chicks learned aversions to bad-tasting foods . Henri Fabre (1918), a French observer, demonstrated the surprisingly automati c nature of insect behavior frequently considered to be rational . C . O . Whitman (1898) using pigeons and doves and O . Heinroth (1910) using ducks provided evidence that motor patterns were similar to morphology in characterizing different species . Finally, Sherrington (1906) summarized his extensive work examining the elicitation and control of reflexes . During the first half of the twentieth century, two causal approaches came to dominate the experimental study of behavior : American behaviorism as outlined


24 9

by Watson and Thorndike, and modified by many others (especially Skinner, 1938, 1974) ; and ethology, a biological view developed conceptually by Tinbergen (1951) and Lorenz (1937, 1950), and added to by others (see Baerends , 1988 ; Thorpe, 1979) . In many ways, the two approaches were remarkably similar. Both rejected anthropomorphism, vitalism, and purposivism, attempted t o distinguish instinct and learning, and divided the determinants of behavior int o sensory-motor organization and motivation . Both subscribed to the idea that science was a public endeavor and that data should show intra- and interobserver reliability such that any knowledgeable researcher should be able to repeat well controlled observations and experiments and get the same results . Despite these remarkable similarities, representatives of each discipline initially focused on th e few differences, some having political overtones (Glickman, 1985 ; Lehrman, 1953) . American behaviorism was heavily influenced by physics, engineering, an d medical physiology. It used laboratory-based techniques to assess and press the limits of what individual animals could do, perhaps echoing the American focu s on remaking humans through self-improvement and education (Boring, 1968) . Because the experimenter's manipulations simultaneously construct behavio r and provide its function, I refer to this general approach as manipulation-centered behaviorism . It focused on developing abstract laws relating the experimenter's manipulations to an index of behavior. Necessary information about the sensory motor organization, motivation, and functioning of animals was incorporated into the apparatus and procedures, but was not specified in the laws . Ethology, on the other hand, arose from naturalistic biology and was concerned primarily with the determinants of the typical behavior of a species in a natural setting . Although ethologists attempted to generalize models of motivation and sensory-motor organization (e .g ., Tinbergen, 1951), their focus was more on the specific motivational and perceptual-motor organization of particular species . I refer to this aspect of their approach as animal-centered behaviorism . It combines observation and experimentation to create a functional mode l of a particular animal . I consider these two approaches more fully in the nex t two sections . III . STRENGTHS AND LIMITATIONS O F MANIPULATION-CENTERED BEHAVIORIS M To escape the close embrace of anecdotal evidence, anthropomorphic interpretation, and introspection, American behaviorists stressed an objective experimental approach patterned on the techniques and conceptual analysis of the physical and medical sciences . The focus was on the experimental manipulatio n of presumed causal variables and documentation of the effect in terms of simpl e measures . For example, experimenters documented the effect of amount of reward by measuring the speed with which a rat ran from the start box of a straight

250

TIMBERLAK E

alley to the reward in the goal box . Functional relations between variables were produced by presenting multiple levels of the independent variable and measuring the effects on the primary measure . Control groups were used to rule ou t alternative explanations . The manipulation-centered attribute of traditional behaviorism was critica l to its successful development . It was in many ways a powerful model of idealized science that produced an instant methodological respectability . At the same time there are reasons to question whether manipulation-centered behavioris m has limitations that may interfere with a comprehensive analysis of behavior. A . STRENGTHS OF THE MANIPULATION CENTERED APPROAC H The strengths of manipulation-centered behaviorism have to do with its scientific respectability, replicability, and generality. 1. Standardized Equipmen t A key for ease of communication and the confirmation and further analysis o f results was the development of standard types of apparatus in the form of multiple choice mazes, straight alleys, discrimination boxes, jumping stands, and en closed test chambers for leverpressing in rats and keypecking in pigeons (e .g . , Warden, Jenkins, & Warner, 1935) . Experimenters carefully isolated the behavior of the experimental subject from any influences except the manipulations of the experimenter. To increase the generality of the results, experimenters set arbitrary (usuall y human-related) tasks for animals, tasks in which the response to be learne d seemingly was well removed from the possibility of an instinctive solution . For example, Thorndike (1911) carefully examined how cats learned to open a "door" separating them from food, a type of problem mentioned frequently i n the cat learning anecdotes reported by Romanes (1884) . To open the door, the cats were required to use an unfamiliar response, such as pulling a string, or flip ping up a latch . The result of this care was a common language of manipulation s and results that enabled growth of a community of scientists who could readil y examine, replicate, and extend any finding . 2. Universal Measures Accompanying the emphasis on standardized apparatus was a tendency to emphasize measures that readily fit the physical universe, such as speed, rate , and quantity. In physics entities can be characterized in terms of universal measures like mass and speed . In conjunction with the general laws of physics, thes e few measures can be used to predict the behavior of diverse-appearing element s like rocks, water, and bird feathers . For example, in a gravitational field, rocks , water, and containers of feathers all behave the same. Manipulation-centered


25 1

behaviorism tried to attain a similar level of generality by standardizing on physical measurements such as speed and quantity. 3. Experimental Paradigm s Particular combinations of apparatus, procedures, and measures came to de fine the experimental paradigms that soon dominated the study of learning . Thus, experimenters used the paradigms of operant learning (Skinner, 1938) , Pavlovian conditioning (Pavlov, 1927), and assorted trial-and-error maze problems to trace functional relations between independent variables (manipulate d by the experimenter) and dependent response measures . For example, the spee d of traversing a runway to food was shown to be monotonically related to th e amount and delay of food in the goal box . In Pavlovian conditioning the amount of salivation was directly related to the intensity of the unconditioned food stimulus (see Kimble, 1961 ; Mackintosh, 1974) . Experimental paradigms furthe r clarified the types of variable that controlled behavior and provided a commo n currency in which to phrase and evaluate hypotheses . 4. Critical Test s After a period of assembling their results into general laws relating independent and dependent variables, investigators began to focus on distinguishin g among alternative causal explanations by using critical experiments . Researcher s attempted to create "strong inference" experiments that pitted alternative causa l explanations against each other in a way that would determine which was correc t (see Platt, 1964) . The ideal set of experiments was a series of competitions at th e end of which only the correct theory was undefeated . B . LIMITATIONS OF THE MANIPULATIONCENTERED APPROAC H Without question, the manipulation-centered approach gave the study of behavior the respectability of scientific procedures and a philosophical foundatio n borrowed from physics and logical positivism . However, there are reasons to think these tools may have been borrowed too easily . A manipulation-centered approach appears to work best when dealing with simpler and better understoo d phenomena, such as the 92 stable elements and 3000 stable minerals found i n the readily accessible physical universe (Mountjoy, 1980) . In contrast there ar e millions of more or less stable species, and probably billions of semistable behaviors sensitive to a remarkable array of variables including differences amon g species . Like any robust approach, traditional behaviorism can in theory account for all of these effects after the fact, but we are interested here in an approach that makes the complexities of behavior easier to predict and account for . Manipulation-centered behaviorism has had difficulties with oversimple operationalism, unexpected behavior in standard paradigms, the generalization of

252

TIMBERLAK E

procedures and laws, and the entanglement of date, procedures, apparatus, an d theory. 1. Operationalis m Operationalism is a primary example of psychology borrowing from th e physical sciences to get a head start on scientific procedures and respectability . A difficulty was that physics and psychology were at two different places in their development. Physics had converging ways of using and measuring concept s like distance that were quite satisfactory under most circumstances. But in tryin g to define distance in the twentieth century physics of relativity and quantum mechanics, it became important to know how a measurement was made . A differen t measurement procedure could produce a markedly different result . Operationalism allowed distance to have a defensible meaning in these different contexts . Because psychologists lacked a history of satisfactory laws relating concept s and their measures, operationalism in psychology served primarily to ensure explicit definitions of concepts . Unfortunately, without a background of relate d concepts and converging measures, there was no prescribed way to assess th e reasonableness or coherence of these definitions . It was too easy to confuse precise definitions with adequate models . For example, defining drive as hours o f food deprivation yielded a precise definition, but not necessarily an adequate o r valid model of drive effects . Many animals usually eat in meals restricted to th e active period of their rest-activity cycle, so the particular circadian hours of de privation are important . Further, rats at least do not change their subsequent in take as a function of the length of deprivation periods in the 1 to 6 hour range , requiring at minimum a complex function relating deprivation and feeding . In short, the use of operationalism in psychology established an important traditio n of explicit definitions, but frequently failed to encourage the work necessary t o establish their general validity . 2. Unexpected Behavior in Standard Paradigm s A second limitation in using the manipulation-centered approach is the complexity of species-typical behavior that can emerge unexpectedly in the contex t of a seemingly standard experimental paradigm. For example, merely deliverin g free food periodically produces complex behavior in pigeons (Skinner, 1948 ; Timberlake & Lucas, 1985) . The simple addition of a water bottle to a rat operant conditioning chamber resulted in a remarkable increase in water intake unde r interval food schedules to levels well above the typical daily intake (Falk, 1971) . Other such "adjunctive behaviors" readily appeared in slightly different environmental conditions (Staddon, 1977) . Similar problems can arise when new species are used in a standard apparatus . For example, neither pigeons, gerbils , nor raccoons take particularly well to solving mazes with walls (see Timberlake, 1983b, 1990) .


25 3

3. Difficulties in Generalizing Procedures and Behavioral Laws A third constraint is the difficulties often encountered in transporting th e technology and laws discovered in the laboratory to real world settings (Timber lake, 1995, 1997) . This has been an area of some notable successes in operan t conditioning, but there are notable problems as well . To the extent that experimenters are able to ensure laboratory-like environments in real world settings , typically by pretraining subjects and closely controlling stimulus conditions , contingencies, and responses, the outcome is frequently successful . For example, programmed teaching and behavior therapy have shown highly successful applications of laboratory-based principles . But even with both knowledge and control, surprises readily occur. For example, the Brelands, in the process of trainin g animals for display, discovered many examples of inappropriate responses (misbehavior) in their subjects that emerged despite their careful use of standard operant training procedures and their laboratory experience (Breland & Breland , 1961) . A more complete account of behavior should predict what and when difficulties will occur in generalization . 4. Entanglement of Data, Experimental Paradigms, and Theor y Finally, manipulation-centered research has been marked by considerable en tanglement of data, procedures, apparatus, and theories . Many data were produced by manipulations focused on discovering, illustrating, and testing general principles and theories . When such a theory inevitably was shown to be less tha n accurate, researchers too often abandoned not only the theory, but also the experiments that supported it . The experimental ideal of strong inference may hav e exacerbated this problem by encouraging experimenters to pit embryonic general theories against each other in critical experiments before their ties to specific local behavior were sufficiently clear. In several instances in the behavioris t study of learning, a series of critical tests either left no theory standing or faile d to find a distinction among the alternatives, in both cases providing the field nowhere to go . Such impasses favored simpler models that focused on more restricted data sets instead of encouraging the longterm development of comple x models of behavior. IV. AN ANIMAL-CENTERE D BIOLOGICAL BEHAVIORIS M The preceding discussion of the limitations of manipulation-centered behaviorism in no way argues that operationalism and experimental analysis ar e wrong, or that experimental manipulation of variables in carefully controlle d conditions should be stopped. To the contrary, the manipulation-centered approach has been a critical contributor to the emergence of the scientific study o f behavior and remains a cornerstone of its success . Rather, the argument here is

254

TIMBERLAK E

that oversimple dependence on manipulation for definition and analysis has contributed to the limitations outlined above . For longterm progress I believe operational definitions and manipulation-based analyses must be embedded within th e larger framework provided by the structure and processes of a functioning organism . Such a functional framework necessarily exists prior to the experimenter's manipulations, helps determine their effects, and provides a basis fo r interpreting and assimilating the results . It is instructive to compare the extent to which the importance of behavioral phenomena change with the rise and fall of particular causal theories against th e relatively steady accumulation of knowledge about animals by experienced hunters and trainers . The problem is not with the talents of individual scientist s because they, too, show steady improvement in roles such as trapper and trainer . The contrast appears to turn on differences in the dominant type of model that scientists and trainers assemble and use to interpret behavior. For most scientists , data are organized by type of manipulation and the abstract law to which it relates . For example, a researcher may manipulate the intensity of the unconditioned stimulus (US) and state the results in terms of a general law of conditioning. In contrast, a trainer's data are more often organized within a less formal bu t animal-centered framework that aims to specify the relevant perceptual-moto r organization and motivational processes of a particular animal and how this organization interacts with the purposes and manipulations of the trainer. I believe there is much to be gained by explicitly developing an animal-centered approach for behaviorism that combines observational and interactiv e knowledge with the technology of experiments in generating a causal syste m model. Such a framework has been shown to be suited for combining contributions of neurophysiology, evolution, and development with the use of experimental tools able to explore and test such a model (e .g ., Fanselow, 1994 ; Hogan , 1994 ; Timberlake, 1994) . An animal-centered approach also facilitates contac t between laboratory experimentation and applied settings (Timberlake, 1995 , 1997), In the following I clarify the nature of animal-centered observation an d how experimental manipulations can be used to develop and test an animal-centered model . In the subsequent section on causal systems, I illustrate the construction of a specific animal-centered model . A . ANIMAL-CENTERED VERSU S ANTHROPOMORPHIC OBSERVATION An important technique in developing an animal-centered model of behavior is for the observer to attempt to assume the place of the animal (Timberlake & F . J . Silva, 1994) . This approach was used to good effect by protoethologists, suc h as von Uexkull (1934) and Craig (1918), in developing models of how the fit of the animal with the stimulus environment directed and regulated behavior. Animal-centered observation was also used by American behaviorists in designin g specific apparatus, procedures, and stimuli. For example, Munn (1950, pp . 146-


25 5

151) summarized research on concept formation showing that rats have a hierarchy of cues in discriminating geometric forms, with the bottom edge of the cu e being differentially salient . To ensure appropriate manipulation of the cues confronting a rat, experimental psychologists recognized the importance of beginning with a rat's-eye view of the apparatus . However, despite their practical sensibilities, American behaviorists avoided emphasizing an animal-centere d approach, and it is not difficult to see why . 1 . Anthropomorphism Revisited Most observers who have focused on an animal's view of the world have appeared eager to fall into anthropomorphic speculation . In recent times, this eagerness has taken the form of an argument that anthropomorphism is central t o good research (e .g ., Bekoff & Jamiesen,1991 ; Cheney & Seyfarth, 1990; Dun bar, 1984 ; Griffin, 1981) . Popular nature films about animals almost invariabl y are overwhelmingly anthropomorphic . For example, the narrator invites the viewer to assume the role of a hungry lioness hunting for a meal on the savannah . Viewers (basically, humans in lion's clothing) are asked to imagine skulkin g through the tall dry grass of the veldt feeling the brush of the stalks, the looming heat of the sun, insects biting, our stomachs growling . Finally in place, w e exchange significant looks with our sisters and so coordinate a successful am bush to procure a fresh meal from a nearby herd of zebras . Perhaps not surprisingly, the narrator doesn't invite us to imagine how it feel s to break the back of the zebra, eat its entrails while it lives, and shatter its bones . There are few positive referents to these behaviors in our personal experienc e and it could be argued that our human emotional reactions would interfere wit h understanding these behaviors . In reality, I suspect that our emotional reaction s interfere with most anthropomorphic-based understanding of behavior . We simply are more aware of the intrusion of aversive reactions because we feel uncomfortable . The difficulty with explaining behavior by inferring human-like perceptions , feelings, and motivations is that it is not sufficiently grounded in a causal analysis of behavior. Imagining the feelings and motivation of a hunting lioness is ap t to reveal no more of the determinants of her behavior than imagining the feelings of a trampled blade of grass helps us understand why and how it spring s back up . Following the anthropomorphic style of the movie commentator, we could imagine the blade's shock as it was suddenly crushed by the immense weight of the lioness's foot . Cut off from the life-giving light and fighting despair, it might tense to spring back toward the sun when the oppressive weight was lifted, celebrating its rebirth by straining toward the light . But this account adds nothing to simple description in helping us understand the behavior of th e grass . To be sure, anthropomorphic speculations sometimes suggest experimenta l questions of potential relevance to predicting specific behavior (Cheney & Seyfarth, 1990 ; Dunbar, 1984 ; Martin & Bateson, 1993) . However, researchers typi-

256

TIMBERLAK E

cally fail to realize that anthropomorphism is not the key to this issue . What is critical is the observer's implicit causal model of the animal built up over hour s of functional observation . For example, if anthropomorphism were the key to understanding and predicting the behavior of vervet monkeys, then, as a practiced anthropomorphizer, I should be as good at prediction as Cheney and Seyfarth (1990) . But I am not, because I lack the animal-centered model of vervet behavior that Cheney and Seyfarth have developed . Without this model, my anthropomorphism provides a woefully incomplete, if not crudely false, sense o f understanding . 2. Theromorphic Observation The animal-centered approach assumes the point of view of the lioness, but a s a lioness, not a human. Perceptually, what can a lioness see at eye height on th e veldt? What do we know about the sensitivity of the visual system of big cats t o movement and the ability to resolve adjacent edges accurately? What sort o f stimulus configurations produce the greatest attention? Which way is the win d blowing, and how sensitive is the olfactory system to relevant types of molecule ? Behaviorally, what are a lion's reactions to moving stimuli? What is its top spee d relative to that of the prey? What killing techniques have evolved in its repertoire? What is the lion's experience with this prey? With this location? What i s its body state? What changes in behavior occur related to the presence of other members of the pride? In other words, do not ask what you would see if you as a human were in the place of the lion, ask how the lion perceives, what sorts o f motor and strategic alternatives are available ; and what social context, experience, and evolutionary history link the animal to the present environment. A domestic rat in a maze is a less dramatic subject than a hunting lioness, bu t similar questions arise . What can a rat see from the height of its eyes above th e substrate and their placement on the head? How is its retina differentiated ? What patterns and lines can it visually resolve? What can it smell and hear? How does the stimulus situation relate to its response organization and ecological niche? What cues typically control its navigation? Do rats follow trails in natura l settings? Are social stimuli important in trail-following behavior ? At first glance, theromorphism may appear to raise the same epistemologica l question as anthropomorphism how can a human ultimately see or know wha t another animal sees or knows? But theromorphic seeing differs from anthropomorphism in that it is based on convergent information from behavior, physiology, and the results of experimental manipulations . By examining and testin g relevant neurophysiological mechanisms, we can often induce and demonstrat e stimulus filtering . By examining behavior while manipulating stimulus characteristics (e .g ., Tinbergen, 1951 ; von Uexkull, 1934), or response contingencie s (Skinner, 1938), researchers can discover and/or differentiate functional relation s among stimuli, neurophysiological organization, and behavior . In an animal-centered approach, these relations should be referenced to and instantiated within a model of the animal that can make further predictions, and incorporate subsequent additions and corrections .


25 7

B . ANIMAL-CENTERED MANIPULATION S

To perform animal-centered manipulations it is necessary to have at least a preliminary model of the functioning animal . In the twentieth century, protoethologists such as Craig (1918) and von Uexkull (1934) led the way by extensively observing the behavioral sequences of animals, and making guesses a s to the controlling stimuli and regulatory processes involved . The ethologists, le d by Tinbergen (1951), added experiments in which the manipulations were base d on exploring and testing models developed through observation . Many of th e early studies explored the stimulus selectivity of the perceptual-motor organization that produced "fixed action patterns" (e .g ., Tinbergen & Perdeck, 1950) . Others examined the sequential and hierarchical organization underlying regulatory behavior (Baerends, 1988 ; Wilz, 1967) . Progress was also made in linkin g physiological analyses to an animal-centered model (e .g ., Nelson, 1964 ; vo n Holst & von St . Paul, 1963) . Animal-centered manipulations were used in America prior to the extensiv e development of Watsonian behaviorism (e .g ., Small, 1899 ; see Galef, 1988 ; Timberlake, 1983a), and regained a degree of importance during the constraint s on learning controversy of the 1970s . In the latter controversy researchers argued about the extent to which traditional learning theory predicted deviations from equipotential associations between all stimuli and responses (e .g ., Selig man, 1970) . Those who propounded acknowledgment of specific constraints i n essence were arguing for an animal-centered approach, but they lacked an animal-centered framework . In the absence of an alternative to the manipulation centered approach, the evolutionary and ecological aspects of the constraint s movement were reduced to differences in equation parameters describing learning speed (e .g ., Logue, 1979) . An opportunity for developing more specific models of animals was largely missed (although not entirely, see Johnston, 1981 ; Rozin & Schull, 1988) . The most productive attempts to link laboratory manipulations with functional animal-centered concerns have involved the use of laboratory circumstances to study naturally occurring problems. This approach uses laboratory apparatus to simulate specific problems encountered by a species in natura l conditions . For example, Pietrewicz and Kamil (1979) presented slides of moth s on trees to study the formation of foraging search images in blue jays . Mellgre n (1982) studied foraging in rats by burying food in sand in tubs located at different heights on climbing poles . A great deal of recent research on the determinants of memory in food-storing birds has used laboratory conditions to simulat e circumstances faced by the animal in a natural setting (e .g ., Balda & Kamil , 1989 ; Shettleworth, 1990, 1994) . Work with chickadees has used artificial tree s with drilled holes covered by velcro patches to discover the strategies and cue s used in recovering stored food. The result of this research has been the development of a more specific model of cue hierarchies (e .g ., Brodbeck, 1994) and the specific neurophysiological circuitry associated with memory for stored foo d (e .g., Sherry, Vaccarino, Buckenham, & Herz, 1989) .

258

TIMBERLAK E

A second example of research focused on the way the animal typically functions concerns the abilities of animals to navigate in space and time events . Navigation research by zoologists, particularly the work on bees and ants, focuse s directly on problems related to foraging and homing (Dyer & Gould, 1983 ; Gallistel, 1990 ; Wehner & Wehner, 1990) . Some of the timing research as well, including Biebach, Gordijn, and Krebs's (1989) work on time and place learnin g in garden warblers, clarifies how foraging relates to circadian cycles . In contrast , the research on timing by psychologists (e .g ., Gibbon & Church,1984 ; Killeen & Fetterman, 1988) and navigation (Cheng & Gallistel, 1986) has concentrated o n reaching traditional species-irrelevant conclusions, yet there remains a focus o n the contribution of the organism rather than solely the experimenter's manipulations (see also the experimental exploration of the concept of time horizon s e .g ., Timberlake, Gawley, & Lucas, 1987) . In short, the key to an animal-centered manipulation is its basis in knowledge about the animal . Manipulations are imposed based on their potential effect o n the structure and processes of the animal functioning in a particular context . For example, research on the stimulus control of patch choice or circadian feeding anticipation should be based on information about the organization of the animal's feeding system, including sensitivity to stimuli, organized response components, foraging strategies, the role of social interactions, and the characteristics of motivational processes . C. AN ANIMAL-CENTERED APPROAC H TO TRADITIONAL EXPERIMENTATION

Where does adoption of an animal-centered approach leave the traditiona l study of learning based on standardized laboratory apparatus and operant and Pavlovian procedures? Some scientists have argued that to produce an evolutionary-ecological model it is necessary to abandon traditional research and its artificial environments in favor of a predominately naturalistic approach (e .g ., Johnston, 1981) . Although I strongly agree that we need a functional basis for th e study of learning grounded in evolution and ecology, I think that the technolog y of the traditional laboratory approach already has provided and can continue to make important contributions to animal-centered ecological models . My belief i s based on three assumptions about how the animal, the experimenter, and the experimental environment interact . 1 . Arbitrary Environments and Evolutio n The first assumption is that animals have been selected for perceptual-moto r organization and motivational processes related to specific functional tasks, suc h as foraging, mating, and avoiding predation . Thus, when researchers place an animal in an environment for the first time, the animal has no choice but to "filter" or "interpret" the environmental stimuli within an evolved and developed functional framework . Evolution can produce selection for responding to arbi-

1O BIOLOGICAL BEHAVIORISM

25 9

trary stimuli only to the extent that they can be discriminated and related to survival and reproduction . But such restrictions necessarily make stimuli nonarbitrary. As a result, even in a presumably arbitrary laboratory environment, behavior will be based on the same evolution-derived mechanisms that underla y behavior in the selection environment . Lorenz (1981) made a related point in arguing for the advantages of raising wild animals in domestic environments . Because the domestic environment does not adequately support many species specific activities, the resultant compromise behaviors can more clearly reveal controlling stimuli and organized response elements . 2 . Tuning Tuning refers to the iterative process of developing procedures and apparatus , identifying and refining measures, and selecting independent variables that facilitate the emergence of interpretable behavior in laboratory tests . My second assumption about the interaction of animal, experimenter, and environment is that common laboratory paradigms reflect the repeated efforts of experimenters t o design apparatus and procedures to contact the framework of perceptual-motor organization and regulatory processes that characterize their subjects . Most experimenters argue they are simply trying to obtain reliable, vigorous, and orderl y behavior, but these very response characteristics appear to be reliable indicators of adjusting experimental apparatus and manipulations to the animal's functional framework . It is certainly no accident that the common instrumental response s used in the laboratory study of learning (keypecking in pigeons, manipulatio n responses in rats, and maze arm following in rats) appear to involve component s of naturally occurring responses and stimulus sensitivities related to foragin g (Roche & Timberlake, 1998 ; Timberlake, 1990 ; Timberlake & Lucas, 1989) . An unusually well-documented example of the importance of tuning was provided by Skinner (1938, 1959) in discussing his invention of the lever press a s the predominant operant response in rats . He left an extensive account of successive changes in his apparatus and measurement procedures as he struggled t o make the response easier to measure and more reliable (Skinner 1938, 1959 ; Timberlake, 1990) . Skinner chose to emphasize serendipity and the convenienc e of the response to the experimenter (although he did note with approval that th e temporal pattern of lever pressing directly reflected the temporal pattern of feeding) . But anyone who has tried to construct a rat lever from "scratch" rapidl y recognizes the skill and care that went into its design and placement . It is disconcertingly easy to build levers that rats will more likely gnaw than press, nos e up rather than press down, ignore, or tug so hard they will prevent its retractio n (see also Boakes, 1979) . It seems evident that the final design of Skinner's lever reflected a careful tuning to the foraging repertoire of the rat . If the reader believes that lever pressing is an unusual case, please see Willard Small's (1899 ) explanation of how he constructed his mazes to resemble the rodent runways h e saw under the floor of a beach house . Tuning, then, is a singular testimony to the ability of experimental psycholo-

260

Ti MBERLAK E

gists to fit tasks to their subjects, a critical contributor to the development o f functional laws and theories, and one of the best methods for exploring the perceptual-motor and regulatory organization of an organism. Arguably, every effective experimental procedure and laboratory apparatus has been carefull y tuned by previous experimenters to facilitate and clarify the results of their manipulations . If a manipulation does not make good contact with the animal's perceptual-motor organization and regulatory processes, there is little chance th e data will be sufficiently regular to be interpreted in terms of a fundamental law . Most importantly, the critical role of tuning in making laboratory researc h work is a compelling argument for the importance of an animal-centered approach to manipulations . Fundamentally, tuning is a method for exploring the perceptual-motor and motivational organization of the animal . What differentiates tuning from explicit model building is that when tuning is the focus, th e model of the animal is contained in the procedure and apparatus . When mode l building is the focus, tuning is specifically referenced to the concepts of th e model. Using tuning to construct a functional model of an animal raises the issu e of apparatus and procedure design to the realm of conceptual analysis . Recognizing the importance of tuning allows researchers to take better advantage o f the work of experimenters who embedded specific models in their apparatus an d procedures . Making more explicit the animal-related design of procedures an d apparatus should facilitate the modeling and understanding of behavior . 3 . Paradigms and Animal-centered Models A last advantage of an animal-centered approach is that it provides a model of an animal to test and build on rather than summarizing the results in a set of abstract general laws that depend in an unspecified way on appropriate tuning o f the apparatus and procedures . This does not mean that the general laws are unimportant. General principles are an essential strategy both in summarizin g the results of science and in serving to indicate the types of manipulations an d phenomena that are possible in a situation . But the present general laws of behavior do not include important variables reflected in the tuning of apparatus an d procedures . To generalize adequately among all animals, it seems essential t o embed general laws within models of specific species . Does this mean researchers should give up manipulations for observation ? Not at all . There is considerable support for the argument that one cannot develop an accurate model of an animal by observation alone (Lorenz, 1981) . Careful, well-thought-out manipulations are usually the key to critically testin g our understanding of the mechanisms underlying behavior . What is needed is to interpret the manipulation within the framework of a more comprehensiv e model . Standard experimental paradigms, such as Pavlovian and operant conditioning, are profitably viewed as producing their effects by constraining and facilitating the operation of an animal-centered model . In this view, experimental paradigms are important tools for revealing an d testing the organization and processes of animal-centered models . For example,


26 1

by manipulating the nature of the predictive stimulus (the conditioned stimulus , or CS) and the interval between the CS and the US, a Pavlovian procedure ca n be used to infer the organization of stimulus sensitivities and response components preceding and following food . Thus, rats treat another rat (a CS) predictin g food (a US) as a welcome colony member, increasing their approach and socia l interaction over what would occur without food pairings . Hamsters, on the other hand, avoid or attack another hamster predicting food, decreasing their social approach . The organization of the causal system related to food differs betwee n these two species (Timberlake, 1983a) . D. CONSTRUCTING AN ANIMAL-CENTERED CAUSAL SYSTEM MODEL Although the potential usefulness of an animal-centered model may be clea r by now, how to construct one has not been spelled out . Observing animals an d reading examples of naturalistic and ethological work provides a good back ground, along with general summaries of work on perceptual motor organizatio n (Dawkins, 1993 ; Land, 1983), regulation of energy balance and behavior (Stephens & Krebs, 1986 ; Toates, 1986), development (Hogan & Bolhuis, 1994) , and evolution (Slater & Halliday,1994) . There is no one way to proceed, but I have found the following steps useful (summarized in Table 10 .1) . An initial step is to choose a question about how an animal works in an ecologically functional context . For example, how does a rat find food? Or a more

TABLE 1 O . 1 : Suggested Steps in Developing a Causal System Model of Behavior 1 . Pose a question about how an organism works in a functional context. 2 . Assemble behavioral observations into a preliminary model of the structure and processes that the animal brings to the context . 3 . Integrate the preliminary model and previous experimental results into a causal system relevant to the functional context of the question . Recognize that a. The choice of apparatus design and procedures in laboratory experiments contain a grea t deal of information and many assumptions about the perceptual-motor organization an d regulatory processes of the animal . b. Typical stimuli and response measures were not selected arbitrarily, but quite likely are related to particular components of the causal system . 4 . Design an experiment that clarifies or tests a prediction of the causal system model . Recal l that experimental manipulations, especially those within the traditional paradigms of Pavlovian and operant conditioning, can be usefully viewed as techniques for exploring the contro l and functioning of a causal system. 5 . Interpret the outcome of the experiment, insofar as it fits, as the result of the interaction of a causal system with the experimental circumstances, rather than as the simple and direct resul t of an independent variable manipulated by the experimenter. 6 . As appropriate, store the results as additions to and changes in the causal system model .

262

TIMBERLAK E

specific question about rats what are the mechanisms involved in the proces s of shifting diet and foraging strategies during weaning (Alberts, 1991)? Suc h questions call attention to both the potential adaptive significance of the behavior and the mechanisms involved . A second step is to assemble the relevant observations made on this specie s into a preliminary model of the structure and processes of the animal (Timber lake & Lucas, 1989 ; Timberlake & F. J . Silva, 1994) . The major point of such a model is to provide information about the initial characteristics the animal bring s to the circumstances under consideration . The model is unlikely to be perfect , and will, of necessity, leave open further analysis of development and expression . However, it is necessary to start with a model of appropriate complexity t o allow analysis and testing in multiple converging ways . A simple model can limit consideration of potentially important determinants, while starting with a model that is too complex can make analysis impossible . In the case of rats , good sources of background information are Barnett's (1975) book on the ra t and the monograph of J . B . Calhoun (1962) summarizing his extensive observational work with the Army Signal Corps on the behavior of free-living rats . The latter work is especially strong on rat's social behavior and general activity in a natural setting . A third step is to integrate the assembled observations with previous experimental work into an initial causal system . It is important to recognize that previous experiments contain two sorts of information in addition to the reporte d results . First, the apparatus design and procedural choices in laboratory experiments, if interpreted correctly, contain a great deal of data about perceptual-motor organization and the regulatory processes of the organism (for a very usefu l general source including apparatus drawings, see Munn, 1950) . Second, th e stimuli presented and the response measures chosen are not usually arbitrary, bu t generally relate well to particular components of the causal system (see Timber lake & Lucas, 1989) . A fourth step is to design an experiment to clarify or test a prediction o f the causal system model . The more explicit the model, the larger role it ca n play in structuring the experiment, and the greater contribution of the experiment to testing the model. Recall that the traditional paradigms of Pavlovian and operant conditioning are usefully viewed as tools for exploring the contro l and functioning of a causal system . For example, the assumption that the food-related behavior of animals is organized around a sequence of searc h states leads to several predictions about the effects of the CS —US interva l on conditioning . Thus, a long CS —US interval may poorly condition a repertoire of responses related to food handling, but relatively strongly condition a repertoire of more general search behaviors (Akins, Domjan, & Gutierrez , 1994 ; Timberlake, 1994) . A short CS—US interval would be expected to show the reverse result strong conditioning of a response repertoire related to food handling, and poorer conditioning of a response repertoire related to more general search . A well-designed experiment should consider the

10 BIOLOGICAL BEHAVIORISML

26 3

nature of the predictive stimulus, the response measure, and the CS –U S interval . A fifth step is to interpret the outcome of the experiment as the result of the interaction of a preorganized causal system with the experimental environmen t and manipulations rather than as the simple and direct causal result of the value s of the independent variable . Thus, presenting a food reward does not produce an isolated strengthening effect on a particular response, but perturbs and engage s several aspects of a causal system related to feeding, including behaviors related to food handling, ingestion, area restricted search, and regulation of local an d overall intake . Phenomena such as adjunctive behavior, misbehavior, sensitization, and pseudo conditioning are no less a part of the effects of presenting food than more "normal"operant reinforcement outcomes . They are all related to the interaction of organization and processes of the animal with the experimental environment and manipulations . Pavlovian conditioning is a particularly powerful manipulation because it al lows the experimenter to control the temporal relation of a cue-stimulus to a re ward . Because no response is specified, the animal must "fill in" a sequence o f behavior, thereby revealing something of the initial organization it brings to th e experimental situation . On the other hand, operant conditioning has the advantage of allowing the experimenter to explore the availability and limitations o f response forms and sequences as well as the contribution of the stimulus circumstances (see Stokes & Balsam, 1991) . By using such manipulations as tools, i t should be possible to test and clarify the organization of an animal-centere d causal model. A final step is to store the results of an experiment in the form of additions o r corrections to the basic causal system model . In this way the information from an experiment is more likely to be used and preserved rather than discarded a s concepts and theories change . V. THE FEEDING BEHAVIOR SYSTEM I N RATS : A CAUSAL SYSTEM MODE L Perhaps the most difficult step in using an animal-centered causal system approach is how to represent what is known in a form that can be tested . Here follows a specific example of one approach in the form of a proposed feeding system for a rat . This approach resembles Tinbergen's (1951) hierarchica l motivation model in calling attention to both the structure and regulation of behavior. Structural concepts focus on the different stimulus sensitivities and repertoires related to particular functions, and on the sequences of varied and stereo typed behavior that precede and follow receipt of critical stimuli . Regulatory concepts remain a useful way to deal with the course of compensatory behavior s over time . Based on familiarity, I have picked a behavior system used in my ow n research, but there are many other examples . Similar models have been developed by ethologists (Baerends, 1988), neuroscientists (Davis, 1984 ; Fanselow,

264

TIMBERLAK E

1994 ; Heiligenberg,1991), and psychologists (Blanchard & Blanchard, 1990 ; Domjan, 1994 ; Hogan, 1994) . What makes a model animal-centered is that it attempts to represent characteristics of the animal that are engaged by the apparatus, procedures, and any experimental contingency. The present model is neither the final form nor the onl y form of an animal-centered model of the feeding system of a rat. But it doe s summarize a good deal of what is known about rat stimulus sensitivities, response components, and motivational structure as they relate to foraging, and i t can be used to predict the results of presenting particular stimuli or imposing particular stimulus-response contingencies in an experiment . A. STRUCTURE OF THE FEEDING SYSTE M The motivational levels of Figure 10 .1 should be read horizontally from th e most general on the left to the most specific on the right . The sequential aspects of motivational substates and behavior should be read vertically from top to bottom . Down is closer to food ; up is farther away. Out of the picture to the left i s the system level, which refers to common aspects of behavioral expression related to a particular function, feeding in this case . The subsystem level reflects

SUBSYSTEM

MODE

MODULE

ACTIO N

Travel

Locomot e Sca n

Capture Test Ingest

Reject Hoard

~

Crawl ove r Sniff Nose Paw Track Cut off Immobile Pounc e Grab Bite Gnaw Hold Che w Swallo w Spit out — Wipe off Carr y

FIGURE 1 O . 1 . The structure of a behavior system includes four levels : system, subsystem, motivational mode, and perceptual-motor modules . This figure focuses on the motivational modes and perceptual-motor modules of the predatory subsystem in the feeding system of the rat. The actions potentially controlled by the system components are shown at the far right .


26 5

the next lower level, the overall strategies for finding particular types of food in this case, live prey . The perceptual-motor modules (the boxes on the right) represent the perceptual motor organization present, while the motivational modes to their left capture the sequential aspects of behavior in the form of different sets of stimulu s sensitivities and motor repertoire as a function of proximity to food . Hungry animals typically begin in general search mode with attention to cues that contro l behavior related to locomotor search general inspection of the environmen t combined with increased sensitivity to cues that predict more proximal food . I n the presence of cues more specifically predicting the spatial/temporal proximit y of food, the focal search mode becomes stronger . Finally, in the presence of food, the handling-consuming mode is expressed (Timberlake & Lucas, 1989 ; Timberlake & Silva, 1995), followed by return to a form of focal search . The sequence of motivational modes captures the overall goal-seeking aspects of behavior by focusing on the conditions of transition between sets o f stimulus sensitivities and related response repertoires . Mode transitions occur as a function of receiving stimuli related to the next mode . Thus, goal-seeking behavior does not necessarily condition a particular sequence of responses or stimuli, or even a fixed end point, although all of those aspects of goal-directed behavior can occur with appropriate stimulus support and repetition . Further, th e sequence of modes places no requirements on how stereotyped or variable particular modules may be . There is a tendency for responding to decrease in variability and stimulus filtering to increase in specificity as one goes from general search to handling-consuming . However, there are stereotyped searching behaviors as well as variable consummatory responses . Again this is a product of th e specifics of an animal, not the conceptual dictates of the system . An advantage of representing the motivational hierarchy from left to right rather than top to bottom is that the sequence of actual behavior (reading top to bottom) is separated from the regulatory motivational aspects (reading horizon tally) . In other words, the model does not require that behavior must be per formed in a particular order because it is both motivated and released by performance of a previous behavior. Instead, motivation is present in the lower level s to the right when it is present at a higher level on the left or when stimuli directly affect lower states (see Timberlake & Silva, 1995) . The sequence of actua l behavior is determined by the interaction of environmental and animal-base d stimuli with the perceptual-motor structures on the right . Whether a particular behavior can occur out of sequence or not is a function of the specific system under consideration and the particular stimulus support present . It is not a functio n of the general framework . The "real world" lies to the right in this diagram . Stimuli enter the system as input to specific perceptual-motor modules and, in a more general way, as input to particular modes and systems . In the laboratory the stimulus typically has a particular relation to a response or a particular timing with respect to presentation of an unconditioned stimulus . Learning is assumed to occur in many forms,

266

TIMBERLAKE

SYSTEM

A v J Q Z_

SUBSYSTE M

Z O cc cn > z w

Y

A

w .J MODE

A V

POSITIV E FEEDBACK

MODUL E ACTIO N PATTERN S

[

Handbook of Behaviorism

Behavior and Personality: Psychological Behaviorism

B.F. Skinner and behaviorism in American culture

Understanding Behaviorism: Behavior, Culture, and Evolution

Radical Behaviorism: The Philosophy and the Science

Toward A New Behaviorism: The Case Against Perceptual Reductionism

Oxford American Handbook of Urology (Oxford Handbook)

Handbook of Psychoeducational Assessment: A Practical Handbook

Oxford American Handbook of Urology (Oxford Handbook)

Handbook of Thermoplastic Elastomers (Pdl Handbook)

The War Between Mentalism and Behaviorism: On the Accessibility of Mental Processes (Scientific Psychology Series)

Handbook of Turbomachinery

Handbook of Electronic Weighing

Lange's Handbook of Chemistry

CRC Handbook of Thermoelectrics

Handbook of Inorganic Chemicals

The handbook of integration

Handbook of photochemistry

Handbook of spatial logics

Handbook of constraint programming

Handbook of Corynebacterium glutamicum

Handbook of algebra

The Handbook of Dramatherapy

Springer Handbook of NanoTechnology

Handbook of ornament

Practical Handbook of Photovoltaics

Lange's Handbook of Chemistry

Handbook of VLSI Microlithography

Handbook of Inorganic Chemicals

Handbook of Behaviorism

Behavior and Personality: Psychological Behaviorism

B.F. Skinner and behaviorism in American culture

Understanding Behaviorism: Behavior, Culture, and Evolution

Radical Behaviorism: The Philosophy and the Science

Toward A New Behaviorism: The Case Against Perceptual Reductionism

Oxford American Handbook of Urology (Oxford Handbook)

Handbook of Psychoeducational Assessment: A Practical Handbook

Oxford American Handbook of Urology (Oxford Handbook)

Handbook of Thermoplastic Elastomers (Pdl Handbook)

The War Between Mentalism and Behaviorism: On the Accessibility of Mental Processes (Scientific Psychology Series)

Handbook of Turbomachinery

Handbook of Electronic Weighing

Lange's Handbook of Chemistry

CRC Handbook of Thermoelectrics

Handbook of Inorganic Chemicals

The handbook of integration

Handbook of photochemistry

Handbook of spatial logics

Handbook of constraint programming

Handbook of Corynebacterium glutamicum

Handbook of algebra

The Handbook of Dramatherapy

Springer Handbook of NanoTechnology

Handbook of ornament

Practical Handbook of Photovoltaics

Lange's Handbook of Chemistry

Handbook of VLSI Microlithography

Handbook of Inorganic Chemicals

Recommend Documents