Lay Epistemics and

Human Knowledge
Cognitive and
Motivational Bases
PERSPECTIVES IN SOCIAL PSYCHOLOGY

ASeries o[ Texts and Monographs • Edited by Elliot Aronson

Recent volumes in this series:

THE DILEMMA OF DIFFERENCE: A Multidisciplinary View of Stigma
Edited by Stephen C. Ainlay, Gaylene Becker, and Lerita M. Coleman

ELIMINATING RACISM: Profiles in Controversy

Edited by Phyllis A. Katz and Dalmas A. Taylor

HUMAN AGGRESSION
Robert A. Baron

INTRINSIC MOTIVATION AND SELF-DETERMINATION

IN HUMAN BEHAVIOR
Edward L. Deci and Richard M. Ryan

LAY EPISTEMICS AND HUMAN KNOWLEDGE:

Cognitive and Motivational Bases
Arie W. Kruglanski

NONVERBAL BEHAVIOR AND SOCIAL PSYCHOLOGY

Richard Heslin and Miles Patterson

THE PHYSICAL A TTRACTIVENESS PHENOMENA

Gordon L. Patzer

REDEFINING SOCIAL PROBLEMS

Edited by Edward Seidman and Julian Rappaport
SELF-DISCLOSURE: Theory, Research, and Therapy
Edited by Valerian J. Derlega and John H. Berg

SCHOOL DESEGREGATION
Harold B. Gerard and Norman Miller

SCHOOL DESEGREGATION: Past, Present, and Future

Edited by Walter G. Stephan and Joe R. Feagin

UNIQUENESS: The Human Pursuit of Difference

C. R. Snyder and Howard L. Fromkin

A Continuation Order Plan is available for this series. A continuation order will bring delivery
of each new volume immediately upon publication. Volumes are billed only upon actual ship-
ment. For further information please contad the publisher.
Lay Epistemics and
Human Knowledge
Cognitive and
Motivational Bases
Arie w. Kruglanski
University of Maryland
College Park, Maryland

Springer Science+Business Media, LLC

 Library of Congress Cataloging in Publication Data

Kruglanski, Arie W.
Lay epistemics and human knowledge.

(Perspectives in sodal psychology)

Bibliography: p.
Includes index.
1. Sodal perception. 2. Knowledge, Sodology of. 3. Epistemics. 4. Attribution
(Sodal psychology) I. Title. IL Series.
BF323.S63K78 1989 302'.12 89·16057

ISBN 978-1-4899-0926-8 ISBN 978-1-4899-0924-4 (eBook)

DOI 10.1007/978-1-4899-0924-4

© 1989 Springer Science+Business Media New York

Originally published by Plenum Press, New York in 1989.
Softcover reprint of the hardcover 1st edition 1989

All rights reserved

No part of this book may be reproduced, stored in a retrieval system, or transmitted

in any form or by any means, electronic, mechanical, photocopying, microfilming,
recording, or otherwise, without written permission from the Publisher
 Ta Hannah,
my most canstructive critic
Foreword

Whatever your reasons, kind reader, for reading these words,-what-

ever your premises about forewords, whatever the epistemic motivation
with which you approach them-Iet me urge you to turn immediately to
Kruglanski's first chapter and skim it. If any enthusiasm for sodal psy-
chology flows in your veins, you will certainly proceed then to read
further in this important book.
It represents some dozen years of Arie's thought and of his and his
colleagues' research. Its intellectual scope covers 50 years of sodal
psychology-from attitudes and attitude change, to balance, disso-
nance, and the various other cognitive consistency theories, to causal
attribution, and to current cognitive sodal psychology. Sodal psycholo-
gists have recently begun to leave the fireside coziness of scribbling
textbook catalogues of our field and to venture out into the cold, outdoor
adventure of detecting (or creating?) its underlying structure. Of these
attempts at providing scope plus order, Kruglanski's must surely be the
most ambitious. For his is no mere overarching theory, which, like a
circus tent over a diverse set of sideshows, covers everything but does
little to provide thematic structure. Rather, Kruglanski tries to produce a
basic reorganization of our thinking about sodal psychology. To use his
LEGO blocks metaphor for the modification of knowledge structures, he
attempts to dismantle the current assembly of elements of our field and
reassemble them into a simpler and more coherent configuration.
The coherence is produced by a theory of "lay epistemics"-a theo-
ry about "the formation of all knowledge." This theory grew out of
Kruglanski's early work on causal attribution, when he became dissatis-
fied with the prevailing distinctions among the categories of causes and
between cognitive and motivational processes. As he generalized his
perspective, he was able to locate attribution as a spedal ca se within a
broader "epistemic" framework. This applies to all knowledge, includ-
ing that about noncausal properties and categories, and analyzes moti-

vii
viü FOREWORD

vation in terms directly relevant to the course of the knowing process.

He was able then to apply this epistemic framework to the panoply of
areas mentioned above, producing the sweeping reinterpretation of old
concepts and results presented in this book.
One consequence of this effort, one that I especially value, is that it
reveals the cumulative nature of work in our field. When viewed from
Kruglanski's perspective, it seems that we have not been going around
in circles or following trendy "fads." Rather, as it now seems, we have
simply shifted our perspective and emphasis-often with confusing
changes in jargon and methods-but always with an implicit focus on
the same core phenomena of how people form, test, and act upon their
knowledge.
Be prepared, kind reader, to leam a new set of concepts. Theyare
not difficult and they are located in an intuitively appealing model of the
acquisition and validation of knowledge. And be prepared to leam to
think about cognitive processes, motivational forces, and informational
content-all "in the same breath," so to speak. Kruglanski is clearly
located within social cognitive psychology, but (as this book's title im-
plies) his epistemic theory attaches no more importance to the logi-
callinformational facets of the knowing process than to the motives that
trigger, direct, and terminate it. And above all, be prepared to cast old
problems in new terms. As in any good new theory, in its LEGO-blocks-
like reconfiguration of our field, things we presently have stuck together
are pulled apart (e.g., imbalance may create discomfort through the
logical contradiction it implies, but stronger negative affect may or may
not occur, depending on the relevant epistemic motives); presently con-
joined things are put together in different ways (e.g., the differences
among the consistency theories are ones of content, not process), and
elements we have kept separate are now linked together (e.g., source
and message effects in attitude change are shown to follow the same
basic principles). A recurrent theme is that many instances of what we
have regarded as strong regularities have exceptions. (Perhaps we have
too often confused "statistical" generalizations with "laws" by failing to
recognize the exceptions.)
Beyond revealing the inconsistencies and contingencies in the
knowing process, Kruglanski's theory provides a reconciliation among
them. This is achieved by raising the level of our analysis by several
degrees of abstractness. Our categories of analysis are raised from the
many specific conditions, motives, and mechanisms that have character-
ized research in such areas as attribution, social comparison, and atti-
tude change to the more abstract level of a few basic types of epistemic
needs, the contents of information and their accessibility, and-the in-
variant aspect of the system-the logical nature of knowledge forma-
tion, restructuring, and use.
FOREWORD ix

A restructuring of this magnitude will surely meet resistance and

raise objections. I myself confess to a discomfort with a view of our
knowledge structures in which, because they are Httle constrained by
the external world, in Arie's words, "nearly anything goes." On the
other hand, I find the reinterpretation of attribution research to be ex-
ceedingly useful, and I expect that you, kind reader, will pick up on
some of the many impHcations for new lines of work in some of these
well-tilled but still fertile sections of our field.

Harold H. Kelley
Department of Psychology
University of California at Los Angeles
Los Angeles, California
Preface

Writing apreface for a book is arguably the most delightful moment of

the entire enterprise. First, it signals that one is dose to the "light at the
end of a tunnel." In the case of the present volume, the "tunnel" has
been long indeed. Second, it provides an opportunity to thank all those
friends and colleagues who selflessly took the time to read) comment,
and generally help improve various portions of the manuscript. In the
case of this volume, the help has been considerable.
The idea for the present work germinated sometime in the summer
of 1980, and the plan at that time was to have the entire volume com-
pleted within two years at the outmost. What prevented this seemingly
reasonable plan from materializing was another "scheme," notably, to
take advantage of opportunities to write invited artides and chapters as
"sketches" for sections of the present work. In retrospect, this strategy
turned out to be a mixed blessing. On the positive side, the time and
effort put into those extraneous ventures hopefully have paid off as far
as the quality of the final product is concemed. I believe I would not
have been able to cover the various topics this book addresses quite as
thoroughly, had I written it in an uninterrupted "sweep." The cost,
however, was the considerable length of time it took to complete the
project. The sketches often bore only a remote resemblance to the final
chapters, and the reworking involved was not nearly as straightforward
as I had hoped.
The subject matter of this volume is twofold, combining as it does
the topics of human knowledge and social cognitive psychology. The present
volume introduces an epistemic perspective to various sodal psycholog-
ical issues, based on the assumption that much of our sodal conduct,
affect, or cognition is related to what we know (or think we know) about
pertinent matters. In asense, the thesis of this book is that sodal cogni-
tive psychologists may have been "talking knowledge" all along without
quite realizing it. At the same time, making explidt the knowledge per-

xi
xii PREFACE

spective promises to cast new light on diverse social psychological topics

and offers suggestions for numerous new lines of future research.
It is a pleasant task indeed to thank all those many persons whose
involvement at one stage or another contributed to the evolution of this
work. Thus, I remember with affection and gratitude the contributions
of my former graduate students, Muli EIlis, Tali Freund, Chilik Klar, and
Ofra Mayseless, whose dedication, talents, and enthusiasm fueled the
"lay-epistemic" enterprise and kept it locomoting toward completion.
Early work on the book was greatly facilitated by a research pro-
fessorship at the Center for Decision Making (Sonderforschungbereich
24) at the University of Mannheim, West Germany, and I am grateful to
Martin Irle for extending me the invitation and for his friendship and
hospitality.
Chapters in this book benefited from a elose reading by numerous
colleagues whose insights, criticism, and good sense were of great help
in preparing the present version. I am particularly indebted to Icek
Ajzen, Reuben Baron, Daniel Bar-Tal, Lee Cronbach, Russel Fazio, Ken-
neth Gergen, John Harvey, Tory Higgins, Dave Jamieson, Harold Sigall,
Wolfgang and Margaret Stroebe, Shelagh Towson, and Mark Zanna for
reading and making useful comments on separate chapters. In addition,
Chapter 10, on lay and scientific epistemics, benefited considerably from
several discussions and friendly debates with Donald Campbell. Last,
but of utmost importance, Judson Mills read and made highly valuable
comments on the entire manuscript. Although I made extensive use of
many of the suggestions I received, it is more than a mere formality to
state that the final judgment and responsibility for the volume's content
are my own.
I also wish to extend special thanks to EIliot Aronson, the series
editor, for his initial encouragement and continued support. Eliot
Werner of Plenum Press was a paragon of patience, and his surprise on
the project's final completion was a pie asant one. My heartfelt gratitude
goes to EIlie Daino, our executive secretary at the psychology depart-
ment at the University of Maryland, and to Wilma Holdway for their
considerable help in managing the endless typing jobs associated with
various versions of the manuscript. Their cheerful goodwill and efficien-
cy during the final stages of this project will not be forgotten.
Finally, my wife Hannah has my greatest thanks. She not only
provided unwavering support at moments of doubt, but also coura-
geously told me what she thought when I asked for her opinion.
Contents

1. The Lay Epistemic Framework: Its History and Scope ......... 1

The Endogenous-Exogenous Partition and the

Process-Contents Distinction .................. 2
Logical Consistency and Inconsistency .............. 3
Motivational Bases ................................ 4
Content Domains of Application: Attitude Formation
and Change .................................. 4
Sodal Comparison and Minority Influence .......... 5
The Issue of Accuracy ............................. 5
Applications to Cognitive Therapy . . . . . . . . . . . . . . . . . . 6
Sdentific and Lay Modes of Knowledge Acquisition .. 6

2. Knowing All: A Theory of Lay Epistemics . . . . . . . . . . . . . . . . . . . 9

The Psychological Properties of Knowledge . . . . . . . . . . 9

The LEGO Blocks and the Searchlight Metaphors .... 10
The "Searchlight" Metaphor ....................... 11
The Process of Knowledge Acquisition .............. 13
Hypothesis Generation .......................... 13
The Process of Hypothesis Validation: Person as a
Natural Logidan .............................. 22
Encoding of Stimulus Information as Hypothesis
Validation. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Functions of Knowledge ........................... 30
Summary........................................ 32

xiii
xiv CONTENTS

3. Empirical Research in the Lay Epistemic Framework .......... 35

Motivational Effects in Epistemic Behavior . . . . . . . . . . . 35

The Context of Knowledge Formation . . . . . . . . . . . . . 35
The Context of Knowledge Utilization ............ 42
The Context of Social Behavior ................... 45
The Logical Element in Epistemic Behaviors ......... 48
Cognitive Balance Effects ........................ 48
Inconsistency Resolution via Informational Search . . 49
Hypothesis Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
Content-Bound Research in the Lay Epistemic
Framework ................................... 52
Self-Ascribed Epistemic Authority ................ 53
Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 55

4. Unique and Nonunique Aspects o[ Attributions .............. 57

The Logical Aspect of Attribution . . . . . . . . . . . . . . .. . . . 57

Kelley's ANOVA Model......................... 58
Cognitive Capability Notions: Availability,
Accessibility, and Saliency ..................... 66
Motivational Bases of Attributions .................. 68
Epistemic Motivations Revisited: Needs for
Nonspecific and Specific Closure and for the
Avoidance of Closure ......................... 69
Implications of the Epistemic Reanalysis of
Motivational Effects in Attribution .............. 76
Concluding Comments: Unique and Nonunique
Aspects of Attributions ........................ 78
Implications for Further Research . . . . . . . . . . . . . . . . . 81
Summary........................................ 82

5. A Bridge to Consistency Theories .......................... 85

Deductive, Motivational, and Content-Related

Elements of Epistemic Behavior ................ 86
The Logical Aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
Festinger' s Dissonance Theory ................... 87
Heider' s Balance Theory ......................... 89
The Congruity Model ........................... 92
CONTENTS xv

Rosenberg' s Affective-Cognitive Consistency

Theory ..................................... 93
Abelson and Rosenberg's Psycho-Logic ........... 93
Aronson' s Disconfirmation of Expectancy
Formulation ................................ 95
Irle' s Revised Dissonance Theory ................. 95
McGuire's Consistency Theory ................... 95
Recapitulation .................................. 95
The Mode of Resolution Problem ......... . . . . . . . . 95
The Deductive Element in Cognitive Consistency
Theories: A Summing Up .. . . . . . . . . . . . . . . . . . . 99
The Content Aspect of Cognitive Consistency
Theories ..................................... 100
Content-Specific Definitions of Inconsistency ...... 100
Mapping Out Attempts . . . . . . . . . . . . . . . . . . . . . . . . . . 101
The Motivational Aspect . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
Inconsistency as a Universal Drive . . . . . . . . . . . . . . . . 102
Inconsistency' s Motivational Properties as
Derived from Other Needs .... ... . .. .... .... . 103
Different Types of Inconsistency . . . . . . . . . . . . . . . . . . 103
When Inconsistency Leads to Affect: A Lay
Epistemic Interpretation ..................... 104
Lay Epistemic Interpretation of the Consistency
Models: A Recapitulation .................... 105
Bridging the Consistency and Attribution
Paradigms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
Research Implications of the Present Work ........ 107
Summary ........................................ 109

6. Attitudes as Knowledge Structures . . . . . . . . . . . . . . . . . . . . . . . . . 111

The Definition of an Attitude ...................... 111

The Derivation of Attitudes ...................... 113
Attentional Focus Factors in Attitude Research . . . . . 120
Epistemic Motivations and Attitude Change ....... 123
Unitary or Disjunctive Processes of Attitude
Formation and Change ...................... 129
Central versus Peripheral Routes to Attitude
Change: A Unified Perspective ............... 131
An Epistemology of Actions and the Attitude-
Behavior Relation ........................... 134
Summary and Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . 139
xvi CONTENTS

7. Further Domains 01 Application: Social Comparison

Processes and Minority-Influence Phenomena ................ 143

Sodal Comparison Processes ....................... 143

The Two Senses of Sodal Comparison ............ 144
Sodal Comparison as the Content of a Question ... 146
Sodal Comparison as Use of Sodal Sources of
Information ................................ 148
The Motive Base of Sodal Comparison
Phenomena ................................ 154
Domain Differentiation of Sodal Comparison
Phenomena: The Content-Process
Distinction ................................. 155
Sodal Comparison Phenomena: The Epistemic
Perspective ................................... 158
Proposed Reorientations ......................... 159
Minority Influence ................................ 161
Popular "Reputations" of Minorities and
Majorities .................................. 161
Categories of Factors Affecting Judgment Formation
and Change. .... .... . ..... .... .... .... .... . 162
Conformity versus Independence in Minority and
Majority Contexts ........................... 165
Compliance versus Internalization in the Majority
and Minority Paradigms ..................... 166
Conclusion ..................................... 168
Summary ........................................ 168

8. The Issue 01 Accuracy in Social Perception and Cognition ...... 171

Themes in Recent Accuracy Research. .... . ... . ... .. 172

Defining Accuracy in Sodal Perception and
Cognition .................................. 172
Accuracy versus Accurades ...................... 176
A Process-Based Analysis of Situational Accuracy .... 187
The Accuracy Framework .... . . . . . . . . . . . . . . . . . . . . 187
The Realist Paradigm .. . . . . . . . . . . . . . . . . . . . . . . . . . . 188
Phenomenal Accuracy ........................... 192
Summary and Conclusions .. . .... . .... .... .... .... . 194
CONTENTS xvü

9. Knowing How to Cure: Implications for Cognitive Therapy 197

The Contents of Dysfunctional Beliefs:

The Frustrative Hypothesis .................... 197
Modifying the Frustrative Hypothesis:
Assessment Shifts, Goal Shifts, and
Means Shifts ............................... 198
Applying Lay Epistemic Principles to the Practice of
Cognitive Therapy ............................ 200
Knowledge Structures and Their Validity . . . . . . . . . . 200
Epistemic Authority ............................. 201
Freezing and Unfreezing Mechanisms .. .... ....... 205
Contents of Therapeutic Change ................. 208
Lay Epistemics and Cognitive Therapies:
A Comparative Appraisal ...................... 209
The Dysfunctional Misconception (Distortion)
Hypothesis ................................. 210
Are Depressives Wiser? . . . . . . . . . . . . . . . . . . . . . . . . . . 213
Contents of Irrational Beliefs ..................... 216
Primacy of Personal Experience as the Mediator of
Cognitive Change..... . .... ...... ...... .... . 217
Implications of the Lay Epistemic Analysis for the
Practice of Cognitive Therapy ................ 219
Summary ........................................ 221

10. The Social Psychology of Science: On the Lay Epistemic

Underpinnings of Research Methodology .... . . . . . . . . . . . . . . . . 223

Introduction ...................................... 223

Official and Unofficial Research Methodologies ...... 224
Official Methodological Concepts ................. 225
Beyond Official Methodologies: Science as a Social
Cognitive Enterprise ........................ 235
Summary ........................................ 245

References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 247

Author Index .............................................. 269

Subject Index 275

 1
The Lay Epistemic Framework
Its History and Scope

Several years ago I began to formulate a set of ideas that promised a

fresh perspective on the diverse body of causal attribution models pro-
posed in recent years by social psychologists. The fresh perspective was
contained in my theory of lay epistemics, and the attributional models
were assumed to represent special cases of that particular framework.
The theory of lay epistemics addresses the process whereby people ac-
quire knowledge about themselves and the surrounding world. It de-
picts the invariant principles that govern the formation and revision of
beliefs in various domains of life as disparate in content as science,
religion, or politics.
Lay epistemic analyses of various attributional topics have appeared
in several papers published in recent years by myself and my colleagues
(e.g., Kruglanski, 1980; Kruglanski and Ajzen, 1983; Kruglanski, Hamei,
Maides, and Schwartz, 1978; Kruglanski, Schwartz, Maides, and Harne!,
1978). But as this work progressed, I realized that the relevance of the
lay epistemic paradigm transcends the domain of causal attribution. The
attributional domain is restricted to the knowledge of causal relations,
whereas the epistemic paradigm applies to an knowledge, including
knowledge of noncausal relations and categories (Bar-Tal and Bar-Tal,
1988). Because of its breadth, the lay epistemic framework affords a new
way of looking at numerous classical topics in social cognition, such as
social perception, cognitive consistency models, social comparison pro-
cesses, attitude formation and change, and cognitive aspects of social
influence. The lay epistemic framework promises to synthesize an those
heretofore separate topics under a common set of principles.
The main facets of the present epistemic theory include (1) the
content element that refers to an kinds of propositions and hypotheses
that persons might entertain, (2) the logical element whereby beliefs are

1
2 CHAPTER 1

proven or disproven on the basis of relevant evidence, and (3) the moti-
vational element that sets the process in motion, terminates it at a given
point, and provides the base for the individual' s cognitive and affective
reactions to validating and/or invalidating information. A full descrip-
tion of the lay epistemic theory is given in Chapter 2 and the main lines
of empirical evidence for the theory are described in Chapter 3.

THE ENDOGENOUS-EXOGENOUS PARTITION AND THE

PROCESS-CONTENTS DISTINCTION

My first, quite unintended, step toward the lay epistemic frame-

work was taken in a 1975 Psychological Review article that argued in favor
of replacing in some cases the distinction between internal and external
causes of human action by a distinction between endogenous and ex-
ogenous causes. An endogenously attributed action denotes the case in
which the action is regarded as an end in itself, and an exogenously
attributed action, one in which the action is considered a means to some
further end. Prior to my endogenous-exogenous distinction, the inter-
nal-external distinction was regarded as the fundamental dimension on
which persons were assumed to order their perceived causes for various
events.
In fact, the internal-external distinction was assumed to be so basic
that it was commonly considered a central feature of attribution theory,
a framework presumably dealing with the way people generally go
about ascribing causes to effects. But after it became apparent that occa-
sionally persons might be interested in another causal distinction, nota-
bly that between endogenous and exogenous causes, I began to suspect
that the internal-external partition may not be as exclusive as had been
implied thus far.
This reaHzation posed for me something of a dilemma as to a fruitful
research strategy to pursue from then on. A natural possibility wOl.j.ld
have been to try and make an extensive empirical case for the notion that
the endogenous-exogenous distinction should occupy a central place in
attribution theory, alongside the internal-external distinction. What
stopped me from pursuing this tack was the following haunting
thought: In the same way that I identified the endogenous-exogenous
distinction, it should be possible to enumerate further and further causal
distinctions in which lay attributors may occasionally have an interest.
There seemed no compelling reason to believe that across the wide
variety of effects that persons might wish to explain, and the wide
variety of cultural, historical, or situational circumstances people might
experience, they would pose the very same causal questions and be
LA Y EPISTEMIC FRAMEWORK 3

concerned with the very same causal categories. Such doubts suggested
the disturbing possibility that the quest for universally basic causal di-
mensions might be doomed apriori for failure.
The alternative approach which I chose in the end was to distin-
guish between the process of causal attribution and the contents of causal
categories. The process of causal attribution refers to the set of invariant
mIes whereby all causes are inferred. It seemed to me that attribution
theory, if it is to be a general theory of human psychology, ought to
concern itself with process rather than contents. The latter, although of
considerable interest and predictive utility, is necessarily spedfic to
given persons in given situations and at given times. But what might the
attributional process consist of? A elose analysis of extant attributional
models quickly diselosed an important process element. It was embed-
ded in the notion of "attributional criteria" (Kelley, 1967), assumed to
represent patterns of information on which basis confident causal as-
criptions may be reached. On careful examination, it turned out that
once they are "stripped" of the contents elements, the attributional crite-
ria reflect the prindple of "logicai" derivability or consistency. An at-
tribution starts with an "if then" premise whereby a given pattern of
data is judged to be indicative of a given causal category. When this
particular data pattern is then encountered, the causal category is as-
signed. A detailed analysis of the attribution process in general and the
role of logical consistency in particular is undertaken in Chapter 4.

LOGICAL CONSISTENCY AND INCONSISTENCY

The notion of logical consistency (or derivability) that seems to play

a central (if implidt) role in attribution theory brings back memories of
another day and age, that of the cognitive consistency theories (like
dissonance or balance) that virtually dominated sodal psychological re-
search in the 1960s. Could logical consistency be involved there as weH?
It turns out that it could be and was. Diselaimers that cognitive con-
sistency theories pertain to "psychological" rather than "logical" con-
sistency may rest on a falladous distinction. In every case of psychologi-
cal inconsistency there seems to exist a core of logical inconsistency.
What may be psychological or "subjective" are the contents of persons'
beliefs and the reasons those individuals subscribe to those particular
beliefs. On the other hand, the notions of logical consistency or inconsis-
tency pertain to the formal relations among beliefs rather than to their
content. In this sense, any experience of an inconsistency contains both
psychological and logical aspects.
4 CHAPTER 1

MOTIVATIONAL BASES

More important perhaps, the epistemic effect of inconsistency is op-

posite to that of consistency. Consistency enhances the individual's con-
fidence in his or her knowledge (e.g., his or her causal attribution)
whereas inconsistency undermines the individual's confidence. Could
this epistemic function of inconsistency be at the root of the affective and
cognitive phenomena of concern to cognitive consistency theories? Ap-
parently so, but only under some conditions. Inconsistency-based un-
dermining of confidence may give rise to negative affect only when the
individual desired confident knowledge to begin with. Furthermore, it is
only under those conditions that the individual may be expected to
engage in extensive cognitive work (e.g., attitude change) aimed at the
restoration of consistency. In short, we must posit an epistemically rele-
vant motive base for an inconsistency-based unpleasantness to be expe-
rienced and to motivate cognitive change. These issues are discussed in
Chapter 5.
From the present perspective, attributional and cognitive consisten-
cy models appear to address the same epistemic phenomena, albeit from
different angles. Attribution theories look at the effects of cognitive con-
sistency (fulfillment of the attributional criteria) on subjective confidence,
whereas cognitive consistency theories look at the effects of inconsisten-
cy on affective states and subsequent cognitive change. Such differences
in emphasis notwithstanding, both types of theory refer to separate
aspects of the single process whereby individuals acquire, modify, and
test their knowledge. It is in that sense that both types of theory are
special cases of the lay epistemic model.

CONTENT DOMAINS OF ApPLICATION:

ATTITUDE FORMATION AND CHANGE

The general knowledge-acquisition process outlined in this volume

is assumed to apply across different possible contents of knowledge.
Two chapters of the present volume, 6 and 7, illustrate the heuristic
value of considering, from the lay epistemic perspective, classical social
psychological topics that revolve around unique conte nt categories of
knowledge. Chapter 6 considers attitudes as knowledge structures
which content is evaluative. From this vantage point, attitudes are a
subclass of beliefs or judgments; hence, the general epistemic process
that governs the formation and change of all judgments also applies to
the formation and change of attitudes. This analysis implies a unitary
attitudinal process that may serve to integrate previous distinctions,
such as those between "central" and "peripheral" (cf. Petty and Caciop-
LAY EPISTEMIC FRAMEWORK 5

po, 1981) or uheuristic" versus usystematic" (cf. Chaiken and Stangor,

1987) routes to attitude change. The epistemic approach to attitude
change has novel implications for several central problems in the atti-
tudes domain, for example, conditions leading to attitude polarization,
and those under which the attitude-behavior relation may or may not
hold.

SOCIAL COMPARISON AND MINORITY INFLuENcE

Social comparison processes (cf. Festinger, 1954; Goethals, 1987)

pertain to comparative knowledge that persons form about their various
attributes. Therefore, a theory about the formation of all knowledge
should be compatible with a conception of social comparison processes.
Surprisingly, however, consideration of the social comparison theory
from the present epistemic perspective suggests several major incom-
patibilities. Contrary to previous notions, it will be concluded that (1)
there seems to exist no general tendency or drive toward social com-
parison; (2) the similarity to oneself of the other being compared seems
less critical than has been assumed, that is, the comparative question of
interest may or may not refer to similar others, nor is the similarity
dimension of general significance in choosing one' s source of com-
parison information; and (3) the same process of social comparison
seems applicable across different content domains of comparison. Argu-
ments relevant to those conclusions are developed in Chapter 7.
Chapter 7 also explores the possibility of a lay epistemic synthesis of
minority and majority types of influence. Findings in the literature on
minority influence (e.g., Moscovici, 1985; Nemeth, 1986) are classified in
terms of underlying epistemic variables. Accordingly, some effects (e.g.,
behavioral consistency of minority members) are considered information-
al, whereas others are classified as cognitive or motivational. The epistemic
analysis is also employed to identify conditions under which opinion
discrepancy between individuals is resolved via passive conformity to
one of the opposing views or via active and independent assessment of
the positions involved. Finally, the epistemic model affords a new look
at the phenomena of Uinternalization" and ucompliance" (cf. Keiman,
1958) in minority and majority contexts.

THE IssuE OF ACCURACY

The present epistemic framework has implications concerning the

issue of accuracy in social perception and cognition. Following an inte-
grative review of recent social psychological analyses of this topic, it will
6 CHAPTER 1

be suggested that general questions concerning accuracy cannot be

meaningfully answered; for instance, whether persons are generally ac-
curate or inaccurate, what the boundary conditions for accuracy are, and
the process (or "method") whereby accuracy may be improved. Instead,
a situationally specific approach to accuracy is outlined. Accuracy as-
sessment is depicted as a metajudgment comprising three subcompo-
nents: the judgment, the situational criterion, and the correspondence
between them. It is further assumed that each of those subcomponents
is itself a judgment governed by all the cognitive and motivational fac-
tors that affect the epistemic process in general. These notions are dis-
cussed in Chapter 8.

APPLICATIONS TO COGNITIVE THERAPY

Whereas major contemporary schools of cognitive therapy assume

that dysfunctional thought is characterized by remediable biases and
distortions in clients' inference processes, the present analysis suggests
that biases (e.g., of a motivational or a cognitive nature) are an insepa-
rable aspect of all inference process, "normal" as weH as "neurotic." It is
proposed that what sets adaptive and maladaptive thoughts apart is
their contents rather than the process whereby they have been reached.
Maladaptive or dysfunctional thoughts' contents are those with suffer-
ing-inducing properties and are assumed to derive from some type of a
"frustrative hypothesis," that is, the notion that the individual's impor-
tant goal has been or will be thwarted. Our conception of the epistemic
process has implications for "unfreezing" the maladaptive thought pat-
terns and "refreezing" in their stead alternative, more adaptive notions.
These topics are addressed in Chapter 9.

SCIENTIFIC AND LAY MODES OF KNOWLEDGE ACQUISITION

The theory of lay epistemics stresses the similarity of the process

whereby lay and scientific inferences are reached. This is not necessarily
meant to imply a lay person's essential accuracy or "rationality." In a
sense, our approach portrays the acquisition of scientific knowledge as
subject to similar cognitive and motivational biases as the acquisition of
knowledge by lay persons. This portrayal is hardly unique to the present
approach. Rather, it characterizes major contemporary trends in the
philosophy and sociology of science (cf. Popper, 1972; Weimer, 1979)
that eschew efforts at demarcating science from nonscience and stress
the essential continuity of scientific inference and common sense.
LAY EPISTEMIC FRAMEWORK 7

The present comparison of lay and scientific judgment has several

implications for research methodology. Those are considered in Chapter
10, in particular with reference to prevalent methodological precepts in
the behavioral sciences. It is argued that the notion of "validity threats"
(CampbeH and Stanley, 1963) and the related idea that research designs
can be apriori ordered as more or less efficient are of a relative rather
than an absolute import. Threats are relative to a given state of knowl-
edge in a domain of inquiry. Furthermore, one research design may be
considered superior to another only if it controls better for threats that
are currently plausible. As the plausibility of threats in a domain of
science may change over time, the value of research designs, too, may
change correspondingly.
The lay epistemic analysis of the scientific process further implies
the need to expand the notion of scientific methodology and to include it
in the motivational and cognitive factors known to affect the knowl-
edge-acquisition process at large. In order to get ideas across to a given
scientific audience, it may be weH to take into account the validity
threats of concem to members of that community and to be sensitive to
their unique needs and motivations.
Concluding our overview of issues dealt with in this volume, let us
now move on to a detailed exposition of these topics, starting with the
epistemic model itself.
 2
Knowing All
A Theory of Lay Epistemics

As we go about our daHy affairs, interacting with other people or attend-

ing to our tasks, there are a variety of things that we know. We might
know that it is Christmas Eve, that the concerto being played is by
Vivaldi, that we are late with our heating bill, or that two and two is
four. In the following pages, I outline a psychological theory that spec-
ifies how a1l such knowledge comes about. I call this theory a theory of
lay epistemics. The term lay is not meant to imply a break with "sophisti-
cated," "normative," or "scientific" epistemics. To the contrary, I feel
that the continuities between common sense and science significantly
outweigh the discontinuities. Accordingly, this book is devoted to the
general knowledge acquisition process that is manifest alike in activities
of scientists and laypersons. What I me an to suggest is that all epistemic
activity is in a sense lay; it shares the same process, is influenced by the
same factors, and, above a1l, suffers from the same frailties and foibles.
This is not to say that lay and scientific epistemics are fu1ly alike.
After a1l, even the epistemic processes (methods, procedures) employed
by the separate sciences differ considerably. Physicists may use the nu-
clear reactor, biologists the electronic microscope, and sociologists the
participant observer. Nor do I mean to imply that the differences be-
tween common sense and science are inconsequential or that science
should abrogate its centrality in human affairs. Still, throughout the
present work, the emphasis is on psychological properties that lay and
scientific modes of knowledge acquisition share in common. Let us see
what those commonalities may be.

THE PSYCHOLOGICAL PROPERTIES OF KNOWLEDGE

"Knowledge" is assumed to consist of propositions (or bodies of

interrelated propositions) in which a person has a given degree of confi-
9
10 CHAPTER2

dence. Such characterization suggests that the acquisition of knowledge

contains two aspects: (1) Propositional contents must be generated
somehow. This implies an aspect of hypothesis generation. (2) A given
degree of confidence must be bestowed on the hypotheses generated.
This implies an aspect of hypothesis validation. Let me note in passing that
different versions of the hypothesis-generation-validation model have
been used in reference to knowledge acquisition on levels of everyday
perception (Bruner, 1951, 1973; Gregory, 1970, 1973), concept forma-
tion (Levine, 1975), problem solving (Boume, Dominowski and Loftus,
1979) (Newell and Simon, 1972); and sdentific discovery (Popper, 1972).
This implies the generality of the epistemic process and its unity across
levels of analysis. But before dwelling more extensivelyon the process
of knowledge acquisition, it may be well to comment on the unstable
nature of our knowledge. This is attempted in the paragraphs to
follow.

THE LEGO BLOCKS AND THE SEARCHLIGHT METAPHORs

In thinking about the properties of human knowledge, I find it
convenient to use two metaphors, those of the LEGG blocks and of the
searchlight, respectively. The first metaphor applies to our general store
of knowledge available in "long-term memory" (Higgins, King, and
Mavin, 1982). The second applies to momentarily accessible knowledge in
"short term memory" (Higgins and King, 1981).
The LEGO metaphor connotes the considerable diversity in types of
things that we may "know." Our general repertory of knowledge is
likened to a set of temporarily stable configurations. Any bit of knowl-
edge ("Sara contracted the cold because she did not dress properly,"
"John received a raise in salary," "Ross affectionately caressed Laura") is
a composite of interlinked concepts (objects, properties, relations). And
just as the LEGO blocks may be combined to yield a manifold of differ-
ent configurations so may the concepts in our mind.
In keeping with the sodal cognitive tradition (Fiske and Taylor,
1984) I assume that our knowledge is actively constructed by ourselves
in the course of our interaction with others, and that the constraints set
on its various forms are subjective (having to do with things we regard as
facts) rather than objective. This means that in principle nearly "any-
thing goes," and hardly any conceptual configuration is prohibited a
priori. In someone's mind any concept can be perceived as relevant to
any other concept (the weather in Malta to the price of sugar in Cuba or
the gravitational field of Mars to Escher' s graphical ideas). Consequently,
the variety of things that we may know (i.e., confidently believe in) is
vast indeed.
A THEORY OF LAY EPISTEMICS 11

The putatively unbridled nature of conceptual configurations in

which persons may have faith implies the considerable relativity of hu-
man knowledge. Different cultures may subscribe to divergent belief
systems, and the potential heterogeneity in what different persons
"know for a fact" may be considerable. Furthermore, just like the LEGO
constructions, the conceptual ones, too, may be occasionally dissembled
and turned into alternative constructions. This betokens the tremendous
potential for change in our knowledge. Under the appropriate condi-
tions, many kinds of knowledge could be revised, modified, or aban-
doned. Thus, one could presumably change one's knowledge of basic
values (i.e., one's knowledge of good and bad), one's views of physics
(Kuhn, 1962), one's fundamentallikes and dislikes, or one's self-identi-
ty. Furthermore, to the extent that our personalities are largely com-
posed of preferences, values of ideologies, they, too, can change in
significant ways. A rationalist could turn into a mystic, a devout be-
liever into a skeptic, and a zealous Communist into an enthusiastic
capitalist.
Nor is the potential for change restricted to the individual knower.
It may extend to whole groups, communities, and cultures. Contempo-
rary philosophers and historians of science (Kuhn, 1962, 1970, or Feyera-
bend, 1976, for example) demonstrated how drastic shifts in world
views of entire scientific communities represented a major feature of
progress in science. According to these conceptions, science does not
advance by accumulating objective truths about the world. Regardless of
how popular or deeply venerated, today's truths or laws of nature could
be overthrown at some future date and replaced by alternative truths.
Even yesterday's "facts" can not be counted on in days to come. In an
important sense facts are as conceptual as hypotheses (Popper, 1959,
1972; Weimer, 1979); they, too, can be reinterpreted and reconcep-
tuaHzed in significant ways.
In short, the LEGO metaphor depicts our knowledge as relatively
unconstrained and dynamically evolving. It conveys a tremendous po-
tential for variability and change both between and within persons,
societies, or cultures. But the dynamic character of our knowledge ex-
tends to momentary shifts, as weH as the relatively enduring ones repre-
sented by our LEGO metaphor; those are depicted by our searchlight
metaphor.

THE "SEARCHLIGHT" METAPHOR

At any given moment, only a restricted portion of the knowledge

available in our memory is lit by the beam of cognizance; the rest stays
12 CHAPTER2

shrouded in the dark blanket of unawareness. Similarly, only a small

portion of momentarily available external information is focalized by our
mind. This idea is central to contemporary cognitive theory and is as-
signed a key role in conceptions of selective attention Games, 1890;
Kahneman, 1973; Shiffrin, 1975), limited information-handling capacity
(Newell and Simon, 1972), notions of saliency and vividness (Fiske and
Taylor, 1984), or of category accessibility (Higgins and King, 1981).
All the foregoing formulations imply the potential for considerable
personal inconsistency (or instability) due to shifts in the portion of
one's cognitive field currently at the center of attention. Thus, a person
can regard a given conclusion as warranted because a momentarily sa-
lient schema (i.e., a bit of knowledge or a nucleus of interrelated beliefs)
happens to strongly suggest it. But the same individual could be quite
unaware that a directly opposite conclusion is warranted by a different
schema generally available in his or her stable store of knowledge yet
currently inaccessible.
According to the present portrayal, the capacity for mental disjunc-
tion and personal inconsistency are, to a degree, part and parcel of
normal cognitive functioning rather than mental pathology. Because of
the constantly mobile flux of cognitive structures, it should be possible
to love and hate the very same person on different occasions, feel eu-
phorie or dejected about one's life, or be pleased or extremely displeased
by the same outcome (Higgins, Strauman, and Klein, 1985). All may
depend on the mental accessibility of particular knowledge configura-
tions momentarily lit by our wandering searchlight of consciousness. In
this sense, we may all have multiple and possibly disparate personalities
variously manifest on different occasions (Gergen, 1968, 1977). We
might occasionally be Hydes even if normally we are Jekylls; we might
resemble to a degree "Eves with many faces."
These notions are recognized by major theories of the human na-
ture. For instance, Freud's tripartite division of the psyche into the id,
the ego and the superego portrays abasie disjunctivity presumed com-
mon to all human beings. According to the present interpretation, how-
ever, Freud's trichotomy represents merely one among many possible
ways of categorizing that part of our epistemic repertory concerned with
knowledge of our "wants," "desires," or "duties." A more diverse type
of disjunctivity is stressed by the role theorists (e.g., Cooley, 1956; Goff-
man, 1959), whereby persons at different times assurne different identi-
ties in accordance with various social roles presumed to fit the circum-
stances. But let us consider more closely the process whereby the
various knowledge configurations get constructed out of our mental
LEGO pieces.
A THEORY OF LA Y EPISTEMICS 13

THE PROCESS OF KNOWLEDGE ACQUISITION

HYPOTHESIS GENERATION

(1) Cognitive Capability

As noted earlier, the process whereby all our knowledge is acquired
may be characterized in terms of its two aspects: hypothesis generation
and hypothesis validation. It is not altogether clear how a given concep-
tual configuration (a hypothesis) arises out of the manifold of concepts
and ideas flowing through our mind (for an excellent historical survey of
thinking about hypothesis generation, see the chapter by Campbell,
1974, on Evolutionary Epistemology). But it seems safe to assume that
the component pieces must be (1) available in our general store of LEGO
bricks and (2) be momentarily accessible or Ht by our searchHght of
awareness.
Consider a person whose car suddenly stalled in the middle of a
highway. If the driver happened to be an expert in auto mechanics and
had available a large repertory of knowledge concerning automotive
concepts and ideas, he or she might be capable of producing corre-
spondingly numerous hypotheses to account for the car's failure. By
contrast, a person lacking an appreciable degree of automotive expertise
may produce fewer potential explanations, and possibly may despair of
locating the source of the trouble. Furthermore, to the extent that a
particular subset of our individual's automotive notions was recently
primed (say, by an article just skimmed about carburetors), they are
more likely to figure in this person's hypotheses than alternate, less-
accessible notions.

(2) Epistemic Motivations: Closure Seeking versus Avoidance and Specificity

Nonspecificity Dimensions
In addition to the long-term and momentary capability to engender
ideas on a given topic, it is important to consider one' s motivations for
doing so. No epistemic activity is likely to take place without a modicum
of motivation, that is, a minimal degree of interest in a topic. But beyond
the degree of motivation necessary to get the epistemic process under
way, it is weIl to distinguish between several types of pertinent
motivation.
According to the present proposal, epistemically relevant motiva-
tions can be classified on two orthogonal dimensions; of closure seeking
versus avoidance and of specificity versus nonspecificity. The first dis-
14 CHAPTER2

Table 1. A Two-Dimensional Classification of Epistemic Motivations

Disposition toward closure

Type of motivating closure Avoidance Seeking

Nonspecific Need to avoid Need for

(nonspecific closure) (nonspecific closure)
Specific Need to avoid Need for a
a specific closure specific closure

tinction asks whether the individual desires cognitive closure on a topic

or wishes to avoid· closure and keep an open mind. The second distinc-
tion asks whether the desired or avoided closure is of a special kind, or
whether any closure or absence of closure will do. The two orthogonal
distinctions define a 2 x 2 matrix, depieted in Table 1. This yields a
typology of four motivational orientations: The needs for nonspecific or
specific closure and the needs to avoid nonspecifie and specific closure.
Let us examine them in turn.

Need tor Nonspecific Closure. The need for nonspecifie closure repre-
sents the desire for adefinite answer on some topie, any answer as
opposed to confusion and ambiguity. Such need thus represents a quest
for assured knowledge that affords predietability and a base for action.
A heightened need for nonspecifie closure is likely to trigger an intense
epistemic activity, where no initial knowledge existed, and to bring such
activity to a halt once some plausible hypothesis was advanced and
supported by extant evidence. In this sense, the need for nonspecific
closure is said to promote epistemie "freezing" (Kruglanski and Ajzen,
1983; Kruglanski, and Freund, 1983; Freund, Kruglanski and Schpitza-
jzen, 1985).
Four additional assumptions further characterize the need for non-
specifie closure construct. This need is assumed to be (1) topie bound, (2)
capable of originating from diverse possible motives, (3) aroused in spe-
cifie situational contexts, and (4) essentially unbiased or nondirectional
in its judgmental effects. These properties are now discussed in greater
detail.
Topie Boundedness. It is assumed that persons do not seek just any
knowledge, only knowledge in whieh they have some special interest.
Such interest may arise from "intrinsie" curiosity concerning specific
aspects of the environment (Berlyne, 1960), or from perceived "extrin-
sie" utility of such knowledge for mastery, or control over one's out-
comes (Berscheid, Graziano, Monson, and Dermer, 1976; Erber and
A THEORY OF LAY EPISTEMICS 15

Fiske, 1984; Fiske and Neuberg, 1988; Kelley, 1967, 1971; Neuberg and
Fiske, 1987; Weiner, 1985a).
Whereas a need for nonspecific dosure is likely to vary across topics
and situations, stable individual differences in such a need mayaiso
exist. Thus, persons might reliably vary in the disposition to apprehend
their world in dear-cut, unambiguous terms. Such individual differ-
ences, presumably general across topics, have been identified by various
authors (e.g., Frenkel-Brunswik, 1949; Rokeach, 1960) and have been
explained by diverse socialization experiences or cultural practices
(Hofstede, 1980).
Diverse Origination. The need for nonspecific dosure could spring
from heterogenous motivational sources. One's intrinsic interest in foot-
ball, or in child rearing might derive from a motivation for strong gender
identity; an interest in finances might derive from a motivation for se-
curity. Self- esteem and social approval concerns might elicit one's inter-
est in various topics (e.g., the arts, or world affairs) and hence to poten-
tiate needs for nonspecific dosure in the corresponding domains.
Extrinsic interest in a given knowledge could arise from diverse
types of outcomes that such knowledge might help control. In turn,
those outcomes might cater to a variety of different motives: Achieve-
ment strivings could arouse a need for nonspecific dosure, where
knowledge was believed to facilitate excellence. A need for nonspecific
dos ure could be similarly based on power strivings or a craving for
affecfion.
Situational Arousal. The need for nonspecific dosure could be
aroused by various situational conditions in two categories: (1) those
implying the benefits of possessing firm knowledge on a given topic,
having to do with the guidance and predictability that dosure may af-
ford, and (2) those implying the costs of lacking closure on a topic of
interest, having to do with the need to invest time and effort in further
information processing, or an inability to pursue salient alternative in-
terests. For instance, the need to make adecision or to undertake an
action may augment the need for orienting cognitive dosure in the rele-
vant domain. Similarly, granting one's initial interest in a topic, compet-
ing interests or time pressure generally may intensify one's need for
nonspecific dosure: Confusion or ambiguity could be time-consuming,
as it involves the cognitive juggling of alternate possibilities. By contrast,
dosure is time-efficient, because of its coherence and unitary nature. An
empirical relation between (subjectively imposed) time pressure and the
tendency to render extreme and confident judgments (assumed likely
under high need for dosure) has been noted in research on the A-type
personality (Humphries, Carver, and Neumann, 1983).
Unbiasedness. Persons are often interested in an answer to a ques-
16 CHAPTER2

tion, without prejudiee as to the answer's special properties. A school

psychologist might administer an IQ test to a student without predilec-
tion for a specific outcome. A computer user might wish to be informed
about how new hardware operates without leaning toward a particular
mode of operation, and an air traveller might wish to know the depar-
ture gate of a plane, without being partial to a specifie gate. The "non-
specifie" qualifier of the present need-for-closure concept connotes such
an unbiased desire for definite knowledge on a particular topie.

Need for Specific Closure. At times, however, persons are partial to-
ward particular answers to their questions, whieh represent their needs
for specific closure. A need for specific closure could relate to any of the
answer's properties; for instance, to its particular contents that may be
flattering or otherwise desirable (Kruglanski and Ajzen, 1983; Kunda,
1986), to its novelty (Berlyne, 1960; Klar, Bar-Tal, and Kruglanski, 1988),
to its apparent creativity, in short, to any content-related, structural, or
formal feature that might appear attractive in given circumstances.
If the need for nonspecific closure may generally promote epistemie
freezing, that for a specific closure might promote either freezing or
unfreezing depending on whether current knowledge was congruent
with one's particular desires. For example, someone who entertains the
possibility of having contracted a dangerous illness or of being about to
lose an attractive job might seek out new evidence inviting more desir-
able conclusions, and/or be ready to accept alternative, more positive
interpretations of extant evidence. By contrast, a person who surmises
that he or she is about to receive a desirable job offer or that he or she
has done weIl on an exam might often be refractory to contrary evidence
and be quiek to dismiss alternative, less sanguine interpretations of
available information.
Thus, the term need for specific closure refers to a directionally biased
influence on the epistemie process. Such influence may steer the process
toward desirable knowledge structures. Furthermore, needs for specific
closures are assumed to be topie bound in accordance with individuals'
particular interests and inclinations. A high school student who aspires
to become an actor might care little about passing or failing a mathema-
ties exam, whereas a fellow student who hopes to become a physicist
may entertain definite preferences between the two outcomes.
Needs for specific closure may originate from diverse possible mo-
tives in reference to whieh specifie features of knowledge structures may
look attractive. For instance, esteem concerns might render ego-
protective or ego-enhancing attributions more attractive than their alter-
natives (Miller, 1976; Ross and Sicoly, 1979). The need for control might
induce preferences for controllable or important features of causal
categories (cf. Kelley, 1971, p. 22).
A THEORY OF LAY EPISTEMICS 17

Finally, needs for specific closure may be aroused situationally: The

apparent desirability of a given judgment may often derive from the
total situational context that lends it meaning and significance. For in-
stance, the knowledge that one is the last to leave the sinking ship may
have different implications for a person's esteem if he or she happened
to be the captain or the cook. The preference for an idea because of its
novelty or creativity depends on the prior context in reference to which
novelty and creativity are defined.

Need to Avoid (Nonspecijic) Closure. A third motivational force as-

sumed relevant to the epistemic process is the need to avoid closure. In
terms of its desired end state, this need is opposite to the need for
nonspecific closure. The need to avoid closure pertains to situations
where judgmental noncommitment is valued or desired. Under somecir-
cumstances, this might be occasioned by a person's fear of invalidity,
due to the appreciable expected costs of amistaken judgment (Freund et
al. , 1985; Kruglanski and Ajzen, 1983; Kruglanski and Freund, 1983;
Mayseless and Kruglanski, 1987). Where the erroneous commitment to a
judgment threatens to prove costly, a person might value a lack of
closure, for example, occasioned by indecision among multiple compet-
ing hypotheses and/or inconsistent items of information.
At times, lack of closure might be valued temporarily as an interim
stage or as a means that may ultimately reduce the likelihood of a mis-
take Gones and McGillis, 1976). On alternative occasions, however,
where closure may not appear essential, its avoidance may appear as a
permanently desirable state. For instance, in attitude measurement,
provision of an ambiguous response alternative (a "neutral" midpoint
on a scale) often reduces the proportion of definite (e.g., yes or no)
responses (Aronson, Brewer, and Carlsmith, 1985). This may indicate an
occasional preference for cognitive noncommitment possibly prompted
by a fear of invalidity.
A need to avoid closure mayaiso arise from the unwanted restric-
tion of potential that adefinite judgment may impose. Snyder and Wick-
lund (1981) furnished an illuminating discussion of cases where definite
self-knowledge might connote unattractive predictability and dullness;
in such instances, persons might weIl prefer to maintain cognitive non-
commitment and eschew closl,lre. In other cases discussed by Snyder
and Wicklund, "mystery" might be compatible with a cherished, roman-
tic worldview. On those occasions, too, a sanguine ambiguity might be
preferred over a sombre closure. Wisconsin's Senator William Proxmire
articulated such sentiment clearly. In critically commenting on social
psychological investigations of romantic love he said: "I believe that 200
million ... Americans want to leave some things in life a mystery, and
right at the top of things we don't want to know is why a man falls in
18 CHAPTER2

love with a woman and vice versa" (cited in Wisconsin State Journal,
March, 20, 1975, p. 2).
The need to avoid closure may instigate intense epistemic activity
where closure was "in danger" of forming. This might induce increased
sensitivity to new information possibly inconsistent with the current
hypothesis and/or increased tendency to generate competing alterna-
tives to the hypothesis. Furthermore, with absence of closure to start
with, a need to avoid closure might suppress further epistemic activity
that "threatens" to lead to unwanted closure. Generally, then, the need
to avoid closure may promote epistemic unfreezing.
As with the needs for (nonspecific or specific) closure the need to
avoid closure may often be topic bound. Thus, one might prefer to abstain
from closure in regard to some issues only and not to others. Similarly,
the need to avoid closure could originate from diverse possible motives,
for example, tied with the manifold possible costs of judgmental errors.
Thus, error may threaten to lower one's self-esteem, inflict economic
losses, or jeopardize one's physical well-being. Control strivings could
enhance one's need to avoid closure where judgmental commitment
might entail the restriction of control over potentially desirable out-
comes (e.g., adecision to accept a given job offer may nullify the desir-
able outcomes of other potential positions). The need to avoid closure is
assumed to be unbiased in its effects on the epistemic process: It may
steer epistemic activity away from closure in general rather than away
from particular closures. Finally, the need to avoid closure is assumed to
be aroused situationally. As with the epistemic needs discussed earlier,
the perceived negative valence of closure may depend on the situational
context that lends it aversive hedonic significance.

Need to Avoid Specific Closure. Often, persons may be motivated to

avoid specific closures because of the undesirable properties of such
closures. The need to avoid specific closure may occasionally represent
the need for the opposite closure. For instance, the need to avoid a belief
that the weather conditions on Sunday will be poor may be indis-
tinguishable from the need to believe that they will be good. However,
occasionally the need to avoid a specific closure is not tantamount to the
need for the opposite closure. The difference may reside in the individu-
al' s . focus of interest: So me persons may focus on the closure to be
avoided and not be particularly preoccupied with its positive opposite.
Others may be strongly attracted to the positive closure and not be
particularly concerned about its negative counterpart. In a similar vein,
workers in the domain of achievement motivation (e.g., Atkinson, and
Birch, 1970) distinguish between individuals motivated to avoid failure
and those who strive to attain success. In present terms, the former
A THEORY OF LAY EPISTEMICS 19

individuals are motivated to avoid a specific closure that asserts they

have failed on some task; by contrast, the latter individuals are moti-
vated by the need for a closure that asserts they have succeeded.
As with the need for a specific closure, the need to avoid a specific
closure may occasionally effect epistemic freezing and at other times
bring about epistemic unfreezing. Thus, when the person's current no-
tions represented the closure to be avoided (e.g., that one has failed an
exam), a strong need to avoid closure would promote unfreezing,
whereas when one's current notions (e.g., that one has passed the
exam) did not represent a negatively valenced closure, a strong need to
avoid it would promote freezing.
The need to avoid a specific closure could be aroused situationally
and could stern from diverse possible motives. For example, reactance
against an attempt to usurp one's freedom (Brehm, 1966) could render
the usurper's recommendation a closure to be avoided. Thus, in a study
by Worchel and Brehm (1971), subjects reduced their preference for a
task after a confederate threatened their freedom by stating that the task
should obviously be chosen over its alternative. In addition, the effects
of commitment could occasionally induce a need to avoid a closure at
odds with the committed position. Indeed, Kiesler and Sakamura (1966)
found that commitment to a position did not render the position itself
more attractive but it increased subjects' resistance to an attack on the
position. In present terms, the commitment manipulation did not seem
to increase the need for a specific closure (the committed position) but
rather the need to avoid the specific closure entailed in the attack. In a
clever study that combined commitment and reactance effects Schwartz,
Frey and Kumpf (1980) found that either committment or reactance was
sufficient to induce resistance to a counterattitudinal communication
(or, in present terms, to arouse a need to avoid a specific closure con-
tained in the communication).
According to the present analysis, a need to avoid a specific closure
could arise from numerous additional motives beyond reactance or com-
mitment. For instance, a fear of rejection could lead one to avoid the
belief that one is unpopular, and a fear of high places, that one would
have to travel by air.

Compatibility among the Epistemic Motivations

It is of interest to consider whether the four epistemic motivations
are mutually compatible. Simply put, it should be possible for a person
to des ire simultaneously the various epistemic end states but not to attain
them simultaneously. For instance, one cannot have closure simultane-
ously and avoid it. Furthermore, the ends of nonspecific and specific
20 CHAPTER2

c10sure might occasionally c1ash. For instance, a need for nonspecific

c10sure might be equally gratified by success or failure information,
since either disambiguates a performance outcome. By contrast, a need
for specific c10sure might be gratified only by success information and be
frustrated by failure information.
Yet one might simultaneously desire to attain and avoid c1osure,
wish for a specific c10sure as well as for a generalized c1osure, in short,
strive to "eat the cake and have it too." For example, a scientist might
desire cognitive c10sure on a problem of interest yet may concomitantly
experience a high fear of invalidity stemming from an anticipated crit-
icism of the work by colleagues, inducing a need to avoid c1osure. A
request for feedback on a manuscript might be motivated by the writer' s
genuine interest in a colleague's opinion (representing a need for non-
preferential c1osure), as weH as by a strong wish for an approving reac-
tion (representing a need for specific c1osure).
Simultaneous existence of incompatible epistemic goals may en-
gender conflict and tension presumably resolvable in the direction of the
stronger of the opposing forces (Lewin, 1938; 1951). In fact, in natural
settings persons might only rarely have a singular epistemic goal and
more typically might experience a mix of epistemic motivations, for ex-
ample, of ultimately craving c10sure while temporarily avoiding it. In
such a case, the relative magnitudes of epistemic forces operative at the
moment might affect the extent, direction, and outcome of information
gathering. Thus, it is important to understand the effects of each epis-
temic motivation separately, even if in natural circumstances it often
comes blended with other motivations.

Summary: The Epistemic Motivations Compared and Contrasted

The four motivational forces described above are assumed to fulfill
the same general function of initiating and/or terminating the epistemic
activity. However, they are assumed to differ in their desired end states.
The need for nonspecific c10sure is gratified where any c10sure (on a
given topic) is attained, the need for specific c1osure, where some desir-
able c10sure is formed, the need to avoid c1osure, where no c10sure is
formed, and the need to avoid a specific c1osure, where some undesirable
c10sure fails to form.
The ends of the motivations for c10sure are illustrated in Figure I,
and those of the motivations to avoid c1osure, in Figure 2. In Figure 1 the
left circ1e represents a set of c10sures with property A and the right circ1e
a set of c10sures with property B. As indicated by the appropriate ar-
rows, the need for nonspecific c10sure is gratified by structures in either
set (Le., A or B). By contrast, the need for a specific c10sure (say, with
A THEORY OF LAY EPISTEMICS 21

Needtor Needtor
Speci!ic Oosure Speci!ic Oosure
(A) (ß)

Need tor Nonspeci!ic Oosure

Figure 1. Desired end states of motivations for closure.

Need to Avoid Non specific Oosure

Need to Avoid Specific Oosure (A)

Figure 2. Desired end states of motivations to avoid closure.
22 CHAPTER2

property A or property B) is gratified exc1usively via structures in the

corresponding set (A or B, respectively).
In Figure 2, absence of c10sure is represented either by X, the over-
lap between the eirc1es (where structures of types A and B are in com-
petition), or by Y, the area around the circ1e, signifying the total lack of
eognitive struetures (hypotheses) on a topic. As shown, a need to avoid
(nonspeeific) c10sure is gratified where either of the above cases obtains.
By contrast, a need to avoid a specific c10sure A is gratified either by a
c10sure of the B type or by the absence of c10sure (X or Y).
All four epistemic needs are assumed to be topic specific, and capa-
ble of originating from diverse motives or goals. This suggests that par-
ticular motivational origins do not afford unambiguous predictions as
far as the epistemic process is concerned. For example, a heightened
need for self-esteem, or for control, might exert diametrically opposed
effects on the information gathering process depending on the kind of
epistemic motivation it engendered in specific circumstances.
Furthermore, the needs for specific c10sure or the avoidance of spe-
cifie c10sure are assumed to exert biased or directional effects on the
epistemic process, whereas the needs for nonspecific c10sure or the
avoidance of c10sure are assumed to be directionally unbiased in their
effects. Finally, although the various epistemic ends often are incompat-
ible, they could be simultaneously entertained by persons in various
situations. Such a mix of epistemic motivations might give rise to tempo-
rary tension, which particular resolution may affect the extent, direc-
tion, and outcome of the informational search.

THE PROCESS OF HYPOTHESIS VALIDATION:

PERSON AS A NATURAL LOGIClAN

An individual motivated to attain knowledge on a given topic may

often wish to validate his or her hypotheses. According to the present
theory, this is accomplished in a subjeetively logical fashion: The indi-
vidual departs from apremise that links, in an "if-then" form, a given
category of evidence with a given hypothesis and proceeds to infer the
hypothesis on affirming the evidence. For example, a person could vali-
date the hypothesis that it is raining by observing drops of water falling
from the sky. According to the present interpretation, this presupposes
the conditional premise: if falling drops, then rain. Thus, "falling drops"
could be considered evidence for the "rain" hypothesis.
In some situations, the same bit of evidence could be linked with
several alternate hypotheses. For instance, a guest's compliment on the
quality of one' s cooking could be interpreted as evidence for one' s culin-
A THEORY OF LAY EPISTEMICS 23

ary success or for the guest's politeness and good manners. To decide
between the alternate hypotheses one would have to construct an addi-
tional premise (again conditional in form) involving diagnostic (Jones and
Davis, 1965; Trope and Basok, 1983) or noncommonly deducible (Krug-
lanski, 1980) evidence. For instance, if one's own children, who are
known not to suffer from excess politeness, also praised the cooking,
this would support the "culinary success" but not the "sheer politeness"
hypothesis.
In validating or "proving" a given idea we thus transmit our subjec-
live confidence from our premises to our conclusions. Of course, per-
sons may vary greatly in the types of premises to which they subscribe.
Accordingly, they could prove their ideas via markedly different types
of evidence. A cultist, for example, might believe that a statement is
worth believing only if endorsed by the guru, and an experimental
scientist might believe that hypothesis x is true only if an empirical
outcome y was observed in a controlled experiment. Note that although
the contents of these premises differ substantially, they both share the
conditional (if-then) form that serves as a basis for deductive validation.
The experimentalist may accept the hypothesis if he or she experimen-
tally observed the relevant datum, and the cultist may accept a state-
ment if it was endorsed by the guru. Thus, people's premises may vary
as may the evidence they consider relevant or legitimate. However, the
logical process of proving one's ideas seems uniform across disparate
instances of inference. These issues will be dealt with more extensively
in a subsequent section.

ENCODING OF STIMULUS INFORMATION AS HYPOTHESIS VALIDATION

It is tempting to speculate that the encoding of sensory information

in terms of durable categories involves the logical hypothesis-testing
process previously described. Suppose a bundle of colors, shapes, and
sounds is encoded as a dog. This requires that the perceiver have pre-
vious knowledge of the category "dog" and of the visual and auditory
qualities prototypically associated with it. In logical terms, the perceiver
may depart from the premise whereby only if a stimulus pattern is a dog
will it display a specific constellation of shapes, sounds, and move-
ments. The inference "dog" would logically follow if the pattern at issue
was observed.
In some instances, the same stimulus pattern might fit more than
one encoding category: Suppose an approaching dark figure was equally
congruent with the "man" and "woman" labels. In other words, none of
the currently noticed features (longish hair, slacks) seemed sufficiently
diagnostic to set apart the"man" from the "woman" hypothesis. Sup-
24 CHAPTER2

pose now a glimpse of a mustache was sighted. This may allow a ready
encoding of the stimulus as "man", based on the premise that only a
man and not a woman can sport a mustache. Not in all instances is
diagnostic information available. In such cases ambiguity might prevail.
Perceptual examples of such ambiguity would be the various reversible
figures (like the Necker cube or the face-vase pair) in which the available
stimulus pattern does not afford adecision among rival encoding
hypotheses.
The foregoing discussion suggests that perceptual encoding phe-
nomena might be governed by the same epistemic principles as those
ruling the more conceptuallevels of knowledge acquisition. This possi-
bility has several researchable implications. For instance, under high
need for elosure; individuals might exhibit a lesser tendency to vacillate
between members of a reversible pair of figures and settle quicklyon a
given member. The opposite tendency might be manifest under a high
need to avoid elosure.

"Unfounded" Knowledge
Not every single bit of confident knowledge needs to be linked in a
person' s awareness with other knowledge from which it has been de-
rived. Once a given proposition was accepted as true, a person may
forget its evidential basis. For instance, we may be certain that the area
under a cirele is the largest for a given perimeter, or that the earth rotates
around the sun, without quite recalling the logical support for our be-
liefs. In this sense, persons may not be conscious of the reasons for their
behavior, which (if pushed a bit) could be taken to mean that some of
our actions may have unconscious determinants.
Furthermore, some of our beliefs might have never been derived
from evidence. A person might "simply know" that something is the
case, having "strong intuition" or "premonition" about it without expe-
riencing a need to prove such knowledge in any way. In lay epistemic
terms, this could occur because of strong motivational forces suppressing
the generation of alternative hypotheses and preventing the recognition
of evidence inconsistent with one's current beliefs. Alternatively, a per-
son may lack the capability to come up with alternatives to a given hy-
pothesis, and be unable to imagine that things could be other than what
they presently seem. Thus, although positive proof may enhance our
confidence in a proposition, not every confidently held proposition
needs to have been systematically proven.

Logical Contradiction: A Reason for Doubt

In the same way that logical derivability or consistency of one' s
conelusions with one's premises fosters confidence, logical inconsisten-
A THEORY OF LAY EPISTEMICS 25

cy or contradiction may foster doubt. However, in order to have such an

effect, the contradiction needs to be recognized by the individual. In turn,
what is recognized as contradiction may vary from one person to an-
other, and from one context to the next. For instance, the terms love and
hate are typically thought of as antonyms. Hence, loving and hating the
same person at the same time may see m a contradiction in terms. Yet for
some persons in some situations loving and hating may be a faithful
representation of an emotional state rather than a contradiction. Thus,
contradiction is subjective or is in the eyes of the beholder.

Resolving a Contradiction
Given a subjectively recognized contradiction, a sense of knowl-
edge may only be attained if the contradiction was resolved. This may be
accomplished by denying one of the contradictory propositions. P and
not-p may not both be true; therefore, one of them must be recognized
as false, hence, denied. Which of the two contradictory statements
would be subject to denial may depend on the initial confidence with
which they are held. A proposition held with lower initial confidence
may be 4enied more readily. For instance, in an encounter between a
fact and a hypothesis the latter usually gives way to the former. Accord-
ing to the present interpretation, this is because hypothesis often repre-
sents a lower degree of belief than fact. A review of some empirical
evidence in Chapter 3 supports the thesis that in resolving a cognitive
inconsistency denial attempts are often directed as the less firmly be-
lieved of the contradictory propositions.

The Limitations of Human Logicality: Form versus Content of Inferences

It is presently assumed that the logical mode of validating or prov-
ing one' s inferences is highly prevalent and that it generally character-
izes the thinking of normally functioning persons (children as weIl as
adults, native peoples as weIl as sophisticated Westemers, lay persons
as weIl as scientists). This is not to imply thatpeople are familiar with
the intricacies of formallogic or that they do not frequently err on simple
logical tasks. Indeed, there exists by now an extensive body of empirical
evidence that they do (for extensive discussion and reviews, see Wason
and Johnson-Laird, 1972; Bourne et al., 1979). According to the present
interpretation, however, such logical "errors" stern from various cogni-
tive and motivational factors that affect the contents of subjects' premises
and/or conclusions; theyare compatible with the assumption that the
form of human reasoning remains deductive.
Consider the following example of a common logical error. Subjects
are presented with the premises that all males possess the XY chromo-
26 CHAPTER2

some pair and that P possesses the XY pair. From these premises, they
erroneously conclude that P is a male. This conclusion is not entailed in
the premises: That all males have the XY pair does not imply that all
persons who have the XY pair are male, and only the latter assumption
logically warrants the conclusion P is a male. Possibly, in this instance,
subjects mistook the first assumption for the second and proceeded in a
logically proper fashion to derive their conclusion. Such "conversion" of
premises (Chapman and Chapman, 1959) could occur for a variety of
reasons. For instance, subjects might have subscribed to the indepen-
dent belief that all XY chromosomes' possessors are male and this might
have facilitated the conversion. The fact that subjects' independent be-
liefs and implicit assumptions affect their tendency to favor some deduc-
tions, whether correct or incorrect, over others has been realized in
some of the earliest investigations of human inference (Wilkins, 1929).
Alternatively, a conversion might occur for motivational reasons.
Thus, subjects under high need for nonspecific closure may be strongly
motivated to reach adefinite conclusion. Given the information that P
has the XY chromosome pair, adefinite conclusion can only be reached
if the initial premise states that all XY possessors are males. By contrast,
the original (unconverted) premise, all males are XY possessors, does
not afford adefinite conclusion concerning P's identity. A conversion of
premises mayaiso be prompted by a need for specific closure. Suppose a
considerable material benefit was to accrue to the inference maker if P
were a male. This might facilitate a conversion of premises allowing
such a conclusion to be reached. On the other hand, if P' s maleness ran
counter to the inference maker's interests, the likelihood of a conversion
might be lower. The effects of preferential closures on subjects' tenden-
cy to derive erroneous or correct conclusions has been widely demon-
strated in empirical research (McGuire, 1960b; Morgan and Morton,
1944).
As our "searchlight" metaphor suggests, subjects' tendency to
make a specific deduction may depend on the degree to which its com-
ponent categories are cognitively accessible. For instance, in Wason's
(1966) four-card task, subjects test the validity of the proposition, if there
is E on one side of the card there is 4 on the other side. Subjects are then
presented with four cards bearing respectively an E, a K, a 4 and a 7.
Subjects' task is to overturn the minimal number of cards necessary for
assessing the proposition. Under those circumstances, subjects pre-
dominantly choose the E and the 4 cards; that is, those with categories
"primed" by the statement of the task objective (Evans and Lynch,
1973).
Increased tendency to favor deductions involving categories primed
by the logical task was demonstrated in early research by Woodworth
and Sells (1935) and Sells (1936). These authors dubbed this phenome-
A THEORY OF LAY EPISTEMICS 27

non the "atmosphere effect," referring to subjects' pervasive tendency

to prefer conclusions phrased in terms of categories contained in (hence,
presumably primed by) the premises over conclusions phrased in alter-
native categories. For instance, if the premise contained a universal
quantifier (e.g., all) a conclusion with such quantifier tended to be fa-
vored over a conclusion with an existential quantifier (e.g., some). Sim-
ilarly, if the premise contained a positive (all, some) or a negative (none,
few) quantifier, conclusions with the corresponding quantifiers tended
to be favored over their alternatives.
Note that the various cognitive or motivational effects on the in-
ference process are compatible with the notion that, irrespective of con-
tents, the form of our reasoning remains deductive. Conversion of the
initial premises due to such effects is thus compatible with a correct
derivation of conclusions from the new premises (Chapman and Chap-
man, 1959). In fact, persons could reason deductively from premises
whose content contradicts the very meaning of logical concepts. Thus,
in Wason's (1966) four-card selection task, some subjects attempt to
verify the if Ethen 4 conditional by seeking the E, 4 conjunction. Such
subjects appear to subscribe to the erroneous premise if E and 4 then (if E
then 4). Yet although erroneous from the logical standpoint the latter
statement itself is a conditional. Thus subjects' verifying conclusion (that
the if Ethen 4 statement is true because the E, 4 pair was observed)
follows logically from their premises even if the content of these premises
betrays a lack of insight into the connotations of the conditional as a
logical term (Johnson-Laird and Wason, 1970).
The foregoing analysis is thus compatible with the view that per-
sons are "partial logicians" (cf. Bourne et al., 1979). According to the
present interpretation, they are logicians in form of their reasoning only,
not necessarily in content. It is noteworthy that traditional critiques of
people's putative illogicality often addressed the unacceptable contents
of their inferences. In the following seetions, let us briefly revisit two
major such critiques: the Piagetian notion of a prelogical stage in human
development and Levy-Bruhl' s notion of illogical reasoning among na-
tive tribes.

Prelogical Developmental Stage. Consider the reasoning of the pre-

operational child characterized in Piaget's theory as prelogical (cf. Fla-
vell, 1963). The preoperational period extends approximately between
the ages of a year and a half to seven. Ouring that time, the child's
reasoning is said to involve "flagrant contradictions with earlier cogni-
tions," presumably because of an inability to master the notion of rever-
sibility. The above limitations are often illustrated by the water-Ievel
problem testing the child' s ability to conserve quantity.
On the water-Ievel task, the child may be presented with two differ-
28 CHAPI'ER2

ently shaped beakers, say, a wide and short one and a narrow and tall
one. At this point one of the beakers, say, the tall one, is filled with
water up to a certain level while the second is left empty. The water is
then poured into the second beaker and the chiId is asked whether now
the water is as much, more, or less than before. A preoperational child
might say that it is less because the water level in the short and wide
beaker is lower than that in the tall and narrow beaker. According to
Piaget, the child' s judgment in this case involves a straightforward logi-
cal contradiction: A given amount of water is not this amount after all.
And the reason for such a judgmental error, according to Piaget, is the
child's inability to conceptually reverse the act of pouring so as to rein-
state in imagination the water to the original beaker. Presumably, had
the child been capable of mental reversibility, it would immediately
realize that the water level in the original beaker remained the same;
hence, that the water quantity could not have changed.
But note that a faiIure to yield the correct answer to the water-Ievel
problem hardly refutes the possibility that the preoperational child rea-
sons logically, as far as the the form of its inferences is concemed. The
contradiction in the waterlevel problem may go unrecognized by the
child, because it depends on an adult view of the relation between
quantity and shape: Most adults in our culture subscribe to the view that
the quantity of matter does not vary with a change in shape. But a young
child may not have yet adopted such a view; after all, there are numer-
ous views that the adult has not yet adopted. Furthermore, it is possible
to be a "nonconserver" on the water-Ievel task and yet have perfect
command of the reversibility concept. Thus, one could envisage the
same water-Ievel being reinstated in the original beaker and yet believe
that quantity is correlated with shape and that the amount of water
varies back and forth between the differently shaped beakers. In short,
an ability to "conserve" in this case seems to depend on a whole set of
beliefs and assumptions regarding the relation between quantity and
shape. A young child may lack some of those beliefs; yet granting his or
her particular beliefs, his or her reasoning may be implicational in form
and in this sense 10gicaI.
To see this, answer a young child's query "Where is Mommy?" by
pointing to the kitchen and watch it immediately proceed in that direc-
tion. How else may one account for such commonplace occurrence other
than by assuming that the chiId reasons logically from apremise where-
by if mother is in the kitchen, getting there will bring me in her pres-
ence. AItematively, try answering the chiId's query by asserting that
mother is at once in two separate places (e.g., the living room and the
kitchen) and notice the confusion, uncertainty, and doubt manifest in
those circumstances. Examples such as these suggest that the young
A THEORY OF LAY EPISTEMICS 29

child reasons logically from conditional premises. And this is perfectly

compatible with the variety of "errors" the child may commit because
the contents of its premises contradict those of an adult.

The Reasoning of Native Peoples. The thesis that native peoples think
in a logically inferior way to sophisticated Westerners was put forth by
Levy-Bruhl (1910) in his book, How Natives Think. In this work, Levy-
Bruhl discusses
a form of thinking rare among us but dominant among primitives, governed
by a law 01 participation. Under this law, phenomena have the attribute of
being themselves and yet partaking of other phenomena as weil, in contrast
to the dominant form of Western thinking governed by the logical law 01
contradiction.

The "following beliefs are instances of pre-Iogical thinking: a group of

Brazilian Indians claim that they are also parrots; Bororo believe that
portraits possess some of the qualities of life of their models" (Cole and
Scribner, 1974, p. 142).
But note that the law of contradiction as such does not prohibit a
Brazilian Indian to be also in some sense a parrot or a portrait, to possess
the life qualitites of its model. The law of contradiction does not specify
which proposition contradicts another. All it asserts is that given a sub-
jectively recognized contradiction, a person would not have the experi-
ence of knowledge. Instead, the person would experience doubt, uncer-
tainty, and confusion.
Actually, at this late date, Levy Bruhl's hypothesis is hardly a viable
position. It has been vigorously rejected by major anthropologists start-
ing with Boas (1911). A.F.C. Wallace, for example, demonstrated that
previously baffling kinship terminologies and other conceptual systems
of native peoples have their underlying logic, and this conclusion is
shared by many of today's cultural anthropologists (Cole and Scribner,
1974, p. 143). Following the works of Luria (1971) and Henle (1962),
Scribner (in Cole and Scribner, 1974) conducted studies of syllogistic
reasoning among the Kpelle tribe in Liberia. Cole and Scribner's conclu-
sions drawn from this and other similar research are instructive regard-
ing the issue of native logicality. In their words
with all the varied materials and procedures that have been used, we have
encountered no examples that violate the logicallaw of contradiction; analy-
sis shows that the major sources of error in Kpelle performance are of the
same kinds as those involved in the reasoning of American college students.
(p. 168)

To reiterate, those latter errors also pertain to the contents of premises

and inferences. They do not impugn the notion that the form of people's
reasoning remains deductive.
30 CHAPTER2

The Problem of Rationality. The foregoing discussion of human log-

icality is not meant to paint a rosy picture of the human being as a
predominantly "rational" creature. Nor is it meant to suggest that hu-
man inference effectively minimizes the inddence of judgmental errors
and biases. Thus, there is more than a kernel of wisdom in Braude's
(1965) definition of logic as "an organized procedure for going wrong
with confidence and certainty" (p. 48). The lay epistemic conceptions of
bias and accuracy in human judgment are dealt with extensively in
Chapter 8.

FUNCTIONS OF KNOWLEDGE

The knowledge contained in people's repertories importantly af-

fects the way they interact with their sodal and physical environment.
Recent sodal cognitive research on the functions of schemata (see Fiske
and Taylor, 1984, for a review) attests to the influence of existing knowl-
edge structures on the way persons interpret the present, remember the
past, and predict the future. Consider an instance in which a new item
of information is encountered. An interpretation of this information that
is consistent with an accessible knowledge structure (schema) is likely to
be favored over'an inconsistent interpretation. Suppose that at an art
exhibit you sight a painting interpretable alternately as the work of
genius and the primitive doodle of a child. Suppose now you access the
knowledge that the exhibit features the works of a famous modernist,
for example of Picasso or Klee. Trus knowledge represents evidence
consistent with the former evaluation of the painting and inconsistent with
the latter; hence, it may exert a discriminative influence on the interpre-
tation you are likely to adopt.
Recent research on schemata attests that our existing knowledge
structures promote in various ways "going beyond the information
given" (Bruner, 1973). For instance, an existing schema may bias our
memory toward the recall of schema-consistent information not origi-
nally contained in the acquisition set (Cantor and Mischel, 1977;
Markus, Crane, Bernstein, and Siladi, 1982). Furthermore, existing sche-
mata afford inferences (postdictions or predictions) about formerly un-
spedfied drcumstances (Markus, 1977).
According to the present theory, going beyond the information
given is an inevitable feature of all knowledge, whether lay or scientific.
In asense, this is analytically implied by the respective meanings of the
"evidence" and "hypothesis" concepts insofar as these are distinct from
each other (in an if X then Y proposition X is distinct from Y): evidence
A THEORY OF LAY EPISTEMICS 31

(or information) istor a hypothesis (or ascherna); it is not tantamount to

the hypothesis (or the schema). Hence, by definition the hypothesis (or
the schema) has meaning (or implications) nonoverlapping with evi-
dence on which it is based. The obverse is also true in that the evidence
for a hypothesis contains features (meanings, implications) not con-
tained in (or derivable from) the hypothesis as such. In particular, each
item of evidence contains numerous "local" aspects irrelevant to the
hypothesis. Those may be "leveled" or neglected when the evidence is
used instrumentally in order to deduce a given hypothesis (Boume et al.,
1979).
Contemporary literature on schemata tends to emphasize their rela-
tively enduring nature. Persons are assumed to seek or remember infor-
mation that confirms their preexisting notions, which suggests that
schemata generally are self-perpetuating. By contrast to this somewhat
static view of knowledge structures the present analysis projects a dy-
namic image in which, just like LEGO constructions, our knowledge
structures or schemata, while often stable and enduring, can also be
modified under the appropriate conditions. Such conditions include the
accessibility of information inconsistent with an existing schema and the
appropriate motivational welcoming conditions: a lowered need for non-
specific closure, a heightened need to avoid closure, or a need for a
specific closure with which the existing schema is incompatible. Such
motivational conditions may increase the likelihood that the incon-
sistency would be recognized and/or resolved by abandoning the
schema.
Recent research findings are compatible with the foregoing analy-
sis. Thus, experiments by Lord, Lepper, and Preston (1984) suggest that
judgmental biases can be "corrected" when subjects are induced to con-
sider the opposite hypotheses. Similarly, research by Rose (1981) and
Wilder (1981) suggests that stereotypes and prejudice may be reduced
when alternative schemata are made salient. Furthermore, research by
Sentis and Bumstein (1979) suggests that when ample time for informa-
tion processing is allowed, hence, presumably lowering subjects' need
for closure, the tendency to remember schema-inconsistent information
is increased (see also Brewer, Dull, and Lui, 1980; Hastie, 1980). Further
relevant work conceming motivational conditions under which existing
schemata may persevere or be modified is reviewed in Chapter 3.
Last but not least important, an extant knowledge structure may
guide persons' actions or overt behaviors. This may be accomplished via
the derivation of behavioral intentions from relevant knowledge struc-
tures. This issue is discussed more fully in Chapter 6, which deals with
attitudes and the attitude-behavior relation.
32 CHAPTER2

SUMMARY

This chapter has outlined a theory of lay epistemics concerned with

psychological factors affecting the formation and modification of human
knowledge. The potential instability of our knowledge was depicted via
the LEGO blocks and the searchlight metaphors. The LEGO metaphor
conveys the notion that our knowledge structures can be highly varied.
Furthermore, they can be dismantled under the appropriate drcum-
stances and be replaced by alternative structures. This signifies the po-
tential for considerable personal (or sodal) change in beliefs on numer-
ous topics. In sdence, it holds out the possibility of revolutions in which
reigning world views are toppled and supplanted by alternative views
(Kuhn, 1962).
The searchlight metaphor conveys the notion that at any moment
only a slight portion of one' s total repertory of knowledge is at the focus
of awareness. Our thoughts are in constant flux (James, 1890) so that at
any given moment we may be consdous of slightly different notions. To
the extent that we act or react on the basis of those notions, our actions
may differ according to currently accessible schemata. Thus, the search-
light metaphor implies the potential for inconsistency in our momentary
beliefs and actions in accordance with the momentarily accessible struc-
tures illuminated by our searchlight.
Collectively, the LEGO and the searchlight metaphors imply a dy-
namic view of human knowledge whereby its contents are infinitely
varied and potentially changing. This contrasts with the static view of-
ten implidt in previous sodal cognitive formulations. For instance,
rather than positing a relatively stable drive for the sodal comparison of
abilities, opinions, and emotions (Festinger, 1954), the sodal com-
parison question is presently regarded as one merely among many, its
appearance in one' s mind depending on the ebbs and flows of one' s
stream of thoughts and assodations. Similarly, rather than assuming
that people are perennially concerned about ascribing causality (Heider,
1958), and that, moreover, they couch their causal questions in terms of
invariant categories (in particular, of internal or external causes), it is
proposed that people' s causal categories are highly diverse and that the
very concern with causality may be fleeting.
The process whereby knowledge structures crystallize or change is
assumed to contain the interwoven aspects of hypothesis generation
and validation. Hypothesis generation has to do with individuals' capa-
bility to come up with ideas on a given topic and their motivation to do
so. Epistemic motivations may be classed into four major categories: (1)
the need for nonspedfic closure that prornotes the early freezing of
beliefs or the attainment of early closure; (2) the need to avoid (non-
A THEORY OF LAY EPISTEMICS 33

specific) closure expressing a reluctance to commit oneself to adefinite

judgment and a disposition to unfreeze one' s existing beliefs; (3) the
need for specific closure that may promote the freezing or unfreezing of
beliefs depending on their degree of desirability to the individual; and
(4) the need to avoid spedfic closures because of their undesirable
properties.
People were assumed to validate their hypotheses by use of subjec-
tive logic in which conditional (if-then) premises are socially constructed
and used as a basis for various conclusions. A sense of "proven knowl-
edge" may result from a realization that one's conclusions are consistent
with (or derivable from) one's evidence; and a sense of doubt, from a
realization that they are inconsistent. Such portrayal of human logicality
is compatible with a variety of logical errors that people may often
commit.
A central aspect of the present epistemic theory is the assumption
that logic and motivation are the inevitable parameters of any inference
or judgment. This is in contrast to previous sodal cognitive views (like
attribution or dissonance theories) that disjunctively stressed either the
logical/informational or the motivational aspects of human reasoning.
According to the present theory, no inference including scientific in-
ference is devoid of a strong motivational component that affects the
conclusions ultimately rendered. Laypersons as weIl as sdentists may
differ in their epistemic motivations but there exists no best or most
rational motivation that may improve one' s chances for a correct in-
ference. Some of these issues are taken up systematically in Chapter 10,
where a comparison of lay and scientific modes of knowledge is at-
tempted. But now let us systematically review some lines of empirical
research inspired by hypotheses and predictions derived from the lay
epistemic model.
 3
Empirical Research in the Lay
Epistemic Framework

The lay epistemic model recently has been used to generate new re-
search concerning several topics in cognitive social psychology. In accor-
dance with the integrative character of the present conceptual frame-
work, the pertinent empirical research spans a broad range of previously
explored areas to which it introduces a new point of view. The work to
be presently reviewed can be classified into three categories dealing with
different factors assumed to affect judgmental behavior: (1) epistemic
motivations, (2) the "logic" of hypothesis validation, and (3) the con-
tents of knowledge structures. In the sections that follow, these lines of
research are considered in turn.

MOTIVATIONAL EFFECTS IN EPISTEMIC BEHAVIOR

THE CONTEXT OF KNOWLEDGE FORMATION

(1) Extent of Information Processing

Motivational effects in epistemic behavior have been recently stud-
ied in the context of knowledge formation. A major question raised by
the present model concerns the extent of information processing before a
firm judgment is reached. The epistemic analysis suggests that often the
extent of processing may be negatively affected by the need for (non-
specific) closure and positively affected by the need to avoid closure. In
particular, a heightened need for closure may enhance the individual's
tendency to freeze on an early hypothesis, whereas a heightened need
to avoid closure may effect a retardation of freezing. These notions are
examined in several experimental studies described below.

35
36 CHAPTER3

Primacy Effects in Impression Formation. Primacy effects in impression

formation (Asch, 1946) refer to persons' tendency to base impressions of
a target more on the early than on the late information. Such effects
could represent the phenomenon of epistemic freezing, because they
signify early closure and insensitivity to subsequent information. If so,
primacy effects should be augmented by a heightened need for non-
specific closure and reduced by a heightened need to avoid closure.
In an experiment designed to test those predictions, Kruglanski and
Freund (1983) presented subjects with serial information about a target
and had them rate this person's attractiveness as a job candidate. In
some cases, positive target information came first, followed by negative
information; in other cases, the order was reversed. Need for non-
specific closure was manipulated via degrees of time pressure. We had
assumed that time-pressured individuals would experience a height-
ened need for nonspecific closure because of the costs of extensive infor-
mation processing under time constraints.
The need to avoid closure was manipulated via fear of invalidity
aroused through evaluation apprehension. In the high fear-of-invalidity
condition, subjects expected their judgments to be compared with those
of professional psychologists. In the low condition, subjects were told
that the evaluation method was at a pilot stage and its validity was
unknown. Primacy effects were significantly higher under high (versus
low) need for (nonspecific) closure and significantly lower under high
(versus low) need to avoid closure (fear of invalidity).

Replicating the Effects of Epistemic Motivations on Primacy Effects.

Freund et al. (1985) replicated the effects of epistemic motivations on
primacy effects in impression formation with different operational def-
initions of the needs for nonspecific closure and the avoidance of
closure. In one replication, the need for closure was operationalized via
demands for unidimensional (hence, global and undifferentiated) ver-
sus multidimensional judgments, and the need to avoid closure (fear of
invalidity) via potential costs to the target person (rather than to the
subject) of subject's judgmental mistake. In the second replication, need
for closure was manipulated as in the first experiment and need to avoid
closure via degrees of evaluation apprehension. In the third replication,
need to avoid closure was manipulated via potential costs to the target
person and need for closure via time pressure. All three designs strongly
replicated the original findings of Kruglanski and Freund (1983) in
which primary effects were consistently more pronounced under high
(versus low) need for closure and under low (versus high) need to avoid
closure.
EMPIRICAL RESEARCH 37

Anchoring Effects. Tversky and Kahneman (1974) identified persons'

tendency to anchor numerical estimates in initial values without suffi-
cient adjustments made to possible subsequent information. According
to the present analysis, "anchoring" could represent epistemic freezing
in which information processing stops after an initial estimate is made
and is somewhat (albeit insufficiently) adjusted in light of further infor-
mation. If so, anchoring should be strengthened by a heightened need
for (nonspecific) closure and weakened by a heightened need to avoid
closure. These predictions were tested by Kruglanski and Freund (1983,
Experiment 3).
In estimating the probabilities of compound (conjunctive or disjunc-
tive) events, subjects may often use as anchors the probabilities of their
component simple events (cf. Bar-Hille!, 1973). If so, the anchoring ten-
dency could lead to an overestimation of the probabilities of conjunctive
events because these are smaller than probabilities of their constituent
simple events. Similarly, anchoring could result in an underestimation of
the probabilities of disjunctive events that are larger than probabilities of
the component simple events. Following a procedure developed by Bar-
Hillel (1973), Kruglanski and Freund (1983) asked subjects to choose
within pairs of events the one more likely to occur. Some pairs con-
trasted conjunctive events with simple events and other pairs contrasted
disjunctive events with simple ones. Need for closure was manipulated
via degrees of time pressure and need to avoid closure via evaluation
apprehension (fear of invalidity). Subjects' tendency to overestimate the
likelihood of conjunctive events and to underestimate the likelihood of
disjunctive events, both indicative of anchoring, increased, as predicted,
under high (versus low) need for c10sure and decreased under high
(versus low) need to avoid closure.

Hypothesis Generation. Although our impression-formation studies

demonstrated that the extent to which persons may process external
stimulus information may vary as a function of their epistemic motiva-
tions, the anchoring study just described suggests that the same motiva-
tions would also affect the extent of processing internal information, in
this case related to mathematical concepts retrieved from memory. Our
epistemic model additionally predicts that the epistemic motivations
would also affect the extent of generating new ideas in reference to a
judgmental problem of interest. Specifically, hypothesis generation
should be less extensive under high need for closure than under high
need to avoid closure. These effects were investigated in arecent study
by Mayseless and Kruglanski (1987, Experiment 3).
Subjects were presented with unusual-Iooking photographs of parts
38 CHAPTER3

of common objects. As in previous studies, the need to avoid closure

was induced via a fear of invalidity. SpecificaIly, subjects were told that
a correct identification of objects depicted in the photographs will earn
them extra credits. The need for closure was induced by stressing the
value of definite, clear-cut opinions. A neutral condition in which no
motivational induction was attempted was also included. All subjects
were encouraged to engender as many hypotheses as possible concern-
ing the objects' identity and subsequently indicate what they thought
was the answer. As expected, subjects in the need-to-avoid-closure con-
dition generated on the average more hypotheses concerning the ob-
jects' identity than subjects in the neutral condition who, in turn, gener-
ated more hypotheses than subjects in the need-for-closure condition.

Interactive Effects of Epistemic Motivations and Initial Confidence. In the

research thus far, the need for (nonspecific) c10sure appeared to inhibit
epistemic activity, but the need to avoid c1osure, induced via a fear of
invalidity, appeared to facilitate such activity. However, such effects
may be moderated by the degree to which an individual starts out with a
fairly crystallized knowledge structure versus a relative ambiguity: A
person under high (verus low) need for c1osure, who starts out with
crystallized knowledge, already has what he or she desires, and, hence,
might be unmotivated to pursue a further informational search. Accord-
ingly, in studies described earlier, curtailment of epistemic activity has
typically occurred after some preliminary knowledge (e.g., an ethnic
stereotype, an early impression, or an initial estimate) had already been
formed.
The case in which the individual starts out with a relative ambiguity
might be different. In such circumstances, a person under high (versus
low) need for (nonspecific) closure might be considerably motivated to
conduct an informational search that might hopefully lead to the covet-
ed state of knowledge. The opposite might be observed with persons
with high (versus low) need to avoid c1osure. SpecificaIly, the individual
who starts out with a fairly crystallized knowledge, when what he or she
prefers is an absence of c1osure, might weIl be motivated to conduct a
further informational search. However, if the same person starts out
with a lack of c1osure, he or she might be unmotivated to jeopardize it by
an additional search. The above notions were investigated in a pair of
recent studies by Peri, Kruglanski, and Zakai (1986).
In one experiment, subjects were presented with several series of
drawings. Each series contained (either 2 or 4) standard drawings on a
given topic (a man, a woman, or a tree), each produced by a different
person, and a criterion drawing (invariably a house) drawn by one of the
EMPIRICAL RESEARCH 39

persons who had prepared the standard drawings. The subjects' task
was to identify for each series the particular standard drawing by the
individual who had done the criterial drawing. Before stating their final
judgment, subjects were given the opportunity to obtain sodal com-
parison information concerning chokes made by previous subjects in
the experiment. A high need for (nonspedfic) closure was created by
giving subjects no guidelines whatsoever for assessing the drawings'
similarity. We assumed (and empirically confirmed) that, in the absence
of guidelines, subjects would feel unpleasantly confused, and, hence,
would experience a high need for closure regarding the exact nature of
their task. In contrast, subjects in the low-need-for-closure condition
were provided with guiding criteria for assessing the similarity of the
drawings. They indeed found the task less confusing and more pleasant
than did their high-need-for-closure counterparts. Initial confidence in
the hypothesis was manipulated by varying the number of possible
choice alternatives. In the high-confidence condition there were two
alternatives and in the low-confidence condition, four alternatives.
The relevant results supported our predictions. In the high initial-
confidence condition, extent of information processing (indexed by the
amount of information inspected and the latency of inspecting the first
informational item) manifested by subjects under high need for closure
was lower than that manifested by subjects under low need for closure.
This trend was reversed for the low initial-confidence condition, in
which subjects under high need for closure manifested the higher extent
of information processing.
In the second experiment by Peri et al. (1986), the variable of initial
confidence was orthogonally manipulated to the need to avoid closure.
The procedure of this second study was similar to that of the previous
experiment as was the manipulation of initial confidence (by the number
of choice alternatives). Unlike the first study, however, subjects were
not required to commit themselves to any particular choke but instead
could distribute their guesses across the various choke alternatives. In-
deed, subjects in the high-need-to-avoid-closure condition were actively
discouraged from making a strong commitment to any given alternative,
for, if it turned out to be wrong, they stood to loose a11 points for that
particular trial. Subjects in the low-need-to-avoid-closure condition were
not similarly discouraged from making a commitment. Thus, the need to
avoid closure was again manipulated via degrees of the fear of
invalidity.
The results of the study supported our predictions. In the high
initial-confidence condition, subjects under high (versus low) need to
avoid closure manifested a greater extent of information processing,
40 CHAPTER3

whereas in the low initiaI-confidence condition, subjects in the high-

(versus low) need-to-avoid-closure condition manifested a lesser extent
of processing.
In sum, the findings of Peri et al. (1986) suggest that epistemic
motivations may interact with the epistemic point of departure in affect-
ing persons' extent of information processing. With a fairly confident
knowledge to start with, high (versus low) need for closure may curtall
information processing and high need to avoid closure may intensify it.
By contrast, with ambiguous knowledge to start with, high (versus low)
need for closure may intensify information processing, whereas high
(versus low) need to avoid closure may suppress it.

(2) Type of Information Processed

The impact of epistemic motivations on extent of information pro-
cessing mayaiso affect indirectly the type of information processed.
SpecificaIly, under a heightened need for (nonspecific) closure, the indi-
vidual may be less sensitive to possible alternatives to a given hypoth-
esis. Hence, he or she may search for information that is relevant ex-
clusively to that particular hypothesis. The case may be different under a
heightened need to avoid closure where the individual is assumed to be
rather sensitive to alternative hypotheses. Hence, his or her information
search might be directed toward items that are diagnostic with respect to
the hypothesis and its alternative(s), and in this sense are relevant to
both. In a recently completed study that bears on those notions,
Kruglanski and Mayseless (1988) had subjects test the hypothesis that an
interviewer is a painter. In an offhand manner, subjects were also told
that the target could, of course, be a member of a different profession;
for Instance, he or she could be an architect. Need for closure was manip-
ulated by informing the subjects that the experiment assessed persons'
ability to make quick decisions and need to avoid closure by stressing
the value of accuracy. In addition, a neutral control condition was in-
cluded where no motivational induction was attempted. Subjects' task
was to select questions to be included in the interview from a list that
included items about prototypical features of painters' (e.g., Bohemian
life-style, artistic ability), as weIl as items about features diagnostic with
respect to painters versus architects (e.g., Bohemian life-style, mathe-
matical ability). The results confirmed the predictions derived from the
lay epistemic model. Under the need for (nonspecific) closure, subjects
predominantly selected items about prototypical features of painters
and showed only a weak tendency to select diagnostic questions rele-
vant to the hypothesis and the alternative. The tendency to select diag-
EMPIRICAL RESEARCH 41

nostic questions was more pronounced under the motivationally neutral

condition and was most pronounced under the need to avoid cIosure.

(3) Subjective Confidence

Besides their effects on the extent of information processing and the
type of information processed, epistemic motivations may influence per-
sons' subjective confidence in their hypotheses. Oue to the discounting
tendency (cf. Kelley, 1971), an individual who considers multiple com-
peting hypotheses may have lower confidence in each hypothesis than
an individual who considers fewer hypotheses or only a single hypoth-
esis. According to the lay epistemic model, an individual under high
need for cIosure may consider fewer hypotheses than one under a high
need to avoid cIosure. It follows that in the former (versus the latter)
condition the individual should experience higher subjective confi-
dence. The above notions were tested in aseries of studies by Mayseless
and Kruglanski (1987a).
Subjects in an experiment performed a tachistoscopic task of identi-
fying barely visible digits flashed on the screen. The need to avoid
cIosure was manipulated via a fear of invalidity. SpecificaIly, subjects
were promised extra credits for correct1y identifying 9 of the 10 digits.
The need for cIosure was manipulated through verbal instructions
stressing the importance of forming unambiguous, cIear-cut opinions,
an ability said to correlate with "high mental concentration and intelli-
gence." In addition, a neutral condition was incIuded in which no moti-
vational induction was attempted.
As predicted, subjects' confidence in their initial hypotheses was
highest in the need-for-cIosure condition, lowest in the need-to-avoid-
cIosure condition, with the neutral control condition falling in between
these two. Conversely, subjects' extent of informational search (Le., the
number of times they operated the tachistoscope) was highest in the
need-to-avoid-cIosure condition, lowest in the need-for-closure condi-
tion, with the neutral condition occupying, again, the middle ground.
Furthermore, to the extent that subjects under need for closure
more so than subjects under the need to avoid closure may interpret
each tachistoscopic flash in relatively unambiguous terms, the impact of
such stimulus presentation on their confidence should be greater. When
the new stimulus appeared to contradict subjects' previous hypothesis,
such confidence shift could be negative, whereas when it appeared to
support subjects' previous hypothesis, such confidence shift would be
positive. Indeed, we find that subjects' magnitude of confidence shifts
(positive as weIl as negative) occasioned by each successive stimulus
42 CHAPTER3

presentation was highest in the need-for-dosure condition, lowest in the

need-to-avoid-dosure condition, with the neutral condition falling in the
middle. These findings support the lay epistemie analysis.

The Context of Knowledge UtiIization

Our results thus far suggest that the epistemic needs for dos ure and
for the avoidance of dosure exert a range of interrelated effects in the
knowledge-formation context. They seem to influence the extent of in-
formation processing prior to dosure, the type of information attended
to, as weIl as the subjective confidence in one's hypotheses. But beyond
their impact on the formation of knowledge, the epistemie motivations
are expected to play'an equally important role in the context of knowl-
edge utiIization. This context should be of particular relevance to sodal
psychologists because it relates to the impact of such previously stored
knowledge structures as attitudes, prejudices, or stereotypes on current
perceptions, judgments, and behaviors. EssentiaIly, in forming a judg-
ment or dedding a course of action, the individual may consult two
sources of "wisdom": preexisting knowledge in long-term memory and
current stimulus information about the case. An interesting situation
may arise when these two sources have conflicting implications for cur-
rent judgment. The way the conflict is resolved may depend on the
person's epistemic motivations.
SpedfieaIly, under a high need for closure, persons may be disin-
clined to confront their preconceptions with new evidence. Rather, they
might be disposed to follow those preconceptions uncritically and in this
sense be "theory driven" rather than "data driven." By contrast, under a
high need to avoid closure, persons might be more prepared to consider
current evidence that may be inconsistent with their prior opinions and
in this sense be more data than theory driven. Support for these notions
comes from several recent studies described below.

(1) Stereotype-Based Judgments

Kruglanski and Freund (1983, Experiment 2) made use of two prev-
alent ethnic stereotypes in Israel: Sephardic Jews and Ashkenazic Jews.
The Sephardic Jews come predominantly from the Middle East and
North Africa and their stereotypie image in domains of academic perfor-
mance and achievement is generally negative. By contrast, the Ash-
kenazie Jews come mostly from Europe and/or North America and their
corresponding image or stereotype is generally positive.
Students in a graduating dass of an Israeli teachers college assigned
a grade to a literary composition ostensibly written by an 8th grader. All
EMPIRICAL RESEARCH 43

subjects responded to exactly the same composition, whieh differed

only in the alleged writer's name. In one condition, the name (Blu-
menthal) suggested an Ashkenazie origin whereas in another condition
the name (Aboutbull) suggested a Sephardie origin. The need for closure
was manipulated via degrees of time pressure, and the need to avoid
closure via degrees of evaluation apprehension.
In general, existing ethnie stereotypes exerted substantial effects of
grading. In all but one experimental condition grades assigned the Ash-
kenazie writer were considerably higher on the average than those as-
signed the Sephardie writer. This stereotypie effect disappeared almost
entirely, however, in the experimental cell that combined a low need for
closure with a high need to avoid closure. Thus, it appears that individu-
als' tendency to base their judgments on preexisting knowledge may
depend on their epistemie motivations. In particular, stereotype driven
judgments may be more likely to occur under a high need for (non-
preferential) closure or a low one to avoid closure, and less likely to
occur under a low need for closure and a high one to avoid closure.

(2) Biases in Social Differentiation: From Prejudice to Discrimination

The study just described addressed the stereotypieally biased eval-
uation of a single target. However, a similar process may enter into the
inappropriate discrimination among several targets based on the per-
ceivers' prejudices about the sodal categories of which those targets are
members. In arecent study pertinent to those ideas, Bechtold, Zanna,
and Naccarato (1985) first measured subjects' prejudiee toward women
in management and subsequently (months later) had them rank re-
sumes of men and women on their attractiveness for a given managerial
position. Based on pilot work with a similar subject population it had
been ascertained that the two types of resumes, did not systematically
differ in attractiveness. A very different judgment was reached by the
prejudieed subjects when their need for closure was experimentally
aroused by time pressure. Under those drcumstances, prejudiced per-
sons rated the men' s resurnes as signifieantly more attractive than the
women's, exhibiting considerable discrimination in line with their preju-
didal attitudes. Virtually no sex-typed discrimination, however, was
manifested by prejudieed subjects in the absence of time pressure. Fi-
nally, nonprejudiced subjects did not significantly differ in their rank-
ings of men's and women's resurnes as a function of time pressure.
Those data are consistent with the notion that the need for closure may
affect the degree to which prejudiee may lead to discrimination, or,
more generally, the degree to whieh spedfie judgments are dietated by
preexisting knowledge structures.
44 CHAPTER3

(3) Retrieval of Global Impression or Specific Information

Studies reviewed thus far juxtaposed "internai" knowledge struc-
tures retrieved from long-term memory to "external" stimulus informa-
tion. But our motivational analysis could be relevant to another central
concern of social cognition research, notably, the question of what is
retrieved. Previous research (cf. Hamilton, 1981) suggested that global
impressions are often retrieved more readily than the specific stimulus
information on which they were based. However, with some effort the
specific information, too, may be retrieved. In turn, persons' inclination
to put in the required effort may depend on their epistemic motivations.
SpecificaIly, under a high need for closure, the person may base his or
her judgments on the more readily retrievable global impressions,
whereas under a high need to avoid closure, he or she may be willing to
expend the effort and search for relevant specific information as weIl.
These notions were examined in arecent study by Sanbomatsu and
Fazio (1986).
Subjects were given information about two department stores: the
Smith department store and the Brown department store. The informa-
tion was such that it generally depicted the Smith department store as
superior in most relevant respects to the Brown department store. But
there was one exception. The camera department was portrayed as
clearly superior in the Brown store as compared to the Smith store. This
was conveyed by three items of information interspersed throughout
the entire informational sequence. Subjects viewed the information un-
der an "impression formation" set, and, as expected, they generally
formed a more positive impression of the Smith than of the Brown
store.
Now came the critical task. Subjects were asked to decide in which
of the stores they would decide to purchase a camera if they needed one.
This task was performed under one of four conditions defined by a 2 x 2
factorial design in which two degrees of the need for closure, manipu-
lated by time pressure, were orthogonally crossed with two degrees of
the need to avoid closure manipulated by evaluation apprehension. The
results were clear-cut. Specifically, subjects' tendency to erroneously
choose the Smith over the Brown camera department was stronger un-
der high (versus low) need for closure and was weaker under high
(versus low) need to avoid closure. Thus, the tendency to base one's
judgments on the readily retrievable global information rather than on
the less readily retrievable specific information seems to be positively
related to the need for closure and negatively related to the need to
avoid closure.
EMPIRICAL RESEARCH 45

THE CONTEXT OF SOCIAL BEHAVIOR

To social psychologists the epistemic process becomes of particular

interest where it appears to mediate socially relevant behavior, or to
guide social interaction. Studies examined below explored, therefore,
some derivations of the lay epistemic model conceming various behav-
ioral effects of social significance.

(1) Behavioral Consequences of Disconfirmed Expectancies

Persons' expectancies about significant others, their own selves, or
the course of impending social events may offer important guides for
decision making and action. Therefore, a disconfirmation of an expec-
tancy may occasion considerable upset and prompt behavioral efforts to
reaffirm the expectancy (Aronson, 1968). This also may be one of the
reasons why persons may often strive to confirm their hypotheses rather
than test them in an impartial manner (Snyder and Swann, 1978). Ac-
cording to the present epistemic analysis, however, persons' reactions
to disconfirmed expectancies may be moderated by their epistemic
motivations. As already implied, a disconfirmed expectancy may und er-
mine a person's sense of knowledge. Thus, if persons' need for (non-
specific) closure in a given domain is high and/or if their need to avoid
closure is low, they may attempt to reaffirm the expectancy more than if
the need for closure is low and the need to avoid closure is high. Fur-
thermore, if the disconfirmation is congruent with a person's need for
specific closure, this individual should not strive as strongly to reaffirm
the expectancy as in the case where the disconfirmation was not so
congruent. These notions were tested in arecent study conducted at Tel
Aviv University.
Based on the procedure developed by Aronson and Carlsmith
(1962), subjects experienced aseries of failures followed by a success.
Specifically, they viewed sets of three drawings and identified the one
drawing in the set that belonged to a person with some unique property
(either a mentally disturbed or an artistically gifted individual). On the
first 4 se ries of 10 trials, subjects received failure feedback of 3,2,2, and
3 correct identifications, respectively, out of a possible 10. On the fifth
series of trials, subjects received success feedback, that is, they made
correct identifications on 8 out of the 10 trials.
Half of the subjects were placed in a combined high-need-for-
closure/low-need-to-avoid-closure condition, whereas the remaining
half were in a combined low-need-for-closure/high-need-to-avoid-
closure condition. Furthermore, half the subjects in each condition were
46 CHAPTER3

led to regard success on the task as highly desirable, whereas the re-
maining half regarded it as less desirable.' Thus, for the former (versus
the latter) subjects judgment that they are successful was assumed to
represent to a greater extent a specific, desired, closure.
The experimental manipulations were accomplished as follows. All
subjects expected to fill out a questionnaire concerning their perceived
ability to make the requisite indentifications. In the high-need-for-
closure/low-need-to-avoid-closure condition, subjects expected to have
limited time in which to respond to the questionnaire, and were not
given any specific instructions concerning accuracy. In the high-need-to-
avoid-closure/low-need-for-closure condition, subjects were urged to
reach the most accurate assessments possible and expected to have un-
limited time at their disposal.
In the high-success-desirability condition, subjects were told that
the task consisted of identifying drawings prepared by mentally dis-
turbed individuals. It was expected that, for psychology students, who
served as our subjects, success on such a task would appear highly
desirable. By contrast, subjects in the low-success-desirability condition
were told that the task consisted of identifying the drawings of artis-
tically gifted persons. Success on such a task was not expected to appear
particularly important or desirable to psychology students.
Under the pretext that the experimenter had mistakenly failed to
time the fifth series of trials, this series was repeated and the number of
answers that the subjects changed was recorded. It was found that the
numberof changes was significantly higher in the high-need-for-
closure/low-need-to-avoid/closure condition than in the low need-for-
closure/high-need-to-avoid-closure condition. Furtherrnore, the number
of changes was significantly lower with high (versus low) success de-
sirability. Thus, it appears that the degree to which a disconfirmation of
an expectancy is rejected and the expectancy is behaviorally reconfirmed
may depend on persons' episternic motivations with which the discon-
firmed expectancy may or may not be incongruous.

(2) Comparisons with Agreeing and Disagreeing Others

The present analysis has implications for conditions under which
persons might wish to compare with others whose opinions are similar
or dissimilar to their own. According to Festinger's (1954) theory of
sodal comparison processes, similarly minded others are often preferred
for comparison purposes over dissimilar others. This general prediction
may be qualified by motivational considerations suggested by the lay
epistemic theory. In particular, a comparison preference for similarly
minded others may be manifest when the individual is high on a need
EMPIRICAL RESEARCH 47

for nonspecific closure and/or a need for specific closure with which a
current opinion is congruent. However, the opposite preference, that is,
for dissimilarly minded others, might be manifest when the individual is
under a high need to avoid closure and/or a high need for specific
closure with which a current opinion is incongruent. For instance, a
person who just heard that the value of his or her real estate investment
is about to depreciate may seek the company of people who believe the
contrary to be the case.
These notions were recently tested in aseries of studies by
Kruglanski and Mayseless (1987). In one study, subjects whose fear of
invalidity was experimentally aroused by an induced expectation of a
monetary prize for a correct answer chose to compare with dissimilarly
minded others. By contrast, subjects whose need for self-confirmation
(or defensiveness) was aroused, hence, lending their opinion the status
of a specific preferred closure, chose to compare more with similarly
minded others. FinaIly, in a neutral control condition, subjects com-
pared to an equal degree with others whose opinions were dissimilar
versus similar to their own.
Two further studies tested the combined effects on choice of the
other being compared with the need for (nonspecific) closure and the
need to avoid closure. The experiments differed considerably in the
tasks employed. In one study, the task involved the resolution of a
cognitive inconsistency, whereas in the other study it involved percep-
tual judgments of relative length. Furthermore, in one study, the need
to avoid (or postpone) closure was manipulated by an expectation of a
prize (a candy bar) for an accurate answer, whereas in the other study it
was manipulated by degrees of evaluation apprehension. Also, in both
studies the need for closure was manipulated in various ways via de-
grees of time pressure. The results of the two studies were highly simi-
lar. In both cases subjects under high (versus low) need for closure
tended to choose for comparison those whose opinions were similar to
their own. By contrast, under high (versus low) need to avoid closure,
subjects preferred to compare with dissimilar others (for details, see
Kruglanski and Mayseless, 1987).
In sum, it appears that the epistemic motivations for closure and the
avoidance of closure exert a wide variety of effects in contexts of (1)
knowledge formation, (2) knowledge utilization, and (3) socially rele-
vant action. In the context of knowledge formation, those motivations
appear to affect the extent of information processing and hypothesis
generation, the type of information processed, as weIl as the subjective
confidence in the judgments reached. In the context of knowledge utiliz-
ation, these same motivations may affect the tendency to base judg-
ments on preexisting knowledge structures (hence, being theory driven)
48 CHAPTER3

rather than on the case information at hand (hence, being data driven).
Furthermore, these motivations may affect subjects' tendency to base
judgments on the more readily retrievable global impressions versus the
less readily retrievable stimulus information serving as the basis for
those impressions. In the context of socially relevant action, epistemic
motivations were shown to affect the tendency to behaviorally reaffirm
disconfirmed expectancies and to moderate the tendency to choose sim-
ilarly versus dissimilarly minded others as objects of comparison.
It will be noted that the various studies reviewed in the foregoing
sections manipulated the epistemic motivations in a variety of ways, yet
the results consistently supported the predictions of the lay epistemic
model. Thus, even if any one study was susceptible to alternative inter-
pretations of one sort or another, it is unlikely that any such interpreta-
tion would be comprehensive enough to account for the wide range of
effects that the epistemic analysis predicted.
Furthermore, it is noteworthy that many of the effects described
above contain aspects that are characteristically motivational, that is, are
not readily interpretable in cognitive terms. For instance, extent of infor-
mation processing is akin to the degree of behavioral persistence, a typ-
ically motivational phenomenon. Similarly, choice of similar or dissimilar
others for comparison and the apparent investment of effort in retrieval
of original stimulus information are characteristic motivational effects
addressed by major theories of motivation (Weiner, 1980).
Other aspects of the present findngs are more clearly cognitive in
nature, for example, the degree of subjective confidence in one's hy-
potheses or the tendency to base judgments on preexisting knowledge
structures. Together, the motivational and the cognitive aspects of the
present findings confirm that we have manipulated motivational vari-
ables of a special kind, namely those whose goals are uniquely tied to
the epistemic process.

THE LOGICAL ELEMENT IN EPISTEMIC BEHAVIOR

The bulk of research in the lay epistemic framework has focused on

the effects of epistemic motivations. However, the remaining aspects of
the model have researchable implications as well. In the following sec-
tion, some studies are reviewed about the logical aspect ofepistemic
behavior.

Cognitive Balance Effects

Consider Heider's (1958) balance theory. According to the present
interpretation (discussed more fuHy in Chapter 5), the classical triad "p
EMPIRICAL RESEARCH 49

likes 0", "p likes x," and "0 does not like x" is imbalanced because it
contains a logical contradiction. The sentence "p likes 0" may imply to
many people that p "approves of" or "agrees with" o's preferences,
attitudes, and opinions. Thus, to find out that p likes 0 yet disapproves
of o's preferences regarding x contains a contradiction that could foster
the experience of unpleasantness, the judgment of unreasonableness,
and a sense of instability, all predicted by balance theory.
However, not all positive statements may imply approval or agree-
ment. And when they do not, imbalance may not result in upset. These
notions were tested in a study by Klar and Pol (1980) performed at Tel
Aviv University. In the first research phase, native speakers of English
judged the extent to which different sentiment terms implied agree-
ment, that is, the extent to which a given sentiment feIt by p toward 0,
implied that p and 0 would share the same sentiment toward x. Exam-
pIes of sentiments rated as having strong agreement implication are
"approves of," "venerates," or "is proud of." Sentiments rated to have
medium agreement implication are "likes," "cherishes," "is impressed
by," and those with weak agreement implications are "fancies," "longs
for," or "is charmed by."
In the second phase, a different group of subjects was exposed to
balanced and imbalanced triads with sentiment terms having respec-
tively strong, medium, and weak implications for agreement. The vari-
ous sentiment terms employed in the study were matched on their
perceived intensity such that sentiments in the three implication catego-
ries did not systematically differ in their apparent strength. The depen-
dent variables used by Klar and Pol were those employed traditionally in
balance theoretic research: (1) rated pleasantness of the situation de-
picted, (2) its rated reasonableness, and (3) its rated stability. As ex-
pected, significant implication by balance interactions appeared on all
the dependent measures. More specifically, balance effects (i.e., differ-
ences between balanced and imbalanced triads) were strongest with
sentiments in the strong implication category, weaker with sentiments
in the medium implication category, and weakest with those in the low
implication category. These findings support the notion that the effects
of cognitive inconsistency, in this case, the effects of cognitive im-
balance, are mediated by the epistemic consequences of logical contra-
diction.

INCONSISTENCY RESOLUTION VIA INFORMATIONAL SEARCH

The present analysis has implications for ways of resolving a cogni-

tive inconsistency: A logical contradiction can be resolved by denying or
rejecting one of its components. By definition, A and not-A cannot both
be true. One of them has to be false. If an inconsistency is to be resolved,
50 CHAPI'ER3

one of the contradictory cognitions must therefore be rejected or denied

(see Chapter 5 for details). This analysis also points to the likely direc-
tion of the rejection or denial. The cognition more likely to be denied is
that which is less dearly "undeniable" or is held with the lesser
confidence.
But what about the case where both inconsistent cognitions are held
with approximately equal confidence? In such a situation, the individual
might be at a loss as to which of the incompatible cognitions is to be
denied. Nonetheless, if motivated to reduce the ambiguity, the individ-
ual might conduct an informational search hopefully suggesting a way
to resolve or explain the inconsistency (Hastie and Kumar, 1979). Thus,
the epistemic model predicts that individuals whose confidence is dis-
tributed evenly over the two contradictory cognitions would be more
likely to engage in an informational search than individuals whose confi-
dence is distributed unevenly.
The above ideas were supported by findings of the Kruglanski and
Mayseless (1987) research on sodal comparison processes. In one experi-
ment (Study 3), some subjects were given ample evidence supporting a
target hypothesis, creating high initial confidence. Other subjects were
given considerably less supportive information, creating low initial con-
fidence. All subjects were then presented with an item of information
that contradicted the target hypothesis. The main dependent variable
was subjects' tendency to seek the opinions of others before reaching a
judgment. The results were dear-cut: 65% of subjects in the high-
confidence condition and only 30% of subjects in the low-confidence
condition expressed the wish to consult with others. In accordance with
the present analysis, it appears that when subjects have low confidence
in a proposition, they have less difficulty resolving the contradiction
between it and another (credible) statement (presumably by rejecting
the less confidently held cognition); hence, they have less of a need to
seek the opinions of others than when they have high confidence in
both contradictory propositions.

HYPOTHESIS TESTING

The notion that lay persons test their hypotheses deductively has
implications concerning the question whether they predominantly em-
ploy confirmatory (Snyder and Swann, 1978) or diagnostic (Trope and
Basok, 1983) strategies of searching for information. A confirmatory strat-
egy refers to a search for features that are highly probable or prototypical
under a given hypothesis (e.g., a search for instances of extrovert be-
havior when testing the hypothesis that someone is an extrovert). In
contrast, a diagnostic strategy refers to a search for features that best
EMPIRICAL RESEARCH 51

discriminate between the hypothesis and some alternative (e.g., search-

ing for behaviors characteristic of a politician but not an actor when
testing the alternative hypotheses that a target is a politician or an actor,
but not both).
According to the present analysis, whether persons seek prototypi-
cal ("confirmatory") or diagnostic information should depend on the
implicational linkages that, in their cognitive structure, connect a given
type of information with a given hypothesis of interest. Among other
factors, this could depend on whether they considered the target hy-
pothesis in isolation or in the context of a competing hypothesis (or
hypotheses). For instance, when testing the hypothesis that a given
celebrity is a politician, the person may look for information about be-
haviors characteristic of a politician (e.g., public appearances, interest in
affairs of state). However, when testing the competing hypothesis that
the celebrity is a politician or a military commander, the person may
look, instead, for information about diagnostic features that may help
discriminate between the two alternatives (e.g., interest in weapons or
in military tactics).
The above notions were tested in a recently conducted study by
Kruglanski and Mayseless (1988). Subjects tested the single hypothesis
that a given target is an architect, the single hypothesis that the target is
a painter, or the competing hypotheses that the target is an architect or a
painter. As predicted, in the single-hypothesis conditions, subjects ex-
hibited strong preferences for prototypical (versus nonprototypical) in-
formation. Subjects testing the architect hypothesis preferred questions
about features prototypical of an architect over those about non-
prototypical features. Similarly, subjects testing the painter hypothesis
preferred questions about features prototypical of a painter over ques-
tions about less prototypical features. Finally, subjects testing the paint-
er versus architect hypotheses did not exhibit preferences for prototypi-
cal features of architects or painters. Instead, they exhibited a significant
preference for diagnostic (versus nondiagnostic) features demarcating a
painter from an architect.
Thus, it appears that the information that persons seek in order to
test their hypotheses depends in part on the informational context in
which a given hypothesis is evaluated. This may depend on the implica-
tional representations that, for a given individual, render specific item(s)
of information relevant to the hypothesis (or hypotheses) being tested. It
is because of such considerations of implicational or deductive relevance
(Kruglanski, 1980) that prototypical information may be sought after
when a hypothesis is tested singly, whereas diagnostic information may
be preferred when a hypothesis is tested against definite competing
alternatives.
52 CHAPTER3

CONTENT-BoUND RESEARCH IN THE LAY EPISTEMIC FRAMEWORK

The lay epistemic model suggests a two-pronged approach to sodal

cognitive research. On the one hand, a general model of the judgmental
process is proposed and its relevance to a variety of sodal psychological
problems is explored. All the preceding research belongs in that catego-
ry, induding studies that applied the model to social comparison phe-
nomena, ethnic stereotyping, the prejudice-discrimination relation, or
sodal hypothesis testing. Such a broad range of implications derives
from the assumption that the same epistemic or judgmental process
applies to all contents of knowledge. Thus, the general principles sug-
gested in the lay epistemic model should apply to seemingly uncon-
nected social psychological phenomena whereas differences should
stern from the separate content domains that they represent.
In addition, however, the present approach strongly implies that
content-bound research, focused on contents of given knowledge struc-
tures, is of considerable value for predicting persons' responses in par-
ticular situations. Rather than confounding separate levels of analysis
and mistaking special categories of judgments for the judgmental pro-
cess, it is important to study separately particular contents of knowledge
structures of spedal consequence or utility.
Arecent, unpublished study by Muzfir (1985) attempted to "uncon-
found" judgmental process and contents in reference to the relation
between attitude extremity and resistance to change. According to some
theories (in particular, Osgood and Tannenbaum, 1955), an attitude's
resistance to change is positively related to its extremity. One explana-
tion for such a correlation may be that extreme attitudes are held with
greater confidence than less extreme ones (Insko, 1967; Kelley and
Lamb, 1957). However, although attitudinal extremity and confidence
may often be correlated, the two variables are conceptually separate.
Extremity refers to the attitude's content, as a more-extreme attitude
differs in its content from a less-extreme attitude. By contrast, confi-
dence is a parameter of the judgmental process: One could be more or
less confident in propositions of all possible contents induding more- or
less-extreme attitudes. Furthermore, resistance to change should de-
pend on one' s confidence in a proposition and not on the proposition' s
content.
Muzfir's (1985) experiment employed a 2 x 2 design in which atti-
tude extremity (high and low) was manipulated orthogonally to attitudi-
nal confidence (high versus low). Subjects in the low-extremity condi-
tion were presented a vignette depicting a mild crime (pickpocketing),
whereas those in the high-extremity condition were presented a vignette
depicting a severe crime (rape and murder). Subjects in all conditions
were presented with a sketch of a suspect as he appeared in court and
EMPIRICAL RESEARCH . 53

with a composite drawing of the criminal based on a witness' s descrip-

tion. In the high-confidence condition, the composite drawing was high-
ly similar to the sketch and in the low-confidence condition it was some-
what dissimilar.
The initial attitude measure concerned subjects' punitivity toward
the defendant. As expected, subjects in the high-extremity condition
recommended higher punishment on the average than subjects in the
low-extremity condition. No difference in punitivity separated the high-
and low-confidence groups. Also as expected, subjects in the high-
confidence condition were significantly more confident in the recom-
mended punishment than subjects in the low-confidence condition,
whereas no differences in confidence separated the high- and low-
extremity groups.
To induce attitude change, subjects in a11 groups were presented
with evidence inconsistent with the defendant's culpability, notablyan-
other composite drawing portraying a face highly unlike the defendant's
and presumably based on a second witness's testimony. Subjects were
then asked to make a final recommendation as to the appropriate pun-
ishment. The degree of change from the first to the second recommen-
dation served as a measure of attitude change. As predicted, subjects in
the high-confidence condition exhibited lesser degree of change than
those in the low-confidence condition, whereas subjects in the high-
versus low-extremity conditions did not significantly differ in the
amount of attitude change. Thus, it seems that an attitude's resistance to
change depends on a judgmental process element, notably confidence,
rather than on a judgmental content element, notablyextremity.
According to the lay epistemic perspective, attitudes refer to a par-
ticular content domain of knowledge structures (see Chapter 6, this
volume) that specifically deals with evaluative statements. Attitude for-
mation and change, therefore, should be affected by the same factors
that govern the formation and change of innumerable alternative knowl-
edge structures unrelated to evaluation. Subjective confidence in a
knowledge structure, as studied by Muzfir (1985), is one such prominent
factor.

SELF-AsCRIBED EPISTEMIC AUTHORITY

Muzfir's research demonstrates the utility of drawing a dear distinc-

tion between epistemic contents and processes, although it does not in
itself address epistemic contents per se. Yet the specific contents of per-
sons' knowledge structures should importantly affect behavior and per-
formance in various situations. Arecent dissertation by EIlis (1984) illus-
trates a possible direction that such content-bound research could take.
The construct of epistemic authority (Kruglanski and Jaffe, 1988)
54 CHAPTER3

relates to a classical topic in the sodal psychology of communication,

notably the characteristics of the communication source, like expertise
and trustworthiness. Spedfically, epistemic authority refers to the de-
gree to which a given source is likely to be considered authoritative on a
given topic. Eilis (1984) assumed that individuals may differ in their
beliefs concerning their own epistemic authority in given domains and
that those beliefs may affect the persons' reactions to various sodal
influence situations.
More spedfically, Eilis (1984) hypothesized that an individual's be-
liefs concerning his or her own epistemic authority in a domain may
moderate the benefits to him or her of experientiallearning versus a frontal
lecture on a topic delivered by another individual. In an experiential
learning situation, persons interpret the meaning of their experiences by
themselves. Those with low self-ascribed epistemic authority may have
little confidence in their interpretations. For instance, a person with little
perceived self-authority in the domain of cars may have little confidence
in his or her impressions after test-driving a new vehicle at a car dealer-
ship. Consequently, low perceived self-authority individuals may have
difficulty in deriving confident knowledge from personal experiences
and may benefit from them less than from learning situations in which
the same topic is interpreted by an external authority. In contrast, indi-
viduals whose self-ascribed epistemic authority is high may derive
greater benefits from experiental (versus externally interpreted) learning
situations.
To test the above ideas, Eilis constructed questionnaires of subjects'
own perceived epistemic authority in two separate domains, mathema-
tics and interpersonal skills. In each case, the questionnaires tapped the
degree to which subjects trusted their own ability to make sense of and
confidently interpret information in the domain at issue. In course of the
research, subjects who scored above or below the median on the mea-
sure of their own epistemic authority were exposed to experientiallearn-
ing or a frontallecture delivered by an external authority on a relevant
topic. Thus, in the mathematical study, high or low scorers on the mea-
sure of their own epistemic authority in mathematics were assigned to
experientiallearning or frontallecture conditions designed to teach the
same set of mathematical prindples. Similarly, in the interpersonal skills
study, persons high or low on their own epistemic authority in the
interpersonal domain were assigned to experiential learning or frontal
lecture conditions designed to teach a set of emphatic skills.
When subsequently tested, the high (self-ascribed) epistemic au-
thority individuals who had been previously exposed to experiential
learning (of mathematics or empathy) did consistently better than their
counterparts in the frontallecture condition. In contrast, in the mathe-
EMPIRICAL RESEARCH 55

matics study those low in their own epistemic authority did about equal-
ly well in the two learning situations, whereas in the interpersonal skills
study the lows actually did better in the frontallecture condition. Thus,
the contents of subjects' beliefs about their own authority affected their
reactions to various knowledge acquisition situations.

SUMMARY AND CONCLUSIONS

This chapter reviewed several categories of recent empirical re-

search guided by hypotheses derived from the lay epistemic model. The
research reviewed included studies concerning the motivational under-
pinnings of judgmental phenomena, as well as studies on aspects of the
subjective logic of hypothesis testing and on the spedfic psychological
implications of given epistemic contents.
Research on epistemic motivations supported the present theoreti-
cal predictions concerning a wide range of effects in contexts of knowl-
edge formation, knowledge utilization, and of sodally relevant action.
Research on the deductive logic of hypothesis testing addressed the
cognitive consequences of logical contradiction and the conditions un-
der which it may be more versus less readily resolvable. Further re-
search suggested that it is judgmental process (confidence with which a
given hypothesis is held) rather than content (e.g., attitude extremity)
that affects cognitions' resistance to change. Furthermore, exploring the
contents of particular belief clusters (e.g., concerningpersons' own epis-
temic authority in a domain) was shown to be of value in predicting
individuals' psychological reactions (e.g., learning) in relevant situ-
ations.
In general, research examined in this chapter illustrates the epis-
temic model' s breadth of applicability. The topics studied included phe-
nomena as diverse as impression formation, sodal comparison, ster-
eotyping, learning, and attitudes. These findings are thus compatible
with a major tenet of the lay epistemic approach whereby diverse topics
of sodal psychological inquiry may differ in the contents of lay knowl-
edge to which they pertain, yet may be similarly affected by parameters
of the knowledge-acquisition process such as its underlying logic or its
motivational basis. The integrative character of the lay epistemic frame-
work will now be further demonstrated through the identification of
commonalities among various, heretofore separate, areas of sodal cogni-
tive research, starting with attribution theory.
 4

Unique and Nonunique Aspects of

Attribution

In a review chapter published several years ago, Kelley and Michela

(1980) noted that the available attributional notions are piecemeal and
greatly in need of synthesis. The present chapter attempts such a syn-
thesis, based on the present epistemic theory. In many ways this work
continues my earlier attempts to integrate the field of attribution (Krug-
lanski, 1980; Kruglanski, Hamei, Maides, and Schwartz, 1978) yet it
differs from those initial endeavors in two principal respects: (1) by
taking into account relevant research findings pubIished since and (2) by
explicitly incorporating the motivational dimension of epistemic be-
havior barely mentioned in those earlier analyses.
The central thesis explored in the following sections is that attribu-
tional activity represents a special case of the general epistemic or judg-
mental process. SpecificaHy, attributions can be thought of as judgments
whose contents are causal. Thus, to the extent that the general judgmen-
tal or epistemic process applies to judgments of aH contents, by defini-
tion it applies to attributions as weH. In other words, attributions may
differ in contents only from possible alternative judgments and may
partake in the same process whereby all judgments are rendered and
modified. In this sense, the present analysis stresses the nonunique as-
pects of attribution, those which it shares in common with aH judgment.
This chapter substantiates these notions by considering, in turn,
three major parameters of attributional activity, related to its logical,
cognitive and motivational aspects.

THE LOGICAL ASPECT OF ATTRIBUTION

As elaborated in Chapter 2, the validation of alllay (orscientific)

judgments may be accomplished by logical derivation. According to the
57
58 CHAPTER4

present theory, the knower departs from an implicational (if then) link
assumed to exist between a given X and a given Y and proceeds to infer
Y upon noting X. A central concern of attribution theory is with the
process whereby causal hypotheses are validated; it is therefore of inter-
est to see whether such validation is essentially depicted as logical in
nature.
A particularly lucid portrayal of the process whereby causal hypoth-
eses are validated is given in Kelley's (1967) ANOVA model of attribu-
tion. Let us therefore revisit that particular model and examine closely
the proposed validational mechanism it describes.

KELLEY'S ANOVA MODEL

According to Kelley (1967), confident causal ascription of some

effect to an external entity is made if the following four criteria are
fulfilled:
(1) Distinctiveness: The impression is attributed to the thing if it uniquely
occurs when the thing is present and does not occur in its absence; (2)
Consistency over time: Each time the thing is present, the individual's reaction
must be the same or nearly so; (3) Consistency over modality: His reaction must
be consistent even though his mode of interaction with this thing varies; (4)
Consensus: Attributes of extemal origin are experienced the same way by all
observers. (p. 197)

To realize the deeper significance of these criteria examine them

more closely. Consider first the distinctiveness criterion. Essentially it
asserts that the effect (the impression, in Kelley's terms) covaries with
the entity: It is present when the entity is present and is absent other-
wise. Consider now consistency over time. What it signifies is that the
effect does not covary with any particular time. Instead, it is uniformly
present across times, when the entity is present, or is uniformly absent
across times, when the entity is absent. In an identical manner con-
sistency over modality signifies that the effect does not covary with a
particular modality, and consensus, that it does not covary with any
particular person.
Generalizing from the above analysis we see that Kelley's attribu-
tional criteria embody the principle of covariation as the mediator of
causal inferences: The category with which the effect covaries (in this
example, the external entity) is recognized as its cause, whereas catego-
ries with which the effect does not covary (in this example, the time, the
modality, or the particular person) are excluded as possible causes. This
interpretation raises two issues of interest: (1) Why should covariation
evidence mediate ascriptions of causality? (2) With covariation as the
underlying principle, what is the remaining significance of the attribu-
tional criteria? Let us consider these issues in turn.
UNIQUE AND NONUNIQUE ASPECTS 59

Covariation as Evidence for Causation

The reason why covariation evidence may mediate causal attribu-
tions is straightforward enough. As commonly understood, causality
denotes two major properties: Temporal precedence of the cause relative
to the effect, and covariation of the two such that given a presence of the
cause, the effect, too, is present and given an absence of the cause, the
effect is absent. More formaHy this me ans that the cause is conceived of
as both a sufficient and a necessary condition for the effect.
These commonly accepted connotations of causality render covaria-
tion as weH as temporal-ordering information relevant to casual in-
ferences. Covariation (or temporal-ordering) information is evidence for
causality because most people probably start with apremise whereby if
X (invariably) precedes Y in time as weH as covaries with it, X is the
cause of Y. Based on such apremise, and assuming temporal prece-
dence, the additional presence of covariation information warrants the
inference of a causallink between X and Y. In other words, granting the
temporal precedence of X in relation to Y, an attributor could assurne
that if X also covaries with Y, it is Y's probable cause. To summarize,
people may derive causality from covariation, based on the commonly
shared understanding of the causality concept.

Beyond Covariation and Temporal Ordering: Alternative Types of

"Causal Evidence"
Although covariation and temporal precedence are important evi-
dence categories for causality, they are not the exclusive kinds of evi-
dence. Our LEGO metaphor, outlined in Chapter 2, suggests that in
somebody' s mind any notion could be implicationally linked with any
other notion. Thus, anything could occasionally constitute evidence for
any other thing, including causality. For example, an individual could
infer with confidence that Xis the cause of Y (e.g., that the Russian flu is
cause of one's upset stornach) upon being told so by an expert (e.g., the
family physician), or if the group consensus asserted this to be the case.
In the last examples, too, causality is inferred deductively albeit from
evidence other than covariation, notably from assumptions concerning
expert or group authority (for a fuller discussion of various epistemic
authorities see Kruglanski and Jaffe, 1988).

The Attributional Criteria Reconsidered: Causal Process or Contents?

Consider what the foregoing analysis implies for the attributional
criteria of distinctiveness, consensus, and consistency. Specifically, it
suggests that these comprise two separate elements: the covariation
60 CHAPTER4

principle as such and the particular causal categories (or dimensions) to

which it is conjoined. For example, the distinctiveness criterion conjoins
positive covariation evidence to the entities category, the consistency
over times criterion conjoins negative covariation evidence to the times
category, and the consistency over modalities criterion conjoins negative
covariation evidence to the modalities category.
As noted earlier, covariation evidence can establish the causality of
any X, that is, of any conceivable category (see also Kelley, 1971). Thus,
Kelley's (1967) ANOVA matrix depicts a situation in which a person is
concerned about the causal role of the categories: (1) entity (or stimulus),
(2) person, (3) time, and (4) modality. Now, there is no compelling
reason to assurne that persons should be pervasively concerned with the
causal role of those particular four categories to the exc1usion of all
others. Depending on the circumstances, any set of categories could
figure in persons' causal hypotheses. In such a case, the covariation
principle might be conjoined to entirely different sets of concepts, giving
rise to an entirely different set of attributional criteria or data patterns
implying a given causal attribution.
For example, consider an economist attempting to determine
whether the cause of a soaring inflation was governmental spending,
public consumption, liberal imports policy, or subsidies for certain prod-
ucts. Our economist might welllook for evidence (e.g., from other times
or places) to indicate whether any of those causal factors actually
covaried with inflation. A set of criteria pointing to govemmental spend-
ing as the cause would involve distinctiveness (or covariation) of infla-
tion with periods of greater or lesser spending, and consistency of infla-
tion (i.e., an absence of covariation) across periods of lesser or greater
consumption, a more or less liberal imports' policy, and the presence
and absence of subsidies.
Attributional research conducted over the last two decades attests
to the variability of possible causal dimensions or categories that lay
persons might occasionally employ. Thus, Weiner and his associates
(e.g., Weiner, 1979; 1985; Weiner and Litman-Adizes, 1980) proposed
that, in achievement settings, persons may ascribe their successes and
failures to causes that are internal or external, stable or unstable, and
controllable or uncontrollable. This three-dimensional analysis identifies
eight possible causal categories to which a success or a failure can be
imputed: for example, whereas the "effort" category is internal, unsta-
ble, and controllable, the "task-difficulty" category is external, stable,
and uncontrollable.
Further attributional work identified additional causal dimensions
of presumed interest to lay persons. Thus, Abramson, Seligman, and
Teasedale (1978) identified the attributional dimension of globality (refer-
UNIQUE AND NONUNIQUE ASPECTS 61

ring to a cause' s range of applicability), in addition to intemality and

stability, Orvis, Kelley, and Butler (1976) identified the attributional di-
mension of intentionality (intentional or unintentional), as well as 13
categories of explanation used by young couples to account for their
own or their partner's behaviors. Kruglanski (1975) distinguished
among endogenous and exogenous attributions of actions, the former refer-
ring to cases in which the action is perceived as an end in itself, and the
latter to cases where it is perceived as means to a further end.
Anderson (1983) suggested that different types of situations invite
different types of plausible causal candidates and proceeded to demon-
strate this in research crossing interpersonal versus noninterpersonal
with success versus faHure situations. In this vein, Fincham (1985) writ-
ing on attributions in elose relationships noted that
attempts to categorize attributions in relationships elucidate the !imitations of
the categories utilized to date ... {and that} the intemal-extemal dimen-
sion . . . is tricky to operationa!ize and has created . . . confusion among
attribution researchers.

To reflect more faithfully the type of attributions occurring in rela-

tionships, Newman (1981, p. 63) proposed a category of interpersonal
attribution stressing "one's perception of self in regard to other and
other in regard to self." Fincham (1985, p. 225) proposed to differentiate
even further the interpersonal attribution category into an "interactive
attribution which maintains a focus on the partner (e.g., She does not
trust me," Partner has a poor attitude about me) and one which is truly
relational (e.g., There is a lack of trust between uS," lack of mutual
affection). "
However diverse, these attributional categories and dimensions
might merely reveal a tip of the iceberg. Depending on the topic, attribu-
tors may be capable of invoking a vast array of disparate causal hypoth-
eses to account for (the comparably heterogeneous) effects of possible
interest. And, to reiterate, each set of causal hypotheses might give rise
to a corresponding set of attributional criteria in which the general
covariation principle is conjoined to the specific causal categories in-
voked in a given instance. This implies that the specific attributional
criteria of distinctiveness, consensus and consistency might be of rela-
tively little interest to attributors interested in causal categories other
than person, entity, time, or modality to which the foregoing criteria are
not relevant.
Research to date supports the above suggestions. In a study de-
scribed in Kruglanski, Hamel, Maides, and Schwartz (1978) subjects
rated the relative usefulness of (1) consensus and distinctiveness infor-
mation in which covariation evidence is respectively conjoined to the
62 CHAPTER4

categories of persons and entities, and (2) covariation information in-

volving alternative conceptual categories (e.g., of ability and effort). As
predicted, consensus and distinctiveness information was chosen only
where subjects' attributional problem involved the person and entity
categories. Where the problem was phrased in terms of alternative
categories, however, consistency and distinctiveness information were
rated less useful than covariation information specifically relevant to
those latter categories.
In another study described in Kruglanski, Hamei, Maides and
Schwartz (1978), subjects were found to express differential interest in
different types of criterial information depending on categories impli-
cated in plausible causal hypotheses of interest. Consensus information
was preferred over distinctiveness, and consistency where person fig-
ured in the hypothesis of interest; distinctiveness information was most
preferred where entity figured in that hypothesis, and consistency infor-
mation was most preferred where time did.
These findings suggest that attributors' interest in different types of
information is affected by the information' s perceived relevance to
causal hypotheses considered by those attributors. Relevance, in turn,
can be conceived of as an assumed implicationallink (of an if-then kind)
between the evidence and the hypotheses that affords the derivation of
the hypotheses from the evidence. Besides accounting for conditions
under which different types of criterial information would be sought out
or preferred by attributors, the notion of relevance might explain the
conditions under which criterial information would or would not impact
on these persons' judgments.
Consider the recurrent finding (for reviews, see Kassin, 1979;
Harvey and Weary, 1985) whereby the magnitude of consensus deter-
mines whether consensus information has the predicted effect on at-
tributions. As noted earlier, consensus signifies an absence of covaria-
tion between any particular person and the effect. Hence, affirmation of
consensus allows the deductive exclusion of person as cause of the
effect. According to this analysis, the magnitude of consensus variable
bears on whether consensus is, in fact, affirmed. Thus, for an attributor,
consensus might mean "almost everyone exhibiting a behavior," but
only 70% of the people, as compared with, say, 95%, might not quite
qualify as "almost everyone," failing to affirm consensus and to conse-
quently affect attributions. Somewhat similar relevance considerations
might underlie the finding that perceived representativeness of consen-
sus information determines its degree of impact on attributions (cf. Ka-
ssin, 1979). To the extent that the sampie is nonrepresentative, sampie
consensus fails to affirm population consensus, correspondingly reduc-
ing the effect on attributions.
UNIQUE AND NONUNIQUE ASPECTS 63

The case discussed by Kassin (1979), in whieh one type of consen-

sus (e.g., the implicit one based on the attributor's expectancies) contra-
dicts another (e.g., the explicit one based on the actual behavior of
others), has clear logieal significance as it defines a situation where two
items of evidence are logically inconsistent with each other, forming a
pair of contradietory cognitions of the A, not-A type. As noted earlier,
such a situation shouldundermine the knower's confidence in either
cognition, weakening the impact of either type of consensus informa-
tion. Note that for this to occur consensus has to be perceived as log-
ieally relevant to the judgment at hand. This particular conc1usion has
been stressed recently by several authors (Borgida and Brekke, 1981;
Kruglanski, Friedland, and Farkash, 1984).

Alternative Attributional Criteria

Of the three attributional criteria specified by Kelley (1967), consen-
sus received by far the most research attention. Yet the principle of
perceived relevance should apply to distinctiveness and consistency in-
formation as weIl. For instance, the magnitude of consistency informa-
tion should exert an effect similar to the magnitude of consensus. Specif-
ieaIly, an attributor might reason that if only in a given situation P, but
not other persons, exhibited a given behavior on most occasions the
cause must reside in the P-situation combination. For instance, if almost
every time John takes an airtrip he feels siek, it might seem reasonable to
assurne that the siekness was caused by John's taking the airtrip. Infor-
mation that the siekness occurred on only 60% of the trips might fail
appreciably to affect the attributions as 60% might fall to qualify as most
occasions. Thus, as in the case of consensus, strong (versus weak) con-
sistency information affirms consistency more compellingly and, hence,
provides firmer logical base for the appropriate attributions.
Relevance considerations should also apply to distinctiveness infor-
mation. For instance, if such information is stated more (versus less)
confidently or is consistent (versus inconsistent) across informants, it
might affirm distinctiveness (i.e., covariation) more compellingly, exert-
ing a correspondingly greater impact on attributions. Suppose a mother
remembered confidently (rather than vaguely) that her child was weIl
before eating a souffle and was sick immediately afterward. Such dis-
tinctiveness information might be feIt more strongly to warrant an at-
tribution of the siekness to the souffle. In sum, it appears that all criterial
information would have greater impact if it unambiguously affirmed the
antecedent term in apremise specifying the information's relevance to
the attribution; such unambiguous information establishes a firm deduc-
tive base for inferring the attribution from the evidence.
64 CHAPTER4

Logical Aspects of the Discounting Principle

Just as the individual's confidence may be bolstered by evidence
deductively consistent with a proposition, it may be undermined by
inconsistent evidence. A special case of this is represented by a major
attributional concept known as the discounting principle whereby "the
role of a given cause in producing a given effect is discounted if other
plausible causes are also present" (Kelley, 1971, p.S). In other words, an
attributor's confidence in a given cause should wane if alternative plaus-
ible causes were also apparent. According to the present interpretation,
discounting would occur only to the extent that the alternative plausible
causes represented competing alternative causes, that is, statements in-
consistent with, or contradictory to the original causal proposition. In a
sense, such competing causes can be regarded as evidence or informa-
tion inconsistent with, hence, effecting a withdrawal of confidence
from, the orginal proposition.
For instance, consider an attributor wondering whether an actor' s
reason for performing a task resided in the task'5 intrinsic attractiveness or
in an extrinsic gain contingent upon task performance. First note that
intrinsic attractiveness and extrinsic gains represent distinct conceptual
categories. Secondly, if the reason for an act is assumed to constitute its
exhaustive explanation, the two possibilities just outlined are inconsis-
tent: If intrinsic attractiveness was the complete reason, extrinsic gains
could not have been involved and vice versa. This argument does not
exclude the possibility that some combination of intrinsic and extrinsic
properties was the reason. To the extent that such a possibility appeared
plausible it would qualify as a third and separate explanatory hypothesis
inconsistent with the former two (of intrinsic attractiveness and extrinsic
gains). Faced with two or more contradictory explanations of some
effect, the knower' s confidence might weIl wane just as implied by the
discounting rule.
The lowering of confidence might not occur where the alternative
explanations were not conceptually separate, for example, where one
was hierarchically nested within the other. For instance, imagine a pa-
tient told by one physician that her symptoms are the consequence of a
flu, and by another that they are the consequence of a virus. It seems
that the second suggestion would not effect a discounting of the first
because no inconsistency seems involved. After all, some types of influ-
enza are viral, so the virus and influenza explanations are not mutually
exclusive.
Thus, the discounting rule seems to represent a special case of
reacting to inconsistent information via a loss of subjective confidence.
According to this analysis, discounting might not be unique to causal
UNIQUE AND NONUNIQUE ASPECTS 65

attributions but might apply to noncausal judgments as weIl. For exam-

pIe, if Jim assurnes that the plane for Indianapolis departs from Phoenix
at 2:30 P.M., he might lose some confidence in that proposition if told by
a friend that it departs at 2:30 A.M., instead. Thus, the introduction of
an incompatible alternative might reduce the person's confidence in a
hypothesis even though the propositions involved may not be causa!.

Discounting versus Augmentation

The foregoing discussion sheds a new light on the relation between
the principles of discounting and augmentation usually presumed to repre-
sent logical opposites. According to the augmentation rule
if for a given effect both a plausible inhibitory cause and a plausible facilita-
tive cause are present, the role of the facilitative cause in producing the effect
will be judged greater than if it alone were present as a plausible cause for the
effect. (Kelley, 1971, p. 12)

The discounting and augmentation principles both refer to the case

in which an additional cause is introduced into the situation beyond a
given facilitative cause. In the case of discounting, the additional cause
is also facilitative, and its introduction is assumed to lower confidence in
the initial cause. In the case of augmentation, the additional cause is
inhibitory, and its introduction is assumed to increase the perceived role
of the initial cause. So far, the discounting and augmentation principles
may seem as perfeet mirror images of each other. In fact, however, the
two rules might bear on quite distinct phenomena.
As emphasized earlier, the discounting principle relates to the phe-
nomenon of subjective confidence and it represents a special case of the
more general rule whereby logical inconsistency between propositions
occasions a loss of confidence. In this sense, the broad discounting
principle is content-free. It applies to any conjunction of logically incom-
patible propositions, whether causal or not, and irrespective of specific
topic.
By contrast, the augmentation principle does not see m to bear on
subjective confidence as such or to be tied with the effects of logical
consistency or inconsistency. Instead, the augmentation principle seems
tied to the concept of magnitude. More specificaIly, it expresses the idea
that the greater the magnitude of a counterforce (inhibitory cause) the
greater must have been that of the prompting force (facilitative cause) if an
effect has, in fact, occurred. In that sense, the augmentation principle is
content-bound. Itsscope seems restricted to magnitude inferences con-
cerning various casual forces.
These notions were tested empirically in a study by Kruglanski,
66 CHAPTER4

Schwartz, Maides, and Hamel (1978). In that experiment the number of

facilitative causes (one versus two) concerning some effect was manipu-
lated orthogonally to the presence (versus absence) of an inhibitory
cause. Two dependent variables, subjects' confidence in the identity of
an event's cause, and estimates of the cause's magnitude were included.
As implied by our previous analysis, subjective confidence in identity of
cause varied inversely with the number of facilitative causes, but not
with presence versus absence of the inhibitory cause. Conversely, causal
magnitude estimates varied directly with presence versus absence of the
inhibitory cause but not with number of facilitative causes. These results
support our analysis whereby the discounting and augmentation princi-
pIes are not simple mirror images of one another. Instead, the discount-
ing phenomenon seems to represent the broader notion that inconsis-
tency between propositions of whatever content reduces subjective
confidence, whereas augmentation refers to the specific idea that judg-
ments of causal magnitude are affected by the presence of counterforces
(and/or their size).
In sum, the higher order notion that persons validate their hypoth-
eses via logical derivation seems capable of illuminating a wide variety
of concepts and findings in the attribution area. Among other things, it
provides the basis for separating the principle of validation as such from
the contents of hypotheses being validated, and from evidence whereby
such hypotheses might be tested. It also serves to explicate the notion of
relevance that may often determine whether given information is sought
in the first place, or, if provided, whether it impacts on attributions.
Finally, it serves to highlight the connection between attributional con-
structs tied to confidence gain, in particular, the attributional criteria and
those tied to confidence loss, in particular, the discounting notion. In the
sections that follow further higher order epistemic notions in attribution
are considered, starting with notions of cognitive capability.

COGNITIVE CAPABILITY NOTIONS:

AVAILABILITY, ACCESSIBILITY, AND SALIENCY

Early attributional models reflected a rather static image of the lay

knower. Implicit in those frameworks was the notion that the attributor
is generally concerned with the same basic causal categories (e.g., of
internal or external causes, or of endogenous and exogenous causes)
and with the same basic patterns of information (e.g., of consensus,
distinctiveness, and consistency information). However, the present
epistemic analysis suggests that both the attributions rendered and the
information responded to should depend on considerations of cognitive
UNIQUE AND NONUNIQUE ASPECTS 67

capability. This includes both long-term considerations of construct

availability (Higgins et al., 1982) in a person's box of epistemic LEGO
blocks, and short term considerations of construct accessibility (cf. Hig-
gins and King, 1981) or stimulus saliency (cf. Taylor and Fiske, 1975,
1978) alluded to in the present searchlight metaphor, described in Chap-
ter 2.
The latter category of short-term capability effects has received a fair
amount of research attention in recent years. For instance, in a study by
Enzle and Schopflocher (1978), only subjects explicitly asked to render
attributions about dispositional qualities of a confederate responded dif-
ferentially to information suggesting an external cause of this person's
behavior versus information suggesting an absence of external cause. A
possible interpretation of this result is that only when rendered access-
ible (e.g., via explicit instructions) do dispositional categories channel
the attributor's attention to pertinent information (e.g., suggesting the
competing situational categories).
Diener and Dweck (1978) found that children classified as helpless
on the Intellectual Achievement Responsibility scale tended to ascribe
their failure on an experimental task to low ability, possibly because low
ability is a particularly accessible construct to those children. Similarly,
in research by Brunson and Matthews (1981) Type-A individuals ver-
balized low-ability attributions in a condition where failure was made
salient. Again, it is possible that for A-types low ability ascriptions are
particularly accessible given failure.
One way of drawing attention to specific causal categories is by
increasing the saliency or vividness of the external simulus representing
the categories at issue. Several studies reported evidence that salience
manipulations increase the tendency to make causal attributions to the
salient or vivid stimulus (cf. McArthur and Ginsburg, 1981; Pryor and
Kriss, 1977; Smith and Miller, 1979; Taylor and Fiske, 1975; Taylor and
Thompson, 1982). Although the specific theoretical interpretation of sa-
liency effects varies across investigators (for a review, see Harvey and
Weary, 1985), there seems to exist a widespread agreement that such
effects often occur; hence, that persons' tendency to invoke specific
causal categories is not static but rather depends on internal (ac-
cessibility) or external (saliency or vividness) factors drawing an individ-
ual's attention to specific categories.
Furthermore, individuals' sensitivity to information relevant to spe-
cific causal categories seems to depend on the accessibility of such infor-
mation. Thus, Ferguson and Wells (1980) found that attributors' respon-
siveness to consensus, consistency, and distinctiveness information was
increased by priming manipulations designed to enhance informational
accessibility. And in a review of conditions under which consensus in-
68 CHAPTER4

formation is likely to have impact, Kassin (1979) mentioned the saliency

of such information as one prominent factor.
The tendency to form a specific causallinkage may depend on the
accessibility of information conceming the effect as weH as the cause.
Thus, Anderson (1983) provided subjects with information about the
relation between risk-taking behavior and fire-fighting effectiveness.
The information was provided either concretely by a description of few
ca se studies or abstractly by a summary of numerous cases. It was found
that the tendency to causally link the two variables was more apparent
with concrete rather than abstract data. According to Anderson, this
could occur because concrete information is more readily accessible from
memory and more imaginable.
Accessibility effects might underlie the finding by Moore et al. (1979)
of a "dispositional shift" in causal memory. Specifically, subjects asked
to account for an action in which they just engaged tended to invoke
situational demands. Yet when asked to account for the same action
several weeks later, subjects tended to invoke dispositional explana-
tions. According to the present interpretation, the spedfic situation
category might be particularly accessible immediately foHowing the act,
because of its recent priming. But as the time elapses, the dispositional
category might become more accessible because of its frequent priming
(for a discussion of recency and frequency effects on accessibility, see
Higgins, Bargh, and Lombardi, 1985).
In sum, the type of causal categories assigned as explanations,
and/or the type of information affecting causal assignments, are often a
function of persons' short-term cognitive capability to cognize various
concepts that are affected by factors of accessibility and saliency. This
implies that persons might not be generally invoking the same causal
categories (e.g., situational, drcumstantial, or dispositional categories)
or seeking the same causaHy relevant information (e.g., consensus, dis-
tinctiveness, or consistency information) but rather vary considerably in
doing so, depending on the vicissitudes of their memory or shifts of
their attention. Assuming in addition that the lay persons' content uni-
verse of causal categories is quite vast, a rather dynamic image of the lay
attributor is suggested in which the explanations rendered vary con-
siderably across situations and occasions. A dynamic view of attribution
is also suggested by various motivational effects examined in the next
section.

MOTlVATlONAL BASES OF ATTRIBUTIONS

Initial work on attribution tended to deemphasize the motivational

dimension. The current popularity of attribution theory among sodal
UNIQUE AND NONUNIQUE ASPECTS 69

psychologists (for reviews of recent research, see Harvey and Weary,

1984, 1985) can be traced back to the mid and late 1960s and the early
1970s, when social psychologists seemed to grow weary of the motiva-
tional flavor of cognitive consistency theorizing. In this context, attribu-
tion theory might have been welcome as a refreshing informational al-
ternative to the prevalent motivational paradigm. ActuaIly, early con-
ceptions of attribution (e.g. Iones and Davis, 1965; KeIley, 1967) did
contain important references to motivational influences. However, the
informational turn in the Zeitgeist relegated them to a secondary place
and cast them as "ground" as opposed to "figure" in leading attribution-
al analyses. Nonetheless, as time went by a considerable body of moti-
vationally relevant attribution research has accumulated. The purpose in
the following pages is to discuss this research and integrate it within the
several motivational categories of the lay epistemic model.

EPISTEMIC MOTIVATIONS REVISITED: NEEDS FOR NONSPECIFIC AND

SPECIFIC CLOSURE AND FOR THE A VOIDANCE OF CLOSURE

The basic epistemic function presently assigned to motivational

constructs is to initiate as weIl as terminate the potentially endless se-
quence of hypothesis generation and validation, accounting for the for-
mation, maintenance, and change of knowledge structures. As de-
scribed in Chapter 2, the model identifies four general motivational
categories. Those are respectively the needs for nonspecific and specific
closure and the needs for avoidance of specific and nonspecific closure.
In the section that follows, extant motivational findings in attri-
bution are ordered in accordance with the foregoing epistemic need
categories.

Nonspecitic Need tor Closure Effects

The attributionalliterature contains references to persons' need to
cognitively structure their environment as a powerful motivational force
behind attributional activity. Thus, Kelley (1967, p. 193) in a seminal
paper highlighted the individual's motivation "to attain cognitive mas-
tery of the causal structure of the environment" as a pervasive motive
base for attributions. This reference and further similar ones allude to an
unbiased nonspecific need for causal closure rather than denoting a
motivational bias toward particular types of causal explanation. Accord-
ing to the present conception, nonspecific need for cognitive closure is
topic or situation specific rather than general. In other words, the indi-
vidual may strongly strive to attain answers to some questions and may
be quite content to remain ignorant about others. Recent motivational
research on attributions supports this view of need for closure effects.
70 CHAPTER4

Berscheid et al. (1976) found that female subjects in a dating study,

expecting to be highly outcome dependent on their date, paid more
attention to him, evidenced better memory of his characteristics and
behavior, and rendered more extreme as well as confident evaluations of
him. Outcome dependency on another person might well induce a need
for cognitive dosure concerning that person' s properties and indina-
tions to serve as a base for one' s own behavioral decisions in the interac-
tion. This much is implied by Berscheid et al. (1976, p. 978), who view
the above effects as "manifestations of an underlying motivation to pre-
diet and control the social environment./I Expected interactions with
(hence, possible outcome dependency on) another person were also
found to increase attributional activity in a study by Miller, Norman,
and Wright (1978). Similar effects were obtained in outcome-
dependency research by Erber and Fiske (1984) and Fiske and Neuberg
(1988).
In aseries of experiments by Harvey, Yarkin, Lightner, and Town
(1980), various ways of manipulating subjects' involvement in an ob-
served event led to increased attributional activity concerning the event.
Harvey et al. (1980, p. 552) defined involvement as a condition that
increases subjects' " attempts to understand the interpersonal dynamies
of the event./I In present terms, involvement might be conceived of as a
factor increasing subjects' need for cognitive dosure concerning the ob-
served event. SimiIar involvement effects on causal attributions were
also manifest in studies by Heller (1972) and Pittman, Schemrer, and
Wright (1977).
Often, the need for closure might be induced by an explicit demand
to answer a specific attributional question. Such might have been the
case in numerous attributional experiments in which subjects were di-
recily requested to assign an effect to specific causal categories (e.g.,
McArthur, 1972; Orvis, Cunningham, and Kelley, 1975). A study by
Enzle and Schopflocher (1978), referred to earlier, suggested that direct
attributional questions may give rise to an attributional activity that
would not have taken place otherwise. In that experiment, subjects
explicitly asked to render dispositional attributions appropriately dis-
counted the role of internal factors in which possible external explana-
tions were present. Discounting effects failed to emerge for subjects
given no explicit attributional instructions.
A need for cognitive dosure might be aroused under conditions in
which prior closure proved inadequate for some reason. Research on
spontaneously initiated attributional activities corroborates that sup-
position. In arecent review of the relevant literature, Weiner (1985a)
conduded that such activity is particularly likely when an expectancy is
disconfirmed, or when failure is encountered. Both conditions might
UNIQUE AND NONUNIQUE ASPECTS 71

suggest that the individual's existing knowledge structure was inade-

quate. Disconfirmation of an expectancy constitutes a cognitive inconsis-
tency (Kruglanski and Klar, 1987) likely to undermine persons' confi-
dence in the expectancy at issue. Similarly, failure of an action might
imply that the knowledge structure on which the particular action was
based was flawed. Thus, findings of spontaneous attributional activity
following unexpected events or failures are interpretable in terms of the
need-for-c1osure construct.
It is of interest to note Weiner's (1985a, pp. 82-83) admonition that
not every disconfirmed event or failure is likely to give rise to attribu-
tional activity. This recalls the point made earlier whereby need for
c10sure is likely to be topic specific rather than general. As Weiner put it,

attributional search ... requires time and may place cognitive strain on an
organism. Hence, important outcomes such as academic failure, job ineffec-
tiveness, politicalloss, marital problems, and for some the loss of a World
Series baseball game by a favorite team, seem especially likely to give rise to
"why" questions. (p. 83)

Specific Closure and of Avoidance Closure Effects

Motivational effects reviewed in the preceding section were de-
scribed as nonspecific. By this is meant that attributors exhibiting such
effects were assumed motivated to attain cognitive c10sure on a topic
without having apriori preferences among types of c1osure. Often, how-
ever, attributors might harbor strong such preferences. They may not be
motivated to attain or avoid just any causal closure, only closure with
some special properties. In such cases, one might speak of needs for
specific c1osures, or for the avoidance of specific c1osures.
In a sense, most motivational research in attribution falls in the
"specific effects" category. This is research investigating the desirability
of given attributional conc1usions (e.g., flattering or damaging to one's
self-esteem) as it may affect the process and outcomes of attributional
activity. Heider (1958, pp. 120-121) seemed to have such specific effects
in mind when he wrote on "subjective forces of needs and wishes"
possibly affecting attributions. Similarly, the notion of hedonic relevance
introduced by Jones and Davis (1965, p. 237), refers to individuals' pref-
erences for specific attributions that fulfill rather than obstruct a pur-
pose. Jones and Davis (1965) implied that the existence of a hedonically
relevant explanation (e.g., perception of an actor in positive terms,
when the action serves our interests, and in negative terms, when it
hinders our interests) serves to reduce the number of alternative expla-
nations (noncommon effects). This recalls the lay epistemic notion of
72 CHAPTER4

freezing on a preferred knowledge structure and disregarding possible

alternative structures.
Kelley (1967) alluded to specific needs for closure or the avoidance
of closure in a discussion of ego-protective or enhandng biases, which
he grouped among "less rational" attributional tendendes. In a subse-
quent paper, Kelley (1971, p. 22) identified an additional category of
such preferential tendendes occasionally deriving from the motivation
to control one's environment. According to Kelley, environmental con-
trol may involve the balance between controlling the "controllable" and
controlling the "important." In this sense, perceived controllability of a
given factor or its perceived importance represents preferential features
that possibly attract attributions to the factor in question.
Vigorous research on ego-protective and enhandng attributions
continued under the label of self-serving biases, amidst controversy
whether such effects are explicable in motivational or informational
terms (Ajzen and Fishbein, 1975; Miller and Ross, 1975). Despite those
early reservations, recent reviews (e.g., Harvey and Weary, 1984; Kelley
and Michela, 1980; Zuckerman, 1979); uniformly concluded in favor of
motivational interpretations (but see Tetlock and Levi, 1982). Such as-
sessments are based in part on experimental evidence in which attribu-
tional differences were observed among conditions equated on informa-
tional inputs, and varying among themselves in motivational
implications only (e.g., Miller, 1976; Ross and Sicoly, 1979).
In a typical self-serving study (e.g. Johnson, Feigenbaum, and
Weiby, 1962), subjects attributed or success to themselves and failure to
external drcumstances. However, recent research identified conditions
under which the pattern can be reversed. For example, Arkin, Apple-
man, and Burger (1980) found that subjects accepted less credit for suc-
cess than for failure when they expected their behavior to be evaluated
by a group of experts. Tetlock (1980) provided subjects with descriptions
of a teacher who either succeeded or failed and had made either self-
serving or counterdefensive attributions. Subjects evaluated more high-
ly the counterdefensive than the self-serving teacher, suggesting that
counterdefensive attributions might be rendered in order to gain sodal
approval.
According to the present analysis, both self-serving and sodal ap-
proval needs might occasionally render some attributions preferable
over others. In some instances, the preferential implications of one
motivation might conflict with those of the other. In such cases, attribu-
tions could probably be biased in the direction of the stronger motiva-
tion. Presumably, in research by Arkin et al. (1980) approval motivation
prevailed over esteem motivation, resulting in a preponderance of coun-
terdefensive attributions. In other instances, however, the relative mag-
UNIQUE AND NONUNIQUE ASPECTS 73

nitudes of the motivations involved might be reversed with correspond-

ing effects on attributions.

Self-Handicapping
In addition to interpreting existing evidence in preferred terms, a
person may actively manufacture evidence in support of a preferential
interpretation. At least one such tendency has been demonstrated re-
cently in research on self-handicapping (for a review, see Arkin and
Baumgardner, 1985). In the ground-breaking experiment on this phe-
nomenon (Berglas and Jones, 1978; see also Jones and Berglas, 1978;
Jones and Pittman, 1982), subjects who had previously experienced suc-
cess, but did not expect it to recur, ingested what they perceived as a
debilitating drug. Presumably, this might increase the likelihood of as-
cribing the past success to the individuals' ability and possible future
failure to the drug. This finding has been since replicated in several
studies (e.g., Gibbons and Gaedert, 1984; Kolditz and Arkin, 1982).
Furthermore, subjects who had succeeded in the past but were
expecting to fail in the future withdrew effort from their performance
(Frankel and M. L. Snyder, 1978), or chose a situation with an impedi-
ment to performance (Rhodewalt and Davison, 1984). Verbal reports of
handicaps, such as test anxiety, shyness, depression and hypochondria-
cal complaints also occurred in situations in which future perfomance
evaluation was expected (Snyder, Smith, Augelli, and Ingram, 1983).
The issue can be raised whether self-handicapping behavior strat-
egy reflects a preferential effect of motivations on private attributions or
merely a self-presentational strategy intended for public consumption.
In their initial demonstration of the effect, Berglas and Jones (1978)
included a "private" condition in which the experimenter was unaware
of the subject's performance; the degree of self-handicapping in this
condition was comparable to that in a public condition including an
aware experimenter.
Kolditz and Arkin (1982) contested this interpretation by arguing
that even in the private condition of Berglas and Jones's (1978) study
subjects might have been trying to impress the experimenter' s assistant,
a young woman. In a condition intended to establish a more complete
degree of privacy, Kolditz and Arkin (1982), indeed, reported an almost
total elimination of self-handicapping. However, arecent comprehen-
sive review of the self-handicapping literature by Arkin and
Baumgardner (1985) included studies in which self-handicapping ap-
peared to affect subjects' own self-judgments. Thus, Shrauger and Os-
berg (1980) found that low self-esteem subjects who had moderately
high- performance expectancies (hence, resembled persons with prior
74 CHAPTER4

successes yet concomitant doubts about their competence) appeared to

self-handicap by spending little time preparing for a task; in line with
the presumptive logic of the self-handicapping strategy, these subjects
attributed their failures to lack of effort and successes to ability. The
latter findings suggest that self-handicapping might be occasionally un-
dertaken in order to influence subjects' own self-views (for a similiar
conclusion, see Snyder and Wicklund, 1981, p. 215). In that sense, self-
handicapping might reflect a biased motivational influence on attribu-
tions, ascribable to subjects' wish to protect their self-esteem.
In sum, diverse concepts and research findings in the attribution
literature are compatible with the notion that needs for specific closure,
or the avoidance of specific closure, may often exert biasing effects on
causal ascriptions. Although most of the research to date has examined
biasing effects due to esteem or approval motivations, the present analy-
sis implies that such effects might be expected to emerge with numerous
alternative motivations as weIl. To reiterate, KeIley's (1971) construct of
control motivation suggests that some causal explanations might be pre-
ferred over others because they appear controllable or important. Ex-
trapolating further, it might be possible to imagine situations in which
preferred explanations are based on one' s need for physical safety or
comfort, or one's need for novelty or complexity, or, in fact, on any
conceivable need or goal. In this sense Jones and Davis's (1965) concept
of hedonic relevance comes dosest to capturing the broad flavor of the
notion of needs for specific closure, or for the avoidance of specific
closure as it refers to the relation of an attribution to a fullfilment or an
obstruction of a perceiver' spurpose of whatever kind.

Effects of the Need to Avoid Nonspecific Closure

A fourth type of motivational force that may occasionally affect
attributional activity is what I earlier labeled as the need to avoid non-
specific closure: (1) In its pure form, a heightened degree of such need
defines an absence of closure as a desired end state. (2) Often, however,
a need to avoid closure may appear concomitantly with a need for
closure; in those conditions, it may serve to prolong temporarily the
tolerance for ambiguity, extending the duration of an attributional search
until some closure is ultimately attained. Both types of situation have
been captured by attributional researchers and are now considered in
turn.

Attribute Ambiguity. Situations in which attributional ambiguity rep-

resents a preferred state of affairs have been dealt with by Snyder and
Wicklund (1981). These investigators discussed two general cases under
UNIQUE AND NONUNIQUE ASPECTS 75

which the absence of c10sure might be preferred over c1osure. In one

case, the preference sterns from the constraining implications of any
c1osure, whereas a lack of c10sure connotes unpredictability, hence, free-
dom. Studies by Goodstadt (1971), Rozen (1970), and Hannah, Hannah,
and Wattie (1975) suggest that in some situations individuals behave in
ways apparently aimed at enhancing the unpredictability of their
behavior.
According to the present analysis, with an absence of c10sure as the
desired goal, individuals are likely to engage in considerable attribution-
al search if they initially possessed cognitive c1osure. Under such condi-
tions, persons are likely to genera te competing hypotheses to a current
attribution and/or seek out or actively manufacture evidence inconsis-
tent with such attribution. In doing so, attributors are guided by the
deductive logic of hypothesis validation and in particular by the notion
that cognitive inconsistency undermines one's confidence in a proposi-
tion, promoting ambiguity. In fact Snyder and Wicklund (1981, p. 211)
described research illustrating such tendencies as instances of "using the
traditional attribution logic" in the interest of attaining ambiguity.
Furthermore, with lack of c10sure as the desired state, individuals
should refrain from attributional activity if they had little c10sure to start
with. Snyder and Wicklund (1981, pp. 205-207) discussed numerous
research findings demonstrating just such tendencies, which they label
as "leaving the field" mode of securing ambiguity.

Accountability of Conduct. TetIock (1985a,b) recently argued that ac-

countability of one' s conduct to significant others is a pervasive feature
of many judgments and decisions made in real-life social settings. Ac-
cording to TetIock (1985b), accountability may often: "motivate people to
become more vigilant, complex and self-critical information processors"
(p. 314). In present terms, accountability may arouse a need to avoid
premature c1osure, stemming from a fear of invalidity. It is such need to
avoid c10sure that may increase the thoroughness with which subjects
attend to contextually available information. In a study relevant to those
predictions (Tetlock, 1985a), subjects succumbed less to the "fundamen-
tal attribution error" of neglecting possible situational causes of an ac-
tor' s behavior and ascribing it exc1usively to the actor' s personality.

Control Deprivation. Research by Pittman and his colleagues (for a

review, see Pittman and 0'Agostino, 1985) on the effects of control
deprivation may represent a case in which "fear of invalidity" (Freund et
al., 1985; Kruglanski and Freund, 1983; Kruglanski and Mayseless, 1987;
Mayseless and Kruglanski, 1987a) may induce temporary avoidance of
c1osure. In a typical experiment, subjects in the high-control-
76 CHAPTER4

deprivation, condition receive failure feedback on a concept-formation

task. It is found that as compared with base line or low-deprivation
controls they tend more to utilize relevant information on a subsequent
attributional task, process such information more carefully, and take
more time before reaching a judgment. According to the present inter-
pretation, subjects in the high-deprivation condition, assuming to have
already failed once, might be particularly concerned about the possibil-
ity of a repeat failure and its consequences for their self-esteem. Such
fear of invalidity might weIl enhance subjects' need to postpone c10sure
and prompt the in-depth consideration of multiple items of information
apparently relevant to the judgment at hand.

IMPLICATIONS OF THE EPISTEMIC REANALYSIS OF MOTIVATIONAL EFFECTS

IN ATTRIBUTION

The reanalysis of motivational effects in attribution has several im-

plications of interest. Two major c1asses of such implications are now
considered in turn. One follows from the implied irrelevance to the
knowledge-acquisition process of the origin of given epistemic motiva-
tions. The second relates to the proposed integration of motivational and
informational aspects of attribution.

lrrelevance-of-Origin Assumption
As noted in Chapter 2, each epistemic motivation might derive from
diverse points of origin, and the same point of origin might, under the
appropriate circumstances, give rise to any epistemic motivation. For
instance, self-esteem concerns could occasionally serve as origin for the
need for nonspecific c10sure (as in the case of an expert whose reputa-
tion hinges on having an answer to a problem); in other instances, they
might serve as origin for the need for preferred c10sure (e.g., a conc1u-
sion with a self-serving content); and in yet others, as origin for the need
to avoid c10sure (cf. Snyder and Wicklund, 1981). Also as Kelley (1971,
p. 22) noted, the need to exercise control can occasionally serve as origin
for needs for specific c1osure, notably for causal structures appearing
controllable or important, whereas on other occasions the same need for
control might give rise to strivings for nonspecific or unbiased structures
(which Kelley, 1971, labels as more rational attributional tendencies).
Thus, in counterdistinction to previous motivational research in at-
tribution that often stressed the origin of motivational tendencies (e.g.,
in esteem concerns, the need for control, or for social approval), the
present analysis focuses on the epistemic end state (of nonspecific or
specific c10sure or the absence of nonspecific or specific c1osure) aimed at
UNIQUE AND NONUNIQUE ASPECTS 77

in each particular instance. The reason for this shift of focus is the
assumption that different possible origins for the same epistemic
motivation can be functionally equivalent as far as the attributional process
is concerned. For example, motivation for a nonspecific elosure might
arise from self-esteem needs, from the need to feel in control, or to make
adecision under haste. In a11 those cases alike, individuals' hypothesis-
generation tendencies and sensitivity to inconsistent information might
be affected in a similar way, notably by being suppressed or curtailed.
Similar examples could be cited in connection with motivations for a
specific elosure or for the avoidance of (specific or nonspecific) elosure.
In short, what seems to matter as far as the attributional or the epistemic
process is concerned is the end state being striven for rather than the
origin of those strivings in particular needs. The critical question to pose
in this connection appears to be what is desired rather than why.

Integrating the Informational and Motivational Aspects of Attribution

Previous motivational research on attribution has not been elearly
integrated with the informational aspects of the process. For instance,
the various ego-enhancing or protective tendencies were usua11y con-
sidered as less rational than the informational principles assumed to
guide attributions (e.g., the criteria of consistency, consensus and dis-
tinctiveness or rules like covariation, discounting, or augmentation).
Furthermore, motivational and informational accounts of self-serving
biases were often juxatposed (cf. Ajzen and Fishbein, 1975; Miller and
Ross, 1975; Zuckerman, 1979), a11 of which implies a disjunctive model
comprising two distinct routes to attribution, an informational one, typ-
ica11y regarded as also the more rational and a motivational one, re-
garded as less rational.
In contrast, the present epistemic model suggests that the motiva-
tional dimensionconstitutes an inseparable part of attributional activity,
and it assigns adefinite function to motivation in all instances of causal
assignment. In asense, this proposal explicates and possibly refines
what has been already implicit in the ground-breaking attributional
analyses of Heider (1958), Jones and Davis (1965), and Kelley (1967), all
of whom posited a motivational substratum without which no attribu-
tional activity is likelyto occur.
To reiterate, an essential function of motivation in attribution is to
initiate and/or bring to a elose the sequence of hypothesis generation
and validation by which a11 knowledge may form. Thus, the need for
nonspecific elosure is assumed to effect an initiation of attributional
activity and its quick freezing, once a plausible hypothesis has been
generated and found consistent with extant evidence. Needs for specific
78 CHAP'fER4

c1osure, or the avoidance of specific c1osure, are assumed to effect freez-

ing or unfreezing depending on whether a currently entertained hy-
pothesis corresponded with the desired state of affairs. Finally, the need
to avoid nonspecific c10sure is assumed to effect the unfreezing of cur-
rent knowledge structures and a cessation of attributional activity where
ambiguity prevailed.
All the foregoing tendencies are assumed fully coextensive with the
application of various informational principles elaborated by attribution
theorists. For instance, generation of an alternate hypothesis, driven by
a need to avoid c1osure, may result in confidence decrements in accor-
dance with the discounting logic (cf. Kelley, 1971). The motivation to
attain nonspecific or specific c10sure might lead to increased or de-
creased sensitivity to various types of information (e.g., consistency,
consensus, or distinctiveness information) depending on whether such
information appeared consistent or inconsistent with the desired
c1osure. Thus, motivational forces are assumed to affect the extent and
direction of information processing during attributional activity. How-
ever, such processing may still be affected by the same informational
principles, so a disjunction between motivational and informational
modes of attribution making may not be warranted.

CONCLUDING COMMENTS: UNIQUE AND NONUNIQUE ASPECT5 OF

ATTRIBUTIONS

From the present perspective it is useful to conceptually partition

factors affecting variance in causa1 attributions into two categories: (1)
those having uniquely to do with contents of causality or, even more
specifically, with contents of particular causal categories and (2) those
affecting nonuniquely causal and noncausal judgments alike.
In the first, unique category belong variables Iike covariation (in 50
far as it represents evidence for causality but not for other judgmental
contents), or the attributional criteria (of consensus, consistency, and
distinctiveness) that conjoin covariation to specific types of entities (no-
tably, person, time/modality, and external stimulus, respectively). A
notion like augmentation (Kelley, 1971) relates to causality or, more spe-
cifically, to the notion of causal magnitude as discussed earlier. The
concept of discounting relates to causality, as it is defined in reference to
competing causal hypotheses. A uniquely causal notion surprisingly
neglected in attributional writings is that of temporal ordering, (but see
Kelley, 1971), denoting the temporal precedence of cause with respect to
effect.
In the second, nonunique category belong concepts like logically
based hypothesis testing insofar as it pertains to the validation of causal
UNIQUE AND NONUNIQUE ASPECTS 79

and noncausal hypotheses alike. As demonstrated earlier, the notion of

logical testing underlies such attributional concepts as covariation in its
function as evidence for causation, the attributional criteria (of consen-
sus, consistency, and distinctiveness) serving as evidence for special
causal contents (e.g., of entity causation), or discounting as it represents
an invalidation of a hypothesis by a logical contradiction. To reiterate,
although causal hypotheses may be validated logically from relevant
evidence (like covariation), so may noncausal hypotheses of various
kinds. For example, one might deduce the hypothesis that it is time to
get up from relevant evidence indicated by one's alarm dock. Similarly,
although causal hypotheses might be invalidated by a logical contradic-
tion, so may noncausal hypotheses. Thus, one might lose confidence in
the hypothesis that the snow conditions in Vermont are good after lis-
tening to a contradictory report by a local TV station.
Beyond deductive reasoning, a group of nonuniquely causal concepts
often referred to in the attributionalliterature have to do with stimulus
saliency (cf. Taylor and Fiske, 1975, 1978) or construct accessibility (cf.
Higgins and King, 1981). Admittedly, rendering a stimulus salient or a
construct accessible may, under some conditions, enhance the likeli-
hood of accepting a causal hypothesis concerning the stimulus or con-
struct at issue. But similar effects are likely to be manifest with a non-
causal hypothesis as well. For instance, in a study by Higgins and Lurie
(1983), subjects evaluated a judge's verdict as lenient, if it fell below a
contextually accessible assessment standard, and as harsh, if it fell above
the accessible standard (for a review of similar such studies, see Higgins
and Stangor, 1988). Thus, accessibilty or saliency effects might affect
nonuniquely the acceptance of causal as weH as noncausal hypotheses.
Furthermore, the various motivational forces shown to affect at-
tributions might all be dassed in the category of nonunique factors as
they appear to affect similarly noncausal judgments as weIl. Consider,
for example, the need for nonspecific dosure assumed to effect an initia-
tion of attributional search and its cessation after attaining dosure on a
given causal assignment. According to the present analysis, need for
dosure would exert similar effects on judgments whose contents are
noncausal. For example, in research by Kruglanski and Mayseless (1987)
need for dosure heightened via an informational inconsistency in-
creased the extent of epistemic search (manifest via social comparison)
in regard to noncausal judgments concerning a target' s emotional
maturity.
Arousal of the need for dosure by cognitive inconsistency recalls
the effects of disconfirmed expectancies that also involve an inconsisten-
cy of sorts (notably between prior beliefs and present information). As
Weiner's (1985a) review suggests, such disconfirmed expectancies often
 CHAPTER4

resuIt in the initiation of an attributional search for causes of the unex-

pected events. According to the present analysis, disconfirmed expec-
tandes should instigate informational search concerning noncausal mat-
ters as weIl. Suppose that on the way to the airport one realized that
one's watch was discrepant with that of the cab driver. This might weIl
represent an unexpected occurrence giving rise to a frantic informational
search. The latter might not necessarily aim at locating the cause for the
discrepancy, but rather at finding out which of the two contradictory
pieces of information was, in fact, correct. Thus, one might check an-
other passenger' swatch or glance at a dock tower to ascertain the cor-
rect time.
Arousal of a need to avoid dosure might exert similar effects on the
judgmental process irrespective of whether the contents of such judg-
ments were causal or noncausal. For example, to the extent that manip-
ulation of the "control deprivation" in Pittman's research (cf. Pittman
and 0' Agostino, 1985) aroused subjects' "fear of invalidity," we might
expect the intensified information-processing activity observed on a sub-
se quent task to replicate with noncausal tasks as weIl. Intensified infor-
mation processing under a heightened fear of invalidity was manifest in
aseries of recent studies by Mayseless and Kruglanski (1987) employing
noncausal, in fact, perceptual recognition tasks.
Effects of needs for specific dosures should affect noncausal judg-
ments in similar ways to causaIones. For instance, self-serving needs
might lead to an interpretation of an ambiguous remark as a comple-
ment rather than a criticism, to an overestimation of one's popularity or
physical attractiveness, or to optimistic expectations regarding the out-
come of a job interview. None of the latter judgments qualifies as causal
yet they might weH be biased toward upholding a positive view of
oneself just as has been demonstrated with causal attributions (Miller,
1976; Ross and Sicoly, 1979).
The present view has been that causal attributions represent a con-
tent domain of the general judgmental process; hence, that they are gov-
erned nonuniquely by the same prindples as other judgmental contents.
Recent experimental support for this view has been obtained by
Hamilton (1988). Using recall memory as the major dependent variable,
Hamilton was able to demonstrate that inconsistent sets of attributionally
relevant information (e.g., some items in the set implying a person
attribution and others implying an object attribution) were better re-
called than consistent sets (all items in the set implying either a person or
an object attribution).
According to Hamilton this is because the inconsistent (versus the
consistent) items require more extensive thought relating them to each
other before an attributional judgment can be made. As a consequence,
UNIQUE AND NONUNIQUE ASPECTS 81

these items become linked with others in a well-developed network as

this information is represented in memory. Such manner of organization
in memory is assumed to facilitate retrieval of information in general
(Hastie and Kumar, 1981) and is not presumed specifically restricted to
the retrieval of causally relevant information. Thus, Hamilton's research
illustrates a sense in which attributional activities are special cases of the
general cognitive process that differ in contents only from numerous
alternative judgments that persons might make.

IMPLICATIONS FOR FURTHER RESEARCH

The present epistemic analysis of attributional activity has ample

implications for further research. To mention just two, consider briefly
the implications of the present emphasis on (1) the uniform epistemic
function of various motivations, and (2) the special case relation between
attribution and the general epistemic process.
According to the present analysis, the several types of motivation
serve essentially the same epistemic function of initiating and/or termi-
nating the sequence of hypothesis generation and evaluation. Hence,
divergent motivations should lead to similar phenomena as far as the
attributional process is concerned. This suggests the possible inter-
changeability of dependent variables to be studied in conjunction with
different motivational forces.
For instance, in past research, effects of needs for specific closures
(e.g., self-serving ones) were demonstrated primarily via attributional
products, that is, by the contents of causal ascriptions ultimately made.
By contrast, effects of needs for nonspecific closure were demonstrated
via the initiation of an attributional activity and/or the extent of such an
activity. But, according to the present analysis, extent of attributional
activity should also be impacted by specific closure needs; specifically, it
should vary inversely with the desirability of an initial attribution.
Similarly, attributional products might often be affected by the es-
sentially nondirectional motivations for nonspecific closure or for clo-
sure avoidance. For instance, to the extent that a given causal category
was readily accessible (say, the person category) and the individual was
under high need for nonspecific closure, such a category might likely be
assigned as cause of the event. By contrast, the likelihood that a less-
accessible category would be ultimately assigned as a cause should in-
crease under conditions in which need to avoid closure was enhanced,
for example, via induction of a fear of invalidity.
The present assumption that attribution is a special case of broader
judgmental processes suggests that it will reflect phenomena demon-
strable also with alternative judgmental contents. Hamilton's (1988)
82 CHAPTER4

aforementioned work on attributional memory is relevant to such pre-

diction. Another line of simiIarly minded research might extend the
work on epistemic freezing and unfreezing (cf. Freund et aZ., 1985;
Kruglanski and Freund, 1983) to the domain of causal attribution. Thus,
under a high need for nonspecific closure, or a preferential one con-
gruent with an initial attribution, persons might be insensitive to infor-
mational patterns inconsistent with the attribution in question. Con-
versely, under a high need to avoid closure or for a preferred closure
incongruent with an initial attribution, persons might be particularly
sensitive to inconsistent information and tend to come up with an alter-
nate causal hypothesis competing with the intial attribution. Those im-
plications of the present analysis could be profitably explored in future
attributional research.

SUMMARY

The notion that attribution theory constitutes a special case of lay

epistemics was examined in this chapter. Thus, a distinction was drawn
between various causal categories in which lay persons may occasionally
be interested (e.g., the internal and external categories, or the endoge-
nous and exogenous categories) and the way in which causality is as-
cribed to those categories. The former may be specific to particular at-
tributions, whereas the latter is general across attributions. However,
from the present perspective even the broad domain of causality is con-
sidered a content area of inference governed by the same epistemic
process whereby all inferences (including noncausal inferences) may be
reached.
Conceptual and empirical evidence for the operation of such pro-
cess in various attributional instances was considered. Thus, it was
noted that the attributional criteria (Kelley, 1967), whereby persons are
assumed to validate their causal assignments, involve the logical deriva-
tion of given causes from premises concerning the nature of causality.
Whereas logical consistency of a causal hypothesis with a pattern of
relevant evidence may facilitate a gain in confidence concerning a
category' s role in producing an effect, logical inconsistency between
competing causal hypotheses may produce a confidence loss. The latter
notion is implicit in Kelley's (1971) discounting principle, whereby the
attributor' s confidence in the causal role of a given category wanes if
alternative plausible categories are also present. Consistency and incon-
sistency notions in attribution theory are seen to represent the general
logical element in the validation of all hypotheses.
Whereas attribution theory is sometimes considered a cognitive for-
UNIQUE AND NONUNIQUE ASPECTS 83

mulation in which motivation plays no major part, considerable motiva-

tionally relevant research on attribution has recently been carried out.
Such research was presently interpreted in terms of the present notions
of epistemic motivations. In those terms, studies that stress persons'
need to understand the causal structure of their environment allude to a
need for nonspecific closure on a causal topic; studies of defensive (or
ego-enhancing) attributions deal with needs for specific closure or the
avoidance of specific closure; whereas studies on the desirability of at-
tribute ambiguity, for example, allude to a need to avoid nonspecific
closure. Thus, the present epistemic framework integrates the motiva-
tional and informational aspects of attribution. Such a unified concep-
tion has heuristic implications for further research on various attribu-
tional problems.
 5
A Bridge to Consistency Theories

Attribution theory is sometimes considered as an informational or a

"cold cognition" reaction to cognitive consistency models of the 1950s
and 1960s, such as dissonance (Festinger, 1957) or balance (Heider,
1958), whose flavor was predominantly motivational, hence, depicting
"hot cognition." Indeed attribution theory and the cognitive consistency
models project contrasting images of the person. Attribution theory pro-
jects a highly rational image that likens the lay individual to the sophisti-
cated scientist or to a dispassionate seeker after truth. By contrast, cogni-
tive consistency models convey a largely irrational image in which
human judgments are strongly biased by needs, wishes, and desires.
Because of such differences, the attribution and consistency paradigms
have often been considered incompatible and have been pitted against
each other in experimental research. A prominent example of such an-
tagonism was the heated controversy between proponents of Bem' s
(1967) self-perception theory and of Festinger's (1957) dissonance theory
(see Bem, 1972, and Wicklund and Brehm, 1976, for a review of this
particular dispute).
Yet despite their differences, the consistency and attribution para-
digms shared an important similarity: Each spawned a variety of sepa-
rate formulations whose commonalities and differences have not been
c1early spelled out. Moreover, there have been few attempts to interre-
late the attribution and the consistency frameworks and to explore their
mutual relevancies and implications (but see Kruglanski and Klar, 1987;
Zanna and Cooper, 1976). By now, social cognitive psychology has come
to inc1ude a plethora of theoretical formulations c1assifiable into the
consistency and attribution categories. But a lack of synthesis within
paradigms and between paradigms has lent this area of study a complex
and unwieldy character. The lay epistemic theory promises to furnish
such a c1arifying synthesis.
The previous chapter has attempted to integrate the various attribu-

85
86 CHAPTER5

tional notions within the general epistemic model. This chapter extends
the integrative analysis to the consistency theories as weIl. Specifically,
it will show that they, too, are special cases of the lay epistemic frame-
work and are therefore not as remote from the attributional formulations
as has been typically supposed.

DEDUCTIVE, MOTIVATIONAL, AND CONTENT-RELATED ELEMENTS

OF EPISTEMIC BEHAVIOR

Dur epistemic reanalysis of the consistency paradigm will be carried

out by extracting from the epistemic process described in Chapter 2
three important aspects: (1) deductive logic, assumed to represent the
validation criterion of all judgments, (2) cognitive contents, assumed to
comprise the substance of judgments and inferences, and (3) motivation,
assumed to affect the initiation and termination of the epistemic se-
quence and to account for affective phenomena associated with fulfill-
ment or frustration of the individual's epistemic needs. These aspects
are regarded as conceptuaIly separate, functionally complementary, and
jointly essential to understanding human epistemic behavior. The dis-
tinction between the deductive, motivational, and content components
of epistemic behavior will therefore guide the subsequent integrative
analysis of the consistency models.

THE LOGICAL ASPECT

A major and largely unresolved theme in cognitive consistency the-

orizing has concerned the nature of the inconsistency being addressed:
Is it a logical inconsistency or a psychological inconsistency? Is there a
unitary type of inconsistency variously tapped by different consistency
models, or do those models depict different types of inconsistency? The
following analysis suggests that the juxtaposition of logical and psycholog-
ical inconsistency is misleading. In part, it sterns from a confusion of the
logical, motivational and content-related aspects of cognitions. In this
chapter, an attempt will be made to show that any inconsistency in the
cognitive consistency literature has a logical aspect: Invariably, it in-
volves a simple contradiction of the A and, not-A types. But the reasons
why a given person arrives at a given inconsistency might well be la-
beled psychological. These may concern (1) the idiosyncratic cognitive
contents that populate a particular individual's mind; and (2) the specific
epistemic motivations that affect this person' s tendency to believe in
A BRIDGE TO CONSISTENCY THEORIES 87

specific propositions or subscribe to specific premises. As shall be seen,

the different "types" of inconsistency featured in the literature vary
among themselves in such "psychological" elements (i.e., in the con-
tent-related or motivational elements). However they share in common
the "logical" element of inconsistency as simple contradiction. In the
present section, the focus is, in particular, on the inconsistency element
in major consistency models.

FESTINGER'S DISSONANCE THEORY

According to Festinger (1957, p.13), "two elements are in a disso-

nant relation if, considering these two alone, the obverse of one element
would follow from the other." Subsequent commentators, however
(e.g., Insko, 1967) faulted the imprecision of the terms obverse and follows
from, for obfuscating the meaning of "dissonance." According to the
present analysis, the expression "follows from" signifies an implication-
al if-then relation. To say that B follows from A is tantamount to assert-
ing one's belief in the statement if Athen B. Furthermore, the term
obverse in Festinger's definition can be interpreted as a negation, the
obverse of B being not-B, for example. The above interpretation suggests
that a dissonant relation represents a simple contradiction; that is, a case
in which a person has reasons to consider seriously both Band not-B.
Specifically, A and not-B would be in a dissonant relation if an individu-
al subscribed to a previous belief whereby if Athen B: taken together, if
Athen B, and A implies B. But the individual observes that not-B is the
case; this results in a contradictory pair of B (as implied) and not-B (as
observed).
It is of interest to note that following Festinger (1957), leading disso-
nance theorists considered "logical inconsistency" as one merely among
several possible sources of dissonance rather than as the essential core in
all dissonance. Let us now examine those various sources and consider
carefully the type of inconsistency they exemplify.
In an initial statement, Festinger (1957) listed the following four
sources of dissonance: (1) Logical inconsistency. If a person believed that
human beings will reach the moon in the near future and also believed
that they will not be able to build a device that can leave the atmosphere
of the earth, these two cognitions are dissonant with each other. (2)
Inconsistency with past experience. If a person were standing in the rain
and yet could see no evidence of getting wet, these two cognitions
would be dissonant with one another because the person knows from
experience that getting wet follows from being out in the rain. (3) Cultur-
al mores. If a person at a formal dinner uses his or her fingers to' eat a
piece of chicken, the knowledge of what he or she is doing is dissonant
88 CHAPTER5

with the knowledge of formal dinner etiquette. (4) Inclusion of a specific

opinion in a more general opinion. If a person is a loyal Democrat, but in a
particular election prefers a Republican candidate, the two opinions are
dissonant with each other because being a Democrat inc1udes, as part of
the concept, favoring Democratic candidates.
Now I should like to argue that in all the above cases there exists an
essential core of logical inconsistency or contradiction. In the last case,
for example, the reference seems to be to a belief that if someone is a
Democrat (A) that person votes for the Democratic candidate (B). This
belief is contradicted by the knowledge that A is affirmed (an individual
is a known Democrat) implying B, but in fact not-B is known to be the
case (the person voted for a Republican candidate). A similar reanalysis
can be applied to the rain-without-wetness example in (2): An individual
believes that if someone is standing in the rain, he or she is sure to get
wet. In other words, getting wet (B) is implied by standing out in the
rain (A), so standing in the rain (A) without getting wet (not-B) involves
a logical contradiction of the B, not-B type.
Perhaps most troublesome to our logical reanalysis is the chicken-
eating example that represents dissonance based on cultural mores. It
could be argued that no logical inconsistency is involved here: A person
noted that he or she failed to uphold the rules of proper etiquette, but
then nobody has implied that he or she would uphold such rules; so
where is the logical inconsistency? On reflection, the hypothesis that our
individual was in fact an etiquette-observing type, or was considered
such by other people, must have been at least entertained by this per-
son. Someone who firmly believed he or she is a perfect slob would
hardly be affected by a new item of information suggesting that he or
she iso Thus, the chicken-eating example may not be expected to work or
create the aversive sense of dissonance without an element of logical
contradiction: notably, a contradiction between the knower's belief that
he or she is, or is perceived by others to be, a well-bred, etiquette-
observing person, and the finger licking evidence that he or she is not.
In sum, I am proposing that in all of Festinger' s examples there
exists an element of logical inconsistency or contradiction. The examples
may differ in the sources of the contradictory beliefs. In some cases,
these have to do with past experience (the wetness-rain example); in
other cases, with connotative meanings of concepts (like Democrat); in
yet other cases, with cultural norms or mores. Furthermore, the state-
ments that make up the examples may vary in their believability to
different individuals: Some persons might believe that in order to reach
the moon one needs, first, to leave the atmosphere of the earth, whereas
others (badly misinformed, most of us would say) might believe it possi-
ble to reach the moon without leaving the earth's atmosphere.
A BRIDGE TO CONSISTENCY THEORIES 89

Both the source of a given statement and its degree of believability

can be properly regarded as psychological or subjective: They may vary
widely from one individual to the next depending on the idiosyncratic
contents of aperson' s phenomenology and/or the various epistemic
motivations operative for this individual at a time. But possible psycho-
logical differences among cognitions should not obscure the fact that
any cognition can be involved in a logical contradiction. Indeed, in a
subsequent section, it is demonstrated further that the aversive proper-
ties of dissonance stern in part from the essential epistemic consequence
of contradiction: the undermining of belief. Before doing so, however,
another major cognitive consistency formulation, Heider' s balance theo-
ry, will be considered.

HElDER' S BALANCE THEORY

Balance theory addresses three categories of elements: The person

(p) in whose phenomenology balance type of effects occur, another per-
son (0) about whom p is thinking, and (x), an idea, event, or entity to
which p and 0 relate in p's perceptions. Between any two of the elements
may exist relations of two kinds: a sentiment relation and a unit relation.
The term sentiment refers to such affects as liking or disliking, loving or
hating, approving or disapproving. The term unit refers to the tendency
of elements to be "perceived as belonging together" (Heider, 1958, p.
176). Examples of unit relations would be "owns," "is similar to," "be-
longs to or with," and "is in proximity to." Both sentiment and unit
relations can be positive (likes, owns), or negative (dislikes, disowns).
Heider' s discussion of balance phenomena addresses two separate
contexts: the triad (p, 0 ,x) and the dyad (e.g., p, 0 or p, x). According to
Heider (1958), balance is "a harmonious state ... in which entities com-
prising the situation and feelings about them fit together without stress"
(p. 180). Just like dissonance, imbalance is assumed to induce psycho-
logical tension and to dispose cognitive reorganization aimed at the
restoration of balance.
It is the sign of a relation (plus or minus) and not its type (sentiment
or unit), or context (triad or dyad) that is theoretically significant in the
balance formulation. Specifically, the states of balance and imbalance
can be determined by applying the product rule (Cartwright and Ha-
rary, 1956), wherebya system is in balance when the product of its signs
is positive, and not in balance when the product is negative. For exam-
pIe, a triad in which p likes 0 (p, L ,0), p likes x (p, L ,x) and 0 does not like
x (0 -L x) is imbalanced because it contains an odd number of negative
signs that yield a negative product. Similarly, a dyad in which p L 0 and 0
L P is balanced because of a positive product, whereas the dyad p L x and
90 CHAPTER5

p -u x (p is "not in a unit" relation with x) is imbalanced because of the

negative product.

Exceptions to the Product Rule

From the earliest days of balance theory, exceptions to the product
rule were apparent. Consider the humorous example of chicken feed
(imputed by Zajonc, 1960, to Festinger): A person could like chicken and
dislike chicken feed, which, in turn, is craved very much by the chicken.
On strict application of the product rule, this state of affairs represents
an imbalance; therefore, it should lead to unpleasantness and tension.
Yet we all sense that no such effects are likely.
Another famous exception to the product rule is the romantic tri-
angle. Imagine a balanced triad in which Paullikes Mary, Peter likes
Mary, and Paullikes Peter. If in the specific context Paul's and Peter's
affections toward Mary could be interpreted as romantic love, it is
doubtful whether the balanced triad would represent a paragon of har-
mony or stability contrary to a strict balance-theoretic prediction.

When Imbalance Creates Tension: The Case of Logical Inconsistency

Several authors noted the insufficiency of "sign imbalance" in pre-
dicting imbalance effects (cf. Eiser, 1980, p. 33; Pepitone, 1966, p. 266;
Stroebe, Thompson, Insko, and Reisman, 1970) and suggested that, in
addition, the cognitive elements in a dyad or a triad need to be perceived
as mutually relevant. I concur but take the additional necessary step of
defining relevance logically as an implicational relation that interprets
imbalance in a relevant structure as a logical contradiction. To illustrate
this point consider the study by Heider (1958, pp. 176-177) in his at-
tempt to induce imbalance: Subjects were presented with the following
imbalanced situation that contained a positive sentiment relation, a neg-
ative sentiment relation, and a positive unit relation:
Bob thinks Jim is very stupid and a first dass bore. One day Bob reads some
poetry he Iikes so weIl that he takes the trouble to track down the author in
order to shake his hand, he finds that Jim wrote the poems. (p. 179)

Subjects' responses to the vignette were pretty much as balance theory

would predict. Some individuals thought that Bob would change his
mind about Jim, others thought that he would change his opinions of
the poems, others, that he would question Jim's authorship, and others
thought that Bob would feel confused and not know what to do.
Now a elose look at the foregoing vignette reveals a critical element
of logical contradiction. Furthermore, subjects' responses to the vignette
A BRIDGE TO CONSISTENCY THEORIES 91

reflect reactions to the main epistemic consequence of contradiction: the

undermining of belief. Specifically, most subjects probably reasoned
that (1) if somebody is stupid (A), he is incapable of writing excellent
poetry (B); (2) Jim is stupid (A), yet (3) he does apparently write excellent
poetry (not-B). Thus, we have a straightforward contradiction between
B as entailed by (1) and (2) and not-B, as given in (3). In keeping with the
epistemic analysis it is thus natural for subjects to react to the contradic-
tion by assuming that Bob is confused and does not know what to do
and to feel that Bob will modify his opinions so that the inconsistency is
removed.
Our analysis can readily handle the various exceptions to the bal-
ance rule considered earlier. For instance, the liking of chieken does not
implyan agreement with chieken's gastronomie preferences. Hence, in-
formation that one likes chicken and dislikes chieken feed does not
involve a contradiction and should not result in instability or confusion.
Similarly, we can account without difficulty for the romantic-triangle
exception to the balance rule. The concept of romantic love may imply to
manya strong dislike for one's rivals. The knowledge that p is o's riyal
yet, at the same time, likes 0 contains a contradietion that may foster
uneasiness and instability. A highly similar interpretation of the roman-
tic triangle was given by Abelson (1968, p. 127).
Klar and Pol (1980) recently tested the more general hypothesis
based on the foregoing line of reasoning. According to these authors,
when a given sentiment term implies the subject's agreement with the
object's tastes and preferences, "imbalanced" triads based on such senti-
ment convey a disagreement and, hence, contain a contradiction. How-
ever, when a sentiment term has no agreement implications, the perti-
nent imbalanced triads do not involve a contradietion. Therefore, if
imbalance effects like unpleasantness and instability are mediated by a
contradietion, imbalanced triads using high-agreement sentiments
should result in imbalance effects, whereas imbalanced triads using low-
agreement sentiments should not.
In the first phase of Klar and Pol' s research, native speakers of
English judged the extent to whieh different sentiment terms implied
agreement; that is, the extent to which a given sentiment term feIt by p
toward 0, and 0 toward x, implied the same sentiment by p toward x.
Based on this preliminary stage, various sentiment terms were classed
into the categories of strong-, medium-, and weak-agreement implications.
Examples of sentiments with strong-agreement implications are "ap-
proves of," "venerates," "is proud of"; sentiments with medium-agree-
ment implication are "likes," "cherishes," "is impressed by"; and those
with weak-agreement implication are "fandes," "longs for," or "is
charmed by."
92 CHAPTER5

In a second phase of the study, a different group of subjects was

exposed to balanced triads with sentiment terms in each of the implica-
tional categories. The dependent variables were those traditionally em-
ployed in balance-theory studies: (1) the rated pleasantness of the situa-
tion depicted, (2) its rated reasonableness, and (3) its stability. As
expected, significant implication x balance interactions appeared on all
the dependent measures. More specifically, balance effects, that is, dif-
ferences between balanced and imbalanced triads, were strongest with
sentiments in the strong-implication category and weakest with those in
the low-implication category. In other words, imbalanced triads tended
to be perceived as unpleasant, unreasonable, and unstable to the extent
that they contained an element of contradiction that, according to our
epistemic analysis, fosters confusion and undermines subjective
knowledge.
To conclude, we argue that the usual balance-theory analysis, in
which psychological tension, unpleasantness, or instability are pre-
dicted from the signs of relations connecting elements in a cognitive
structure, often falls to accomplish the task. As we have seen, this typ-
ically occurs when an imbalanced set of relations does not contain a
logical contradiction. On the other hand, where imbalance results in
tension, contradiction seems invariably involved (see also Runkel and
Peizer, 1968; Insko, 1984). At this juncture, I should like to emphasize
what I am not saying: I am not suggesting that contradiction as such is a
sufficient condition for tension or unpleasantness. In addition, a motive
base needs to be present and frustrated by contradiction. This point will
be discussed more fully in a subsequent section of this chapter.

THE CONGRUITY MODEL

In Osgood and Tannenbaum's (1955) congruency model, the cogni-

tive elements of interest include: a communicative source (s), a concept
(c), and an associative (positive, approving) or a dissociative (negative,
dissapproving) assertion made by the source about the concept. Astate
of congruency obtains in four cases: (1) when a positively valued (i.e., a
liked) source makes an associative assertion about an equally positive
concept, or (2) makes a dissociative assertion about an equally negative
concept; (3) when a negatively valued source makes an associative asser-
tion about an equally negative concept, or (4) makes a dissociative asser-
tion about an equally positive concept.
It is of interest to note that the inconsistency that underlies in-
congruency effects is logical inconsistency or contradiction. Consider a
case in which a positively valued source, say, the Chicago Tribune,
makes an associative (i.e., an approving) assertion about a negatively
A BRIDGE TO CONSISTENCY THEORIES 93

valued concept, say, a nuc1ear arms race. The contradiction here is be-
tween the subject' s belief that a nuc1ear arms race is bad and the propo-
sition that it is good (i.e., not bad) made plausible via an endorsement
bya positive source. By counterdistinction to the original formulation of
congruency theory, the present emphasis is on the reliability (or epis-
temic authority) of the source, rather than on the sheer positivity of
sentiment (or attitude) one might feel toward it. For instance, one could
have an extremely positive attitude toward one' schild yet consider it
rather inexpert in the domain of insurance policies. Consequently, an
associative assertion made by one' schild about a negatively valued con-
cept in the domain of insurance is likely to occasion little upset because
of its failure to make plausible a belief contradictory to one's own
judgment.

ROSENBERG' 5 AFFECTIVE-COGNITIVE CONSISTENCY THEORY

According to Rosenberg (1956, 1960), attitudes consist of affect to-

ward objects believed to be instrumentally related to other objects of
affective significance, such that change in the affective component
should produce change in the cognitive or instrumental components
and vice versa. For example, consider a medical student whose negative
attitude toward socialized medicine also involved the belief that it leads
to debasement of medical standards and a loss of professional freedom.
According to the theory, a shift in one of the latter beliefs, for instance,
that no debasement of standards is involved, is also likely to occasion a
shift in the attitude.
These notions are readily interpretable in terms of logical inconsis-
tency: A positive evaluation of a given entity may imply to many people
that the entity has desirable effects (e.g., a good person is one who
produces good deeds). If so, it would be simply contradictory to hold
simultaneously that x is extremely worthwhile yet all its effects are ex-
tremely negative. Such inconsistency could weIl occasion reassessment
and cognitive change of the type referred to by Rosenberg.

ABELSON AND ROSENBERG'S PSYCHo-LoGIC

The issue of logical versus psychological implication appears with

acuity in Abelson and Rosenberg's (1958) discussion of "psycho-logic."
This theory contains three general categories of cognitive elements (i.e.,
elements that may enter into cognitive organizations of various types):
(1) actors, inc1uding oneself, other people, and groups; (2) means, inc1ud-
ing actions and instrumental responses; and (3) ends, that is, outcomes.
The elements can be linked by positive, negative, null and ambiva-
94 CHAPTER5

lent relations. An element pair linked by a relation institutes a basic

sentence, and there are eight mIes governing implications among the
basic sentences. Such mIes are referred to as psycho-logic, a label intend-
ed to set them apart from more purely logical algorithms. For example,
consider the rule: A n Band B n C implies A p C, in which n and p refer to
negative and positive relations, respectively. This mle can be illustrated
by propositions of the following content:
(1) India (A) opposes (n) the United States' Far Eastern policy (B). (2) The
United States' Far Eastern policy (B) is directed against (n) Communism (C).
Therefore, (3) India (A) is in favor of (p) Communism. (Zajonc, 1968, p. 322)

According to Abelson and Rosenberg (1958),

such reasoning would mortify a logidan, yet it can be found in much this
form inside of millions of heads. Thus, we speak of the formal system as
psycho-logical rather than as logic. (p. 159)
But, according to the present interpretation, these mIes of psycho-Iogic
are not incompatible but rather irrelevant to logic. SpecificaIly, the psy-
chological mIes derive from a person's semantics, that is, with the way
in which he or she interprets various concepts. Thus, the first two prop-
ositions in Zajonc's example could readily imply the third for an individ-
ual who happened to assume that (1) being "anti-anti"means being "in
favor", (2) being "directed against" means being "anti," as does being
"opposed to." The first proposition states that A is anti Band the sec-
ond, that B is anti C. Combining and substituting, it follows deductively
that A is anti-anti C which, by assumption, is in favor of C, which is just
the conclusion obtained. Of course, a person could semantically dis-
agree with our subject, whether being "anti-anti" is really equivalent to
being "in favor," or whether being "directed against" really means being
"anti." It could even be said that such aperson' s cognitions are more
refined and differentiated than our subject's. But one should be careful
not to confuse differences in the contents of peoples' cognitions with the
appropriateness of their deductions. In particular, our subject's deduc-
tions in the example just given are quite appropriate or logical, and the
semantic disagreements one might raise are entirely extralogical.
Abelson and Rosenberg's (1958) system makes it particularly appar-
ent that cognitive imbalance actually represents a logical contradiction.
Consider the case in which, for some reason, the sentences A p C and A
n C are both implied. According to Zajonc (1968), "if this occurs, those
two contradictory implications are combined into one, in which A is
connected with C by an ambivalent (a) relation (A a C)" (p. 322). In
interpreting this, note first that A p C and A n C are contradictory in a
strict logical sense: p represents a non-n, therefore, p and n are a contra-
dictory pair. Such astate of affairs could weIl give rise to an ambivalence
relation; that is, to unclarity, confusion, and doubt whether p or n are
A BRIDGE TO CONSISTENCY THEORIES 95

the case. As noted throughout, logical inconsistency undermines the

experience of knowledge, hence, it fosters ambiguity or ambivalence
represented by the (a) relation discussed by Abelson and Rosenberg. To
conc1ude, in the psycho-10gical system imbalance is logical inconsisten-
cy. This point was noted by Pepitone (1966), who remarked that "the
Abelson and Rosenberg imbalances approach being logical contradic-
tions of the type A implies non-A" (p. 265).

ARONSON'S DISCONFIRMATION OF EXPECTANCY FORMULATION

Aronson (1968, p. 9) proposed a rule of thumb whereby dissonance

is aroused whenever an expectancy is violated. Now it should be c1ear
that violated expectancy contains an element of logical contradiction:
One expects A, and not-A turns out to be the case. Thus, Aronson also
sensed that disconfirmed expectancies capture the essence of dis so-
nance, which, according to the present analysis, requires the element of
contradiction.

IRLE' 5 REVISED DISSONANCE THEORY

In Irle's (1975; Irle and Montman, 1978) revised theory of cognitive

dissonance, a dissonant state is explicitly assumed to result from logical
contradiction. According to Irle, whenever the occurrence of two cogni-
tions X and non-Y contradicts a hypothesis held by a given person,
cognitive dissonance will occur (1975, p. 312). Granting that what Irle
means by a hypothesis is an implicational statement whereby if X then
Y, the occurrence of X and non-Y logically contradicts the hypothesis.

MCGUIRE' S CONSISTENCY THEORY

McGuire's (1960a,b,c) theory of cognitive consistency strongly sug-

gests that cognitive consistency and inconsistency are logical in charac-
ter. According to the theory, if several inconsistent beliefs are elicited
together, the individual will become sensitized to the inconsistency and
change his or her beliefs so that a greater degree of logical consistency
will result. McGuire calls the whole process of movement toward a
greater degree of logical consistency, following the relatively simultane-
ous elicitation of contradictory beliefs, the Socratic elfect (Insko, 1967, p.
103).

RECAPITULATION

The consistency literature contains positions that explicitly portray

cognitive inconsistency in terms c10sely approaching logical inconsisten-
96 CHAPTER5

cy, as, for example, Abelson and Rosenberg's (1958) imbalances, Aron-
son's (1968) disconfirmations of expectancies, Irle's (1975) dissonances,
and McGuire's (1960a) inconsistencies. Alternative positions interpret
cognitive inconsistency as fundamentally psychological, a term usually
taken to denote incompatibility with a logical interpretation. But accord-
ing to the present analysis, any cognitive inconsistency is at once logical
and psychological. The logical element pertains to the contradictory
structure (of the A, not-A type) necessarily involved in any cognitive
inconsistency. The psychological element pertains to the origins of be-
liefs in the contradictory propositions: the accessible semantic assump-
tions from which such propositions might be derived, and the motiva-
tional factors that inhibit the generation of possible alternative
propositions. In other words, cognitive inconsistency can exist only if
the logically contradictory elements are believable or plausible to the
individual and believability or plausibility are determined by psycholog-
ical forces. Hence, logical and psychological elements, far from being
incompatible, are in fact inextricably coexistent in any case of cognitive
inconsistency .

THE MODE OF RESOLUTION PROBLEM

Contributors to the cognitive consistency literature viewed it as a

task of some importance to specify different modes of inconsistency
resolution and the conditions affecting the choice of a particular mode.
According to the present approach, the cognitive inconsistency to be
reduced is actually a logical contradiction of an A, not-A type. Such
problem definition makes it obvious that the only way in which an
inconsistency can be resolved is by negating or denying at least one of its
horns. A and not-A cannot both be true. At least one must be false;
therefore, one must be negated or denied. In this sense denial underlies
all modes of inconsistency resolution rather than constituting, as has
been often implied, merely one of several alternate modes. Our analysis
also suggests the direction that denial attempts are likely to take, notably
toward the less strongly believed (and, in this sense, the more "denia-
ble") of the contradictory propositions. In turn, the degree of belief
strength in a proposition is likely to be affected by the type of motivation
and cognitive capability factors outlined in the lay epistemic theory. I
will now attempt to substantiate these notions by reference to the rele-
vant theoretical and empiricalliterature.

The Place of Denial among Modes of Inconsistency Resolution

Abelson (e.g., 1959, 1968) proposed a set of resolutional modes,
including "denial," "bolstering," "differentiation," "transcendence,"
A BRIDGE TO CONSISTENCY THEORIES 97

and "rationalization." In this scheme, denial is considered one among

several resolutional modes of equal status. Instead, I shall argue that
denial (or negation) underlies each and every mode in Abelson's classifi-
cation. Consider differentiation, for example. This mode consists of "a
splitting of an element into two parts: A 'good' part and a 'bad' part
(Abelson, 1968, p.120). Consider a person exposed to an inconsistency-
producing statement that the Pope endorses LSD. One way of dealing
with this situation would be by positing that "the concept of LSD had
some good aspects ... (say, medical use in certain appropriate cases)
and that "the Pope endorsed these good aspects" (Abelson, 1968, p.
120). But note that the original inconsistency in this ca se must have been
between the assumption that the Pope condemned LSD universally, and
the newly introduced cognition that he endorsed LSD at least on some
occasions. The differentiation resolution whereby the Pope endorses
some LSD aspects and condemns others amounts to a denial of the
former proposition in which the Pope condemns the drug totally.
A similar analysis applies to the "transcendence" mode, whereby
the seemingly contradictory propositions, for example, that killing is
unacceptable or killing is acceptable, are derived from the same over-
arching principle, for example, defending one's society. This renders the
killing of the enemy at war acceptable and the killing of fellow citizens
unacceptable, thus denying the original implication that all killing is
unacceptable.
The "bolstering" mode of inconsistency resolution involves the ad-
dition of cognitive elements consistent with one of the contradictory
cognitions. For example, a smoker aware of the relation between smok-
ing and cancer, hence, cognizant of the inconsistency between smoking
and the hope of staying in good health, may persuade hirnself that
"smoking is extremely enjoyable, good for his nerves, and sociaHy nec-
essary" (Abelson, 1959, p. 345). To recognize here the element of denial,
note first that the smoker's arguments are irrelevant to the relation be-
tween smoking and cancer and that they therefore do not actually deny
it. But they do deny one very unpleasant implication of such a relation;
that smoking makes one out to be an irrational, unthinking person
whose actions lack justifiable reasons. Such an unflattering cognition is
denied via bolstering, in which one convinces oneself and/or others
that, despite the possibly harmful aspects of smoking, one's choice to
keep the habit is based on legitimate considerations and reflects a weH-
reasoned decision rather than an unthoughtful foHy.
Finally, resolving inconsistency via the "rationalization" mode in-
volves the construction of "some links of relations with special proper-
ties" (Abelson, 1968, p. 121). Consider the prototypical imbalanced sen-
tence, "A Good Actor engaging in a Bad Action." This could be resolved
by three subvarieties of rationalization: (a) "Reinterpret final goal," (b)
98 CHAPTERS

"Accidental by-product," and (c) "Find the prime mover." The three
differ in the location in a chain at which new material is added. In (a),
the Bad Action is seen to lead ultimately to a Good Goal, thus yielding a
construction in which the Good Actor is pursuing a Good Goal. In (b),
the Good Actor has had bad luck while pursuing a Good Action. In (c), a
Bad Actor is controlling the good actor with the upshot that the Bad
Actor is really responsible for the Bad Action (p. 121).
To identify the element of denial embedded in rationalization, con-
sider carefully the inconsistency in the foregoing example. The sentence
11A Good Actor engaging in a Bad Action" does not, in and of itself,
signify an inconsistency, but it could be considered evidence for the
sentence A Good Actor intending a Bad Action." This already suggests
11

an inconsistency for the notion of a good actor normally implies a well-

intentioned actor; hence, a good actor whose intentions are bad is a
contradiction in terms. From this perspective, the three varieties of ratio-
nalization can be viewed as denials of the statement that the gaad actor
intended a bad thing. In (a), the badness of the actor's intention is denied,
and the intention is depicted instead as ultimately good. In (b), the
action's intentional nature is denied, instead the action is depicted as
accidental. In (c), the good actor's bad intention is denied, instead the
intention is ascribed to a bad actor whose control over the good actor
cannot be helped.
In short, denial is essential to a11 inconsistency resolution. The vari-
ous resolutional modes in the cognitive consistency literature may repre-
sent a possible c1assification of ways in which a proposition (sentence,
cognition) can be denied. More interestingly, the present epistemic the-
ory suggests the conditions under which a particular mode of resolution
is likely to be chosen. Generally, this may depend on considerations of
cognitive capability and epistemic motivation as previously outlined.
For instance, whether differentiation, transcendence, or bolstering
would be adopted as a mode of resolution may depend on the mental
accessibility of ideas (Higgins and King, 1981) relevant to each mode.
When the individual' s dominant epistemic motivation is a need for non-
preferential c1osure, accessibility may be the primary determinant of
mode selection. Under such conditions the individual may care only that
the inconsistency be resolved, and not particularly haw. For instance, a
person interested in knowing the location of a supermarket in a strange
town might initially start out with contradictory cognitions but qUickly
discard or deny some of those, in light of newly accessible information.
The case may be very different when the individual' s predominant
motivation is need for specific c1osure. This may render some cognitions
hard to abandon, and others hard to accept. For instance, a given cogni-
tion may pertain to one' s area of specialization, or qualify as a topic on
A BRIDGE TO CONSISTENCY THEORIES 99

which any sophisticated person should be knowledgeable. Total denial

of such cognition may therefore carry unpleasant implications for one' s
self-esteem and may likely not take place. Under the circumstances, one
may opt for a face-saving or a compromise resolution (e.g., differentia-
tion or transcendence), whereby one's original cognition was not totally
wrong, albeit possibly less general than originally presumed.
In other circumstances, there may exist strong motivational reasons
to abandon one's initial cognition. For example, one may value conver-
sion to a new idea as such and may experience pride on having irn-
proved oneself (Ross and McFarland, 1988) or having seen the light.
Such needs for specific closure, concerning conversion or change, might
render total denial more attractive than partial denial or a compromise
solution. These notions, although presently speculative, could be pur-
sued profitably in future research.
An additional point needs to be made: It is not assumed that lay
persons are consciously aware (Nisbett and Wilson, 1977) of the cogni-
tive or motivational factors affecting their preferences among modes of
inconsistency resolution. What they may be aware of is having denied
the less subjectively valid or credible of the contradictory propositions.
Indeed, the cognitive consistency literature contains ample suggestions
that, in resolving an inconsistency, persons change (i.e., deny) the cog-
nition that is less resistant to change, or that is held with the lesser
degree of belief strength (Walster, Berscheid, and Barclay, 1967; Wick-
lund and Brehm, 1976, p. 125).
In turn, belief strength in a proposition may be affected, probably
unconsciously, by such cognitive factors as the accessibility of plausible
alternative propositions (Gotz-Marchand, Gotz, and IrIe, 1974), or moti-
vational factors like the importance to oneself of having knowledge on a
given topic (representing one's need for nonspecific closure) (Hardyck
and Kardush, 1968, p. 687), or the desirability to oneself of a given
knowledge, representing one's need for specific closure (Wicklund and
Brehm, 1976, pp. 134-135).

THE DEDUCTIVE ELEMENT IN COGNITIVE CONSISTENCY THEORIES:

A SUMMING Up

Just as in attribution theory, the deductive aspect of hypothesis

validation and invalidation plays a prominent, if an insufficiently recog-
nized, role in cognitive consistency theories. In attribution, however,
the primary emphasis has been on the validation of causal hypotheses
from consistent evidence. By contrast, in cognitive consistency theories,
the emphasis has been on the consequences of inconsistent evidence, or
logical contradiction that can only be resolved if at least one of the
100 CHAPTER5

contradidory elements is denied. Indeed, denial was shown to underlie

each of the previously identified modes of inconsistency resolution,
rather than representing merely one resolutional mode among many.
Denial has been conceived of as a withdrawal of belief from a given
statement. Therefore, in resolving a cognitive inconsistency, denial at-
tempts are likely to be aimed at the less believable of the contradictory
propositions. In acordance with the present theory, the degree of belief
strength is affected by various cognitive and motivational fadors of epis-
temic significance. Indeed, the cognitive consistency literature impli-
cated just such factors in cognitions' resistance to change (or to denial) in
the course of attempts at inconsistency resolution.

THE CONTENT ASPECT OF COGNITIVE CONSISTENCY THEORIES

My discussion has emhasized that the various cognitive consistency

models address the same kind of inconsistency, notably, logical contra-
diction. In the sections that follow, I will suggest that differences among
such models are ones of cognitive content. Conte nt specificity of con-
sistency formulations has been manifest in two separate features: (1)
content-specific conditions for inconsistency outlined in the different
models and (2) attempts to map out different content schemata in a lay
person' s repertory.

CONTENT-SPECIFIC DEFINITIONS OF INCONSISTENCY

Major consistency formulations accord special status to content

categories assumed to constitute the defining preconditions of the in-
consistent state. In balance theory, for example, the defining precondi-
tions of inconsistency (or imbalance) concern cognitions about one's
own seH (p), another person (0), an impersonal entity (x), and about
positive or negative sentiments and unit relations.
Although balance theory was content-bound to begin with, dis so-
nance theory was originally stated in content-free terms. In the original
formulation of the theory (Festinger, 1957), a dissonant relation was
defined as one in which the obverse of one cognitive element follows
from the other, without specifying the nature or contents of the ele-
ments involved. However, Brehm and Cohen (1962) interpreted disso-
nance in terms of cognitions associated with "the decision context," an
interpretation seconded by Festinger (1964). In addition, it became dear
that simply making adecision is not enough. One has to possess the
additional cognitions of "having exercised one's volition," of "being
committed to the decision," and of "the decision's yielding highly nega-
A BRIDGE TO CONSISTENCY THEORIES 101

tive consequences" (cf. Brehm and Cohen, 1962; Collins and Hoyt, 1972;
Cooper and Fazio, 1984; Festinger, 1964). These conceptual develop-
ments in the dissonance framework signify the linkage of an initially
content-free theory to a specific cognitive cluster concerned with unfor-
tunate decisions.

MApPING OUT ATTEMPTS

Several contributors to the cognitive consistency literature seemed

weIl aware of the content specificity of major consistency formulations.
Rather than focusing on any one formulation, these authors opted,
therefore, for lists of content schemata that presumably map out a lay
persons' phenomenology. In this vein, DeSoto and Albrecht (1968) con-
sidered beside the ''balance schema" such cognitive schemata as "tran-
sitivity" or "linear order." Although these have not been generally
thought of as instances of the consistency principle, it is easy to show
that they are. For example, DeSoto and Albrecht (1968) discussed the
problem of disproving the hypothesis of linear ordering by logically
adducing inconsistent evidence. Alternative minisystems operated by
the principle of logical consistency are the "implicational molecules"
(e.g., romantic rivalry) of Abelson (1968, pp. 133-139) and the "cogni-
tive structures" (as the virtue-is-rewarding rule) of Pepitone (1966, p.
324).
In short, any such cognitive molecule can be thought of as a knowl-
edge structure with a particular content. That content defines what is
unique about the molecule. In addition, however, the molecule abides
by principles that apply to all knowledge structures; that is, are non-
unique to this particular molecule. Take the balance case, for example.
As I have tried to show, in many instances an imbalanced situation
represents a logical inconsistency that may produce the typical im-
balance effects, such as confusion, tension, or instability (Klar and Pol,
1980). In those instances, the effects of imbalance are nonunique; simiIar
effects should manifest themselves upon inconsistency among any set of
cognitive elements, not only the specific elements (p, 0 ,x) used to define
balanced and imbalanced states.
But in another sense, the balance schema represents a particular
knowledge structure whose unique content may be "good things going
with other good things, and bad things going with other bad things."
Such a schema may represent a specifically preferred closure, that is, a
state of affairs most persons would find desirable and satisfying (per-
haps because of its connotation of a "just world" (Lerner, 1980). In the
latter sense, an imbalance need not refer to logical inconsistency, but
rather to a (consistent) state of affairs that is contrary to a desirable
102 CHAPTERS

closure, say, a good thing linked with a bad thing. In such a case, one
would expect the situation to engender negative affect, but not neces-
sarily confusion and doubt of the sort that may characterize cases in
which imbalance represents a logical inconsistency.
It should be noted that not all cases of imbalance (as formally de-
fined) may represent equally undesirable closures. For instance, a triad
with three negative signs (p does not like 0,0 does not like x, and p does
not like x) seems less unpleasant than one with two positive and one
negative sign (p likes 0, 0 likes x, p does not like x); in fact, the former
situation may be quite pleasant because it implies that one of a person's
foes dislikes the other, and, hence, might inflict harm upon hirn or her,
which person might find desirable. Thus, two situations that are equally
imblanced by a strict application of the product role may not represent
equally undesirable closures.

THE MOTIVATIONAL ASPECT

The various consistency theories are generally considered motiva-

tional in flavor. An inconsistency is assumed to result in an aversive
experience and, therefore, to prompt its own reduction. But beyond this
common denomination, considerable debate has evolved around the
motivational status of the inconsistency construct. Some of the central
issues raised were: (1) Whether inconsistency is a universal drive and, if
so, what might be its underlying nature and its intensity? (2) Whether
inconsistency is an autonomous drive, or do its motivating properties
stern from conjunction with other needs or motives? (3) Whether incon-
sistency is a unitary motivational concept, or can there be different types
of inconsistency with varying motivational properties?

INCONSISTENCY AS A UNIVERSAL DRIVE

Heider (1946, 1958) is generally counted among those who regard

inconsistency as a universal drive. In accordance with the Gestaltist
position, Heider located imbalance in the stimulus field and viewed
inconsistency as a motivation in itself (Wicklund and Brehm, 1976, p.
253). Conceming the intensity of balance strivings, several writers (e.g.,
Freedman, 1968, p. 502; Zajonc, 1968, p. 347) agreed that they are of a
rather weak intensity and resemble preferences rather than driving
forces.
Assuming that consistency striving represents a universal drive,
some writers attempted to reduce it to more basic needs, for example,
the need toward "perfection" (Heider, 1958, pp. 212-217) or toward
A BRIDGE TO CONSISTENCY THEORIES 103

"simplicity" (Osgood and Tannenbaum, 1955, p. 43). Other writers (in

particular, Wicklund and Brehm 1976, p. 254) regarded consistency as a
secondary reinforcer or as a learned, functionally autonomous drive.
Festinger (1957) implied that cognitive dissonance is a general need
or drive to be included among a "psychologist's inventory of drives or
motives" (Wicklund and Brehm, 1976, p. 252). This does not necessarily
mean that any inconsistency among cognitions would be aversive. In
Festinger's (1957) formulation, the magnitude of dissonance is a func-
tion of the importance of the cognitions involved. According to this
view, inconsistency, as such, need not be very stressful; it would not be
if the inconsistent cognitions were unimportant to the individual.

INCONSISTENCY'S MOTIVATIONAL PROPERTIES AS DERIVED FROM OTHER

NEEDS

Festinger's formulation does not clearly specify what "importance"

iso A possible solution is to define importance in terms of other needs
(Brehm and Cohen, 1962, pp. 228-231; Wicklund and Brehm, 1976, p.
256). Accordingly, not just any inconsistent cognitions would possess
motivating properties, only those that are "psychologically relevant" for
an individual; that is, those that frustrate some prior motive associated
with the cognitions. Although intuitively appealing, this suggestion
leaves unspecified the reason why inconsistency should frustrate a prior
motive or how it could do so. I will return to this issue later.
Pepitone (e.g., 1966, 1968) proposed to construct a list of motives
that could be frustrated by an inconsistency. Prominent on that list is the
need of validation or reality testing. As Pepitone (1966) put it,
A compelling possibility is that unexpected events threaten the individual's
hold on reality. When the expectation and the disconfirming event represent
a contradiction of the type A, not-A, the individual becomes confused and
disturbed as to what is real. (p. 272)

However prominent, the need for validation is assumed to be only one

among several motivational ba ses for inconsistency effects: "Inconsis-
tency arouses other motives as weH as cognitive validity tensions, and
the research problem in the consistency model field is to isolate and deal
with the interrelations of these motives" (Pepitone, 1968, p. 326).

DIFFERENT TYPES OF INCONSISTENCY

Some writers implied the existence of different types of incon-

sistency with different motivating properties. According to Singer (1968,
p. 395), "there are different kinds of inconsistency. Some of these have
a non-cognitive component, they are disturbing, bothersome, and ac-
104 CHAPTER5

companied by negative affect." A somewhat similar position was

taken by Berkowitz (1968, p. 402), who mentioned the existence of "per-
ceptual" incongruencies as weIl as "defensive" dissonances. But let us
see what our epistemic model has to say about the motivational status of
inconsistency .

WHEN INCONSISTENCY LEADS TO AFFECT: A LAY EPISTEMIC

INTERPRETATION

First, I concur with Wicklund and Brehm (1976) that inconsistency

could derive its motivating properties from any conjoined motive. Sec-
ond, I also accept Singer's (1968) and Berkowitz's (1968) suggestions that
inconsistency with a non-cognitive component can be bothersome,
whereas perceptual inconsistency without such a component may have
no aversive consequences (see also Abelson, 1983). Third, I concur with
Pepitone's emphasis on validational needs. In fact, the reason why incon-
sistency may be capable of frustrating any motive sterns precisely from
its invalidating properties mentioned by Pepitone.
The key to the present proposal is a dear separation between the
deductive and motivational aspects of the epistemic process. As noted
earlier, inconsistency represents a logical contradiction that fosters a loss
of confidence in the contradicted proposition. Confidence loss is thus
conceived of as an inherent property of any inconsistency, one which
does not depend on the "importance" or "psychological relevance" of
the contradictory cognitions, or on motivational considerations in gener-
al. Suppose a resident of New York received two contradictory reports
about the weather in southern Florida. Under those circumstances, the
person may not be very sure about the actual state of affairs. This may be
due to the deductive consequence of logical inconsistency, notably, a
lowering of confidence.
However, confidence loss as such need not occasion upset. Our New
Yorker may care little about Florida and may plan, instead, a journey to
Mexico. In order for upset to be feIt, a motivational aspect needs to be
additionally engaged. According to the present analysis, two separate
motive categories can be frustrated by an inconsistency. (1) A person
with a high need for nonspecific dosure on a topic would be upset by an
inconsistency which undermines structure and fosters ambiguity. Our
New Yorker may still have a strong interest in Florida weather, if he or
she intended to visit there, and was considering the appropriate ward-
robe for the trip. Everyone would agree that confusing weather reports
in such a case are likely to be quite irritating. (2) A person with a high
need for preferential dosure on a topic would be upset by information
inconsistent with such a dosure. The New Yorker may hope for sunny
A BRIDGE TO CONSISTENCY THEORIES 105

weather in Florida, to allow for lovely afternoon walks along the beach.
Areport about torrential rains in the area would be inconsistent with
such a desirable prospect, would undermine one' s faith in it, and could
be, accordingly, quite upsetting.
In fact, the list of alternative motives possibly frustrated by an in-
consistency can be conceived of as a list of specifically preferred dos ures
that inconsistency may undermine. Contemporary dissonance research,
for example (Cooper and Fazio, 1984; Greenwald and Ronis, 1978) has
largely moved away from concern with cognitive inconsistency as such
to an almost exdusive foeus on the circumstances that produce upset in
the "forced compliance" situation (in partieular, a perceived personal
responsibility for negative consequences) and/or the underlying motive
for such an upset (e.g., self-esteem). Such a shift away from interest in
inconsistency is understandable if one' s focus is on dissonance as a
source of aversive affect: As previously noted, logical inconsistency only
serves to undermine one's confidence in a knowledge structure. Wheth-
er this should generate negative, positive, or no affect may depend
entirely on how desirable the undermined knowledge was to begin
with.
As we have seen, if a given bit of knowledge provided the desired
nonspecific or specific dosure, its undermining by inconsistent informa-
tion may indeed engender negative affect. However, if the undermined
knowledge was undesirable, negative affect should be reduced rather
than increased by the inconsistency. More specifically, inconsistency
should induce positive affect where individuals experienced high need
for the avoidance of dosure or for the specific avoidance of current
c1osure. Suppose our hypothethical New Yorker was loath to decide
between vacationing in Florida versus Mexico because each choice might
insult a dose friend. Under those circumstances, the person might expe-
rience ~onsiderable relief on receiving information inconsistent and,
hence, likely to ambiguate an impending decision. Alternatively, sup-
pose that the would-be vacationer had a strong bias toward Florida yet
the gloomy weather forecast counterindicated this choice. Under those
circumstances, the traveler would likely welcome an opposite (hence,
an inconsistent) forecast that may extend some hope to the desired alter-
native.

LAY EPISTEMIC INTERPRETATION OF THE CONSISTENCY MODELS:

A RECAPITULATION

In preceding seetions, I attempted to demonstrate that the lay epis-

temic model may furnish simple solutions to several persistent issues in
cognitive consistency theorizing. This analysis suggests that the incon-
106 CHAPTER5

sistency in cognitive consistency models is logical inconsistency or con-

tradiction. Accordingly, heretofore obscured properties of inconsistency
resolution become apparent; notably, that (1) it necessarily involves the
denial of at least one of the contradictory cognitions and that (2) denial is
likely to be aimed at the least strongly believed of those cognitions.
Furthermore, we have seen that the various consistency models are
content-speeific: They exemplify with different cognitive clusters the
same underlying regularities of the epistemic process. Although the
study of content may be useful and important, so is that of the process
that the contents commonly obey.
Finally, our epistemic analysis elueidates the motivational status of
the inconsistency construct. As such, inconsistency is seen to have no
motivational consequences; rather, its consequences are cognitive and
involve the undermining of confidence in a knowledge structure. The
affective consequences often assoeiated with inconsistency derive from
the epistemic motivations behind the undermined structure. Specifi-
cally, inconsistency may be aversive if it fosters confusion where definite
knowledge was strongly desired, and/or if it undermines a speeific
knowledge structure that the individual favored. Inconsistency may be
enjoyable if it undermines a distasteful closure, or helps one to generally
avoid closure when desired.

BRIDGING THE CONSISTENCY AND ATTRIBUTION PARADIGMS

We may now assess and consider the sense in which our lay epis-
temic theory integrates the attribution and consistency paradigms. By
integration I mean identifying the common elements addressed by both
paradigms and interpreting their major differential foei. The com-
monality of process underlying the phenomena dealt with by both the
attribution and cognitive consistency models is particularly striking in
regard to the logical component of epistemic activity. As previously
shown, the validational criterion of consistency or derivability plays a
central role in both frameworks. However, although attribution theory
focuses predominantly on the consequences of the consistent informa-
tion that enhances one's confidence in causa! assignments, cognitive
consistency theories focus on the consequences of the inconsistent infor-
mation that undermines judgmental confidence.
Furthermore, a major phenomenon of interest to attribution theo-
ries (e.g., Kelley, 1967) has been attributional confidence per se, assumed
to be high when a given informational pattern, for example, of distinc-
tiveness, consistency, and consensus, is consistent with a given attribu-
tion, such as attribution to an external entity. By contrast, the cognitive
consistency models professed no explieit interest in the inconsistency-
A BRIDGE TO CONSISTENCY THEORIES 107

engendered lowering of confidence. Instead, those models focused on

the affective consequences of inconsistency which were, in turn, of Httle
recognized relevance to attribution theories.
However, to the extent that an individual desired knowledge on a
given topic, under high need for nonspecific closure, or desired particu-
lar knowledge, under high need for specific closure, he or she would
presumably experience positive affect, given an informational pattern
consistent with such knowledge. Thus, attributional models could prof-
itably tap the positive, affective consequences of informational con-
sistency. Moreover, to the extent that the affective consequences of in-
consistency are indeed mediated by waning confidence, cognitive
consistency models could profitably incorporate confidence measures to
check on the postulated causal path from inconsistency, to confidence de-
cline, to affect (granting the appropriate motivations).
Thus, the present analysis suggests that the cognitive consequences
of inconsistency for subjective confidence should be distinguished con-
ceptually from the affective consequences. To reiterate, inconsistency
seems to have no unique affective consequences. Instead, these may
derive from the undermining of knowledge one might be motivated to
hold. However, both the logical element (of consistency/inconsistency)
and the motivational element seem essential to an understanding of the
epistemic process common to the attribution and consistency frame-
work.
Both the attribution and the cognitive models allude to several moti-
vational elements. However, attribution models fall to incorporate them
within the deductive (informational) elements of the process, whereas
cognitive consistency frameworks fai! to separate them sufficiently from
the deductive element of inconsistency that, per se, seems devoid of
motivational properties. Present elaboration of the motivational bases of
epistemic behavior aims to redress deficiencies in both types of frame-
work. This may afford a meaningful synthesis of recent motivational
research in attribution (Kruglanski, 1989), as weIl as the specification of
conditions under which inconsistency would be upsetting.
FinaIly, from the present perspective, differences between the vari-
ous attributional models and between the various consistency theories
are predominantly differences of specific cognitive content. Although of
considerable psychological interest, these are conceived of as "surface"
differences that conceal important "depth" commonaHties.

RESEARCH IMPLICATIONS OF THE PRESENT WORK

The present epistemic analysis does not merely relabel in more

abstract terms previous ideas and formulations; rather, it suggests new
108 CHAPTER5

research avenues for the cognitive consistency paradigm. In the final

section of this chapter, I will indicate in broad strokes what some of
those novel directions might consist of.
As already noted, inclusion of the variable of subjective confidence
allows an exploration of the notion that affectively aversive conse-
quences of inconsistency are mediated by lowered confidence in desired
knowledge. Furthermore, it should be possible to specify the deductive
and motivational conditions under which a given inconsistency, for ex-
ample, an imbalance, would yield the postulated cognitive and affective
phenomena. As Klar and Pol (1980) indicated, to produce cognitive "in-
stability" and perceived "unreasonableness," a set of imbalanced rela-
tions may have to involve a logical contradiction. To further result in
affective upset, such contradiction should frustrate the relevant motive
base for knowledge (i.e., a need for nonspecific or specific closure).
If the lay epistemic analysis is correct, needs for specific closure
other than those based on self-esteem (Cooper and Fazio, 1984; Green-
wald and Ronis, 1978) ought to result in arousal, upset, and cognitive
change typically associated with dissonance. This ought to occur wher-
ever knowledge structures compatible with those alternative needs were
undermined by an inconsistency. Furthermore, the epistemic model
may impose boundaries on the recognized conditions for dissonance
effects, notably the actor's self-perceptions as "personally responsible"
for "highly negative consequences" (Collins and Hoyt, 1972; Cooper and
Fazio, 1984). In particular, exceptions may occur where personal respon-
sibility and negative consequences failed to represent the underlying
epistemic constructs of inconsistency with a specifically preferred
closure. Thus, some individuals might find it highly desirable to be
responsible for negative consequences, for example, the devotees of a
Satanic cult, or persons who are motivated to arouse fear or appear
dangerous. Other persons may fail to be affected by variables that gener-
ally imply personal responsibility (e.g., choice, lack of behavioral con-
sensus); for instance, individuals with a strongly fatalistic outlook in
which all outcomes are predestined anyway.
The present epistemic analysis may specify the motivational condi-
tions under which selective exposure to information may or may not
occur (Frey, 1986; Kruglanski, 1989), or under which a given mode of
inconsistency resolution would be chosen (Abelson, 1968). Finally, the
epistemic analysis may suggest the conditions under which inconsistent
information will have a recall advantage. According to recent proposals
(Crocker, Hannah, and Weber, 1983; Devine and Ostrom, 1988; Hastie,
1984) items of information that are inconsistent with a prior impression
would lead to attempts at causally explaining the inconsistency (see also
Weiner, 1985). Such causal elaboration would establish pathways be-
A BRIDGE TO CONSISTENCY THEORIES 109

tween the inconsistent items and numerous other concepts that could
subsequently serve as retrieval cues thus improving recall. However,
according to the present analysis, various motivational and information-
al conditions might determine whether such elaborative process would,
in fact, occur. For example, in the absence of needs for nonspecific or
specific c10sure or under a high need to avoid c10sure on a topic, an
individual might be unmotivated to resolve the inconsistency and thus
be unlikely to engage in causal elaboration. Furthermore, even if moti-
vated to resolve an inconsistency, the individual might be able to access
information that affords an easy denial of one of the inconsistent items
(e.g., by derogation of the source). In those circumstances, too, no
causal elaboration, with consequent improvement in recall, is predicted
to take place. Thus, the epistemic model suggests novel approaches to
both c1assic and contemporary issues in the cognitive consistency
domain.

SUMMARY

The present chapter applied the lay epistemic analysis to various

consistency formulations. It was noted that all cognitive consistency
models contain an element of logical inconsistency. According to the
present approach, such inconsistency may undermine the individual's
confidence in a knowledge structure. Whether or not such inconsisten-
cy-based decrement in confidence would be aversive or upsetting would
depend on the individual's epistemic motivations. Specifically, inconsis-
tency may occasion negative affect where the individual either desired
nonspecific c10sure or a specific one undermined by the inconsistency.
In contrast, inconsistency may occasion positive affect whenever it un-
dermined a specifically undesirable c10sure or fulfilled aperson' s need
to generally avoid c1osure. From the present perspective, cognitive con-
sistency models and attributional formulations address different facets
of the same epistemic process. Consistency models focus on the affec-
tive consequences of inconsistent information, whereas the attributional
models stress the cognitive consequences of consistent information.
 6
Attitudes as Knowledge Structures

The notion of attitude has been among the most central and influential
constructs in contemporary sodal psychology. Writing in 1935, Allport
called attitude a "most distinctive and indispensable concept." Ever
since, active interest in attitudinal phenomena has been represented
in the sodal cognitive literature, and recent literature reviews (cf. Cial-
dini, Petty, and Cadoppo, 1981; Cooper and Croyle, 1984) attest that
attitude research today is among the most vital areas of inquiry in sodal
psychology.
Conceptual attempts to comprehend attitude formation and change
have taken many forms, and numerous theories of attitude change exist,
addressing different subsets of attitudinal phenomena. In this present
chapter, I propose a unified approach to the study of attitudes, based on
the theory of lay epistemics. My central thesis is that attitude can be
conceived of as a specific content category of judgment or a specific
knowledge structure. Therefore, the lay epistemic theory that addresses
the process, whereby all judgments (knowledge structures) are formed
and modified, is also a theory of attitude formation and change. I will
also attempt to show that such an approach orders a host of heretofore
disparate findings and generalizations in the attitude-change literature,
suggesting intriguing new directions for future research.

THE DEFINITION OF AN ATTITUDE

Anyone attempting to study the voluminous research literature on

attitudes is likely to be struck by the diversity of definitions that attempt
to pinpoint what an attitude is (cf. Fishbein and Ajzen, 1975; Him-
melfarb and Eagly, 1974; Kiesler, Collins, and Miller, 1969; Petty and
Cadoppo, 1981). Some authors propose multilevel definitions, whereby
an attitude consists of cognitive, affective, and behavioral aspects

111
112 CHAPTER6

(Breckler, 1984; Rosenberg and Hovland, 1960). Perhaps more typically,

attitude is taken to refer to the affective component (Thurstone and
Chave, 1929) sometimes assumed to derive from the cognitive compo-
nent (Fishbein and Ajzen, 1975).
The present approach treats attitude as subjective knowledge or
belief which content is affective or evaluative. Thus, when a person
makes an attitudinal statement whereby he or she feels positive about a
given object, or considers it to have positive qualities, he or she actually
is expressing an opinion or belief of an affective or evaluative type.
Viewed in those terms, attitudes are a subclass of beliefs and the generic
properties of beliefs should also apply to attitudes. This does not mean
that the distinction between beliefs with evaluative or affective content
versus those with nonevaluative or nonaffective content (the distinction
between "attitudes" and "beliefs") is irrelevant or useless, or that it
should be abolished. Indeed, differences in belief contents can have
important implications (e.g., for overt actions), and different content
categories of beliefs can be differentially related to other concepts in
people's phenomenologies. Thus, the present viewpoint is compatible
with cases in which attitudinal beliefs are based on nonattitudinal ones,
or with situations in which affective beliefs describing one' s emotional
states and/or experiences differ in their implications (e.g., for behavioral
intentions, or physiological reactions) from evaluative beliefs (Zanna and
Rempel, 1988). Such linkages or implications follow from the contents of
beliefs or from their subjective meanings to the individual. Even though
contents of attitudinal and nonattitudinal beliefs may differ, both are
belief types nonetheless, and, hence should exhibit the generic proper-
ties of beliefs as a category.
The notion that attitudes are a subtype of opinions or beliefs (i.e.,
they constitute kinds of subjective knowledge) is implicit in much of the
literature on the topic. Allport (1935, p. 806) in an early paper stressed
the guiding function of attitudes without which "the individual is con-
fused and baffled." Typically, review articles of attitudinal research in-
clude the terms attitudes and opinions in their titles and refer indis-
criminately to both. Other pertinent discussions lump together attitudes
and persuasion (e.g., see Petty and Cacioppo, 1981), a term which accord-
ing to arecent dictionary definition refers to the act of "causing someone
to believe something" (italics added). In arecent chapter, Fazio (1988)
explicitly argued that attitudes primarily fulfill a "knowledge function"
and represent summary evaluative judgments of objects or events which
aid individuals in structuring their complex social environments. In-
de~d, Fazio's programmatic research on attitudes (e.g., Fazio, Lenn, and
Effrein, 1983; Fazio, Powell, and Herr, 1983; Fazio, Sanbonmatsu,
Powell, and Kardes, 1985; Powell and Fazio, 1984) treats attitudes as
ATIITUDES AS KNOWLEDGE STRUCTURES 113

cognitive or semantic structures and employs methodologies developed

in the context of studying cognitive structures in general (Neely, 1977).
A central implication of the proposal that attitudes are a subcatego-
ry of beliefs or judgments is that they are formed and modified accord-
ing to rules governing the general judgmental process. The remainder of
this chapter purports to demonstrate that this is so.

THE DERIVATION OF ATTITUDES

According to the lay epistemic theory, judgments are validated in a

Iogical fashion; that is, a proposition tends to be believed if it can be
derived from previously accepted premises. Similarly, belief in a propo-
sition would be weakened or undermined if the latter is contradicted by
another credible notion. A glance at the literature on attitudes demon-
strates that the Iogical element has been persistently accorded a central
role in attitude formation and change.
Thus, in an early classic, Hovland, Janis, and Kelley (1953) pro-
posed that
a major basis for aeeeptanee of a given opinion is provided by arguments or
reasons which, aeeording to the individual's own thinking habits, eonstitute
"rational" or "logieal" support for the eonclusions. (p. 11)

More contemporary views, such as the "cognitive response" approach

(Petty, Ostrom, and Brock, 1981) or the "central versus peripheral route"
approach (Petty and Cacioppo, 1981), accord an equally important role
to Iogical processes. According to the cognitive response approach
when a person anticipates or reeeives a persuasive eommunieation, an at-
tempt is made to relate the information in the message (or the expected
message) to the pre-existing knowledge that the person has about the top-
ie .... In doing this, the person will eonsider a substantial amount of infor-
mation ... {which} may agree with the proposals being made in the mes-
sage, disagree, or be entirely irrelevant to the eommunication. (Petty and
Cacioppo, 1981, p. 225)

More formally, an agreeing thought (supporting argument) constitutes

evidence which, given the person's belief system, Iogically implies the
advocated position, whereas a disagreeing thought (counterargument)
is evidence which, given the person's beliefs, Iogically contradicts the
advocated position.
It is of particular interest to note that logical processes may playas
important a role in mediating various context effects on responses to
persuasion as they do in mediating responses to message contents. For
instance, consider such source attributes as a communicator's perceived
trustworthiness or expertise. Their pervasive impact on persuadability
114 CHAPTER6

could be mediated logically, proceeding from the premise that if credible

expert X proposed a given conclusion, the conclusion is probably valid.
Consistent with this interpretation is Eagly, Chaiken, and Wood's (1981)
attributional analysis of source-credibility effects. For instance, Eagly,
Wood, and Chaiken (1978) found that persuasion was less likely to occur
when a communicator' s advocacy of an attitudinal position could be
ascribed to audience pressures, suggesting a reporting bias, or to a strong
prior commitment to the position, suggesting a knowledge bias. Both
types of bias raise inconsistent alternatives to the hypothesis that the com-
municator' s statements reflected his or her true beliefs. Thus, source
effects are shown to follow the same deductive path as message effects,
in which a conclusion (e.g., concerning communicator's credibility) is
derived from consistent evidence and is undermined by inconsistent
evidence.
Further context effects could be mediated logically, given the sub-
ject's assumptions about the communication situation. For example, in a
study by Miller, Maruyama, Beaber, and Valone (1976), subjects were
more persuaded by a communicator who talked somewhat faster than
normal than by one talking at a normal rate. Miller et al. argued that
subjects inferred that a fast talker knows what he is talking about and this
is what accounts for the enhanced persuasion. In present terms, subjects
may assurne that if a communicator talks rapidly he or she probably is
knowledgeable, which, in turn, yields the next deduction that the mes-
sage is valid. Petty and Cacioppo (1981) viewed such context effects as
representing the peripheral route of pel'suasion. This is distinguished
from the central route, where contents of the message have the persua-
sive impact. By contrast, the present discussion stresses a commonality
between the central and peripheral routes, notably, that a deduction 01
attitudes occurs in both. A fuller epistemic analysis of the central versus
peripheral distinction is undertaken later this chapter. But let us now
consider further attitudinal phenomena explicable in terms of the notion
of deductive reasoning.

The Classical Conditioning of Attitudes

The acquisition of attitudes via classical conditioning (Staats, 1975;
Staats and Staats, 1958; Zanna, Kiesler, and Pilkonis, 1970) can be un-
derstood as the deduction of attitudes. In the classical conditioning para-
digm, an initially neutral stimulus is paired with an affectively laden
stimulus, and consequently comes to evoke an affective response or
attitude. For instance, in the study by Zanna et al. (1970), an initially
neutral word, "light," was paired with the onset or offset of an electric
shock. Under such circumstances, the bodily reaction to shock or the
ATIITUDES AS KNOWLEDGE STRUCTURES 115

annoyance experienced could be missattributed to the word "light," and

used as evidence from which a negative attitude toward this word could
be deduced. It is as if the person was saying, "I am annoyed whenever
'light' is flashed; somehow this word must really be annoying." Similar
attributional interpretations of classical conditioning phenomena have
been recently offered for a variety of findings incongruous with a purely
behavioristic interpretation (Testa, 1974).

Effects of Rewards
The positive effect on attitude change of rewards for statements
consistent with an attitude (cf. Scott, 1957) could also reflect adeduction
of one's attitude from relevant evidence. Such evidence could be the
experimenter' s own apparent attitude expressed by the reward he or she
sees fit to dispense. Imagine a person induced to articulate a given
position and rewarded by the experimenter for doing so. The reward
could be taken to signify the experimenter's approval for the position.
To the extent that he or she is perceived as a credible authority, a subject
could deduce that the position is valid and proceed to adopt it or sub-
scribe to it with greater force.

Logical Bases of Attitudes

Given that attitude is conceived of as evaluative or affective knowl-
edge, the question can be raised what evidence serves as a basis for the
deduction of attitudes. Recently, Zanna and Rempel (1988) proposed
that attitudes can be based on three classes of information: (1) cognitive
information, (2) affective information, and (3) information concerning
past behaviors. This classification is broader than Bem's (1972), in which
attitudes are deducible from overt behaviors performed under specific
reward contingencies and from internal cues, presumably of an affective
nature. Similarly, Zanna and Rempel's classification encompasses Fish-
bein and Ajzen' s (1975) proposal whereby attitudes are based on or
deduced from primarily cognitive information, notably utilitarian be-
liefs.
Zanna and Rempel' s suggestion that attitudes can be deduced from
a variety of sources is in keeping with the present approach, albeit, the
present proposal goes even further. Specifically, our LEGO metaphor of
human cognition, whereby anything can be implicationally linked with
anything else, suggests that, in principle, attitudes can be deduced from
any type of evidence, and that no classification or taxonomy, however
broad, is likely to do justice to the inexhaustible variety of evidential
categories from which different persons might infer their attitudes. For
116 CHAPTER6

instance, one could deduce one's attitude from the source's characteris-
tics (like credibility or trustworthiness), from a variety of message argu-
ments unrelated to utilitarian beliefs (e.g., those invoking absolute ethi-
cal standards), and from reactions of a referent audience to the
communicator's appeals.
In general, the way an attitude is subjectively defined, that is, the
meaning that the attitude concept connotes to a given person or group
ofpeople should constrain the types of evidence from which it can be
inferred. For instance, defining an attitude as an evaluative judgment of
the object (e.g., science is worthwhile, or nouvelle cuisine is tasty) should
render different evidence relevant to an attitude inference than defining
an attitude as the subject's affective response (e.g., I am thrilled by sci-
ence, or I have a passion for nouvelle cuisine). In other words, the use of
different terms in a given attitudinal proposition renders it likely that the
proposition would enter into different deductive linkages with other
concepts or categories; hence, that it would be derivable from different
types of evidence. These issues could be profitably explored in future
research.

The Relative Efficacy of Attitudinal Bases

Several authors have recently implied that some deductive bases of
attitudes are more effective in producing firm, resistant, and/or lasting
attitudes than other bases. For example, Petty and Cacioppo (1981) sug-
gested that attitudes acquired by consideration of message contents
(persuasion via the "central route") are more persistent than attitudes
based on the communication context (persuasion via the "peripheral
route"). From the present perspective, the firmness or persistence of
judgments, including attitudinal judgments should not universally de-
pend on specific contents of evidence from which the judgments are
deduced. Rather, judgmental firmness and persistence should depend
on the degree to which the evidence for the judgment is subjectively
compelling for the individual, uncontested by alternative evidence, or
the degree to which the individual is motivated to form a confident
judgment based on the evidence.
Thus, different types of evidence (e.g., associated with a message
content or message source) could be differentially compelling for differ-
ent individuals, thus giving rise to correspondingly less or more confi-
dent attitudes. Furthermore, different persons could have different de-
grees or types of epistemic motivation concerning a given topic. Such
persons should react to the same evidence, regardless of the nature of its
content or context, by forming differentially confident attitudes. For
instance, based on the same information, persons high on the fear of
AlTITUDES AS KNOWLEDGE STRUCTURES 117

invalidity should continue to exhibit low attitudinal confidence (Mayse-

less and Kruglanski, 1987a).
The present approach can be used to reinterpret findings typically
cited in support of the notion that persuasion based on the contents of a
message is more lasting than that based on the message context. For
example, in a study by Chaiken (1980), subjects read a persuasive mes-
sage portrayed as having high or low personal consequences for those
individuals. Some subjects received numerous cogent message argu-
ments, whereas other subjects received only few such arguments. Fi-
nally, some subjects were led to believe that the source of the message
was likeable, and others that the source was unlikeable. It was found
that in the high-consequences condition, subjects' attitudes were deter-
mined primarily by the number of issue-relevant arguments in the mes-
sage, and almost not at all by source likeability. By contrast, in the low-
consequences condition, attitudes were determined primarily by source
likeability, and the number of issue-relevant arguments produced no
significant effects. Furthermore, when retested ten days later, subjects
in the high-conseqences condition retained significantly more of the
original attitude change than subjects in the low-consequences condi-
tion.
But before concluding that Chaiken's (1980) findings reflect a great-
er persistence of attitudes induced by issue-relevant argument rather
than by contextual cues (like source likeability), it seems desirable to
explore the possibilities that (1) the degree of motivation to form an
attitude was greater in the high- versus low-consequences condition; (2)
the type of motivation was different (notably, to form a valid attitude in
the high versus low-consequences condition and to be liked by the source
in the likeable versus unlikeable source condition); and (3) the motiva-
tion present during retest was more similar to that during induction for
subjects in the high- (as compared to low-) consequences condition.
Each of the above motivational differences could account for the greater
endurance of attitude change in the high- versus low-consequences
condition.
It is noteworthy that the various motivational factors are, in princi-
pIe, unrelated to (or independent of) the question whether one's atti-
tude is deduced from message contents or contexts. For example, a
religious person might be highly motivated to form a valid attitude on a
given issue, yet proceed to derive such an attitude from an authoritative
religious source (e.g., Holy Scripture or a priest). Intuitively, at least, it
appears that attitudes formed in such a way may be as enduring as
attitudes derived via the processing of message contents. A fuller con-
sideration of motivational influences on attitude change is undertaken in
a later section of this chapter.
118 CHAPTER6

The Factor of Perceived Epistemic Authority

According to the present approach, processing information on
one's own, that is, attempting to evaluate the relative merits of various
content messages might not prove a very efficient means of acquiring an
attitude if one doubted one's ability to validly assess the evidence. Un-
der such conditions, ascribing a position to a revered authority might
prove a more efficient means of acquiring a confident and lasting atti-
tude on a given issue. Relevant to this contention is the experimental
work by Ellis (1984) described in Chapter 3. In that research, subjects
rated high on self-ascribed epistemic authority, that is, persons who
believed that they possessed considerable expertise in a given domain
(of mathematics or of interpersonal skills) benefited from experiential
learning more than subjects low on self-ascribed epistemic authority. By
contrast, in some conditions, subjects with low (versus high) perceived
authority benefited more from a "frontal lecture" in which identical
material was delivered by an external source.
According to the present interpretation, subjects with high-ascribed
authority in a given domain can readily deduce their conclusions from
their experiences (analyses, observations), whereas subjects with low-
perceived authority may prefer to base their deductions on external
experts. Although Ellis's dependent variable (efficacy of performance)
was measured immediately following the learning experience, the ob-
served differences might persist over time as weIl. At any rate, at least
short-term learning seems to be determined by one' s subjective confi-
dence in one's evidence rather than by the content-versus context based
nature of the evidence as such.
A somewhat similar implication emerges from early research on the
relative efficacy of drawing explicit conclusion in the message as op-
posed to allowing the recipients to draw their own conclusion (Hovland
and Mandell, 1942; Thistlewaite, de Haan, and Kamenetzky, 1955). It
appears that when the message is sufficiently understandable, or the
recipients are sufficiently intelligent to comprehend it, greater persua-
sion results when the communicator refrains from drawing an explicit
conclusion. However, when the message is not sufficiently understand-
able, greater persuasion results when an explicit conclusion is drawn. In
sum, because of an inability to make deductions from issue-relevant
arguments, and/or because of a lack of confidence in the ability to make
such deductions, one might prefer to derive one's attitudes from contex-
tual information, for example, concerning the expertise or trustworthi-
ness of the communication's source. In principle, such conditions need
not be correlated with low motivation to form attitudes on an issue, with
low personal involvement, or with lower resulting confidence in, or
persistance of, the attitudes formed.
ATIITUDES AS KNOWLEDGE STRUCTURES 119

Evidential Consistency and Attitudinal Confidence

According to the lay epistemic model, inconsistency among items of
evidence concerning a given judgment should undermine the person' s
confidence. Recent research suggests that this generalization holds for
attitudes as well. For instance, Norman (1975) demonstrated that atti-
tudes high in affective-cognitive consistency predicted behavior better
than attitudes low in affective-cognitive consistency. The issue of pre-
dicting behavior from attitudes is dealt with in a subsequent portion of
this chapter. For now, suffice it to note that lower predictability of be-
havior in the case of the affective-cognitive inconsistency could stern
from the lower confidence in one's attitude under those circumstances.
Similarly, research by Chaiken and Baldwin (1981) suggested that indi-
viduals low in affective-cognitive consistency may hold weaker attitudes
or be more "aschematic" with respect to the attitude object than individ-
uals high in such consistency. Finally, research by Lydon, Zanna, and
Ross (1985) suggested that attitude change is more persistent when cog-
nitive and behavioral bases for the attitudes are in agreement. Recently,
Zanna and Rempel (1988) theorized that persistence should be greatest
when cognitive, affective and behavioral bases of attitudes are consis-
tent (versus inconsistent).
The lay epistemic model suggests that the more confident one is
about one's attitude, for example, because of consistency among infor-
mational sources or as a result of the appropriate epistemic motivations
the greater one's resistance to persuasive communication. That will oc-
cur because inconsistency between cognitions is likely to be resolved by
denying, and in this sense "changing, " the less confidently held of the
contradictory cognitions. Persuasive communication directed against
one's initial attitude provides cognitions inconsistent with it. Hence, the
lower one's confidence in the attitude, relative to confidence in the new
cognitions, the more likely one is to change it in order to resolve the
inconsistency. The study by Muzfir (1985), reviewed in Chapter 3, is
consistent with this implication. As will be recalled, this study explored
the notion that the negative relation assumed to exist between attitude
extremity and resistance to change (Osgood and Tannenbaum, 1955) is
actually mediated by greater confidence with which more (versus less)
extreme attitudes are often held. Manipulating extremity and confidence
orthogonally, Muzfir (1985) found that more (versus less) confidently
held attitudes were more resistant to change, whereas more (vs. less)
extreme attitudes were not.
If a confident attitude or opinion is particularly resistant to change,
a person holding it might have greater difficulty in resolving an inconsis-
tency when such an attitude is contradicted by new information. In
turn, difficulty in resolving an inconsistency might prompt the individu-
120 CHAPTER6

al to engage in additional search for relevant information. Precisely such

a result obtained in research by Kruglanski and Mayseless (1987), de-
scribed in Chapter 3. In one experiment, Kruglanski and Mayseless
(1987) manipulated confidence in an attitude in favor of enrolling in
elementary school a child below the required age subsequently con-
fronted by inconsistent evidence concerning bed-wetting. As expected,
when subsequently given reason to question their attitudes subjects
with a highly confident initial attitude sought social comparison considera-
bly more than subjects with a less-confident attitude.
In sum, the notion that attitudes are validated logically orders in a
new way a wide variety of concepts and findings in the attitude change
area. These include message as weIl as source factors in persuasion,
effects of rewards and classical conditioning, behavioral, affective, and
cognitive bases for attitudes, and other variables. In particular, the pres-
ent approach suggests that factors often identified with the efficacy and
persistence of persuasive communication, such as the content of one's
evidence, having to do with message arguments or communicator char-
acteristics, may not play the role they are usually assigned. Instead,
alternative factors are identified as critical to persuasion and attitude
change, especially the individual's confidence in the attitude and vari-
ables affecting it. These novel implications of the present approach
could be profitably explored in subsequent research, but now let us turn
to factors affecting the grist for the individual' s logical mill, having to do
with saliency or accessibility of attitudinally relevant cognitions.

ATTENTIONAL Focus FACTORS IN ATTITUDE RESEARCH

According to present theory, the judgment finally reached depends

on the kinds of supportive or refuting evidence accessed by the knower
in the course of the epistemic sequence. Numerous phenomena in the
attitude-change literature attest to the effects of informational saliency or
accessibility, both of which may bring specific constructs to the individu-
al' s focus of attention. Consider the notion of "biased scanning" invoked
to explain the occasional positive effects on attitude change of incentives
offered for making a counterattitudinal statement. According to Janis
and Gilmore (1965) (pp. 17-18)
when a person accepts the task of improvising arguments in favor of a point
of view at variance with his own personal convictions, he becomes tem-
porarily motivated to think up all the good arguments he can, and at the
same time supresses thoughts about the negative arguments .... This "bi-
ased scanning" increases the chances of acceptance of the new attitude
position.

In present terms, by a selective retrieval process biased scanning in-

creases the momentary accessibility of evidence consistent with an atti-
ATIITUDES AS KNOWLEDGE STRUCTURES 121

tude and decreases the accessibility of inconsistent evidence. Both

effects should increase the likelihood of deriving the attitude, given the
apparent evidence.
Attentional focus effects may underlie the efficacy of refutational as
compared to supportive defenses that induce resistance to attack on
one's position. A refutational defense consists of an attack on a person's
position and a refutation of the attack. By contrast, a supportive defense
consists of arguments consistent with the person' s position. In a study
by McGuire and Papageorgis (1961), subjects were exposed to messages
attacking various cultural truisms (e.g., you should brush your teeth
after every meal, or mental illness is not contagious). Prior to that,
subjects were given an "inoculation" or a refutational defense consisting
of paragraphs presenting and refuting arguments against one of the
truisms to be attacked and a supportive defense with arguments in favor
of another truism. That subjects were significantly more resistant to
attack following the refutational as opposed to the supportive defense
may reflect the fact that the former defense helped either to retrieve
from memory or to currently form arguments inconsistent with the at-
tack; both ways may render such arguments more focalized in the per-
sons' attention.
In a study by Papageorgis and McGuire (1961), a refutational de-
fense successfully inoculated the subject against an attack unmentioned
in the defense. This may have occurred because the inoculation brought
to the subject's attention the general idea that attacks on one's truisms
are easily refutable.
Somewhat similar to the distinction between refutational and sup-
portive defenses is that between one-sided versus two-sided com-
munications (Hovland, Janis, and Kelley, 1953). In an early experiment
by Hovland, Lumsdaine, and Sheffield (1949), two-sided communica-
tions, in which counterarguments to the position advocated were men-
tioned and refuted, proved more effective than one-sided communica-
tions in producing attitude change for members of the audience initially
opposed to the position. By contrast, one-sided communications were
more effective for members of the audience initially in favor of the posi-
tion. Both findings can be explained in terms of the attentional focus
notion: The two-sided communication may have drawn attention to
refutations of subjects' position of which they may have been previously
unaware. This would tend to sway the initially opposed subjects toward
the position advocated, and the initially favorable subjects away from
that position.
A well-known attitude-change phenomenon explicable in terms of
the accessibility notion is the "sleeper effect" (Hovland, Lumsdaine, and
Sheffield, 1949). When a compelling message is linked with a "discount-
ing cue" (e.g., an untrustworthy source), its effect is suppressed result-
122 CHAPTER6

ing in decreased attitude change. However, if after aperiod of time a

dissociation occurs between the message and the discounting cue, great-
er attitude change may be evidenced. Opposite effects occur when the
message is initially linked with an "augmenting cue" (e.g., a highly
credible source). If dissociation between such a cue and the message
occurs, lowered attitude change may be evidenced at the delayed test. In
terms of the present discussion, dissociation between the cue and the
message renders inaccessible important (contextual) evidence against (in
the case of a discounting cue) or in favor of (in the ca se of an augmenting
cue) the initially advocated position, respectively resulting in increased
or decreased tendencies to accept the position.
Further attitude-change phenomena are explicable in terms of the
focus of attention. Thus, increased persuadability under conditions in
which the subject is distracted from listening to a persuasive com-
munication (Allyn and Festinger, 1961; Festinger and Maccoby, 1964) is
often interpreted as reflecting the prevention of subjects from coun-
terarguing against the position or, in present terms, as reflecting the
momentary inattention to refuting evidence that subjects might have
noted, if not for the distraction.
Attentional effects could underlie the findings of much recent re-
search (for a review, see Cialdini, Petty, and Cacioppo, 1981) that infor-
mation which people generate by themselves is a more important deter-
minant of persuasion than information furnished by others. By
definition, self-generated arguments are momentarily attended to by the
person, whereas arguments presented by others might not be (for perti-
nent research, see Eiser and Pancer, 1979; Eiser and Ross, 1977; Watts,
1977). Finally, consistent with the accessibility notion and the atten-
tional focus argument generally is Greenwald's (1968) proposal whereby
persistence of persuasion depends on the extent to which subjects' pro,
or con thoughts, elicited by the communication, remain salient over
time.
In sum, a variety of findings in the attitude-change literature can be
ordered in terms of the attentional focus construct: An attitude would be
held to the extent that evidence from which the attitude can be deduced
was mentally focal to the person, and evidence from which the opposite
attitude can be deduced was not. Thus, pervasive findings in the re-
search literature on attitudes are compatible with the present search-
light metaphor of human consciousness, whereby only a subportion of
our cognitive LEGO blocks is illuminated at any point in time. Our
beliefs, including the attitudinal ones, are determined by notions con-
tained in this illuminated portion of the field; that is, by ideas mentally
focalized at the moment. This view is compatible with the recent pro-
posal by Salancik (1982) that an individual can have more than one
AITITUDES AS KNOWLEDGE STRUCTURES 123

attitude toward the same issue or object. Different attitudes toward the
same object can be manifest on different occasions when different types
of attitudinally relevant information are temporarily salient or accessible
(Salancik and Conway, 1975; see also Zanna and Rempel, 1988).
The present section considered the extent to which accessibility of
confirming or refuting evidence may strengthen or weaken one' s atti-
tude on an issue. The extent to which the attitude itself is accessible may,
in turn, depend on its degree of prior consolidation or integration in
memory (cf. Fazio, Lenn, and Effrein, 1983; Fazio, Powell, and Herr,
1983; Powell and Fazio, 1984). This issue is discussed in a later section
dealing with the attitude-behavior relation.

EPISTEMIC MOTIVATIONS AND ATTITUDE CHANGE

The lay epistemic theory accords a central role to motivational influ-

ences on the formation and modification of beliefs. Specifically, persons'
tendency to freeze upon a given belief or to unfreeze it and be open to
alternative viewpoints is assumed to be affected by the several epistemic
motivations: the need for nonspecific elosure, the need to avoid non-
specific elosure, and the needs for specific elosure and the avoidance of
specific elosure. To the extent that attitudes are a subelass of beliefs or
knowledge structures they should be similarly affected by the same
epistemic motivations. Perusal of the attitude- change literature attests
that this may be so.

Need for Nonspecific Closure

Motivational effects occupy a prominent place in the functional the-
ories of attitude change (cf. Katz, 1960; Smith, Bruner, and White, 1956;
Sarnoff, 1960). Katz (1960), for example, distinguished four general
functions of attitudes: (1) knowledge, (2) instrumental (or utilitarian), (3)
ego-defensive, and (4) value-expressive. The knowledge function elose-
ly resembles our need for nonspecific elosure. This is quite elear in the
following quote from Katz (1960, pp. 175-176) whereby
individuals ... seek knowledge to give meaning to what would otherwise
be an unorganized, chaotic universe. People need standards or frames of
reference for understanding their world, and attitudes help to supply such
standards.

In elose affinity to our need for elosure construct, Katz's "need for
knowledge" is situation specific rather than general across sitations and
topics. Thus, people are not assumed to seek just any kind of knowledge
or attain a general"understanding of the world." Instead, they do want
to understand the events that impinge directly on their own life (Katz,
124 CHAPTER6

1960, p. 171). He further postulates that the need for knowledge is

aroused in circumstances in which existing attitudes seem inadequate
to deal with new and changing situations. Any situation which is ambiguous
for the individual is Iikely to produce attitude change. [A person's] need for
cognitive structure is such that [he] will either modify his beliefs to impose
structure or accept some new formulation presented by others. (Katz, 1960,
pp. 190-191)

Something like the need for dosure is implicit in the various theo-
ries of cognitive consistency dealt with in Chapter 5 (Abelson, Aronson,
McGuire, Newcomb, Rosenberg, and Tannenbaum 1968, Berkowitz,
1978). In those formulations, a breach of cognitive consistency is as-
sumed to result in an aversive experience, motivating its reduction via
attempts to restore consistency. Unlike the present theory, or Katz's
approach, however, the cognitive consistency formulations, as a body,
do not typically link consistency strivings to adesire for dear-cut knowl-
edge. As noted earlier, the motivational base stressed by the cognitive
consistency theorists (dissonance theorists, in particular) has more to do
with specific dosure needs (such as the need to maintain a positive self-
esteem); those are considered more fully in a subsequent section of this
chapter.
Empirical findings, interpretable in terms of the need for dosure
construct, were obtained in aseries of experiments by Tesser and his
associates (Sadler and Tesser, 1973; Tesser, 1976, 1978; Tesser and Con-
ley, 1975; Tesser and Leone, 1977) on the effects of " mere thought." A
major condusion drawn from this research is that, given the explicit
instruction to think about an object or an issue, subjects' attitudes be-
come more polarized. Thus, in one experiment (Sadler and Tesser,
1973), subjects were introduced to either a likeable or a dislikeable part-
ner. Some subjects were then instructed to think about the partner,
whereas others were given an irrelevant distraction task to work on. It
turned out that subjects under the "thought instruction" rated the like-
able partner in more positive terms and the dislikeable partner in more
negative terms than the distracted subjects.
In present terms, the instruction to "think" could be interpreted by
subjects as a demand to form an opinion about a given object; this could
induce a high need for dosure concerning the topic. In turn, increased
need for dosure might lead to a polarized attitude as found in Tesser's
research. The reason that a high need for dosure may lead to a polarized
attitude is that it may dispose a person to be biased in favor of informa-
tion consistent with early impressions or dosures (see also Mills, 1968;
Mills and Q'Neal, 1971; Q'Neal and Mills, 1969). Indeed, Sadler and
Tesser's (1973) subjects in the thought condition generated more posi-
ATTITUDES AS KNOWLEDGE STRUCTURES 125

tive thoughts about the likeable partner and negative thoughts about the
dislikeable partner than subjects in the distraction condition.
Consider some of the novel implications of the epistemic analysis of
the mere thought phenomenon. If the instruction to think arouses a
need for closure, alternative modes of arousing this need should have
similar effects. For example, making an action or adecision contingent
on an attitude should increase the degree of polarization, as should
enhancing subjects' need for closure via exposing persons to an unpleas-
ant confusion. Furthermore, individuals with a stably high need for
closure (i.e., displaying a need for closure as a personality trait) should
exhibit generally more polarized attitudes toward objects and be more
likely to exhibit polarization when instructed to think about an object.
Finally, for subjects lacking a preexisting attitudinal schema on a given
topic, the instruction to think should interact with the situational avail-
ability of attitudinal schemata. Thus, when instructed to think subjects
might adopt a situationally available schema or norm with greater as-
surance than in an absence of such an instruction.
Perhaps of greater interest, the epistemic interpretation suggests
that a different sense of increased thinking about a topic could result in a
less-polarized attitude toward an issue. This is predicted for conditions in
which the thinking instructions are interpreted in ways that may induce
a need to avoid or postpone closure; for example, where they imply the
desirability of judgmental accuracy. In such circumstances, the individu-
al is expected to generate numerous thoughts concerning the attitudinal
object, including those inconsistent with earlier schemata (cf. Tetlock,
1985a). Consequently, the person's final attitude may become less rather
than more polarized as compared to someone with a lower need to avoid
closure. Furthermore, individuals with a stably high (versus low) level
of the fear on invalidity might exhibit less-polarized attitudes toward
various issues, and be less likely to polarize their attitudes under the
instruction to think. These possibilities might be profitably explored in
further research.

Effects of Needs for Specific Closure or the Avoidance of Specific Closure

A veritable plethora of effects in the attitude-change literature is
subsumable under the present notion of specific closure biases, so much
so that the generic notion of motivational influences on attitudes has
been typically assumed to refer to specific closure effects. For example,
dissonance theory (Festinger, 1957) is considered a central motivational
theory of attitude change and stresses the tendency to adjust one' s
attitudes in order to maintain an esteem supporting or enhancing
126 CHAPTER6

closure (e.g., that one is an intelligent person who does not knowingly
make regrettable decisions).
With the exception of the "knowledge function," interpreted pre-
viously in terms of the need-for-closure notion, Katz' s (1960) remaining
functions of attitudes are readily coordinated to various preferred
closure needs. Thus, his ego-defensive function suggests that such atti-
tudes are likely to be adopted which avoid conclusions damaging to
one's ego. (Note here the similarity to the esteem-maintaining function
implicit in some versions of dissonance theory.) Katz' s value-expressive
function suggests that an attitude is more likely to be adopted if it
implies the desired conclusion that one would be able to express one's
values. For instance, if an antiwar demonstration is being planned, and
if one's values are highly pacifist, one mayadopt a positive attitude
toward participating in street demonstrations.
The "social-adjustment" function of attitudes stressed by Smith,
Bruner, and White (1956) suggests that attitudes are more likely to be
adopted if they imply the desirable conclusion that one is similar to,
hence, is likely to be approved of by, one's idols or significant others.
Quite similar is the notion of ego-involved attitudes stressed by Sherif
and Hovland (1961). An ego-involved attitude represents a social value
with which an individual strongly identifies (Sherif and Cantril, 1947,
pp. 126-127). Adopting such an attitude affirms the desirable conclusion
that one is a member of a valued group (e.g., a black, a Baptist, a liberal).
Sherif's notion that ego involvement increases the resistance to attitude
change is thus similar to the present proposal that a specific closure need
that may be satisfied by some conclusions and not by others can increase
the tendency to adhere to a wish-congruent belief.
Note, however, that a need for the avoidance of a specific closure
can lower the resistance to attitude change, under conditions in which a
given attitude is manifestly contrary to a desirable conclusion. Thus, an
ego-involved member of the peace movement, with a strongly negative
attitude toward conventional weapons, might quickly modify his or her
views if informed that the movement's leadership announced such
weapons to be a preferred alternative to nuclear armament, ultimately
increasing the prospects of peace.
Specific closure effects could underlie "anticipatory shifts" in atti-
tude change; that is, shifts resulting not from the persuasive com-
munication per se but rather from a mere expectation of persuasive at-
tacks. Thus, Cialdini and Petty (1981) note that a large self-
presentational component is involved in the occurrence of such shifts,
and that the shifts appear to be attempts to move toward the "defensible
and admirable moderate positions of the attitude scale" (cf. Cialdini,
Petty, and Cacioppo, 1981, p. 393). In present terms, one seems to adopt
ATIITUDES AS KNOWLEDGE STRUCTURES 127

such an attitude that renders more likely the conclusion that one would
gain the approval or admiration of significant others.
The effects of fear-arousing communications could represent specif-
ic closure effects on attitude change. In some relevant studies (e.g., Janis
and Feshbach, 1953; Janis and Terwillinger, 1962; Nunally and Bebsen,
1959) high-fear communication was less effective in producing attitude
change than moderate- or low-fear communication. This was explained
in terms of the "defensive-reactance" hypothesis whereby the highly
threatening message evokes considerable anxiety defended against by
misperception and denial of the persuasive arguments. In other research
(Leventhal, 1970) the high- (versus low-)fear communication proved a
more effective means of inducing attitude change. This was explained in
terms of the "incentive effect," whereby the incentive to change is great-
er the higher the threat. Both defensive reactance and the incentive
effects may be interpreted as different instances of a specific closure
effect on attitudes: The highly negative conclusion, concerning one's
physical welfare, for instance, conveyed by the threatening message
seems best avoidable, in the one case, by directly denying the validity of
the message; and, in the other case, by accepting the message and
deciding to adopt the preventive measures it outlines. Increased tenden-
cy to accept high-fear appeals when the preventive measures are clear
and compelling was manifest in several studies (Hass, Bagley, and
Rogers, 1975; Mewborn and Rogers, 1979; Rogers, 1975; Rogers and
Mewborn, 1976).

Avoidance of Nonspecific Closure

Several recent studies in the attitude change area are interpretable
in terms of the need to avoid closure. Those are studies manipulating
the personal consequences of a given attitude and observing the effects
on subjects' tendency to consider thoughtfully arguments relevant to
the attitudinal issue. Petty, Cacioppo, and Goldman (1981) had college
students listen to a persuasive message advocating that seniors be re-
quired to take a comprehensive exam in their major prior to graduation.
In the high-consequences condition the communicator advocated that the
policy commence the next year, whereas in the low-consequences condi-
tion that it commence ten years hence. For half the subjects the com-
munication contained eight highly persuasive arguments and for the
other half, eight weak arguments. Finally, for half the subjects, the
source was described as an expert (a professor of education at Princeton
University), and for the remaining half, as a nonexpert (a junior at a local
high school). It was found that, in the high-consequences condition,
subjects' attitude change was determined primarily by quality of the
128 CHAPTER6

arguments (strong versus weak), whereas in the low-consequences con-

dition, by source expertise.
Inpresent terms, the high-consequences condition in the Petty,
Cacioppo, and Goldman (1981) study might have induced in subjects a
need to postpone closure based on a fear of invalidity. According to the
lay epistemic theory this should lead to extensive consideration of all the
available evidence, including arguments presented in the message, per-
tinent to the issue discussed. By contrast, in the low-consequences con-
dition subjects' concem with validity might be rather low. Their pre-
dominant motivation might be a need for closure stemming from the
experimental demand to form some kind of pertinent opinion. To do so,
subjects may use the shortest possible route to an opinion and derive it
from the single most general bit of relevant evidence, in this case source
expertise.
The foregoing interpretation of Petty, Cacioppo, and Goldman's
(1981) findings yields several yet untested predictions. For instance, it
follows that if the message arguments were inconclusive (rather than
downright weak, i.e., counterindicative of the advocated position), sub-
jects in the high-consequences' or high fear-of-invalidity condition
might also base their attitude on source expertise, just as would subjects
in the low-consequences condition. Furthermore, manipulating the
need for closure by means other than low consequences, for example, by
time pressure, should also increase subjects' tendency to use the short-
est possible route to an attitude inference, hence, increasing their re-
liance on source expertise. Finally, if our analysis is correct, it is not
essential that the consequences of an attitudinal position be personally
involving to the subjects, only that the perceived costs to oneself or
others of an invalid position be high. For example, in the study by
Freund, Kruglanski, and Schpitzajzen (1985), fear of invalidity was ma-
nipulated by perceived costs of subjects' judgmental mistake to the tar-
get person, and this significantly mitigated subjects' tendency to base
their judgments predominantly on early information (primacy effects).
Further attitudinal studies manipulated personal consequences in
ways that lend themselves to a closure-avoidance interpretation. In a
study referred to earlier, Chaiken (1980) exposed subjec~s to either six or
three arguments in favor of a given position, and those arguments were
alleged to originate with a likeable or a dislikeable source. Subjects in the
high-consequences condition expected to be questioned about the issue
at a later time, whereas subjects in the low-consequences condition ex-
pected to be interviewed on an irrelevant issue. From the present per-
spective, the high- (versus low-)consequences condition may induce a
fear of invalidity having to do with evaluation apprehension (Kruglanski
and Freund, 1983). This may increase subjects' need to avoid closure.
ATIITUDES AS KNOWLEDGE STRUCTURES 129

Furthermore, a likeable communicator may induce a preferred closure

need to gain approval from this person, and a dislikeable source a need
to dissociate oneself from hirn or her. As the epistemic analysis would
predict, under high consequences, attitude change was primarily deter-
mined by the number of arguments presented (six versus"three) withthe
likeability manipulation exerting no significant effects. However, under
low consequences, attitude change was determined primarily by source
likeability, with the number of arguments making no significant dif-
ference.
To conclude, in the preceding sections I have reviewed evidence for
the motivational component in attitude formation and change. It ap-
pears that the numerous motivational effects on attitudes reported in the
literature are classifiable into categories suggested by the epistemic anal-
ysis. Particularly often evidence was encountered for specific closure
biases; however, effects interpretable in terms of the needs for closure
and closure avoidance were also found. It is worth reiterating that,
according to the lay epistemic model, motivational interpretations of
attitude-change phenomena are not incompatible or mutually exclusive
with the effects of cognitive factors, such as the use of logical deduction
or the processing of mentally accessible informational categories. In
other words, the lay epistemic theory provides a unified view of atti-
tude-change phenomena in which motivational and cognitive influences
complement each other. In the following section, this unified view is
explicitly contrasted with disjunctive formulations that distinguish be-
tween putatively disparate attitude-change processes.

UNITARY OR DISJUNCTIVE PROCESSES OF ATTITUDE FORMATION AND

CHANGE

The Motivation, Information-Processing Debate

The issue whether attitude change observed in dissonance experi-
ments is due to the postulated motivational factors as opposed to purely
informational ones was forcefully raised by Bem (cf. 1965, 1967, 1972).
His major point was that a person's attitudes, just as the attitudes of
other people, are often inferrable from overt behavior and the control-
ling circumstances in which it occurs. For example, a subject in a forced-
compliance experiment who expresses a positive attitude toward an
issue or an object in the presence of a large inducement (e.g., a $20
reward) can infer that he or she is "manding," and that his or her public
behavior is not indicative of a private attitude. By contrast, a subject
expressing the same attitude in the presence of a small inducement
(e.g., a $1 reward) can infer that he or she is "tacting," and that his or
130 CHAPTER6

her public behavior does reflect the private attitude. Thus, a greater
attitude change, following a counterattitudinal statement expressed for
a low (versus a high) inducement (Festinger and Carlsmith, 1959), need
not indicate the reduction of a motivationally aversive state of dis so-
nance but could indicate a "cool" information-driven process of attitude
inference from behavioral and situational cues.
To tease apart the self-perception from the dissonance interpreta-
tion, Bem conducted aseries of "interpersonal replications" in which
subjects received information concerning the behavior of participants in
a given dissonance experiment, and the external circumstances under
which this occurred. Subjects' task was to postdict participants' relevant
attitudes. Presumably, the interpersonal replicators are not motivation-
ally involved in the situation, nor do they experience dissonance of any
sort. Thus, if they succeed in inferring the actual participants' attitudes,
the latter persons, too, may be inferring their attitudes from relevant
information rather than reducing dissonance.
For example, one of Bem's studies interpersonally replicated the
dassical forced-compliance experiment by Festinger and Carlsmith
(1959), in which subjects for high ($20) or low ($1) inducement reported
to another individual that the tedious experimental task that they had
just performed was actually "fun, exciting, and enjoyable." According to
dissonance theory, making such counterattitudinal statement for low
(hut not high) justification arouses astate of dissonance reducible by
adopting a more positive attitude toward the issue. Indeed Festinger
and Carlsmith (1959) predicted and found an inverse relation between
the magnitude of inducement and attitude change. This above result
was successfully mimicked by Bem's (1967) interpersonal replicators
who received information about a subject in either the $1 or the $20
condition of the original experiment. Bem took this and other similar
results (Bem, 1972) to mean that attitude-change phenomena, at least
those demonstrated in typical dissonance studies, reflect informational
rather than motivational processes.
But from the present perspective, Bem's interpersonal replicators
did not actually render their judgments in a motivational vacuum; more
likely, they were motivated by a need for dosure, notablya desire to
form a dear answer to the experimental question about another person' s
attitude. Admittedly, this may differ from actual subjects' motivation
that might have been a specific dosure need to infer that they do not
make public statements without good internal or external reasons. Ac-
cording to this analysis, the success of Bem's interpersonal replicators
does not necessarily indicate that motivational factors have no role to
play in attitude-change phenomena. Rather, it could indicate that differ-
ent motivations may occasionally result in the same attitudinal in-
ATIITUDES AS KNOWLEDGE STRUCTURES 131

ferences. Given the available information, the cognitive structure that

most readily forms in subjects' minds mayaiso happen to represent the
most desired conclusion. Thus, subjects with a high need for closure
(e.g., Bem's interpersonal replicators) might, in the appropriate circum-
stances, make the same attitudinal inference as subjects with a high
need for specific closure (e.g., actual participants in a dissonance study).
This analysis yields the further prediction that in other circumstances
the most readily suggested or accessible closure may be quite opposite to
the most desirable conclusion. Under those conditions, subjects under a
high need for closure should reach a very different conclusion than
subjects under a high need for specific closure. These possibilities could
be explored in future research.
The foregoing analysis implies that motivational explanations of
attitude-change phenomena are not mutually exclusive with informa-
tional explanations. In each instance of an attitudinal inference, there is
the informational base from which the attitude is deduced, and there is
also the motivational base without which no inferential activity would
occur, and that may determine the direction and the duration of the
informational search.

CENTRAL VERSUS PERIPHERAL ROUTES TO ATTITUDE CHANGE:

A UNIFIED PERSPECTIVE

According to Petty and Cacioppo (1981), attitude change can take

place via two separate routes: the central and the peripheral route. The
central route emphasizes the information that a person has about the
attitude object or issue under consideration (p. 255). Such issue-relevant
information could be contained in the persuasive message or could be
generated by the recipient in response to the message. By contrast,
attitude change via the peripheral route
is determined by such factors as the rewards and punishments with which
the message is associated . . . or the simple inferences that a person draws
about why a speaker advocated a certain position. The picture that these
peripheral approaches paint of the persuasion process is not a very thought-
ful one. If a message is associated with a pleasant smell or attractive source, it
is accepted ... people "observe" their own behaviors or physiological re-
sponses and infer what their attitudes "must" be. (p. 255)

According to Petty and Cacioppo (1981),

the difference between the central and peripheral routes to attitude change is
not that the former actually is rational and logical whereas the latter is not
[rather] the difference has to do with the extent to which the attitude change
that results from a message is due to active thinking about either the issue or
the object-relevant information. (p. 255)
132 CHAPTER6

When attitude change occurs via the central route it is determined by

issue-relevant information, and when it occurs via the peripheral route it
is the result of contextual"persuasion cues," like very attractive sources
or an ability to obtain areward. Such cues are sufficient to produce
attitude change "without engaging in any extensive cognitive work rele-
vant to the issue under consideration" (p. 255).
A major difference assumed to separate attitude change obtained
via the central and the peripheral routes is that the former is more
enduring than the latter. In the study by Chaiken (1980) referred to
earlier, subjects in a condition of high personal consequences were per-
suaded more when the number of cogent arguments contained in the
message was large (six) rather than small (two) and were rather un-
affected by source likeability. This pattern represents an attitude change
via the central route. By contrast, subjects in a condition of low personal
involvement were persuaded more when the source was likeable (ver-
sus unlikeable), and were rather unaffected by the number of cogent
arguments contained in the message. This pattern typifies an attitude
change via the peripheral route. Relevant to the present concern, the
decay in attitude change as measured 10 days later was less for subjects
in the high personal consequences condition, persuaded via the central
route than for subjects in the low personal consequences condition per-
suaded via the peripheral route.

A Lay Epistemic Reanalysis o[ the Central-Peripheral Routes Distinction

The central-peripheral distinction links together several factors
which, according to the lay epistemic theory, are essentially unrelated.
Thus, the extent of information processing or of cognitive work is linked
with the specific contents of information processed, having to do with
the issue as such rather than the context of the communication, illus-
trated by variables as source credibility or expertise or one's own phys-
iological responses and overt behaviors toward the attitude object. Fur-
thermore, the concern for the validity of one's opinions (or fear of
invalidity), probably involved in the manipulation of "personal conse-
quences" in Chaiken's (1980) and in other similar studies, is likewise
linked with issue relevant information. By contrast, various specific
closure needs (e.g., the need to reach a conclusion likely to result in
approval from an attractive person or in receiving areward) are dissoci-
ated from the processing of issue-relevant information.
According to the epistemic approach, none of these linkeages are
necessary. For example, we could have a situation in which the individ-
ual experiences a need to avoid closure (induced, say, by manipulation
ATIITUDES AS KNOWLEDGE STRUCTURES 133

of personal consequences) and this could motivate him or her to process

information extensively concerning the source characteristics or his or
her physiological or behavioral reactions to the attitude object.
Furthermore, what constitutes issue-relevant information may be
quite subjective and in the eye of the beholder. For some persons source
credibility or expertise could be quite relevant to an attitude, in the sense
of being deductively linked, via an if-then bond, to it in the person's
mind. Finally, various preferred closure needs (e.g., of approval by an
attractive communicator or of self-esteem) could be linked with the pro-
cessing of endogenous message arguments. Imagine a person whose
position was insultingly attacked by a communicator, threatening a po-
tentialloss of face in front of a significant audience. Such a person might
be strongly motivated to restore his or her threatened respectability by
thinking up numerous endogenous message arguments contrary to the
attacker's allegations. Thus, a specific closure need of ego protection or
enhancement could be associated with the processing of endogenous
message information, just as the fear of invalidity could be associated
with the processing of exogenous message information.
Furthermore, the extent of information processing, or cognitive
work invested in reaching a given attitude inference, may be quite unre-
lated to whether the information processed was endogenous or ex-
ogenous to the message. For instance, a high need for closure might
result in abrief information-processing sequence, yet the attitude ar-
rived at might be based exclusively on endogenous message informa-
tion. Similarly, a high need to avoid nonspecific or specific closure might
result in extensive information processing of available evidence, irre-
spective of its endogenous or exogenous nature.
In short, the present epistemic analysis features a unified view of
attitude change that stresses the fundamental commonalities rather than
the differences between the central and the peripheral routes. In both
cases, the attitude is derived from subjectively relevant evidence that
could relate to endogenous or exogenous information. Also, in both
cases, the extent of information processing as weIl as its directionality
would depend on (1) the relative magnitudes of the several epistemic
motivations experienced by the individual and (2) the momentary ac-
cessibility of various categories of evidence that the individual deerns
relevant to the attitude or the opinion. These implications of the present
analysis could be explored empiricaIly, as could the additional implica-
tion that attitude change, associated with peripheral features (e.g.,
source characteristics, one's bodily or behavioral reactions to the attitude
object, or a specific closure need of some sort), could be as enduring as
that associated with central features (consideration of endogenous mes-
sage arguments, or the need for nonspecific closure).
134 CHAPTER6

AN EPISTEMOLOGY OF ACTIONS AND THE ATTITUDE-BEHAVIOR

RELATION

The persistent interest of sodal psychologists in attitudes is trace-

able to the assumption that they are the antecedents of behaviors. How-
ever, research in the area (for reviews, see e.g., Fishbein and Ajzen 1975;
Petty and Cadoppo, 1981; Wicker 1969) suggests that attitude-behavior
correlations can be rather weak. A conceptual advance in this domain
was accomplished by Fishbein and Ajzen (1975; Ajzen and Fishbein,
1977) who suggested that behaviors do not directly follow from attitudes
but rather from behavioral intentions. The latter, in turn, can be pre-
dicted from specific attitudes toward a behavior; that is, attitudes toward
behaviors at particular times toward particular targets and in particular
contexts.
The present epistemic analysis of the attitude-behavior relation ac-
cepts Fishbeins and Ajzen's (1975) proposal that (voluntary) behavior
follows from a behavioral intention. In fact, behavior could be described
as the translation of a contemporaneous intention into some sort of
physical or mental change. Furthermore, intention is presently conceived
of as knowledge of sorts. To intend to go to a theater on Sunday night is to
be consdous of one's will to do so; in present terms, it is to "know" it or
to feel subjectively certain about it. The conceptualizing of intentions as
types of knowledge suggests that a general epistemic theory should be
applicable to the explication of conditions under which intentions are
formed or modified. Furthermore, such a theory should aid in eluddat-
ing the attitude-intention relation.

Knowing What to Do: An Epistemic Analysis of Behavioral Intentions

How can one be sure that one actually intends to perform a given
behavior? According to the present analysis, just as other types of
knowledge, intentions too are derived from relevant evidence. For in-
stance, John may be certain that he intends to attend the Saturday night
dance, because he knows Paula will be there, and he has long dedded to
see Paula whenever possible. Furthermore, John's intention to attend
the dance might be undermined by contradictory evidence. Suppose an
alternative consideration became apparent, for example, adesire to do
well at school that implies an incompatible intention of staying at horne
and studying for an exam. Under such drcumstances, John might be
confused about the nature of his intentions virtually "not knowing what
to do."
It follows that, under some drcumstances, attitudes might be rele-
vant to behavior because the behavioral intention might be implied by the
ATTITUDESASKNOWLEDGESTRUCTURES 135

attitude. For example, an individual might believe that a positive atti-

tude toward peace implies participating in astreet demonstration, or
that a negative attitude toward processed foods implies declining to
breakfast on artificially flavored cereals. The behavioral implications of a
given attitude could be highly subjective; some persons could see how
these follow from the attitude whereas other persons might not. Consis-
tent with the foregoing argument is Ajzens and Fishbein's (1977) notion
(for empirical support, see Weigel and Newman, 1976) that a general
attitude measure might not predict a specific behavioral instance but
should predict a general behavioral index based on a variety of attitude-
consistent behaviors. In present terms, this should be the case if the
general attitude implied different behaviors for different persons. Fur-
thermore, in a study by Bagozzi (1981), attitudes toward blood donation
influenced behavior only through their effects on intentions. This, too,
suggests that only when an attitude implies a behavioral intention will it
have a determinative influence on behavior.

Nonattitudinal Predictors of Behavior: The Concept of Action Schemata

The discussion thus far suggests that an attitude does not invariably
predict a behavior. Individuals may vary in their perceived behavioral
implications of an attitude and the same individuals may be aware of
different attitudinal implications on different occasions. Furthermore,
behaviors may often be implied by nonattitudinal antecedents. In a re-
cent chapter, Kruglanski and Klar (1985) proposed the notion of action
schemata, referring to the implicational (if-then) premises in which con-
sequent terms are behavioral intentions. According to this concept, an
attitudinal action schema is an if-then statement with attitude as the
antecedent term and a behavioral intention as a consequent. As an ex-
ample, if my attitude toward a social issue say, ecology or birth control is
positive, I will try to express it in whichever way I can.
However, the present LEGO metaphor suggests that any concep-
tual category could be linked implicationally with a behavioral intention.
Thus, we should have numerous action schemata in which antecedent
terms are not attitudes. For example, a person may employ an action
schema whereby if aleader issues a command, I will immediately ex-
ecute it. If the action mediates a worthwhile goal and is within his or her
ability to perform it, then he or she intends to do so (Atkinson and Birch,
1970).
An example of a nonattitudinal action schema could underlie the
finding of Abelson, Kinder, Peters, and Fiske (1982) that voting prefer-
ences were reliably predicted by affects evoked by a political candidate,
indexed by a response to the question whether the candidate made one
136 CHAPTER6

feel happy or sad. In such a case, the action schema could state some-
thing such as if a person makes me feel happy, I will try to help him or
her in whichever way I can. In fact, Abelson et al. found that affect was a
better predictor of voting preferences than attitudes evaluations of the
candidate as competent or incompetent. However, according to the
present approach, whether an affective or an attitudinal predictor of
behavior would be better should depend entirely on the idiosyncratic
action schemata in a given person's repertory. Some individuals (e.g.,
males with a conservative upbringing) might be socialized not to act on
their affects but rather on their "cool" calculations.
The phenomenon of modeling (cf. Bandura, 1969) can be explicated
in terms of specific (nonattitudinal) action schemata from which behav-
ioral intentions are. deduced. For instance, a small child may proceed
from the premise whereby, if an older sibling or a parent is performing a
given action, it intends to do the very same thing as weIl. In sum,
attitudinal and nonattitudinal action schemata may function as decision
rules that guide the person's actions on various occasions.
This analysis, however, has several implications concerning condi-
tions under which a behavior would follow from an attitude. First, the
attitudinal cognition would have to be available in the person's reper-
tory or stored in his or her memory. Degree of belief, or subjective
confidence, in the attitudinal statement should be determined in part by
the perceived quality of supporting evidence for the statement at issue.
For instance, in some cases direct experience with the attitude object
may constitute cogent evidence for one's attitude, in particular, where
one has a high degree of "perceived epistemic authority" (EIlis, 1984),
allowing one to interpret the experience with confidence.
Evidence that direct experience with the attitude object increases
the attitude-behavior correlation was reported by Fazio and Zanna
(1981), who argued that attitudes based on direct experience are particu-
larIy good predictors of behaviors, presumably because direct experi-
ence is a more cogent source of evidence than indirect1y attained infor-
mation. As noted earlier, however, direct experience may constitute
cogent evidence for knowledge for some people only rather than as a
rule. As EIlis's (1984) research indicates, individuals with low perceived
epistemic authority in a given domain may not be able to benefit from
direct experience; however, they might arrive at confident inferences by
using alternate sources of evidence. The possibility that individual dif-
ferences in epistemic authority would moderate the impact of direct
experience on attitudes as weIl could be profitably pursued in further
research.
In addition, findings suggesting that attitudes based on direct (ver-
sus indirect) experience are better predictors of behavior can often be
AITITUDES AS KNOWLEDGE STRUCTURES 137

interpreted altematively in terms of motivational differences resulting in

more confident attitudes in the direct (versus indirect) experience condi-
tions. For instance, a study by Regan and Fazio (1977) investigated atti-
tudes toward a housing crisis by freshmen who actually experienced it
(having been assigned a cot in the lounge of a dormitory) and those who
dip not (having been assigned to permanent quarters). As it tumed out,
subjects in the first (versus the second) group tended to display more
attitude-consistent behaviors, such as agreeing to sign a petition, writ-
ing a letter to the University Housing Office.
Regan and Fazio' s finding might weIl reflect the greater attitudinal
confidence that was attained by subjects in the direct (versus indirect)
experience conditions. But this might not stern from experiential direct-
ness per se; rather, the subjects in the direct experience condition might
have been more motivated to form a confident attitude on the housing
issue (in present terms, they might have experienced a higher need for
c1osure) than subjects in the indirect experience condition. After all,
personal suffering from the housing crisis might induce the need to
develop a guiding orientation (Le., to know what one should do or not
do) in the circumstances, whereas a lack of personal involvement might
render an ambiguous position less tolerable. Note that experiencing
high versus low need for c10sure is in principle independent of having
direct or indirect experience with a topic. For example, a person who
cares very little about theater might fai! to form a confident attitude
about a Broadway play which he or she "directly experienced." By con-
trast, an involved theatre lover might form a rather strong attitude to-
ward the play, leading to the decision to attend or not attend it, based on
a review of an admired critic.

Attitude Accessibility
The degree to which a confident or well-consolidated attitude is
formed may affect its subsequent accessibility from memory. This prop-
osition was recently illustrated in aseries of well-designed studies by
Fazio and his colleagues. Using response latency as the dependent mea-
sure, these investigators showed that (1) increasing the number of at-
titudinal expressions (Fazio, ehen, McDonel, and Sherman, 1982;
Powell and Fazio, 1984), (2) presenting a cue suggesting a future interac-
tion with the attitude object or a future questioning about the attitude
object (Fazio, Lenn, and Effrein, 1983), or (3) asking the subject to re-
spond to a standard attitude scale (Fazio et al., 1982) reduced the latency
of respo~se to an attitudinal inquiry, thus reflecting increased ac-
cessibility. Each of the above manipulations could bolster the formation
of a confident attitude: Increasing the number of attitudinal expressions
138 CHAPTER6

could increase the amount of attitudinally relevant information subjects

may retrieve from memory, thus augmenting their confidence in the
final attitude.
Expecting future interaction with, or future questioning about, the
attitude object may induce a high need for closure leading again to
enhanced confidence, as may the request to respond (versus not to
respond) to an attitude scale. In turn, a confident attitude may be weIl
integrated in memory or linked with numerous retrieval cues that would
render it more readily accessible. The need for closure mayaiso be
aroused when subjects are asked to take some time to consider their
attitudes toward an issue. Precisely such manipulation was introduced
by Snyder and Swann (1976), resulting in an increased attitude-
behavior correlation.
To be sure, formation of a confident attitude is not the sole determi-
nant of accessibility. Another such determinant may be the recency with
which the attitudinal cognition was deposited in long-term memory (cf.
Higgins, Bargh, and Lombardi, 1985; Wyer and Carlston, 1979) or the
extent to which the attitude was recently primed. Research by Zanna
and Mohr (1985), for example, demonstrated that sexist attitudes that
were primed via exposure to a pornographie film led to sexually oriented
behaviors toward a female experimenter, including attempts at getting
physically closer to her and attending to her appearance rather than to
what she had to say.
Our epistemic analysis further suggests that even if the attitude is
momentarily accessed from memory, its particular behavioral implica-
tion might not be. Thus, one might recall one's positive attitude toward
physical fitness without at the same time recalling that it implies re-
duced alcohol intake or a low-cholesterol diet. Such implications might
fail to be recalled because they are incongruous with one's momentary
wishes or desires. In present terms, they might be contrary to one's
needs for specific closure and, in this sense, less likely to be cognized.
Other epistemic motivations mayaiso affect the attitude-behavior rela-
tion. Suppose a given behavioral implication of an attitude was momen-
tarily accessible; the likelihood that the behavior will be executed may
depend on the formation of competing behavioral intentions. This, in
turn, may depend on the individual's motivations, in particular on his or
her needs for closure or for the avoidance of closure. A high need to
avoid closure should increase the likelihood of cognizing competing
intentions and thus reduce the likelihood of executing the attitude-
consistent behavior. In contrast, a high need for closure should desensi-
tize the individual to competing intentions and thus increase the likeli-
hood of executing the behavior.
Evidence for the last prediction was obtained in recent research by
ATTITUDES AS KNOWLEDGE STRUCTURES 139

Bechtold, Zanna, and Naccarato (1985), described in Chapter 3. In this

study, subjects under time pressure, assumed to induce a high need for
closure, acted more in accord with their sexist prejudices and in viola-
tion of the situational norm of evaluative fairness than subjects under no
time pressure.

SUMMARY AND CONCLUSIONS

In this chapter I have tried to demonstrate that the lay epistemic

theory can serve as an integrative framework for the study of attitudes.
According to this approach, attitude is a special type of knowledge,
notably knowledge of which content is evaluative or affective. The lay
epistemic synthesis brings together various, heretofore separate ap-
proaches to attitude change: (1) the learning theoretical approaches con-
cerned with the conditioning of attitudes and the effects of rewards and
incentives; (2) the gestalt theoretical approaches and, in particular, the
cognitive consistency models dealt with in Chapter 5; (3) the attribution-
al approaches (e.g., Bem's self-perception theory that stresses the occa-
sional function that overt behavior may serve as evidence for one' s
attitudes); and (4) the central versus peripheral approaches discussed by
Petty and Cacioppo (1981).
The lay epistemic framework orders in a new way a variety of find-
ings in the attitudes area and suggests commonalities unattended to
previously. For instance, according to the present approach, derivability
or consistency with other firm beliefs is the underlying mechanism of
source as weIl as the message effects on persuasion; and it mayaiso
underlie classical conditioning effects or reward-mediated effects. The
construct of need for specific closure may underlie dissonance-mediated
attitude change, functionalistic effects, or the effects of fear appeals. The
accessibility construct may link such heretofore disparate phenomena as
biased scanning of arguments, refutational defenses, or sleeper effects.
Besides providing a broad integrative framework for known at-
titudinal phenomena, the lay epistemic model has several implications
for further research. Two main classes of novel research implications can
be discerned and are referred to as negative and positive. By negative
implications, I mean the kind of research questions that might be futile
to ask or the type of research that might be futile to execute. To the
extent that our lay epistemic theory indeed depicts the essential process
whereby attitudes are formed and/or are modified, those variables in the
attitude-change literature that do not systematically represent the epis-
temic constructs should also fail to display systematic effects on
attitudes.
140 CHAPTER6

As an example, consider the behavioristic variable of rewards. A

persistent issue in the attitudes area has been whether rewards for coun-
terattitudinal advocacy, for instance, have positive or negative effects on
attitude change. But from the present perspective, such a question
should not be expected to yield a c1ear-cut answer. Simply, under differ-
ent circumstances, rewards could represent different episternic factors
with contrasting irnplications for attitude change. For instance, rewards
could invite biased scanning, enhancing the activation and considera-
tion of arguments supporting a given attitude. Yet, in slightly different
circumstances, rewards could be construed as evidence that behaviors
or utterances consistent with a given attitude were exogenously moti-
vated (Kruglanski, 1975). In turn, this could lead to the deduction that
one's private attitude was contrary to those overt indications. In yet
other circumstances, rewards could induce a positive affect which, if
"rnisattributed" to the attitude being expressed, might enhance the in-
ference of an authentically positive attitude. In sum, one should not
expect a general relation between rewards and attitude change. The
plethora of inconsistent findings on that issue (for a review, see Petty
and Cacioppo, 1981) support this conc1usion.
For another example, consider the sleeper effect, referred to earlier,
in which source (versus message) effects on attitude change exhibit a
faster decay. From the present perspective, however, such effects
should not be expected to exhibit much generality. That is because link-
eage of a given conc1usion with' the source of the message should not
invariantly show greater decay over time than linkeage with message
arguments. According to the present approach, the reason why a given
cognition is remembered or forgotten is determined by factors having to
do with the general cognitive process rather than with the specific cogni-
tive contents (e.g., having to do with the source or the message). For
instance, some individuals might repress a particular communicator's
name because it is associated with unpleasant events, but other people
rnight fail to exhibit such bias. Research literature on sleeper effects (for
a review, see Petty and Cacioppo, 1981) suggests, indeed, that the phe-
nornenon shows little generality and occurs only under narrowly de-
fined circumstances (Cook, Gruder, Hennigan, and Flay, 1979; Green-
wald, Baumgardner, and Leippe, 1979).
Perhaps of greater interest than the negative research implications
of the lay episternic analysis are the positive implications. In previous
sections of this chapter, references were made to novel research pos-
sibilities offered by the lay epistemic model in connection with such
topics as (1) the mere thought effects on attitude polarization and the
possible interactions of such effects with the individual' s epistemic
motivations; (2) the integration of the central versus the peripheral
ATTITUDES AS KNOWLEDGE STRUCTURES 141

routes to attitude change under a common set of epistemic principles;

and (3) the identification of conditions under which the attitude-
behavior relation may or may not be evident. These implications and
possible additional ones suggest that the lay epistemic framework may
be useful for thinking about, and further studying, a broad variety of
attitudinal phenomena.
 7
Further Domains of Application
Sodal Comparison Processes and Minority-
Influence Phenomena

The lay epistemic analysis may be applied to further sodal psychological

topics. In the sections that follow two such additional domains are spe-
dfically discussed, notably those of sodal comparison and minority-
influence phenomena. Let us consider them in turn.

SOCIAL COMPARISON PROCESSES

Although investigators' interest in sodal comparison issues has

waxed and waned over the years (Goethals, 1987) the primary concep-
tual framework to guide research in the area has remained Festinger' s
(1954) theory of sodal comparison processes. The theory of sodal com-
parison has rested on the assumption that people possess a drive to
evaluate their opinions and abilities. This may take place by reference to
physical reality (as in assessing one's ability to lift a weight by actually
trying to do so) or, where objective means of evaluation are unavailable,
by comparison with other people. A major tenet of the theory was that
similar rather than dissimilar others provide the more efficient standard
for comparison. Thus, in order to predsely evaluate their opinions and
abilities, persons were assumed to prefer in general comparisons with
individuals who are similar to themselves. Over the years, sodal com-
paris on research has led to a number of revisions and extensions in the
original theory. In particular, new domains of comparison were pro-
posed (e.g., emotions, cf. Schachter, 1959; or values, cf. Goethals
and Darley, 1977). Similarly, additional motives relevant to comparison
processes were enumerated, such as the needs for self-enhancement
(Gruder, 1971), for validation (Goethals and Darley, 1977), for mainte-

143
144 CHAPTER 7

nance of a positive self-evaluation (Tesser and Campbell, 1982) as weH as

other needs (Kruglanski and Mayseless, 1987). Furthermore, situations
were identified in which dissimilar rather than similar others are pre-
ferred as objects of comparison (Goethals and Nelson, 1973; Kruglanski
and Mayseless, 1987; Mettee and Smith, 1977; Wills, 1981). Finally, a
reconceptualization of the critical similarity dimension Was proposed in
terms of similarity in attributes related to an opinion or a performance
rather than similarity in opinion or performance per se (Goethals and
Darley, 1977).
Despite those revisions, fundamental aspects of Festinger's original
conception have remained intact, notably that (1) another person's sim-
ilarity to oneself (however defined) is a crudal parameter of sodal com-
parison; (2) if not exactly of a "drivelike" quality, there exists a fairly
general and pervasive tendency toward sodal comparison; and (3) the
processes of sodal comparison differ across content domains of com-
parison. For instance, a "unidirectional drive to do better and better"
exists when comparing one's abilities but not one's opinions (Festinger,
1954a), whereas a preference for similar others is more pronounced in
the domain of values and less in that of opinions (Goethals and Darley,
1977).
Generally speaking, sodal comparison processes pertain to compara-
tive judgments that people make about their various attributes. Therefore,
most sodal psychologists would probably agree that if one applied what
is known about the judgmental process to the sodal comparison domain
the fit should be good. Surprisingly, however, examination of the sodal
comparison theory from a general judgmental perspective contained in
the present epistemic theory suggests several major discrepandes be-
tween the two approaches.

THE Two SENS ES OF SOCIAL COMPARISON

A lay epistemic reexamination of social comparison theory reveals

two separate senses in which the term social comparison has been used. In
the first sense, sodal comparison refers to a spedfic content 01 a question
in which persons may have an interest (Smith, 1981), and which con-
cerns one's standing on some dimension relative to others. Iones and
Gerard (1967) referred to this type of question as the problem of "com-
parative appraisal." In the second sense, sodal comparison denotes the
use of social sources of information in evaluating any question of interest,
for instance, where "physical sources" of information are unavailable
(Festinger, 1954a).
It is noteworthy that studies of ability comparisons typically re-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 145

ferred to the former sense of sodal comparison, whereas studies of

opinion comparisons generally referred to the latter sense. Thus, in
ability comparison research, a typical question of interest concerned the
loeus of eomparison. This refers to some characteristic of others with
whom one may wish to compare one's own attributes or performance.
In present terms, locus of comparison research addresses the specifie
eontent of comparative knowledge in which persons may be interested:
00 people wish to know how they stand in reference to the extremes or
the center of a score distribution? 00 they wish to know by how much
they are trailing those ahead or surpassing those behind? Such ques-
tions, typical of ability comparison research, dearly pertain to the con-
tents of comparative appraisals that persons may seek.
By contrast, in studies of opinion comparison, the eomparison others
of interest have been generally treated as a potential source of informa-
tion rather than as an object of comparative appraisal. Unlike research
on abilities comparison, in opinion comparison research the issue was
not to determine the extent of opinion discrepancy between oneself and
another individual; instead, it was to react to such a discrepancy when
assessing the validity of one' s own opinion.
Accordingly, in studies of opinion comparison, the typical research
questions concerned the role of the other persons in validating an indi-
vidual's various notions irrespective of content: Would a supportive
other be sought after more when subjects' confidence is low versus
high? Would a supportive other be liked more after subjects encoun-
tered several previous disagreements (versus agreements)? Would an
agreement from a similar other raise subjects' confidence more than an
agreement from a dissimilar other? Those were the typical questions in
opinion comparison research. They dearly refer to other people as a
validating souree rather than a eomparative eontent of knowledge.
It should be noted that the "content" and "source" senses of sodal
comparison are unrelated: The same content of comparative knowledge
can be conveyed via a variety of informational sources and viceversa.
For instance, the knowledge that one has scored at the top of one's dass
can be obtained by an inquiry among one's dassmates, by asking the
instructor, or by inspecting the grade list. Similarly, a knowledgeable
source (one's parent, instructor, or therapist) might be able to furnish
information about one's comparative standing in regard to a variety of
comparison contents (e.g., intelligence, sodability, persistence) and/or
concerning a variety of reference groups (one's siblings, dassmates, or
other undergraduates). But let us see what the content and source no-
tions imply more spedfically as far as sodal comparison theory is
concerned.
146 CHAPTER 7

SOCIAL COMPARISON AS THE CONTENT OF A QUESTION

From the lay epistemic perspective, the contents of questions in

which persons may have interest are quite heterogeneous. In reference
to social comparison questions, persons might often be interested in
knowing how they measure up against others with whom they share an
attribute (Goethals and Darley, 1977; Suls, Gastorf, and Lawhon, 1978).
Yet, occasionaHy, they might wish to compare with others who are
similar to them in performance (Wheeler, Koestner, and Driver, 1982),
with others who are dissimilar in performance (Gruder, 1971; Wheeler
and Koestner, 1984), or ignore an attribute's relevance to performance
altogether (MiHer, 1984). The heterogeneity of possible self-evaluations
has been noticed by early workers in the field. In 1966, Singer summed
up the special issue of the Journal of Experimental and Social Psychology,
which was devoted to social comparison, by commenting that "If a
person wishes to evaluate himself on an ability X there are several kinds
of evaluations he can make" (p. 107). Indeed, specific studies in that
issue of the journallisted some of the types of possible comparison, for
example, comparison with a positive example (Thornton and Arrowood,
1966), or with someone doing beUer than oneself (Wheeler, 1966), or
doing worse (Hakmiller, 1966).
Subsequent research demonstrated yet other types of possible com-
parison. For instance, Wheeler et al. (1969) found that most of their
subjects exhibited a "range-seeking" tendency. Arrowood and Friend
(1969) found a comparison preference for the "positive instance," Wil-
son and Benner (1971) found a preference for the results of the "best
off," whereas Friend and Gilbert (1973) demonstrated subjects' interest
in both the best-off and the worst-off cases. Similarly, Gruder (1977)
presented evidence that people may seek various kinds of comparative
information:
They want very much to Iearn about the highest scoring other .... Theyare
also interested ... in the other who best represents a definition of the di-
mension, the "positive instance." And they want to compare themselves
with others who are better off than they are on the dimension. (pp. 37-38)

Taken as a whole, the foregoing research seems to contradict Fes-

tinger's original proposal and to suggest that people are often interested
in comparative knowledge about dissimilar as weH as similar others.
However, the similarity hypothesis was restated in a modified form in
Goethals and Darley's (1977) revision of the sodal comparison theory. In
this conceptualization, the critical similarity dimension was assumed to .
be similarity on attributes relevant to the performance rather than sim-
ilarity of the performance itself. In support of this line of reasoning,
Zanna, Goethals, and Hill (1975) and Suls, Gastorf, and Lawhon (1978)
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 147

reported a preference for information about the results of similar sex and
similar age groups regardless of similarity in performance.
Goethals and Darley's (1977) influential revision inspired further
research which revealed, nonetheless, that people may desire diverse
kinds of comparative information (Miller, 1984; Wheeler et al., 1982;
Wills, 1981). Although information about others who are similar in at-
tributes or performance is often sought, people may be occasionally
interested in other kinds of information including information about
others who are dissimilar to themselves in attributes or performance.
These findings are typically depicted as inconsistent with the notion that
persons uniformly prefer information about similar others.
According to the present approach, the types of comparative ap-
praisals people might seek are quite diverse and unlimited. As with
knowledge in general, any content of comparative knowledge might be
of interest to some people in some drcumstances. Thus, in given condi-
tions, persons might be interested in comparing with similar others,
whereas in other conditions, they might be interested in comparing with
dissimilar others. Occasionally, persons might be interested in informa-
tion about successful and affluent persons (the rieh and famous) irre-
spective of their degree of similarity to themselves, whereas in other
situations they might be interested in comparing with failing or destitute
persons. In short, if any knowledge could be occasionally of interest to
some persons, the possible dimensions of sodal comparison questions
should be quite unconstrained and open-ended.
Furthermore, comparative appraisals need not refer to other peo-
pIe. Instead, they could refer to various standards of conduct or perfor-
mance. Such standards are classifiable in accordance with hetero-
geneous possible dimensions, for example, standards of duty (ought)
and standards of desire (want), one's own standards or those of signifi-
cant others (cf. Higgins et al., 1985), and past or present standards (cf.
Albert, 1977). The importance of internalized standards in sodal com-
parison contexts was explidtly noted by Gruder (1977).
The present framework also highlights the cultural and situational
factors that may determine both the extent of interest in comparative
issues and the type of comparison questions of interest. Thus, a com-
petitive culture may foster pervasive interest in comparison issues, as
maya competitive profession (e.g., athletics) or a competitive situation
(a contest). In alternative cultures, professions, or situations, compara-
tive appraisals may be of lesser interest, and the tendency to engage in
sodal comparison may be less extensive.
In short, rather than viewing sodal comparison as propelled by a
universal human drive, the present epistemic analysis depicts it as a
particular category of epistemic questions that may or may not be of
148 CHAPTER 7

interest. Moreover, wherever of interest, the types of comparative ques-

tions may vary greatly across individuals and situations. Hence, an at-
tempt to list the comparisons in which people are generally interested
could be unproductive. The present analysis suggests that this question
does not have a general answer, only situationally spedfic ones.

SOCIAL COMPARISON AS USE OF SOCIAL SOURCES OF INFORMATION

Let us now consider sodal comparison as the use of soda! sources of

information. Here questions raised by sodal comparison theory (Fes-
tinger, 1954) concerned the type of comparison others to which people
typically turn for information, and the impact that information from
such different sodal sources has on people's opinions. Sodal com-
parison theory answered those questions by highlighting again the role
of the similar other. Spedfically, individuals were presumed to exhibit
general preference for information from similar (versus dissimilar) others,
as well as to derive greater confidence and accuracy from such infor-
mation.
Very different implications follow from our lay epistemic theory. In
particular, it is suggested that informational sources would be preferred
and have an impact on individual's opinions to the extent that they are
perceived as relevant (versus irrelevant) to the opinion topic, and/or the
information they convey is perceived as compatible with individuals'
epistemic motivations. Neither relevance nor motivational compatibility
seems generally correlated with the similarity to oneself of an informa-
tional source. For example, a person may consider hirnself or herself
rather nonexpert in medical affairs. In such a case, a relevant informa-
tional source may be a qualified physidan, that is, a dissimilar other. In
an alternative case, only a similar other may possess relevant informa-
tion. For example, for an American in Europe, only another American
might be expected to have relevant information on the latest develop-
ments in the World Series.
In regard to motivation, a similarly minded other might fulfill one's
desire for c10sure or stability. However, a dissimilar other could fulfill
one's wish for a disconfirmation (e.g., where one's current knowledge
was unpleasant or undesirable), or one's wish to avoid c10sure that may
allow one to suspend one' s commitment and/or refrain from carrying
out an unwanted action (Brickman and Bulman, 1977; Kruglanski, 1989;
Snyder and Wicklund, 1981).
The foregoing discussion suggests that we should not expect a gen-
eral preference for similar (versus dissimilar) others as sources of informa-
tion, nor should we expect that information from similar (versus dissimi-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 149

lar) sources will have a generally greater impact. Let us consider the
relevant evidence for those assertions.

Motivational Effects and Comparison Similarity

Early studies of opinion comparisons interpreted the critical sim-
ilarity dimension as that of the opinion itself. This research has generally
identified moderating variables of the extent to which similar opinions
are preferred over dissimilar ones. For example, Brodbeck (1956) re-
ported that the tendency to choose a similar other for comparison varied
as an inverse function of subjective confidence. Worchel and McCor-
mick (1963) found greater liking for an agreeing other after the subject
encountered several previous disagreements as compared to several
agreements. Similarly, Shrauger and Jones (1968) found more liking for
agreers and disliking for disagreers when the answer to a judgmental
issue was unclear (versus clear) to subjects.
A possible basis for these effects could be motivational. For in-
stance, subjects lacking in confidence (e.g., after having encountered
several disagreements or when an answer was unclear) may feel a
stronger need to have their views confirmed. This may mediate a strong-
er comparison preference for similarly minded others (Brodbeck, 1956)
as weIl as a greater liking for those persons (Shrauger and Jones, 1968;
Worchel and McCormick, 1963).
In studies described in Chapter 3, Kruglanski and Mayseless (1987)
found that subjects made defensive concerning their opinion prepon-
derantly chose to compare with similar others, whereas subjects under a
fear of invalidity typically chose to compare with dissimilar others. Fur-
ther experiments reported by Kruglanski and Mayseless replicated the
finding that high (versus low) fear of invalidity induces a tendency to
compare with dissimilar others. In addition, they demonstrated that
high (versus low) need for cognitive closure (induced via time pressure)
causes a tendency to compare with similar others.
Thus, a tendency to choose an agreeing (similar) or a disagreeing
(dissimilar) other as a source of information seems to depend in part on
aperson' s motivational state: Agreeing or disagreeing others may be
chosen for comparison to the extent that the information they are ex-
pected to provide is compatible with the choosers' motivation. Similar
motivational effects emerged in studies that referred to similarity of
general attitudes or attributes related to the opinion rather than to the
similarity of opinion as such (cf. Goethals and Nelson, 1973). For exam-
pIe, Reckman and Goethals (1973) found that most subjects concerned
with accuracy preferred to compare with others who are dissimilar
150 CHAPTER 7

rather than similar to themselves on related attributes. The reverse pat-

tern obtained for subjects concerned with congenial association. Fazio
(1979) combined both agreement and similarity in one experimental de-
sign and found that if subjects' goal was belief confirmation (the "mo-
tive-for-validation" condition), their tendency to choose similar and/or
agreeing others was higher than if their goal was accuracy (the "motive-
for-construction" condition).
The generalization that a choice of an information source would be a
function of the chooser' s motivational state should apply to numerous
additional variables beside agreement or attribute similarity. For in-
stance, an individual under high need for nonspecific c10sure may prefer
a source characterized as authoritarian and expected, therefore, to pro-
vide a c1ear-cut opinion. In contrast, an individual under high need to
avoid c10sure may prefer a source characterized as creative and uncon-
ventional and expected, therefore, to provide alternatives to prevalent
points of view. A concerned relative may elect to consult with a physi-
cian who is expected to espouse an optimistic prognosis concerning the
patient's condition, whereas a competitor may selectively attend to ex-
perts who are pessimistic with regard to a rival's chances of success.
To conc1ude, the specification of conditions under which various
possible source characteristics (agreement, authoritarism, and opti-
mism) render it preferable for an individual seems highly complex and
laborious. Rather than constructing long lists of situationally specific
contingency rules, a more parsimonious approach may be to retain the
(transsituational) generalization that any source characteristic is prefer-
able to the extent that it promises to yield motivationally compatible
information. This generalization may then be applied to a specific situa-
tion of interest via a supplement of correspondence mIes specifying
what characteristic is motivationally compatible under the circumstances
at issue (Kruglanski and Klar, 1987).

The Role of Relevance

In a study on the effects of comparison, Goethals (1972) found that
agreement from a dissimilar other raised subjects' confidence more than
agreement from a similar other. This finding was qualified by results of
Goethals and Nelson (1973) who found that, in the realm of values,
subjects' confidence rose more following agreement from a similar (ver-
sus a dissimilar) other, whereas, in the realm of beliefs, confidence rose
more following an agreement from a dissimilar (versus a similar) other.
Based on those findings, Goethals and Darley (1977) theorized that,
when people make value judgments, agreement from a similar other has
greater impact than that from a dissimilar other, because someone with
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 151

divergent values may not seem relevant as a source of information. On

the other hand, when people evaluate their beliefs, agreement (but not a
disagreement) from a dissimilar other is given greater weight because it
is more informative than agreement from a similar other. The last pre-
diction is derived from Kelley's (1967) attributional analysis whereby
similar others may share one's own biases. In such a case, they may not
provide diagnostic information that would indicate whether the judg-
ments represent bias or the true nature of the entity. Broadly speaking,
the dissimilar other in this case has more relevant information than a
similar other, where relevance denotes informational diagnosticity
(Trope and Basok, 1983) that allows one to make a given judgment with
confidence.
In other words, both the greater impact of the similar other, when
the topic is values, and the greater impact of the dissimilar other, when
the topic is beliefs, seem mediated by the general judgmental variable of
perceived relevance. Specifically, a source seen to possess more relevant
information on a topic has, therefore, more impact on the individual's
confidence than a source with less apparently relevant information. This
analysis raises two issues for consideration: (1) whether similarity and
dissimilarity are uniformly related to relevance in the areas of values and
beliefs, and (2) whether relevance may not play the same role across
other possible classifications of judgmental topics.
00 similar (versus dissimilar) others generally convey the more rele-
vant information in the domain of values and preferences? This would
seem to depend on the particular dimension of similarity. Consider the
example discussed by Goethals and Oarley (1977):
Someone wants to decide which of two novels to take with hirn on a short
vacation. He is familiar with the authors of both books from other works they
have published. He makes the tentative evaluation that he would enjoy the
book by author A more than the one by author B, but would like to check it.
(p. 471)

Goethals and Oarley compellingly argue that in this case one would be
more likely to seek the advice of a similar (versus a dissimilar) other who
is likely to share one's literary tastes and preferences. In other words, in
evaluating what is good or worthwhile, we are likely to consult with
persons who share our criteria for goodness. Hence, we are more likely
to consult with a friend who shares our taste for political biography (i.e.,
a similar other), than one for whom good reading is primarily science
fiction. Relevant in this connection is an early study by Brock (1965) in
which a salesman who reported his own magnitude of paint consump-
tion. as similar (versus dissimilar) to a purchaser' s was more effective in
getting the purchaser to switch to a different price level. Apparently, the
similar salesman was seen as sharing the purchaser's criteria for (the
152 CHAPTER 7

magnitude of) a paint job, hence, having more relevant information to

the purchaser' s problem.
However, where similarity of criteria may be assumed, we may seek
the advice of someone with special ability to determine how a given
object measures up against those criteria. Thus, in selecting the films we
are about to view this summer, we may often heed the advice of the New
York Times critic (a dissimilar other) over and above that of our friend
who, though sharing our criteria for good entertainment, may be less
reliable in determining whether those are met by a particular movie.
Occasionally, we might worry that our preference for a given entity
has been biased by contextually induced factors. In such a case, we
might prefer to consult with someone who has not been subjected to the
same kind of influence, and in this sense is a dissimilar other. For exam-
pIe, we might suspect that our liking for an expensive sports car has
been unduly affected by the charisma and techniques of the sales person
at our neighborhood dealership. To check up on our impressions, we
might prefer to consult with a friend who has not accompanied us on
our car-shopping visit than with one who has.
These examples suggest that in the domain of values the relation of
similarity to relevance is not uniform. Depending on the particular di-
mension of similarity, a similar or a dissimilar other could be seen to
have the more relevant information and thus be selected for comparison
purposes. As we shall see, the same conclusion applies to the domain of
beliefs. Admittedly, when concerned about his or her own bias, a per-
son may prefer to seek the opinion of someone who does not share the
same bias, that is, a dissimilar other (Goethals and Darley, 1977). How-
ever, the hypothesis of bias may not be invariably accessible to the
person (Higgins and King, 1981). When it is not, a person perceived to
have a different bias need not be preferred over one with a similar bias.
A similar (versus a dissimilar) other may be perceived to have the
more relevant information in other ways beyond the issue of bias, and,
therefore, be preferred for comparison purposes. For an expert, other
(similar) experts may be perceived as more relevant sources of informa-
tion than nonexperts. Similarly, a person with an interest in some ac-
tivity (e.g., football or the stock market) may expect others who share
his or her interests (Le., similar others) to keep up with developments in
that field and prefer them as informational sources over others who do
not share such an interest.
This discussion suggests that similarity versus dissirnilarity is not
uniformly related to informational relevance in the domains of values or
beliefs. Furthermore, it seems that relevance plays the same role across
other possible classifications of interest domains. For example, a rele-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 153

vant other may be preferred as a source of information in the domain of

emotions (Schachter, 1959) or of motor behaviors (e.g., the tennis pro, the
ballet instructor). Although not presently attempted, it can be readily
shown that, in those alternative domains as weIl, there does not seem to
hold a uniform relation between similarity and relevance.

Accuracy o[ Comparisons
An early study by Radloff (1966) claimed support for the prediction
that comparison with similar others yields greater judgmental accuracy
than comparison with dissimilar others. In this research, subjects en-
gaged in a variable-speed pursuit motor task and received feedback
about their relative performance. Subjects who were informed that they
occupied extreme positions on the performance scale and thus were
dissimilar from most other subjects were subsequently less accurate in
their self-evaluations than subjects who were told they are average,
hence, similar to others.
However, Radloff's results are open to an alternative interpretation
whereby lowered accuracy was not caused by the extremity of subjects'
position per se but rather by a surprise at finding themselves at an ex-
treme position away from everyone else. Such information could be
inconsistent with subjects' expectancy, thus undermining their confi-
dence in self-evaluations. In turn, lowered confidence may foster confu-
sion and doubt adversely affecting accuracy.
Those notions were tested in arecent study by Mayseless and
Kruglanski (1987) that manipulated orthogonally similarity of the com-
parison others (operationally defined via performance extremity) and
consistency with expectancy. As predicted, inconsistency with expec-
tancy exerted adverse effects on the accuracy of self-evaluations,
whereas dissimilarity from the comparison other did not. Thus, internal
consistency rather than similarity seems related to evaluative accuracy.
Our analysis of sodal comparison as the use of sodal sources of
information suggests the following conclusions: A sodal source is likely
to be preferred and to exert impact to the extent that its information is
compatible with the individual's motivations and is considered relevant
to the topic at hand. Moreover, the epistemic analysis suggests that the
internal consistency of information may affect the accuracy of pertinent
judgments. Neither motivational compatibility nor relevance or con-
sistency seems uniformly related to the similarity to oneself of the com-
paris on other. Hence, the traditional variable of interpersonal similarity
may playa less systematic or unique role in sodal comparison phenome-
na than has been previously assumed.
154 CHAPTER7

THE MOTIVE BASE OF SOCIAL COMPARISON PHENOMENA

In the original version of sodal comparison theory, Festinger (1954,

p. 124) postulated the existence "in the human organism of a drive to
evaluate his opinions and abilities," presumably in order to function
effectively in one's environment (Festinger, 1954). In addition, Festinger
(1954, p. 124) postulated a "unidirectional drive upward in the case of
abilities." Whereas the need for evaluation seems nondirectional and
unbiased, the unidirectional drive upward appears biased toward com-
parisons that allow the inference of one's superior ability. Further simi-
lar motivations assumed to directionally bias sodal comparisons of abil-
ity were subsequently suggested and studied; for example, the need for
self-enhancement (Gruder, 1971), the need to increase subjective weIl··
being (Wills, 1981), the motivation to maintain a positive self-evaluation
(Tesser and Campbell, 1982) or to feel positively distinctive and unique
(Campbell, 1986).
Although most studies on biased motivational effects centered on
the comparison of abilities, Goethals and Darley (1977, pp. 264-265)
identified a directional motive in the domain of opinions as weIl, notably
the "desire to validate them and to obtain evidence that they are in fact
correct." This motive is assumed to exist in addition to the largely non-
directional desire to "objectively evaluate (one's) opinions."
The lay epistemic analysis elaborates the motivational base of sodal
comparison phenomena in two ways: (1) by refining the notion of non-
directional evaluative needs assumed basic to the comparison process,
and (2) by expanding the conception of directional biases in sodal com-
parison. Let us consider these topics in turn.

Nondirectional Motives tor Social Comparison: Opposing Closure and

Closure-Avoidance Effects
A person who wishes to evaluate his or her opinions or abilities may
desire actually two separate things: he or she may wish to have definite
knowledge on the topic at issue and may wish the knowledge to be
accurate. The first wish represents what we have earlier referred to as
the need for nonspedfic c1osure. The second wish may arouse the need
to avoid or postpone c1osure. The needs for c10sure and c10sure avoid-
ance may often pull the sodal comparison process in opposite direc-
tions. Need for c10sure may motivate comparisons that enable one to
c1ing to a preexisting opinion, whereas need to avoid premature c10sure
may motivate comparisons that promise to undermine such an opinion
(Freund et al., 1985; Kruglanski and Freund, 1983; Kruglanski and May-
seless, 1987). Thus, depending on the relative magnitude of the c10sure
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 155

and c1osure-avoidance concerns in a given instance, sodal comparison

tendendes may assume different forms.
According to the present conception, self-evaluative motivations
are topic spedfic (Kruglanski, 1987). They are not necessarily thought of
as drives (Festinger, 1954; Schachter, 1959) apt to arise inevitably in
cases of low confidence or ambiguity. Thus, not every lack of knowledge
is assumed to represent adesire for knowledge. Occasiona11y, it may
fulfill adesire for the avoidance of c10sure and afford a welcome suspen-
sion of commitment (Brickman and Bulman, 1977; Kruglanski, 1989;
Snyder and Wicklund, 1981). In sum, persons may or may not wish to
know various things about themselves or the world, and such needs for
c10sure may vary across topics, occasions, and situations (Cottrell and
Epley, 1977; Goethals, 1987; Decarufel and Insko, 1979).

Directional Biases in Social Comparison Processes: Needs tor Specific Closure

The present construct of need for spedfic c10sure is a general term
that covers a variety of directional biases assumed to affect sodal com-
parisons. Some spedal cases of such a need are (1) the unilateral drive to
do better and better, (2) the desire to increase subjective we11-being, or to
obtain evidence that one is in fact correct and good. Two possible the-
oretical benefits of using a general motivational term instead of spedfic-
need rosters are parsimony and generative value. Thus, the construct of
need for spedfic c10sure identifies a functional equivalence of the sepa-
rate directional motivations in the sodal comparison process: They are
a11 assumed to facilitate comparisons likely to yield the desired types of
evaluative knowledge. Furthermore, the concept of specific closure al-
lows one to generate numerous directional needs previously unre-
corded, yet which, under the appropriate drcumstances, may similarly
bias sodal comparisons toward desired knowledge. For example, a need
for physical safety may bias comparisons in the direction of safety-
asserting conc1usions; a need for affection, in the direction of affection-
asserting conc1usions.

DOMAIN DIFFERENTIATION OF SOCIAL COMPARISON PHENOMENA:

THE CONTENT-PROCESS DISTlNCTlON

A characteristic implication of previous work has been that sodal

comparison processes differ across domains of comparison. According-
Iy, domain distinctions were proposed to reflect the putative differences
in process. Festinger (1954) drew the widely known distinction between
the comparison of "opinions" and "abilities," Schachter (1959) identified
the further domain of "emotional" comparisons, and Goethals and
156 CHAPTER 7

Darley (1977) distinguished between the comparison of "values" and

"beliefs" .
Interestingly, alongside the trend of domain differentiation an op-
posite trend toward integration was apparent. Specifically, several anal-
yses addressed social comparison processes occurring across domains
and treated them from more general perspectives. For example, Singer
(1966) proposed that people compare with others in order to evaluate
generally their self-esteem rather than assess only an isolated opinion or
ability. Pettigrew (1967) proposed including sodal comparison theory
within a general theory of social evaluation. Mettee and Smith (1977)
suggested integrating the theory of social comparison within a more
comprehensive theory of self-knowledge. Goethals (1987) proposed that
social comparison theory should deal with the general question of self-
evaluation.
The present lay epistemic analysis of sodal comparison phenomena
continues the latter integrative trend. Not only are social comparison
processes viewed as a special case of self-evaluation or even of self-
knowledge, but also of knowledge acquisition in general, including
knowledge of objects and events unrelated to self. All cases of knowl-
edge acquisition are assumed to deal with persons' attempts to form
some opinions. In this sense, abilities' comparison is a special case of
opinions' comparison, namely, of opinions about one's abilities Gones
and Gerard, 1967). In the same way, emotional comparisons (Schachter,
1959) refer to people's opinions about their emotions, whereas value
comparisons (Goethals and Darley, 1977) refer to people's opinions
about their values. Let us now consider what this characterization im-
plies for specific distinctions in the process that is assumed to separate
the different comparison domains.

Abilities versus Opinions

(1) The Unidirectional Drive Upward. Festinger (1954a, p. 124) as-
sumed the existence of "a unidirectional drive upward in the case of
abilities which is largely absent in opinions." As already noted, this
drive upward is presently assumed to constitute a special case of the
need for specific closure. If so, (1) appropriate counterparts of the drive
upward should be represented outside the abilities domain, because
needs for specific closure may arise with respect to different content
domains of knowledge. (2) Even within the abilities domain, the drive
upward need not be invariably present, because needs for specific
closure are not necessarily linked in a stable fashion to any knowledge-
acquisition situation. (3) Directly opposite tendencies to the drive up-
ward may be occasionally present in the abilities domain, because op-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 157

posite contents of c10sure may be occasionally preferred in different

circumstances.
In support of the first implication, "false-consensus" research
(Campbell, 1986; Marks, 1984) suggests that a counterpart of the biasing
drive upward, namely, a biased preference for the conc1usion that one
was right, may affect the comparison of one's opinions. Relevant to the
second implication, work by Trope (1975, 1980) has demonstrated that,
in the abilities domain, people may often be interested in accurate as-
sessment, rather than in exhibiting an upward bias. As for the third
implication, it seems plausible that occasionally persons might find it
desirable to conc1ude that their ability in some domain is lower than that
of others. For instance, the belief that one is helplessly confused in math
may invite the sympathy and assistance of one' s friends; the conc1usion
that one is a poor golfer may excuse one from competing with one' s boss
on Sunday. I am not referring here to mere self-presentational overtures
designed to elicit desirable outcomes from others, but to motivational
biases in one' s authenthic beliefs that, once adopted, may promote desir-
able consequences without being accompanied by the unpleasant feel-
ing that one has been manipulative and insincere.
In passing, it is noted that the remaining epistemie motivations,
notably the nonspecifie need for elosure and the need to avoid non-
spedfie c1osure, are also assumed to affect the formation of opinions in
general and thus to be equally pertinent to various possible domains of
sodal comparison. For example, under a high need for c1osure, a person
may freeze on a contextually accessible evaluation of his or her ability
even if it is an unflattering one (Ross, Lepper, and Hubbard, 1975).
Similarly, under a high need to avoid or postpone closure (induced,
e.g., by accuracy concerns), a person may be ready to unfreeze a posi-
tive self-assessment and pay elose attention to information inconsistent
with such an assessment.

(2) Malleability of Abilities and Opinions. According to Festinger

(1954a), "there are non-sodal restraints whieh make it diffieult or even
impossible to change one's ability. These non-sodal restraints are largely
absent for opinions" (p. 125). Note that this juxtaposition of abilities and
opinions does not really apply if the sodal comparison of abilities actu-
ally concerns opinions about one' s relative abilities. An alternative inter-
pretation of the distinction at issue might, therefore, be that opinions
about one's abilities are generally less malleable than opinions about
other objects. On reflection, however, this may depend strictly on the
ability or the object in question. For instance, Ross et al. (1975) effected a
quiek change in subjects' opinion about their ability to identify accu-
rately authentie suidde notes by simple experimental feedback. In con-
158 CHAPTER7

trast, opinions about issues other than abilities, for example, about ethi-
cal and aesthethic values or ideological doctrines, may be extremely
rigid and resistant to change. In general, a readiness to change one's
opinion is presently assumed to depend on cognitive and motivational
factors (e.g., the accessibility of alternative opinions or a need for
c1osure) that affect the epistemic process at large, rather than on the
contents of spedfic opinions (e.g., having to do with abilities or with
alternative issues).

The Value-Belief Distinction

From this perspective, recall Goethals and Darley's (1977) distinc-
tion between the comparison of values versus beliefs. As already im-
plied, it is possible to think of value comparisons as referring to beliefs
about values. In this context, the proposal that, in the domain of values,
similar others are often preferred as a source of information over dis-
similar others was shown to represent the generalization that a source
with more apparently relevant information concerning a belief domain is
preferred over a less-relevant source. Furthermore, recall that similarity
does not seem to be related systematically to relevance in any given
belief domain, inc1uding that of values. In short, the process of sodal
comparison (e.g., the role of informational relevance or of motivation)
seems uniform across content domains of comparison, for example, do-
mains of values, abilities, or other topics.

SOCIAL COMPARISON PHENOMENA: THE EPISTEMIC PERSPECTIVE

It is time now to take stock and collect the three main conc1usions of
our epistemic analysis. First, according to the present approach, it does
not seem justifiable to assume that individuals exhibit a pervasive ten-
dency or drive toward sodal comparison. Instead, the term social com-
parison may be conceived of as a general category of self-appraisal ques-
tions in which persons may or may not be interested in given
circumstances.
Second, the similarity to oneself of the comparison other does not
appear as central as previously assumed. The content of the comparison
question seems generally independent of the similarity dimension: Even
though persons occasionally might wish to know how they compare to
similar others, often they may be interested in comparing with dissimi-
lar others, and on yet other occasions the comparison dimension of
interest may be unrelated to the others' similarity or dissimilarity.
Nor does the similar other appear to playa unique role as a source
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 159

of information. Rather than be affected by the similarity variable, the

preference for an informational source and its impact on subjects' opin-
ion seem mediated by such general judgmental variables as perceived
·relevance, internal (logical) consistency, and motivational compatibility with
one's needs for information provided by the source.
Third, the process of sodal comparison appears uniform across dif-
ferent content domains of comparison. Biased (unidirectional) and un-
biased motivational tendendes are assumed possible with abilities' com-
parisons as weIl as with other possible comparison topics. Furthermore,
no apriori relation seems to exist between the content of an opinion
(e.g., having to do with ability) and its resistance to change. FinaIly,
considerations of source relevance appear equally applicable to various
comparison domains.

PROPOSED REORIENTATIONS

The present distinction between the content of the comparison

question and the judgmental process underlying sodal comparison sug-
gests some potentially useful reorientations for research in the area.

Content Issues
The spedfic conte nt of the comparison question may have impor-
tant psychological implications. For instance, many comparisons in-
volve an implidt evaluative dimension (Pettigrew, 1967; Singer, 1966).
Often, individuals do not just wish to know how they measure up
against others but rather how weil they are doing in reference to those
persons (e.g., how correct is their opinion, how impressive their perfor-
mance, or how appropriate their emotion). This suggests that com-
parisons may often include an affective dimension. An individual who
does as weIl as or better than the others would presumably experience
positive affect, whereas an individual who does not do as weIl as others
would experience negative affect.
Furthermore, the type of affect experienced may relate to the con-
tent of the comparison. Recent work by Higgins and colleagues (e.g.,
Higgins et al., 1985) suggests that discrepandes from different com-
parison standards may induce such distinct affects as agitation versus
dejection. Thus, depending on whether the comparison other repre-
sents a moral standard of duty (or "ought") or a desired attainment (Le.,
an "ideal"), various comparison outcomes may lead to correspondingly
different affective states.
Beyond their affective implications, different contents of com-
parison may have diverging implications for action: Discrepandes from
160 CHAPrER7

some standards (e.g., of skilled performance) may appear bridgeable

and inspire the investment of considerable effort aimed at their removal.
Other discrepandes (e.g., from standards of appearance) may appear
stable in character and effect the withdrawal of effort. The affective and
behavioral consequences of different comparison contents thus appear
to suggest fruitful directions for future sodal comparison research.

Proeess Phenomena
It may be of interest to investigate the conditions under which sodal
comparison questions of whatever content may get activated. Contex-
tual priming could account for cases in which strong interest in given
sodal comparison outcomes is momentarily aroused, without the per-
son being necessarily aware of the origins of his or her comparative
concerns (cl. Bargh, 1984). Another issue, of possible applied signifi-
cance, is whether comparison tendendes may be brought under volun-
tary control. Suppose that "buming" sodal comparison interests were
aroused at a professional meeting, resulting in dysfunctional tension
and unpleasant affect. By what mechanism may one divert one' s atten-
tion away from such comparative obsessions and focus it on more adap-
tive pursuits? Such questions, too, may be fruitfully explored in future
comparison research.
FinaIly, it is of interest to consider the implications of viewing the
answers to previous sodal comparison questions as relatively stable
knowledge structures. For instance, one's feelings of "inferiority" or
"superiority" (Adler, 1964) may reflect the availability of the correspond-
ing knowledge structures conceming one' s relative abilities or at-
tributes. Future research may examine the conditions under which such
structures are activated and/or modified under new relevant evidence.
In short, the various process issues noted above seem weIl deserving of
research attention in the sodal comparison area.

Dynamie versus Statie Views 01 Social Comparison

Early conceptions of sodal comparison were relatively static in char-
acter. People were seen as nearly always wishing to compare, doing so
mostly with similar others, and nearly always striving to attain evalua-
tive accuracy. The view of sodal comparison suggested by the present
analysis (see also Brickman and Bulman, 1977; Goethals and Darley,
1977; and Metee and Smith, 1977) is very different. According to this
portrayal, persons may not always desire to compare;moreover, even
when they do, this need not be in reference to similar others. Similarly,
accuracy, although often an important concern, may not be the exclusive
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 161

or predominant basis of sodal comparisons. Thus, comparison ten-

dendes are viewed as highly fleeting and dynamic, contingent as they
seem on the overall psychological context affecting the person' s
judgments.

Conclusion
Festinger's (1954a) theory of sodal comparison processes has had
considerable impact and has genera ted a wealth of valuable research
findings. In this sense, it has been of considerable historical utility and
importance (Hall and Lindzey, 1978; Shaw and Constanzo, 1982). How-
ever, the many conceptual and empirical difficulties with some of the
theory' s central postulates suggest the benefits of a reorientation where-
by sodal comparison phenomena are treated as a special case of the
general epistemic process.

MINORITY INFLUENCE

Although the social cognitive approach is generally considered as

an intrapersonal perspective, we have just examined its relevance to
such an interpersonal process as social comparison. Moreover, the ap-
proach may be similarly applied to issues in group psychology. In the
sections that follow, this is illustrated by an epistemic comparison of
minority and majority influence.

POPULAR "REPUTATIONS" OF MINORITIES AND MAJORITIES

The term minority often primes such characteristics as depressed,

underprivileged, or exploited whereas the term majority is typically
taken to connote such attributes as dominant, ruling, or oppressive.
From that perspective, minority and majority situations appear poles
apart and in need of vastly different psychologies for their explication.
The question is whether these considerable surface dissimilarities might
not conceal deeper similarities, and whether what seems manifestly
different is not latently the same.
A lay epistemic analysis of majority and minority phenomena as-
sumes that sodal influence (whether of a majority or a minority or vice
versa) involves, at least in part, the formation and/or shaping of knowl-
edge structures by forces emanating from the influence source. There-
fore, variables that affect the knowledge-formation process in general
should prove relevant also to social influence phenomena.
162 CHAPTER 7

CATEGORIES OF FACTORS AFFECTING JUDGMENT FORMATION AND

CHANGE

The epistemic theory suggests three categories of variables that af-

fect the formation of knowledge: these are the informational, cognitive,
and motivational variables.
By informational variables, I me an the contents of evidence for or
against given knowledge. For example, "covariation" and "temporal
precedence" are informational contents that imply a causal knowledge.
Other types of knowledge may be supported or refuted by alternative
contents of information, all in accordance with the perceived relevance
of the information to the judgments at issue.
In contrast to informational variables that refer to contents, cogni-
tive and motivational variables belong with the process whereby all
knowledge is formed. In previous chapters, I frequently referred to such
cognitive factors as availability or accessibility, and to such motivational
factors as needs for nonspecific and specific closure or the need for the
avoidance of closure.
Let us therefore consider some informational, cognitive, and moti-
vational antecedents of minority and majority types of influence. Begin-
ning with the informational category, we immediately encounter the
major factor assumed to be involved in minority-influence phenomena:
the variable of minority consistency.

Consistency Effects
Workers in the minority-influence area agree that minority influ-
ence is carried predominantly via its behavioral style. In turn, a major
element of an influential style is considered behavioral consistency. The
central premise of minority-influence theory is that "consistent behavior
by minorities will exert influence, whereas inconsistent behavior will fail
to bring about any change of the majority's attitudes and perceptions"
(Maass and Clark, 1984, p. 429). The reason that consistency effects are
presently classified as informational is that they are assumed to produce
attitude change via the attribution of certainty and competence to mem-
bers of the minority (Maas and Clarke, 1984). Consistency is thus con-
sidered to represent specific information suggesting the conclusion that
the source, in this case, the minority, is "confident" and/or "compe-
tent." It is the latter conclusion that presumably mediates minority influ-
ence, because many people seem to subscribe to the belief that a compe-
tent and a confident source possesses a valid knowledge and, hence, is
worth listening to.
It seems plausible to assurne that assumptions of confidence and
competence would mediate the influence of any source, not just a mi-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 163

nority source. In fact, support for such an assumption does exist in the
area of majority influence. Already in the Asch paradigm (1952, 1956), it
was found that an intemally inconsistent majority (i.e., a majority with a
defector) produced less conformity than a fully consistent majority. Not
only was this true where the defector supported the subject' s percep-
tions but also where the defector deviated from those even further than
the majority (the extreme-dissenter condition). Furthermore, where the
majority' s rate of incorrect items increased, suggesting greater consis-
tency of the innovative response, majority influence increased as well.
More recently, importance of consistency among members of a majority
has been further demonstrated by Allen and Levine (1968), Doms and
van Avermaet (1980), and Doms (1984). Thus, consistency effects appear
unrelated to whether the influence source is in a minority or in a
majority.
The present analysis suggests that content of information indicating
that judgments espoused by a source are valid would enhance persons'
tendency to render similar judgments. Not only does this general prind-
pIe appear applicable to both minority and majority cases, but the specif-
ic contents of information, indicating source competence (notably, infor-
mation conceming consistency), also yield similar effects in both cases.
At least thus far then, minority and majority effects seem capable of
being mediated by the same judgmental variables.

Motivational Factors in Minority-Influence Research

Beyond various informational effects, minority-influence research
illustrates the role of epistemic motivations in the judgmental process.

Needs for Specific Closure

(1) Zeitgeist Effects. Paicheler (1976, 1977) noted that a minority
whose advocacy was consistent with the (profeminist) norm trend
(Zeitgeist) was more effective than a minority whose advocacy ran coun-
ter to the Zeitgeist (see also Maass, Clark, and Haberkom, 1982;
Paicheler, 1979). One interpretation of such Zeitgeist effects is that a
new idea ts more likely to be influential, the more desirable it is to the
individual or the more congruent it is with his or her preferential
closures. The sodal norm of Zeitgeist defines what is sodally desirable;
hence, all else being equal, advocacy consistent with the Zeitgeist
should be more effective than advocacy inconsistent with the Zeitgeist.
Again, it seems that Zeitgeist effects, or specific closure effects in
general should not be unique to minority sources; rather, they should
apply alike to all sources of influence, including majority sources. For
instance, a major sodal psychological finding in the 1950s has been that
164 CHAPTER 7

the degree to which an individual is attracted to a group positively

affects the degree to which he or she tends to accept the group' s norm
(cf. Back, 1951; Festinger, 1950; Festinger, Schachter, and Back, 1950;
Schachter, 1951). Being attracted to the group renders acceptance by the
group more desirable and rejection less desirable; hence, it renders
group views or norms more desirable as weH, lending them the status of
specific closures. In this sense, group cohesion effects on the acceptance
of influence may represent the workings of the need for preferred
closure in majority contexts.

(2) Closure-Avoidance Effects. In some minority-influence studies

(e.g., Nemeth, Swedlund, and Kanki, 1974), serious consideration of
minority ideas may have been enhanced by subjects' need to avoid
closure (cf. Kruglanski, 1987) aroused by the experimenter's instructions
to remember the other participants' opinions. This may have increased
subjects' tendency to attend to the minority views and lessened their
readiness to discount them in the interest of closure (Anderson and
Hubert, 1963). However, needs for closure or the avoidance of closure
should exert their effects irrespective of whether the inconsistent idea
belonged to a minority or a majority. In sum, various motivational medi-
ators of minority-influence phenomena appear generaHy applicable to
the epistemic process and, hence, equaHy relevant to cases where the
influence source is in a minority or a majority.

Cognitive Factors
(1) Accessibility. Finally, some minority-influence effects may be in-
terpreted in terms of cognitive variables affecting the process at large.
For instance,' consistent repetition of the minority position might exert
its effects via enhanced accessibility of the minority viewpoint, for, as
we know, accessibility of a construct is a positive function of its frequen-
cy of activation (Higgins et al., 1985; Wyer and SruH, 1981).

(2) Conflict and lnconsistency. Another cognitive variable, often in-

voked in reference to minority-influence phenomena, is cognitive con-
flict or the inconsistency that a minority position may introduce
(Moscovici and Mugny, 1983). It is noteworthy that the introduction of
conflict or inconsistency is not unique to minorities. Thus, in the Asch
conformity paradigm, it is the bogus majority rather than the minority
that introduces the conflict.
Furthermore, whenever introduced, an inconsistency presumably
creates uncertainty and undermines confidence. Thus, unique minority
influence may not stern from the break with existing reality, or the
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 165

uncertainty engendered by the inconsistency, but rather from the subse-

quent cognitive process that may unfold in reaction to the inconsistency.
According to Moscovici (1980) minority-induced conflict triggers a "val-
idation" process in which the validity of the contradicting opinions is
assessed by a elose examination of the object in question. Such active
information processing is likely to produce a modification of the pre-
viously held attitudes (Maass and Clark, 1984, p. 444). By contrast, a
majority in the conformity paradigm was assumed to trigger a com-
parison process, in which the subject will simply compare the contra-
dicting opinions with each other and, presumably, decide in favor of the
majority opinion.
Thus, our problem is (1) to identify the conditions under which an
introduction of cognitive inconsistency will promote passive conformity
to one of the opposing views, or an active and independent assessment
of the issues involved, and (2) to decide whether such conditions are
correlated with minority or majority contexts. A renewed consideration
of motivational factors in the epistemic process should help us in ad-
dressing those tasks.

CONFORMITY VERSUS INDEPENDENCE IN MINORITY AND MAJORITY

CONTEXTS

First, note that cognitive inconsistency undermines closure. How

the individual reacts to such an event may thus depend on his or her
motivations in regard to elosure. Under a high need for (nonspecific)
elosure he or she might either vehemently reject the inconsistent idea as
false or immediately accept it as true, hence, relinquishing the former
position (Kruglanski and Mayseless, 1987). Both solutions would
promptly resolve the inconsistency and should therefore be equally ac-
ceptable to the individual. In other words, und er high need for closure,
an individual may accept the more plausibly appealing idea irrespective
of whether it belonged to a minority or a majority.
Often, however, persons' motivation may prompt them to prefer a
solution because of its source. For instance, the individual may emo-
tionally identify with, or desire acceptance by, a majority. This may
arouse a need for specific closure that may render majority solutions
particularly attractive. In such circumstances, an individual may tend to
accept majority solutions and creatively generate alternatives to minor-
ity solutions. However, an individual might also be attracted to a minor-
ity (e.g., because of its power, status, deviancy, or originality) and may
be repulsed from a majority (e.g., because of its banality, anonymity, or
conservatism). This might arouse a motivation to accept the minority
solution and oppose the majority solution.
166 CHAPTER 7

In yet other circumstances, the individual may be disinclined to

accept either the majority solution that might brand hirn or her as an
unimaginative conformist, or the minority solution that might brand
hirn or her as an opportunistic "bandwagon jumper." Under those con-
ditions, the individual might be strongly motivated to generate his or
her own ideas and ways to resolve the problem.
Thus, whether an individual accepts or rejects solutions proposed
by others or chooses to generate his or her own solutions should depend
on the person's epistemic motivations. Nor is it obvious that those moti-
vations would be systematically correlated with minority versus major-
ity contexts. Further research seems needed to explore the specific con-
ditions under which minorities or majorities may induce motivations
that prompt cognitive conformity or independence.

COMPLIANCE VERSUS INTERNALIZATION IN THE MAJORITY AND

MINORITY PARADIGMS

A major line of evidence for the presumptive difference in the pro-

cess of mediating the influence of minorities and majorities refers to the
putatively disparate outcomes of that process. In particular, it has been
proposed (Maass and Clark, 1984; Moscovici, 1980) that minority influ-
ence prornotes private change or "intemalization" (Keiman, 1958),
whereas majority influence promotes public change or "compIiance"
(Keiman, 1958).
The distinction between intemalization and compIiance resembles
that between "informational" versus "normative" influence (Deutsch
and Gerard, 1955; Jones and Gerard, 1967). Informational influence has to
do with the impact of others' perceived opinions on one's own private
views. In this sense, informational influence may induce an authentic
acceptance of others' suggestions and thus an intemalization. By con-
trast, normative influence hinges on the rewards and punishments the
group can mete out to the individual and/or withhold as sanctions for
nonconformity. In this sense normative influence may predominantly
foster public compliance with the source of pressure, aimed at averting
the implicit sanctions, and will not necessarily be accompanied by
intemalization.
The question is whether informational influence is generally re-
served for minority contexts and normative influence for majority con-
texts. To begin with, studies in the majority (or conformity) paradigm
seem to demonstrate pervasively the occurrence of informational influ-
ence as weIl as of normative influence Gones and Gerard, 1967). Infor-
mational influence was evidenced in situations in which conformity
rates were enhanced by the introduction of difficult, complex, or sub-
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 167

jective judgments (Asch, 1952, Coleman, Blake, and Mouton, 1958;

Deutsch and Gerard, 1955; Kelley and Lamb, 1957), or in which subject's
conformity was inversely proportionate to his or her degree of relative
expertise as compared to the disagreeing majority. In contrast, norma-
tive influence was evidenced primarily in studies in which conformity
rates increased if subjects' responses were public versus private (Argyle,
1957; Deutsch and Gerard, 1955; Mouton, Blake, and Olmstead, 1956;
Raven, 1959).
Thus, the empirical evidence seems to indicate that majorities are
capable of exerting informational as well as normative influence. This
implies that, in certain contexts, majorities ought to induce internaliza-
tion. But can minorities induce compliance?
To answer this question, consider the compliance and internaliza-
tion phenomena more c1osely. In asense, they seem to represent two
distinct contents of judgment persons might variously wish to make. In
case of compliance, the person seems interested in judging or deciding
what to do. Let us call this type of judgment action appraisal. In case of
internalization, the person seems interested in judging what to think.
Let us call this type of judgment issue appraisal. Whether an individual
has an interest in an action appraisal or an issue appraisal may depend
on a variety of factors (Kruglanski, 1980) as, for instance, the relative
subjective utility of the two types of judgment. A person subjected to a
congressional inquiry may wish to know what to say in response to
various questions. Thus, he or she may be interested in action appraisal.
However, members of the investigating committee who may wish to
know what to think may be interested in opinion appraisal.
Again, the question is whether minority contexts generally promote
issue appraisals, and majority contexts, action appraisals. Apriori, no
such contingency seems probable. An individual may be effect depen-
dent on a minority member (e.g., he or she may desire the identity of a
rebel or of a revolutionary, which only acceptance by a minority might
confer); alternatively, he or she may fancy the gratitude and admiration
of minority members (the status of a big fish in a small pond, as it were).
Such a person might be quite interested in action appraisal, notably in
finding out what to do in order to curry favor with the minority. Sim-
ilarly as we have seen, an individual who is informationally dependent
on a majority may be quite interested in issue appraisal.
The implied independence of minority and majority influence from
concerns with issue versus action appraisal raises the possibility that the
contingency reported in past research may be telling only apart of the
story. Indeed, in previous studies (e.g., Moscovici and Lage, 1978),
majority-influence manipulations might have typically introduced con-
cern with action appraisal, for example, by arousing a fear of embarass-
168 CHAPTER 7

ment by a group of unanimous peers. In contrast, minority innovations

might have typically produced concern with issue appraisal, by an intro-
duction of a puzzling variability. However, it should be possible to
design situations in which action appraisals are aroused in minority
contexts and issue appraisals by majority contexts. Admittedly, the
above suggestions are presently speculative and in need of a careful
empirical corroboration.

CONCLUSION

It seems, in conclusion, that a general epistemic framework may

provide a useful way of looking at phenomena of minority and majority
influence. In the foregoing application, the epistemic perspective has
tended to support the view of a unitary mediating process of these two
types of influence, although the final verdict must await further inves-
tigation. On the side of the unitary view, several minority-influence
effects were interpreted in terms of informational, cognitive, and moti-
vational factors likely to moderate similarly majority-influence effects.
Furthermore, our analysis suggested that phenomena of conformity ver-
sus independence and of internalization versus compliance are orthogo-
nal to the minority or majority status of an influence source. More im-
portantly, the epistemic analysis suggests the conditions under which
those phenomena may be manifested; hence, it may provide a useful
guide for further relevant research. Analyzing the phenomena of minor-
ity and majority influence from an epistemic perspective may thus fur-
nish an integrative framework for empirical findings and theoretical
constructs that workers in the area (e.g., Maass and Clark, 1984) have
found lacking.

SUMMARY

This chapter applied the lay epistemic perspective to two further

sodal psychological topics: sodal comparison processes and minority-
influence phenomena. Spedfically, the lay epistemic analysis implies a
reorientation in thinking about and in investigating sodal comparison
issues. It has been noted that persons do not seem to exhibit a pervasive
drive toward sodal comparison; rather, they may or may not pose vari-
ous comparative questions depending on the drcumstances. In addi-
tion, the similarity to oneself of the comparison other does not appear as
central to the comparison process as has been assumed. Finally, the
motivations that affect sodal comparison tendendes seem to be the
same ones that affect all judgmental phenomena; other variables of
COMPARISON PROCESSES AND INFLUENCE PHENOMENA 169

judgmental significance, for instance, informational relevance, seem to

affect sodal comparison in the same way.
The lay epistemic analysis suggests that future sodal comparison
research could be profitably refocused on pertinent content and process
issues. For instance, it would be of interest to investigate the types of
affective phenomena that accompany different contents of sodal com-
parison. Furthermore, it would seem important to investigate the cogni-
tive and motivational variables that facilitate sodal comparison ten-
dendes and those that inhibit them.
Applying the lay epistemic analysis to the topic of minority influ-
ence highlights the fact that it may be influenced by the very same
factors that underlie majority influence. Further research on both types
of influence may be guided by constructs and hypotheses derived from
the present concept of the epistemic process. The phenomena of both
sodal comparison and minority influence pertain to interpersonal effects
of considerable interest to sodal psychologists. The present discussion
demonstrates, therefore, that a sodal cognitive level of analysis is not
necessarily restricted to intrapersonal phenomena. 1

IFor a lay epistemic analysis of international conflict, for example, see Bar-Tal, Kruglanski,
and Klar (1988).
 8
The Issue of Accuracy in Social
Perception and Cognition

Interest among researchers in the accuracy of sodal perception has

never waned for long. Admittedly, early accuracy work came to a halt
following the publication of critiques of Cronbach and Gage (Cronbach,
1955, 1958; Gage and Cronbach, 1955; Gage, Leavitt, and Stone, 1956)
and was supplanted by research on the judgmental process Gones, 1985;
Schneider, Hastorf, and Ellsworth, 1979). The process models, however,
though initially descriptive, soon acquired prescriptive or "normative"
overtones (Funder, 1987). Accordingly, researchers' interest in recent
years has centered on persons' tendency to stray from optimal or pre-
scribed modes of judgment (e.g., as embodied in models of statistical
inference), and the emphasis shifted from the study of process per se, to
the study of bias or inaccuracy (Nisbett and Ross, 1980; Tversky and
Kahneman, 1974).
For over a decade, research on judgmental biases and errors has
exerted a dominant influence on views of the human cognitive process.
Recently, however, several authors took issue with the conclusion that
lay persons are incorrigibly inaccurate and error prone (cf. Einhorn and
Hogarth, 1981; Funder, 1987; Hastie and Rasinsky, 1988; Kenny and
Albright, 1988; Kruglanski and Ajzen, 1983; McArthur and Baron, 1983;
Nisbett, Krantz, Jepson, and Kunda, 1983; Swann, 1984). Furthermore,
these analyses raised several fundamental issues that a comprehensive
treatment of accuracy in sodal perception and cognition may do weIl to
address.
These new developments warrant areexamination of the accuracy
problem focusing on the major themes stressed by recent accuracy anal-
yses. The present chapter pursues this objective and outlines a concep-
tual framework for the study of human accuracy, based on the lay epis-
temic approach featured in this volume.

171
172 CHAPTERS

THEMES IN RECENT ACCURACY RESEARCH

Psychologists' interest in the accuracy of people' s judgments is not

hard to fathom because accuracy seems a most valuable asset to possess.
Primarily, accuracy affords predictability that may help persons cope
with their sodal and physical environments: A spouse who can accu-
rately perceive his or her partner' s needs may avoid unpleasant conflicts
and confrontations; a teacher who can accurately diagnose a pupil's
difficulties can take the appropriate remedial steps; the accurate person-
nel officer can select the best candidate for the job; and an accurate
tennis player can place his or her shots at just the correct spot on the
court to win the point.
Given the considerable importance of accuracy, it is of interest to
ask whether persons are generally accurate or inaccurate in their every-
day judgments, what are the conditions under which they are accurate
or inaccurate, and what are the psychological factors that fadlitate or
hinder accuracy. But before those questions are examined, it seems
worthwhile to consider how accuracy may be defined.
DEFINING ACCURACY IN SOCIAL PERCEPTION AND COGNITION

In parallel to major philosophical conceptions of truth (Popper,

1959), recent sodal psychologicalliterature contains three separate no-
tions of judgmental accuracy. Perhaps the most prevalent definition is
that of a correspondence between a judgment and a criterion (Funder,
1987; Hastie and Rasinsky, 1988; Kenny and Albright, 1988). This defini-
tion closely resembles the philosophical "correspondence theory" of
truth that compares a "perception" to a "reality."
Also common is the definition of accuracy as consensus or the inter-
personal agreement between judges (Funder, 1987; Hastie and
Rasinsky, 1988). Consensus represents consistency within an interper-
sonal array of judgments; hence, it resembles the philosophical
"coherence" theory of truth as the internal consistency of beliefs. Fi-
nally, recent analyses of sodal perception (McArthur and Baron, 1983;
Swann, 1984) defined accuracy in terms of judgments' adaptive value.
Such a view of accuracy as subjective utility is explidtly indebted
(Swann, 1984, p. 461) to James's (1907, 1909) "pragmatic theory" of
truth, whereby the veridicality of an idea resides in its apparent ability to
work to a person's benefit. These three major views of judgmental ac-
curacy will now be examined in some detail.
Accuracy as Correspondence between Judgment and Criterion
A major problem for the correspondence view of accuracy is dedd-
ing on the criterion for accurate judgments. In the realm of social percep-
ISSUE OF ACCURACY 173

tion, this may not be easy. Hastie and Rasinski (1988) tellingly noted
why:
When the object of judgment is as intangible as a personality trait, emotional
state, ability, attitude, or intention, or as fleeting as a behavior, the establish-
ment of the researcher's criterion itself [mustj involve a high degree of sub-
jectivity and intersubjectivity. (p. 4)

This means that a subject's "inaccurate judgment" boils down to "a

simple competition between the subject' s judgment and the researcher' s
judgment" (p. 22).
Does this imply pitting one "privileged" group of judges (the re-
searchers) against another "defenseless" group (the subjects)? In a sense
yes, but the preference for the researchers' judgment is not arbitrary. It
is supported by well-reasoned argument and cogent evidence. Such
justification can be "a lot of work and shares much with the common-
sense manner in which everyday beliefs are justified with reference to
subjective degree of support and intersubjective consensus" (Hastie and
Rasinski, 1987, p. 4). On those grounds, researchers often have more
evidence and better evidence than subjects. They use sophisticated mea-
surement methods, based on well-validated psychological theory, and
can muster more social support for the validity of their judgments than
the subjects. Consequently,
in most research on sodal judgment even the subjects agree, when the basis
for the researcher's judgment is explained, that the researcher has more
reasons and better reasons for his or her judgment. (Hastie and Rasinski,
1987, p. 22)

In short, criteria for accurate judgments are not invariably self-evi-

dent. Often, they need to be justified by complex argument and/or
indirect evidence. Similarly, the adequacy of any given criterion (e.g.,
supported by a current scientific theory) is constantly open to criticism
(as is the theory, cf. Popper, 1959). In this vein, Einhorn and Hogarth
(1981) argued that the statistically "normative" models of inference may
not invariably represent the criteria for accurate judgments. For in-
stance, the finding that people are insufficiently regressive in their pre-
dictions (cf. Tversky and Kahneman, 1974) and, hence, that they stray
from values derived from the normative model assumes that the model
itself is applicable in a given situation. This may or may not be the case:
If a fluctuation in outcomes represents random variability around a sta-
ble parameter, the regressive model may indeed apply. However, fluc-
tuation may represent systematic shifts rather than random variability.
For instance,
if you think that Chrysler' s . . . los ses are being genera ted by a stable pro-
cess, you should predict that profits will regress up to their mean level.
However, if you take the ... los ses as indicating a deteriorating quality of
174 CHAPTER8

management and worsening market conditions, you should be predicting

even more extreme losses. (Einhorn and Hogarth, 1981, p. 56)

The notion that standards for accurate judgments are themselves

judgments, contingent on argument and evidence, has further implica-
tions. It suggests that the type of psychological factors that affect indi-
viduals' readiness to accept a given set of arguments and/or a given
body of evidence may similarly influence the standard setters' deter-
mination of the criteria. For example, to the extent that notions of "ran-
dom fluctuation" were more mentally accessible to the standard setter
(Higgins and Bargh, 1987; Higgins and King, 1981) than notions of "sys-
tematic shifts," they may playa correspondingly greater role in defining
the criterion (e.g., in terms of a statistical model). Furthermore, if be-
cause of motivational reasons a given argument (e.g., concerning ran-
dom variability) was aversive to the judge (e.g., because of the disquiet-
ing unpredictability it connoted) the referent criterion would be less
readily accepted. Similarly, arguments propitious from the judge's per-
spective may be more readily accepted.

Accuracy as Interpersonal Consensus

Hastie and Rasinsky (1988) as well as Funder (1987) examined inter-
personal consensus as a standard for accuracy. Whereas consensus does
not positively establish accuracy (after aU, collective faUacies are always
possible), a lack of consensus indirectly implies someone's inaccuracy. In
other words, consensus may constitute a necessary but not a sufficient
condition for accuracy.
The consensus definition of accuracy does not actually compete
with the correspondence definition. A lack of consensus only indicates
that in a set of judgments some unidentified judgment has been inaccu-
rate. It does not tell us whether a specific judgment was accurate, or
how inaccurate it was. Answers to the latter questions may be deter-
mined by applying the correspondence criterion. On the other hand, as
noted earlier, an accuracy criterion may be thought of as the standard
setter's judgment of what is veridical. Thus, in asense, the correspon-
dence standard of accuracy is a special case of the interpersonal consen-
sus standard, notably a consensus between the subject' sand the stan-
dard setter's judgments. Admittedly, the consensus definition does not
imply a given judgment's superiority to the others whereas the corre-
spondence definition conveys a standard setter' s superiority. The above
considerations imply an "asymmetrical consensus" definition of ac-
curacy. A rather different interpretation is implied by the pragmatic
definition of accuracy that is considered next.
ISSUE OF ACCURACY 175

Accuracy as Pragmatic Usefulness

Recently, Swann (1984) and McArthur and Baron (1983) proposed
"pragmatic usefulness" as a standard of accuracy. Interestingly, the
pragmatist standard is open to different interpretations with corre-
spondingly varied implications for accuracy assessment. Let us examine
the issues involved.
According to the pragmatist doctrine, a judgment is deemed accu-
rate if it is useful. But what exactly is "useful"? It seems reasonable to
define as useful any action or belief that facilitates goal attainment, that
is, that brings about a desired reward. Such definition allows two sepa-
rate senses of usefulness; a subjective sense and an objective sense. In the
subjective sense areward need not be an external object. Instead, it
could be an internal experience. For example, one might choose to be-
lieve in the existence of God because this put one's mind at rest, and in
that way it was useful. Thus, usefulness could be highly subjective and
reflect a motivational bias toward beliefs whose contents are comforting,
pleasing, or otherwise desirable.
More broadly, the "subjective usefulness" sense of pragmatism de-
notes the notion that needs for specific c10sure (Kruglanski, 1988) may
bias our judgments and dispose us to believe statements compatible
with our wishes and desires. This refers more to subjective validity or
judgmental confidence rather than to accuracy as typically defined. Spe-
cifically, the subjective usefulness interpretation alludes to a motivation-
al influence on judgment in general, whereas accuracy more appropri-
ately refers to a particular type of judgment, notably that of a relation
between a perception and a criterion. This topic will be discussed more
fully later.
Beyond its subjective sense, usefulness could be defined objec-
tively, for example, in terms of successful task accomplishment by an
actor. For example, if a student believed that effort mediates academic
success and exerted effort and attained success, the accuracy of the
original belief would be corroborated. It is such objective sense of useful-
ness that social psychologists generally mean by the pragmatic standard
of accuracy. For instance, Baron (1988) noted that the organism's percep-
tion of a crack's width is accurate if it can crawl through it. Similarly,
Swann (1984) suggested that a perception is pragmatically accurate if the
target upholds the perceiver's expectations.
Strictly speaking, however, what defines accuracy in the foregoing
examples is correspondence between a judgment and a criterion rather
than usefulness per se. The fact that an organism succeeds in crawling
through the crack is criterial evidence that corroborates the perception of
the crack's sufficient width. Moreover, the judgment that the crack was
176 CHAPTER8

wide enough is not restricted to the crawling organism as such but

presumably would be reached by any observer exposed to the same
information, irrespective of whether such judgment was useful or useless
to the observer. Similarly, the target's behavior in accordance with the
perceiver's expectations establishes the expectations' accuracy indepen-
dently of their utility to actors or uninvolved observers.
In sum, usefulness can be variously defined in terms of a (subjective)
motivational bias in judgment or in terms of a correspondence between
a judgment and (objective) goal accomplishment. More generally, the
present discussion suggests the utility of distinguishing between two
categories of factors affecting accuracy assessments: those unique to the
accuracy construct as such and those of general judgmental relevance.
Uniquely, accuracy can be construed as a compound, or metajudgment
that compares a given simple judgment with a criterion. This definition
of accuracy as a unique construct refers to the correspondence/
consensus (or assymetrical consensus) conception referred to earlier.
General judgmental factors are assumed to affect each component
in the compound determination of accuracy; namely, (1) the simple
judgment or perception, (2) the criterion, and (3) the correspondence
between them. Such factors may include biasing motivational variables
that are implicit in the subjective usefulness notion of accuracy, cogni-
tive variables such as availability or accessibility (cf. Higgins et al., 1982),
and informational factors, such as evidence for a given judgment. This
analysis is developed further later.

ACCURACY VERSUS ACCURACIES

Besides attempts to define accuracy broadly (e.g., as correspon-

dence, consensus, or utiIity), some authors have proposed taxonomies
of accuracies. Widely referred to is the taxonomy advanced by Cronbach
and his colleagues (Cronbach, 1955, 1958; Gage and Cronbach, 1955;
Gage et al., 1956). This taxonomy refers to the case in which each judge
rates a set of targets on a set of traits. For each judgment there is a
criterion score and accuracy is defined as a correspondence, indexed
inversely by a discrepancy, between the judgment and the criterion.
Cronbach criticized the use of a single discrepancy score as a measure of
accuracy. Instead, he recommended to partition the score into "compo-
nent accuracies" of which only some were deemed of interest.
Specifically, both the judgment and the criterion scores can be sub-
divided into the following component parts: (a) constant, (b) trait, (c)
target, and (d) uniqueness (Kenny and Albright, 1988). Constant is the
tendency for the judge to rate all targets on all the traits in the same
direction. This ordinarily reflects the judge's general positivity or nega-
ISSUE OF ACCURACY 177

tivity. It representsa very general response set because it affects the

judge's rating of every target on every trait. On the criterion side, con-
stant refers to all targets' average standing across all traits. "Elevation
accuracy," as used by Cronbach denotes the correspondence between
the constant of the judgment with the constant of the criterion. Accord-
ing to Cronbach, elevation accuracy does not reflect true accuracy be-
cause it involves a match between the subjects' response set and the
criterion. Hence, it should be of relatively little interest to psychologists.
A trait effect represents the judge' s tendency to view a particular
trait as high or low across targets relative to other rated traits. On the
criterion side, trait effect refers to the tendency of all targets to score
high or low on a given trait relative to other traits. "Stereotype accuracy"
concerns the degree of correspondence between the trait effects of the
judgment and the trait effects of the criterion. In Cronbach's scheme,
stereotype accuracy, too, is "artifactual" on grounds of subjects' bias or
response set, notably their preexisting positivity or negativity toward
some trait (relative to others).
Target effect represents a judge's tendency to view a specific target
as high or low across the set of traits relative to other targets. On the
criterion side, a target effect refers to an actual tendency of some targets
to score high or low across traits relative to other targets. "Differential
elevation" reflects the degree of correspondence between the target
effects on the judgment and criterion side. According to Cronbach, this
takes into account the individual target and, hence, it constitutes "gen-
uine" accuracy.
Finally, uniqueness effect represents the judge's view of the target on
a particular trait after constant, traH, and target effects were removed.
This reflects how the target is uniquely evaluated on a particular trait by
the judge. On the criterion side, the uniqueness effect similarly refers to
the target's standing on a particular trait relative to other targets. "Dif-
ferential accuracy" concerns the correspondence between the unique-
ness components on the judgment and criterion sides. This, too, is
considered genuine accuracy as it reflects the judge' s sensitivity to the
target-trait combination.
Kenny and Albright (1988) proposed a taxonomy of accuracies
somewhat akin to those of Cronbach. The context they address is one in
which subjects both form impressions of others and predict others' im-
pressions of them. Both judgment (prediction) and criterion (impres-
sion) scores are partitioned into four components: constant, actor effect,
target effect, and relationship effect. As in Cronbach's scheme, these
four components give rise to four types of accuracy: (a) elevation accuracy
concerns the ability of judges in general to know the criterion scores of
targets in general; (b) response-set accuracy concerns the ability of judges
178 CHAPTER8

in general to predict a given target's criterion score; (c) individual ac-

curacy concerns a given target' s ability to predict others' view of him or
her; and (d) dyadic accuracy pertains to a given target's ability to predict a
specific judge's view of the target.
The partitioning approach to global accuracy scores seems tied to
the notion that a global score "confounds" several separate sources of
accuracy, only some of which are of theoretical interest. Thus, constant
or trait components of the global accuracy score are of little interest in
Cronbach's scheme, whereas constant and actor effects are of little inter-
est in Kenny and Albright's scheme.
Overall, the partitioning approach to accuracy scores raises two
issues of interest: (1) Whether the taxonomies of accuracies proposed so
far are exhaustive (or nearly so), or whether it might be possible to
construct numerous alternate taxonomies? (2) How general (or specific)
is any ordering of accuracies by their interest value? With regard to the
first issue, a comparison between the partitions of Cronbach and Kenny
and Albright is instructive. The two taxonomies, though related, are not
identical. A careful examination reveals that "individual accuracy" in the
Kenny scheme corresponds somewhat to "differential elevation" in the
Cronbach scheme and "dyadic accuracy" in Kenny corresponds some-
what to "differential accuracy" in Cronbach's. However, the similarity is
only approximate because Kenny and Albright address the accuracy for
a trait (an impression) across a set of judges and targets, whereas Cron-
bach examines the accuracy of a judge across a set of targets and traits.
Thus, Kenny and Albright's scheme affords an assessment of a target's
accuracy (individual or dyadic), whereas Cronbach's scheme is not
meant to assess targets' accuracy.
In essence, a given discrepancy score may be partitioned in various
ways, based on the different conceptual confounds one may identify
within such a score (Le., different potential sources of the overall dis-
crepancy). In turn, this may vary with the content domain of the particu-
lar discrepancy and with the types of factors assumed to affect judg-
ments in that domain. For example, an additional "confound" untapped
by either Cronbach or Kenny and Albright could be "situation," if one
surmised that the global discrepancy scores could be affected by the
judges' situation-specific biases. To estimate such accuracy one needs to
include a situations dimension and compare the judge's tendency to
yield differential or uniform judgments across situations. Indeed,
Swann (1984) implied just such a classification in his distinction between
"circumscribed" (or situation-specific) versus global (hence, transsitua-
tional) accuracy. Further possible confounds could be "occasion" (refer-
ring to time of judgment), "judge's state" (e.g., excited versus calm),
and target's state.
ISSUE OF ACCURACY 179

Furthermore, an accuracy of interest to one researcher may not be

so to another researcher. For instance, the tendency of judges to have
generally accurate perception of targets (Le., "elevation accuracy" of low
interest value in Cronbach's scheme) may reflect judges' veridical theo-
ries of people in general that some researchers may keenly wish to
demonstrate and explain. Similarly, human perceivers may have accu-
rate perceptions of the relative magnitudes of given traits in a popula-
tion, giving rise to "stereotype accuracy," and a student of stereotypes
may be interested in understanding the bases for such accuracy.
Depending on the investigator's purpose, it is possible to classify
accurades in numerous ways. Furthermore, an ordering of accurades by
their interest value also seems relative to the investigator's perspective.
In Cronbach's scheme, a particular interest revolved around "stimulus-
driven" accuracy. That is why differential elevation and differential ac-
curacy are considered the only true or genuine types of accuracy. In both
cases, subject's judgments are responsive to the individual target, and
thus to the external stimulus being presented. By contrast, "theory-
driven" accurades (notably, elevation accuracy and stereotype accuracy)
are deemphasized in Cronbach's framework and considered inauthen-
thic or spurious. However, recent sodal cognitive interest in people's
personal theories, scripts, schemata, or heuristics (Fiske and Taylor,
1984) may render theory-driven types of accuracy of considerable in-
terest.

Relative Attainability 01 Different Accuracies

Besides an ordering by degree of putative interest, types of accuracy
were occasionally compared on their presumed attainability. In particu-
lar, Swann (1984) has recently argued that "circumscribed accuracy" is
often higher than "global accuracy." As noted earlier, global accuracy
refers to transsituationally generalizable beliefs, whereas circumscribed
accuracy refers to situationally delimited ones. According to Swann,
"professional person perceivers," such as clinidans, for example, are
typically interested in global accuracy (e.g., in patients' personality
traits), whereas everyday perceivers are often interested in situationally
circumscribed accuracy (e.g., how is one's boss likely to behave in the
office).
Circumscribed accuracy is considered easier to attain than global
accuracy because
forms of circumscribed accuracy usually require only that perceivers predict
target behaviors in a lirnited range of sodal situations ... in which per-
ceivers themselves may often be present .... This not only reduces the
number of potential influences on target behavior that perceivers must con-
180 CHAPTER8

sider, it also raises the probability that perceivers will be familiar with these
influences and with how they are apt to affect the behavior of targets.
(Swann, 1984, p. 462)

From the lay epistemic perspective the number of potential influ-

ences that perceivers may consider is not necessarily constrained by the
spedfidty or generality of a situation. Such potential influences are per-
sons' hypotheses about the situation. Their number may vary in accor-
dance with psychological factors, such as accessibility and type of infor-
mation, and/or the individual's motivational orientation, that affect
hypotheses generation in general; it should not be generally affected by
the object of the hypotheses (e.g., its content or degree of spedfidty).
Thus, a person could entertain numerous interpretative hypotheses
about a spedfic situation and only a few hypotheses about a general
situation depending on (1) the number of relevant constructs momen-
tarily accessibZe to this individual; (2) his or her stable pool of availabZe
constructs on the topic (cf. Higgins et aZ., 1982); (3) his or her (epistemic)
motivation to generate or search for further relevant notions (cf. Mayse-
less and Kruglanski, 1987); and (4) his or her creativity or constructive
ability. In sum, the idea that some stimulus situations (e.g., drcum-
scribed ones) generally invite fewer interpretative notions than other
situations, which afford greater likelihood of accuracy, may be debated
from the present theoretical perspective.
The argument that accuracy increases with perceivers' familiarity
with (Le., their amount of information about) the situation also may
need to be qualified. Information may not invariably improve the ac-
curacy of our judgments and occasionally may lead us astray; when this
occurs, Zess information may actually mean a more accurate judgment
than more information. According to the lay epistemic viewpoint, infor-
mation, too, is conceptual or constructed, rather than objectively given.
If so, it could be potentially invalid, with reference to some criterion,
and thus misleading: A person who starts out with a "correct" hunch
and subsequently modifies it on the basis of incorrect information may
end up in error. Thus, openmindedness to new information or the avail-
ability of ample information (familiarity) need not improve the accuracy
of one's judgments and may occasionally be detrimental to accuracy.
Of unique sodal psychological interest is Swann's (1984) argument
that drcumscribed accuracy is often higher (or easier to attain) than
global accuracy because of processes of "identity negotiation" and pro-
cesses in sodal perception. Two separate processes are distinguished,
"behavioral confirrnation" and "self-verification." Behavioral confir-
mation refers to targets' tendency to behave in ways that confirm the
expectandes of perceivers (Snyder, Tanke, and Berscheid, 1977). Self-
ISSUE OF ACCURACY 181

verification refers to targets' tendency to behave in ways that confirm

their self-concepts. According to Swann,
if target self-conceptions and perceiver expectancies differ, both global and
circumscribed forms of accuracy are likely to be higher if self-verification
occurs, because in such circumstances perceivers are induced to bring their
actions and beliefs into harmony with target self-conceptions [hence] they
will be equipped to predict how targets will behave in the future because
targets will theoretically continue to behave in accordance with their self-
conceptions. (1984, p. 466)

By contrast, processes of behavioral confirmation could contribute to

circumscribed accuracy while detracting from global accuracy as "targets
may behave quite differently once they escape the constraining influ-
ence of the perceiver" (p. 466)
Although Swann' s conclusions may often hold, behavioral confir-
rnation processes need not invariably detract from global accuracy. First,
the target could occasionally use his or her own behavior as a cue to his
or her properties (attitudes, traits) and/or come to internalize the per-
ceiver's expectandes via dissonance reduction or self-perception pro-
cesses (Bem, 1972). If that occurred, a perceiver who expected the target
to behave similarly across situations could be right. Second, the per-
ceiver could take into account the situational constraints, including his
or her own influence on the target' s actions and not expect such be-
haviors to generalize to alternative contexts. Such perceiver sophistica-
tion might help hirn or her avoid pitfalls to global accuracy.

Common and Distinguishing Features 01 Accuracy Types

The subtypes of accuracy discussed above all refer to a correspon-
dence between some judgment and a criterion. This basic notion is
explidt in Cronbach's and in Kenny and Albright's analyses and is im-
plicit in Swann's (1984) analysis, in which "drcumscribed accuracy" de-
notes a within-situation behavioral criterion and "global accuracy," a
between-situation criterion. Thus, the correspondence definition of ac-
curacy seems common to the various types of accuracy distinguished in
the sodal perception literature.
As noted earlier, the differences between accuracy types refer to
possible dimensions on which judgments may vary. To reiterate,
Swann's (1984) drcumscribed and global distinctions refer to the spedf-
icity (versus generality) dimension; Cronbach's elevation accuracy refers
to judgments of targets across traits, whereas stereotype accuracy refers
to judgments of traits across targets. It should be possible to classify
judgments on further dimensions and to generate a corresponding num-
ber of further accuracy types.
182 CHAPTER8

The interest value of a particular form of accuracy seems contingent

on the researcher's perspective. Furthermore, there are reasons to doubt
that some types of accuracy are generally easier to attain than other
types. The potential multiplicity of accuracy types as weIl as the appar-
ent difficulty of ordering them systematically (e.g., in terms of a general
interest value or attainability) suggest that an elucidation of the accuracy
issue in social perception may have to center on what the various types
of accuracy share in common. Taxonomies of accuracy may be very
useful for specific research purposes; however, they are unlikely to yield
us a general theory of judgmental accuracy.

Are People Generally Accurate or Inaccurate in Their Social Judgments?

Several recent papers have addressed the question whether lay per-
sons' judgments generally are accurate or inaccurate. Research on biases
and errors (cf. Kahneman, Slovic, and Tversky, 1982; Nisbett and Ross,
1980) underscored the high incidence of judgmental mistakes perpe-
trated by the use of "sub-optimal" heuristics. In contrast and, possibly,
in reaction to this unflattering portrayal of the common person, several
recent analyses argued that human judgments generally are accurate.
Thus, McArthur and Baron's (1983) ecological perspective emphasized
the essential accuracy of perception-based knowledge. Although errors
can occur, they are variously explained by a lack of sufficient motivation,
insufficient information in the extemal stimulus array, or insufficient
attention paid to what the stimulus does in fact afford.
Besides "errors of ommission" there may exist "errors of commis-
sion," in which the object's affordances are attended to but are mistaken.
This may typically occur in artificial environments and is often indicative
of inappropriate overgeneralizations of highly adaptive attunements.
For instance, a shark might react to an electrical field inappropriately as
if it were a prey. However, in natural environments, electricity would be
correlated with food; hence, responding to it would hardly count as an
error.
The notion that persons are more accurate in "natural" versus "ar-
tificial" settings is seconded in papers by Swann (1984) and Funder
(1987). According to Swann "the errors that people make in the labora-
tory ... occur only rarely in everyday contexts" (1984, p. 460). A social
perceiver who is asked, in a laboratory context, to make global trait
inferences about a target may base his or her impressions predominantly
on current evidence and fall to take into account possible transsituation-
al variability in behavior. In a sense, such failure of global accuracy is a
special case of lacking sufficient information (in this case, between-
contexts information) discussed by McArthur and Baron (1983). Presum-
ISSUE OF ACCURACY 183

ably, in natural contexts, perceivers have more relevant information at

their disposal than in artifidal contexts.
Furthermore, in everyday settings, persons may exerdse considera-
bly greater control over their sodal reality than in laboratory settings
(Swann, 1984). In asense, they have a degree of control over the criteri-
on (e.g., the target's behavior) as weIl as over the judgment (the predic-
tion of the behavior). Perceivers' influence over targets based on the
targets' motivation to agree with perceivers may often enhance the ac-
curacy of sodal perception.
The desire for shared reality is unlikely to constitute the sole, natu-
rally occurring motivation relevant to accuracy. The notion of "behavior-
al confirmation" of Snyder et al. (1977) implies that targets occasionally
may be motivated to uphold the perceivers' expectandes contrary to
what they themselves believe. For instance, a salesperson may agree
with the customer' s expressed preference irrespective of his or her own
tastes or inclinations. In yet other instances, an individual may be moti-
vated to deceive his or her interaction partners and promote, therefore,
disparate rather than shared realities. Spies, undercover agents, or con-
fidence tricksters may work hard to disguise their true identities. If they
succeed, perceivers may end up with grossly inaccurate predictions.
On occasion, interaction partners may have motivations that "pulI"
their respective beliefs in opposite directions. For instance, antagonists
in a bitter feud may be motivated to hold negative opinions of each
other: Such unfavorable views might conveniently legitimize otherwise
unacceptable hostilities against one's riyal (Bar-Tal, 1985). For the most
part, the negative view held by aperson' santagonist is unlikely to be
shared by the relevant target. In turn, an absence of shared reality may
result in predictive inaccurades. For instance, if B's antagonist believed
hirn to be cruel and callous, whereas B believed instead that he is warm
and kind, the antagonist may err in predicting B's behavior toward a
neutral C.
This discussion suggests that several motivational factors may de-
termine whether interacting persons subscribe to a commonly shared
reality and thus whether they are accurate in predicting each other' s
behavior. The diversity of possible motivations highlights the difficulty
of estimating people's general accuracy in naturally occurring situations.
Rather than guessing whether persons generally are accurate or inaccu-
rate, it might be useful to investigate the motivational, cognitive, and
informational conditions under which they mayor may not be.
Occasionally, it may be difficult to determine just what constitutes a
natural environment for a perceiver. This problem is particularly acute
with human perceivers in today's rapidly changing world that may ren-
der change itself (from old to new environments) a "natural" state of
184 CHAPTER8

affairs (Einhorn and Hogarth, 1981; Slovic, Fischoff, and Lichtenstein,

1977). An absence of stringent criteria for what is natural poses a danger
of conceptual circularity. In particular, one may be able to argue after the
fact that a situation was artificial or natural, given that an erroneous or
an accurate judgment has occurred. For instance, succumbing to a car
salesman' stricks may be excused on grounds of the "artificial" nature of
the car lot situation, whereas being able to see accurately through such
legerdemains may be explained by the considerable "naturalness" of
such settings to members of the "car generation."
The foregoing discussion underscores the many difficulties in as-
sessing the prevalence of judgmental accuracy. In this vein, Funder
(1987) noted that the very same process that may promote errors in one
setting may mediate accurate judgments in another setting (see also
Kruglanski and Ajzen, 1983). Hence, statements that people in general
are rather accurate or rather inaccurate reduce to a difference between a
half full-glass and a half-empty one. Instead, Funder recommended the
adoption of clear-cut criteria (interpersonal consensus, behavioral pre-
diction), in order to study whether in a specific situation (say, in an
organizational setting) rather than generally, perceivers are accurate or
inaccurate.

The Accuracy-Mediating Process

As we have seen, the issue of persons' general accuracy often was
handled via an attempt to elaborate the boundary conditions for ac-
curacy (e.g., of naturalistic versus artificial conditions or circumscribed
versus global conditions). In turn, the putative boundary conditions for
accuracy typically hinted at the general process assumed to be con-
ducive to accurate judgments.

Amount of Information Considered

Several authors implied that the more information a perceiver takes
into account, the greater the likelihood of his or her being accurate. The
suggestion of McArthur and Baron (1983) that active perceivers are typ-
ically more accurate than passive ones could be interpreted in terms of
the greater amount of information active exploration may afford. Sim-
ilarly, their notion of "sins of ommision" refers to cases in which the
perceiver misses part of what is afforded because of attentional selec-
tivity or because the stimulus array is impoverished. In both cases,
errors are therefore traced to informational deficiencies.
Einhorn and Hogarth's (1981) suggestion that feedback from the
environment enhances accuracy implies that the ability to receive suffi-
ISSUE OF ACCURACY 185

cient amounts of relevant information improves aeeuracy. Finally, the

suggestion of Swann (1984, p. 462) that "cireumseribed aeeuraey" is
easier to attain than "global aeeuraey" is premised in part on the as-
sumption that, in the former ease, pereeivers have more relevant infor-
mation at their disposal.
On elose examination, however, the relation between the amount of
information and aeeuraey seems eomplex. First, the terms informational
sufficiency or insufficiency, which are often used to suggest that more
information means better aeeuraey, are cireular. Sueh terms typically fail
to specify in advanee adefinite informational quantity, and are used
after the faet; that is, after an aeeurate or inaeeurate judgment has
oeeurred.
Seeond, as several authors have stressed (cf. Campbell, 1969;
Weimer, 1979) any amount of information is eompatible with multiple
alternate hypotheses. Thus, one may continue to hold to an inaceurate
judgment despite eonsiderable amounts of information that though eon-
sistent with the "eorrect" alternative, that the individual may have failed
to entertain, are equally eonsistent with the "ineorrect" hypothesis. In
other words, eonsiderable amounts of information may be nondiagnos-
tie (Trope and Basok, 1983) with respect to the eorreet and the ineorreet
hypotheses.
Finally, as noted earlier, extensive informational seareh may oeea-
sionally undermine eorreet judgments if the information reeeived was
itself "biased" or "distorted. " For instanee, the military eommander who
believes in a high likelihood of a surprise attaek may modify his opinion
on the basis of new information. Yet he eould have been eorrect in the
first place, and the new information may have been fabricated by foreign
agents and intended deliberately to mislead hirn.
One might argue that information may improve aceuracy only if it is
the "eorreet" information. However, if we aeeept the notion that any
information (fact or empirieal observation) is, at least in part, a eoneep-
tual eonstruetion (Kuhn, 1962, 1970; Lakatos, 1970; Popper, 1959;
Weimer, 1979), we ean regard it as a judgment whose aeeuraey must be
determined in referenee to a further eriterion, pushing the issue indefi-
nitely baekward. In other words, one may not know whether one's
information is eorreet, anymore than whether the judgment based on
the information is eorreet.
All of which suggest that proeessing more and more seemingly
relevant information does not neeessarily improve one's ehanees of
reaehing a eorrect judgment: Even if eorreet, the information eould be
nondiagnostic with regard to the erroneous hypothesis and the eorreet
alternative. Oeeasionally, the information eould be ineorreet or more
eompatible in fact with the erroneous than with the correct alternative,
henee, leading the pereeiver astray.
186 CHAPTER8

Motivational Factors
Beyond informational deficiencies, inaccurate judgments have been
occasionally linked to motivational deficiencies. Thus, McArthur and
Baron (1983, p. 230) suggested that perceivers' inability to detect decep-
tion may have been due to a lack of sufficient motivation. Again, the term
sufficiency in this context could be begging the question. The issue, how-
ever, is whether motivation may be linked to accuracy in any simple
way.
Consider the distinction between degree of motivation and type of
motivation. According to the present theory, some motivations (e.g.,
high need for nonspecific or specific dosure) may lead the individual to
freeze on particular judgments. To the extent that such judgments devi-
ate from the criterion, a high degree of motivation could result in error.
Even a high degree of the motivation to avoid dosure (prompted, e.g.,
by a high fear of invalidity) need not increase judgmental accuracy: As
noted earlier, one's initial hunch might be correct; the tendency to aban-
don it on the basis of further, possibly invalid, information might in-
crease the likelihood of error. The foregoing arguments suggest that
increasing incentives for correct judgments need not reduce the inci-
dence of error. Reviews of the relevant empirical literature (e.g.,
Einhorn and Hogarth, 1981) corroborate this condusion.

Normative Models
It has been implied that judgmental accuracy may be enhanced by
following the normative models of inference (Kahnerman et al., 1982;
Nisbett and Ross, 1980). However, normative models, too, can be con-
ceived of as constructed representations of reality (Kruglanski and
Ajzen, 1983), as fallible in principle as alternative, nonnormative bases
for judgment (e.g., the various judgmental heuristics). As Einhorn and
Hogarth (1981) noted, a normative model may be inapplicable to given
circumstances (e.g., the model's assumptions may not hold). When this
occurs, the judgments derived from the model will be inaccurate (i.e.,
deviate from the criterion). Thus, use of normative models may not
safeguard accurate judgments, at least not in any absolute sense.
In sum, the accuracy literature thus far has identified several pro-
cess variables implied generally to improve accuracy. However, a dose
examination of the issues involved indicates several difficulties. It ap-
pears that neither high degrees of motivation, large amounts of informa-
tion, nor the use of various "normative" models may reliably improve
accuracy. As of now, no compelling analysis exists concerning the gen-
eral process whereby accuracy is obtained. Indeed, recent proposals in
ISSUE OF ACCURACY 187

the accuracy domain (e.g., Funder, 1987; Kenny and Albright, 1988)
suggest methodologies for the assessment of specific accuracies rather
than advance a general theory of perceptual accuracy.
Our discussion thus far warrants the following main conclusions:
The most prevalent definition of accuracy has been that of correspon-
dence between judgment and criterion. Assuming that the criterion,
too, represents someone's judgment, the correspondence definition
shades into the consensus definition. The various forms of accuracy,
identified in the literature, typically share the correspondence notion
and differ with respect to particular judgmental aspects of interest (e.g.,
the judgmental object at issue or judgmental specificity versus gener-
ality). As judgmental aspects are potentially quite numerous and varied,
so are possible forms of accuracy. Finally, no compelling answers seem
available to the question of whether persons are generally accurate or
inaccurate, nor are we presently capable of identifying the boundary
conditions for accurate judgments of specifying the general process con-
ducive to accuracy.
The foregoing conclusions serve as a basis for the following re-
analysis of the accuracy problem in social perception and cognition. A
general framework for conceptualizing accuracy is outlined and applied
to two separate paradigms for possible accuracy research. One paradigm
addresses the study of accuracy from arealist perspective, and the sec-
ond, from a phenomenal perspective. It will be suggested that both sets of
accuracy issues may be discussed in terms of the process whereby all
judgments are reached. Thus, although historically the study of the
judgmental process supplanted that of accuracy (Jones, 1985; Funder,
1987), according to the present analysis the two perspectives may be
fruitfully combined.

A PROCESSED-BASED ANALYSIS OF SITUATIONAL ACCURACY

THE ACCURACY FRAMEWORK

Consider the definition of accuracy as correspondence between a

judgment and a criterion, assumed also to constitute someone's (e.g.,
the researcher's) judgment. In the discussion that follows, the term
standard setter is used to refer to the person whose judgment is con-
sidered the criterion. As noted earlier, the foregoing conception of ac-
curacy describes a compound or meta-judgment that consists of (1) the
target judgment, (2) the judgment of the criterion, and (3) the judgment
of correspondence between the judgment and the criterion.
When the standard setter is someone other than the subject, one
188 CHAPTER8

may investigate accuracy from a realist perspective; that is, measure or

prediet the degree to whieh a subject' s judgment coincides with the
external criterion. When the standard setter also happens to be the
subject, one may investigate phenomenal accuracy, that is, measure or
predict the degree to which a subject regards two of his or her own
judgments, the target judgment and the criterion, as correspondent. In
other words, depending on whether standard setting is external or inter-
nal, the study of accuracy may be approached from two different
perspectives.

THE REALIST PARADIGM

The realistic study of accuracy stresses the psychological factors that

influence subjects to render judgments whose content coincides with the
criterion. In such a framework, the standard setter' s own judgmental
process is neglected. The only focus is on the subject's judgment and on
its degree of correspondence to the criterion or the externally defined
reality. Most social cognitive research on error and bias falls into this
category; for example, the study of cognitive heuristics (for reviews, see
Kahneman et al., 1982) in which lay judgments are compared with a
priori assumed normative criteria.

Accuracy by Guessing
Accurate judgments may be rendered for a variety of reasons. First,
subjects may guess at the correct answers, thus attaining accuracy by
happenstance. Guessing may appear of little practical value because of its
unpredietable nature; a correct guess may not be expected to recur on
subsequent occasions. However, it is still possible to ask whether guess-
ing may not occasionally represent individuals' lack of awareness of the
systematic factors behind their responses (Nisbett and Wilson, 1977; Wil-
son, 1985). Recent interest in automatie processing (Bowers, 1987) may
open up new avenues for research on guessing.
Although previous studies of unconscious influences on judgments
have emphasized their error-inducing potential (Bowers, 1987; Nisbett
and Wilson, 1977), unconscious effects may similarly be involved in the
production of accurate responses (Kruglanski, 1987). As noted earlier,
the accurate or criterial response is defined in terms of its content that in
some situations coincides with the standard setter' s judgment. Current
cognitive theory (e.g., Schneider and Shiffrin, 1977) implies that auto-
matic or unconscious processing is not restricted to specific judgmental
contents. Hence, unconscious influences may occur with judgments
ISSUE OF ACCURACY 189

whose contents are situationaHy defined as accurate as weH as inac-

curate.

Reasoned Accuracy
Whereas guessing typicaHy refers to judgmental responses ren-
dered without apparent reasons, the issue of accuracy extends to rea-
soned judgments as weH. Because accuracy is defined in terms of the
judgment' s content, the issue is how that particular content is confi-
dently derived. According to the present analysis, judgments are typ-
icaHy inferred from relevant evidence. This requires (1) prior formation
of a connective schema, that is, a "rule" that links a given category of
evidence and a given judgment, (2) momentary accessibility of the rule
as weH as the evidence, and (3) appropriate motivational conditions
whereby the evidence is accepted or rejected and/or alternative types of
evidence are sought.

Rule Learning
Prior formation of connective schemata may involve learning partic-
ular rules in various ways. For instance, some people may have acquired
various judgmental heuristics (Kahneman, et al., 1982) that imply judg-
ments at variance with normative statistical predictions. To the extent
that the latter are assumed to represent the accuracy criterion, use of
heuristics may be said to pro pagate errors (Nisbett and Ross, 1980).
Recent work by Nisbett and his colleagues (Nisbett et al., 1983; Fong,
Krantz, and Nisbett, 1986; Jepson, Krantz, and Nisbett, 1983; Nisbett,
Cheng, Fong, and Lehman, 1987) suggests that the teaching of statistical
rules can increase the likelihood of statistical reasoning and, hence, of the
rendition of statisticaHy accurate judgments. For instance, Fong et al.,
(1986) found that subjects' likelihood of giving statistical answers to
simple problems increased with the amount of their previous statistical
education. Similarly, subjects' tendency to give statistical answers was
higher at the end of a course in statistics than at the beginning.
The foregoing discussion implies that the connective schemata rele-
vant to judgments in a domain may vary between individuals. We have
already seen that some individuals may rely on heuristics, whereas oth-
ers, on statistical rules. Somewhat similarly, some individuals may uti-
lize negative self-schemata in reference to given classes of judgment
whereas others may use more positive schemata in those domains. Spe-
cifically, Beck's theory of depression (Beck, Rush, Shaw, and Emery,
1979) holds that negative seH-schemata may influence the judgments of
190 CHAPTERB

depressed as compared to nondepressed individuals. In this connection,

Dykman, Abramson, Alloy, and Hartlage (1987) found that depressives
(versus nondepressives) erred more in making inappropriately negative
self-judgments in which their negative schemas were relevant to the
experimental task. Both depressives and nondepressives exhibited posi-
tive errors, negative errors, and accurate responses, all depending on
the way in which their preexisting schemata related to the situational
criteria of correct responding. The Dykman et al. (1987) research fur-
nishes important evidence that "depressive realism" (cf. Abramson and
Alloy, 1981; Alloy and Abramson, 1988) is situationally specific rather
than general. It may depend on whether the depressed or nondepressed
subjects' judgmental process yields outcomes that correspond with a
situational criterion of accuracy.
In some connective schemata, judments may be conditionally
linked with informational sources rather than contents. For example,
a statement linked with a high (versus Iow) authority source is more
likely to be accepted by an individual and be confidently adopted as his
or her own judgment. In attitude research (for a review see Chaiken and
Stangor, 1987), such reliance on various source heuristics has been
shown often to mediate subjects' judgments. Occasionally, the relevant
informational source may be one's own self. In such a case, one's own
epistemic authority, that is, one's self-perception as a reliable source of
information (cf. Kruglanski and laffe, 1988), may affect the degree to
which one may confidently base judgments on one's own experiences.
The foregoing line of reasoning is iIIustrated in the study by EIlis
(1984) described in Chapter 3. In that research, subjects who ascribed to
themselves high (versus low) epistemic authority (in mathematics or in
empathy) rendered more accurate judgments in those domains after
being exposed to "experimental learning" of the appropriate judgmental
rules. Furthermore, subjects for whom the external instructor con-
stituted a higher authority rendered more accurate judgments following
exposure to a "frontal presentation" of those rules than subjects for
whom the instructor represented a less authoritative source.
Finally, from the present perspective, the very same judgment may
be derived from diverse types of evidence. In other words, the connec-
tive schemata from which the same judgment is inferred may vary con-
siderably across individuals and situations. Thus, if our objective is to
increase the likelihood that an individual would render a given accurate
judgment x, we may have a choice among different connective schemata
from which x may be inferred. Our choice in such a case may be deter-
mined among others by the ease of imparting a given rule to a given
category of persons, and/or by the intended uses of the rule. For in- '
stance, it may be easier to impart to psychology students various "cook-
ISSUE OF ACCURACY 191

book" approaches to correct statistical inferences rather than derive

those answers from complex mathemathical proofs.

Accessibility
In order that it be utilized in a given judgmental situation, a rule (or
a connective representation) must not only be available in the individ-
ual' s long-term memory but also momentarily accessible (Higgins and
Bargh, 1987; Higgins and King, 1981; Higgins et al., 1982). In turn, ac-
cessibility may be determined by frequency as weIl as recency of activa-
tion (Higgins et al. , 1985; Wyer and SruIl, 1981). For example, in Luchins'
(1942) work on cognitive sets, an inappropriate judgmental rule may
have become highly accessible because of activation recency and may
have therefore led to erroneous solutions on subsequent judgmental
tasks. The same research, however, suggests that heightened rule ac-
cessibility facilitated performance accuracy on a previous task to which
the rule was appropriate.
Accessibility of relevant constructs is possibly involved in interpre-
tation of situational evidence in ways that highlight the applicability of
given inferential rules. Support for this notion comes from studies in
which specific "packaging" of the evidence increased subjects' tendency
to employ otherwise underutilized principles. Thus, in research by
Ajzen (1977), base-rate information was more likely to be utilized when
it was interpreted to have causal significance. In research by Kruglanski
et al. (1984), the appropriate use of evidence for regression~to-the-mean
effects increased where such evidence was couched in familiar everyday
examples. Similarly, research on logical reasoning (Griggs and Cox,
1982) suggests that subjects' erroneous failure to falsify hypotheses (by
the modus tollens) is much reduced when the examples are couched in
familiar terms.

Motivation
Individuals' tendency to utilize available evidence mayaIso depend
on their motivational condition. To the extent that the evidence supports
an undesirable conclusion, subjects may downgrade the value of the
evidence and base their judgments on alternate evidence instead (Lord,
Ross and Lepper, 1979; Lord, et al., 1984). Furthermore, under high
need for cognitive closure (Kruglanski, 1988; Kruglanski and Freund,
1983) an individual may be disposed to reject evidence inconsistent with
an initial judgment, whereas under high need to avoid closure he or she
may be disposed to accept inconsistent evidence. Thus, to the extent
that a person's initial hunch was accurate, high need for closure would
192 CHAPTERB

strengthen his or her tendency to render a correct judgment, whereas

high need to avoid closure would increase the likelihood of error. Con-
versely, to the extent that the initial hunch was inaccurate, high need for
closure would decrease the likelihood of a correct judgment, whereas
high need to avoid closure would increase such likelihood.

Accurate /udgments: General or Specific?

Suppose p rendered an accurate judgment in situation x. What is
the likelihood that p will also be accurate in situation y? The answer to
this question would seem to depend on two categories of factors having
to do with the situation and the person, respectively. The situational
factors concern the extent of similarity between the criteria for x and y
and the personal factor, and the degree to which p's psychological states
(i.e., the various motivational and cognitive variables affecting p's judg-
ment) are similar in the two cases. If both the situations and the psycho-
logical states are highly similar, the generalizability of an accurate re-
sponse may be high. For instance, ready (psychological) accessibility of
statistical notions may prompt p to give a correct (stochastic) answer to a
question about the behavior of a random device (situation x) as weH as
about regression to the mean (situation y). However, if either the ac-
cessibility of statistical notions declined or the problem shifted to one
about a systematically unstable process, p's accuracy might wane.
Assuming that shifts in subjects' psychological states are largely
uncorrelated with shifts in judgmental criteria, no rigorous answer
seems possible to the question whether a given individual is generally
accurate or inaccurate in his or her judgments, let alone whether people
at large are accurate or inaccurate.
In sum, persons' tendency to render accurate judgments in a situa-
tion may depend on factors that determine their tendency to render
judgments of a given content. Prominent among such factors are (1)
previously formed or learned reasoning rules or connective schemata,
(2) momentary mental accessibility of a given rule and the pertinent
evidence, and (3) motivationally determined readiness to accept or reject
given types of information. Thus far, relatively little research exists on
attempts to vary systematically, rather than merely asess, persons' ac-
curacy. Thus, the present emphasis on the judgmental bases of accuracy
phenomena may suggest fruitful new directions for further accuracy
research.

PHENOMENAL ACCURACY

Our discussion of the realist approach to accuracy centered on fac-

tors that may influence the subject' s forming a given content of judg-
ISSUE OF ACCURACY 193

ment. In turn, we·as researchers (or standard setters) may compare the
judgment with an accuracy criterion (e.g., a statistically warranted in-
ference) that we deern appropriate ör real for the situation. By contrast,
in the phenomenal approach to accuracy, the comparison of judgment
with criterion is carried out by the subject. Thus, subjects' perception of
accuracy is likely to be affected not only by (1) what they perceive to
constitute a given judgment (e.g., someone else's opinion) but also by (2)
what they perceive to be the case (subjects' criterion) and by (3) their
perception of the degree of correspondence between the two.
Perceptions of accuracy may pertain to one' s own judgments and
those of others. In the case of self-perception, error may be admitted, for
example, when an inconsistency (or a lack of correspondence) is noted
between one' s own judgments rendered on two separate occasions, and
the inconsistency is resolved by adecision whereby one judgment is
correct and the other, incorrect. In the case of another's perception, the
inconsistency could be between one's own (criterion) and the other per-
son's judgment. To the extent that one adhered to one's own criterion,
the other person' s judgment would be deemed in error.
The same variables that may affect subjects' judgments (Le., pre-
viously leamed rules of inference, considerations of accessibility, and
motivational factors discussed earlier) mayaIso affect their perception of
the criterion and of the correspondence between judgment and criteri-
on. For instance, if information given to one individual differed from
that given to another individual, this may make accessible to those indi-
viduals different rules of inference and/or prompt the formation of dis-
crepant beliefs about the case (Le., discrepant subjective criteria). These
divergent beliefs may lead each individual to assurne the other is in
error. Such process was illustrated by Back's (1951) early experiment in
which two subjects' received allegedly the same, yet actually different,
sets of pictures as a basis for creating a story. Although not specifically
tapped in Back' s research, his procedure could readily lead to mutual
perception of error by the subjects involved.
Perception of error could also arise for various motivational reasons.
In their c1assic study, Hastorf and Cantril (1954) demonstrated how
Princeton and Dartmouth fans perceived the very same game between
their respective teams in diametrically opposite ways. The Princeton
fans perceived their game with Dartmouth as brutal and provocative and
their response as measured and appropriate. The Dartmouth fans had
very different perceptions. Presumably, if ever confronted with the oth-
er group's perception, each group of fans would consider it in gross
error.
Thus far we have considered subjects' perception of the criterion
(Le., what they perceive to be the case). However, similar psychological
factors, of a cognitive or a motivational nature may affect one's percep-
194 CHAPTER8

tion of the judgment. For instance, in arecent study by Vallone, Ross,

and Lepper (1985) pro-Arab viewers perceived the content of the major
networks' coverage of the Beirut massacre (in 1982) very differently from
pro-Israeli viewers. Vallone et aZ. suggested that such divergent percep-
tions of the media position or its judgment could be one mechanism
responsible for the perception of media bias. Another mechanism impli-
cated by Vallone et aZ. were differences between pro-Arab and pro-Israeli
viewers' judgment of the truth or, in present terms, their perception of
the criterion.
Besides influendng subjects' perceptions of the criterion or the
judgment, psychological factors may underlie their perception of the
correspondence between the two. Thus, in some cases, individuals may
be motivated to perceive a correspondence between a judgment and a
criterion, whereas, in opposite cases, they may be motivated to perceive
a discrepancy. For instance, it is possible that the tendency to perceive
the media as biased (Vallone et aZ., 1985) is motivated by the desire to
sway them to one's own side on subsequent occasions. Similarly, the
perception of one' s own past errors could be motivated by the desire to
view oneself as having grown wiser or as having seen the light (Ross
and McFarland, 1988). The various informational and motivational fac-
tors that may enter into subjects' perception of the criterion, the judg-
ment, and their correspondence could be profitably explored in future
research on the perception of accuracy and error.

SUMMARY AND CONCLUSIONS

The present chapter attempted to analyze recent concepts of ac-

curacy in sodal perception and cognition. Based on this review, it has
been concluded that several major questions typically posed in the ac-
curacy domain have not been satisfactorily resolved thus far, nor do
they see m capable of resolution in the foreseeable future. To reiterate, it
has not been possible to answer meaningfully whether people are gener-
ally accurate or inaccurate or to estimate reliably the proportion of times
at which they may be. Furthermore, it has not been possible to delineate
general classes of boundary conditions for accurate and inaccurate judg-
ments. Finally, it has not been possible to eluddate the process (or
method) for reaching generally accurate judgments.
Instead of attempting to answer general questions about the ac-
curacy of sodal perception, a situationally spedfic approach has been
outlined. Such an approach adopts the correspondence definition of
accuracy and assurnes that the criterion for accuracy represents the stan-
dard setter's judgment as to the true state of affairs. It follows that the
ISSUE OF ACCURACY 195

criterion is relative to the situation and to the perceptions of the standard

setter rather than absolute in any strong sense of the term.
The present interpretation suggests that, despite historical separa-
tion, a study of the judgmental process is highly relevant to the study of
accuracy. According to this analysis, determination of accuracy con-
stitutes a metajudgment comprising three components: the simple judg-
ment, the criterion, and the correspondence between them. Each of
these components is a judgment governed by various cognitive and
motivational factors that affect the judgmental process in general.
The foregoing analysis suggests two general paradigms for accuracy-
related research. In the realist approach, the situational criterion (the
reality) is assumed to be known, and the research addresses the condi-
tions under which subjects' judgments approximate the criterion. In the
phenomenal approach, the focus is on subjects' compound perception
comprising the judgment, the criterion, and their correspondence. Both
research paradigms have been represented in recent accuracy research.
By systematically outlining their judgmental underpinnings, the present
analysis may help extend those lines of work in previously unexplored
directions.
 9
Knowing How to eure
Implications for Cognitive Therapy

The theory of lay epistemics addresses the general process of cognitive

change. In contrast, cognitive therapy deals with a specific type of
change: the substitution of functional cognitive elements for dysfunc-
tional ones. Thus, cognitive therapy may be thought of as an application
of general epistemic principles in order to effect change in specific cogni-
tive contents characterized as maladaptive for the individual.
It will be noted at the outset that the lay epistemic approach to
cognitive therapy challenges several assumptions prevalent in the field
today. In particular, this approach questions the notions that "veridi-
cal," "undistorted," or "realistic" conceptions are necessarily functional
for patients and contribute to their sense of well-being. Similarly, issue
is taken with the notions that "normals" are generally more accurate in
their judgments than "neurotics," and that veridical judgments are more
likely to be rendered if one adopts the appropriate method of inference
based on logical and empirical principles. In contrast, it will be sug-
gested that the functionality or adaptability of a belief derives from its
particular content rather than from its truth value.
This chapter has the following major objectives: (1) to offer a general
conception of the functional-dysfunctional dimension of belief contents,
(2) to show how the principles of lay epistemics can be applied fruitfully
to the modification of dysfunctional cognitions and their replacement by
functional ones, and (3) to compare and contrast the lay epistemic ap-
proach to some major schools of cognitive therapy practiced today.

THE CONTENTS OF DYSFUNCTIONAL BELIEFS:

THE FRUSTRATIVE HYPOTHESIS

According to the present proposal, suffering and displeasure result

generally from the perceived thwarting of one's goals, needs, or desires.
197
198 CHAPTER9

Cast in cognitive terminology, suffering results from the belief that a

desirable state of affairs is unlikely to be reached, or that an undesirable
state of affairs has been or is likely to be reached. Thus, belief in or
knowledge of a past or a future frustration is assumed critical to the
experience of unhappiness or demoralization (Frank, 1973). Henceforth,
I refer to this type of belief as a frustrative hypothesis.
It is important to note that the term frustration is used here in its
broadest sense to include states of deprivation and loss, for example,
occasioned by the death of a spouse or a parent as weIl as the blocking of
goal-directed actions. For example, if a person's objective was to excel in
absolutelyeverything, he or she would be likely to suffer upon failing on
a given task. Similarly, if an individual' s objective was to be weIlliked
and adored by aIl, he or she would be likely to suffer upon being re-
jected by at least some people. The foregoing are examples of beliefs or
types of knowledge pertaining to what Beck et al. (1979) have described
as states of loss leading to the depression syndrome. SpecificaIly, de-
pression may be conceptualized to evolve from past frustration or a
"loss" already experienced. In contrast, an anticipatory frustration could
lead to what Beck has termed a "danger to domain" experienced as
anxiety or fear. For instance, a person might fear situations in which the
delivery of a speech in public is required because of a belief that in those
particular situations he or she is likely to be ridiculed by members of the
audience. The anticipatory fear in this example represents the belief that
at some future time the person's objective of maintaining positive self-
esteem would be thwarted. It is possible to extend this analysis to alter-
native cases of psychological disorders (e.g., to the variety of phobias)
and to show that in all those instances the disorder refers to a sense of
pain and malaise caused by a belief that some important personal objec-
tives have been or will be frustrated or some important personal values
have been denied.
Therefore, the object of cognitive therapy is to persuade the patient
to modify his or her frustrative hypotheses. The different possible ways
of accomplishing such modifications are discussed in the next section.

MODIFYING THE FRUSTRATIVE HYPOTHESIS: ASSESSMENT SHIFTS, GOAL

SHIFTS, AND MEANS SHIFTS

The therapeutically relevant changes one could perform on the frus-

trative hypothesis can be naturally classified in accordance with the
conceptual structure of such a hypothesis. A frustrative hypothesis con-
tains the elements of a goal, a means to that goal, and a relation of
ineffectiveness connecting the two, whereby the means is an ineffective
way of getting to the goal. Accordingly, the frustrative hypothesis could
be variously modified via (1) an assessment shitt, that is, a change in the
KNOWING HOW TO eURE 199

perceived relation of ineffectiveness between the current means and

goal to one of perceived efficiency; (2) a goal shift, that is, a change in the
person's goal to a more readily attainable objective given the individu-
al's resources; and (3) a means shift, that is, a change in the person's
means of getting to the goal to a more effective one.
An assessment shift concerns the ca se in which it may be possible to
produce the required change in the frustrative hypothesis by getting the
dient to take a new look at the relevant evidence, induding data not
sufficiently attended to heretofore. In such a case, no attempt is made to
change the person' s goal or his or her particular means of reaching it.
Instead, an attempt is made to help the person perceive that, contrary to
former assessment, the current means is an effective rather than an
ineffective way of attaining the goal in question. Just as an erstwhile
ugly duckling may discover that it is actually a beautiful swan, a person
whose suffering sterns from a belief that he or she is disrespected by
others or is failing at his or her job may be introduced to evidence
suggesting the contrary to be the case.
Often, however, all evidence available to the dient as well as the
therapist may unequivocally support the dient's frustrative hypothesis.
It may appear that this person is indeed failing to earn the approval of
important others, is ignored, disliked, or ridiculed, or is poorly en-
dowed to cope with a given category of tasks. Under these circum-
stances, it would seern futile, beside being unethical, to convince the
dient of the opposite or to induce hirn or her to execute an assessment
shift. This leaves two principal avenues for inducing the individual to
abandon his or her frustrative hypothesis. They are a means shift and a
goal shift. To reiterate, a means shift refers to the substitution of a new,
efficient means to the desired goal for the old, inefficient one. A means
shift could promote the belief that the end may indeed be attained and
could thereby eliminate the suffering founded on the contrary belief.
Unlike an assessment shift in which the dient is encouraged to operate
in the cognitive mode, that is, to think about evidence and arguments
presented by the therapist, in a means shift one often needs to enter the
action mode, that is, to engage in novel behaviors presumed to yield
ultimately evidence inconsistent with the frustrative hypothesis. For
example, a person who fails to attain his or her goals because of a lack of
assertiveness may be induced to undergo a course in assertiveness train-
ing and thus improve his or her interpersonal relations; a person with
frustrations in the sexual domain may be induced to attend a course in
sexual techniques and ultimately abandon the negative beliefs about his
or her sexual inadequacies.
However, a means shift may occasionally be inappropriate for
changing a person's frustrative belief. First, regardless of the attempted
means, some goals may appear unrealistic for, or unattainable by, al-
200 CHAPTER9

most anyone. For instance, the goal of succeeding at every task, or of

being admired by every person, see m to belong in that category. Second,
some goals may seem unreachable for particular individuals who lack the
unique qualities necessary for their attainment. Many people could not
hope to become successful writers, musicians, or business persons be-
cause of a lack of talent in those areas. In cases where the goal may lie
beyond the person's reasonable resources, an efficient way of dealing
with frustration and suffering might be to induce the individual to aban-
don the goal in question and substitute for it a new, more readily achiev-
able objective. The task of facilitating a goal shift may require (1) a
criticism of the original goal as unworthy of pursuit or wholly unrealistic,
and (2) an advocacy aimed at "selling" the individual on the new goal.
To summarize, I have argued that negative affect may stern from a
person's frustrative belief, that is, a belief that a desirable state of affairs,
a goal, is unlikely to be attained or maintained. To alleviate the pain
attendant on such a belief, it seems necessary to modify it, which may be
accomplished in one of three general ways: (1) Leaving intact the goal
and the means employed toward its accomplishment, a person may
execute an assessment shift, that is, reconsider in light of new evidence
and ultimately abandon the frustrative hypothesis. (2) A person may
execute a means shift whereby a new effective means of getting to the
goal is substituted for the old ineffective means. (3) Lastly, a person may
execute a goal shift, that is, abandon the goal that is hopelessly frus-
trated and substitute for it a more readily attainable objective. Each of
these approaches is assumed to be potentially helpful in appropriately
modifying the patient's frustrative belief, and, in this sense, each is
considered to constitute a mode of cognitive therapy. Which of the three
approaches should be used in a given ca se might depend on its cost-
effectiveness or the ease with which it may produce the desired results
under the specific circumstances. Regardless of the approach finally
selected, the way in which therapy is actually implemented will depend
on the persuasion techniques available to the therapist. It is in this
connection that the principles of lay epistemics may become useful, as
the following sections attempt to illustrate.

ApPLYING LAY EPISTEMIC PRINCIPLES TO THE PRACTICE OF

COGNITIVE THERAPY

KNOWLEDGE STRUCTURES AND THEIR V ALIDITY

Through cognitive therapy, dysfunctional knowledge structures

(frustrative hypotheses) are replaced by functional ones. Frequently this
may be accomplished by criticizing the dysfunctional structures as inval-
id and by establishing the validity of the functional structures. A noted
KNOWING HOW TO eURE 201

in Chapter 2, validity arguments are based on the principle of logical

consistency. To attack the validity of a given proposition, it is necessary
to show that it is inconsistent with other beliefs held by the individual.
Similarly, to establish the validity of an alternative proposition, it is
incumbent to show that it is consistent with those beliefs. Such uses of
the consistency principle are frequent among practitioners of cognitive
therapy. For instance, therapists such as Albert Ellis or Aaron Beck
typica11y pinpoint logical contradictions in the patients' belief systems
and use this as a leverage to effect the desired cognitive change. In
almost a11 schools of cognitive therapy the dient is induced to engage in
a kind of hypothesis testing in which the evidence obtained favors the
substitute functional hypothesis over the original dysfunctional one. For
instance, success experienced at some task in the course of therapy
(Bandura, 1977) is inconsistent with a dient's notion that he or she is a
hopeless failure. Acceptance of the dient by the therapist or members of
a therapeutic group (Carl Rogers's notion of "unconditional positive
regard") is inconsistent with the dient' s belief that he or she is objection-
able to everyone.
For therapy to be successful, the cognitive structures imparted to
the client in the course of treatment should not encounter 10gica11y in-
consistent (hence, invalidating) evidence outside the therapeutic set-
ting. No matter how pleasing or functional a given belief might be for
the dient, it would hardly last if contradicted by readily discoverable
"facts." Even if the therapist succeeded in sheltered sessions in persuad-
ing the dient of his or her considerable charms or talents or of the
desirability of a given behavioral norm (e.g., of openness and frankness
in interpersonal relations), these beliefs might be quickly shattered by
inconsistent evidence in the outside world.
Accordingly, a frequent criticism leveled at the variety of group
therapies has been that the subcultures created in group sessions con-
tained norms or beliefs inconsistent with those outside the group. In
such cases possible beneficial effects achieved in the group atmosphere
may quickly dissipate when the individual is reexposed to his or her
routine surroundings and activities. The idea that the cognitive change
that the dient undergoes in the course of therapy should be compatible
(i.e., consistent) with, and constantly checked against, events and opin-
ions in the outside world has been stressed by Strong (1970). It seems
also implicit in the "open" or ambulatory treatment of psychotic disor-
ders (Raush and Raush, 1968).

EPISTEMIC AUTHORITY

In supporting the functional beliefs which the therapist may wish to

impart to the dient, he or she may make extensive use of the individu-
202 CHAPTER9

al's subjective categories of legitimate evidence. In what follows, I con-

sider one such evidential category related to the notion of epistemic au-
thority. As noted in Chapter 2, epistemic authority relates to the person's
beliefs regarding the validity of pronouncements emanating from a
given source. In the following section, the generic construct of epistemic
authority is illustrated by three exemplary subdasses: expert authority,
group authority, and self-authority.

Expert Authority
An approach based on an expert authority involves the therapist' s
invoking his or her professional expertise, scientific status, or pre-
sumptively superior intellect in order to influence the dient in the direc-
tion of a desired change (for a thorough discussion of this point, see
Strong, 1978, pp. 106-111). Reliance on the therapist's personal authori-
ty is prominent in certain major systems of cognitive treatment.
Meichenbaum (1977a), for example, commented on the forceful, au-
thoritative manner in which Albert Ellis conducts therapy. In Meichen-
baum's words, Dr. Ellis's approach
is forceful, so forceful that on one occasion I was moved to suggest rather
tongue in cheek that RET [rational emotive therapy] as conducted by Ellis
would only be successful with New Yorkers. (p. 190)

Beck (1976) explicitly recognized that psychotherapy can have the great-
est impact on the resolution of emotional problems because of the con-
siderable authority attributed to the therapist (p. 216). Such authority
also figures prominently in Raimy's (1975) "method of explanation," in
which the therapist presents an alternative account of the evidence
"which is more valid than that upon which the [patients] misconception
is based" (p. 44).
Finally, although not usually discussed in these terms, the thera-
pist's personal authority probably plays an importaht role in therapy as
practiced by the "cognitive behavior modifiers" (e.g., see Mahoney,
1985, for a review of the literature). Thus, in teaching the dient to talk to
him- or herself or to self-instruct (Meichenbaum, 1977b), the therapist
suggests the content of things to be said in specific circumstances. These
statements are dearly more than just behavioral acts of a verbal nature.
They are better regarded as hypotheses that the therapist is proposing to
the dient, the credibility of which has a great deal to do with the thera-
pist's epistemic authority in the dient's eyes. For example, in inducing a
child to be creative, the therapist might suggest that he or she utter self-
directives such as the following: "If you push yourself you can be cre-
ative" or "Think of something no one else will think of." Both state-
KNOWING HOW TO eURE 203

ments are intended as credible and valid hypotheses regarding effective

means to the end of creativity. Undoubtedly, the dient's readiness to
believe in such hypotheses has much to do with the therapist's per-
ceived authority.

Group Authority
In addition to the therapist's personal authority, the collective au-
thority of a therapeutic group can be employed to persuade a dient to
make desired changes. A respected reference group might induce the
individual to attempt various new means of attaining the desired goal
(i.e., to execute a means shift). For example, a feminist group might lend
legitimacy to a woman's initiative in striking up an acquaintance with a
male, or a gay group might lend legitimacy to public openness about an
individual's gay identity. Altematively, an individual might substitute
one set of goals for another (i.e., execute a goal shift) based on the
influence of a respected reference group. For example, after having
joined a commune or the Unification Church organization, a person
might quickly adjust his or her life objectives to the prevailing group
norms.
Indeed, the use of groups is a frequent weapon in the arsenal of
many cognitive therapists. Ellis's RET often employs groups in which
the collective authority of an individual's peers, appropriately led by the
therapist, is utilized to weed out maladaptive beliefs. Beck (e.g., 1976)
also recommends that persuasion be accomplished by interaction with
other people in certain organized settings, that is, through the collective
authority of one's peers. Finally, cognitive therapists with a more behav-
ioral bent (e.g., Meichenbaum, 1977b) frequently utilize groups in order
to facilitate the desired cognitive change. For instance, in some cases the
therapist might make a behavioral contract with the group as a whole.
The group' s acceptance of the contract as worthwhile and desirable
could serve as an important influence on the individual members.

Seif as Authority
The dient' s own perceived authority can be fruitfuHy employed to
modify dysfunctional beliefs in the desired ways. Frequently, a person's
own experience may constitute a powerful epistemic authority whereby
beliefs are induced or modified. Seeing often is believing; hence one's
own experience may often be a more effective means of communication
and persuasion than the use of expert or group authority. Reliance on
the dient's perceived authority is prominently featured in many schools
204 CHAPTER9

of cognitive therapy. Ellis, for example, makes frequent use of behavior-

al homework assignments, such as asking a member of the opposite sex
for a date. This exercise is intended to demonstrate to the dient that
such an experience, even if unsuccessful, does not have the devastating
consequences that might have been feared. In terms of the concepts
discussed here, the homework assignment technique represents a re-
liance on the dient's own perceived authority as means of persuasion:
the dient experiences the events; he or she sees them with his or her
own eyes, so to speak. This can be very convincing, providing the dient
trusts his or her own ability to make sense of the experience; that is,
perceives him- or herself as a reliable epistemie authority in the domain
at issue.
Beck's mode of cognitive therapy also often utilizes the dient's self-
ascribed authority in situations in whieh his or her hypotheses are expe-
rientiaIly tested; for instance, in situations in whieh a dient "in response
to his psychotherapist' s warmth and acceptance . . . modifies his ste-
reotyped conception of authority figures" (Beck, 1976, p. 215). FinaIlY'
the cognitive behavioral approach to psychotherapy often reHes on the
patient's self-ascribed authority as a means of effecting the desired cog-
nitive change. For example, in the technique of modeling (RosenthaI
and Bandura, 1978), the client by observing role models tests the hy-
pothesis that people like him- or herself can reach a desired end by
following a certain route. The frequent use of imagination in cognitive
behavioral therapy may be viewed as representing personal hypothesis
testing. For instance, in the technique of desensitization (see, e.g., Mur-
ray and Jacobson, 1978; Yates, 1975), the client discovers that he or she
can think of a given object or situation without necessarily getting upset.
Whatever else this particular procedure might accomplish, it certainly
refutes the hypothesis that a mere thought about the phobie object nec-
essarily brings about considerable tension and disturbance.
It should be emphasized that the use of personal experiences as a
means of inducing cognitive change assumes that the client's self-
ascribed epistemie authority in a domain is high. When it is low, change
may be more effectively induced by alternative epistemie authorities that
the dient holds in high regard. These notions received empirical cor-
roboration in the study by EIlis (1984), described in Chapter 3, in whieh
subjects high on self-ascribed epistemic authority (in mathematics or
interpersonel skills) did better than those who were low in an experiential
learning condition (on mathematieal or interpersonal topies, respec-
tively) but did equaIly weIl (in the mathematies study) or less weIl (in the
interpersonal study) than those who were low in a frontal lecture
condition.
KNOWING HOW TO eURE 205

FREEZING AND UNFREEZING MECHANISMS

(1) Cognitive Capability Effects

The replacement of dysfunctional cognitions by functional ones
through cognitive therapy requires that the former be "unfrozen" and
the latter "refrozen" in their stead. To accomplish this, one might use
some of the unfreezing mechanisms discussed earlier, under the cogni-
tive capability and epistemic motivation categories. Thus, the therapist
may impart to the dient useful ways of thinking about his or her prob-
lem that may have been otherwise unavailable or inaccessible. The dient
may be surprised to leam that his or her "glass" can be regarded as half-
full rather than half-empty, that a goal (e.g., of making it in the corpo-
rate rat race) is not as sacrosanctas he or she has previously supposed,
or that a relatively brief course of studies may effectively rid hirn or her
of what previously was considered as an unalterable personal defidency
(such as soda! awkwardness, lack of assertiveness, or shyness). In
short, therapist-engendered accessibility of adaptive notions may con-
stitute an important precondition to the dient's unfreezing of his or her
frustrative hypotheses.
Instead of, or in addition to, attempting to increase the accessibility
of functional notions, the therapist might try to decrease the accessibility
of dysfunctional ones. Occasionally, this might mean a recommendation
to avoid situations or persons in whose presence such maladaptive no-
tions are likely to be activated. For instance, the dient might be advised
to limit his or her participation in activities of an elitist dub whose
snobbish atmosphere invokes notions of one's inadequacy or personal
failure. A dient may be advised to avoid violent films if they activate
severe fears and anxieties, or shun the company of cynical friends
whose presence may activate dysfunctional notions about life' s futility
and purposelessness. As noted in Chapter 2, following prolonged in-
activity, the dysfunctional notions may become dissodated in the per-
son's mind from their evidential basis and, hence, become less securely
entrenched and more tractable or amenable to alteration.

(2) Motivational Influences: Specific Closure Needs and Persuasive Efficacy

To enhance persuasive efficacy, a therapist might take advantage of
the idea that an individual's needs and desires may exert a selective
infIuence on the person's inferences, leading him or her to engage in
wishful thinking. Two complementary therapeutic strategies derive
from this notion: (1) The therapist may aim to demonstrate that the
206 CHAPTER9

dysfunctional belief (the frustrative hypothesis) is incongruent with the

person' s specific needs and desires and/or that this belief has conse-
quences contributing to the frustration of other needs or desires. Alter-
natively, (2) the therapist may aim to convince the dient that a given
substitute belief is indeed congruent with a need or adesire and so
enhance the likelihood of it being accepted. For example, the therapist
might suspect that a dient's depressive belief that he or she is an utter
failure is prompted by this person' s need to elicit compassion and sym-
pathy from others. Under those circumstances, the therapist could at-
tempt to persuade the dient that, far from eliciting compassion, presen-
tation of oneself as a failure invites instead disdain and rejection. The
therapist could further suggest that a self-presentation as an average
person with a share of failures and successes is more conductive to
widespread sympathy.
Typicaily, cognitive therapists eschew explicit references to motiva-
tional variables. Occasionally, however, they see m to utilize such vari-
ables in the course of therapy. For example, in one case reported by Beck
(1976), the dient's preferred dosure needs were enlisted in the aid of
therapeutic persuasion. The case concerned modification of an elderly
man' s belief, that his wife was carrying on an affair with his young
physician (p. 225), by pointing out to hirn the negative consequences of
such a belief induding the possibility of his wife ultimately leaving and
ceasing to care for hirn. In this particular instance, a demonstration that
the content of a given belief may ultimately lead to the frustration of an
important need (to be cared for by one's wife) may have discouraged the
dient from holding the dysfunctional belief.

Needs for Closure and the Avoidance of Closure

A heightened need for nonspecific dosure (e.g., aroused by an
impending decision) could facilitate the adoption of any plausible con-
ception relevant to this decision. For example, a person might be more
readily persuaded by the therapist's argument extolling a training
course in social skills if this person expected to have to decide soon
whether or not to enroil in the course at issue. The need for cognitive
dosure may thus be utilized as a consolidating influence that may freeze
the individual's belief on a given hypothesis (ideally, the one proposed
by the therapist) and suppress the dient's tendency to generate riyal
alternative hypotheses.
The use of homework assignments (Shelton and Ackerman, 1976)
may capitalize on the client's dosure needs. A belief proposed by the
therapist (e.g., that the dient will not be devastated by social rejection) is
likely to be accepted by the dient more readily if he or she is about to
KNOWING HOW TO eURE 207

make a commitment to some definite action (e.g., the striking up of a

conversation with a stranger) in the immediate future. The contracts and
commitments typical of more behaviorally oriented brands of cognitive
therapy (e.g., Kanfer, 1975) make effident use of the dosure need in
order to consolidate in the client's mind the new, desired belief. The
pressure on a client to make a firm commitment or enter a contract
regarding a course of action renders it likely that he or she will quickly
formulate a conception relevant to such an action. Typically, this would
be the conception provided by the therapist. If the dient might have
been hesitant about accepting the conception at issue, the pressure of a
required commitment may enhance the probability of acceptance.
Mobilization of the dient's need for dosure may be useful at the
stage of consolidating new beliefs assumed therapeutic in their conse-
quences. In contrast, enlistment of the need to avoid dosure may be
useful at the stage of uprooting the old, dysfunctional beliefs assumed
responsible for pain and suffering. The need to avoid dosure is particu-
larly likely to arise when the price of making amistake is adjudged
severe. In this situation, the individual may be more than ordinarily
ready to question and reassess extant beliefs and sympathetically con-
sider the alternatives offered. For example, a person might be indined to
abandon a belief about his or her need to be approved by everyone if
that belief was ridiculed by the therapist and/or by members of a
therapy group. As a useful persuasive technique, the therapist might
therefore question the validity of dysfunctional beliefs while making
salient the personal price of such an "invalidity."
Ellis's therapeutic approach frequently invokes the client's fear of
invalidity in order to unfreeze his or her maladaptive beliefs and re-
freeze their more adaptive substitutes. I refer here to the somewhat
aggressive mode of questioning typical of RET in which the dient' s
dysfunctional ideas are challenged and ridiculed. According to Ellis
(1977),
RET largely consists of the use of [the]logical-empirical method of scientific
questioning, challenging and debating ... this kind of debating really con-
sists of rhetorical questions designed to dispute and rip up the false belief.
(p. 20)

Under those circumstances, the dient may indeed become quite moti-
vated to be rid of his or her beliefs and accept in their stead the more
"rational, sdentifically provable, empirically validatable" notions pro-
posed by the therapist. Invocation of the dient' s need to avoid dosure
through intellectual challenge is a prominent feature of the therapeutic
procedures developed by Beck, but is less frequently used by cognitive
therapists of the behavioral tradition (such as Bandura, 1977, or
Meichenbaum, 1977b).
208 CHAPTER9

CONTENTS OF THERAPEUTIC CHANGE

The foregoing discussion demonstrated how numerous elements of

cognitive therapy may be explicated in terms of the present analysis of
the epistemic process. This section will show that the content of desired
change represents a change in dients' frustrative beliefs and is accom-
plished via shifts in their assessments, goals, and means as outlined earlier.
In Bandura's conception (1977), for example, the aim of cognitive
therapy is enhancement of the dient's sense of self-efficacy. This is the
belief in one' s ability to attain through one' s own efforts important per-
sonal objectives. Stated in those terms, it is dear that self-efficacy repre-
sents a generalized denial of the frustrative belief that asserts just the
opposite, notably one's inefficacy. A similar type of change is striven for
in systems proposed by EIlis and Beck. Thus, EIlis (1978) described a
hypothetical case of an individual fired from his job who considers this
loss unfortunate and disadvantageous. EIlis commented that had the
individual believed that it really did not matter whether he kept or lost
his job he would hardly feel depressed or sad on the occasion. This
represents an instance in which a shift in the person's goals would have
a suffering redudng effect. Indeed, EIlis mentioned among possible ob-
jectives of the RET a change in the dient's core values, that is, a shift in
ultimate goals the individual deerns worth pursuing.
Ellis's therapeutic techniques frequently aim to bring about a means
shift in the clients' attempts to pursue their goals. Specifically, a dient
may be shown how to stay on a long term diet, be given training in
assertiveness without hostility, and generally be taught various new
means of getting to the desired ends. Finally, EIlis's techniques occa-
sionally indude induction of assessment shifts regarding the relation of
the means to the goal. For example, EIlis views full acceptance or toler-
ance of the dient by the therapist as a way of demonstrating that the
dient is acceptable even with his unfortunate present traits. In terms of
the present analysis, the dient's goal of being acceptable remains con-
stant, as does the person's means of attaining acceptability through a set
of personality traits, induding those the dient considers unworthy of
acceptance by others. Instead of encouraging the dient to modify his or
her goals or means, the therapist presents new evidence for the dient's
essential acceptability; this invites a reassessment and arevision of the
individual's former beliefs.
The categories of goal, means, or assessment shifts are readily ap-
plicable to the kinds of therapeutic techniques utilized by Beck. Thus, in
one reported instance, Beck (1976) portrayed as ridiculous a dient's
concern about strangers' opinions of hirn. This represents the therapist's
attempt to dis suade the dient from considering sodal acceptance by
KNOWING HOW TO CURE 209

strangers a worthwhile goal (hence, to facilitate a goal shift). Encourage-

ment of a means shift is apparent in Beck's (1976) reference to tech-
niques, such as "assertiveness training ... that may enhance the pa-
tient's self-confidence and other behavioral methods" (p. 214). An
assessment shift in the perceived means-ends relations is also fre-
quently referred to in Beck's reports about psychotherapy. Thus, in one
example, "a patient's response to his therapist's warmth and acceptance
often modifies his stereotyped conception of authority figures" (Beck,
1976, p. 214). In present terms, this signifies a shift in the dient's former
assessment that his or her goal of being accepted by authority figures
was frustrated.
The elements of means, goal, or assessment shifts are apparent also
in the works of cognitive therapists of a behavioral orientation. Thus,
Meichenbaum (1977b) reviewed various skill-training programs that of-
fer the dient new means of coping with difficulties in areas, such as
speech anxiety, test anxiety, sodal incompetence, sodal withdrawal,
and other problems. In other cases, the self-instructional techniques
practiced by behaviorally oriented cognitive therapists indude elements
of goal shifting. For example, a dient who feels anxious about being
interrogated by a date on a topic about which he knows very little may
be encouraged to utter to himself a statement such as "it is not such a big
deal to admit that I don't know something. There are probably lots of
things I know that she doesn't" (Meichenbaum, 1977b, p. 132). In this
case, the individual is dearly encouraged to abandon the goal of appear-
ing all knowledgeable. The possible failure to live up to this objective
appears now not nearly as devastating as might have been originally
feared.
Finally, proponents of self-instructional methods occasionally incor-
porate elements of assessment-shift techniques. For example, in a study
on the control of obesity, Mahoney and Mahoney (1976) viewed as
therapeutically appropriate such statements as "Eating does not satisfy
psychological problems, it creates them," or "My eating schedule is not
worse than anyone else's," that is, statements asserting relations be-
tween behavioral means and desired ends not heretofore apparent to (or
saHent for) the client.

LAY EPISTEMICS AND COGNITIVE THERAPIES:

A COMPARATIVE ApPRAISAL

As the foregoing discussion demonstrates, the lay epistemic frame-

work provides an integrative scheme for the various methods and tech-
210 CHAPTER9

niques employed by cognitive therapists of various persuasions. It is of

interest that many of the cognitive therapeutic techniques highlighted
by the lay epistemic analysis are typically presented in an offhand man-
ner as if they were the "ground" rather than the "figure" of the
therapeutic process. For instance, major contemporary systems of cogni-
tive therapy treat as obvious and undeserving of special comment the
criterion of logical consistency of belief assessment, the use of epistemic
authorities, and the reliance on epistemically relevant needs. Instead,
such systems highlight as conceptually critical several notions and hy-
potheses with which the present epistemic analysis is sharply at odds.
One might say that the lay epistemic framework can account for the kind
of things that cognitive therapists actually do though it differs from them
in what they say they do. In the following sections, I explore some such
differences and consider their implications for the practice of cognitive
therapy.

THE DYSFUNCTIONAL MISCONCEPTION (DISTORTION) HYPOTHESIS

To the question of why one should attempt to change through

therapy the client's conceptions and beliefs, many cognitive therapists
would probably reply that those particular ideas are dysfunctional; that
is, are maladaptive or irrational, and that they ultimately cause pain and
suffering for the client. The following passage from EIlis (1977) indicates
this viewpoint dearly:
by irrationality I mean any thought, emotion, or behavior that leads to self-
defeating or self-destructive consequences that significantly interfere with
the survival and happiness of the organism. (p. 15)

Thus far, there is no disagreement; in the present analysis, too, dysfunc-

tional thoughts are defined as pain inducing and are generally sub-
sumed under the label of frustrative hypotheses.
Where the lay epistemic analysis diverges from several major
schools of cognitive therapy is in conceptualizing the relation between
dysfunctional ideation and misconceptions about reality. In most cur-
rent approaches to cognitive therapy, the two are equated. Such, for
example, is a comerstone of EIlis's (1977) approach: "RET holds that
virtually all serious emotional problems with which humans are beset
directly stern from their magical, superstitious, empirically invalidatable
thinking" (p. 172). EIlis's list of "irrational" attributes inc1udes "sense-
less," "illogical," "unreasonable," "unrealistic," and/or "absolutistic,"
"overgeneralizations" that "distort reality." Hence, an essential rhetori-
cal question that EIlis would pose to his client in order to "dispute and
rip up the false belief is: In which way does it have truth or falseness?"
(EIlis, 1977, p. 20).
KNOWING HOW TO eURE 211

In Beck's system of cognitive therapy, dysfunction and distortion

are similarly equated. In Beck's (1976) terms, "psychological problems
may result from making incorrect inferences, and not distinguishing
between imagination and reality" (p. 19); thus "a profound or chronic
discrepancy between the internal and external systems may ultimate in
psychological dis orders (p. 24). Further, "problems are derived from
11

certain distortions of reality based on erroneous premises and miscon-

ceptions" (p. 20). Accordingly, the pivotal endeavor in Beck's therapy
aims at rectifying departures from reality. A major cognitive therapy
manual by Beck and his associates (Beck et al., 1979) asserts this in the
following terms:
The therapeutic techniques are designed to identify, reality-test, and correct
distorted conceptualizations and the dysfunctional beliefs (schemas) under-
lying these conditions .... The cognitive therapist helps the patient to think
and act more realistically and adaptively about his psychological problems
and thus reduce symptoms. (p. 129)

The equation of dysfunctionality with cognitive disortion is espe-

cially salient in Raimy's (1975) writings where this issue is extensively
treated under the heading of the "misconception hypothesis." Raimy
claimed that "successful psychological treatment occurs when faulty
ideas or beliefs are modified or eliminated" (p. 4). As Lazarus (1971)
asserted, "the bulk of psychotherapeutic endeavors may be said to cen-
ter around the correction of misconceptions" (p. 165).
In contrast to these views, typical of several major approaches to
cognitive therapy, our analysis takes issue with the dysfunctional mis-
conception hypothesis, that is, with the notion that erroneous thoughts
are pain inflicting. Life seems replete with examples in which uncom-
promising "truth" is anything but salutary in its affective consequences:
the betrayed lover, the bankrupt businessman, or the terminally ill pa-
tient may not only suffer terribly upon the discovery of the truth about
their condition but may sometimes be driven to self-destruction in such
circumstances. This is not to say that truthful ideas may not often be
beneficial and false ones pernicious in their impact. But, in general,
there seems little basis for concluding that a relation exists between the
veridicality of a belief and the happiness it may promote.
Buttressing this argument are commentary and data presented by
such personality researchers as Lazarus and Mische!. Lazarus (1980)
suggested that cognitive therapists conceptualize pertinent cognitive
processes in terms of faulty conceptions or irrationalities because of their
use of psychopathology as a reference framework for the understanding
of human functioning in genera!. Reminding us that mental health pro-
fessionals have for long been "monogamously wedded to reality testing
as the main hallmark of psychological well-being," Lazarus particularly
212 CHAPTER9

objected to the prevailing cognitive therapy view that "only accurate

perceptions of reality (or correct assumptions about life) can be adapta-
tionally effective and desirable" (p. 3).
Based on his own research on stress and coping, Lazarus (1980)
argued that well-functioning people may sometimes have to improve
their lot by "bending reality a bit to maintain certain illusions" (p. 3). In
fact, there are stress conditions (e.g., terminal illness, incapadtation,
aging) that essentially preeIude the direct action, problem-solving mode
of coping, leaving accessible primarily the intrapsychic (cognitive) pallia-
tive mode. This palliative function "regulates the distressing emotion
itself whieh arises from a harm or threat appraisal," and ineIudes "mini-
mizing, avoiding, denying, or detaching" (p. 4)-all representing what
many cognitive therapists have downgraded as maladaptive concep-
tions about reality. Lazarus therefore coneIuded that:
Although people often can be helped by getting them to see the hard realities
and the illusory and irrational nature of their assumptions about living, we
often also need the luxury of some illusions, and therapy in some instances
might better revolve around helping the person to think more positively
about his or her plight rather than fixating on the painful truth. So much of
life is ambiguous that there is much room for variation in appraisal. One's
cup can be half full rather than half empty. (p. 5)
Difficulties with a reliance on the cognitive distortion hypothesis in
explaining affective dis orders have been highlighted in recent self-
perception studies conducted with depressed patients. Thus, Lewin-
sohn, Mische!, Chapin, and Barton (1980) found that, for such individu-
als, initial self-evaluations of performance in group interaction, as well
as evaluations of such performance by independent observers, were
more negative as compared to self-evaluations rendered by non-
depressed psychiatrie controls and by normal controls. Of particular
pertinence is the finding that the self-evaluations of depressed patients
coindded more eIosely with the observers' evaluations than did evalua-
tions of subjects in both control groups; the laUer subjects' self-
perceptions were more positive than the observers' evaluations. In other
words, the depressives were more realistie than were the controls in
evaluating their own sodal competendes: They were eIoser to the sodal
reality as defined by the judgment of objective observers.
As Mischel (1979) pointed out, the findings of Lewinsohn et al.
(1980) are inconsistent with Beck's (1976) view regarding the etiologieal
importance in depression of an "unrealistically negative view of seIf."
Also inconsistent with this view are findings that depressed individuals
were surprisingly accurate in their recall of feedback (Nelson and
Craighead, 1977) and in their judgment of the degree of contingency
between their own responses and the outcomes obtained (Alloy and
KNOWING HOW TO eURE 213

Abramson, 1979, 1988). It is particularly interesting to note that, in the

Lewinsohn et al. (1980) study, a second assessment of sodal compe-
tence, made at a later point in treatment, revealed that self-perception of
the depressed patients became more positive and so more similar to that
of the controls, who did not change much over time; that is, with treat-
ment the depressed patients had become increasingly less rather than
more realistic. Mischel (1979) aptly concluded
It is tempting to conjecture that a key to avoiding depression is to see oneself
less stringently and more favorably than others see one. If so, the beliefs that
unrealistic self-appraisals are a basic ingredient of depression and that real-
ism is the crux of appropriate affect may have to be seriously questioned. To
feel good about ourselves we may have to judge ourselves more kindly than
we are judged. SeIf-enhancing information processing and biased self-
encoding may be both a requirement for positive affect and the price for
achieving it. (p. 752)

Finally, the hypothesized positive relation between realism and

well-being is at odds with the results of several misattribution studies
demonstrating that negative emotional behavior can be reduced if the
subject is "falsely" led to believe that his autonomie arousal was caused
by an extraneous stimulus (Nisbett and Schachter, 1966; Ross, Rodin,
and Zimbardo, 1969; Storms and Nisbett, 1970). On the other hand,
positive behavior falsely ascribed to oneself can encourage effective cop-
ing. Thus, in one study, Davison and Valins (1969) had subjects believe
that their alleged increased tolerance of pain is attributable to them-
selves or to a drug. The former subjects perceived the electrie shocks
they received as less painful than did the latter. Note that the beliefs of
both groups of subjects were essentially false because no increased pain
tolerance was involved (rather, shock intensities were surreptitiously
reduced); yet one group of subjects perceived their experiences as less
painful than the other group in accordance with the contents (rather than
the veridieality) of their attributions.

ARE DEPRESSIVES WISER?

The preceding discussion suggests that alleviation of the depressed

client' s suffering may depend on his or her success in relinquishing the
frustrative hypothesis rather than on the realism of his or her inferences.
Moreover, as argued in Chapter 8, the reality of our cognitions and
perceptions is dependent on a criterial judgment of some standard set-
ter; hence, it may not be objectively established anyway. In this sense,
the presumptive realism of the depressives in the Lewinsohn et al. (1980)
study is to be questioned. From the present vantage, it appears less
plausible that depressives are generally more realistic than normals than
214 CHAPTER9

that they may have shared the specific biases of the external observer,
whereas the normals might have possessed a different set of biases. For
example, normals may have had strong esteem needs disposing them to
believe in their considerable competence. Depressives might lack such
needs for some reason and in this sense be doser in avantage point to
impartial observers. Alternatively, the depressives' esteem needs might
be equally strong as those of the normals, yet the effects of these needs
might be counteracted by highly negative bellefs about themselves to
which the depressives subscribe. The net judgmental effect would be
the same as for an external observer lacking both the negative beliefs
and the preferential dosure need to protect the actor's ego.
Thus, according to our analysis, to agree with outside observers, a
prevailing consensus, or an expert opinion is merely to adopt a different
judgmental bias, rather than to avoid bias altogether or to attain ultimate
realism. Essentially, the therapist may never be certain that the cognitive
change which he or she advocates is any doser to the truth than the
client's original inferences. At the very best, the therapist may hope to
demonstrate that the beliefs he or she advocates, besides being consis-
tent with past evidence, are also consistent with new evidence contrary
to the dient's original ideas. This does not me an that the advocated
beliefs are immune to the possibility of being, in turn, contradicted by
further, as yet unexamined evidence. But the cllent's original beliefs
might also have been consistent with past evidence, so the dient' sand
the therapist' s evidence are more or less on par as far as their ultimate
truth status is concerned.
For example, consider a young male cllent whose depression is
apparently founded on the belief that all women despise him. Assume
further that this particular belief was prompted by two consecutive ex-
periences in which he was rejected by a woman. A therapist may quickly
challenge the dient' s belief on grounds of an overgeneralization: One is
not really justified in conduding that all women have a certain attitude if
only two women have exhibited it. Although possibly incorrect, the
cllent' s condusion is at least consistent with the facts: If all women
indeed despised him, any woman would tend to reject him, induding
the two who actually did.
Now suppose that, in an effort to overthrow this dysfunctional
generalization, the therapist arranged for a situation in which some
female members of the therapy group admitted to finding our man
attractive. Also suppose that, based on those pleasant experiences, the
dient was persuaded to abandon his original, depressing conclusion
regarding his sex appeal and to exchange it for the more optimistic view
that at least to some women he is not repellent. The latter condusion is
certainly more pleasing and therefore more functional for the cllent than
KNOWING HOW TO eURE 215

the former one. But is it therefore "truer"? Not necessarily. The individ-
ual may soon encounter several successive rejections outside of the
therapeutic group that might lead him to question the sincerity of the
apparent acceptances. Under those circumstances, he may feel justified
in reverting to the gloomy outlook with which he first came to therapy.
In other words, any interpretation, induding that offered by the
therapist, may at best be consistent with past evidence but could be
overthrown by future evidence. Any standard setter's inferences (i.e.,
criteria of accuracy) are subject to potential change. Thus, it does not
seem justifiable to regard the inferences the therapist wishes to impart to
the dient as in any objective sense more veridical than the dient' s origi-
nal judgments.
From this perspective, it seems inappropriate to characterize a
dient's conceptions uniquely as selective abstractions (Beck, 1976), be-
cause any abstraction is selective in that the same pattern of evidence is
compatible with a vast number of alternative interpretations (Campbell,
1969). Nor would it be justifiable to characterize the patient's inferences
as uniquely arbitrary, as in jumping "to a condusion when evidence is
lacking" (Beck, 1963, p. 94). Any inference is in a sense arbitrary: It may
not justifiably daim to be grounded in objective proof or to exdude all
potential alternative inferences consistent with the extant evidence.
The foregoing analysis is not meant to imply that, in attempting to
bring about cognitive change, the therapist should not depict the client's
beliefs as invalid and his or her own substitute propositions as of a
superior validity. To imply this would be to deny that people evaluate or
validate their hypotheses, whereas the present theory accords a central
place to the validational process. Even though at a given moment the
therapist's interpretation might appear more valid than the client's, this
should not be taken to mean that it is objectively true and unalterable.
Such a belief may render the therapist dangerously dosed minded
and impervious to various alternative suggestions and possibilities,
some of which might be potentially very useful in solving the dient's
problem.
Further, a therapist's belief in the infallibility of the scientific meth-
od or the necessarily erroneous nature of the client' s dysfunctional
thoughts might heavily load the therapeutic process toward the induc-
tion of cognitive change via a fear of invalidity. As we have seen, how-
ever, numerous alternative induction modes are open to the therapist.
For instance, where the dient' s adherence to a belief sterns from its
representing a specific preferred closure, a frontal attack on the validity of
his or her notions might fall or boomerang. Persuasive attempts aimed
at weakening the link between those notions and the underlying wishes
and desires might work more efficiently in those circumstances.
216 CHAPTER9

CONTENTS OF IRRATIONAL BELIEFS

As discussed earlier, a dysfunctional belief may represent the per-

ceived frustration of an important goal. This means that the contents of
dysfunctional beliefs can vary considerably in accordance with the diver-
sity of possible goals th:lt persons may have and with those individuals'
perceived capability to attain the goals in question. At the same time, it
should be possible to identify within a given cultural context a set of
relatively common, and yet impossible to attain goals. Those goals are
likely to represent prevalent frustrative hypotheses shared by large
numbers of people. Their explicit identification may profitably guide the
activities of a cognitive therapist. Indeed, several influential approaches
to cognitive therapy, such as those of Ellis (1978), Beck et al. (1979), and
Raimy (1975), specify lists of "irrational beliefs" or "themes" assumed to
underlie the suffering of numerous patients. Statements, such as "I
must be loved by absolutely everybody" or "I must succeed at every task
I attempt" are widely cited as examples of these irrational beliefs. Con-
trary to common assumption, the dysfunctionality of such beliefs does
not reside in their falsity, which is ultimately indeterminate, but rather
in their inclusion of impossible goals, likely to give rise to frustrative
hypotheses.
In this connection, it is important to note that personal goals may
shift, as may individuals' capacities to attain those goals. Thus, an irra-
tional and/or a dysfunctional goal may, with a shift in context, cease to
represent a goal or, to the contrary, come to represent a readily attain-
able goal. In other words, it is important to recognize the relative status
of dysfunctional themes. Unlike truth or falsity that are absolute and
constant, goals and relevant attainment capabilities may change in the
course of time. According to this analysis, no belief is inherently dys-
functionali it is dysfunctional only to the extent that it represents an
unattainable goal.
For example, Bandura's (1977) stress on the importance for the indi-
viduals' psychological welfare of beliefs in self-efficacy may be related to
the cultural goal of personal competence. Such a goal may be important
for the Western cultures, committed to the value of individualism, but
less important to alternative, communal cultures (e.g., to India's Bud-
dhist communities). At any rate, even in Western cultures, self-efficacy
is not the only goal that persons might have and whose frustration
might induce considerable suffering. Thus, a belief that every inhabitant
of the town in which one lives is an inferior dullard with whom there
can be no meaningful interaction might frustrate one's goal of having
fulfilling social relationships. A belief in an impending pain or danger to
one' s survival represents the frustration of one' s goal of physical well-
KNOWING HOW TO eURE 217

being. None of the latter goals seems to belong in the self-efficacy

category, yet a belief in their frustration would seem capable of engen-
dering considerable pain and in this sense would be a proper target for
cognitive therapy.
Finally, approaches to cognitive therapy with conceptual roots in
attribution theory (Heider, 1958; Kelley, 1967) often deal with the specif-
ic contents of hypotheses without sufficient concern about how these
relate to important goals. For instance, the preoccupation of attribution-
al researchers with internal or external ascriptions of causality represents
an interest in specific cognitive contents that are classified as generically
functional or dysfunctional without regard to the relation of such con-
tents to individuals' goals. Thus, internal ascription of behavioral
changes is usually considered beneficial for aperson, because it is as-
sumed to suggest persistence and stability, paralleling the assumed sta-
bility of the internal self. But according to the lay epistemic analysis,
internal attribution should be considered beneficial only if the changes
represent attainment of formerly frustrated goals. In time, those goals
could shift, in which case behavioral persistence as such could become
irrelevant or even contrary to the individual's well-being. For example,
behavioral toughness and violence could serve well the member of a
street gang, yet their persistence would turn out to be counterproduc-
tive if this individual aspired to an acceptance by the middle-class so-
ciety. Ultimately then, whether a content of a given attribution or in-
ference (like the internal attribution just referred to) could be beneficial
to the individual would depend on whether or not it represented an
abandonment of the frustrative hypothesis. The latter thus remains the
essential content category of relevance to cognitive therapy.
To summarize, although the lay epistemic approach acknowledges
the utility of constructing apriori lists of pervasive dysfunctional themes,
it stresses that (1) the dysfundionality of the themes derives from their
inclusion of impossible goals rather than from their falseness, and that
(2) pervasive cultural goals may shift in time, and an impossible goal may
cease to be so in the course of time. This implies a cautiously relativistic
approach to lists of dysfunctional themes, as opposed to an absolutistic
approach based on their identification as false.

PRIMACY OF PERSONAL EXPERIENCE AS THE MEDIATOR OF COGNITIVE

CHANGE

The present epistemic approach differs from those analyses that

regard personal experience of performance outcomes as a uniquely
effective media tor of cognitive change. For example, Kopel and
Arkowitz (1975) comment that
218 CHAPTER9

the therapeutic effects brought about by observing one's own behavior may
be more powerful than the effects obtained from a verbal interchange which
conveys the basic information to the dient. (p. 195)

Bandura (1977) similarly implied that expectancies or inferences arrived

at via actual performance are superior to those based on alternative
modes of persuasion (e.g., verbal) or on vicarious experience, because
the latter methods "do not provide an authentic experiential base for
them." (p. 198). Therefore, "it is performance-based procedures that are
proving to be most powerful for effecting psychological changes [and]
cognitive events are induced and altered most readily by experiences of
mastery arising from effective performance" (p. 191).
Within the present conceptual framework, personal experience rep-
resents the ca se in which it is necessary for the knower to rely on his or
her own perceived epistemic authority in judging the validity of an
inference. For example, a success experience on a task is unlikely to
uplift one' s feelings of self-efficacy if one doubted one' s own qualifica-
tion for pronouncing the experienced outcome a success. Bandura (1977)
did note that the efficacy of experiential method is constrained by the
interpretations imposed on the experiences. In the present framework,
any "experience" is considered an interpretation; hence, the capacity to
learn from experience should depend on one's assumed capability to
lend meaning to or impose an interpretation on the experience. Directly
relevant to these notions is the study by Ellis (1984), described in Chap-
ter 3, in which subjects high in self-ascribed authority benefited more
from experientiallearning (of mathematics or empathic skills) than sub-
jects low in self-ascribed authority.
According to the present analysis, the knower's self-ascribed au-
thority is merely one among several factors affecting his or her tendency
to accept a given hypothesis. Other relevant factors are alternative au-
thorities linked with this particular hypothesis, authorities associated
with alternative hypotheses, and the motivational significance of the
hypotheses. From our perspective, the question of which of these fac-
tors is the most effective mediator of cognitive change cannot be settled a
priori. The answer would depend on the relative strengths or levels of
these various factors in a particular set of circumstances.
In summary, the lay epistemic scheme, though capable of account-
ing for many of the activities that actually constitute cognitive therapy,
differs from major schools of such therapy in the interpretation imposed
on those activities. The theory of lay epistemics does not assurne that, in
the course of therapy, warped and distorted beliefs are replaced by
veridical ones, but rather that suffering-promoting beliefs are replaced
by satisfaction-producing ones. Nor is it assumed that teaching an indi-
KNOWING HOW TO eURE 219

vidual the scientific way of reasoning is an effective avenue of curing his

or her ills. As will be argued in Chapter 10, the lay and scientific mo des
of reasoning are fundamentally similar, and neither guarantees the con-
tent of beliefs ultimately arrived at. Therefore, application of the scien-
tific way of reasoning cannot, in and of itself, have therapeutic effects.
Unlike some major systems of cognitive therapy, the lay epistemic
approach stays relatively uncommitted regarding the contents of beliefs
in need of change. Its only emphasis is that cognitive therapy should
concern itself with beliefs whose content is frustrative in the sense elabo-
rated above. Finally, unlike some behaviorally oriented approaches, the
present analysis does not assurne the primacy of behavior or personal
experience as the mediators of cognitive change. Both are assumed to
represent a case in which the knower' s self-ascribed authority is brought
to bear on the inference process. The extent to which this facilitates the
arrival at confident inferences should depend on the degree to which
such self-ascribed authority is high or low in a specific situation.

IMPLICATIONS OF THE LAY EPISTEMIC ANALYSIS FOR THE PRACTICE OF

COGNITIVE THERAPY

If the present epistemic analysis were accepted, current practices of

cognitive therapy would be more likely to succeed in cases in which they
embodied the present theoretical constructs rather that in alternative
cases. For example, the therapist's critique of the dient's dysfunctional
thoughts on the grounds of their being "unprobable generalizations"
might successfully unfreeze such thoughts to the extent that it would
arouse the person' s fear of invalidity and thus enhance the need to avoid
nonspecific dosure. But, in some cases, the same critique might fail to
arouse such a fear: The dient could be a philosopher of science firmly
convinced that no beliefs are provable and that all statements are over-
generalizations anyway. It is doubtful that the therapist's approach
would have the expected curative effect in this case. Under the circum-
stances, fear of invalidity might be aroused by alternative means, such
as by pointing out to this dient that his or her dysfunctional views are at
variance with a philosophical authority he or she admires.
Similariy, a therapist's attempt to uproot a dient's belief on the
grounds of its being induded in an apriori roster of "irrational contents"
would fail to have a therapeutic effect if the belief in question did not, in
fact, represent a fmstrative hypothesis. For example, an individual
might believe that unless all women adored hirn he would undergo a
nervous breakdown, but might fortunately also believe that all women
did, in fact, adore hirn. A modification of these beliefs is unlikely to
enhance this person's satisfaction and well-being. In the same vein, the
220 CHAPTER9

use of group therapy might fail or boomerang if a dient had a negative

view of groups as epistemic authorities. Similarly, exposure to personal
experiences would presumably fail in cases in which an individual dis-
trusted his or her own epistemic authority in a given domain (EIlis,
1984).
According to this argument, current practices of cognitive therapy
though often effective, often might not be. Which would be the case
would depend on the degree to which they "operationa11y defined" the
constructs of the present epistemic analysis. These constructs may be
thought of as generative principles from which it should be possible to
derive a variety of operational techniques. For instance, the construct of
epistemic authority suggests that people might vary in the perceived
persuasiveness of different communication sources: In administering
cognitive therapy to a member of astreet gang, it may be more effective
to enlist the assistance of the gang leader, rather than rely on the profes-
sional status of the therapist.
Let us now consider the general image of cognitive therapy sug-
gested by the present lay epistemic analysis. First and foremost, it
should be a highly contingent therapy that derives particular treatment
procedures from the dient' specific circumstances. There would be no a
priori postulation of specific irrational beliefs or a commitment to an
invariant sequence of treatment techniques to which a11 dients are uni-
formly exposed. Instead, the therapist would aim to establish (1) what
frustrated goal apparently underlies a given dient's distress, and (2)
why the dient perceives the goal as frustrated. The therapist would then
attempt to determine whether the goal appears unrealistic, given the
client's means and capabilities, or whether, given the appropriate
means, the goal appears attainable. Answers to these questions would
help the therapist in choosing the direction in which he or she may take
the process of therapeutic persuasion. The therapist would decide
whether the dient' s goal might best be shifted, whether the dient might
be profitably induced to try a new means of getting to the goal, or
whether the dient might best reassess his or her frustrative belief in the
light of new evidence.
The therapist should also aim to characterize the dient's unique
epistemic sensitivities and predispositions. An attempt should be made
to identify the client's positive and negative epistemic authorities and to
ascertain his or her self-perception as an epistemic authority. The
strength of the client's needs for closure and the avoidance of closure
should be appropriately assessed, and specific dosures relevant to the
dysfunctional beliefs should be identified.
Subsequent treatment should be dosely tailored to the dient's
"epistemic style" and to the contents of the cognitive shift that seems
KNOWING HOW TO eURE 221

indicated. For example, a dient with a particular respect for expert au-
thority might profitably be exposed to individual sessions with the
therapist, whereas a dient with high regard for group authority might
be more profitably exposed to therapy in group settings, and a dient
with high self-ascribed authority might be exposed to relevant personal
experiences. Dysfunctional beliefs should be addressed from the stand-
point of the needs they might be tacitly serving. An attempt might be
made to convince the dient that those same beliefs may be served more
efficiently by alternative, more functional beliefs. An appropriate use
should be made of the dient' s dosure and dosure-avoidance needs at
the respective stages of unfreezing and refreezing the appropriate
beliefs.

SUMMARY

The lay epistemic framework provides a perspective for viewing the

process of cognitive therapy in which dysfunctional or frustrative judg-
ments are replaced by more adaptive alternatives. Although sharing
with several alternative approaches the assumption that cognitions are
important mediators of suffering and distress, the lay epistemic frame-
work differs in important respects from the major schools of cognitive
therapy. In particular, it locates dysfunctionality in the content of indi-
viduals' judgments rather than in their truth status or veridicality. It also
offers novel suggestions for carrying out the therapeutic enterprise,
based on an identification of the dysfunctional contents in need of
changing and on the dient' s unique features of the epistemic process
that may be utilized to facilitate the desired change.
 10
The Social Psychology of Science
On the Lay Epistemic Underpinnings of Research
Methodology

Leading social psychologists, such as Heider (1958), Jones and Davis

(1965), and Kelley (1967), have likened the processes of lay judgment to
the inferential procedures used by sophisticated scientists. In this final
chapter, I will attempt a comparison of naive and scientific inference in
order to show how important aspects of the scientific method may in-
deed reflect the processes of lay epistemics.

INTRODUCTION

Modern times rightly deserve to be called the "Age of Science."

Never before have human lives been governed to such a degree by
artifacts, technologies, and prescriptions derived from scientific re-
search. The hopes of contemporary individuals for surmounting the
remaining difficulties (cancer, pollution, energy shortages, AIDS) also
are bound, for better or worse, to the hopes of future scientific suc-
cesses. This is a major reason why immense material resources are allot-
ted to the advancement of science, and why countless women and men
the world over toll day and night in devoting major portions of their
lives to the work of science.
The glory of science is generally assumed to come from its method.
Rather than relying on mere intuition, the scientist is thought to have at
his- or her disposal a set of superior inferential aids-the controlled
experiment, the statistical analysis, the computer program-which, if
not completely guaranteeing the validity of scientific conclusions, are at
least thought to assure their high probability.

223
224 CHAPTER 10

Yet the ways in which science is viewed have changed drastically in

the last few decades. The works of Popper (in particular, 1972), Feyera-
bend (1976), Kuhn (cf. 1962, 1970), and Laudan (cf. 1977) have seriously
underrnined the conception that had previously lent science much of its
prestige and influence. These analyses have cast doubt on the assump-
tion that science has a unique method of discovery or of proof. Similarly,
the demarcation criteria assumed to set apart scientific from nonscien-
tific inference now appear problematic (Weimer, 1979). Instead of view-
ing it as a natural consequence of applying a rational method of inquiry,
the growth of scientific knowledge has been reconstrued in terms of
historical or sociopsychological processes (Feyerabend, 1976; Kuhn,
1962; Weimer, 1979). According to this interpretation, (1) all scientific
knowledge consists of shared conjectures whose objective validity can-
not be ascertained; (2) any scientific knowledge can be undermined or
toppled in the course of future "scientific revolutions."
These notions are not exactly novel and, in recent years, have be-
come a central topic of discussion among philosophers and sociologists
of science (Feyerbend, 1976; Kuhn, 1962, 1970; Lakatos, 1970, 1974;
Laudan, 1977; Merton, 1973; MitroH, 1974; Mulkay, 1979; Popper, 1966,
1972). Nor have they remained foreign to psychologists (cf. Cronbach,
1982; Gergen, 1982; Gholson and Barker, 1985; Weimer, 1979; Weimer
and Palermo, 1973). But it is one thing to accept a set of ideas in the
abstract, and another to recognize their implications for everyday scien-
tific affairs. The purpose of this chapter is to bridge the gap and apply a
social cognitive concept of science, embodied in the present epistemic
theory, to major methodological precepts and injunctions in the behav-
ioral sciences.
According to the present concept, the process whereby scientists
and lay persons arrive at their "knowledge" is essentially the same. This
fact renders a theory of lay epistemics simultaneously a framework for
understanding scientific epistemics or research methodology. In the sec-
tions that follow, I will attempt to substantiate the foregoing thesis by
showing how major methodological constructs in psychology represent
aspects of the general epistemic process characterized earlier. From this
perspective, prevalent concepts of research methodology will be ex-
panded and supplemented by various social cognitive variables of epis-
temic relevance.

OFFICIAL AND UNOFFICIAL RESEARCH METHODOLOGIES

It is possible to discern two types of methodologies that are prac-

ticed in a field of science. One, the official methodology, consists of ex-
plicitly stated rules and techniques for conducting research. This type
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 225

usually consists of "rationally" sounding maxims often expressed in

formal (e.g., mathematicalor logical) terms. In psychology, for example,
official methodology consists of suggested approaches to research de-
sign as well as data collection and analysis. Official methodology typ-
ically constitutes an obligatory part of graduate curricula. It is thought to
impart to students the tools of their trade and, in this way, to initiate
them to their chosen profession. By contrast, the unofficial methodology is
rarely taught in graduate school. It consists of "scientific folk wisdom"
garnered by researchers via years of personal experience. It bears on the
extent of one' s involvement in the social and institutional activities of
one's discipline (indexed, e.g., with attendance at congresses and con-
ferences), criteria for choosing the outlets for one's scientific publica-
tions, the style of presenting one's ideas to scientific audiences (Day,
1979) and of relating to one's colleagues and graduate students (Zanna
and Darley, 1980).
According to the present analysis, the latter, unofficial methodol-
ogy may be quite relevant to one's scientific efficacy and in this sense at
least may complement the official methodology. Both the official and the
unofficial methodologies are considered facets of the process whereby
persons (lay persons and scientists alike) acquire their beliefs on various
topics. Let us first consider in those terms several official constructs in
psychological research methodology.

ÜFFICIAL METHODOLOGICAL CONCEPTS

Whether or not they state so explicitly, most methodological con-

structs in psychology refer to maximizing the validity of interpretations
imposed on data. According to present theory, the validation of all
inferences is carried by implicational logic. Thus, it is interesting to
discover that official methodological constructs reduce, upon dose ex-
amination, to a conjunction of such logic to particular content categories
of inference. Note that, strictly speaking, the contents of scientific in-
ferences should not be considered part of the scientific method. After
all, the contents of physics differ widely from those of biology, for exam-
pIe, yet the scientific method as such is, by definition, common to all
fields of science. Nevertheless, if we strip inferential contents from vari-
ous methodological constructs, much of what remains is the principle of
logical implication. Let me demonstrate this by examining a few promi-
nent methodological notions, starting with the scientific experiment.

The Scientific Experiment

According to the present argument, the controlled scientific experi-
ment conjoins deductive logic to the content category of causal ques-
226 CHAPTER 10

tions. Typically, causality is assumed to denote (1) the temporal prece-

dence of the cause relative to the effect and (2) the covariation of the
two, such that, in a given situation, appearance of the cause paralleis
appearance of the effect, and disappearance of the cause paralleis disap-
pearance of the effect as weIl. 2 These assumptions conceming the mean-
ing of causality serve as a major premise in the experimenter' s syIlo-
gism. On their bases the experimenter may postulate that if x
(invariably) precedes y in time and also covaries with y, then x is the
cause of y. Execution of the experiment proper serves to establish the
minor premise that by arranging (1) a situation in which x is actually
made to precede y, (2) observational conditions for noting whether x and
y covary in those circumstances. If so, the antecedent in the major prem-
ise is affirmed and one may derive the consequent that x is the cause
of y.
Often one might identify a "confounded" variable, z, that has not
been separated properly from x in previous work; this may make it
difficult to tell whether z or x was the cause of y. In such circumstances,
we typically conduct a controlled experiment separating, in some condi-
tions, the x from the z. The compound deductive logic of such an experi-
ment is based on the assumption that if x preceded y, and y covaried
with it while z preceded y, and y did not so covary, then x (rather than z)
is cause of y. The various experimental designs employed by behavioral
scientists (Campbell and Stanley, 1963) attempt to control in this fashion
for a variety of spurious variables that may pose threats to hypothesized
causes. More specifically, the differences among experimental designs
lie in the content categories of threats being controlled for. Yet the un-
derlying principle of testing is identical in all experimental instances: it
consists of implicational reasoning conjoined to the more general content
category of causality. More about this matter will be said in later seetions.

Inferential Statistics
In· essentially these terms, one might think of inferential statistics
that, with the controlled experiment, count among the more important
scientific tools of the psychological researcher. In this case, too, logical
reasoning is conjoined to a specific content of a hypothesis one typically
wishes to control for, notably the statistical null hypothesis: A central

20ften (cf. Cook and Campbell, 1975) a third denotation is recognized, notably (3) an
absence of alternative causes of the effect. An alternative cause or an alternative explana-
tion represents a hypothesis inconsistent with the statement that Xis the cause of Y. But an
inconsistent hypothesis undermines the confidence in any inference, not just a causal
inference. Thus, the third denotation above turns out to be nonunique to the notion of
causality.
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 'l27

concern of inferential statistics is with the possibility that a seeming

"effect" is explicable by a sampling variability rather than a systematic
relation between variables. This possibility is tested in a characteris-
tically logical fashion, by deriving its specific implications, suggested by
an assumed statistical model, and probing them in reference to relevant
data.

Test-Theoretic Notions: Reliability versus Validity

In psychological testing, a fundamental distinction is often drawn
between the reliability of a test and its validity. According to Anastasi
(1969, p. 71),
Reliability refers to the consistency of scores obtained by the same individu-
als when re-examined with the same test on different occasions or with
different sets of equivalent items. . . . In its broadest sense, test reliability
indicates the extent to which individual differences in test-scores are attribu-
table to "true" differences in the charaeteristics under consideration ... [by
contrast] the validity of a test concerns what the test measures and how weIl
it does so. (p. 99)

Often test validity is also defined by the question of whether the test
measures what it is supposed to, and test reliability, whether it does so
consistently.
But on a dose inspection, it turns out that essentially both validity
and reliability revolve about the former question, whether the test mea-
sures what it is supposed to. In other words, the construct supposedly
measured by a test has certain implications that issue from its presumed
meaning. To the extent that test scores contradict those implications,
the test is judged at fault. In these terms, high test-retest reliability is
desired because psychological tests are typically designed to measure
stable traits (such as intelligence). In other words, we assurne that the
trait is constant across time. Low test-retest reliability suggests that the
test is unresponsive to this fundamental stability. Hence, the test does
not seem to be tapping the target trait, or not measuring what it is
supposed to.
Alternatively, consider reliability defined as an internal consistency
across item aggregates (Epstein and O'Brien, 1985). To the extent that
the trait being measured is uniform, one would expect different ways of
measuring the same trait, or different item aggregates to yield similar
results. The notion of trait uniformity implies consistency across items,
just as the notion of trait stability implies consistency across time. Low
internal-consistency reliability indicates that the test falls to respond to
the fundamental uniformity of the trait. This may again indicate that
the test is not tapping the target trait or not measuring what it is sup-
posed to.
228 CHAPTER 10

To summarize, notions of test reliability and validity pertain to the

same underlying problem, whether the test measures what it is meant to
measure (Cronbach, 1972). To the extent that this notion defines the
issue of validity, reliability also concerns validity of sorts: The construct
of the test has various implications regarding answers to test items. Valid-
ity (or reliability) checks probe whether those irnplications are upheld.
In other words, validity (reliability) notions apply implicationallogic to
contents yielded by a given trait construct. Just as the notion of causality
implies covariation, the notion of trait implies stability and uniforrnity.
This suggests that the experiment and the psychological test share an
important commonality: Both apply implicationallogic to specific con-
tents of inferences. But the commonality of methodological constructs
extends even further , as shown below.

Taxonomies o[ Validity
The methodological literature in psychology contains several tax-
onomies of validities. For example, Campbell (1957; Campbell and
Stanley, 1963) distinguished between internal and external validities. In a
later work, Cook and Campbell (1979) expanded this distinction into a
fourfold classification by adding two additional validity types: construct
validity and statistical conclusion validity. In a previous influential essay,
Cronbach and Meehl (1955) identified alternative validity types relevant
to psychological tests; notably, concurrent, predictive, construct, and
content validities.
Although undeniably useful for various purposes, such taxonornies
could cloud the idea that all validity types share a fundamental com-
monality, the quintessence of the validity notion. Validity refers to sci-
entific propositions. These are considered valid when compelling evi-
dence is marshalled in their support. In turn, scientific evidence for a
proposition has to do with what the proposition implies for specific em-
pirical circumstances. In short, a proposition is considered valid if it can
be derived from the observed evidence. 3 To put it differently, the essen-
tial notion of validity is that of derivability. And validity types differ
among themselves in the content categories of propositions and the types

3Contrary to the widely accepted notion that hypotheses can only be falsified by evidence
and not verified, positive deduction of hypotheses from evidence is presently considered
consistent with the canons of logic. In particular, the researcher could assurne that "only if
hypothesis H is true would datum D be observed." Such a statement of material equiv-
alence renders logically legitimate a straightforward derivation of the hypothesis from the
data. A juxtaposition of this analysis with Poppers (1959) falsificationism is given in
Kruglanski and Mayseless (in press).
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 229

of evidence from which such contents are believed derivable. Let us

examine a few examples.
Consider the notions of internal and external validity mentioned
earlier. Internal validity, on the one hand, was defined as the question
of whether the treatment within the specific experiment made a differ-
ence (Campbell and Stanley, 1963, p. 5). External validity, on the other
hand, was defined as the issue of generalizability, notably across "set-
tings, treatment variables and measurement variables" (p. 5). But what,
precisely, constitutes the "specific experiment" in the definition of inter-
nal validity? Presumably, this is a conjunction of boundaries, concerning
time, place, and personnel, that encases a given research instance
(Cronbach, 1982; Kruglanski, 1975; Kruglanski and Kroy, 1976). Such
boundaries qualify the presumed treatment effects in the relevant causal
proposition. An example would be this proposition: In an experiment by
Smith performed in January, 1985, at VCLA, monetary payment was
negatively associated with subjects' tendency to work on a puzzle dur-
ing a free period. In other words, the issue of internal validity pertains to
a specific content dass of statements, those that explicitly specify the
boundary conditions of an experiment.
By contrast, external validity pertains to the dass of "experimentally
unbounded" statements ab01~t treatment effects. An example would be
this proposition: When offered material rewards, people's intrinsic moti-
vation for a task dedines. Vnlike the proposition in our first example,
this statement pertains to numerous further times, situations, and per-
sons. That is why external validity is considered to pertain to gener-
alizability, as noted earlier.
In other words, internal and external validities may be interpreted
to refer to distinct content categories of statements. In this sense, inter-
nal or external validities do not pertain to data as such, only to findings,
that is, interpretations imposed on data: Internal validity refers to "ex-
perimentaIly" bounded interpretations, whereas external validity refers
to "experimentally unbounded" ones. Admittedly, the latter interpreta-
tions are more general or indusive than the former ones because an
effect that obtains across all situations surely obtains in a given situation
as weIl. In this sense, doubts concerning internal validity threaten exter-
nal validity as weIl, whereas confidence about external validity applies
also to that about internal validity.
But beyond content differences in level of generality, internal and
external validities are assessed in exactly the same logical way; that is, by
drawing the implications of the particular statement being validated and
noting whether these are empirically affirmed. In the case of internal
validity, for example, the experimenter might reason that only if treat-
ment X caused the effect would subjects exposed to the same experi-
230 CHAPTER 10

mental history except for X exhibit systematically different behavior.

Based on such premises, the experimenter would consider the causal
role of the treatment (or the internal validity of the study) established, if
the postulated empirical implications were affirmed.
Similarly in the case of external validity, the experimenter might
reason that only if treatment X causes the effect in people generally,
would the results replicate across age levels and sexes. Once again, one
would consider the external validity of the findings established if the
postulated implications were empirically supported. In short, the method
of establishing internal and external validities is simple logical deriva-
tion. And the differences between the two validity types derive from the
content categories of statements to which they respectively pertain.
Note that while experimentally unbounded statements are more
general than experimentally bounded ones, either statement type can
have unlimited implications. According to our LEGO metaphor, any
idea can be implicationally linked with any other idea. Hence, any
datumsupporting a treatment effect could be interpretable in terms of
numerous competing hypotheses. Let me elaborate.
Suppose one used an experimental design of the pretest posttest,
experimental-control-group type (cf. Campbell and Stanley, 1963). If the
mean difference between the pre- and postmeasures in the experimental
group significantly exceeded the corresponding difference in the control
conditions, one might conclude that treatment X had an effect, if by X
we mean the only difference that separates the experimental and control
conditions. But here, precisely, lies the catch. Are we objectively justi-
fied in stating that X constitutes the only such difference? In principle
we are not. Suppose our X was a sum of ten dollars handed to subjects
in the experimental condition but not in the control conditions. "Ten
dollars" is a conceptual category, whereas what we had in the experi-
ment was a particular ca se of that category inevitably contaminated by
endless contextual specifics, for example, the number of monetary bills,
their size and color, the way they were handed to subjects, or their
particular weight. For all we know, each of those contextual correlates of
X, rather than X as a category, might have produced the observed effect.
Now the notion that the color, say, of a ten-dollar bill had any effect
on subjects' motivation might strike us as unreasonable and farfetched.
Yet this is precisely because of our presumed familiarity with the psy-
chology of our college sophomore subjects. Imagine the very same study
conducted with Martians. How sure would one be in those circum-
stances that color had nothing to do with our subjects' reactions? Proba-
bly not very sure, because our "knowledge" of Martian psychology does
not apriori exclude color as a determinant of behavior.
In short, neither internal nor external validity can be established
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 231

securely in any objective sense of the term. This point is stressed these
days by major philosophers of science (as Feyerabend, 1976; Kuhn, 1962;
Laudan, 1977; or Popper, 1959). Cronbach (1982) crisply expressed this,
in his statement that "validity is subjective rather than objective: The
plausibility of a conclusion is what counts. And plausibility, to twist a
cliche, lies in the ear of the beholder" (p. 108). From this perspective, it is
impossible to determine whether internal or external validity is more
attainable in a research design (Campbell and Stanley, 1963). Any datum
could be implicationally linked to any number of hypotheses, and no
amount of testing could conclusively prove that one of those and not the
others represented the objective reality. Thus, it is always possible in
principle, if not in practice, to raise reservations about the internal valid-
ity of a study just as it is about external validity.
Similar considerations may apply to other types of validity identi-
fied in the methodological literature. In all those cases, too, validity
pertains to deductibility conjoined to different content classes of state-
ments. For example, "concurrent validity" concerns implications about
contemporaneous events; "predictive validity" concerns implications
about future events, whereas "construct validity" concerns implications
that derive from abstract concepts (Cronbach and Meehl, 1956). In the
case of "content-validity," the issue is whether a given trait construct
indeed implies a specific response pattern to items on a test (Cronbach
and Meehl, 1956).
Generalizing from the discussion thus far, it appears that as a group
"official" methodological concepts in psychology (the controlled experi-
ment, validity and reliability types, and inferential statistics) conjoin
implicationallogic to different content types of inference. Furthermore,
if one subscribes to the assumption that the notion of " scientific meth-
od" only makes sense if kept conceptually separate from local inferential
contents, then much that remains of "official scientific method" is im-
plicationallogic. The above analysis turns out to have important conse-
quences for the designing of research, a topic I will address in the next
seetion.

Research Designs and Their Hierarchies

The concept of a research design denotes a plan for setting up the
investigative situation with the aim of maximizing the validity or inter-
pretability of the data collected. The accepted methodological doctrine in
the behavioral sciences espouses the thesis of design hierarchies. This is
the notion, that with respect to certain purposes, some research designs
are inherently and universally superior to others. Such a hierarchy, now
well-known, was introduced by Campbell (1957; Campbell and Stanley,
232 CHAPTER 10

1963) in the chapter referred to above. This particular hierarchy intro-

duces a distinction between "true" and "quasi" experiments, in which
the former are depicted as generally preferable or superior to the latter.
A true experiment is noted by the full control an experimenter has
over "the scheduling of experimental stimuli ... and the ability to ran-
domize exposures" (Campbell and Stanley, 1963, p. 34). In a quasi ex-
periment, the investigator "can introduce something like experimental
design into his scheduling of data collection procedures" (p. 34). This,
however, may not fully alter the fact that "the results of quasi experi-
ments are less interpretable than the results of true experiments" (cf.
Cook and Campbell, 1975, p. 224).
It is not that true experiments are presumed to yield thoroughly
valid conclusions. Indeed, Campbell and Stanley (1963) explicitly note
that the risk of misleading the research community is not unique to
quasi experiments. But this is considered a risk shared in kind only and
not in degree by true experiments (p. 35). The risk in kind is assumed to
exist because "in a very fundamental sense experimental results never
'confirm' or 'prove' a theory" (p. 35). This seems to weaken the dicho-
tomy between quasi and true experiments. Yet without more explicit
disclaimers, the basic distinction is retained and is thought to relate to
the relative degrees of validity presumably afforded by the two classes of
research design.
On eloser examination, the idea of a design hierarchy presupposes a
commitment to a finite list of "threats to validity." Campbell and Stanley
(1966), for example, list the following threats to internal validity: (1)
history, (2) maturation, (3) testing, (4) instrumentation, (5) statistical
regression, (6) selection biases, (7) experimental mortality, and (8)
selection-maturation interaction. Cook and Campbell (1975) added to
this list the following further "threats": (9) diffusion or imitation of the
treatment, (10) compensatory equalization of treatment, (11) compensa-
tory rivalry, and (12) local history. A research design would occupy a
high place in the hierarchy the greater the number of such threats it
controlled for. Thus, the true experiment was presumed to control for a
greater number of threats than a case study. Hence, it was accorded a
higher place in the hierarchy (Campbell and Stanley, 1966; but see
Campbell, 1975).
But the question is whether any list of threats can be considered
finite or exhaustive. For if it turns out to be unlimited or open-ended
instead, the notion of a preordained design hierarchy would lose much
of its attraction. Let us consider these issues more elosely. To begin
with, observe that, in the absolute, the notion of "threat to validity"
presupposes a stable distinction between theoretically interesting vari-
ables as opposed to "threats." Yet it appears that this distinction must be
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 233

relative rather than absolute. For example, "maturation" could con-

stitute a threat in one research context but a theoretically interesting
variable in another context, for instance, to a developmental psycholo-
gist. Thus, one person' s threat could be another' s treatment variable and
vice versa.
A blurring of the distinction between threats and variables suggests
that a list of possible threats could be none other than that of possible
variables. But the latter list surely can be quite vast and open-ended.
After all, variables are constructs invented by creative scientists. Hence,
the limits of a variable list approaches the limits of human inventiveness
and creativity. These are unlikely to be approximated by a list including
eight, twelve, or even twenty variables. This implies that any list of
threats could be highly relative and temporary: It could be relative to the
state of knowledge within a field of research, and it could vary con-
siderably from one field of science to another.
Examples of such relativity are not hard to find. For instance, a
sociologist might be concerned about possible "sodal-class" confounds
that are of a relatively lesser interest to a sodal psychologist. In turn, a
sodal psychologist might worry about the "demand characteristics" in-
herent in a set of experimental procedures, whereas a cognitive psychol-
ogist might be less concerned about "demand-based artifacts." A devel-
opmental psychologist with an interest in "life phases" might be
skeptical about the validity of sodal psychological research conducted
with college freshmen. In short, the answer to the validity question may
often depend on whom you ask.
Furthermore, what qualifies as a threat could shift over time with
changes in state-of-the-art knowledge. For instance, Campbell
(Campbell and Stanley, 1963) at one point expressed concern about the
reactive effects of the pretest, but his concerns were apparently allayed
(Campbell, 1969) by Lana's (1969) reassuring review of empirical re-
search on the topic. Occasionally, spedfic threats could be rendered
saHent and/or plausible to a community of sdentists by political or ideo-
logical trends in the society at large. For instance, with the advent of the
feminist movement in the United States, there has been a noticeable
increase in the inddence of developmental studies that include gender
as a control variable (Deaux, 1985). All of which goes to show that a static
list of threats is incompatible with the dynamic nature of sdence.

Implications tor Design Hierarchies

The present argument is not intended to imply that a list of threats
is totally devoid of utiIity; rather , it suggests that its utiIity is relative to a
given sdentific audience at a given point in time. A list of threats could
234 CHAPTERIO

hardly furnish the comprehensively general guide to research design

with which it has been often credited (Saxe and Fine, 1981). But an even
more irnportant irnplication of our "methodological relativism" concerns
the notion of design hierarchies.
RecalI that a design hierarchy derives its rationale from a list of
threats. The place of a given design in the hierarchy corresponds to the
number of threats on a list for which it successfully controls. Thus, if
lists of threats are relative, so are the design hierarchies. This means
that, contrary to the common view, no research design is generally supe-
rior or inferior to other designs; a "true" experiment could be occasion-
ally less interpretable (i.e., less valid) than a "quasi" experiment, for
example, a correlational design or even a "lowly" case study (Campbell,
1975). For example, Lewin's (1943) experimental studies of effecting
change in American food habits were open to a variety of alternative
interpretations (Bennett, 1955). In contrast, Wrightsman's (1960) correla-
tional finding that groups with intermediate (versus high or low) vari-
ances in initial anxiety exhibited greatest convergence following group
discussion does not readily adrnit a direction of causality reinterpreta-
tion, because anxiety was assessed prior to convergence.
In fact, most developmental studies are correlational in nature; this
hardly rnakes them less valid than many experimental studies. In devel-
opmental research, the direction of causality issue does not pose a se-
rious interpretative threat. Based on our commonsense knowledge, it
does not seem very reasonable to assurne that a child first learns to walk
and then to crawl, for example, or acquires the concept of quantity and
then matures. More generally, then, the issue is not whether a given
research design is of particular type (e.g., whether it is a true or a quasi
experiment), but whether it controls for currently plausible interpretations
of the data. In principle, the number of threats or alternative interpreta-
tions one could invoke in reference to any body of data is vast. A re-
search design that controls for specific threats out of a fixed list could be
controlling for the "wrong" or implausible ones, while failing to control
for those of real concern. There can be an experiment that invites several
alternative interpretations of the data, and a single-shot case study that
invites none. For example, a physician may be qttite confident that a
patient was poisoned, say, byantimony, and be unable to engender any
alternative explanations for the observed symptoms.
To conclude, the ruling methodology in a field of study represents
the factors that a community of scientific practitioners deerns relevant to
this particu1ar domain. These factors are not of an absolute or invariant
interest. They rnight appear of lesser relevance to alternate scientific
communities concerned with different topics, and/or working at a differ-
ent time or place. Thus, it is irnportant to gain mastery over currently
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 235

accepted research methods and to be aware of currently accepted threats

while, at the same time, keeping in mind their relative status. Other-
wise, we might be overly dogmatic in our approach to research and, in
addition, be unable to communicate our notions effectively to different
scientific audiences. In this vein, Cronbach (1982, p. 109) recently im-
plied that the methodology of inference may weIl be moved "from the
realm of pure logic into the realm of rhetoric, or persuasion."
To sum up then, official methodological concepts in psychology
conjoin deductive logic to various content categories of inference. In this
connection, note that some methodological precepts (e.g., threats to
internal validity as "maturation") pertain to rather localized interpreta-
tive concerns, whereas others (e.g., the statistical null hypothesis) per-
tain to much more general such problems, common to a wide spectmm
of social science investigations.
Some methodological precepts may be so general in fact, as to be
relevant to scientific inquiry as such. Consider the official canons con-
cerned with the in-principle reproducibility of reports, or the use of
standardized instruments to give some assurance that independent ob-
servers would have produced a similar protocol. To be sure, those, too,
refer to contents of alternative interpretations (e.g., that the data are
biased by the specific observer) that need to be mIed out by deducing
their implications and confronting them by negative findings. Thus, in
some ca ses, the contents of the alternative interpretations may be so
general as to qualify properly as integral to the scientific method as such.

BEYOND OFFICIAL METHODOLOGIES: SCIENCE AS A SOCIAL COGNITIVE

ENTERPRISE

In its amalgamation of implicational logic with specific content

categories of inference, official scientific methodology remarkably re-
sembles areas of social cognitive psychology, like attribution, dis so-
nance, or attitude change, in which contents of lay inference have been
similarly linked with the principle of logical validation.
In neither case may deduction necessarily afford objectively valid
inferences: Any deduction is as valid as the premises from which it
departs. Premises, in turn, are beliefs concerning relations of an if-then
type between concepts. Such beliefs can be highly labile and replaceable
by alternative beliefs. Thus, the present notion of "logical" derivation, in
scientific and lay inference, is a descriptive term solely, not meant to
imply the objective validity of the inferences reached.
Furthermore, official methodology constrains in specific ways the
types of deductions that are warranted or the types of evidence (e.g.,
empirical observations) from which conclusions are legitimately deriv-
236 CHAPTER 10

able. According to the present approach, however, the LEGO metaphor,

in particular, numerous additional types of evidence could serve as the
bases for inferences. In science, too, derivations are often made from
officially unrecognized premises. Thus, the logical element enters into
unofficial methodology even as it does into official methodology. This
point is discussed in a subsequent section.
However, according to present theory, logical derivation alone is
insufficient to account for the inference process. In addition, it is neces-
sary to specify the various cognitive and motivational factors affecting
the generation of alternate hypotheses (Popper, 1972) and the search for
relevant information. Finally, in order for an idea to become part of
scientific knowledge, it must be accepted not just by an individual scientist
but by a whole community of scientists. This suggests that beyond the
principle of logical derivation, a complete explication of scientific in-
ference must specify the psychological and sociopsychological processes
that affect the acceptability of conclusions to individuals and groups;
these are the psychological and sociopsychological processes that gov-
ern persuasion (Weimer, 1979). And if scientific methodology as a field
of knowledge addresses factors that facilitate one's efficacy as a scientist,
it ought to include cognitive, motivational, and social principles that
affect one's ability to communicate effectively with one's scientific au-
dience. This, too, is part and parcel of unofficial scientific methodology.
In the sections that follow, unofficial scientific methodology is dis-
cussed in terms of constructs and propositions derived from the lay
epistemic theory. It is argued that notions like "evidential relativity,"
"category accessibility," or "epistemic motivations" may significantly
affect the understandability and acceptability of one' s scientific argu-
ments. In the next sections, I will discuss these matters, occasionally
drawing on examples from the history of Western science.

Evidential Relativity and the Epistemic Authority of Scientific Sources

Earlier, we have seen that what counts as compelling evidence is
relative to the perspective of a particular scientific audience. What con-
stitutes cogent evidence to one audience might be unacceptable evi-
dence to another audience, possibly concerned with alternative threats
to validity. But the notion of evidential relativity extends even further . It
suggests that empirical data are but one source of evidence on which one
may base one's scientific deductions.
This rests on the assumption, widely held in contemporary philoso-
phy of science, that empirical data are conceptual in nature; they repre-
sent mere hypotheses, albeit ones held with considerable degrees of
confidence (for a review of arguments concerning the conceptual nature
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 237

of "facts," see, e.g., Weimer, 1979b). But there exist other firmly held
hypotheses (or beliefs) to which scientists pervasively subscribe. These
constitute important bases of scientific derivations beside empirical data.
A prominent example of such beliefs concerns the source from which a
given scientific conclusion emanates.
A prestigious source, high in "epistemic authority," affords the de-
duction that the conclusion is credible, not so a source low in "epistemic
authority." Indeed, we are all aware at some level that findings pub-
lished in a highly reputable journal carry greater weight than the same
findings published in a less-respected outlet. Similarly, innovative work
has much more chance to gain inroads into the scientific community if
sponsored by established scientific "opinion makers."
The influence of a communication source may affect not only the
impact of a scientist's final conclusion, but also the tendency to follow
the mediating arguments. For instance, research by Norman (1975) sug-
gests that when a message is ascribed to an expert source, provision of
supportive arguments increases persuasive impact. However, when a
message is ascribed to a physically attractive source, provision of sup-
portive arguments has no effect on persuasion. Possibly, then, the qual-
ity of one' s evidence makes greater difference if the report appears in a
prestigious outlet and makes less of a difference if published in a less
prestigious, albeit otherwise attractive (e.g., popular) outlet. Numerous
additional findings on source and message factors (for a review, see
Petty and Cacioppo, 1981) are likely to have direct relevance to the
process of scientific communication.
The importance of source effects is illustrated via several historical
examples of scientific accomplishments. Darwin, for example, published
his theory of natural selection in the Journal of the Linnaean Society in 1859
under the sponsorship of Sir Charles Lyal, a famed geologist. Generally,
Darwin seemed aware of the importance of support from revered "epis-
temic authorities": "he consciously set out to persuade important well-
placed scientists both before he published his Origin of Species and
after. Darwin was interested primarily in the verdict of a dozen or so
men of science, if they came around the rest would follow" (Darwin,
1899, p. 91).
Einstein's general theory of relativity gained popularity in England
following the endorsement by Eddington and the legitimation of quan-
tum optics is generally ascribed to its explicit support by the venerated
Henri Poincare.
Often, however, scientists may fall to secure support for their inno-
vative ideas from prestigious figures in the scientific establishment. A
prominent example is Freud, who is believed to have been prevented for
a lang time from publishing papers on psychoanalysis or presenting
238 CHAPTER 10

them at important scientific meetings. Freud's way of coping with the

rejection was to found his own scientific support network; his own
unique scientific establishment complete with publication outlets. The
psychoanalytic cirdes in Vienna, Berlin, and Zurich united to found the
International Psychoanalytic Society. From then on this allowed psycho-
analysts to train physicians in their methods and to provide supportive
atmosphere for their endeavors at professional meetings.
In 1911, a journal was founded, the Zentralblatt für Psychoanalyse,
with Freud as editor. As the psychoanalytic movement gained force,
further publishing outlets were established applying psychoanalytic no-
tions to art, literature, and philosophy besides psychiatry.
The foregoing examples are consistent with the idea that scientific
audiences often draw their condusions from extra-empirical bases. For
example, one particularly important base seems to be the epistemic au-
thority of the source that supports the condusions at issue. The accep-
tance of some scientific innovations seems to have been facilitated be-
cause of an endorsement by established authorities. In other cases (like
that of psychoanalysis), scientists may have constructed their own epis-
temic authority by founding a cirde of enthusiastic followers. Securing
epistemic authority for one' s proposals or condusions thus seems to
constitute part and parcel of unofficial scientific methodology.

Construct Accessibility

According to the "availability heuristic" (Tversky and Kahneman,

1974), people often make judgments of category frequency (e.g., of out-
standing historical figures being female) in accordance with the ease
with which instances of the category come to mind. In turn, this may
depend on the distinctiveness or imaginability of those instances. The
availability heuristic is typically considered to be a feature of lay in-
ference that sets it apart from scientific inference and is often responsible
for inferential errors that lay persons (but presumably not trained scien-
tists) may commit. But let us take a second look. First, note that the
availability heuristic is a special case of a much broader notion of con-
struct accessibility (Higgins and King, 1981). Specifically, the availability
heuristic is restricted to a particular content dass of judgments, notably
judgments of category frequency (Kruglanski and Ajzen, 1983). By com-
parison, the notion of construct accessibility pertains to judgments of all
possible contents (e.g., judgments of causality, of dass membership, of
identity, and of controllability). Second, it seems that the inferences of
scientists must depend on considerations of accessibility as much as
those of lay persons: The rendition of any inference presupposes first
having a hypothesis, and the generation of any hypothesis depends on
the accessibility of ideas to the knower.
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 239

Einstein, for example, lived in an intellectual ambience in which the

notion of relativity was quite commonplace. On the one hand, he was
strongly influenced by Mach, who emphasized the relativity of space
and time. On the other hand, the political and cultural atmosphere of
Einstein' s milieu contained as a central tenet the relativity of societal
laws. Einstein and his friends discussed this frequently, concluding that
laws are relative to the lawmakers' perspective. Applied to society this
meant that the middle-class perspective was not uniquely essential and
could be replaced by an alternative social order based on different
premises.
Ready accessibility of Machian-Marxist notions of relativity might
have led Einstein to import them to the realm of physics. This might
have resulted in his repudiation of Newton's system and his assertion
that events simultaneous under one system of coordinates need not be
simultaneous under another system of coordinates. In this sense, Ein-
stein' s revolutionary ideas appear to draw on highly accessible notions
frequently "primed" by his intellectual environment.
The notion of accessibility has important implications for one' s
efficacy as a scientist. To the extent that efficacy is measured (among
others) by the impact one has on one's scientific community, one's con-
tribution to the accepted body of knowlege in one' s field may depend on
one' s ability to render one' s ideas continually accessible to one' speers.
The difficulty of this task is compounded by the information explosion
characterizing contemporary science. According to Psychological Abstracts
(1985), in psychology alone an average of 30,000 articles are published
every single year. Even if one corrects for the fact that psychology is
highly specialized, and the actual number of papers relevant to onei s
interests is much lower than that, drawing attention to one's ideas or
enhancing their accessibility to one' s scientific audience could pose a
problem.
The effective communication of scientific ideas can be facilitated in
any number of ways. It might help, for instance, to have a bunch of
talented and motivated graduate students who publish a lot and in good
places, thus amplifying the accessibility of their supervisor's notions,
assuming these had any impact on the students to begin with. Kurt
Lewin' s considerable impact on social psychology is generally ascribed,
at least in part, to the high grade of co-workers he managed to attract
and who went on to become major leaders in the field for decades to
follow (Marrow, 1977).
Some scientists may attempt to render their research noticeable by
choosing problems that are currently in vogue and thus are accessible to
many of their peers. This might lead to an overly quick reactivity to
shifting scientific trends, possibly at the expense of a profound and
long-Iasting commitment to one's chosen topic of study.
240 CHAPTER 10

Epistemic Motivations and Scientific Impact: The Effects of Needs for

Specific Closure

Annals of Western science abound with examples of scientists who

seemed ready to accept and/or formulate conceptions compatible with
their interests and values and to reject the incompatible ones. For exam-
pIe, the staunch opposition that the scientific establishment put up
against Freud' s theorizing is typically accounted for in terms of his em-
phasis on sexuality, which threatened traditional middle-elass values in
the domains of religion, morals, family, and political authority.
To take a different example, Einstein's predilection for revolution-
ary theorizing could be at least somewhat related to his general bias in
favor of "revolutions." Einstein belonged to a Marxist-Machian intellec-
tual cirele in Zurich called the "Olympia Academy." Members of that
group harbored considerable sympathy for revolutionary activities, and
one member, Einstein's elose friend, Friedrich Albert, took matters so
far as to assassinate the Austrian prime minister in 1917. Einstein was
sympathetic to Bolshevik leaders like Lenin and Trotsky, whom he de-
scribed as "guardians of humanity and its innovators." It is just possible
that Einstein' s appreciation for social or political revolutions motivated
hirn to formulate the "revolutionary" theory of relativity that he finally
proposed.
Gregor ]ohann Mendel, the recognized founder of modem genetics,
represents another example of theorizing congruent with a specific-
elosure need. Mendel was a devout Augustinian monk and from 1868
until his death in 1884 the head of the St. Thomas monastery in Brünn. It
is noteworthy that the Augustinian order places particular emphasis on
the notion of original sin; that is, the doctrine that human being's sinful
nature is transmitted from parent to child as a result of Adam's fall. It is
conceivable that, threatened by an evolutionary theory that stressed the
dominant role of the environment, Mendel was motivated to counter
with an alternative whereby traits of character, like the original sin, exist
latently or manifestly irrespective of environmental influences.
The specific-elosure construct suggests that a scientific work stands
a greater chance of gaining acceptance if it is perceived as serving the
predominant needs of the audience. It follows that a scientist's research
would be well received or rewarded if it communicates information
"which the colleagues deern to be useful in the pursuit of their own
studies (Mulkay, 1979, pp. 69-70). Of course, different groups of col-
leagues might find different types of information useful. Theoretically
minded colleagues might value the conceptual implications of a scien-
tific analysis, empirically minded colleagues might be more interested in
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 241

empirical implications, and application-minded colleagues, in POS-

sibilities the analysis extends for applied work. 4
Finally, the specific-closure construct underscores the connection
between one' s interpersonal relations with one' s fellow scientists and
their openness to one' s ideas. Good interpersonal relations involving
liking and trust might enhance the motivation to see a colleague' s point
of view and to find fault with his or her critics. As one example of this
process, significant aspects of scientific communication and persuasion
may be carried by informal social gatherings or over drinks at conven-
tions. In a situation in which camaraderie prevails and the defenses are
down, receptivity to new ideas might be paramount, and the perception
of threat at its lowest. This implies, for example, that good graduate
training would provide students and faculty with ample opportunities
for informal exchanges in a convivial atmosphere, and that considera-
tions of congeniality are not irrelevant to academic hiring decisions.
This is not to say, that, in all instances, personallikes and dislikes
dictate one' s receptivity to a colleague' s work in a simple linear fashion.
The constellation of one' s preferred closure needs might be complex and
involve values and assumptions with contrasting implications for recep-
tivity. For instance, a scientist might be leary of potential bias and "bend
over backward" in assessing a friend's work. Alternatively, the scientist
might impute to a colleague an exploitative motivation and out of " reac-
tance" (Brehm, 1966) reject this person's ideas. Neither example contra-
dicts the general supposition that specific closure needs may bias scien-
tific receptivity. What the examples do suggest is that interpersonal
relations among scientists might engage a variety of contrasting needs.
The momentary response to a colleague' s ideas might depend on the
most salient of the several needs.
The foregoing discussion also suggests that an interpretation of
scientific history in terms of villainous defenders of the status quo cal-
lously fighting courageous martyrs (like Copernicus, Galileo, or Freud)
could be exaggerated. Rather than reflecting an ugly conspiracy of pro-
establishment scientists, who privately recognize the truth yet cynically
refuse to admit it in public, the opposition could often be quite genuine.
It could reflect the principle that the more an idea contradicts our fa-
vored conclusions, the less likely it is to be perceived as the truth and the

4Mulkay (1979, p. 82), for exampIe, notes that " ... the spectrometer became a dominant
research tool Iargely because of its practical advantages: it could be used routinely to
provide precise measurements over a wide range of areas. Yet, once firmly established it
became a major resource for rejecting knowiedge claims which embodied different as-
sumptions." Thus, theoretical assumptions giving rise to a useful scientific instrument
couId be more acceptable to a community of researchers than assumptions devoid of
equally useful implications.
242 CHAPTER 10

more likely it is to be avoided or denounced as a falsehood that warrants

an unwavering rejection.

Nonspecific Closure Effects

Scientists typically require a guiding conceptual closure, or para-
digm, to carry out their work (Conant, 1947; Kuhn, 1962). Such a para-
digm involves basic assumptions about a domain of study and about the
nature and methods of research. In other words, a paradigm constitutes
a body of knowledge scientists subscribe to. Without it, they can make
no claims to expertise and thus no claims to a unique contribution to
society. Small wonder, then, that when a ruling paradigm is crumbling
away and the scientists' sense of closure is slipping, they experience
considerable frustration and upset.
This is vividly documented in Kuhn's (1962) seminal work on scien-
tific revolutions. For instance, he described Copernicus's bitter com-
plaint concerning the inconsistency of astronomers in his day, so much
so that "they could not even explain or observe the constant length of
the seasonal year" (Kuhn, 1957, p. 138). According to Copernicus, "it is
as though an artist were to gather the hands, feet, head and other
members for his images from diverse models . . . and since they in no
way match each other, the result would be monster rather than man" (p.
138).
Similarly, Einstein, referring to the confused state of physics at his
time, wrote that "it was as if the ground had been pulled from under
one, with no firm foundations to be seen anywhere upon which one
could have built" (Einstein, 1949, p. 45). And Wolfgang Pauli wrote to a
friend, months before the epoch-making paper by Heisenberg made its
appearance:
At the moment physics is again terribly confused. In any case, it is too
difficult for me and I wish I had been a movie comedian or something of the
sort and had never heard of physics. (Kuhn, 1962, p. 84)

Given the upset occasioned among scientists by confusion and epis-

temic void, it is not surprising that a paradigm may be clung to until an
alternative is suggested in its place. According to Kuhn (1962),
once it has achieved the status of paradigm a scientific theory is dec1ared
invalid only if an alternative candidate is available to take its
place .... Though they may begin to lose faith [scientists] do not renounce
the paradigm that has led them into crisis. They do not, that is, treat anoma-
lies as counter examples though in the vocabulary of philosophy of science
that is what they are .... The decision to reject one paradigm is always
simultaneously the decision to accept another. (p. 84)
Also, see Conant (1947).
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 243

As Kuhn's analysis implies, there seems to be more to the accep-

tance and rejection of scientific ideas than the deductive testing of theo-
ries. In particular, scientists' need for (nonspecific) c10sure seems to play
a significant role in the shift from one framework to the next. At the
same time, different scientific communities might vary in the degree of
their tolerance for loosely structured formulations. For instance, on the
whole, the sociobehavioral sciences seem more tolerant of ambiguity
than the natural sciences. As one index of such a difference, the use of
formal, precisely defined terms, in particular, mathematical ones, as a
vehic1e of explanation is much more obligatory in physics and chemis-
try, for example, than in psychology or sociology.
Even within the same general field, the degree to which different
subgroups of scientists may crave c10sure and order is likely to vary in
accordance with their task content. Applied scientists as a group might
have astronger need for c10sure than basic researchers. This is because
applied scientists are required to act on their conceptions. They make
recommendations, offer practical advice, and consult in actual decisions
on which a great deal might hinge. To do so, they might need to speak
with authority and confidence; this presupposes a crystallized concep-
tion on which to draw, that is, a c1ear-cut, structured set of notions. In
contrast, basic researchers face no comparable demand for c1ear-cut con-
c1usions and it is quite acceptable for them to offer tentative suggestions
for exploration in further research. As a consequence, their need for
c10sure might be considerably lower than that of their practically minded
colleagues but their need to avoid c10sure might be higher.

Avoidance of Nonspecific Closure

Contemporary interpreters of science highlight the role of criticism
in effecting scientific change, that is, in promoting a shift from one
scientific framework to another (Lakatos and Musgrave, 1970; Weimer,
1979b). Criticism, in turn, consists of evidence and/or arguments incon-
sistent with the originally held position. And as noted earlier, inconsis-
tency, once recognized, casts doubt on a belief system and induces a
sense of its possible invalidity.
Fear of an invalidity based need to avoid c10sure may underlie the
major mechanism Kuhn (1962) posits for explaining scientists' readiness
to abandon a ruling paradigm. Presumably, this is mediated by the
accumulation of "anomalies," that is, findings inconsistent with the par-
adigm. It is easy to appreciate the atmosphere of intense crisis that may
envelop a scientific community on becoming aware of a mounting pile of
such anomalies. Scientists are expected to possess knowledge, and
knowledge is presumed to be true or valid. However much they may
244 CHAPTER 10

crave dosure, scientists are compelled to let go of their conc;epts if these

are seriously suspect of invalidity. Thus, fear of invalidity could act as a
wedge, driving open whatever cracks might be showing in a scientific
edifice.
Just as the need for dosure may vary across scientific contexts, so
maya fear-of-invalidity-induced need to avoid dosure. Two considera-
tions seem relevant here: (1) the extent to which an invalidity is readily
demonstrable and (2) the extent to which it is perceived as costly. De-
monstrability may have to do with audience expertise. For instance, a
professor who lectures on his or her views to a dass of undergraduates,
or an applied psychologist who offers suggestions to a lay dienteIe,
might not consider it very likely that someone in the audience would
discover a fundamental inconsistency in the presentation. By contrast, a
scientist who delivers a speech before a group of expert peers might
view such an eventuality as considerably more likely. In present terms,
the latter situation may invite a higher need for openness than the
former.
Demonstrability might also have to do with darity of the assess-
ment criteria. An opinion poller who wrongly predicts an outcome of an
impending election would be found out quite readily. By contrast, a
geophysicist, predicting a cooling of the earth in the next millenium,
might not be compromised as vividly. Consequently, the opinion poller
might experience a higher need to avoid dosure than the geophysicist.
Finally, fear of invalidity may have to do with the perceived costli-
ness of a scientific error. A scientist who makes a recomendation for a
practical decision that involves considerable human and material re-
sources might experience a relatively high fear of invalidity, so might a
candidate for an academic position about to deliver a job colloquium. In
contrast, a researcher presenting his research findings to a group of
mildly curious laypersons might experience a relatively low fear of inval-
idity and, hence, a relatively low need to avoid dosure.
To condude, I argued in preceding pages that the process whereby
scientific knowledge forms is influenced by several factors typically ne-
glected in official methodological discussions. Official methodological
concepts (the controlled experiment, inferential statistics, notions of va-
lidity and reliability) highlight the logical element of scientific inference
yet neglect certain dasses of derivations (e.g., from premises about
source authority) that often occur in science. In addition, official meth-
odological discussions neglect the factor of mental accessibility, as weIl
as several epistemic motivations (needs for specific and nonspecific
dosure and for the avoidance of dosure) that probably exert an impor-
tant influence on scientific condusions.
As major philosophers of science have argued (e.g., Popper, 1959,
LAY EPISTEMIC UNDERPINNINGS OF RESEARCH 245

1963, 1972), the deductive element of our inference process may not
ensure the validity of scientific conclusions in any objective sense of the
term. In this sense at least, logic is not "superior" to the various so-
ciocognitive influences on scientific inference enumerated above. More
importantly from the present epistemic perspective, it may not be useful
to consider it in disjunction from those alternative psychological influ-
ences. To reiterate, logical, cognitive, and motivational factors are pre-
sumed to be involved in any instance of inference, including any in-
stance of scientific inference.

SUMMARY

In this final chapter, I argued that scientific knowledge is acquired

via the process outlined in the lay epistemic theory. From this perspec-
tive, I distinguished between official and unoffical scientific meth-
odologies. Official methodology was defined as a set of explicitly stated
mIes and norms for conducting research. It stresses the logical element
in our reasoning, whereby we "prove" our knowledge to ourselves and
others. Unofficial methodology has to do with the sodal nature of sci-
ence.1t involves derivations from officially unrecognized premises (e.g.,
concerning the authority of scientific sources) as well as cognitive (e.g.,
accessibility) and motivational factors involved in all judgments. Both
types of methodology pertain to distinct aspects of the general epistemic
process; hence, both affect the ways in which scientific knowledge is
formed.
Thus, the same epistemic process seems to be shared in common by
lay and scientific modes of inference. The present volume has attempted
to substantiate this notion through a review of the theory and research
in diverse content domains of inference studied by sodal cognitive
psychologists.
References

Abelson, R. P. (1959). Models of resolution of belief dilemmas. Journal of Conflict Resolution,

3,343-352.
Abelson, R. P. (1968). Psychological implication. In R. P. Abelson, E. Aronson, W. J.
McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds.), Theories of
cognitive consistency: A source book. Chicago: Rand McNally.
Abelson, R. P. (1983). Whatever became of consistency theory? Personality and Social Psy-
chology Bulletin, 9, 37-54.
Abelson, R. P., & Rosenberg, M. J. (1958). Symbolic psycho-logic: A model of attitudinal
cognition. Behavioral Science, 3, 1-13.
Abelson, R. P., Aronson, E., McGuire, W. J., Newcomb, T. M., Rosenberg, M. J., &
Tannenbaum, P. H. (Eds.). (1968). Theories o[ cognitive consistency: A sourcebook.
Chicago: Rand McNally.
Abelson, R. P., Kinder, D. R., Peters, M. D., & Fiske, S. T. (1982). Affective and semantic
components in political person perception. Journal of Personality and Social Psychology,
42, 619-630.
Abramson, L. Y., & Alloy, L. B. (1981). Depression, nondepression and cognitive illusion:
Reply to Schwartz. Journal o[ Experimental Psychology: General, 110, 436-477.
Abramson, L. Y., Seligman, M. E. P., & Teasedale, J. D. (1978). Leamed helplessness in
humans: Critique and reformulation. Journal o[ Abnormal Psychology, 87, 48-74.
Adler, A. (1964). The practice imd theory of individual psychology (P. Radin, Trans.). London:
Routledge & Kegan PauI.
Ajzen, I. (1977). Intuitive theories of events and the effects of base rate information on
prediction. Journal o[ Personality and Social Psychology, 35, 303-314.
Ajzen, I., & Fishbein, M. (1975). A Bayesian analysis of attribution processes. Psychological
Bulletin, 82, 261-277.
Ajzen, 1., & Fishbein, M. (1977). Attitude-behavior relations: A theoretical analysis and
review of empirical research. Psychological Bulletin, 84, 888-918.
Albert (1977). Temporal comparison theory. Psychological Review, 84, 485-503.
Allen, V. L., & Levine, J. M. (1968). Social support, dissent and conformity. Sociometry, 31,
138-149.
Alloy, L. B., & Abramson, L. Y. (1979). Judgment of contingency in depressed and non-
depressed students: Sadder but wiser? Journal of Experimental Psychology: General, 168,
441-485.
Alloy, L. B., & Abramson, L. Y. (1988). Depressive realism: Four theoretical perspectives.
In L. B. Alloy (Ed.), Cognitive processes in depression. New York: Guilford.
Allport, G. W. (1935). Attitudes. In C. Murchison (Ed.) Handbook of social psychology (Vol.
2). Worcester, MA: CIark University Press.

247
248 REFERENCES

Allyn, J., & Festinger, L. (1961). The effectiveness of unanticipated persuasive communica-
tions. Journal o[ Abnormal and Sodal Psychology, 62, 35-40.
Anastasi, A. (1969). Psychological Testing (3rd Edition). New York: Macmillan.
Anderson, C. A. (1983). The causal structure of situations: The generation of plausible
causal attributions as a function of type of event situation. Journal o[ Experimental Social
Psychology, 19, 185-203.
Anderson, N. H., & Hubert, S. (1963). Effects of concornitant verbal recall on order effects
in personality impression formation. Journal o[ Verbal Learning and Verbal Behavior, 2,
379-391.
Argyle, M. (1957). Social pressure in public and private situations. Journal o[ Abnormal Social
Psychology, 54, 172-175.
Arkin, R. M., & Baumgardner, A. H. (1985). Self-handicapping. In J. H. Harvey & G.
Weary (Eds.) Attribution: Basic issues and applications. New York: Acadernic Press.
Arkin, R. M., Appleman, A. J., & Burger, J. M. (1980). Social anxiety, self-presentation,
and the self-serving bias in causal attribution. Journal o[ Personality and Social Psychol-
ogy, 38, 23-25.
Aronson, E. (1968). Dissonance theory: Progress and problems. In R. P. Abelson, E.
Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum
(Eds.), Theories o[ cognitive consistency: A sourcebook. Chicago: Rand McNally.
Aronson, E., & Carlsrnith, J. M. (1962). Performance expectancy as a deterrninant of actual
performance. Journal o[ Abnormal and Sodal Psychology, 65, 178-182.
Aronson, E., Brewer, M., & Carlsrnith, J. M. (1985). Experimentation in social psychology.
In G. Lindzey & E. Aronson (Eds.). The handbook o[ social psychology (3rd ed.). Reading,
MA: Addison-Wesley.
Arrowood, A. J., & Friend, R. (1969). Other factors determining the choice of a comparison
other. Journal 0/ Experimental Sodal Psychology, 5, 233-239.
Asch, S. E. (1946). Forrning impressions of personality. Journal 0/ Abnormal and Sodal
Psychology, 41, 258-290.
Asch, S. E. (1952). Sodal psychology. Englewood Cliffs, NJ: Prentice-Hall.
Asch, S. (1956). Studies of independence and conformity: I. A rninority of one against a
unanimous majority. Psychological Monographs, 70 (9).
Atkinson, J. W. & Birch, D. (1970). The dynamics o[ action. New York: Wiley.
Back, K. W. (1951). Influence through social communication. Journal o[ Abnormal and Social
Psychology, 46, 9-23.
Bagozzi, R. P. (1981). Attitudes, intentions and behavior: A test of some key hypotheses.
Journal o[ Personality and Social Psychology, 41, 607-627.
Bandura, A. (1969). Principles o[ behavior modification. New York: Holt Rinehart & Winston.
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psycho-
logical Review, 84, 191-215.
Bargh, J. A. (1984). Automatie and conscious processing of social information. In R. S.
Wyer & T. K. Srull (Eds.), Handbook o[ social cognition (Vol. 3, pp. 1-43). Hillsdale, NJ:
Erlbaum.
Bar-Hillel, M. (1973). On the subjective probability of compound events. Organizational
Behavior and Human Performance, 9, 396-406.
Baron, R. (1988). An ecological framework for establishing a dual mode theory of social
knowing. In D. Bar-Tal and A. W. Kruglanski (Eds.), The social psychology o[ knowledge.
Cambridge, England: Cambridge University Press.
Bar-Tal, D. (in press). Group beliefs. New York: Springer-Verlag.
Bar-Tal, D. (1986). The Masada syndrome: a case of central belief. In N. Milgram (Ed.),
Stress and coping in time o[ war. New York: Brunner/Mazel.
Bar-Tal, D. & Bar-Tal, Y. (1988). New perspective on social psychology. In D. Bar-Tal & A.
REFERENCES 249

W. Kruglanski (Eds.), The social psychology of knowledge. Cambridge, England: Cam-

bridge University Press.
Bar-Tal, 0., & Kruglanski, A W. (Eds.). (1988). The social psychology of knowledge. Cam-
bridge, England: Cambridge University Press.
Bar-Tal, 0., Kruglanski, A W., & Klar, Y. (1989). Conflict termination: An epistemological
analysis of international cases. Political Psychology (in press).
Baumgardner, A. H., Lake, E. A, & Arkin, R. M. (1984). Claiming depression as a self-
handicap. Unpublished manuscript, University of Missouri, Columbia.
Bechtold, J., Zanna, M. P., & Naccarato, M. (1985). Unpublished data, Waterloo Univer-
sity.
Beck, A. T. (1963). Thinking and depression. Archives of General Psychiatry, 9, 324-333.
Beck, A T. (1976). Cognitive therapy and the emotional disorders. Madison, CT: International
Universities Press.
Beck, A. T., Rush, A J., Shaw, B. F., & Emery, G. (1979). Cognitive therapy of depression: A
treatment manual. New York: Guilford.
Bem, D. J. (1965). An experimental analysis of self-persuasion. Journal of Experimental Social
Psychology, 1, 199-218.
Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive dissonance
phenomena. Psychological Review, 74, 183-200.
Bem, D. J. (1972). Self-perception theory. In L. Berkowitz (Ed.), Advances in experimental
social psychology (Vol. 6). New York: Academic Press.
Bennett, E. B. (1955). Discussion, decision, commmitment and consensus in "group deci-
sions." Human Relations, 8, 251-274.
Berglas, S., & Jones, E. E. (1978). Drug choice as a self-handicapping strategy in response
to noncontingent success. Journal of Personality and Social Psychology, 30, 405-517.
Berkowitz, L. (Ed.). (1978). Cognitive theories in social psychology. New York: Academic
Press.
Berlyne, D. E. (1960). Conflict, arousal and curiosity. New York: McGraw-Hill.
Berscheid, E., Graziano, W., Monson, T., & Dermer, M. (1976). Outcome dependency:
Attention, attribution and attraction. Journal of Personality and Social Psychology, 34978-
989.
Boas, F. (1911). The mind of primitive man (rev. ed.) New York: The Free Press.
Borgida, E., & Brekke, N. (1981). The base rate falIacy in attribution and prediction. InJ. H.
Harvey, W. Ickes, & R. F. Kidd (Eds.), New directions in attribution research (Vol. 3).
Hillsdale, NJ: Erlbaum.
Bourne, L. E., Dominowski, R. L., & Loftus, E. F. (1979). Cognitive process. Englewood
Cliffs, NJ: Prentice-Hall.
Bowers, K. S. (1987). Revisioning the unconscious. Canadian Psychology, 28, 93-105.
Braude, J. M. (1965). Complete speaker's and toastmaster's library. Englewood Cliffs, NJ:
Prentice-Hall.
Breckler, S. J. (1984). Empirical validation of affect, behavior, and cognition as distinct
components of attitude. Journal of Personality and Social Psychology, 47, 1191-1205.
Brehm, J. W. (1966). A theory of psychological reactance. New York: Academic Press.
Brehm, J. W. & Cohen, A. R. (1962). Explorations in Cognitive Dissonance, New York: Wiley.
Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions of the elderly: Stereotypes as
prototypes. Journal of Personality and Social Psychology, 41, 656-670.
Brickman, P., & Bulman, R. J. (1977). Pleasure and pain in social comparison. In J. M. Suls
& R. L. Miller (Eds.), Social comparison processes. Washington, DC: Hemisphere Pub-
lishing Corp.
Brock, T. C. (1965). Communicator-recipient similarity and decision change. Journal of
Personality and Social Psychology, 1, 650-654.
250 REFERENCES

Broadbeck, M. (1956). The role of small groups in mediating the effects of propaganda.
Journal of Abnormal Sodal Psychology, 52, 166-170.
Bruner, J. S. (1951). Personality dynamics and the process of perceivng. In R. Blake & G. V.
Ramsey (Eds.), Perception-an approach to personality. New York: Ronald Press.
Bruner, J. S. (1973). Beyond the information given: Studies in the psychology of knowing. New
York: W. W. Norton.
Brunson, B. 1., & Matthews, K. A. (1981). Type A coronary-prone behavior pattern and
reactions to uncontrollable stress: An analysis of performance strategies, affect, and
attributions during failure. Journal of Personality and Social Psychology, 40, 906-918.
Campbell, D. T. (1957). Factors relevant to the validity of experiments in social settings.
Psychological Bulletin, 54, 297-312.
Campbell, D. T. (1969). Prospective: Artifact and contro!. In R. RosenthaI and R. L.
Rosnow (Eds.), Artifact in Behavioral Research. New York: Academic Press.
Campbell, D. T. (1974). Evolutionary epistemology. In P. A. Schlipp (Ed.), The philosophy of
Kllrl Popper (Vo!. 14, I & 11: The library of living philosophers). LaSalle, 11: Open Court
Publishing Co.
Campell, D. T. (1975). Degrees of freedom and the case study. Comparative Political Studies,
2,178-293.
Campbell, J. (1986). Similarity and uniqueness: The effects of attribute type, relevance and
individual differences in self-esteem and depression. Journal of Personality and Social
Psychology, 50, 281-294.
Campbell, D. T. & Stanley, J. (1963). Experimental and quasi experimental designs for
research on teaching. In N. L. Gage (Ed.) Handbook of Research on Teaching. Chicago:
Rand McNally.
Cantor, N., & MischeI, W. (1977). Traits as prototypes: Effects on recognition memory.
Journal of Personality and Social Psychology, 35, 38-48.
Cartwright, 0., & Harary, R. (1956). Structural balance: A generalization of Heider's
theory. Psychological Review, 63, 277-293.
Chaiken, S. (1980). Heuristic versus systematic information processing and the use of
source versus message cues in persuasion. Journal of Personality and Social Psychology,
39, 752-766.
Chaiken, S., & Baldwin, M. W. (1981). Affective-cognitive consistency and the effect of
salient behavioral information on the seIf-perception of attitudes. Journal of Personality
and Social Psychology, 41, 1-12.
Chaiken, S., & Stangor, C. (1987). Attitudes and attitude change. Annual Review of Psychol-
ogy, 38, 575-630.
Chapman, I. J., & Chapman, J. P. (1959). Atmosphere effect reexamined. Journal of Experi-
mental Psychology, 58, 220-222.
Cialdini, R. B., Petty, R. E. & Cacioppo, J. T. (1981). Attitude and attitude change. Annual
Review of Psychology, 32, 357-404.
Cialdini, R. B. & Petty, R. E. (1981). (Eds.), Anticipatory opinion effects, Hillsdale, NJ: 217-
237.
Cole, M., & Scribner, S. (1974). Culture and thought: A psychological introduction. New York:
Wiley.
Coleman, J. F., Blake, R. R., & Mouton, J. S. (1958). Task difficulty and conformity
pressures. Journal of Abnormal and Social Psychology, 57, 120-122.
Collins, B. E. & Hoyt, M. F. (1972). Personal responsibility for consequences: an integra-
tion and extension of the "forced compliance" literature. Journal of Experimental Social
Psychology, 8, 558-93.
Conant, J. B. (1967). On Understanding Science. New Haven: Yale University Press.
Cook, T. D. & Campbell, D. T. (1976). The design and conduct of quasi-experiments and
REFERENCES 251

true experiments in field settings. In M.D. Dunnette (Ed.), Handbook of Industrial and
Organizational Psychology. Chicago: Rand McNa!ly.
Cook, T. D. & Campbell, D. T. (1979). Quasi-Experimentation: Design and Analysis Issues for
Field Settings. Chicago: Rand McNally.
Cook, T. D., Gruder, C. L., Hennigan, K. M., & Flay, B. R. (1979). History of the sleeper
effect: Some logical pitfalls in accepting the null hypothesis. Psychological Bulletin, 86,
662-679.
Cooley, C. H. (1902). Human nature and the social order. New York: Scribner's.
Cooper, J., & Croyle, R. T. (1984). Attitudes and attitude change. Annual Review of Psychol-
ogy, 35, 395-426.
Cooper, J., & Fazio, R. H. (1984). A new look at dissonance theory. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 17). New York: Academic Press.
Cottrell, N. B., & Epley, S. W. (1977). Affiliation, socia! comparison, and socially mediated
stress reduction. In J. M. Suls & R. L. Miller (Eds.), Social comparison processes Wash-
ington, DC: Hemisphere Publishing Corp.
Crocker, J., Hannah, D. B., & Weber, R. (1983). Person memory and causal attributions.
Journal of Personality and Social Psychology, 44, 56-66.
Cronbach, L. J. (1955). Processes affecting scores on "understanding of others" and "as-
sumed similarity." Psychological Bulletin, 52, 177-193.
Cronbach, L. J. (1958). Proposals leading to analytic treatment of social perception scores.
In R. Tagiuri & L. Petullo (Eds.), Person perception and interpersonal behavior. Stanford,
CA: Stanford University Press.
Cronbach, L. J. (1982). Designing Evaluations of Educational and Social Programs. San Fran-
cisco: Jossey-Bass.
Cronbach, L. J., Gleser, C. G., Nanda, H. & Nagaswari, R. (1972). The dependability of
behavioral measurements: Theory of generalizability for scores and profiles. New York: Wiley.
Cronbach, L. J. & Meehl, P. E. (1955). Construct validity in psychological tests. Psychologi-
cal Bulletin, 52, 281-302.
Darwin, F. (1899). The life and letters of C. Darwin. New York: Appleton.
Day, R. A. (1979). How to write and publish a scientific paper. Philadelphia: 151 Press.
Davison, G. C. & Valins, S. (1969). Maintenance of self-attributed and drug-attributed
behavior change. Journal 01 Personality and Social Psychology, 11, 25-33.
Deaux, K. (1985). Sex and gender. Annual Review of Psychology, 36, 49-81.
Decarufel, A. c., & Insko, C. A. (1979). Balance and social comparison processes in the
attribution of attraction. Journal of Personality, 47, 432-448.
DeSoto, c., & Albrecht, F. (1968). Conceptual good figures. In R. P. Abelson, E. Aronson,
W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum (Eds.).
Theories of cognitive consistency: A sourcebook. Chicago: Rand McNally.
Deutsch, M., & Gerard, H. B. (1955). A study of normative and informational social
influences upon individual judgment. Journal of Abnormal and Social Psychology, 51,
629-636.
Devine, P. G. & Ostrom, T. M. (1988). Dimensional versus information-processing ap-
proaches to social knowledge: the case of inconsistency management. In D. Bar-Tal
and A. W. Kruglanski (Eds.) The social psychology of knowledge. Cambridge: Cambridge
University Press.
Diener, C. T., & Dweck, C. S. (1978). An analysis of learned helplessness: Continuous
changes in performance strategy, and achievement cognitions following failure. Jour-
nal of Personality and Social Psychology, 36, 451-462.
Doms, M. (1984). The minority influence effect: An alternative approach. In S. Moscovici &
W. Doise (Eds.), Current issues in European social psychology (pp. 1-31). Cambridge,
England: Cambridge University Press.
252 REFERENCES

Doms, M., & van Avermaet, E. (1980). Majority influence, minority influence and conver-
sion behavior: A replication. Journal of Experimental Sodal Psychology, 16, 283-292.
Dykman, B. M., Abramson, L. Y., Alloy, L. B., & Hartlage, S. (1987). Processing of ambigu-
ous and unambiguous feedback by depressed and nondepressed college students: Schematic
biases and their implications for depressive realism. Unpublished manuscript, University of
Wisconsin-Madison.
Eagly, A. H., Chaiken, S., & Wood, W. (1981). An attributional analysis of persuasion. In
J. H. Harvey, W. J. Ickes, & R F. Kidd (Eds.), Newdirections in attribution research. (Vol.
3) Hillsdale, NJ: Erlbaum.
Eagly, A. H., Wood, W., & Chaiken, S. (1978). Causal inferences about communicators
and their effect on opinion change. Journal of Personality and Sodal Psychology, 36, 424-
435.
Edwards, W. (1982). Conservatism in human information processing. In D. Kahneman, P.
Slovic, & A. Tversky (Eds.), Judgment under uncertainty: Heuristics and biases. Cam-
bridge, MA: Harvard University Press.
Einhorn, H. J., & Hogarth, R M. (1981). Behavioral decision theory: Processes of judg-
ment and choice. Annual Review of Psychology, 32, 53-88.
Einstein, A. (1949). Autobiographical note. In P. S. Schilpp (Ed.), Albert Einstein:
Philosopher-Sdentist. Evanston, Illinois:
Eiser, J. R (1980). Cognitive social psychology: A guidebook to theory and research. Maidenhead,
Berkshire, England: McGraw-Hill.
Eiser, J. R, & Pancer, S. M. (1979). Attitudinal effects of the use of evaluatively biased
language. European Journal of Social Psychology, 9, 39-47.
Eiser, J. R, & Ross, M. (1977). Partisan language, immediacy, and attitude change. Euro-
pean Journal of Social Psychology, 7, 477-488.
Ellis, A. (1977). The basic clinical theory of rational emotive therapy. In A. Ellis & R
Grieger (Eds.), Handbook of rational emotive therapy. New York: Springer.
Ellis, A. (1978). The rational emotive approach to counseling. In A. M. Bucks (Ed.), Theories
of counseling. New York: McGraw-Hill.
Ellis, S. (1984). The effect of self-epistemic authority on the relative efficacy of cognitive change
methods in different subject matters. Unpublished doctoral dissertation, Tel Aviv
University.
Enzle, M. E., & Schopflocher, D. (1978). Instigation of attribution processes by attribution
questions. Personality and Social Psychology Bulletin, 4, 595-599.
Epstein, S. and O'Brien, E. J. (1985). The person-situation debate in historical and current
perspective. Psychological Bulletin, 98, 513-537.
Erber, R, & Fiske, S. T. (1984). Outcome dependency and attention to inconsistent infor-
mation. Journal of Personality and Social Psychology, 47, 709-726.
Evans, J. T. St. B., & Lynch, J. S. (1973). Matching bias in the selection task. British Journal
of Psychology, 64, 391-397.
Fazio, R H. (1979). Motives for sodal comparison: The construction-validation distinction.
Journal of Personality and Social Psychology, 37, 1693-1698.
Fazio, R. H. (1988). On the power and functionality of attitudes: The role of attitude
accessibility. In A. R Pratkanis, S. J. Breckler and A. G. Greenwald (Eds.), Attitude
structure and function. Hillsdale, NJ: Erlbaum.
Fazio, RH., & Zanna, M. P. (1981). Direct experience and attitude-behavior consistency.
In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 14). New York:
Academic Press.
Fazio, R. H., Chen, J., McDonel, E. c., & Sherman, S. J. (1982). Attitude accessibility,
attitude-behavior consistency, and the strength of the object-evaluation association.
Journal of Experimental Social Psychology, 18, 339-357.
REFERENCES 253

Fazio, RH., Lenn, T. M., & Effrein, E. A. (1983-1984). Spontaneous attitude formation.
Social Cognition, 2, 217-234.
Fazio, R. H., Powell, M. c., & Herr, P. M. (1983). Toward a process model of the attitude-
behavior relation: Accessing one's attitude upon mere observation of the attitude
object. Journal of Personality and Social Psychology, 44, 723-735.
Fazio, RH., Sanbomatsu, D. M., Powell, M. c., & Kardes, F. R (1985). On the automatie
activation of attitudes. Unpublished manuscript, Indiana University.
Ferguson, T. J., & WeHs, C. 1. (1980). Priming of mediators in causal attribution. Journal of
Personality and Social Psychology, 38, 461-470.
Festinger, 1. (1950). Informal social communication. Psychological Review, 57, 271-282.
Festinger, 1. (1954). A theory of social comparison processes. Human Relations, 7, 117-140.
Festinger, 1. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University
Press.
Festinger, 1. (1964). Conflict, decision and dissonance. Stanford, CA: Stanford University
Press.
Festinger, 1., & Cadsmith, J. M. (1959). Cognitive consequences of forced compliance.
Journal of Abnormal and Social Psychology, 58, 203-210.
Festinger, 1., & Maccoby, N. (1964). On resistance to persuasive communication. Journal of
Abnormal and Social Psychology, 68, 359-366.
Festinger, 1., Schachter, S. and Back, K. (1950). Social pressures in informal groups. New
York: Harper.
Feyerabend, P. (1976). Against method. New York: Humanities Press.
Fincham, F. D. (1985). Attributions in close relationships. In J. H. Harvey & G. Weary
(Eds.), Attribution: Basic issues and applications. New York: Academic Press.
Fishbein, M., & Ajzen, I. (1975). Belief, attitude, intention and behavior: An introduction to
theory and research. Reading, MA: Addison-Wesley.
Fiske, S. T., & Newberg, S. 1. (1988). A continuum of impression formation from category-
based to individuating processes: Influence of information and motivation on atten-
tion and interpretations. In M. P. Zanna (Ed.), Advances in experimental social psychology
(Vol. 23). New York: Academic Press.
Fiske, S. T., & Taylor, S. E. (1984). Social cognition. Reading, MA: Addison-Wesley.
FlaveH, J. H. (1963). The developmental psychology o[ Jean Piaget. Princeton, NJ: Van
Nostrand.
Fong, G. T., Krantz, D. H., & Nisbett, RE. (1986). The effects of statistical training on
thinking about everyday problems. Cognitive Psychology, 18, 253-292.
Frank, J. D. (1973). Persuasion and Healing. Baltimore, MD: Johns Hopkins University Press.
Frankei, A., & Snyder, M. 1. (1978). Poor performance following unsolvable problems:
Leamed helplessness or egotism. Journal of Social Psychology, 36, 1415-1423.
Freedman, J. 1. (1968). How important is cognitive consistency? In R P. Abelson, E.
Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P. H. Tannenbaum
(Eds.), Theories of cognitive consistency: A sourcebook. Chicago: Rand McNally.
Frenkel-Brunswik, E. (1949). Intolerance of ambiguity as emotional and perceptual person-
ality variable. Journal of Personality, 18, 108-143.
Freund, T., Kruglanski, A. W., & Schpitzajzen, A. (1985). The freezing and unfreezing of
impressional primacy: Effects of the need for structure and the fear of invalidity.
Personality and Social Psychology Bulletin, 11, 479-487.
Frey, D. (1986). Selective exposure to information: A review of recent research. In 1.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19). New York: Aca-
demic Press.
Friend, R, & Gilbert, J. (1973). Threat and fear of negative evaluation as determinants of
lorus of social comparison. Journal of Personality, 41, 328-340.
254 REFERENCES

Funder, D. C. (1987). Errors and mistakes: Evaluating the accuracy of sodal judgment.
Psychological Bulletin, 101, 75-9l.
Gage, N. L., & Cronbach, L. J. (1955). Conceptual and methodologica! problems in inter-
personal perception. Psychological Review, 662, 411-422.
Gage, N. L., Leavitt, G. S., & Stone, G. C. (1956). The intermediary key in the analysis of
interpersonal perception. Psychological Bulletin, 53, 258-266.
Gergen, K. J. (1968). Personal consistency and the presentation of self. In C. Gordon & K.
J. Gergen (Eds.), The self in social interaction. New York: Wiley.
Gergen, K. J. (1977). Stability, change and chance in understanding human development.
In N. Datan & H. Reese (Eds.), Life-span developmental psychology: Dialectical perspectives
on experimental research. New York: Academic Press.
Gergen, K. J. (1982). Transformation in social knowledge. New York: Springer.
Gholson, B. & Barker, P. (1985). Kuhn, Lakatos and Laudan: Applications in the history of
physics and psychology, American Psychologist, 40, 755-769.
Gibbons, F. X., & Gaedert, W. P. (1984). Focus of attention and placebo utility. Journal of
Experimental Social Psychology, 20, 159-170.
Goethals, G. R. (1972). Consensus and modality in the attribution process: The role of
similarity and information. Journal of Personality and Social Psychology, 21, 84-92.
Goethals, G. R. (1986). Soda! comparison theory: Psychology from the lost and found.
Personality and Social Psychology Bulletin, 12, 261-278.
Goethals, G. R., & Darley, J. (1977). Sodal comparison theory: An attributional approach.
In J. M. Suls & R. L. Miller (Eds.), Social comparison processes. Washington, DC: Hemi-
sphere Publishing Corp.
Goethals, G. R., & Nelson, R. (1973). Similarity in the influence process: The belief-value
distinction. Journal of Personality and Social Psychology, 25, 117-122.
Goffman, E. (1959). The presentation of self in everyday life. New York: Doubleday.
Goodstadt, M. S. (1971). Helping and refusal to help: A refusal of balance and reactance
theories. Journal of Experimental Social Psychology, 7, 610-622.
Gotz-Marchand, B., Gotz, J., & Irle, M. (1974). Preference of dissonance modes as a
function of their order, familiarity and reversibility. European Journal of Social Psychol-
ogy, 4, 201-228.
Greenwald, A. G. (1968). Cognitive learning, cognitive response to persuasion, and atti-
tude change. In A. G. Greenwald, T. C. Brock, & T. M. Ostrom (Eds.), Psychological
foundations of attitudes. New York: Academic Press.
Greenwald, A. G., & Ronis, D. I. (1978). Twenty years of cognitive dissonance: Case study
in the evolution of a theory. Psychological Review, 85, 53-57.
Greenwald, A. G., Baumgardner, M. H., & Leippe, M. R. (1979). In search of reliable
persuasion affects: IlI. The sleeper effect is dead. Long live the sleeper effect! Unpublished
manuscript, Ohio State University.
Gregory, R. L. (1970). The intelligent eye. New York: McGraw-Hill.
Gregory, R. L. (1973). The confounded eye. In R. L. Gregory & E. H. Gombrich (Eds.),
Illusions in nature and art. London: Duckworth.
Griggs, R. A., & Cox, J. R. (1982). The elusive thematic-materials effect in Wason's selec-
tion task. British Journal of Psychology, 73, 07-42.
Gruder, C. L. (1971). Determinants of sodal comparison choices. Journal of Experimental
Social Psychology, 473-489.
Gruder, C. L. (1977). Choice of comparison persons in evaluating oneself. In J. M. Suls and
R. L. Miller (Eds.) Social comparison processes: Theoretical and empirical perspectives. Wash-
ington, DC: Hemisphere Publishing Corp., 21-4l.
Hakmiller, K. (1966). Threat as a determinant of downward comparison. Journal of Experi-
mental Social Psychology, Supplement 1, 32-39.
Hall, C. S., & Lindzey, G. (1978). Theories of personality. New York: Wiley.
REFERENCES 255

Hamilton, D. L. (1981). Organizational processes in impression formation. In E. T. Hig-

gins, C. P. Herman, & M. P. Zanna (Eds.). Social cognition. The Ontario Symposium
(Vol. 1). Hillsdale, NJ: Erlbaum.
Hamilton, D. L. (1988). The role of the attribution process in memory for sodal informa-
tion. In D. Bar-Tal & A. W. Kruglanski (Eds.), The social psychology of /cnowledge.
Cambridge, England: Cambridge University Press.
Hannah, T. E., Hannah, R, & Wattie, B. (1975). Arousal of psychological reactance as a
consequence of predicting an individual's behavior. Psychological Report, 37, 411-420.
Hardyck, J. A., & Kardush, M. (1968). A modest modish model for dissonance reduction.
In R P. Abelson. E. Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, & P.
H. Tannenbaum (Eds.), Theories of cognitive consistency: A sourcebook. Chicago: Rand
McNally.
Harvey, J. H., & Weary, G. (1984). Current issues in attribution theory and research.
Annual Review of Psychology, 35, 427-459.
Harvey, J. H., & Weary, G. (Eds.) (1985). Attribution: Basic issues and applications. New York:
Academic Press.
Harvey, J. H., Yarkin, K. L., Lightner, J. M., & Town, J. P. (1980). Unsolidted attributions
and recall of interpersonal events. Journal of Personality and Sodal Psychology, 38, 551-
558.
Hass, J. W., Bagley, G. S., & Rogers, R W. (1975). Coping with the energy crisis: Effects of
fear appeals upon attitudes toward energy consumption. Journal of Applied Psychology,
60,754-756.
Hastie, R (1980). Memory for behavioral information that confirms or contradicts a per-
sonality impression. In R. Hastie, T. M. Ostrom, E. B. Ebbesen, R S. Wyer, D. L.
Hamilton & D. E. Carlston (Eds.), Person memory: The cognitive basis of sodal perception.
Hillsdale, NJ: Erlbaum.
Hastie, R (1984). Causes and effects of causal attribution. Journal of Personality and Sodal
Psychology, 44-56.
Hastie, R., & Kumar, P. A. (1979). Person memory: Personality traits as organizing prin-
dples in memory for behavior. Journal of Personality and Social Psychology, 37,
25-38.
Hastie, R., & Rasinsky, K. A. (1988). The concept of accuracy in sodal judgment. In D. Bar-
Tal & A. W. Kruglanski (Eds.), The social psychology of knowledge. Cambridge, England:
Cambridge University Press.
Hastorf, A., & Cantril, H. (1954). They saw agame: A case study. Journal of Personality and
Social Psychology, 49, 129.
Heider, F. (1946). Attitudes and cognitive organization. Journal of Psychology, 21, 107-112,
Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.
Heller, J. F. (1972). Attribution theory: Self and other attributions as a determinant 0/ attitude
change. Unpublished doctoral dissertation, University of Iowa.
Henle, M. (1962). On the relation between logic and thinking. Psychological Review, 69, 366-
378.
Higgins, E. T., & Bargh, J. A. (1987). Sodal cognition and sodal perception. Annual Review
of Psychology, 38369-425.
Higgins, E. T., & King, G. A. (1981). Accessibility of sodal constructs: Information-
processing consequences of individual and contextual variability. In N. Cantor & J. F.
Kihlstrom (Eds.), Personality, cognition and social interaction. Hillsdale, NJ: Erlbaum.
Higgins, E. T., & Lurie, L. (1983). Context, categorization, and memory: The "change of
standard" effect. Cognitive psychology, 15, 525-547.
Higgins, E. T., & Stangor, C. (1988). Context-driven sodal judgment and memory. In D.
Bar-Tal & A. Kruglanski (Eds.), The social psychologyofknowledge. Cambridge, England:
Cambridge University press.
 256 REFERENCES

Higgins, E. T., King, G. A, & Marvin, G. H. (1982). Individual construct accessibility and
subjective impressions and recall. Journal of Personality and Social Psychology, 43, 35-47.
Higgins, E. T., Bargh, J. A, & Lombardi, W. (1985). The nature of priming effects on
categorization. Journal of Experimental Psychology: Learning, Memory and Cognition, 11,
59-69.
Higgins, E. T., Strauman, T., & Klein, R. (1985). Standards and the process of self-
evaluation: Multiple affects from multiple stages. In R. M. Sorrentino & E. T. Higgins
(Eds.), Motivation and cognition: Foundations of social behavior. New York: Guilford.
Himmelfarb, S., & Eagly, A H. (Eds.). (1974). Readings in attitude change. New York: Wiley.
Hofstede, G. (1980). Culture's consequences: International differences in work-related values.
Beverly Hills, CA: Sage.
Hovland, C. 1., Lumsdaine, A A, & Sheffield, F. D. (1949). Experiments on mass communica-
tion. Princeton, NJ: Princeton University Press.
Hovland, C. 1., Janis, I. L., & Kelley, H. H. (1953). Communication and persuasion. New
Haven: Yale University Press.
Hovland, C. I. and Mandell, W. (1942). An experimental comparison of conclusion draw-
ing by the communicator and by the audience. Journal of Abnormal and Social Psychol-
ogy, 4, 581-588.
Humphries, c., Carver, C. S., & Neumann, P. G. (1983). Cognitive characteristics of the
Type A coronary-prone behavior pattern. Journal of Personality and Social Psychology, 44,
177-187.
Insko, C. A (1967). Theories of attitude change. New York: Appleton-Century-Crofts.
Insko, C. A (1984). Balance theory, the Jordan paradigm, and the Wiest tetrahedron. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 19). New York: Aca-
demic Press.
Irle, M. (1975). Lehrbuch der Sozialpsychologie. Göttingen, Federal Republic of Germany:
Hogrefe.
Irle, M., & Montmann, V. (1978). Die Theorie der kognitiven Dissonanz: Eine Resümee
ihrer theoretischen Entwicklung und empirischen Ergebnisse 1957-1976. In M. Irle &
V. Montmann (Eds.), L. Festinger, Theorie der kognitiven Dissonanz. Bezn: Huber.
James, W. (1890). The principles of psychology. New York: Henry Holt.
James, W. (1907). Pragmatism. New York: Longmans-Green.
James, W. (1909). The meaning of truth. New York: Longmans-Green.
Janis, I. (1972). Victims of groupthink. Boston: Houghton Mifflin.
Janis, I. L., & Feshbach, S. (1953). Effects of fear arousing communications. Journal of
Abnormal and Social Psychology, 48, 78-92.
Janis, I. L., & Gilmore, J. B. (1965). The influence of incentive conditions on the success of
role playing in modifying attitudes. Journal of Personality and Social Psychology, 1, 17-27.
Janis, I. L., & TerwiIliger, R. F. (1962). An experimental study of psychological resistances
to fear-arousing communications. Journal of Abnormal and Social Psychology, 65, 403-410.
Jepson, c., Krantz, D. H., & Nisbett, R. E. (1983). Inductive reasoning: Competence or
skill? Behavioral and Brain Sciences, 6, 94-501.
Johnson, T. J., Feigenbaum, R., & Weiby, M. (1964). Some determinants and conse-
quences of the teacher's perception of causation. Journal of Educational Psychology, 55,
237-246.
Johnson-Laird, P. N., & Wason, P. C. (1970). A theoretical analysis of insight into a
reasoning task. Cognitive Psychology, 1, 134-148.
Jones, E. E. (1985). Major developments in socia! psychology during the past five decades.
In G. Lindzey & E. Aronson (Eds.), The handbookof social psychology (3rd ed., Vol. 1, pp.
47-107). New York: Random House.
Jones, E. E., & Berglas, S. (1978). Control of attributions aboutthe self through self-
REFERENCES 257

handicapping strategies: The appeal of alcohol and the role 'öf underaehievement.
Personality anti Social Psychology Bulletin, 4, 200-206.
Jones, E. E., & Davis, K. E. (1965). From aets to dispositions: The attribution proeess in
person pereeption. In L. Berkowitz (Ed.), Advances in experimental sodal psychology
(Vol. 2). New York: Academic Press.
Jones, E. E., & Gerard, H. B. (1967). Foundations o[ social psychology. New York: Wiley.
Jones, E. E., & MeGillis, D. (1976). Correspondent inferences and the attribution cube: A
eomparative reappraisal. In J. H. Harvey, W. J. Ickes, & R. F. Kidd (Eds.), New
directions in attribution research (Vol. 1). Hillsdale, NJ: Erlbaum.
Jones, E. E., & Pitman, T. S. (1982). Toward a general theory of strategie self-presentation.
In J. Suls (Ed.), Psychological perspectives on the self. Hillsdale, Nf Erlbaum.
Kahneman, D. (1973). Attention and effort. Englewood Cliffs, NJ: Prentice-Hall.
Kahneman, D. & Tversky, A. (1973). On the psyehology of prediction: Psychological Review,
80,237-51.
Kahneman, D., Slovic, P., & Tversky, A. (Eds.). (1982). judgment untier uncertainty: Heuris-
tics and biases. New York: Cambridge University Press.
Kanfer, F. H. (1975). Self-management methods. In F. H. Kanfer & A. P. Goldstein (Eds.),
Helping people change. Elmsford, NY: Pergamon.
Kassin, S. M. (1979). Consensus information, prediction, and causal attribution: A review
of the literature and issues. journal o[ Personality and Social Psychology, 37, 1966-1981.
Katz, D. (1960). The functional approach to the study of attitudes. Public Opinion Quarterly,
24, 163-204.
Kelley, H. H. (1967). Attribution theory in social psychology. In D. Levine (Ed.), Nebraska
symposium on motivation. Omaha: University of Nebraska Press.
Kelley, H. H. (1971). Attribution in social interaction. In E. E. Jones, D: E. Kanause, H. H.
Kelley, R. E. Nisbett, S. Valins, & B. Weiner (Eds.), Attribution: Perceiving the causes o[
behavior. Morristown, NJ: General Learning Press.
Kelley, H. H., & Lamb, T. W. (1957). Certainty of judgment and resistance to social
influence. journal o[ Abnormal and Social Psychology, 55, 137-139.
Kelley, H. H., & Michela, J. L. (1980). Attribution theory and research. Annual Review o[
Psychology, 31, 457-501.
KeIman, H. C. (1958). Compliance, identification and internalization: Three processes of
attitude change. journal o[ Conflict Resolution, 2, 51-60.
Kenny, D. A., & Albright, L. (1988). Accuracy in interpersonal perception: A social rela-
tions analysis. Psychological Bulletin, 102, 390-403.
Kiesler, C. A., & Sakamura, J. A. (1966). A test of a model for commitment. journal o[
Personality and Social Psychology, 3, 349-353.
Kiesler, C. A., Collins, B. E., & Miller, N. (1969). Attitude change: A critical analysis o[
theoretical approaches. New York: Wiley.
Klar, Y., & Pol, I. (1980). [Imbalance effects as function of agreement implication]. Un-
published data, Tel Aviv University.
Klar, Y., Bar-Tal, D., & Kruglanski, A. W. (1988). Conflict as a cognitive schema: An
epistemological approach. In W. Stroebe, A. W. Kruglanski, D. Bar-Tal, & M.
Hewstone (Eds.), Social psychology o[ intergroup anti international conflict. New York:
Springer-Verlag.
Kolditz, T. A., & Arkin, R. M. (1982). An impression management interpretation of the
self-handicapping strategy. journal o[ Personality and Social Psychology, 43, 492-502.
Kopel, S., & Arkowitz, H. (1975). The role of attribution and self-perception in behavior
change. Implications for behavior therapy. Genetic Psychology Monographs, 92, 178-212.
Kruglanski, A. W. (1975). The endogenous-exogenous partition in attribution theory.
Psychological Review, 82, 387-406.
258 REFERENCES

Kruglanski, A. W. (1980). Lay epistemo-10gic-Process and contents: Another look at

attribution theory. Psychological Review, 87, 70-87
Kruglanski, A. W. (1987). The taming of the unconscious: If you can't beat it, use it.
Canadian Psychology, 28, 110-112.
Kruglanski, A. W. (1988). Knowledge as a sodal psychological construct. In D. Bar-Tal &
A. W. Kruglanski (Eds.), The social psychology ofknowledge. Cambridge, England: Cam-
bridge University Press.
Kruglanski, A. W. (1989). Motivations for judging and knowing: Implications for causal
attributions. In E. T. Higgins & R. M. Sorrentino (Eds.), Handbook of motivation and
cognition: Foundations of social behavior (Vol. 2). New York: Guilford.
Kruglanski, A. W., & Ajzen, I. (1983). Bias and error in human judgment. European Journal
of Social Psychology, 13, 1-44.
Kruglanski, A. W., & Freund, T. (1983). The freezing and unfreezing of lay inferences:
Effects on impressional primacy, ethnic stereotyping, and numerical anchoring, Jour-
nal of Experimental Social Psychology, 19, 448-468.
Kruglanski, A. W., & Jaffe, Y. (1988). Curing by knowing: A lay epistemic approach to
cognitive therapy. In L. Abramson (Ed.), Social personality perspectives in clinical in-
ference. New York: Guilford.
Kruglanski, A. W. & Klar, Y. (1987). A view from a bridge: Synthesizing the consistency
and attribution paradigms from a lay epistemic perspective. European Journal of Social
Psychology, 17, 211-241.
Kruglanski, A. W., & Klar, Y. (1985). Knowing what to do: On the epistemology of actions.
In J. KuhI & J. Beckman (Eds.), Action control: From cognition to behavior. Heidelberg:
Springer-Verlag.
Kruglanski, A. W. & Kroy, M. (1976). Outcome validity in experimental research: A
reconceptualization. Representative research in social psychology, 7, 166-176.
Kruglanski, A. W., & Mayseless, O. (1987). Motivational effects in the socia1 comparison of
opinions. Journal o[ Personality and Social Psychology, 53, 834-853.
Kruglanski, A. W., & Mayseless, O. (1988). Contextual effects in hypothesis testing: The
role of competing alternatives and epistemic motivations. Social Cognition, 6, 1-20.
Kruglanski, A. W., Hamei, I. A., Maides, S. A., & Schwartz, J. M. (1978). Attribution
theory as a special case oflay epistemology. InJ. H. Harvey, W. J. Ickes, & R. F. Kidd
(Eds.), New directions in attribution research, (Vol. 2). Hillsdale, NJ: Erlbaum.
Kruglanski, A. W., Schwartz, J. M., Maides, S. A., & Harne!, I. A. (1978). Covariation,
discounting and augmentation: Toward a clarification of attributional principles. Jour-
nal of Personality, 46, 176-180.
Kruglanski, A. W., Friedland, N., & Farkash, E. (1984). Laypersons' sensitivity to statisti-
ca1 information: The case of high perceived applicability. Journal of Personality and Social
Psychology, 46, 503-518.
Kuhn, T. S. (1957). The coppernican revolution. Cambridge, MA: Harvard University Press.
Kuhn, T. S. (1962). The structure of scientific revolutions. Chicago: Chicago University Press.
Kuhn, T. S. (1970). Logic of discovery or psychology of research? In I. Lakatos & A.
Musgrave (Eds.), Criticism and the growth of knowledge. Cambridge, England: Cam-
bridge University Press.
Kunda, Z. (1986). Motivated inference: Self-serving generation and evaluation of causal theories.
Unpublished manuscript, Princeton University.
Lakatos, I. (1970). Falsification and the methodology of scientific research programs. In I.
Lakatos & A. Musgrave (Eds.), Criticism and the growth of knowledge. Cambridge, En-
gland: Cambridge University Press.
Lakatos, I. (1974). Popper on demarcation and induction. In P. A. Schilpp (Ed.), The
philosophy of Sir Karl Popper. La SaUe, IL: Open Court.
REFERENCES 259

Lakatos, 1., & Musgrave, A. (1970). (Eds.), Criticism and the growth of knowledge. Cambridge,
England: Cambridge University Press.
Lana, R. E. (1969). Pretest sensitization. In R. RosenthaI and R. L. Rosnow (Eds.), Artifact
in Behavioral Research. New York: Academic Press.
Laudan, L. (1977). Progress and its problems. Berkeley: University of California Press.
Lazarus, A. A. (1971). Behavior therapy and beyond, New York: McGraw-Hill.
Lazarus, R. S. (1980). Cognitive behavior therapy as psychodynamics revisited. In M. J.
Mahoney (Ed.), Psychotherapy process: Current issues and future directions. New York:
Plenum Press.
Lerner, M. J. (1980). The belief in a just world: A fundamental delusion. New York: Plenum
Press.
Leventha!, H. (1970). Findings and theory in the study of fear communications. In L.
Berkowitz (Ed.), Advances in experimental social psychology (Vol. 5). New York: Academ-
ic Press.
Levine, M. A. (1975). A cognitive theory of learning. Hillsdale, NJ: Erlbaum.
Levy-Bruh!, L. (1910). How natives think (Trans.). New York: Washington Square Press.
Lewin, K. (1938). The conceptual representation and measurement of psychological forces. Dur-
harn, NC: Duke University Press.
Lewin, K. (1943). Forces behind food habits and methods of change, Bulletin of the National
Research Council, 108, 35-67.
Lewin, K. (1951). Field theory in social sciences. New York. Harper.
Lewinsohn, P. M., Mische!, W., Chapin, W., & Barton, R. (1980). Sodal competence and
depression: The role of illusory self-perception. Journal of Abnormal Psychology, 89, 203-
212.
Lord, C. G., Lepper, M. R., & Preston, E. (1984). Considering the opposite: A corrective
strategy for sodal judgment. Journal of Personality and Social Psychology, 47, 1231-
1243.
Luchins, A. S. (1942). Mechanization in problem solving: The effects of Einstellung. InJ. F.
Dashiell (Ed.), Psychological Monographs, 54, (WhoIe No. 248).
Luria, A. R. (1971). Towards the problem of the historical nature of psychological pro-
cesses. International Journal of Psychology, 6, 259-272.
Lydon, J., Zanna, M. P., & Ross, M. (1985). Persistence and selective recaU: More evidence that
behavior recaU produces commitment to attitude change. Unpublished manuscript, Univer-
sity of Waterloo.
Maass, A., & Clark, R. D. III. (1984). Hidden impact of minorities: Fifteen years of minority
influence research. Psychological Bulletin, 95, 428-450.
Maass, A., Clark, R. D. III, & Haberkorn, G. (1982). The effects of differential ascribed
category membership and norms on minority influence. European Journal 0/ Social
Psychology, 13, 45-55.
Mahoney, M. (1985). Psychotherapy and human change processes. In M. J. Mahoney & A.
Freeman (Eds.). Cognition and psychology. New York: Plenum Press.
Mahoney, M., & Mahoney, K. (1976). Permanent weight control. New York: W. W. Norton.
Markus, H. (1977). Self schemata and processing of information about the self. Journal of
Personality and Social Psychology, 35, 63-78.
Markus, H., Crane, M., Bernstein, 5., & Siladi, M. (1982). Self-schemas and gender.
Journal of Personality and Social Psychology, 38, 231-248.
Marks, G. (1984). Thinking one's abilities are unique and one's opinions are common.
Personality and Social Psychology Bulletin, 10, 203-208.
Marrow, A." (1977). The practical theorist. The life and work of Kurt Lewin. New York:
Teachers College Press.
Mayseless, 0., & Kruglanski, A. W. (1987a). What makes you so sure? Effects of epistemic
260 REFERENCES

motivations on judgmental confidence. Organizational Behavior and Human Decision

Processes, 39, 162-183.
Mayseless, 0., & Kruglanski, A. W. (1987b). Accuracy of estimates in the sodal com-
parison of abilities. Journal o[ Experimental Social Psychology, 23, 217-229.
McArthur, L. Z. (1972). The how and what of why: Some determinants and consequences
of causal attribution. Journal o[ Personality and Social Psychology, 22, 171-193.
McArthur, L. Z., & Baron, R M. (1983). Toward an ecological theory of sodal perception.
Psychological Review, 90, 215-238.
McArthur, L. Z. and Ginsburg, E. (1981). Causal attributions to salient stimuli: an inves-
tigation of visual fixation mediators. Personality and Social Psychology Bulletin, 7, 547-
553.
McGuire, W. J. (1960a). Cognitive consistency and attitude change. Journal o[ Abnormal and
Social Psychology, 60, 345-353.
McGuire, W. J. (1960b). A syllogistic analysis of cognitive relationships. In C. I. Hovland &
M. J. Rosenberg (Eds.), Attitude organization and change. New Haven: Yale University
Press.
McGuire, W. J. (1960c). Direct and indirect persuasive effects of dissonance produdng
messages. Journal o[ Abnormal and Social Psychology, 60, 354-358b.
McGuire, W. J. & Papageorgis, D. (1961). Journal o[ Abnormal and Social Psychology, 62, 327-
337.
Meichenbaum, D. (1977a). Dr. EIlis, please stand up. The Counseling Psychologist, 7,
43-44.
Meichenbaum, D. (1977b). Cognitive-behavior modification. New York: Plenum Press.
Merton, R K. (1973). The sociology o[ science: Theoretical and empirical investigations. Chicago,
IL: University of Chicago Press.
Mettee, D. R, & Smith, G. (1977). Sodal comparison and interpersonal attraction: The case
for dissimilarity. In J. M. Suls & R. L. Miller (Eds.), Social comparison processes. Wash-
ington, DC: Hemisphere Publishing Corp.
Mewborn, C. R, & Rogers, R W. (1979). Effects of threatening and reassuring compo-
nents of fear appeals on physiological and verbal measures of emotion and attitudes.
Journal o[ Experimental Social Psychology, 15, 242-253.
Miller, C. T. (1984). Self-schemas, gender and sodal comparison: A c1arification of the
related attributes hypothesis. Journal o[ Personality and Social Psychology, 46, 1222-1229.
Miller, D. T. (1976). Ego involvement and attributions for success and failure. Journal o[
Personality and Social Psychology, 34, 901-906.
MilIer, D. T., & Ross, M. (1975). Self-serving biases in the attribution of causality: Fact or
fiction? Psychological Bulletin, 82, 213-225.
MilIer, N., Maruyama, G., Beaber, R. J., & Valone, K. (1976). Speed of speech and persua-
sion. Journal o[ Persona/ity and Social Psychology, 34, 615-625.
Miller, D. T., Norman, S. A., & Wright, E. (1978). Distortion in person perception as a
consequence of the need for effective control. Journal o[ Persona/ity and Social Psychol-
ogy, 38, 598-607.
MilIs, J. (1968). Interest in supporting and discrepant information. In R. P. Abelson, E.
Aronson, W. J. McGuire, T. M. Newcomb, M. J. Rosenberg, and P. H. Tannenbaum
(Eds.), Theories o[ cognitive consistency: A sourcebook. Skokie, IL: Rand McNally.
MilIs, J., & Jellison, J. M. (1968). Effect on opinion change of similarity between the
communicator and the audience he addressed. Journal o[ Personality and Social Psychol-
ogy, 9, 153-156.
MilIs, J. and O'Neal, E. (1971). Antidpated choice, attention and the halo effect. Psycho-
nomic Science, 22, 231-233.
Mische!, W. (1979). On the interface of cognition and personality: Beyond the person-
situation debate. American Psychologist, 34, 740-754.
REFERENCES 261

Mitroff, I. I. (1974). The subjective side of science. New York: Elsevier.

Moore, B. S., Sherrod, D. R., Liu, T. J. and Underwood, B. (1979). The dispositional shift
in attribution over time. Journal o[ Experimental Sodal Psychology, 15, 553-569.
Morgan, J. J. B., & Morton, J. T. (1944). The distortion of syllogistic reasoning produced by
personal convictions. Journal o[ Sodal Psychology, 20, 39-54.
Moscovici, S. (1985). Social influence and conformity. In G. Lindzey & E. Aronson (Eds.),
Handbook o[ social psychology (Vol. 2). New York: Random House.
Moscovici, S. (1980). Toward a theory of conversion behavior. In L. Berkowitz (Ed.),
Advances in experimental social psychology, (Vol. 13), 209-239. New York: Acadernic
Press.
Moscovici, S. & Lage, E. (1978). Studies in social influence IV: Minority influence in a
context of original judgments. European Journal o[ Social Psychology, 6, 365-380.
Moscovici, S. & Mugny, G. (1983). Minority influence. In P. B. Paulus (Ed.), Basic g~oup
processes, (pp. 41-64). New York: Springer-Verlag.
Mouton, J. S., Blake, R. R., & Olmstead, J. A. (1956). The relationship between frequency
of yielding and the disclosure of personal identity. Journal o[ Personality, 24, 339-347.
Murray, E. J., & Jacobson, L. I. (1978). Cognition and leaming in traditional and behavioral
therapy. In S. L. Garfield & A. E. Bergin (Eds.), Handbook o[ psychotherapy and behavior
change: An empirical analysis.
Mulkay, M. J. (1979). Science and the sodology o[ knowledge. London: Allen and Unwin.
Muzfir, M. (1985). Unpublished data. Tel Aviv University.
Neely, J. H. (1977). Semantic priming and retrieval from lexical memory: Roles of inhibi-
tionless spreading activation and lirnited capacity attention. Journal o[ Experimental
Psychology: General, 106, 225-254.
Nelson, R. E., & Craighead, W. E. (1977). Selective recall of positive and negative feed-
back, self-control behaviors, and depression. Journal o[ Abnormal Psychology, 86, 379-
388.
Nemeth, C. J. (1986). Differential contributions of majority and minority influence. Psycho-
logical Review, 93, 10-20.
Nemeth, C. J., Swedlund, M., & Kanki, G. (1974). Patteming of the minority's responses
and their influence on the majority. European Journal o[ Social Psychology, 4, 53-64.
Neuberg, S. L. and Fiske, S. T. (1987). Motivational influences on impression formation:
Outcome dependency, accuracy-driven attention and individuating processes. Journal
o[ Personality and Social Psychology, 53, 431-444.
Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Cliffs, NJ: Prentice-
Hall.
Newman, H. (1981). Communication within ongoing intimate relationships: An attribu-
tional perspective. Personality and Social Psychology Bulletin, 7, 59-70.
Nisbett, R. E., & Wilson, T. (1977). Telling more than we can know: Verbal reports on
mental processes. Psychological Review, 84, 231-259.
Nisbett, R. E., & Ross, L. (1980). Human in[erence: Strategies and shortcomings o[ sodal judg-
ment. New York: Prentice-Hall.
Nisbett, R. E., & Schachter, S. (1966). Cognitive manipulation of pain. Journal o[ Experimen-
tal Social Psychology, 2, 227-236.
Nisbett, R. E., Cheng, P. W., Fong, G. T., & Lehman, D. (1987). Teaching reasoning.
Unpublished manuscript, University of Michigan.
Nisbett, R. E., Krantz, D. H., Jepson, c., & Kunda, Z. (1983). The use of statistical
heuristics in everyday inductive reasoning. Psychological Review, 90, 339-363.
Norman, R. (1975). Affective-cognitive consistency, attitudes, conforrnity and behavior.
Journal o[ Personality and Sodal Psychology, 32, 83-9l.
Nunnally, J. and Bobsen, H. (1959). Variables conceming the willingness to receive com-
munications on mental health. Journal o[ Personality, 27, 38-46.
 262 REFERENCES

O'Neal, E. and Mills, J. (1971). The influence of anticipated choice on the halo effect.
Journal of Experimental Social Psychology, 5, 347-35l.
Orvis, B. R, Cunningham, J. D., & Kelley, H. H. (1975). A closer examination of causal
inference: The roles of consensus, distinctiveness and consistency information. Journal
of Personality and Social Psychology, 32, 605-616.
Orvis, B. R, Kelley, H. H., & Butler, D. (1976). Attributional conflict in young couples. In
J. H. Harvey, W. Ickes, & R F. Kidd (Eds.), New directions in attribution research, (Vol.
1). Hillsdale, NJ: Erlbaum.
Osgood, C. E., & Tannenbaum, P. H. (1955). The principle of congruity in the prediction
of attitude change. Psychological Review, 62, 42-55.
Paicheler, G. (1976). Norms and attitude change: I. Polarization and styles of behavior.
European Journal of Social Psychology, 6, 405-427.
Paicheler, G. (1977). Norms and attitude change: 11. The phenomenon of bipolarization.
European Journal of Social Psychology, 7. 5-14.
Paicheler, G. (1979). Polarization of attitudes in homogeneous and heterogeneous groups.
European Journal of Social Psychology, 9, 85-96.
Pepitone, A. (1966). Problems of consistency models. In S. Feldman (Ed.), Cognitive con-
sistency: Motivational antecedents and behavioral consequents. (257-297). New York: Aca-
demic Press.
Pepitone, A. (1968). The problem of motivation in consistency models. In R P. Abelson et
al. (Eds.), Theories of cognitive consistency: A sourcebook. Chicago: Rand McNally.
Peri, N., Kruglanski, A. W., & Zakai, D. (1986). Interactive effects of initial confidence and
epistemic motivations on the extent of informatinal search. Unpublished manuscript, Tel
Aviv University.
Pettigrew, T. F. (1967). Social evaluation theory: Convergences and applications. In
D. Levine (Ed.), Nebraska Symposium on Motivation. Lincoln: University of Nebraska
Press.
Petty, RE., & Cacioppo, J. T. (1981). Attitudes and persuasion: Classic and contemporary
approaches. Dubuque, IA: Brown.
Petty, RE., Ostrom, T. M., & Brock, T. C. (Eds.). (1981). Cognitive responses in persuasion.
Hillsdale, NJ: Erlbaum.
Petty, RE., Cacioppo, J. T., & Goldman, R (1981). Personal involvement as a determi-
nant of argument based persuasion. Journal of Personality and Social Psychology, 41, 847-
855.
Pittrnan, T. 5., & D' Agostino, P. R (1985). Motivation and attribution: The effects
of control deprivation on subsequent information processing. In J. H. Harvey &
G.,Weary(Eds.),Attribution:Basic issues and applications. New York: Academic Press.
Pittrnan, T. 5., Schemrer, F. W., & Wright, J. B. (1977). The effect of commitrnent on
information utilization in the attribution process. Personality and Social Psychology Bul-
letin, 3, 276-279.
Popper, K. R (1959). The logic of scientific discovery. New York: Harper. (Originally pub-
lished as Loqik der Forschung; Vienna, 1935.)
Popper, K. R (1963). Conjecturesand refutations. New York: Harper.
Popper, K. R (1972). Objective knowledge. Oxford: Oxford University Press.
Powell, M. c., & Fazio, R H. (1984). Attitude accessibility as a function of repeated
attitudinal expression. Personality and Social Psychology Bulletin, 10, 139-148.
Pryor, J. B., & Kriss, M. (1977). The cognitive dynamics of salience in the attribution
process. Journal of Personality and Social Psychology, 35, 49-55.
Radloff, R (1966). Socia! comparison and ability evaluation. Journal of Experimental Social
Psychology (Suppl. 1), 6-26.
Raimy, V. (1975). Misunderstandings of the self. San Francisco: Jossey-Bass.
Raush, H. 5., & Raush, C. L. (1968). The Half Way house movement: A search for sanity. New
York: Appleton Century Crofts.
REFERENCES 263

Raven, B. H. (1959). Social influence on opinions and the communication of related con-
tent. Journal of Abnormal and Social Psychology, 58, 119-128.
Reckman, R. F., & Goethals, G. R. (1973). Deviancy and group orientation as determinants
of group composition preferences. Sociometry, 36, 419-423.
Regan, D. T., & Fazio, R. H. (1977). On the consistency between attitudes and behavior:
Look to the method of attitude formation. Journal of Experimental Social Psychology, 13,
38-45.
Rhodewalt, F., & Davison, J. (1984). Self-handicapping and subsequent performance: The role of
outcome valence and attributional certainty. Unpublished manuscript, University
ofUtah.
Rogers, R. W. (1975). A protection motivation theory of fear appeals and attitude change.
Journal of Psychology, 91, 93-114.
Rogers, R. W., & Mewborn, R. (1976). Fear appeals and attitude change: Effects of a
threat's noxiousness, probabiIity of occurrence, and the efficacy of coping responses.
Journal of Personality and Social Psychology, 34, 54-6l.
Rokeach, M. (1960). The open and c/osed mind: Investigations into the nature of belief systems and
personality systems. New York: Basic Books.
Rose, T. L. (1981). Cognitive and dyadic processes in intergroup contact. In D. L. Hamilton
(Ed.), Cognitive processes in stereotyping and intergroup behavior. Hillsdale, NJ: Erlbaum.
Rosenberg, M. J. (1956). Cognitive structure and attitudinal effects. Journal of Abnormal and
Social Psychology, 63, 367-372.
Rosenberg, M. J. (1960). An analysis of affective-cognitive consistency: In M. J. Rosenberg
& C. I. Hovland (Eds.), Attitude organization and change: An analysis of consistency among
attitude components, (pp. 15-64) New Haven: Yale University press.
Rosenberg, M. J., & Hovland, C. I. (1960). Cognitive, affective, and behavioral compo-
nents of attitude. In M. J. Rosenberg & C. I. Hovland (Eds.), Attitude organization and
change: An analysis of consistency among attitude components. New Haven: Yale University
Press.
RosenthaI, T., & Bandura, A. (1978). Psychological modeling: Theory and practice. In S. L.
Garfield & A. E. Bergin (Eds.), Handbook of psychotherapy and behavioral change: An
empirical analysis. New York: WiIey.
Ross, L., Lepper, M. R, & Hubbard, M. (1975). Perseverance in self-perception and social
perception: Biased attributional processes in the debriefing paradigm. Journal 0/ Per-
sonality and Social Psychology, 32, 880-892.
Ross, L., Rodin, J., & Zimbardo, P. G. (1969). Toward an attribution therapy: The reduc-
tion of fear through induced cognitive-emotional misattribution. Journal of Personality
and Social Psychology, 12, 279-288.
Ross, M., & McFarland, C. (1988). Constructing the past: Biases in personal memories. In
D. Bar-Tal and A. W. Kruglanski (Eds.), The social psychology of knowledge. Cambridge,
England: Cambridge University Press.
Ross, M., & Sicoly, F. (1979). Egocentric biases in avaiIabiIity and attribution, Journal of
Personality and Social Psychology, 37, 322-336.
Rozen, E. (1970). Effects of reactance manipulation on compliance in an experimental situation.
Unpublished master's thesis, State University of New York at Stony Brook.
Runkel, P. S. and Peizer, D. B. (1968). The two-valued orientation of current equiIibrium
theory. Behavioral Science, 13, 56-65.
Sadler, 0., & Tesser, A. (1973). Some effects of salience and time upon interpersonal
hostility and attraction during social isolation. Sociometry, 36, 99-112.
Salancik, G. R., & Conway, M. (1975). Attitude inferences from salient and relevant
cognitive content about behavior. Journal of Personality and Social Psychology, 32, 829-
840.
Sarnoff, I. (1960). Psychoanalytic theory and social attitudes. Public Opinion Quarterly, 24,
251-279.
 264 REFERENCES

Sanbomatsu, D. M. and Fazio, R H. (1986). Unpublished data. Indiana University.

Saxe, L. and Fine, M. (1981). Social experiments: Methods for design and evaluation. Beverly
Hills, CA: Sage.
Schachter, S. (1951). Deviation, rejection and communication. Journal of Abnormal and Social
Psychology, 46, 190-207.
Schachter, S. (1959). The psychology of affiliation. Stanford, CA: Stanford University Press.
Schneider, W., & Schiffrin, R. M. (1977). ControlIed and automatie human information
processing: Detection, search and attention. Psychological Review, 84, 1-66.
Schneider, D. J., Hastorf, A. H., & Ellsworth, P. C. (1979). Person perception (2nd ed.).
Reading, MA: Addison-Wesley.
Schwartz, N., Frey, D., & Kumpf, M. (1980). Interactive effects of writing and reading a
persuasive essay on attitude change and selective exposure. Journal of Experimental
Social Psychology, 16, 1-18.
Scott, W. A. (1957). Attitude change through reward of verbal behavior. Journal of Abnormal
and Social Psychology, 55, 72-75.
Seils, S. B. (1936). The atmosphere effect: An experimental study of reasoning. Archives of
Psychology, 29, 3-72.
Sentis, K. P., & Bumstein, E. (1979). Remembering schema- consistent information.
Effects of a balance schema on recognition memory. Journal of Personality and Social
Psychology, 37, 22oo-221l.
Shaw, M. E. and Constanzo, P. R (1982). Theories of social psychology. New York: McGraw-
HilI.
Shelton, S. L. and Ackerman, J. M. (1976). Homework in counseling and psychotherapy.
Springfield, IL: Charles C Thomas.
Sherif, M. and Cantril, H. (1945). The psychology of attitudes I. Psychological Review,52,
295-319.
Sherif, M., & Hovland, C. I. (1961). Social judgment: Assimilation and contrast effects in
communication and attitude change. New Haven: Yale University Press.
Shiffrin, R M. (1975). Short-term storage: The basis for a memory system. In F. Restle, R
M. Shriffrin, N. J. Castellan, H. R. Lindman, & D. B. Pisoni (Eds.), Cognitive theory
(Vol. 1). Hillsdale, NJ: Erlbaum.
Shrauger, J. S., & Jones, S. C. (1968). Social validation and interpersonal evaluations.
Journal of Experimental Social Psychology, 4, 315-323.
Shrauger, J. S., & Osberg, T. M. (1980). The relationship of time investment and task
outcome to causal attributions and self-esteem. Journal of Personality, 48, 360-378.
Singer, J. E. (1966). Social comparison-progress and issues. Journal of Experimental Social
Psychology (Suppl. 1), 103-110.
Slovic, P., Fischoff, B., & Lichtenstein, S. (1977). Behavioral decision theory. Annual Review
of Psychology, 27, 1-39.
Srnith, E. R, & Miller, F. D. (1979). Salience and the cognitive mediation of attribution.
Journal of Personality and Social Psychology, 37, 2240-2252.
Smith, M. B., Bruner, J. S., & White, R W. (1956). Opinions and personality, New York:
Wiley.
Srnith, W. P. (1981). On the nature of the question in social comparison. In J. H. Harvey
(Ed.), Cognition, social behavior and the environment. Hillsdale, NJ: Erlbaum.
Snyder, C. R, Smith, T. W., Augelli, R W., & Ingram, R E. (1983). On the self-serving
function of social anxiety: Shyness as a self-handicapping strategy. Paper presented at the
annual meeting of the American Psychological Association, Anaheim, CA.
Snyder, M., & Swann, W. B. (1978). Hypothesis testing processes in social interaction.
Journal of Personality and Social Psychology, 36, 1202-1212.
Snyder, M., & Swann, W. B. (1976). When actions reflect attitudes: The politics of impres-
sion management. Journal of Personality and Social Psychology, 34, 1034-1042.
REFERENCES 265

Snyder, M. L., & Wicklund, R. A. (1981). Attribute ambiguity. InJ. H. Harvey, W. Ickes, &
R. F. Kidd (Eds.), New directions in attribution research, (Vol. 3) Hillsdale, NJ: Erlbaum.
Snyder, M., Tanke, E. D., & Berscheid, E. (1977). Sodal perception and interpersonal
behavior. On the self-fulfilling nature of soda! stereotypes. Journal o[ Personality and
Social Psychology, 35, 656-666.
Statts, A W. (1975). Social behaviorism. Homewood, IL: Dorsey Press.
Statts, C. K., & Statts, A W. (1958). Attitudes established by classical conditioning. Journal
o[ Abnormal and Social Psychology, 57, 37-40.
Storms, M. D., & Nisbett, R. E. (1970). Insomnia and the attribution process. Journal o[
Personality and Social Psychology, 16, 319-328.
Stroebe, W., Thompson, V. P., Insko, C. A., & Reisman, S. R. (1970). Balance and differ-
entiation in the evaluation of linked attitude objects. Journal o[ Personality and Social
Psychology, 16, 38-47.
Strong, S. R. (1970). Causal attribution in counseling and psychotherapy. Journal o[ Coun-
seling Psychology, 17, 388-399.
Strong, S. R. (1978). Sodal psychological approach to psychotherapy research. In S. L.
Garfield & A E. Bergin (Eds.), Handbook o[ psychotherapy and behavior change: An empiri-
cal analysis. New York: Wiley.
Suls, J., Gastorf, J., & Lawhon, J. (1978). Sodal comparison choices for evaluating a sex
and age related ability. Personality and Social Psychology Bulletin, 4, 102-105.
Swann, W. B. (1984). Quest for accuracy in person perception: A matter of pragmatics.
Psychological Review, 91, 457-477.
Taylor, S. E., & Fiske, S. T. (1975). Point of view and perception of causality. Journal o[
Personality and Social Psychology, 32, 439-445.
Taylor, S. E., & Fiske, S. T. (1978). Salience, attention and attribution: Top of the head
phenomena. In L. Berkowitz (Ed.), Advances in experimental social psychology (Vol. 11).
New York: Acadernic Press.
Taylor, S. E., & Thompson, S. C. (1982). Stalking the elusive "vividness" effect. Psychologi-
cal Review, 89, 155-181.
Tesser, A (1976). Thought and reality constraints as determinants of attitude polarization.
Journal o[ Research in Personality, 10, 183-194.
Tesser, A. (1978). Self-generated attitude change. In L. Berkowitz (Ed.), Advances in experi-
mental social psychology, Vol. 11, pp. 289-338. New York: Academic Press.
Tesser, A, & Campbell, J. (1982). Self-evaluation maintenance and the perception of
friends and strangers. Journal o[ Personality, 50, 261-279.
Tesser, A, & Conley, M. C. (1975). Some effects of time and thought on attitude polariza-
tion. Journal o[ Personality and Social Psychology, 31, 262-270.
Tesser, A, & Leone, C. (1977). Cognitive schemas and thought as determinants of attitude
change. Journal o[ Experimental Social Psychology, 13, 340-356.
Testa, T. J. (1974). Causal relationships and the acquisition of avoidance responses. Psycho-
logical Review, 81, 491-505.
Tetlock, P. E. (1980). Explaining teacher explanations of pupil performance: A self-
presentation interpretation. Social Psychology Quarterly, 43, 283-290.
Tetlock, P. E. (1985a). Accountability: A sodal check on the fundamental attribution error.
Social Psychology Quarterly, 68, 227-236.
Tetlock, P. E. (1985b). Accountability: The neglected soda! context of judgment and
choice. Research in Organizational Behavior, 7, 297-332.
Tetlock, P. E. and Levi, A (1982). Attribution bias: On the inconclusiveness of the
cognition-motivation debate. Journal o[ Experimental Social Psychology, 1982, 18, 68-88.
Thist!ethwaite, D. L., de Haan, H., & Kamenetzky, J. (1955). The effects of 'directive' and
'nondirective' communication procedures on attitudes. Journal o[ Abnormal and Social
Psychology, 51, 107-113.
266 REFERENCES

Thornton, D., & Arrowood, A J. (1966). SeIf-evaluation, self-enhancement, and the locus
of social comparison. Journal 0/ Experimental Social Psychology (Supp!. 1), 591-605.
Thurstone, L. L. & Chave, E. J. (1929). The measurement 0/ attitude. Chicago: University of
Chicago Press.
Trope, Y. (1975). Seeking information about one's own ability as a determinant of choice
among tasks. Journal 0/ Personality and Social Psychology, 32, 1004-1013.
Trope, Y. (1980). SeIf-assessment, seIf-enhancement and task preference. Journal 0/ Experi-
mental Social Psychology, 16, 116-123.
Trope, Y., & Bassok, M. (1983). Information-gathering strategies in hypothesis testing.
Journal 0/ Experimental Social Psychology, 19, 560-576.
Tversky, A, & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and biases.
Science, 185, 1124-113l.
Vallone, R P., Ross, L., & Lepper, M. R (1985). The hostile media phenomenon: Biased
perception and perceptions of media bias in coverage of the Beirut massacre. Journal of
Personality and Social Psychology, 49, 577-585.
Walster, E., Berscheid, E., & Barciay, A M. (1967). A determinant of preference among
modes of dissonance reduction. Journal 0/ Personality and Social Psychology, 7,211-216.
Wason, P. C. (1966). Reasoning. In B. M. Foss (Ed.) New horizons in psychology. Harmonds-
worth, England: Penguin.
Wason, P. c., & Johnson-Laird, P. N. (1972). Psychology 0/ reasoning: Structure and content.
Cambridge, MA: Harvard University Press.
Watts, W. A (1977). InteIligence and opinion change through active participation as a
function of requirements for improvisation and time of opinion measurement. Social
Behavior and Personality, 5, 171-176.
Weigel, R. H., & Newman, L. S. (1976). Increasing attitude-behavior correspondence by
broadening the scope of the behavioral measure. Journal 0/ Personality mfd Social Psy-
chalagy, 33, 793-802.
Weimer, W. B. (1979). Psychology and the conceptual/oundations of science. Hillsdale, NJ:
Eribaum.
Weiner, B. (1979). A theory of motivation for some ciassroom experiences. Journal of
Educational Psychology, 71, 3-25.
Weimer, W. B. and Palermo, D. S. (1973). Paradigms and normal science in psychology.
Science Studies, 3, 211-244.
Weiner, B. (1980). Human motivation. New York: Holt, Rinehart & Winston.
Weiner, B. (1985b). An attributional theory of achievement motivation and emotion. Psy-
chological Review, 92, 548-573.
Weiner, B. (1985a). "Spontaneous" causal thinking. Psychological Bulletin, 97, 74-84.
Weiner, B., & Litman-Adizes, T. (1980). An attributional expectancy-value analysis of
learned helplessness and depression. In J. Garber & M. E. P. Seligman (Eds.), Human
helplessness: Theory and applications. New York: Academic Press.
Wheeler, L. (1966). Motivation as a determinant of upward comparison. Journal 0/ Experi-
mental Social Psychology (Supp!. 1), 27-3l.
Wheeler, L., & Koestner, R (1984). Performance evaluations: On choosing to know the
related attributes of others when we know their performance. Journal 0/ Experimental
Social Psychology, 20, 263-27l.
Wheeler, L., Shaver, K. G., Jones, R A, Goethals, G. R., Cooper, J., Robinson, J. E.,
Gruder, C. L., & Butzine, K. W. (1969). Factors determining the choice of comparison
other. Journal 0/ Experimental Social Psychology, 5, 219-232.
Wheeler, L., Koestner, R, & Driver, R E. (1982). Related attributes in the choice of
comparison others: It's there but it isn't aII there iso Journal 0/ Experimental Social
Psychology, 18, 489-500.
Wicker, A. W. (1969). Attitudes vs. actions: The relationship of verbal and overt behavioral
REFERENCES 267

responses to attitude objects. Journal o[ Social Issues, 25, 41-78.

Wicklund, R A., & Brehm, J. W. (1976). Perspectives on cognitive dissonance. Hillsboro, NJ:
Erlbaum.
Wilder, D. A. (1981). Perceiving persons as a group: Categorization and intergroup rela-
tions. In D. L. Hamilton (Ed.), Cognitive processes in stereotyping and intergroup behavior.
Hillsdale, NJ: Erlbaum.
Wilkins, M. C. (1929). The effect of changed material on ability to do formal syllogistic
reasoning. Archives o[ Psychology, 102, 83.
Wills, T. A. (1981). Downward comparison prindples in sodal psychology. Psychological
Bulletin, 90, 245-271.
Wilson, S. R, & Benner, L. A. (1971). The effects of self-esteem and situation upon
comparison choices during ability evaluation. Sociometry, 34, 381-397.
Wilson, T. D. (1985). Strangers to ourselves: The origins and accuracy of beliefs about
one's own mental states. In J. H. Harvey & G. Weary (Eds.), Attribution: Basic issues
and applications. New York: Academic Press.
Woodworth, R S., & Seils, S. B. (1935). An atmosphere effect in formal syllogistic reason-
ing. Journal o[ Experimental Psychology, 18, 451-460.
Worchel, P., & McCormick, B. L. (1963). Self-concept and dissonance reduction. Journalo[
Personality, 31, 588-599.
Worchel, S., & Brehm, J. W. (1971). Direct and implied sodal restoration of freedom.
Journal o[ Personality and Social Psychology, 18, 294-304.
Wrightsman, L. S. (1960). Effects of waiting with others on changes in level of feit anxiety.
Journal o[ Abnormal and Social Psychology, 61, 216-222.
Wyer, R S., & Carlston, D. (1979). Social cognition, in[erence and attribution. Hillsdale, NJ:
Erlbaum.
Wyer, R S., Jr., & Srull, T. K. (1981). Category accessibility: Some theoretical and empiri-
cal issues concerning the processing of sodal stimulus information. In E. T. Higgins,
C. P. Herman & M. P. Zanna (Eds.), Social cognition: The Ontario symposium (Vol. 1).
Hillsdale, NJ: Erlbaum.
Yates, A. J. (1975). Theory and practice in behavior therapy. New York: Wiley.
Zajonc, R. B. (1960). The concepts of balance, congruity and dissonance. Public Opinion
Quarterly, 24, 280-296.
Zajonc, R. B. (1968). Cognitive theories in sodal psychology. In G. Lindzey & E. Aronson
(Eds.). Handbook o[ social psychology (Vol. 1). Reading, MA: Addison-Wesley.
Zanna, M. P. and Cooper, J. (1976). Dissonance and the attribution process. In J. H.
Harvey, W. J. Ickes and R F. Kidd (Eds.). New directions in attribution research (Vol. 1).
Hillsboro, NJ: Erlbaum.
Zanna, M. P. and Darley, J. (1980). On managing the faculty graduate student relation,
Catalog o[ selected documents in psychology, 10, 1.
Zanna, M. P. & Mohr, D. (1985). Unpublished data. Waterloo University.
Zanna, M. P., & Rempel, J. K. (1988). Attitudes: A new look at an old concept. In D. Bar-
Tal & A. W. Kruglanski (Eds.), The social psychology o[ knowledge. Cambridge, England:
Cambridge University Press.
Zanna, M. P., Goethals, G. R, & Hill, J. F. (1975). Evaluating a sex-related ability: Sodal
comparison with similar others and standard setters. Journal o[ Experimental Social
Psychology, 11, 86-93.
Zanna, M. P., Kiesler, C. A., & Pilkonis, P. A. (1970). Positive and negative attitudinal
affect established by classical conditioning. Journal o[ Personality and Social Psychology,
14, 321-328.
Zuckerman, M. (1979). Attribution of success and failure revisited, or: The motivational
bias is alive and weil in attribution theory. Journal o[ Personality, 47, 245-287.
Author Index
Abelson, RP., 91, 93-97,
101, 104, 108, 124,
135, 247, 248, 251,
253, 255, 260
Abramson, L.Y., 60, 190,
213, 247, 252, 258
Ackerman, J.M., 206
Adler, A., 160
Ajzen, 1., 1, 14, 16, 17,
72, 77, 111, 112, 115,
134, 135, 171, 184,
186, 191, 238, 247,
253,258
Albert, 147, 247
Albrecht, F., 101, 251
Albright, L., 171, 172,
176-178, 181, 187,
257
Allen, V.L., 163, 247
Alloy, L.B., 190, 212, 247,
252
Allport, G.W., 111, 112
Allyn, J., 122, 248
Anastasi, A., 227, 248
Anderson, c.A., 61, 68,
248
Anderson, N.H., 164, 248
Appleman, A.J., 72, 248
Argyle, M., 167, 248
Arkin, RM., 72, 73, 248,
249,257
Arkowitz, H., 217, 257
Aronson, E., 17, 45, 95,
96, 124, 247, 248, 251,
253, 255, 256, 260,
261,267
Arrowood, A.J., 146, 248, Beck, A.T., 189, 198, 201-
265 204, 206, 208, 209,
Asch, S.E., 36, 164, 167, 211, 212, 215, 216,
248 249
Atkinson, J.W., 18, 135, Beckman, J., 258
248 Bem, D.J., 85, 115, 129,
Augelli, R.W., 73, 264 130, 139, 181, 249
Benner, L.A., 146, 267
Back, K.W., 164, 193, 248, Bennet, E.B., 234, 249
253 Bergin, A.E., 261, 263,
Bagley, G.S., 127, 255 265
Bagozzi, RP., 135, 248 Berglas, 5., 73, 249, 256
Baldwin, M.W., 119,250 Berkowitz, L., 104, 124,
Bandura, A., 136, 201, 249, 251, 253, 256,
204, 207, 208, 216, 257, 259, 261
218, 248, 263 Berlyne, D.E., 14, 16, 249
Barclay, A.M., 99, 266 Berstein, S., 30, 259
Bargh, J.A., 68, 138, 160, Berscheid, E., 14, 70, 99,
174, 191, 248, 255, 180, 249, 264, 267
256 Birch, D., 18, 135, 248
Bar-Hillel, M., 37, 248 Blake, RR., 167, 250, 261
Barker, P., 224, 254 Boas, F., 29, 249
Baron, RM., 171, 172, Bobsen, H., 127, 261
175, 182, 184, 186, Borgida, E., 63, 249
248,260 Boume, L.E., 10, 25, 27,
Bar-Tal, D., 1, 16, 183, 31,249
248, 249, 251, 255- Bowers, K.S., 188, 249
257, 263, 267 Braude, J.M., 30, 249
Bar-Tal, Y., 1, 248 Breckler, S.J., 112, 249,
Barton, R., 212, 259 252
Basok, M., 23, 50, 151, Brehm, J.W., 19, 85, 99-
185,266 104, 241, 249, 266,
Baumgardener, A.H., 73, 267
140, 248, 249, 254 Brekke, N., 63, 249
Beaber, RJ., 114, 260 Brewer, M.B., 17, 31, 248,
Bechtold, J., 43, 138, 249 249
269
270
Brickman, P., 148, 155,
160,249
Brock, T.C., 113, 151,
249, 254, 262
Brodbeck, M., 149,250
Bruner, J.5., 10, 30, 123,
126, 250, 264
Bucks, A.M., 252
Brunson, B.I., 67
Bulman, R.J., 148, 155,
160,249
Burger, J.M., 72, 248
Burnstein, E., 31, 264
Butler, D., 61, 262
Cacioppo, J.T., 4, 111-
114, 116, 122, 126-
128, 131, 134, 139,
140, 237, 250, 262
Campbell, D.T., 7, 13,
185, 215, 226, 228-
234, 250, 251
Campbell, J., 144, 154,
157, 250, 265
Cantor, N., 30, 250, 255
Cantril, H., 126, 193, 255,
264
Carlsmith, J.M., 17, 45,
130, 248, 253
CarIston, D., 138, 255,
267
Carver, C.S., 15, 256
Castellan, N.J., 264
Chaiken, 5., 5, 114, 117,
119, 128, 132, 190,
250,252
Chapin, W., 212, 259
Chapman, I.L., 26, 27,
250
Chapman, J.P., 26, 27,
250
Chave, E.J., 112, 265
Chen, J., 137,252
Cheng, P. W., 189, 261
Cialdini, RB., 111, 122,
126,250
CIark, RD. III, 162, 163,
165, 166, 168, 259
Cohen, A.R, 100, 101,
103,249
Cole, M., 29, 250
AUTHOR INDEX
Coleman, J.F., 167, 250 Dominowski, RL., 10,
Collins, B.E., 101, 108, 249
111, 250, 257 Doms, M., 163, 251, 252
Conant, J.B., 242, 250 Driver, RE., 146, 266
Conley, M.C., 124, 265 Dull, V., 31, 249
Constanzo, P.R, 161 Dunnette, M.D., 251
Conway, M., 123, 263 Dweck, C.S., 67, 251
Cook, T.D., 140, 226, 228, Dykman, B.M., 190, 252
232, 250, 251 Eagly, A.H., 111, 114,
Cooley, C.H., 12, 251 252,256
Cooper, J., 85, 101, 105, Ebbesen, E.B., 255
108, 111, 251, 267 Effrein, E.A., 112, 123,
Copernicus, M., 241, 242 137,253
Cottrell, N.B., 155 Einhorn, H.J., 171, 173,
Cox, J.R, 191, 254 174, 184, 186, 252
Craighead, W.E., 212, 261 Einstein, A., 237, 239,
Crane, M., 30, 259 240, 242, 252
Crocker; J., 108, 251 Eiser, J.R, 90, 122, 252
Cronbach, c.J., 171, 176- Ellis, A., 201-204, 207,
179, 181, 224, 228, 208, 210, 216, 252
229, 231, 235, 251, Ellis, 5., 53, 54, 118, 190,
254 218, 220, 252
Croyle, KT., 111, 251 Ellsworth, P.c., 171, 264
Cunningham, J.D., 70, Emery, G., 189, 249
262 Enzle, M.E., 67, 70, 252
Epley, S.W., 155,251
D' Agostino, P.R, 75, 80, Epstein, 5., 227, 252
262 Erber, R, 14, 70, 252
DarIey, J., 143, 144, 146, Evans, J.T., 26, 252
147, 150-152, 154,
156, 158, 160, 225,
254,267 Farkash, E., 63, 258
Darwin, c., 237, 251 Fazio, RH., 44, 101, 105,
Darwin, F., 237, 251 108, 112, 123, 136,
Dashiell, J.F., 259 137, 150, 251-253,
Datan, N., 254 262-264
Davis, KE., 23, 69, 71, Feigenbaum, R., 72, 256
74, 77, 223, 257 Feldman, 5., 262
Davison, G.c., 213, 251 Ferguson, T.J., 67, 253
Davison, J., 73, 263 Feshbach, 5., 127, 256
Day, RA., 225, 251 Festinger, L., 5, 24, 32,
Deaux, K, 233, 251 46, 85, 87, 90, 100,
Decarufel, A.c., 155, 251 101, 103, 122, 125,
deHaan, H., 118, 265 130, 143, 144, 146,
Dermer, M., 14, 249 148, 154-157, 161,
Desoto, c., 101, 251 164, 248, 253, 256
Deutsch, M., 166, 167, Feyerabend, P., 224, 231,
251 253
Devine, P.G., 108, 251 Fincham, F.D., 61, 253
Diener, C.T., 67, 251 Fine, M., 234
Doise, W., 251 Fischoff, B., 184, 264
AUTHOR INDEX
Fishbein, M., 72, 77, 111,
112, 115, 134, 135,
247,253
Fiske, S.T., 10, 12, 15, 30,
67, 70, 79, 135, 179,
247, 252, 253, 261,
265
Flavell, J.H., 27, 253
Flay, B.R., 140, 251
Fong, G.T., 189, 253, 261
Foss, B.M., 266
Frank, J.D., 198,253
Frankei, A., 73, 253
Freedrnan, J.L., 102, 253
Frenkel-Brunswik, E., 15,
253
Freud, 5., 12, 238, 241
Freund, T., 14, 17, 36, 37,
42, 75, 82, 128, 154,
191, 253, 258
Frey, D., 19, 108, 253, 264
Friedland, N., 63, 258
Friend, R., 146, 248, 253
Funder, D.C., 171, 172,
174, 182, 184, 187,
254
Gaedert, W.P., 73, 254
Gage, N.L., 171, 176, 250,
254
Galileo, 241
Garber, J., 266
Garfield, S.L., 261, 263,
265
Gastorf, J., 146, 265
Gerard, H.B., 144, 156,
166, 167, 251, 257
Gergen, K.J., 12,224,254
Gholson, B., 224, 254
Gibbons, F.X., 73, 254
Gilbert, J., 146, 253
Gilrnore, J.B., 120, 256
Ginsburg, E., 67, 260
Goethals, G.R., 5, 143,
144, 146, 147, 149-
152, 154-156, 158,
160, 254, 263, 267
Goffman, E., 12,254
Goldrnan, R., 127, 128,
262
Goldstein, A.P., 257
Gornbrich, E.H., 254
Goodstadt, M.S., 75
Gordon, c., 254
Gotz, J., 99, 254
Gotz-Marchland, B., 99,
254
Graziano, W., 14, 249
Greenwald, A.G., 105,
108, 122, 140, 252,
254
Gregory, K.L., 10, 254
Griggs, R.A., 191, 254
Gruder, c.L., 140, 143,
146, 147, 154, 251,
254
Haberkorn, G., 163, 259
Hakrniller, K., 146, 254
Hall, C.S., 161, 255
Harnei, I.A., 1, 57, 61, 62,
66,258
Harnilton, D.L., 44, 80,
81,255,267
Hannah, D.B., 108, 251
Hannah, R., 75, 255
Hannah, T.E., 75, 255
Hardyck, J.A., 99, 255
Hartlage, 5., 190, 252
Harvey, J.H., 62, 67, 69,
70, 72, 248, 249, 252,
253, 255, 257, 258,
262, 264, 267
Hass, J.W., 127, 255
Hastie, R., 31, 50, 81,
108, 171-174, 255
Hastorf, A., 171, 193, 255,
264
Heider, F., 32, 48, 71, 77,
85, 89, 90, 102, 217,
223,255
Heller, J.F., 70, 255
Henle, M., 29, 255
Hennigan, K.M., 140, 251
Herrnan, c.P., 255, 267
Herr, P.M., 112, 123, 253
Hewstone, M., 257
Higgins, E.T., 10, 12, 67,
68, 79, 98, 138, 147,
152, 159, 164, 174,
176, 180, 191, 238,
255, 256, 258, 267
271
Hill, J.F., 146, 267
Hirnrnelfarb, 5., 111,256
Hofstede, G., 15, 256
Horgarth, R.M., 171, 173,
174, 184, 186
Hovland, c.I., 112, 113,
118, 121, 126, 256,
260, 263, 264
Hoyt, M.F., 101, 108, 250
Hubbard, M., 157, 263
Hubert, 5., 164, 248
Hurnphries, c., 15,256
Ickes, W., 249, 252, 257,
257, 262, 267
Ingrarn, R.E., 73, 264
Insko, C.A., 52, 87, 90,
92, 95, 155, 251, 256,
264,265
Irle, M., 95, 96, 99, 254,
256
Jacobson, L.I., 204, 261
Jaffe, Y., 53, 59, 190, 258
Jarnes, W., 12, 32, 172,
256
Janis, I.L., 113, 121, 127,
256
Jepson, c., 171, 189, 256,
261
Johnson, T.J., 72, 256
Johnson-Laird, P.N., 25,
27,256,266
Jones, E.E., 23, 69, 71, 73,
74, 77, 144, 156, 166,
171, 187, 223, 249,
256,257
Jones, S.c., 149, 264
Kahnernan, D., 12, 37,
171, 173, 182, 186,
188, 189, 238, 257,
266
Karnenetzky, J., 118, 265
Kanause, D.E., 257
Kanfer, F.H., 207, 257
Kanki, G., 164, 261
Kardes, F.R., 112, 253
Kardush, M., 99, 255
Kassin, S.M., 62, 63, 68,
257
272 AUTHOR INDEX

Katz, D., 123, 124, 126, Kumof, M., 19,264 Lurie, L., 79, 255
257 Kunda, Z., 16, 171, 258, Lydon, J., 119, 259
Kelley, H.H., ix, 3, 15, 261 Lynch, J.S., 26, 252
16, 41, 57, 58, 60, 61,
63-65, 69, 70, 72, 74, Lage, E., 167, 261 Maass, A., 162, 163, 165,
76-78, 82, 106, 113, Lakatos, I., 185, 243, 244, 166, 168, 259
121, 151, 167, 217, 254, 258, 259 Maccoby, N., 122, 253
223, 256, 257 Lake, E.A., 249 Mach, K., 239
Keiman, H.C., 5, 166, 257 Lamb, T.W., 52, 167, 257 Mahoney, K., 209, 259
Kenny, D.A, 171, 172, Lana, R.E., 233, 259 Mahoney, M., 202, 209,
176-178, 181, 187, Laudan, L., 224, 231, 254, 259
257 259 Maides, S.A., 1, 57, 61,
Kidd, R.F., 249, 252, 257, Lawhon, J., 146,265 62,66,258
258, 262, 267 Lazarus, AA, 211, 259 MandelI, W., 118
Kiesler, C.A., 19, 111, Lazarus, R.S., 212, 259 Markus, H., 30, 259
114, 257, 267 Leavitt, G.5., 171, 254 Marrow, A, 239, 259
Kihlstrom, J.F., 255 Lehman, D., 189, 261 Maruyama, G., 114, 260
Kinder, D.R., 135, 247 Leippe, M.R., 140, 254 Marvin, G.H., 10
King, G.A., 10, 12, 67, Lenin, W., 240 Matthews, K.A, 67, 250
79, 98, 152, 174, 191, Lenn, T.M., 112, 123, Mayseless, 0., 17, 37, 40,
238, 255, 256 137,253 41, 47, 50, 51, 75, 79,
Klar, Y., 16, 49, 71, 85, Leone, c., 124, 265 SO, 117, 120, 144, 149,
91, 101, 108, 135, 150, Lepper, M.R., 31, 157, 153, 154, 165, 180,
249, 257, 258 191, 194, 259, 263, 228, 258-260
Klein, R., 12, 256 266 McArthur, L.Z., 67, 70,
Koestner, R., 146, 266 Lerner, M.J., 101, 259 171, 172, 175, 182,
Kolditz, T.A., 73, 257 Leventhal, H., 127, 259 184, 186, 260
Kopel, 5., 217, 257 Levi, A., 72, 265 McCormick, B.L., 149,
Krantz, D.H., 171, 189, Levine, D., 257, 262 267
253, 256, 261 Levine, J.M., 163, 247 McDonel, E.C., 137,252
Kriss, M., 67, 262 Levine, M.A., 10, 259 McFariand, c., 99, 194,
Kroy, M., 229, 258 Levy-Bruhl, L., 27, 29, 263
Kruglanski, AW., vii, 259 McGillis, D., 17, 257
viii, ix, 1, 14, 16, 17, Lewin, K., 20, 234, 239, McGuire, W.J., 26, 95, 96,
23, 36, 37, 38, 40, 41, 259 121, 124, 247, 248,
42, 47, 50, 51, 53, 57, Lewinsohn, P.M., 212, 251, 253, 255, 260
59, 61-63, 65, 71, 75, 213,259 Meehl, P.E., 228, 231
79, 80, 82, 85, 107, Lichtenstein, P., 184, 264 Meichenbaum, D., 202,
108, 117, 120, 128, Lightner, J.M., 70, 255 203, 207, 209, 260
135, 140, 144, 148- Lindman, H.R., 264 Mendel, G.J., 240
150, 153-155, 164, Lindzey, G., 161, 248, Merton, R.K., 224, 260
165, 167, 171, 175, 255, 256, 261, 267 Mettes, D.R., 144, 156,
180, 184, 186, 188, Litman-Adizes, T., 60, 160,260
190, 191, 228, 229, 266 Mewborn, c.R., 127, 260
238, 248, 249, 251, Loftus, E.F., 10, 249 Mewborn, R., 127, 263
253, 255-260, 262, Lombardi, W., 68, 138, Michela, J.L., 57, 72, 257
263,267 256 Milgram, N., 248
Kuhn, T.S., 11, 32, 185, Lord, c.G., 31, 191, 259 Miller, C.T., 77, 146, 147,
224, 231, 242, 243, Luchins, AS., 191, 259 260
254,258 Lui, L., 31, 249 Miller, D.T., 16, 70, 72,
Kuhl, J., 258 Lumsdaine, .A, 121, 256 80,260
Kumar, P.A, 50, 81, 255 Luria, A.R., 29, 259 Miller, F.D., 67, 264
AUTHOR INDEX 273

Miller, N., 111, 114, 257, O'Brlen, E.J., 227, 252 Raush, CL., 201, 263
260 Olmstead, J.A., 167, 261 Raush, H.S., 201, 263
Miller, R.L., 249, 251, O'Nea!, E., 124, 260, 262 Raven, B.H., 167, 263
254, 260 Orvis, B.R., 61, 70, 262 Reckman, R.F., 149, 263
Mills, J., 124,260,262 Osberg, T.M., 73, 264 Reese, H., 254
Mische!, W., 30, 211-213, Osgood, CE., 52, 92, Regan, D.T., 137,263
250, 259, 260 103, 119, 262 Reisman, S.R., 90, 265
Mitroff, I.I., 24, 261 Ostrom, T.M., 108, 113, Rempel, J.K, 112, 115,
Mohr, D., 138, 267 251, 254, 255, 262 119, 123, 267
Monson, T., 14, 249 Restle, F., 264
Montmann, V., 95, 256 Paicheler, G., 163, 26:L Rhodewalt, F., 73, 263
Moore, B.S., 68, 261 Palermo, D.5., 224, 266 Rodin, J., 213, 263
Morgan, J.J.B., 26, 261 Pancer, S.M., 122, 252 Rogers, C, 201
Morton, J.T., 26, 261 Papageorgis, D., 121, 260 Rogers, R.W., 127, 255,
Moscovici, S., 5, 164-167, Pauli, W., 242 260, 263
251,261 Paulus, P.B., 261 Rokeach, M., 15, 263
Mouton, J.S., 167, 250, Peizer, D.B., 92 Ronis, D.I., 105, 108, 254
261 Pepitone, A., 90, 95, 101, Rose, T.L., 31, 263
Mugny, G., 164, 261 103, 104, 262 Rosenberg, M.J., 93-96,
Mulkay, M.J., 224, 240, Perl, N., 38-40, 262 112, 124, 248, 247,
241,261 Peters, M.D., 135, 247 251, 253, 255, 260,
Murchison, C, 247 Pettigrew, T.F., 156, 159, 263
Murray, E.J., 204, 261 262 Rosentha!, R., 250, 259
Musgrave, A., 243, 258, Petty, R.E., 4, 111-114, Rosenthai, T., 204, 263
259 116, 122, 126-128, Rosnow, R.L., 250, 259
Muzfir, M., 52, 53, 119, 131, 134, 139, 140, Ross, L., 77, 157, 171,
261 237, 250, 262 182, 186, 189, 194,
Petullo, L., 251 213, 261, 263, 266
Piaget, J., 27, 28, 253 Ross, M., 16, 72, 80, 99,
Naccarato, M., 43, 139, Pilkonis, P.A., 114, 267 119, 122, 191, 194,
249 Pisoni, D.B., 262 252, 259, 260, 263
Neely, J.H., 113, 261 Pittman, T.S., 70, 73, 75, Rozen, E., 73, 263
Nelson, R.E., 144, 149, 80,262 Runke!, P.S., 92
150, 212, 254, 261 Pol, 1., 49, 91, 101, 108, Rush, A.J., 189, 249
Nemeth, CJ., 5, 164,261 257
Neuberg, S.L., 15, 70, Popper, KR., 6, 10, 11,
253,261 172, 173, 185, 224, Sadler, 0., 124, 263
Neumann, P.G., 15, 256 228, 231, 236, 244, Sakamura, J.A., 19, 257
Newcomb, T.M., 124, 262 Salancik, G.R., 122, 123,
247, 248, 251, 253, Powell, M.C, 112, 123, 263
255,260 137, 253, 262 Sanbomatsu, D.M., 44,
Newell, A., 10, 12 Pratkanis, A.R., 252 112, 253, 264
Newman, H., 61, 261 Preston, E., 31, 259 Samoff, 1., 123, 263
Newman, L.S., 135, 266 Proximire, W., 17 Saxe, L., 234
Newton, 1., 239 Pryor, J.B., 67, 262 Schachter, S., 143, 153,
Nisbett, R.E., 99, 171, 155, 156, 164, 213,
182, 186, 188, 189, Radloff, R., 153, 262 253, 261, 263
213, 253, 256, 257, Raimy, V., 202, 211, 216, Schemrer, F.W., 70, 262
261,265 262 Schilpp, P.S., 250, 252,
Norman, R., 119, 237, 261 Ramsey, G.V., 250 258
Norman, S.A., 70, 260 Rasinsky, KA., 171-174, Schneider, D.J., 171, 264
Nunally, J., 127, 261 255 Schneider, W., 188, 264
 274
Schopflocher, D., 67, 70,
252
Schpitzajzen, A., 14, 128,
253
Schwartz, N., 1, 19, 57,
61, 62, 66, 264
Schwartz, J.M., 258
Scott, W.A., 115, 264
Scribner, 5., 29, 250
Seligman, M.E.P., 60,
247,266
SeIls, S.B., 26, 264, 267
Sentis, K.P., 31, 264
. Shaw, B.F., 189, 249
Shaw, M.E., 161
Sheffield, F.D., 121, 256
Shelton, S.L., 206
Sherif, M., 126,264
Sherman, S.J., 137, 252
Shriffin, RM., 12, 188
Shrauger, J.S., 73, 149,
264
Sicoly, F., 16, 72, 80, 263
Siladi, M., 30, 259
Simon, H.A., 10, 12, 261
Singer, J.E., 103, 104, 146,
156, 159, 264
Slovic, P., 182, 184, 257
Smith, E.R, 67, 264
Smith, G., 144, 156, 160,
260
Smith, M.B., 123, 126,
264
Smith, T.W., 73, 264
Smith, W.P., 144, 264
Snyder, C.R., 73, 264
Snyder, M., 45, 50, 138,
180, 183, 264
Snyder, M.L., 73-76, 148,
155, 253, 264
Sorrentino, RM., 256,
258
SruH, T.K., 164, 191, 248,
267
Staats, A.W., 114, 265
Staats, c.K., 114, 265
Stangor, c., 5, 79, 190,
250,255
Stanley, J., 7, 226, 228-
233,250
Stone, G.c., 171, 254
Storms, M.D., 213, 265
Strauman, T., 12, 256
Stroebe, W., 90, 257, 265
Strong, S.R., 201, 202,
265
Suls, J.M., 146, 249, 251,
254, 257, 260, 265
Swann, W.B., 45, 50, 138,
171, 172, 175, 179-
183, 185, 264, 265
Swedlund, M., 164, 261
Tagiuri, R., 251
Tanke, E.D., 180, 264
Tannenbaum, P.H., 52,
92, 103, 119, 124, 247,
248, 251, 253, 255,
260,262
Taylor, S.E., 10, 12, 30,
67, 79, 179, 253, 265
Teasedale, J.D., 60, 247
Terwillinger, RF., 127,
256
Tesser, A., 124, 144, 154,
263,265
Testa, T.J., 115, 265
Tetlock, P.E., 72, 75, 125,
265
Thistlewaite, D.L., 118,
265
Thompson, S.c., 67, 265
Thompson, V.P., 90, 265
Thomton, D., 146, 265
Thurstone, L.L., 112, 265
Town, J.P., 70, 255
Trope, Y., 23, 50, 151,
157, 185, 266
Trotsky, L., 240
Tversky, A., 37, 171, 173,
182, 238, 257, 266
Vallone, RP., 194, 266
Valone, K:, 114, 260
van Avermaet, E., 163,
252
WaHace, A.F.C., 29
Walster, E., 99, 266
Wason, P.c., 25-27, 256,
266
Wattie, B., 75, 255
Watts, W.A., 122, 266
AUTHOR INDEX
Weary, G., 62, 67, 69, 72,
248, 253, 255, 262,
267
Weber, R., 108, 251
Weiby, M., 72, 256
Weigel, R.H., 135, 266
Weimer, W.B., 6, 11, 185,
224, 236, 237, 243,
266
Weiner, B., 15, 48, 60, 70,
71, 79, 108, 257, 266
WeHs, c.L., 67, 253
Wheeler, L., 146, 147, 266
White, RW., 123, 126,
264
Wicker, A.W., 134,266
Wicklund, RA., 17, 74-
76, 85, 99, 102-104,
148, 155, 264, 266
Wilder, D.A., 31, 267
Wilkins, M.C., 26, 267
Wills, T.A., 144, 147, 154,
267
Wilson, S.R., 146, 267
Wilson, T., 99, 261
Wilson, T.D., 188, 267
Wood, W., 114, 252
Woodworth, RS., 26, 267
Worchel, P., 149, 267
Worchel, S., 19, 267
Wright, E., 260
Wright, J.B., 70, 262
Wrightsman, L.S., 234,
267
Wyer, RS., 138, 164, 191,
248, 255, 267
Yarkin, K.L., 70, 255
Yates, A.J., 204, 267
Zajonc, R.B., 90, 94, 102,
267
Zakai, D., 38, 262
Zanna, M.P., 43, 85, 112,
114, 115, 119, 123,
136, 138, 139, 146,
225, 249, 252, 253,
255,267
Zimbardo, P.G., 213, 263
Zuckerman, M., 72, 77,
267
Subject Index
Ability
comparison processes, 144-145, 154
malleability of, 157-158
opinions versus, 156-158
Accessibility
accuracy, 191
attitudes, 137-139
attribution, 66-68
construct accessibility, 238-239
Accountability, 75
Accuracy, 171-195
comparison processes, 148, 153
definitions, 172-176
framework for, 187-188
lay epistemics and, 5-6
phenomenal aecuracy, 192-194
realist paradigm, 188-192
relative attainability, 179-181
research in, 171
taxonomies of accuracies, 176-187
Accuracy-mediating process, 184
Aetion
attitude-behavior relation, 134-139
comparison processes, 159-160
endogenous-exogenous partition, 2-3
Action appraisal, 167
Advice, 151-152
Affect, 104-105
Affective-eognitive consistency theory, 93
Agreement/disagreement, 46-48
Ambiguity
attributions, 74-75
epistemic motivation, 38-40
nonspecific closure, 14
situational arousal, 15
Anchoring effeets, 37
ANOVA model
covariation, 60
validation, 58
Assessment shifts, 198-200, 208-209
Atmosphere effect, 26-27
Attentional focus, 120-123
Attitude(s), 111-141
aeeessibility, 137-139
attentional foeus factors, 120-123
authority, 118
behavior, 134-139
central versus peripheral routes, 131-
133
classical conditioning, 114-115
content-bound research, 52-53
defined, 111-113
derivation of, 113-120
epistemic motivation and, 123-129
evidential consistency, 119-120
knowledge structures, 4
logical bases of, 115-116
relative efficacy, 116-117
unitary/disjunctive processes, 129-131
Attribution, 57-83
cognitive capabibility, 66-68
cognitive consistency, 4
eonsistency theory, 85-86, 106-107
deductive element, 99
future research implications, 81-82
informationallmotivational aspects inte-
gration, 77-78
internal-external distinetion, 2
logic, 3, 57-66
motivational bases, 68-82
unique/nonunique aspects, 78-81
see also Consistency theory
275
276
Attributional criteria, 3
Augmentation, 65-66
Authority
attitudes and, 118
cognitive therapy, 201-204
comparison processes, 150
research methodology, 236-238
Availability, 66-68
Avoidance. See Closure avoidance
Balance theory, 89-92
Behavior. See Sodal behavior
Beliefs
cognitive therapy, 201
comparison processes, 151
irrational beliefs, 216-217
Belief-value distinction, 158
Bias
cognitive therapy, 6
comparison processes, 152
knowledge utilization context, 43
Bolstering, 97
Causal attribution models, 1
Causation and causality
covariation, 59
research methodology, 226
Central attitude change, 131-133
Circumscribed accuracy, 179-181
Classical conditioning, 114-115
Closure
attitudes, 123-129
attributions, 69-76
cognitive therapy, 205-207
comparison processes, 149, 150, 154-
155
epistemic activity, 38
epistemic motivation, 13-22
knowledge utilization context, 42
minority influence, 165
primacy effects, 36
research methodology, 240-245
sodal behavior, 45, 47
subjective confidence, 41
Closure avoidance
attributions, 71-73
cognitive therapy, 206-207
comparison processes, 150, 154-155
epistemic motivation, 13-22
minority influence, 164
Cognitive balance, 48-49
SUBJECT INDEX
Cognitive capability
attribution, 66-68
hypothesis generation, 13
Cognitive consistency theory. See Con-
sistency theory
Cognitive factors, 164-165
Cognitive psychology, 35. See also Cogni-
tive therapy
Cognitive theory, 12
Cognitive therapy, 197-221
contents of change, 208-209
epistemic authority, 201-204
freezing/unfreezing mechanisms, 205-
207
frustrative hypothesis, 197-200
implications of lay epistemics for, 219-
221
irrational beliefs, 216-217
knowledge structures, 200-201
lay epistemics and, 6, 197, 209-221
personal experience, 217-219
Common sense, 6-7
Comparison processes, 143-161
contents of question, 146-148
domain differentiation, 155-158
dynamic/static views of, 160-161
lay epistemics and, 158-161
motive base of, 154-155
research in, 143-144
senses in, 144-145
sodal sources of information, 148-153
see also Minority influence
Compatibility, 19-20
Compliance, 5, 166-168
Conditioning, 114-115
Confidence. See Subjective confidence
Confirmatory strategy, 50-51
Conformity, 165-166
Confusion. See Ambiguity
Congruity model, 92-93
Consensus
accuracy, 174
logic,58
magnitude of, 62
Consistency
attitudes, 119-120
cognitive therapy, 201
logic,58
minorityinfluence, 162-163
Consistency theory, 85-109
attributions and, 85-86, 106-107
SUBJECT INDEX 277

Consistency theory (cant.) Dissonance theory (cant.)

content aspect of, 100-101 revised,95
deductive element in, 99-100 Distinctiveness criterion, 58
inconsistency, 4 Distortion, 210-215
lay epistemic interpretation of, 105-106 Diverse origination, 15
logic and, 3, 86-100 Domain differentiation, 155-158
motivational aspect, 102-109 Doubt, 24-25
research implications, 107-109 Drives, 154
See also Attribution Drive upward. See Unidirectional drive
Construct accessibility, 238-239 upward
Content Dyadic accuracy, 178
cognitive therapy, 208-209 Dysfunctional beliefs/thoughts
comparison processes, 146-148, 159-160 cognitive therapy, 197-200, 210-215
Content-bound research, 52-55 lay epistemics, 6
Content domains, 4-5
Content-process distinction, 155-158 Ego protection, 72
Content-spedfic definitions, 100-101 Elevation accuracy, 177
Contradiction Empirical research, 35-55
balance theory, 90-92 content-bound research, 52-55
logic, 24-25 knowledge formation, 35-42
Control deprivation, 75-76 knowledge utilization context, 42-44
Correspondence, 172-174 logical element, 48-51
Covariation motivational effects, 35-48
causation, 59 sodal behavior context, 45-48
LEGO model, 59 see also Research methodology
Culture, 147 Encoding, 23-30
Endogenousattribution,61
Deduction, 99-100 Endogenous-exogenous partition, 2-3
Denial Epistemic authority. See Authority
deductive element, 100 Epistemic freezing. See Freezing
inconsistency resolution, 96-97 Epistemic motivation. See Motivation
Depression Epistemic theory. See Attribution; Con-
accuracy, 189-190 sistency theory; Theory
frustration, 198 Error, 25-30. See also Accuracy
truth, 213-215 Ethnic stereotypes, 42-43
see also Cognitive therapy Evaluation. See Self-evaluation
Developmental stage theory, 27-29 Evidential relativity, 236-238
Diagnostic strategy, 50-51 Exogenousattribution,61
Disagreement. See Expectandes
Agreement/ disagreement disconfirmation of expectancy theory,
Disconfirmation of expectancy theory, 95. 95
See also Expectandes sodal behavior, 45-46
Discounting prindple Experimentation
augmentation versus, 65-66 research design, 231-235
logical aspects, 64-65 research methodology, 225-226
Discounting tendency, 41 Expert authority, 202-203
Discrimination, 43 External information, 37
Disjunctive processes, 129-131
Dissonance theory Forced-compliance expriment, 130
attitudes, 129-130 Freezing
logic, 87-89 attributions, 72
278
Freezing (cont.)
cognitive therapy, 205-207
nonspecific c1osure, 14, 19
primary effects, 36
Freudian theory, 12
Frustration
cognitive therapy, 197-200
dysfunctional misconception hypothe-
sis, 210
Global accuracy, 179-181
Global impressions, 44
Goal shifts, 198-200, 208-209
Group authority, 203
Guessing, 188-189
Hedonic relevance, 71
Heuristics
accuracy, 182
construct accessibility, 238
content domains, 4-5
Human logic. See Logic
Hypotheses
knowledge acquisition, 10
knowledge function, 30-31
Hypothesis generation, 13-22
cognitive capability, 13
epistemic motivations, 13-22
knowledge formation, 37-38
Hypothesis-generation-validation model,
10
Hypothesis testing, 50-51
Hypothesis validation process, 22-30
Impression formation, 36
Inconsistency
affect and, 104--105
balance theory, 90-92
c1assification of, 103-104
content-specific definitions, 100-101
knowledge functions, 30-31
minority influence, 164-165
mode of resolution, 96-99
motivation, 4
searchlight metaphor, 12
universal drive, 102-103
Inconsistency resolution, 49-50
Independence, 165-166
Inference, 25-30
Inferential statistics, 226-227
SUBJECT INDEX
Information processing
accuracy, 184-185
attitude, 129-131
empirical research, 35-40
Information sources. See Social informa-
tion sources
Information type, 40-41
Internal-external distinction, 2
Internal information, 37-38
Internalization, 5, 166-168
Irrational belief, 216-217
Irrelevance-of-origin assumption, 76-77
Issue appraisal, 167
Judgmental noncommitment, 17-18
Judgmental process
attribution, 57. See also Attribution
content-bound research, 52
Judgments
accuracy and, 172-174, 192, 193-194
attitudes, 120
comparison processes, 153
consensus, 174
minority influence, 162-165, 167
stereotyped judgments, 42-43
validation of, 113
see also Accuracy
Knowledge
functions of, 30-31
psychological properties of, 9-12
research methodology, 224
Knowledge acquisition process, 13-22
Knowledge formation, 35-42
Knowledge structures, 200-201
Knowledge utilization context, 42-44
Lay epistemics
accuracy, 5-6. See also Accuracy
breadth of, 1-2
cognitive therapy, 6, 197, 209-221. See
also Cognitive therapy
comparison processes, 158-161. See also
Comparison processes
content domains, 4-5
empirical research in, 35-55. See also
Empirical research
endogenous-exogenous partition, 2-3
logical consistency/inconsistency, 3
motivational bases, 4
scientific method and, 6-7
SUBJECT INDEX 279

Lay epistemics (cont.) Motivation (cont.)

social comparison, 5 lay epistemics, 4
theory, 9-33. See also Theory minority influence, 163
LEGO blocks metaphor primacy effects, 36
attitudes, 115, 122, 135 sodal behavior, 48
causation, 59 subjective confidence, 41-42
cognitive capability, 67
described, 10-11
Native peoples, 29-30
hypothesis generation, 13
Neurosis,6
knowledge functions, 31
Nonspecific closure
research methodology, 236
attitudes, 123-125
Logic
attributions, 69-71
attitudes, 113-114, 115-116
research methodology, 242-243
attribution and, 57-66
Nonspecific closure avoidance
consistency theory, 86-100
attitudes, 127-129
contradiction, 24-25
needs of, 74-76
empirical research, 48-51
research methodology, 243-245
lay epistemics, 3
Nonspecificity, 13-22
limitations of, 25-30
Norms
Long-term memory, 10-11
accuracy, 186-187
cognitive therapy, 201
Magnitude concept, 65
Majority influence. See Minority influence
Means shifts, 198-200, 208-209 Opinions
Memory. See Long-term memory; Short- abilities versus, 156-158
term memory comparison processes, 149-150, 154
Methodology. See Research methodology malleability of, 157-158
Minority influence, 161-168 social comparison, 144-145
compliance versus internalization, 166-
168
Performance, 146
conformity versus independence, 165-
Peripheral attitude change, 131-133
166
Personal experience, 217-219
definitions, 161
Persuasive efficacy, 205-206
judgment formation, 162-165
Phenomenal accuracy, 192-194
lay epistemics, 5
Prejudice, 43
reputations, 161
Primacy effects, 36
Modeling, 136
Process-contents distinction, 3
Mode of resolution, 96-99
Process phenomena, 160
Motivation, 13-22
Product role, 90
accuracy, 182-183, 186, 191-192
Progress, 11
attitudes, 123-129
Psycho-Iogic theory, 93-95
attribution, 68-82
classification of, 13-14
cognitive therapy, 205-207 Range-seeking tendency, 146
compared/contrasted, 20-22 Rational emotive therapy (RET), 203, 210-
comparison processes, 148, 149-150 211
compatibility among, 19-20 Rationalization, 97-98
consistency theory, 102-109 Reality
empirical research, 35-48 accuracy, 182-183
information type, 40-41 cognitive therapy, 210-211
initial confidence, 38-40 stress and, 212
280 SUBJECf INDEX

Relevance, 150-153 Social perception, 172-173

Reliability, 227-228. See also Validity Social psychology. See Research
Research design, 231-235 methodology
Research methodology, 221-245
official concepts, 224, 225-235 Specific closure
overview of, 223-224 attitudes, 125-127
social cognitive enterprise, 235-245 attributions, 71-73
see also Empirical Research cognitive therapy, 205-206
Resolution, %-99 comparison processes, 155
Response-set accuracy, 177-178 minority influence, 163-164
Retrieval, 44 research methdology, 240-242
Revised dissonance theory, 95 see also Oosure
Reward,115 Specifidty, 13-22
Role theory, 12 Statistics, 226-227
Rulelearning, 189-191 Stereotyped judgments, 42-43
Subjective confidence
attitudes, 118, 119
Saliency, 66-68 comparison processes, 148, 149
Sdence consistency theory and, 106-107
lay epistemics and, 6-7, 9 discounting princples and, 65
progress and, 11 epistemic motivation and, 41-42
socal cognitive enterprise, 235-245
see also Empirical research; Research
Target effect, 177
methodology
Temporalordering,59
Sdentific experiment, 225-226
Testing, 227-228
Searchlight metaphor
Theory, 9-33
described, 10, 11-12
definitions, 9
hypothesis generation, 13
hypothesis validation process, 22-30
logic limitations, 26
knowledge acquisition process, 13-22
Self-ascribed authority, 53-55, 203-204
knowledge functions, 30-31
Self-esteem, 156
psychological properties of knowledge,
Self-evaluation, 146
9-12
Self-handicapping, 73-74
Topic boundedness, i4-15, 18
Self-perception, 130
Trait effect, 177
Sex discrimination, 43
Transcendence, 97
Short-term memory, 10, 11-12
Truth
Sign imbalance, 90
accuracy, 172
Similar other, 148
depression, 213-215
Situational arousal, 15
Sleeper effect, 121-122
Social behavior Unbiasedness
accountability, 75 nonspecific closure, 15-16
attitude and, 134-139 see also Bias
comparison processes, 147 Unfounded knowledge, 24
empirical research, 45-48 Unfreezing mechanism, 205-207. See also
lay epistemics, 5 Freezing
minority influence, 162-163 Unidirectional drive upward, 156-157
Social comparison processes. See Com- Uniqueness effect, 177
parison processes Unitary process, 129-131
Sodal information sources, 148-153 Usefulness, 175-176
SUBJECT INDEX 281

Validation Validity (cant.)

inconsistency, 104 taxonomies of, 228-231
judgments, 113 Value-belief distinction, 158
logic, 57-58 Values, 150-151, 152
Validity
reliability versus, 227-228 Zeitgeist effect, 163-164