Advances in Group Processes Vol 18

LIST OF CONTRIBUTORS
James W Balkwell
Department of Sociology, University of

Georgia
Ann Branaman
Department of Sociology, Florida Atlantic

University
David Dunning
Department of Psychology, Cornell

University
Sandra Enns
Department of Anthropology and

Sociology, University of British Columbia
Martha Foschi

Michael A . Hogg
School of Psychology, University of

Queensland
Satoshi Kanazawa
Department of Sociology, Indiana

University of Pennsylvania
Gideon Keren
Faculty of Technology Management,

Technical University of Eindhoven
Vanessa Lapointe

George F Loewenstein
Department of Social and Decision

Sciences, Carnegie-Mellon University
David Sally
Department of Economics and

Organizational Behavior, Cornell
University
vii
V ill
Chris Snijders
Department of Sociology/ICS, Utrecht

University
Leaf Van Boven
Faculty of Commerce,University of British

Columbia
PREFACE
EDITORIAL POLICY
Advances in Group Processes publishes theoretical analyses, reviews and
theory based empirical chapters on group phenomena . The series adopts a
broad conception of "group processes ." This includes work on groups ranging
from the very small to the very large, and on classic and contemporary topics
such as status, power, exchange, justice, influence, decision making, intergroup
relations and social networks . Contributors have included scholars from diverse
fields including sociology, psychology, political science, philosophy, mathematics and organizational behavior .
The series provides an outlet for papers that may be longer, more theoretical
and/or more integrative than those published by standard journals . All papers
undergo a peer review process to ensure a consistent standard of quality . We
place a premium on the development of testable theories and theory-driven
research . Chapters in the following categories are especially apropos :
Conventional and unconventional theoretical work, from broad metatheoretical and conceptual analyses to refinements of existing theories and
hypotheses. One goal of the series is to advance the field of group processes
by promoting theoretical work .
Papers that review and integrate programs of research . The current structure
of the field often leads to the piecemeal publication of different parts of a
program of research . This series offers those engaged in programmatic
research on a given topic an opportunity to integrate their published and
unpublished work into a single paper . Review articles that transcend the
author's own work are also of considerable interest.
Papers that develop and apply social psychological theories and research to
macrosociological processes. One premise underlying this series is that links
between macro and microsociological processes warrant more systematic
and testable theorizing . The series encourages development of the macrosociological implications embedded in social psychological work on
groups.
In addition the editors are open to submissions that depart from these
guidelines .
ix
CONTENTS OF VOLUME 18
This volume of Advances in Group Processes examines a broad span
theoretical and empirical work on group-related phenomena. In "From
Prototypicality to Power: A Social Identity Analysis of Leadership," Michael A.
Hogg offers a provocative new model of group leadership based on principles
of social categorization. This model asserts that people often base their
perceptions and evaluations of group leaders not in terms of objective
indicators, but in terms of how well the leader fits a prototypical view of group
membership. After showing where the social identification literature bears on
leadership processes, Hogg reviews empirical evidence and examines the
relation between leadership and power. The next chapter also addresses issues
of group leadership and hierarchy formation. In "Rational and Irrational Bases
of Commitment to Group Hierarchies," Ann Branaman examines the formation
and maintenance of observable power and prestige orders from the perspective
of two highly divergent theories. Using expectation states theory she first
explores "rational" bases of hierarchy that are based on expectations of
competence. Branaman then employs Goffman's theory of self and social
interaction to examine "irrational" bases of hierarchy that are upheld by
emotions, norms, and social identities. Together, both of these chapters do an
especially nice job of articulating new connections between existing bodies of
theory.
The third chapter focuses on the psychology of buying and selling. In
"Egocentric Empathy Gaps in Social Interaction and Exchange," Dunning, Van
Boven and Loewenstein explore how systematic biases in perception lead to
coordination problems for members of exchange and economic relations. Of
particular interest is the endowment effect, a phenomenon that occurs when
sellers overestimate the value of an object just because they own it. Dunning
and colleagues describe a fascinating body of experimental research on
endowment effects, and trace the broader implications of such effects for social
exchange theory and social policy. Focusing more specifically on one-shot
exchange relations, Snijders and Keren examine the individual and situational
factors that lead to trust in their chapter entitled "Do you trust? Whom do you
trust? When do you trust?" They review a substantial number of experiments in
which individuals play trust games under a variety of conditions. They ask if
characteristics of the trustor, features of the trustee, or situational parameters
have the greatest predictive power. An important social psychological finding
is that features of the trustor or trustee appear to matter little; situational
features are the best predictors of trust.
xi
The next two chapters address provocative questions that are seldom
examined in sociological social psychology. First, David Sally asks how the
study of autistic individuals can advance formal understandings of the social
mind . Because autism disrupts the ability to take the role or perspective of
another person, this condition provides a natural experiment for comparing
those who can role-play to those who cannot . In his chapter "Into the Looking
Glass : Discerning the Social Mind Through the Mindblind" Sally asserts that
an adequate theory of the mind is essential to a variety of intrapersonal and
group processes . Next, Satoshi Kanazawa asks "Where do Social Structures
Come From?" in his chapter of the same title . Kanazawa argues that the
emerging field of evolutionary psychology can help explain the origins of
social structures and networks . He then provides two empirical tests by
examining differences between male and female kinship networks . Data from
the 1985 General Social Survey are shown to be consistent with the
evolutionary psychology approach .
The final two chapters address fundamental questions of status organizing
processes . In "Processing Performance Evaluations in Homogeneous Task
Groups : Feedback and Gender Effects," Martha Foschi, Sandra Enns, and
Vanessa Lapointe investigate emergent status structures in homogeneous task
groups . They report data from a six condition experiment that manipulated the
sex of dyad and level feedback for a gender-neutral task . As predicted by
expectation states theory, the findings indicate task feedback is significantly
related the rejection of influence . Although gender differences were not found
to significantly impact behavior, they were related to a number of self-report
measures . Finally, James W. Balkwell reanalyzes data from a classic study in
his chapter entitled "The Camilleri-Berger Model Revisited ." In this chapter
Balkwell applies modem statistical and computation technology to data
collected by Santo Camilleri and Joseph Berger more than 30 years ago and
finds surprisingly supportive results for expectation states theory . Balkwell
shows how theoretical concepts important to social exchange theorizing (such
as sentiment and control) can be incorporated into formal models of status
processing . The broader significance of this work is to show how the social
exchange framework and expectation states theory complimentary one another
for a range of contemporary issues .
Shane R . Thye
Edward J . Lawler
Michael W. Macy
Henry A . Walker
Volume Co-Editors
FROM PROTOTYPICALITY TO POWER :

A SOCIAL IDENTITY ANALYSIS OF
LEADERSHIP
Michael A . Hogg
INTRODUCTION
This chapter describes a social-cognitive analysis of how prototypical group
members can become leaders . The analysis is framed by the social identity
perspective, and the generative processes of social categorization, depersonalization, self-enhancement and uncertainty reduction that are believed to be
associated with collective self-conception, group membership, and group and
intergroup behavior. An embryonic version of some of these ideas appeared in
a chapter on structural differentiation within groups (Hogg, 1996), and was
developed as part of a larger article on social identity processes in
organizational contexts (Hogg & Terry, 2000), and as part of a chapter on
leadership and power (Hogg & Reid, in press) . There is also a number of
published empirical articles directly testing key hypotheses (Fielding & Hogg,
1997 ; Hains, Hogg & Duck, 1997 ; Hogg, Hains & Mason, 1998) . Other studies
are currently underway . The present chapter is an opportunity to present a more
elaborated social identity theory of group leadership .
In many ways this chapter is symptomatic of a revival of social
psychological interest in the social group - a revival that rests on a growing
integration of research on intergroup relations, small group processes, and
social cognition (e.g . Abrams & Hogg, 1998, 1999 ; Hogg & Tindale, 2001 ;
Moreland, Hogg & Hains, 1994 ; Levine, Resnick & Higgins, 1993 ; McGrath,
Advances in Group Processes, Volume 18, pages 1-30 .
Copyright 2001 by Elsevier Science Ltd.
All rights of reproduction in any form reserved .
ISBN : 0-7623-0767-6
1
MICHAEL A. HOGG
1997) . I set the scene by saying a few words about leadership research in social
psychology, and then a few words about the social identity approach . The bulk
of the chapter describes how social identity processes may be implicated in
leadership. I then overview relevant empirical work from my own lab and from
other groups, and go on to speculate about some conceptual and applied
implications of this analysis, in particular focusing on the relationship between
leadership and power .
The key point I will be making is that under certain circumstances,
specifically when group membership is an important basis of self-conception,
people tend to base their perceptions and evaluations of leadership effectiveness on the extent to which a person possesses prototypical properties of group
membership, rather than, say, effective leadership qualities . This implies that to
be an effective leader in such groups, leaders need to pay attention to how
prototypical they are considered to be by members of the group . The analysis
is intended to apply to groups of all sizes and functions, and to emergent as
well as established leaders . Whatever the conditions of group life, prototypicality becomes a strong influence on the dynamics of leadership when
group membership is highly salient and self-conceptually important . The
analysis itself, however, is primarily described in the context of emergent
leadership in relatively transient groups - and the experiments described below
which are specifically designed to investigate the core premises of the theory
are similarly oriented .
RESEARCH ON LEADERSHIP
Allport's classic definition of social psychology as the "scientific study of how
the thoughts, feelings and behavior of individuals are influenced by the actual,
imagined and implied presence of others" (Allport, 1935) places the topic of
leadership center stage for the discipline . Indeed, leadership was an important
focus of research for many years, particularly during the era of small group
dynamics (e .g . Cartwright & Zander, 1968 ; Shaw, 1981) . Leadership was a
component of some of social psychology's classic research programs ; for
example, Bales, 1950 ; Hollander, 1958 ; Lippitt & White, 1943 ; Sherif, 1966 ;
Stogdill, 1974 . This tradition of leadership research culminated in Fiedler's
(1965, 1971) contingency theory - the leadership effectiveness of a particular
behavioral style is contingent on the favorability of the situation to that
behavioral style.
The 1970s witnessed a new emphasis in social psychology on attribution
processes, that led into the 1980s' hegemony of social cognition (e .g . Devine,
Hamilton & Ostrom, 1994 ; Fiske & Taylor, 1991) . There was a well
Social Identity And Leadership
documented decline in interest in groups (e .g . Steiner, 1974, 1986), that carried

across to the study of leadership . The last edition of the Handbook of Social
Psychology had a chapter dedicated to leadership (Hollander, 1985), whereas
the current edition (Gilbert, Fiske & Lindzey, 1998) does not . This is not to say
that group processes and leadership are not scientifically researched, just that
interest within social psychology has waned. The study of small group
processes and of leadership has shifted to neighboring disciplines, most notably
organizational psychology (Levine & Moreland, 1990, 1995 ; McGrath, 1997 ;
Sanna & Parks, 1997 ; Tindale & Anderson, 1998) .
Social Cognition and Leader Categorization Theory
One important exception is Lord and colleagues' leader categorization theory
(e.g . Lord, Foti & DeVader, 1984 ; Nye & Forsyth, 1991 ; Palich & Hom, 1992 ;
Rush & Russell, 1988 ; see also Nye & Simonetta, 1996) . Based on implicit
leadership theory (Hollander & Julian, 1969) and on contemporary social
cognition principles, leader categorization theory states that people have
preconceptions about how leaders should behave in general and in specific
leadership situations . These preconceptions are cognitive schemas of types of
leader (i .e . categories of leader that are represented as person-schemas) that
operate in the same way as other schemas . When someone is categorized on the
basis of their behavior as a leader, the relevant leadership schema comes into
operation to generate further assumptions about behavior . Leadership schemas
vary in situational inclusiveness . Subordinate schemas apply only to specific
situations, whereas superordinate schemas apply to a wide range of situations
and embody very general leadership characteristics . Effective leaders are
people who have the attributes of the category of leader that fits situational
requirements .
This perspective treats leader categories as nominal categories ; that is,
cognitive groupings of instances that share attributes, but do not have any
psychological existence as a real human group . The notion of a social group
composed only of leaders makes little sense ; who would lead and who would
follow? Leadership is viewed as a product of individual information
processing, not as a structural property of real groups nor as an intrinsic or
emergent property of psychological ingroup membership .
Organizational Psychology and New Leadership Research
The main action, however, in leadership research is in organizational
psychology or in organizational settings (e .g . Bass, 1990a, 1998 ; Bryman,
MICHAEL A . HOGG
1992 ; Fiedler & House, 1994 ; Pawar & Eastman, 1997 ; Wilpert, 1995 ; Yukl &
van Fleet, 1992) . The basis of this tradition is the view that leadership is a
dynamic product of transactions between leaders and followers (Bass, 1990b ;
Hollander, 1985 ; Lord & Maher, 1991 ; Nye & Simonetta, 1996) . Because
leaders play a significant role in helping followers achieve their goals,
followers bestow power and status on leaders to restore equity. Relatedly,
followers may try to redress the power imbalance in groups by gaining personal
information about the leader - an attribution process that imbues the leader
with charisma and additional power (Fiske, 1993 ; Fiske & Depret, 1996) .
Leaders may also accumulate "idiosyncrasy credit" with the group by
conforming to group norms - subsequently allowing them to be innovative and
effective leaders (Hollander, 1958 ; Hollander & Julian, 1970) .
Recent transactional leadership perspectives focus on transformational
leadership . Charismatic leaders are able to motivate followers to work for
collective goals that transcend self-interest and transform organizations (Bass,
1990b ; Bass & Avolio, 1993 ; see Mowday & Sutton, 1993, for critical
comment) . This focus on `charisma' is particularly evident in `new leadership'
research (e .g. Bass, 1985, 1990b, 1998 ; Bryman, 1992 ; Bums, 1978 ; Conger &
Kanungo, 1987, 1988) which proposes that effective leaders should be
proactive, change-oriented, innovative, motivating and inspiring, and have a
vision or mission with which they infuse the group . They should also be
interested in others, and be able to create commitment to the group, and extract
extra effort from and generally empower members of the group .
Commentary on the Condition of Leadership Research
The recent study of leadership has largely been conducted outside mainstream
social psychology, and so it has not fully benefitted from some of the
conceptual advances made within social psychology over the last 20 years particularly the development and emerging synthesis of social cognition, group
processes, and intergroup relations research . Although most perspectives on
leadership now acknowledge that leadership is a relational property within
groups (i .e . leaders exist because of followers, and followers exist because of
leaders), the idea that leadership may emerge through the operation of ordinary
social-cognitive processes associated with psychologically belonging to a
group, has not really been elaborated .
Instead, the most recent analytic emphasis is upon : (a) individual cognitive
processes that categorize individuals as leaders - the social orientation between
individuals is not considered, and thus group processes are not incorporated, or
(b) whether individuals have the charismatic properties necessary to meet the
transformational objectives of leadership - leadership is a matter of

situationally attractive individual characteristics rather than group processes .
Both these perspectives have attracted criticism for neglecting the effects of
larger social systems within which people are embedded (e .g . Hall & Lord,
1995 ; Lord, Brown & Harvey, 2001 ; Pawar & Eastman, 1997; see also
Chemers, 2001) . Lord, Brown and Harvey (2001) explain that leadership
cannot be properly understood in terms of a leader's actions, or in terms of
abstract perceptual categories of types of leader, and that a paradigm shift in
how we understand leadership is called for - they pursue a connectionist, or
parallel constraint satisfaction, model .
The approach taken in this chapter is a little different . It integrates two
notions : (a) leadership is a relational property - leaders and followers are
interdependent roles embedded within a social system bounded by common
group/category membership ; and (b) leadership is a process of influence that
enlists and mobilizes others in the attainment of collective goals - it imbues
people with the group's attitudes and goals, and inspires them to work towards
achieving them (e .g . Chemers, 2001 ; Lord, Brown & Harvey, 2001) . The
integration is tackled from a social identity perspective that considers a group
to exist psychologically when people share a collective self definition . This
approach has said much about influence within psychologically salient groups,
but has said less about role differentiation and leadership .
SOCIAL IDENTITY AND SELF-CATEGORIZATION

The social identity perspective has recently been reviewed in detail elsewhere
(e .g. Hogg, 2001 ; Hogg & Terry, 2000; Turner, 1999 ; see also Hogg & Abrams,
1988) . The perspective contains a number of compatible and inter-related
components and emphases ; in particular an original emphasis by Tajfel and
Turner and their associates on social identity, social comparison, intergroup
relations, and self-enhancement motivation (often simply called social identity
theory ; e .g . Tajfel, 1972 ; Tajfel & Turner, 1979 ; Turner, 1982), and a later
cognitive emphasis by Turner and his associates on the categorization process
(called self-categorization theory ; e .g . Turner, 1985 ; Turner, Hogg, Oakes,
Reicher & Wetherell, 1987) .
Research and conceptual developments have focused on self-esteem
motivation (e .g . Abrams & Hogg, 1988), uncertainty reduction motivation
(Hogg, 2000a; Hogg & Mullin, 1999), optimal distinctiveness motivation
(Brewer, 1991), social comparison processes (e .g . Turner, 1975 ; Hogg 2000b),
identity salience processes (Oakes, Haslam & Turner, 1994), stereotyping
(Oakes, Haslam & Turner, 1994), social influence (e .g . Turner, 1991), cohesion
MICHAEL A . HOGG
(Hogg, 1992, 1993), collective action (Reicher, 1982, 2001), language and
ethnicity (e.g . Giles & Johnson, 1987), social belief structures (e .g . Ellemers,
1993), attitudes, norms and behavior (e .g . Terry & Hogg, 1996), deviance (e .g .
Marques, Abrams, Paez & Hogg, 2001), performance motivation (Fielding &
Hogg, in press ; Worchel, Rothgerber, Day, Hart & Butemeyer, 1998), and role
identities (e .g . Hogg, Terry & White, 1995) . There is also a number of recent
edited books (e.g . Abrams & Hogg, 1999 ; Capozza & Brown, 2000 ; Ellemers,
Spears & Doosje, 1999 ; Hogg & Terry, 2001 ; Spears, Oakes, Ellemers &
Haslam, 1997 ; Terry & Hogg, 2000 ; Worchel, Morales, Paez & Deschamps,
1998) . Not surprisingly, the social identity perspective continues to play an
important role in the revival of interest among social psychologists in the study
of groups (Abrams & Hogg, 1998 ; Hogg & Abrams, 1999 ; Moreland, Hogg &
Hains, 1994) .
According to the social identity perspective, people define and evaluate
themselves and others in terms of the groups they belong to : group
memberships define the collective self-concept and thus people's social
identity . Because social identity is evaluative, intergroup relations is a
competitive struggle for evaluatively positive intergroup distinctiveness and
hence positive social identity. The specific strategies that groups and their
members adopt depend on perceptions of intergroup status differences and the
stability and legitimacy of such differences, as well as the permeability of
intergroup boundaries and thus the feasibility of redefining oneself as a
member of a higher status group . The struggle for positive social identity is a
group level manifestation of an underlying human motivation to maintain a
positive sense of self-esteem .
The process underlying social identification is social categorization . Social
categorization segments the social world into ingroups and outgroups that are
cognitively represented as prototypes : context-specific fuzzy sets that define
and prescribe attitudes, feelings, and behaviors that characterize one group and
distinguish it from other groups . Prototypes are stored in memory to be
engaged by social categorization in a particular context to guide perception,
self-conception and action . However they are contextually responsive, and can
even be entirely constructed in situ . The principle governing context sensitivity
is meta-contrast . New prototypes form, or existing ones are modified, in such
a way as to maximize the ratio of perceived intergroup differences to intragroup
differences ; prototypes form to accentuate differences between categories and
similarities within categories .
Social categorization of other people perceptually assimilates them to the
relevant ingroup or outgroup prototype, and thus perceptually accentuates
prototypical similarities among people in the same group and prototypical
differences between people from different groups - a process that perceptually

differentiates groups, and renders perceptions, attitudes, feelings and behaviors
stereotypical and group normative . People are not viewed as unique and
multifaceted individuals but as more or less exact matches to the relevant
ingroup or outgroup prototype . Prototypicality, not individuality, becomes the
focus of attention - a process of depersonalization that refers to change in the
basis of perception, but does not have the negative connotations of terms such
as "deindividuation" or "dehumanization" . Social categorization of self, selfcategorization, has the same effect but more so. It not only depersonalizes
self-perception, but goes further in actually transforming self-conception and
assimilating all aspects of one's attitudes, feelings and behaviors to the ingroup
prototype ; it changes what people think, feel and do .
Social categorization of self and others satisfies an epistemic/self-evaluative
motive to reduce subjective uncertainty (the uncertainty-reduction hypothesis e .g . Hogg, 2000a ; Hogg & Mullin, 1999) . Situational or more enduring
subjective uncertainty, particularly relating to self-conceptualization, motivates
social identification and group formation, particularly with high entitativity
groups (e .g . Campbell, 1958 ; Hamilton & Sherman, 1996) that are distinctive
and have simple, consensual, and prescriptive prototypes . These prototypes
best resolve uncertainty about what to do, what others will do, and about who
one is . Uncertainty reduction and self-enhancement provide the motivational
parameters for the cognitive processes underlying the contextual salience of
self-categorization . People, influenced by self-enhancement and uncertainty
reduction motives, categorize the social context in terms of categories that are
chronically accessible in memory and/or rendered accessible by the immediate
context. That categorization becomes salient which best accounts for relevant
similarities and differences among people in the context (structural/comparative fit), which best accords with the social meaning of the context (normative
fit), and which best satisfies self-enhancement and self-evaluative concerns .
SOCIAL IDENTITY MODEL OF LEADERSHIP

The proposed social identity model of leadership is predicated on a
characterization of leadership as a group process . Effective leadership involves
identifying or defining group goals and being able to inspire followers to
internalize these goals as their own, and to work towards their achievement
(e .g . Chemers, 2001) . Leaders have disproportionate influence over the
attitudes, behaviors and destiny of group members .
MICHAEL A. HOGG
Influence of the Prototype

Where a group is contextually or enduringly salient, members categorize
themselves in terms of the ingroup prototype . Self-categorization-based
depersonalization produces normative behavior - conformity to the ingroup
prototype . In a highly salient group the prototype is likely to be relatively
consensual, and thus the group as a whole appears to be influenced by a single
prototype which prescribes a single norm or goal . Social identity research on
conformity and social influence shows that self-categorization produces
conformity to an ingroup prototype that may capture the central tendency of the
group or may be polarized away from a relevant outgroup (for reviews see,
Abrams & Hogg, 1990 ; Turner, 1991 ; Turner & Oakes, 1989) .
Relative Prototypicality and Influence
Within a salient group some members are more prototypical than others - there
is a prototypicality gradient that, in extreme cases, defines some people as
prototypically marginal and others as prototypically central (e .g . Marques &
Paez, 1994) . Because depersonalization is based on prototypicality, group
members are highly sensitive to prototypicality. Prototypicality is the basis of
perception and evaluation of self and other group members ; thus people notice
and respond to subtle differences in how prototypical fellow members are they are very aware of the prototype, and also of who is most prototypical and
of how prototypical others are (e.g . Haslam, Oakes, McGarty, Turner &
Onorato, 1995 ; Hogg, 1993) .
Within a salient group, then, people who are perceived to occupy the most
prototypical position are perceived to best embody the behaviors to which
other, less prototypical, members are conforming . There is a perception of
differential influence within the group, with the most prototypical member
appearing to exercise influence over less prototypical members . This
"appearance" probably arises due to the human tendency to personify and give
human agency to abstract forces - perhaps a manifestation of the fundamental
attribution error (Ross, 1977) or correspondence bias (e .g . Gilbert & Malone,
1995) . In new groups, this is only an "appearance" because the most
prototypical person does not actively exercise influence; it is the prototype,
which he or she happens to embody, that influences others' behavior . In
established groups the appearance is reinforced by actual influence (see
below) .
Where the social context is in flux, the prototype will likewise be in flux . As
the prototype changes so will the person who appears to be most prototypical
and thus most influential . However, under conditions of enduring contextual

stability the same individual may occupy the most prototypical position over a
long period, and so appear to have enduring influence over the group . In new
groups this person will be perceived to occupy an embryonic leadership role ;
although leadership has not been exercised . There is nascent role differentiation
into "leader" and "followers" .
So far, social identity processes ensure that as group membership becomes
more salient, and members identify more strongly with the group, prototypicality becomes an increasingly influential basis for leadership perceptions .
However it is important to keep this in perspective - prototypicality is not the
only basis of leadership . People also rely on general and more task-specific
schemas of leadership behaviors (what Lord and his colleagues call leader
categories - e.g . Lord, Foti & DeVader, 1984) . However, the importance of
these schemas is either unaffected by self-categorization, or it diminishes as
group prototypicality becomes more important . In either case, leadership
schemas should become less influential relative to group prototypicality as
group membership becomes psychologically more salient .
Social Attraction
Social categorization affects not only perceptions, but also feelings, about other
people . Social identification transforms the basis of one's liking for others from
idiosyncratic preference and personal relationship history (personal attraction)
to prototypicality (social attraction) - ingroup members are liked more than
outgroup members and more prototypical ingroupers are liked more than less
prototypical ingroupers . Where there is a relatively consensual ingroup
prototype, social categorization renders more prototypical members socially
popular - there is consensual and unilateral liking for more prototypical
members . This depersonalized social attraction hypothesis (Hogg, 1992, 1993)
is supported by a series of laboratory and field studies (e .g . Hogg, CooperShaw & Holzworth, 1993 ; Hogg & Hains, 1996, 1998 ; Hogg & Hardie, 1991 ;
Hogg, Hardie & Reynolds, 1995) .
From the point of view of leadership, the person occupying the most
prototypical position may thus acquire, in new groups, or possess, in
established groups, the ability to actively influence because he or she is socially
attractive and thus able to secure compliance with suggestions and recommendations he or she makes . A well researched consequence of liking is that
it increases compliance with requests . If you like someone you are more likely
to agree with them, and comply with requests and suggestions (e .g . Berscheid
10
MICHAEL A. HOGG
& Reis, 1998) . In this way, the most prototypical person is able to actively
exercise leadership by having his or her ideas accepted more readily and more
widely than ideas suggested by others . This empowers the leader, and publicly
confirms his or her ability to exercise influence . Consensual depersonalized
liking, particularly over time, confirms differential popularity and public
endorsement of the leader. It imbues the leader with prestige and status, and
begins to reify the nascent intragroup status differential between leader(s) and
followers . It allows someone who is "merely" prototypical, a passive focus for
influence, to take the initiative and become an active and innovative agent of
influence . In the case of established leaders, it allows them to be more
effectively innovative and influential .
Social attraction may also be enhanced by the behavior of highly
prototypical members . More prototypical members tend to identify more
strongly, and thus display more pronounced group behaviors ; they are more
normative, show greater ingroup loyalty and ethnocentrism, and generally
behave in a more group serving manner . These behaviors confirm prototypicality and thus enhance social attraction . A leader who acts as "one of us",
by showing strong ingroup favoritism and intragroup fairness, is not only more
socially attractive, but is also furnished with legitimacy. Research on justice
considerations in group contexts confirms that although distributive justice is
complicated (intragroup fairness conflicting with intergroup bias), intragroup
procedural justice is critical . According to the group value model of procedural
justice, members feel more satisfied and more committed to the group if the
leader is procedurally fair (Tyler, 1997 ; Tyler & Lind, 1992 ; see Platow, Reid
& Andrew, 1998) .
Attribution and Information Processing
Prototypicality and social attraction work alongside attribution and information
processing to translate perceived influence into active leadership . Attribution
processes operate within groups to make sense of others' behavior. As
elsewhere, attributions for others' behavior are prone to the fundamental
attribution error (Ross, 1977) or correspondence bias (Gilbert & Jones, 1986 ;
see also Gilbert & Malone, 1995 ; Trope & Liberman, 1993) ; a tendency to
attribute behavior to underlying dispositions that reflect invariant properties, or
essences, of the individual's personality . This effect is more pronounced for
individuals who are perceptually distinctive (e .g . figural against a background)
or cognitively salient (e .g . Taylor & Fiske, 1978) .
11
We have seen that when group membership is salient, people are sensitive to
prototypicality and attend to subtle differences in prototypicality of fellow
members . Highly prototypical members are most informative about what is
prototypical of group membership (see Turner, 1991), and so in a group context
they attract most attention . They are subjectively important and are distinctive
or figural against the background of other less informative members . Research
in social cognition shows that people who are subjectively important and
distinctive are seen to be disproportionately influential and have their behavior
dispositionally attributed (e .g . Erber & Fiske, 1984 ; Taylor & Fiske, 1975) . We
have also seen how highly prototypical members may appear to have influence
due to their relative prototypicality, and may actively exercise influence and
gain compliance as a consequence of consensual social attraction .
Together, the leadership nature of this behavior and the relative prominence
of prototypical members may encourage an internal attribution to intrinsic
leadership ability, or charisma. In this analysis charisma is certainly not a
personality attribute that causes leadership . It is an emergent perception on the
part of followers, based on the operation of social-cognitive processes
associated with group membership .
In groups, then, the behavior of highly prototypical members is likely to be
attributed, particularly in stable groups over time, to the person's personality
rather than the prototypicality of the position occupied. The consequence is a
tendency to construct a charismatic leadership personality for that person
which, to some extent, separates that person from the rest of the group and
reinforces the perception of status-based structural differentiation within the
group into leader(s) and followers . This may make the leader stand out more
starkly against the background of less prototypical followers, as well as draw
attention to a potential power imbalance ; thus further fueling the attributional
effect.
There is some empirical support for the idea that followers tend to focus
upon the leader and make dispositional attributions for that person's behavior .
Fiske (1993 ; Fiske & Depret, 1996) shows how followers pay close attention to
leaders, and seek dispositional information about leaders because detailed
individualized knowledge helps redress the perceived power imbalance
between leader and followers . Conger and Kanungo (1987, 1988) describe how
followers attributionally construct a charismatic leadership personality for
organizational leaders who have a "vision" that involves substantial change to
the group . Meindl, Ehrlich, and Dukerich (1985) showed that simplified
dispositional attributions for leadership were more evident for distinctive
leadership behaviors, and under crisis conditions .
12
MICHAEL A . HOGG
Maintaining Leadership
Thus far we have seen how prototype-based depersonalization fairly automatically imbues the most prototypical member of a group with many attributes of
leadership - for example, status, charisma, popular support, and the ability to
influence . These attributes also allow the leader to actively maintain his or her
leadership position . The longer an individual remains in a leadership position
the more they will be socially `liked', the more consensual will social attraction
be, and the more entrenched will be the fundamental attribution effect (but see
below) .
Social contextual changes impact prototypicality - the prototype can change
its content or its consensuality . Thus, over time and across contexts, the leader
may decline in prototypicality while other members become more prototypical ;
opening the door, particularly under high salience conditions, to a redistribution
of influence within the group . An established leader is well placed in terms of
resources to combat this by redefining the prototype in a self-serving manner
to prototypically marginalize contenders and prototypically centralize self . This
can be done by accentuating the existing ingroup prototype, by pillorying
ingroup deviants, or by demonizing an appropriate outgroup . Generally all
three tactics are used, and the very act of engaging in these tactics is often
viewed as further evidence of effective leadership .
Leadership endurance also benefits from consensual prototypicality, because
of the latter's effect on social attraction . In groups with less consensual
prototypes, there is greater dissensus of perceptions of and feelings for the
leader and thus the leader may have less power and may occupy a less stable
position . It is in the leaders's interest to combat proliferation of multiple
conflicting prototypes, and to maintain a consensual prototype . Simple and
more clearly focused prototypes are less open to ambiguity and alternative
interpretations and are thus better suited to consensuality . In addition, ingroup
deviants serve an important function ; by creating and rejecting such deviants
the leader is well able to clarify the self-serving focus of the prototype . Another
strategy is to polarize or extremitize the ingroup relative to a specific "evil"
outgroup . These processes operate in extremist groups with all-powerful
leaders .
EMPIRICAL TESTS OF THE ROLE OF

PROTOTYPICALITY IN LEADERSHIP
The core idea of the social identity model of leadership is that under conditions
of group salience prototypicality becomes the basis of leadership . Automatic
13
processes associated with self-categorization as a group member transform the

most prototypical group member into the group leader, or in the case of
established leaders imbue them with superior perceived leadership . Before
examining additional ramifications of this idea, it is worthwhile at this juncture
describing direct tests of the central prediction that as people identify more
strongly with a group the basis for leadership perceptions, evaluations, and
endorsement becomes increasingly influenced by prototypicality ; prototypical
members are more likely to emerge as leaders, and more prototypical leaders
will be perceived to be more effective leaders .
Hains, Hogg and Duck (1997) conducted a laboratory study of emergent
leadership perceptions and evaluations in ad hoc and relatively minimal groups .
Three independent variables (group salience, group prototypicality, and the
extent to which the leader's qualities were congruent with people's general
schema of good leadership) were manipulated in a 2 x 2 x 2 design. Under
conditions of high or low group salience, participants anticipated joining a
small discussion group formed on the basis of attitude congruence . They were
informed that a randomly appointed group leader was group prototypical or
non-prototypical (group prototypicality) in terms of the attitude dimension, and
had a behavioral style (on the basis of a pretest) that was congruent or
incongruent with a leader schema (leader schema congruence) . Dependent
measures were taken ostensibly in anticipation of the upcoming discussion . In
addition to checks on each of the three manipulations, we also measured group
identification (11-item scale) and perceived leader effectiveness (10-item
scale) . As predicted, when group membership was salient people identified
more strongly with the group, and endorsed the prototypical leader as being
much more effective than the non-prototypical leader ; low salience participants
did not differentiate between prototypical and non-prototypical leaders .
Although leader schema congruent leaders were perceived overall to be more
effective than schema incongruent leaders, we found that this effect
disappeared for high salience participants on one leadership effectiveness item .
Although social attraction for the leader was not explicitly tested, the 10-item
leadership effectiveness scale contained an item measuring liking for the
leader ; thus leadership effectiveness was associated with liking.
To complement this controlled laboratory experiment, we conducted a field
study of leadership in small interactive òutward bound' groups where real
leaders emerged and actually lead the groups in wilderness and outdoor
experiences (Fielding & Hogg, 1997) . There were 13 mixed-sex, approximately 11-person groups of people mainly in their 20s from around Australia .
The groups stayed together for three weeks . We replicated the laboratory
experiment as closely as we could, but of course in a measurement-based
14
MICHAEL A . HOGG
regression format . Leadership schemas, group membership variables, and

leadership effectiveness perceptions were measured a week to 10 days apart . In
this study we were also able to measure social attraction . As predicted, (a)
group identification, perceived leadership effectiveness and social attraction for
the leader increased over time as the group became a more cohesive entity, and
(b) perceived leadership effectiveness was a positive function of social
attraction for and group prototypicality of the leader, and this was amplified
among high identifying participants . Perceived leader schema congruence of
the leader was a predictor of perceived leadership effectiveness, but was
uninfluenced by identification .
We now returned to the laboratory to conduct two minimal group studies
based closely on Hains et al.'s (1997) methodology (Hogg, Hains & Mason,
1998) . The aim of these somewhat complex studies was to treat prototypicality
and leadership as relativistic properties of a comparative frame of reference in
which individuals are perceived and evaluated in relation to other individuals
who are ingroup or outgroup members . The first study had participants
nominate a leader for a small high salience discussion group they were
ostensibly going to join . They were provided with a carefully constructed
transcript of an earlier meeting of the group, that provided leadership schema
congruence and group prototypicality information for each of the other
members . They rated the group, their nominated leader and all other group
members, and their identification with the group . The second study was a
2 x 2 x 2 experiment in which group salience, leadership schema congruence of
the leader, and group prototypicality of the leader were orthogonally
manipulated . The key feature of this experiment was that the prototypicality of
the leader was indirectly manipulated by constructing an intergroup comparative context that, on the basis of metacontrast, influenced the location of the
prototype in precisely calculable ways .
Across these two studies we found that leadership schema congruence
became a less influential, and group prototypicality a more influential,
determinant of leadership endorsement in more cohesive groups with which
people identified more strongly . We also found, as expected, that identification
accentuated perceived prototypical similarities between ingroup members and
thus between leader and non-leaders, and that this weakened the prototypeleadership relationship when leadership and prototypicality were measured
relativistically (leaders relative to non-leaders) . We argued, however, that this
effect was due to the ad hoc and emergent nature of the group, and that over
time the perceptual gulf between leader and followers, described by the theory
(see below), would gradually develop .
15
Platow and van Knippenberg (1999) have recently replicated the finding
from Hains, Hogg and Duck (1997) that prototypicality becomes an
increasingly influential basis for leadership endorsement as group membership
becomes more salient . Duck & Fielding (1999), drawing directly on the social
identity theory of leadership, conducted two laboratory experiments which, in
a relatively minimal way, simulated equal status subgroups nested within a
larger organization . They measured group identification and evaluations of
organization leaders who were randomly appointed from participants' own or
the other subgroup . Ingroup, thus prototypical, leaders were more strongly
supported than outgroup, thus non-prototypical, leaders, and this effect was
more pronounced to the extent that participants identified strongly with their
own subgroup . Again, the basis of leadership perception and endorsement is
more firmly grounded in prototypicality as people identify more strongly with
the group .
Platow, Reid and Andrew (1998) provide some indirect support for the
leadership theory from a laboratory experiment in which they manipulated
group salience (interpersonal vs . intergroup context), and whether a randomly
appointed leader was procedurally fair/unfair and distributively fair/unfair . If it
is assumed that fairness is a general property of leadership schemas, but that
ingroup favoritism is a generally prototypical and socially attractive property of
group membership, then we would predict that distributively and procedurally
ingroup favoring leaders would be more strongly endorsed under high than low
salience conditions . This is precisely what Platow and his colleagues found .
Haslam, McGarty, Brown, Eggins, Morrison and Reynolds (1998) report
three experiments that support the idea that systematically selected leaders may
be less favorably perceived than randomly appointed leaders . They argue, from
social identity theory, that this may be because a systematic selection process
draws attention away from the group and towards individuality . It personalizes
the leader and separates him or her from the group, and thus renders the leader
perceptually less prototypical . It may even also weaken group identification . In
contrast a random process keeps attention on the group as a whole . It does not
personalize the leader, and thus allows him or her to be viewed as a prototypical
group member . It may also strengthen group identification . Again, prototypicality becomes an increasingly important basis of leadership as group
membership becomes more salient .
Finally, there are studies conducted within a social dilemmas tradition .
Drawing on social identity theory and on Tyler and Lind's (1992) group-value
model, van Vugt and de Cremer (1999) conducted two experiments in which
they found that when people strongly identify with a group faced by a social
dilemma they prefer a leader who shares the group's values (i .e . is prototypical)
16
MICHAEL A . HOGG
and that such a leader will actually be more effective . Van Vugt and de Cremer
explicitly view this as supporting the social identity theory of leadership . In
another social dilemma study, de Cremer and van Vugt (in press) manipulated
a number of variables including how much participants identified with the
group and how much a randomly appointed leader ostensibly identified with
the group. Dependent measures focused mainly on cooperation with the leader
as a reflection of leader approval, support and effectiveness . The results showed
that members cooperated more with a high than low identifying leader, and that
this was particularly the case for members who identified strongly with the
group . In addition the effect was clearly mediated by social attraction . De
Cremer and van Vugt explicitly state that these data support the social identity
theory of leadership . Finally, Foddy and Hogg (1999) report some data of their
own that suggest that where there are leaders managing a scarce resource, those
leaders who identify more strongly with the group (and thus consider
themselves to be more prototypical) tend to be more conserving of the scarce
resource, and are thus more effective leaders .
In this section I have reviewed social identity research that directly tests or
indirectly tests the leadership theory presented here . There is consistent and
reliable support for the core idea that as groups become more salient and people
identify more strongly with them, prototypicality of the leader becomes an
increasingly significant basis for leadership perceptions . There is some
evidence that as prototypicality becomes more important, leadership schema
congruence becomes less important . There is also support for the idea that
prototype-based depersonalized social attraction may facilitate leadership .
There is some direct evidence from the studies by Fielding and Hogg (1997)
and de Cremer and van Vugt (in press), whereas in other studies social
attraction is a component of the leadership evaluation measure (e .g. Hains,
Hogg & Duck, 1997 ; Hogg, Hains & Mason, 1998) . The attribution and
associated structural differentiation components of the theory have indirect
support (e .g . Fiske, 1993 ; Fiske & Depret, 1996), but remain to be directly
tested.
LEADERSHIP AND POWER

The social identity analysis of leadership rests on the notion that the leader is
"one of us" - indeed the leader is the most prototypical group member ; the
person who embodies the integrative essence of the group . However, there are
also many instances of leadership where the leader is, or becomes, a relatively
remote and atypical group member who appears to be psychologically
distanced from the life of the group . From a social identity perspective, the
17
reconciliation of these contrasting situations hinges on the relationship between

ingroup roles and intergroup behavior, and between leadership and power (see
Hogg & Reid, in press) .
From Leadership to Power

Prototypical leaders do not need to exercise power over others (i .e . persuade,
gain compliance, coerce, or resort to force) to have influence ; they are
influential because of their position and the depersonalization process that
assimilates members' behavior to the prototype . They, and their suggestions,
are intrinsically persuasive because they embody the norms of the group ; they
have referent power (Raven, 1965) or position power, and therefore do not need
to exercise personal power (Yukl & Falbe, 1991) . Indeed, the influence process
associated with being a highly prototypical leader is the influence process
associated with social identification with a group - the process of referent
informational influence that produces normative behavior in salient groups (e .g .
Hogg & Turner, 1987 ; Turner, 1982) . Followers pay close attention to the
prototype and to those who are prototypical, and engage in systematic,
deliberative, and central route processing of information about the leader and
the leader's "message" . This produces internalized cognitive change - true
attitude change - in members (e .g . Mackie, 1987 ; Mackie & Queller, 2000) .
In addition to not "needing" to exercise power, it is possible that prototypical
leaders may be "unable" to exercise power . High prototypicality is associated
with strong ingroup identification ; self and group are tightly fused prototypically and thus any negative behavior directed against fellow members is
effectively directed against self . There may exist an empathic bond between
leader and followers that protects against the exercise of power over others (cf.
Cadinu & Rothbart, 1996 ; Simon & Hastedt, 1999 ; Smith & Henry, 1996) .
However, as hinted above, there is a paradox . Occupying a highly
prototypical position, particularly in an enduring and stable high entitativity
group with a focused and consensual prototype, makes one gradually appear
enduringly influential, consensually socially attractive, and essentially charismatic . There is a gradual perceptual separation of the leader from the rest of the
group, through structural role differentiation grounded in social attraction and
attribution processes - the leader is gradually perceived as "other" rather than
"one of us" . The person who embodied the essence of the group by being most
prototypical has now become effectively an outgroup member within the group .
An embryonic intergroup relationship begins to emerge between leader (along
with his/her inner clique) and followers .
18
MICHAEL A . HOGG
This intergroup relationship is grounded in a status differential that is

perceived to be relatively consensual, stable, and legitimate - a potent mix that
has potential for a competitive intergroup relationship between leader(s) and
followers, but a competitive intergroup relationship in which the leader has
most of the power. Although the seeds of autocracy are sown, they may not
germinate . Intergroup boundaries may be considered permeable, and the
relationship may still be construed as a mutually beneficial role relationship in
the service of superordinate, non-zero-sum goals - everyone is on the same
team, working for the same goals, but making different contributions to the
greater good of the group . The leader may not be "one of us", but he or she is
certainly working with us, and for us .
However, there are circumstances which may make potential power-based
intergroup behavior a reality . A relatively inevitable consequence of role
differentiation is that the leader gradually realizes that he or she is effectively
treated by followers as an outgroup member - a positive high status deviant, but
none the less a deviant who cannot readily share in the life of the group . The
leader may at this point try to veer away from the abyss by engaging in
behaviors aimed at confirming his or her ingroup prototypicality . If this is
unsuccessful, a sense of rejection by, and distance and isolation from, the group
may occur (possibly also a recognition of reduced influence among followers),
which may "embitter" the leader and, since the empathic bond is severed,
allows the leader to gain compliance through the exercise of power over others .
This may involve coercive behavior, because the interests of the leader and the
group have diverged - the leader is effectively exercising his or her will over
others . The influence process is one that involves coercion rather than attitude
change .
This effect is stronger in hierarchical extremist groups where the leaderfollower role and power differentiation is more tangible, stark, and
impermeable - the potential for coercion is much accentuated in these types of
groups . The effect will also be stronger in groups where there is a leadership
clique rather than a single leader, because a typical intergroup relationship has
effectively emerged and thus the relationship between leader(s) and followers
is an intergroup relationship where one group (the leaders(s)) has disproportionate legitimate power over the other group (the followers) . Such a
relationship will be competitive and potentially exploitative - far removed from
prototype based leadership .
Leaders generally react negatively to perceived threats to their leadership
position. Where a leader is prototypically influential and no intergroup
differentiation has yet emerged, threats to leadership largely come from
prototype slippage - social contextual factors may reconfigure the group
19
prototype and thus reduce the leader's prototypicality. We described above how
leaders then strive to redefine the prototype to better fit themselves - they can
accentuate the existing ingroup prototype, pillory ingroup deviants, or
demonize an appropriate outgroup . These tactics generally do not involve
coercion . However, where an intergroup differentiation is clearly evident,
perceived threats to leadership are automatically perceived in intergroup terms
as collective challenge/revolt on the part of the followers . This makes salient
the latent intergroup orientation between leader(s) and followers, and
engenders competitive intergroup relations between leader(s) and followers competitive relations in which one group has consensually legitimate and
overwhelming power over the other. Under these circumstances leadership
becomes coercion, based on the relatively limitless exercise of coercive power
over others . The dynamic is similar to the way in which a power elite "reacts'
to a perceived challenge to it's privileged position (e.g . Wright, 1997), but
because it occurs within the power-legitimizing framework of a common group
membership the "reaction" is potentially all the more extreme .
From Power to Leadership
The analysis in the previous section suggests a series of steps that transforms
prototype-based leadership into power-based leadership . Highly prototypical
leaders of salient groups, particularly newly-emerged leaders, provide leadership through influence - they do not need to exercise power over followers, and
indeed may not actually be able to behave in this way. Enduring tenure renders
leaders more influential and facilitates normative innovation - leaders still do
not need to exercise power over followers because they now have the capacity
to ensure that they remain prototypical and thus influential . Further tenure
differentiates the leader(s) from the followers . It creates an intergroup
differentiation based on widening, reified and consensually legitimized role and
power differences - the potential to use power is now very real . The conditions
that translate the potential into reality are ones that render salient the latent
power-based intergroup relationship between leaders and followers - for
example a sense of threat to one's leadership position, a feeling of remoteness
and alienation from the group, or a sense of becoming less influential in the life
of the group .
The exercise of leadership through coercion rests on the psychological
reality (based on self-categorization and social identity processes) of a sharp
role, status, and power discontinuity between leaders(s) and followers that
reconfigures cooperative intragroup role relations as competitive intergroup
relations . Such intergroup relations within a group provide ideal conditions for
20
MICHAEL A. HOGG
unilaterally exploitative intergroup behavior. This is because the overarching

common group identity and the diachronic process of leadership emergence
strongly legitimize the status quo - there exists what social identity theory
refers to as a social change belief structure without cognitive alternatives
(Tajfel & Turner, 1979 ; also see Hogg & Abrams, 1988) . Because power and
leadership are attractive to some people, this belief system can be coupled with
a belief in intergroup permeability that encourages followers to try to gain
admittance to the leadership clique - this, of course, marshals support for the
leader(s) and prevents the followers from forming a united front .
The transformation of prototype-based leadership into power-based leadership is by no means inevitable . Leadership through influence is psychologically
and materially less costly all round - it may be much better for the group (but
see below) . However, the challenge is that it is the group, not the leader, that
has to take the initiative in arranging conditions that contain power, and yet the
group is relatively powerless in the face of a leader who is wielding power in
oppressive ways. Nevertheless, anything that inhibits the attribution of
charisma and the process of structural differentiation, and which re-grounds
leadership in prototypicality will inhibit the exercise of power. This may
include quite contrasting conditions - on the one hand, reduced group
cohesion, reduced prototype consensuality, and increased diversity, and on the
other hand any external group threat that refocuses attention on common group
identity. Although the natural course of intergroup relations may create these
conditions, powerful leaders can protect themselves to some extent against
them . The processes may be complicated . For example, if a group becomes less
cohesive, more diverse, and less consensual about its prototype, it is less likely
that followers will agree on and endorse the same person as the leader . The
leader's power base is fragmented, and numerous new "contenders" emerge .
Although this limits the leader's ability to exercise power, it is a threatening
state of affairs, particularly for a leader who has been accustomed to exercise
power - powerful incumbent leaders are likely to "react" in draconian ways .
External threat can make the group so cohesive and consensual that leader
and group become re-fused and the empathic bond re-established - the leader
no longer needs, or indeed is able, to exercise power, particularly in destructive
ways. External threat may also focus the group on promotively interdependent
goals, with the consequence that followers do not grant status to leaders unless
leaders earn such status through an appropriate perceived contribution to group
goal achievement (e .g . Ridgeway, 2001 ; Ridgeway & Diekema, 1989) . Leaders
who exercise power in order to mis-appropriate a share of rewards will face a
resistant coalition of followers . Coercion becomes a less effective or viable
form of leadership - leaders need to reposition themselves to act as prototypical
21
group members who through being prototypical contribute more to the group's
goals than do less prototypical followers .
A DARK SIDE TO PROTOTYPE-BASED LEADERSHIP

Earlier I suggested that prototype based leadership might be better for groups .
This is not, however, always the case . For instance, in decision making groups,
prototype-based leadership can degrade decision making processes . As people
identify more strongly with a group (the group becomes more cohesive),
leadership becomes increasingly based on prototypicality rather than leader
schemas . Rather than basing leadership on leader schemas that generally
contain optimal situation and task specific leadership prescriptions, a situation
can exist where there is a powerful leader who embodies a group prototype that
does not prescribe optimal decision making procedures . This may produce
groupthink (Janis, 1972) ; powerful leaders and the absence of norms for
optimal decision making conspire in highly cohesive groups to produce
suboptimal decision making procedures that lead to poor decisions (Hogg &
Hains, 1998 ; Turner, Pratkanis, Probasco & Leve, 1992) .
Another pitfall of prototype based leadership is that social minorities (e .g .
based on race, ethnicity, gender, disability) may find it difficult to assume
leadership roles in some contexts . For example, if the normative environment
for business, or the organizational culture within specific organizations, renders
social minorities intrinsically less prototypical of the organization than
majorities, then minorities will find it more difficult to achieve and maintain an
effective leadership role (see Hogg & Terry, 2000) .
Finally, prototype-based leadership where the prototype is sharply focused
and consensual may be associated with high entitativity groups that are very
cohesive . These sorts of groups are well suited to subjective uncertainty
reduction through self-categorization (Hogg, 2000a ; Hogg & Mullin, 1999 ;
Reid & Hogg, 2000) - conditions of high societal or personal uncertainty may
motivate identification with these groups or the reconfiguration of existing
groups to be like this . These groups are likely to be "extremist" and prone to
hierarchical leadership structures, with remote and powerful leaders who are
invested with enormous charisma and who can exercise and abuse their position
of power. Under high self-conceptual uncertainty members strive for a simple
and distinct prototype, support witch-hunts to purify the group of deviants,
express consensual social attraction, are highly attuned to prototypicality, and
invest the leader with a highly charismatic leadership personality .
A good example of this is "totalist" groups such as cults (e.g . Curtis &
Curtis, 1993 ; Galanter, 1989) . In these groups, leaders are often distinct and
22
MICHAEL A . HOGG
remote from followers. There is a steep and rigid prestige differential between
leaders and followers that provides a clearly delineated intragroup status
structure . This arrangement contributes further to uncertainty reduction under
conditions of extreme uncertainty, because it imbues the intragroup structural
arrangement, and thus the leader, with perceived legitimacy that protects the
system from change (see Jost & Banaji's, 1994, system justification theory) .
Leaders may also maintain or strengthen their leadership position through
strategic management of uncertainty (see Marris's, 1996, notion that certainty
is power) . Specifically, they can deliberately raise uncertainty (e .g . the specters
of war, economic collapse, cultural disintegration), and at the same time define
a clear social identity predicated on a prototype that closely matches the leader .
Uncertainty reduction automatically endorses the leader .
CONCLUDING COMMENTS
In this chapter I have described a social identity model of leadership .
Leadership is a group process that arises from the social categorization and
depersonalization processes associated with social identity . Prototype-based
depersonalization and the behavior of followers play a critical role ; they
empower individuals as leaders, imbue them with charisma, create a status
differential between leader(s) and followers that has some of the typical
characteristics of uneven status intergroup relations, and set up conditions that
may promote the coercive use of power over others .
Aggregates of individuals mainly base their leadership perceptions and
endorsements on judgements of how well individuals fit the specifications of
situation and task specific schemas of types of leaders . However, as the
aggregate becomes a psychological group - a collection of people who
categorize themselves in terms of the same descriptive and prescriptive ingroup
prototype - leadership perceptions and endorsements are increasingly influenced by how well individuals match the group prototype . In very cohesive
groups with which members identify strongly, leadership processes will be
strongly determined by prototypicality . Prototypical members do not actively
lead but rather appear to have influence ; they act as the attitudinal and
behavioral focus of the group due to self-categorization based prototypical
depersonalization processes .
It is social attraction, an associated effect of depersonalization, that
facilitates active influence . Group membership based social attraction enables
the leader to secure compliance with suggestions and recommendations, and
also endows the leader with status and evaluative differentiation from the rest
of the group . The leader now occupies a high status role and is able to be
23
actively innovative . Finally, there is an attribution through which followers,

influenced by correspondence bias and the fundamental attribution error,
attribute the leader's influence, status, popularity and ability to innovate and
lead, internally to the leader's personality ; thus constructing a charismatic
leadership personality for the leader that further reifies the leader's distinctive
status within the group .
We reviewed empirical support for the model, in particular the key prediction
that as a group becomes more salient, emergent leadership processes and
leadership effectiveness perceptions become less dependent on leader schema
congruence and more dependent on group prototypicality . The final sections of
the chapter are more speculative in describing implications of the model - in
particular a group process through which prototype-based leadership through
influence can mutate into power-based leadership through coercion. We also
explored some of the pitfalls of prototype-based leadership - deficient group
decision making, exclusion of minorities from leadership, and hierarchical
leadership in "totalist" groups . These ideas provide a rich seam of hypotheses
to be mined empirically .
ACKNOWLEDGMENTS
This chapter was made possible by a research enabling grant from the
University of Queensland, and by research grant support from the Australian
Research Council . I would also like to thank Kelly Fielding, Margaret Foddy,
Sarah Hains, Leigh Morris, and Sherry Schneider for their intellectual
contributions to the development of some of the ideas presented in this
chapter.
REFERENCES
Abrams, D ., & Hogg, M. A. (1988) . Comments on the motivational status of self-esteem in social
identity and intergroup discrimination. European Journal of Social Psychology, 18,
317-334 .
Abrams, D ., & Hogg, M. A . (1990) . Social identification, self-categorization and social influence .
European Review of Social Psychology, 1, 195-228 .
Abrams, D ., & Hogg, M. A . (1998) . Prospects for research in group processes and intergroup
relations . Group Processes and Intergroup Relations, 1, 7-20.
Abrams, D., & Hogg, M. A . (Eds) (1999). Social Identity and Social Cognition . Oxford, U .K . :
Blackwell .
Allport, G. W. (1935) . Attitudes . In : C . M . Murchison (Ed .), Handbook of Social Psychology (pp .
789-844) . Worchester, MA : Clark University Press .
Bales, R . F. (1950) . Interaction Process Analysis: A Method for the Study of Small Groups .
Reading, MA : Addison-Wesley.
24
MICHAEL A . HOGG
Bass, B . M . (1985) . Leadership and Performance Beyond Expectations. New York: Free Press .
Bass, B . M. (1990a) . Bass and Stogdill's Handbook of Leadership : Theory, Research and
Managerial Applications . New York: Free Press .
Bass, B . M . (1990b) . From transactional to transformational leadership : Learning to share the
vision . Organizational Dynamics, 18, 19-3 1 .
Bass, B . M. (1998) . Transformational Leadership : Industrial, Military, and Educational Impact.
Mahwah, NJ : Erlbaum .
Bass, B . M ., & Avolio, B . J . (1993). Transformational leadership : A response to critiques . In : M .
M . Chemers & R. A . Ayman (Eds), Leadership Theory and Research : Perspectives and
Directions (pp . 49-80) . London : Academic Press .
Berscheid, E., & Reis, H. T. (1998). Attraction and close relationships . In : D . T. Gilbert, S . T. Fiske
& G. Lindzey (Eds), The Handbook of Social Psychology (4th edn, Vol.2, pp. 193-281) .
New York: McGraw-Hill .
Brewer, M . B . (1991) . The social self: On being the same and different at the same time .
Personality and Social Psychology Bulletin, 17, 475-482 .
Bryman, A. (1992) . Charisma and Leadership in Organizations . London: Sage .
Bums, J . M . (1978) . Leadership. New York : Harper & Row.
Cadinu, M . R., & Rothbart, M . (1996) . Self-anchoring and differentiation processes in the minimal
group setting . Journal of Personality and Social Psychology, 70, 661-677 .
Campbell, D. T. (1958) . Common fate, similarity, and other indices of the status of aggregates of
persons as social entities . Behavioral Science, 3, 14-25.
Capozza, D., & Brown, R . (Eds) (2000) . Social Identity Processes. London: Sage .
Cartwright, D ., & Zander, A . (Eds) (1968) . Group Dynamics: Research and Theory (3rd edn).
London : Tavistock .
Chemers, M . M. (2001) . Leadership effectiveness : An integrative review. In : M . A . Hogg & R. S .
Tindale (Eds), Blackwell Handbook of Social Psychology : Group Processes (pp . 376-399).
Oxford, U.K . : Blackwell .
Conger, J . A ., & Kanungo, R. N. (1987) . Towards a behavioral theory of charismatic leadership in
organizational settings . Academy of Management Review, 12, 637-647 .
Conger, J. A., & Kanungo, R . N . (1988) . Behavioral dimensions of charismatic leadership . In : J.
A . Conger, & R. N. Kanungo (Eds), Charismatic Leadership: The Elusive Factor on
Organizational Effectiveness . San Francisco, CA : Jossey-Bass .
Curtis, J . M ., & Curtis, M. J . (1993) . Factors related to susceptibility and recruitment by cults .
Psychological Reports, 73, 451-460 .
De Cremer, D., & van Vugt, M . (in press) . Why do people cooperate with leaders in managing
social dilemmas? Instrumental and relational aspects of structural cooperation. Journal of
Experimental Social Psychology.
Devine, P. G ., Hamilton, D. L ., & Ostrom, T. M . (Eds) (1994) . Social Cognition: Impact on Social
Psychology. San Diego, CA : Academic Press .
Duck, J . M ., & Fielding, K. S . (1999). Leaders and sub-groups : One of us or one of them? Group
Processes and Intergroup Relations, 2, 203-230 .
Ellemers, N . (1993) . The influence of socio-structural variables on identity enhancement
strategies. European Review of Social Psychology, 4, 27-57 .
Ellemers, N ., Spears, R., & Doosje, B . (Eds) (1999) . Social Identity. Oxford, U .K. : Blackwell .
Erber, R ., & Fiske, S . T. (1984) . Outcome dependency and attention to inconsistent information .
Journal of Personality and Social Psychology, 47, 709-726 .
25
Fiedler, F. E . (1965) . A contingency model of leadership effectiveness . In : L. Berkowitz (Ed .),

Advances in Experimental Social Psychology (Vol . 1, pp. 149-190) New York: Academic
Press .
Fiedler, F . E . (1971) . Leadership. Morristown, N. J . General Learning Press.
Fiedler, F. E ., & House, R . J. (1994) . Leadership theory and research : A report of progress . In: C .
L . Cooper & I . T. Robertson (Eds), Key Reviews in Managerial Psychology : Concepts and
Research for Practice (pp . 97-116) . Chichester, U.K. : Wiley.
Fielding, K. S ., & Hogg, M . A. (1997) . Social identity, self-categorization, and leadership : A field
study of small interactive groups. Group Dynamics : Theory, Research, and Practice, 1,
39-51 .
Fielding, K . S ., & Hogg, M . A . (in press) . Working Hard to Achieve Self-defining Group Goals :
a Social Identity Analysis. Zeitschrift ftir Sozialpsychologie .
Fiske, S . T. (1993) . Controlling other people : The impact of power on stereotyping . American
Psychologist, 48, 621-628 .
Fiske, S . T., & Depret, E. (1996) . Control, interdependence and power : Understanding social
cognition in its social context. European Review of Social Psychology, 7, 31-61 .
Fiske, S. T., & Taylor, S . E . (1991) . Social Cognition (2nd edn) . New York : McGraw-Hill .
Foddy, M ., & Hogg, M . A. (1999) . Impact of leaders on resource consumption in social dilemmas :
The intergroup context. In : M. Foddy, M . Smithson, S . Schneider, & M . A . Hogg (Eds) .
Resolving Social Dilemmas: Dynamic, Structural, and Intergroup Aspects (pp . 309-330) .
Philadelphia, PA : Psychology Press .
Galanter, M . (Ed .) (1989) . Cults and New Religious Movements . Washington, DC : American
Psychiatric Association .
Gilbert, D . T., Fiske, S . T., & Lindzey, G . (Eds) (1998) . The Handbook of Social Psychology (4th
ed .) . New York : McGraw-Hill .
Gilbert, D . T., & Jones, E . E . (1986) . Perceiver-induced constraint : Interpretations of selfgenerated reality. Journal of Personality and Social Psychology, 50, 269-280.
Gilbert, D . T., & Malone, P. S . (1995) . The correspondence bias . Psychological Bulletin, 117,
21-38 .
Giles, H ., & Johnson, P. (1987) . Ethnolinguistic identity theory : A social psychological approach
to language maintenance . International Journal of the Sociology of Language, 68, 66-99.
Hains, S . C ., Hogg, M . A., & Duck, J . M . (1997) . Self-categorization and leadership : Effects of
group prototypicality and leader stereotypicality. Personality and Social Psychology
Bulletin, 23, 1087-1100.
Hall, R . J ., & Lord, R . G . (1995) . Multi-level information processing explanations of followers'
leadership perceptions . Leadership Quarterly, 6, 265-287.
Hamilton, D . L ., & Sherman, S . J. (1996) . Perceiving persons and groups . Psychological Review,
103,336-355 .
Haslam, S . A ., McGarty, C ., Brown, P. M ., Eggins, R . A ., Morrison, B . E., & Reynolds, K . J.
(1998) . Inspecting the emperor's clothes : Evience that random selection of leaders can
enhance group performance . Group Dynamics: Theory, Research. and Practice, 2,
168-184 .
Haslam, S . A ., Oakes, P. J., McGarty, C., Turner, J . C., & Onorato, S . (1995). Contextual changes
in the prototypicality of extreme and moderate outgroup members . European Journal of
Social Psychology, 25, 509-530 .
Heider, F. (1958) . The Psychology of Interpersonal Relations . New York : Wiley.
26
MICHAEL A. HOGG
Hogg, M . A . (1992) . The Social Psychology of Group Cohesiveness: From Attraction to Social
Identity. Hemel Hempstead, U .K . : Harvester Wheatsheaf, and New York : New York
University Press .
Hogg, M . A . (1993), Group cohesiveness : A critical review and some new directions . European
Review of Social Psychology, 4, 85-111 .
Hogg, M. A. (1996) . Intragroup processes, group structure and social identity. In: W. P Robinson
(Ed.), Social Groups and Identities : Developing the Legacy of Henri Taffel (pp. 65-93) .
Oxford, U .K. : Butterworth-Heinemann.
Hogg, M . A. (2000a). Subjective uncertainty reduction through self-categorization : A motivational
theory of social identity processes . European Review of Social Psychology, 11, 223-255 .
Hogg, M. A. (2000b) . Social identity and social comparison . In : J . Suls & L. Wheeler (Eds),
Handbook of social comparison: Theory and research (pp . 401-421) . New York: Plenum .
Hogg, M . A . (2001) . Social categorization, depersonalization, and group behavior. In : M . A. Hogg
& R . S . Tindale, (Eds), Blackwell Handbook of Social Psychology : Group Processes (pp.
56-85). Oxford, U .K . : Blackwell .
Hogg, M. A ., & Abrams, D . (1988). Social Identifications : A Social Psychology of Intergroup
Relations and Group Processes . London: Routledge .
Hogg, M . A ., & Abrams, D . (1999). Social identity and social cognition: Historical background
and current trends . In : D . Abrams & M . A. Hogg (Eds), Social Identity and Social
Cognition (pp. 1-25) . Oxford, U.K. : Blackwell.
Hogg, M . A ., Cooper-Shaw, L ., & Holzworth, D. W. (1993). Group prototypicality and
depersonalized attraction in small interactive groups . Personality and Social Psychology
Bulletin, 19, 452-465 .
Hogg, M . A ., & Hains, S . C . (1996) . Intergroup relations and group solidarity : Effects of group
identification and social beliefs on depersonalized attraction . Journal of Personality and
Hogg, M. A., & Hains, S . C . (1998). Friendship and group identification : A new look at the role
of cohesiveness in groupthink . European Journal of Social Psychology, 28, 323-341 .
Hogg, M . A ., Hains, S . C ., & Mason, I . (1998). Identification and leadership in small groups :
Salience, frame of reference, and leader stereotypicality effects on leader evaluations.
Hogg, M . A., & Hardie, E . A . (1991) . Social attraction, personal attraction and self-categorization :
A field study . Personality and Social Psychology Bulletin, 17, 175-180 .
Hogg, M . A ., Hardie, E . A ., & Reynolds, K. (1995) . Prototypical similarity, self-categorization,
and depersonalized attraction : A perspective on group cohesiveness . European Journal of
Hogg, M . A., & Mullin, B .-A . (1999) . Joining groups to reduce uncertainty : Subjective uncertainty
reduction and group identification . In : D. Abrams & M . A . Hogg (Eds), Social Identity and
Social Cognition (pp . 249-279) . Oxford, U.K.: Blackwell.
Hogg, M . A ., & Reid, S . (in press). Social identity, leadership, and power. In: J . Bargh & A . LeeChai (Eds), The Use and Abuse of Power : Multiple Perspectives on the Causes of
Corruption . Philadelphia, PA : Psychology Press.
Hogg, M . A., & Terry, D . J . (2000) . Social identity and self-categorization processes in
organizational contexts . Academy of Management Review, 25, 121-140 .
Hogg, M. A ., & Terry, D . J . (Eds) (2001) . Social Identity Processes in Organizational Contexts .
Philadelphia, PA: Psychology Press .
Hogg, M . A ., Terry, D. J., & White, K . M . (1995). A tale of two theories: A critical comparison
of identity theory with social identity theory . Social Psychology Quarterly, 58, 255-269.
27
Hogg, M . A., & Tindale, R . S . (Eds) (2001) . Blackwell Handbook of Social Psychology : Group
Processes . Oxford, U.K. : Blackwell .
Hogg, M . A ., & Turner, J. C. (1987) . Social identity and conformity : A theory of referent
informational influence . In : W. Doise & S . Moscovici (Eds), Current Issues in European
Social Psychology (Vol. 2, pp . 139-182) . Cambridge, U .K . : Cambridge University Press .
and Paris : Editions de la Maison des Sciences de l'Homme.
Hollander, E . P. (1958) . Conformity, status, and idiosyncracy credit . Psychological Review, 65,
117-127 .
Hollander, E . P. (1985) . Leadership and power. In : G . Lindzey & E . Aronson (Eds), The Handbook
of Social Psychology (3rd ed ., Vol . 2, pp . 485-537). New York : Random House
Hollander, E . P, & Julian, J . W. (1969) . Contemporary trends in the analysis of leadership
processes . Psychological Bulletin, 71, 387-391 .
Hollander, E . P., & Julian, J . W. (1970) . Studies in leader legitimacy, influence, and innovation . In :
L . Berkowitz (Ed .), Advances in Experimental Social Psychology (Vol . 5, pp . 34-69). New
York : Academic Press .
Janis, I . L . (1972) . Victims of Groupthink: A Psychological Study of Foreign Policy Decisions and
Fiascoes . Boston, MA: Houghton-Mifflin .
Jost, J . T., & Banaji, M. R . (1994) . The role of stereotyping in system justification and the
production of false consciousness . British Journal of Social Psychology, 33, 1-27 .
Levine, J. M ., & Moreland, R. L . (1990). Progress in small group research . Annual Review of
Psychology, 41, 585-634 .
Levine, J . M ., & Moreland, R . L. (1995) . Group processes . In : A . Tesser (Ed .), Advanced Social
Psychology (pp . 419-465) . New York : McGraw-Hill .
Levine, J . M ., Resnick, L. B ., & Higgins, E . T. (1993) . Social foundations of cognition . Annual
Review of Psychology, 44, 585-612 .
Lippitt, R., & White, R. (1943). The `social climate' of children's groups . In : R. G. Barker, J .
Kounin & H. Wright (Eds), Child Behavior and Development (pp. 485-508) . New York :
McGraw-Hill .
Lord, R. G ., Brown, D . J ., & Harvey, J. L . (2001) . System constraints on leadership perceptions,
behavior and influence : An example of connectionist level processes . In : M . A . Hogg & R.
S . Tindale (Eds), Blackwell Handbook of Social Psychology : Group Processes (pp .
283-310) . Oxford, U .K . : Blackwell .
Lord, R. G ., Foti, R. J ., & DeVader, C . L . (1984) . A test of leadership categorization theory :
Internal structure, information processing, and leadership perceptions . Organizational
Behavior and Human Performance, 34, 343-378 .
Lord, R. G ., & Maher, K . J. (1991). Leadership and Information Processing : Linking Perceptions
and Performance . Winchester, MA: Unwin Hyman.
Marques, J . M., & Paez, D . (1994) . The black sheep effect : Social categorization, rejection of
ingroup deviates, and perception of group variability . European Review of Social
Marques, J. M., Abrams, D ., Paez, D ., & Hogg, M . A . (2001) . Social categorization, social
identification, and rejection of deviant group members . In: M. A . Hogg & R. S. Tindale,
(Eds), Blackwell Handbook of Social Psychology: Group Processes (pp .400-424). Oxford,
U.K. : Blackwell .
Mackie, D . M . (1987) . Systematic and non systematic processing of majority and minority
persuasive communications . Journal of Personality and Social Psychology, 53, 41-52 .
Mackie, D . M ., & Queller, S . (2000) . The impact of group membership on persuasion : Revisiting
"who says what to whom with what effect" . In : D . J. Terry & M . A . Hogg (Eds), Attitudes,
28
MICHAEL A . HOGG
Behavior, and Social Context : The Role of Norms and Group Membership (pp. 135-155) .
Mahwah, NJ: Erlbaum .
Marris, P (1996) . The Politics of Uncertainty : Attachment in Private and Public Life. London :
Routledge.
McGarty, C, Turner, J . C., Hogg, M . A ., David, B ., & Wetherell, M . S. (1992). Group polarization
as conformity to the prototypical group member . British Journal of Social Psychology, 31,
1-20.
McGrath, J. E. (1997) . Small group research, that once and future field : An interpretation of the
past with an eye to the future. Group Dynamics: Theory, Research, and Practice, 1, 7-27 .
Meindl, J . R., Ehrlich, S . B ., & Dukerich, J . M . (1985) . The romance of leadership . Administrative
Science Quarterly, 30, 78-102.
Moreland, R . L ., Hogg, M . A ., & Hains, S . C. (1994). Back to the future : Social psychological
research on groups. Journal of Experimental Social Psychology, 30, 527-555 .
Mowday, R. T., & Sutton, R. I . (1993). Organizational behavior: Linking individuals and groups
to organizational contexts . Annual Review of Psychology, 44, 195-229 .
Nye, J. L., & Forsyth, D . R . (1991) . The effects of prototype-based biases on leadership appraisals :
A test of leadership categorization theory . Small Group Research, 22, 360-379 .
Nye, J. L ., & Simonetta, L . G . (1996) . Followers' perceptions of group leaders : The impact of
recognition-based and inference-based processes . In : J . L . Nye, & A. M . Bower, (Eds),
What's Social about Social Cognition : Research on Socially Shared Cognition in Small
Groups (pp . 124-153) . Thousand Oaks, CA : Sage .
Oakes, P. J ., Haslam, S . A ., & Turner, J . C. (1994) . Stereotyping and Social Reality . Oxford, U.K.:
Blackwell .
Palich, L . E ., & Hom, P. W. (1992) . The impact of leader power and behavior on leadership
perceptions : A lisrel rest of an expanded categorization theory of leadership model . Group
and Organization Management, 17, 279-296 .
Pawar, B . S ., & Eastman, K. (1997) . The nature and implications of contextual influences on
transformational leadership . Academy of Management Review, 22, 80-109 .
Peters, L. H., Hartke, D . D ., & Pohlmann, J . T. (1985) . Fielder's contingency theory of leadership :
An application of the meta-analytic procedure of Schmidt and Hunter . Psychological
Bulletin, 97, 287-300 .
Platow, M. J ., Reid, S ., & Andrew, S. (1998) . Leadership endorsement: The role of distributive and
procedural behavior in interpersonal and intergroup contexts . Group Processes and
Intergroup Relations, 1, 35-47 .
Platow, M. J ., & van Knippenberg, D . (1999, July) . The impact of leaders' ingroup prototypicality
and normative fairness on leadership endorsements in an intergroup context . Paper given at
the XII General Meeting of the European Association of Experimental Social Psychology,
Oxford, U .K.
Raven, B . H. (1965) . Social influence and power . In : I. D . Steiner & M . Fishbein (Eds), Current
Studies in Social Psychology (pp . 371-382). New York : Holt, Rinehart & Winston .
Reicher, S . D . (1982) . The determination of collective behaviour. In: H. Tajfel (Ed .), Social
Identity and Intergroup Relations (pp . 41-83). Cambridge, U .K . : Cambridge University
Press.
Reicher, S . D. (2001). The psychology of crowd dynamics . In : M. A. Hogg & R . S . Tindale (Eds),
Blackwell Handbook of Social Psychology : Group Processes (pp. 182-208) . Oxford, U .K. :
Blackwell.
29
Reid, S ., & Hogg, M. A . (2000) . When being deplorable is meaningful : A test of three models
linking self-enhancement and uncertainty reduction with social identification processes.
Manuscript submitted for publication, University of Queensland .
Ridgeway, C . L . (2001) . Social status and group structure . In : M. A. Hogg & R. S . Tindale (Eds),
Blackwell Handbook of Social Psychology: Group Processes (pp. 352-375) . Oxford, U .K . :
Blackwell.
Ridgeway, C . L ., & Diekema, D . (1989) . Dominance and collective hierarchy formation in male
and female task groups . American Sociological Review, 54, 79-93 .
Ross, L. (1977) . The intuitive psychologist and his shortcomings . In : L . Berkowitz (Ed.), Advances
in Experimental Social Psychology (Vol . 10, pp . 174-220) . New York : Academic Press .
Rush, M . C., & Russell, J . E . A . (1988) . Leader prototypes and prototype-contingent consensus in
leader behavior descriptions . Journal of Experimental Social Psychology, 24, 88-104 .
Sanna, L. J., & Parks, C . D . (1997). Group research trends in social and organizational
psychology : Whatever happened to intragroup research? Psychological Science, 8,
261-267 .
Shaw, M . E . (1981) . Group Dynamics : The Psychology of Small Group Behavior (2nd edn). New
York : McGraw-Hill .
Sherif, M . (1966). In Common Predicament: Social Psychology of Intergroup Conflict and
Cooperation. Boston, MA: Houghton-Mifflin
Simon, B ., & Hastedt, C . (1999). Self-aspects as social categories : The role of personal importance
and valence . European Journal of Social Psychology, 29, 479-487 .
Smith, E. R ., & Henry, S . (1996) . An in-group becomes part of the self : Response time evidence .
Personality and Social Psychology Bulletin, 22, 635-642 .
Spears, R., Oakes, P. J ., Ellemers, N ., & Haslam, S . A . (Eds) (1997) . The social psychology of
stereotyping and group life . Oxford, U .K . : Blackwell .
Steiner, I . D. (1974). Whatever happened to the group in social psychology? Journal of
Experimental Social Psychology, 10, 94-108.
Steiner, I . D . (1986) . Paradigms and groups . Advances in Experimental Social Psychology, 19,
251-289 .
Stogdill, R . (1974) . Handbook of Leadership . New York : Free Press .
Tajfel, H. (1972) . Social categorization . English manuscript of `La categorisation sociale' . In : S .
Moscovici (Ed.), Introduction a la Psychologie Sociale (Vol . 1, pp. 272-302) . Paris :
Larousse .
Tajfel, H ., & Turner, J . C . (1979) . An integrative theory of intergroup conflict . In : W. G. Austin &
S . Worchel (Eds), The Social Psychology of Intergroup Relations (pp. 33-47) . Monterey,
CA : Brooks/Cole .
Taylor, S . E., & Fiske, S . T. (1975). Point-of-view and perceptions of causality . Journal of
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,
pp . 249-288). New York: Academic Press .
Terry, D. J ., & Hogg, M . A. (1996) . Group norms and the attitude-behavior relationship: A role for
group identification . Personality and Social Psychology Bulletin, 22, 776-793 .
Terry, D. J ., & Hogg, M . A . (Eds) (2000) . Attitudes, Behavior and Social Context : the Role of
Norms and Group Membership . Mahwah, NJ: Erlbaum .
Tindale, R . S ., & Anderson, E . M . (1998). Small group research and applied social psychology :
An introduction . In : R . S . Tindale, L . Heath, J . Edwards, E . J . Posavac, F. B . Bryant, Y.
Suarez-Balcazar, E. Henderson-King & J. Myer (Eds), Social Psychological Applications
30
MICHAEL A. HOGG
to Social Issues: Theory and Research on Small Groups, (Vol . 4, pp. 1-8) . New York:
Plenum Press.
Trope, Y., &-Liberman, A. (1993). The use of trait conceptions to identify other people's behavior
and to draw inferences about their personalities . Personality and Social Psychology
Bulletin, 19, 553-562.
Turner, J . C . (1975) . Social comparison and social identity : Some prospects for intergroup
behaviour. European Journal of Social Psychology, 5, 5-34.
Turner, J . C . (1982). Towards a cognitive redefinition of the social group . In : H . Tajfel (Ed.), Social
Identity and Intergroup Relations (pp . 15-40) . Cambridge, U .K . : Cambridge University
Press .
Turner, J. C . (1985) . Social categorization and the self-concept : A social cognitive theory of group
behavior. In : E. J . Lawler (Ed .), Advances in Group Processes : Theory and Research (Vol.
2, pp . 77-122) . Greenwich, CT: JAI Press.
Turner, J . C. (1991) . Social influence . Buckingham, U.K.: Open University Press .
Turner, J. C . (1999) . Some current issues in research on social identity and self-categorization
theories . In: N . Ellemers, R . Spears & B . Doosje (Eds), Social Identity (pp . 6-34). Oxford,
U .K. : Blackwell .
Turner, J. C., Hogg, M . A ., Oakes, P. J ., Reicher, S . D ., & Wetherell, M . S . (1987). Rediscovering
the Social Group: A self-categorization Theory . Oxford, U.K. : Blackwell .
Turner, J . C ., & Oakes, P. J . (1989) . Self-categorization and social influence . In : P. B . Paulus (Ed .),
The Psychology of Group Influence (2nd ed ., pp. 233-275) . Hillsdale, NJ : Erlbaum.
Turner, M . E ., Pratkanis, A . R ., Probasco, P., & Leve, C . (1992). Threat, cohesion, and group
effectiveness : Testing a social identity maintenance perspective on groupthink . Journal of
Tyler, T. R . (1997) . The psychology of legitimacy: A relational perspective on voluntary deference
to authorities . Personality and Social Psychology Review, 1, 323-345 .
Tyler, T. R ., & Lind, E . A. (1992). A relational model of authority in groups . In : M. P. Zanna (Ed.),
Advances in Experimental Social Psychology (Vol .25, pp . 115-191) . New York : Academic
Press.
van Vugt, M ., & de Cremer, D. (1999) . Leadership in social dilemmas : The effects of group
identification on collective actions to provide public goods . Journal of Personality and
Wilpert, B . (1995) . Organizational behavior. Annual Review of Psychology, 46, 59-90 .
Worchel, S., Rothgerber, H ., Day, E. A ., Hart, D ., & Butemeyer, J. (1998) . Social identity and
individual productivity within groups . British Journal of Social Psychology, 37, 389-413 .
Worchel, S ., Morales, J . F., Paez, D., & Deschamps, J. -C . (Eds) (1998) . Social Identity :
International Perspectives . London : Sage .
Wright, S . C . (1997) . Ambiguity, social influence, and collective action : Generating collective
protest in response to tokenism . Personality and Social Psychology Bulletin, 23,
1277-1290 .
Yukl, G. A ., & Falbe, C . M . (1991) . Importance of different power sources in downward and lateral
relations . Journal of Applied Psychology, 76, 416-423 .
Yukl, G., & Van Fleet, D . D . (1992) . Theory and research on leadership in organizations . In: M.
D . Dunnette & L . M. Hough (Eds), Handbook of Organizational Psychology (2nd ed ., Vol .
3, pp. 147-197) . Palo Alto, CA : Consulting Psychologists Press .
RATIONAL AND IRRATIONAL BASES

OF COMMITMENT TO GROUP
HIERARCHIES
Ann Branaman
ABSTRACT
This paper draws on two theoretical perspectives - expectation states
theory and Goffman's theory of the self and social interaction - to analyze
the interactional dynamics by which status hierarchies are maintained. In
accounting for the conservative tendencies of social hierarchies, I argue
that the two perspectives rely on different principles. For expectation
states theory, cognition - particularly performance expectations - is key.
Established hierarchies tend to be preserved, according to this perspective, because group members believe that higher-status members are
more competent than lower-status actors and act on these beliefs in ways
that make them self-fulfilling . In Goffman's theory, actors' attachment to
images of self and the normative pressures they feel to affirm others'
images of self and to preserve an orderly flow of social interaction, are the
primary motivational bases of actors' commitment to established social
hierarchies. Whereas expectation states theory holds that hierarchies in
task-group interaction express actual or expected inequalities of competence among group members, Goffman's analyses suggest that interaction
in group processes may be more about affirming or defending established
hierarchies that might otherwise rest on tenuous ground. I distinguish
between "rational" and "irrational" bases of commitment to established
hierarchies. The former refers to hierarchies in which the high status
2001 by Elsevier Science Ltd .
ISBN : 0-7623-0767-6
31
32
ANN BRANAMAN
members are believed by most group members to be the most competent ;

the latter refers to hierarchies in which beliefs about the relative
competence of group members have little bearing on members' standing
within the group, but instead the maintenance of the hierarchy is driven
more by norms, identities, and emotions .
I. INTRODUCTION
Hierarchy is a pervasive feature of social life, seemingly inescapable in taskoriented group interaction . According to Bales' (1950, 1953) research,
inequalities of participation, influence, and prestige quickly emerge in nearly
all decision-making groups . Such hierarchies develop even in groups that are
initially unstructured and that are composed of members who are similar in
terms of social characteristics . When group members bring differing social
characteristics to group interaction, as is typically the case in most actual social
groups, hierarchies emerge even more predictably. In heterogeneous groups,
the hierarchies of participation, influence, and prestige that emerge typically
correspond to the status of the members in the outside society (Strodtbeck,
James & Hawkins, 1957 ; Strodtbeck & Mann, 1957 ; Torrance, 1954) . Once
such hierarchies are established, group members tend to more highly evaluate
the contributions of high-status members (Riecken, 1958 ; Sherif et al., 1955)
and to pressure one another to keep their contributions in line with their
respective statuses in the group (Whyte, 1943) .
Yet, in task-oriented groups that are focused on achieving a collective goal,
task competence is purportedly (at least in modern bureaucratic settings) the
most important basis of status in the group . Despite the strong connection
between group status and status in the outside society, external status does not
itself justify positions of power and prestige in task-oriented groups .
Hierarchies in task-oriented groups tend to mirror societal hierarchies of such
characteristics as race, gender, age, educational achievement, and occupational
status, according to expectation states theory (Berger et al ., 1977), primarily
because generalized expectations of task competence are attached to these
characteristics . Even when there is no actual link between such status
characteristics and task competence in a particular situation, the beliefs tend to
become self-fulfilling because higher-status group members are given more
opportunities to contribute, are evaluated more highly, and are granted more
influence than lower-status group members .
Yet, despite the interactional advantages enjoyed by group members who
bring more highly-valued status characteristics to group interaction, sometimes
Rational and Irrational Bases of Commitment to Group Hierarchies
33
individuals with devalued status characteristics do overcome their disadvantages and demonstrate superiority in competence . This situation is the focus of
my analysis in this study. Whether or not other group members grant
recognition of the competence of the low-status group member and allow this
person to exercise influence on group decision-making and to gain status in the
group, I argue, depends on the extent to which commitment to achievement of
collective goals or emotional attachment to established hierarchies and
identities prevail in the group .
To analyze this situation, I draw on two distinct yet complementary
theoretical perspectives on hierarchy in group interaction : (1) expectation states
theory; and (2) Erving Goffman's analyses of the self and social interaction .
The former is a systematic theoretical research program that explicitly focuses
on understanding the emergence and maintenance of hierarchies in taskoriented group interaction . In comparison, the latter is a quite unsystematic,
often anecdotal, analysis of far more widely-ranging types of social interaction .
Though hierarchy in social interaction may not be Goffman's central focus, his
work has had an important influence on subsequent research exploring the
relationship between everyday interaction and social inequality (e .g. Henley,
1977, 1995 ; Zimmerman & West, 1975 ; Kollock, Blumstein & Schwartz, 1985 ;
Gardner, 1995 ; Derber, 2000) . Yet, following Roger's (1980) reading of
Goffman, I argue that hierarchy in social interaction is, at least implicitly, a
central theme in Goffman's work and one that deserves a thorough
articulation.
Both perspectives provide compelling analyses of the conservative tendencies of social interaction . Though expectation states theory is far more formal
and systematic than Goffman, their descriptions of the mechanisms by which
status hierarchies are preserved are similar . In accounting for the stability of
hierarchy in group interaction, however, the two approaches invoke very
different principles : the former privileges cognition, while the latter privileges
norms and affect. The status structure in collective task-oriented groups,
according to the expectation states approach, is organized primarily according
to the performance expectations that group members hold for themselves and
one another. According to expectation states theories, performance expectations mediate the relationship between social status and group interaction .
Thus, established hierarchies tend to be preserved because group members
generally believe that higher-status members are more competent than lowerstatus members . For Goffman, on the other hand, tendencies to preserve
established hierarchies are driven as much by emotions and norms as they are
by beliefs . In Goffman's theory, actors' attachment to images of self and the
normative pressures they feel to affirm others' images of self and to preserve
34
ANN BRANAMAN
an orderly flow of social interaction, are the primary motivational bases of

actors' commitment to an established social order .
Perhaps one of the main reasons for the enduring attention Goffman receives
from sociologists is the significance that emotions play in his analysis . Until the
relatively recent emergence of the sociology of emotions, emotions have
occupied a marginal place in both social psychology and sociology (Howard,
1995 : 100 ; Williams & Bendelow, 1998 : xvi) . Sociologists of emotion have, in
fact, given Goffman credit for calling attention to the role of emotions in
producing conformity to social order (Smith-Lovin, 1995 : 123 ; Collins, 1990 :
30 ; Scheff, 1990) . The human actor, in Goffman's analysis, is primarily driven
by the desire to "maintain face" much more than to achieve any instrumental
end. In Goffman's analysis, emotions serve as a social glue - a glue that
promotes social solidarity, produces mutually respectful affirmation of
identities on the part of social participants, induces behaviors that make for a
smooth flow of interaction, protects established hierarchies, and locks devalued
members of society into their lowly places in the social order. Goffman's
emphases on the importance of emotions and identity in maintaining social
order is consistent with affect control theory (Heise, 1977, 1979 ; Smith-Lovin
& Heise, 1988), in particular. According to affect control theory, emotions
motivate identity enactment in social life and disconfirmation of identity in
social interaction produces negative emotion that generates effort to reestablish identity (Smith-Lovin, 1995 : 130) . Goffman's approach also bears
affinities to approaches that examine the central role played by emotions in
upholding social stratification and to the emotional impact of changes in power
and status (Collins, 1990 ; Kemper, 1990) .
In order to avoid overstating the emphasis on cognition that I attribute to
expectation states theory, it is important to emphasize that theory and research
within this research program also examines the role that norms and affect play
in stabilizing social hierarchies . Attention to the role of norms is particularly
pronounced in theory and research on legitimation (e .g . Zelditch & Walker,
1984 ; Ridgeway & Berger, 1986, 1988 ; Ridgeway, 1989 ; Meeker & Weitzel
O'Neill, 1985), while attention to affect is evident in the growing body of
literature that explores the relationship between affect and status structures
(e .g . Lovaglia & Hauser, 1996 ; Shelly, 1993 ; Ridgeway & Johnson, 1990 ;
Smith-Lovin, 1995 ; Lawler & Thye, 1999 ; Thoits, 1989 ; Scher & Heise, 1993 ;
Driskell & Webster, 1997 ; Foschi, 1997) . I include a discussion of both of these
areas of research in my paper . Yet, the point I am making in drawing the
distinction that I do between expectation states theory and Goffman is that
there is a fundamental difference between the basic logics of the two
perspectives . I think it is important to emphasize this difference, furthermore,
35
because actual social groups vary in the extent to which their organization is
governed more or less by the respective logics at the root of each of these two
perspectives.
Generally, expectation states theories have limited their analyses to groups
that are task-oriented and are working toward the achievement of collective
goals . While individuals may compete for personal status within such groups,
they do so by striving to demonstrate task competence . Because of the group's
orientation towards collective success, it is in the interests of group members
to favor the contributions of the member(s) they believe to be most competent .
Expectation states theorists acknowledge that there are some social groups in
which members' desires for personal recognition outweigh their commitment
to collective goals . Such groups, however, do not meet their scope conditions
and hence are typically excluded from analysis .
In actual social life, however, many groups are task-oriented and aimed
towards collective goals in principle, while nonetheless group members'
behaviors may be motivated by many other things in addition to or instead of
group success . Unlike Goffman and more like the expectation states tradition,
my concern in this paper is with groups that are task-oriented and aimed
towards collective goals in principle . I am particularly interested, however, in
understanding the differences between groups that act according to those
principles and those that are illicitly driven more by norms, identities, and
emotions .
Undoubtedly, status processes in most actual social groups are affected by
cognitive beliefs about competence and by norms, identities, and affective
processes . Often, these will be difficult to disentangle for a couple of reasons .
First, social psychological research has demonstrated that cognitive beliefs are
often shaped by emotions (Isen, 1987 ; Bower, 1991) . And, second, in taskoriented and collectively-oriented groups where task competence is, in
principle, the most legitimate basis of status, statements of beliefs about
competence may be used as a "smokescreen" to mask more unacceptable
motivations . Despite difficulties in disentangling cognition and emotion,
however, I believe that task-oriented groups do significantly vary in the degree
to which status structures are organized according to beliefs about competence
or norms, identities, and emotions . The degree to which a status hierarchy is
organized according to one or the other set of principles will likely be most
apparent in situations where an established hierarchy is upset by the
demonstration of competence of a low-status actor . For this reason, I focus my
comparison of expectation states theory and Goffman on just such a situation .
I distinguish between "rational" and "irrational" bases of commitment to
established hierarchies . I consider commitment to an established hierarchy to
36
ANN BRANAMAN
be "rational" if the high-status members of the group are also believed to be the
most competent . One of the most significant contributions of the expectation
states theoretical research program is to demonstrate the variety of ways in
which determinations of the relative competence of group members are biased
and hence far less than fully rational . Yet, as long as most group members
believe that performance opportunities, high evaluations, and influence are
allocated according to competence, I consider such a hierarchy to be "rational"
(though the degree of bias in such beliefs may be widely varied .)
Conversely, I say that commitment to an established hierarchy is "irrational"
if beliefs about the relative competence of group members have little bearing
on members' standing within the group, but instead the maintenance of the
hierarchy is driven by norms, identities, and emotions . I argue that rational
justifications may often be used to camouflage more irrational motivations for
resisting an upset of the status quo, particularly in task-oriented groups where
hierarchy is most legitimately defensible when it corresponds to inequalities of
competence and merit. For this reason, the line between rational and irrational
motivations underlying efforts to maintain an established hierarchy will
sometimes be ambiguous and may itself be a point of bitter contention when
established hierarchies are challenged .
In a recent article on the legitimation and delegitimation of power and
prestige orders, Berger et al . (1998) point out that the emergence of
inconsistent performance expectations will put a strain on the legitimacy of an
established power and prestige order and will favor the possibility of its
delegitimation . Yet, they also point out that performance expectations are not
the crucial determinants of whether or not a power and prestige order achieves
legitimacy in the first place . Instead, the diffuse status characteristics of task
leaders are more important determinants . And while inconsistent performance
expectations may strain the legitimacy of a power and prestige order, they may
or may not succeed in actually delegitimating it. Berger et al . (1998) conclude
their article by pointing out that further analyses needs to be done to understand
"how the delegitimation process is affected by the mechanisms groups
sometimes use to cope with situations in which performance expectations
become incongruent with the legitimated structure and threaten to undermine
it" (Berger et al., 1998 : 398) .
According to the basic logic of the expectation states perspective,
legitimated power and prestige orders should become delegitimated when
performance expectations become incongruent with them . When and if they do
not, however, this suggests that norms, identities, and/or affective processes
may have come to override beliefs about competence in determining the
allocation of status in group interaction . In conjunction with recent work on
37
affect and status processes, I believe that Goffman provides a useful analysis of
affectively-driven interaction processes that impede the delegitimation of
legitimated status hierarchies .
The differences I highlight between Goffman and expectation states theory
are by no means absolute . Expectation states and other more systematic
sociological research programs have studied the effects of norms, identities,
and affect on status processes . Beliefs about reality, furthermore, certainly play
an important part in Goffman's analysis . I think it is important, however, to
maintain an analytical distinction between cognitive and normative/affective
determinants of status allocation and hierarchy maintenance . Drawing an
analytic distinction between the basic logics of SCES theories and Goffman is,
I believe, a useful way to conceptualize this distinction - even if neither falls
exclusively on one or the other side of the analytic divide .
I do not hold one logic to be inherently superior to the other in its ability to
account for the emergence and maintenance of status hierarchies . I believe that
both logics operate in most actual task-oriented groups . Furthermore, it seems
to me that the dynamics of some groups are best characterized by the "rational"
logic of the expectation states perspective, while others are best characterized
by a Goffmanian "irrational" logic .
Both perspectives account for common experiences of "subtle discrimination" experienced by persons with devalued diffuse status characteristics in
heterogenous social groups . Both provide analyses of the difficult barriers that
low-status persons face if they attempt to challenge established status
hierarchies . Yet, the types of discrimination implied by the basic logics of each
these two perspectives are qualitatively different . Expectation states theories
imply a cognitively-rooted form of discrimination based on generalized beliefs
that persons with certain status characteristics are generally more competent
than others . Goffman's perspective, on the other hand, implies a more "oldfashioned" form of discrimination rooted in normative and affective attachment
to established hierarchies and identities . Expectations states theories hold that
hierarchies in task- group interaction express or reflect expected inequalities of
competence among group members, while Goffman's analyses suggests that
interaction processes in hierarchical group situations may be more about
affirming or defending established hierarchies that might otherwise rest on
quite tenuous ground .
To address the first type of discrimination requires interventions aimed at
changing the beliefs that people hold about the competence of particular
individuals and categories of individuals . Due to the self-fulfilling interaction
processes set in motion by the initially biased belief, addressing biased beliefs
is difficult enough . It should be pointed out, however, that research within the
38
ANN BRANAMAN
expectation states tradition has suggested strategies for addressing biases that
have been shown to be successful in some situations (Webster & Driskell,
1985 ; Pugh & Wahrman, 1985 ; Cohen & Roper, 1985 ; Meeker & Weitzel
O'Neill, 1985) .
Addressing the more "old-fashioned" form of discrimination, however, is
likely an even thornier problem . In the face of normative and emotional
attachment to established hierarchies, attempts to alter the cognitive beliefs that
members hold about the relative competence of group members may only
increase resistance and intensify efforts to maintain the status quo. Presentation
of contradictory information that should challenge the legitimacy of an
established hierarchy, where there is intense normative and emotional
attachment to it, may only intensify this attachment and generate defensive
efforts to maintain the established hierarchy . This is why it is important to
understand if a hierarchy is fueled primarily by cognitive or by normative and
affective processes .
II. TWO ANALYSES OF THE STABILITY OF

HIERARCHIES IN GROUP INTERACTION :
EXPECTATION STATES THEORY AND GOFFMAN
Both expectation states theory and Goffman identify a number of interactional
processes that favor the maintenance of established social hierarchies . Both
provide analyses of the barriers faced by low-status group members in
challenging the status quo . In this section of the paper, I introduce expectation
states theory and Goffman by focusing on their analyses of the processes by
which established hierarchies are preserved .
Expectation states theory. The most systematic and sustained analysis of
status processes in task-oriented group interaction is the expectation states
theoretical research program . For over three decades, theory and research in
this tradition has analyzed the processes by which social structure shapes social
interaction . This program focuses particularly on how status characteristics
(such as race, sex, age, occupation, physical attractiveness, skills, and abilities)
shape social interaction (Berger et al ., 1977 : 5) . Status characteristics are
defined as "characteristics around which differences in cognitions and
evaluations of individuals or social types of them come to be organized"
(Berger & Zelditch, 1977 : 5) .
In the original formulation of expectation states theory, Berger, Cohen and
Zelditch (1966) argued that groups form expectation states for the performance
of each group member and that these performance expectations shape group
behavior. Research in the expectation states program has established that there
39
are a number of factors that affect performance expectations . These include

task-directed behaviors, nonverbal behaviors, diffuse and specific status
characteristics, evaluations from outside sources, and rewards (Ridgeway &
Walker, 1995 : 289) .
They emphasize, however, that expectation states emerge not primarily from
evaluations of behavior but rather from initial status differentiation based on
diffuse status characteristics (such as race, sex, age, or educational attainment) .
Expectation states derived from diffuse status characteristics generate specific
expectations regarding ability to perform the given task, general expectations
concerning the individual's general competence or intelligence, and social
evaluations in terms of the degree of respect or esteem deserved by the person
(Berger, Wagner & Zelditch, 1985 : 12-13) . According to status characteristics
theory, a major branch of expectation states theory, diffuse status characteristics
such as age, gender, race, educational background, or occupational status form
the basis for performance expectations in collective task-oriented groups when
they are made salient and when they have not been explicitly dissociated from
the task at hand (Berger, Cohen & Zelditch, 1966 ; Berger, Cohen & Zelditch,
1972 ; Webster & Driskell, 1985 ; Webster & Foschi, 1988 ; Meeker & WeitzelO'Neill, 1985) .
Once performance expectations are generated, they affect the observable
power and prestige order of the group . The observable power and prestige
order consists of four components : (1) opportunities to contribute to the group
task (action opportunities) ; (2) attempts to contribute to the group's task
(performance outputs) ; (3) communicated evaluations of one's contributions
(reward actions) ; and (4) changes of opinion after exposure to disagreement
(influence) (Berger, Wagner & Zelditch, 1985 : 6) .
Although research in the expectation states tradition shows that evidence of
ability and enduring interaction mitigate the effects of diffuse status
characteristics on performance expectations, research demonstrates that initial
expectations affect interaction patterns in such a way that they may become
self-fulfilling and endure even in sustained interactions (Berger & Zelditch,
1977 : 10 ; Ridgeway & Walker, 1995 ; Ridgeway, 1991 ; Connor, 1985 ;
Johnston, 1985) . One of the basic postulates of expectations states theory is
that, when a status characteristic is salient in a social group, group members
will use the cultural beliefs about this status characteristic to form beliefs about
the relative competence of group members, even when the status characteristic
is logically irrelevant to the task . The burden-of-proof lies not in demonstrating
the relevance of the status characteristic, but in demonstrating that it is not
relevant to the task (Berger, Wagner & Zelditch, 1985 : 13-14 ; Ridgeway &
Walker, 1995 : 289) . Because initial beliefs that the characteristic is relevant set
40
ANN BRANAMAN
in motion self-fulfilling interaction processes, demonstration of non-relevance

is not always an easy task .
It is difficult for a low-status group member to overcome the group's low
expectations, according to theory and research in this tradition, for a number of
reasons . Most basically, the power and prestige order itself exerts a powerful
impact on interaction within the group and hence affects the performance
expectations that emerge as a result of interaction . In the first place, higherstatus group members are looked to and given more opportunities to contribute
to the group's task. Not encouraged or supported in contributing, a low-status
member will have greater difficulty in gaining the performance opportunities
necessary to demonstrate competence . Secondly, as a result of the discouragement or at least lack of encouragement that they receive, low-status group
members do typically make fewer contributions to the group's task . Obviously,
low expectations cannot be dispelled and competence cannot be demonstrated
unless the low-status group member makes sustained contributions to the
group . Research consistently shows a strong correlation between performance
outputs and perceived competence (Ridgeway & Walker, 1995 : 291) .
Third, even when low-status members do attempt to contribute to the group's
task, their contributions are less likely to be valued as highly as the
contributions of high-status members (Berger, Cohen & Zelditch, 1972 :
253-254) . Since evaluations of past performances shape expectations for future
performances (Berger & Conner, 1974 ; Berger, Conner & McKeown, 1974),
the tendency for low-status persons to receive lesser evaluations perpetuates
their disadvantage . Furthermore, even when objective methods of evaluating
performance are available that reveal the high quality of the low status
member's performance, the contribution will not necessarily enhance performance expectations for the low-status person . The common use of double
standards in evaluation contributes to the perpetuation of initial expectations .
Research on double standards shows that higher-status group members are
evaluated according to more lenient standards for success and failure, while
lower-status group members are evaluated according to stringent standards for
success and failure (Foschi & Foddy, 1988 ; Foschi, 1989, 1991, 1996) . Thus,
even when a low-status member performs well, the contribution may be
perceived to be a matter of chance or luck rather than an indication of the
person's competence . Similarly, an ill-advised contribution to group interaction
will be more quickly taken as substantial evidence of the low-status person's
incompetence than an equally poor performance on the part of a high-status
group member.
Fourth, other group members tend to be more influenced by the contributions
of higher status group members . The inability of low-status members to
41
influence others not only denies the opportunity to demonstrate the potential
efficacy of their suggestions, but the lack of influence itself serves as a marker
of incompetence . Often, the potential value of a contribution may be
ambiguous, and the basis upon which other group members judge the value of
the contribution is by noting the reception it receives from other group
members and the influence it has on them . So even those whose expectations
might otherwise have been unaffected by the diffuse status characteristics of
the low-status group member may form low expectations on the basis of this
person's lack of influence in the group . The effect of influence on perceived
competence has been established by research in the expectation states tradition
(Berger & Conner, 1974 : Berger, Conner & McKeown, 1974) .
By themselves, the self-fulfilling tendencies of power and prestige orders
create a formidable set of barriers for even the most competent low-status
group member to overcome . The well-worn saying - "A woman has to be twice
as good as a man" - expresses the nature of the problem. Yet, an addendum to
this saying that I recently heard adds : "Fortunately, this is not hard ." Obviously,
there are cases in which persons who bring culturally devalued diffuse status
characteristics to group interaction are "twice as good" at the task as other
persons who bring more culturally valued diffuse status characteristics to group
interaction . What then?
One of the basic postulates of expectation states theory is that all salient
status characteristics are combined to form expectations of the competence of
each group member (Berger et al ., 1977 ; Fisek, Berger & Norman, 1991) .
Research demonstrates that it is not the case that diffuse status characteristics
such as race or gender outweigh other relevant status characteristics to an
extent that there is nothing a woman or a minority can do to be perceived to be
competent . Quite to the contrary, evidence clearly demonstrates that actors are
not perceived solely on the basis of their diffuse status characteristics . Both
diffuse status characteristics and specific status characteristics (characteristics
more directly relevant to the particular task at hand) are combined to assign
competence expectations to actors (Webster & Driskell, 1985 : 132-133) . When
group members know their relative task abilities and do not need to rely on
diffuse status characteristics to generate performance expectations, research
suggests that specific, instrumental status characteristics have a stronger effect
on performance expectations than do diffuse status characteristics (Zelditch,
1985 : 104) . Some research on paths of status expectancies suggests that the
more task-connected status characteristics exert a greater influence on a
person's group standing than do less directly relevant diffuse status characteristics (Berger, Wagner & Zelditch, 1985 : 17 ; Zelditch, Lauderdale & Stublarec,
1980) . Other research, however, only partially supports the notion that specific
42
ANN BRANAMAN
status characteristics exert more influence than diffuse status characteristics

(Berger, Wagner & Zelditch, 1985 : 17 ; Moore, 1968 ; Berger, Cohen &
Zelditch, 1972) . In any case, the consistent finding is that all relevant status
characteristics combine to determine the power and prestige order (Berger,
Wagner & Zelditch, 1985 : 18 ; Freese, 1974, 1976; Zelditch, Lauderdale &
Stublarec, 1980) .
Although diffuse status characteristics nearly always influence the power and
prestige order, research suggests that persons who bring devalued diffuse status
characteristics to group interaction may exert considerable influence if they
manage to present sufficient evidence to contradict the negative performance
expectations attached to their diffuse status characteristics . In fact, research
suggests that consistent evidence of task competence by a low-status person
can alter the expectations that other group members attach to the diffuse status
characteristic in general and thereby affect the way they respond to persons
with that same status characteristic in future situations (Ridgeway & Walker,
1995 : 294-295) . Evidence of such a "transfer effect" has been empirically
demonstrated in mixed-sex settings (Pugh & Wahrman, 1985) . Furthermore,
Cohen's research on how to reduce the effects of devalued diffuse status
characteristics shows that the effects "may be modified by assigning low-status
members a high rating on competence on a general performance characteristic"
(Cohen & Roper, 1985 : 374) .
Yet, even though demonstration of competence may offset the effects of
devalued diffuse status characteristics, it does not always . If they can
successfully overcome the self-fulfilling tendencies of the initial power and
prestige order, perhaps the most challenging barrier that low-status persons
face in group situations is the problem of legitimacy. Research shows that the
contributions and resulting status gains of high-status members are viewed as
legitimate and beneficial to the achievement of the group's goals . The
contributions of low-status members, on the other hand, are perceived as selfinterested and illegitimate attempts to gain status in the group (Meeker &
Weitzel-O'Neill, 1985 : 389 ; Ridgeway & Walker, 1995 : 296) .
If a power and prestige order has not been legitimated, i .e . has not become
transformed from a set of merely cognitive expectations about how actors will
likely perform to a system normative and moral expectations about how much
each actor should contribute, deviations frominitial performance expectations
by a low-status actor will alter expectations and lead to a change in the
distribution of power and prestige within the group (Wagner, Ford & Ford,
1986 ; Lockheed & Hall, 1976 ; Pugh & Wahrman, 1983) . Yet, if a power and
prestige order is legitimated, a deviation from expectations will be perceived as
a "status violation" and will be subject to group social control (Wagner, 1988) .
43
A legitimated hierarchy rests not merely on beliefs about competence but on

moral norms . "Legitimation transforms differences in status, influence, or
power into systems of rights and obligations" (Ridgeway & Walker, 1995 : 282) .
In other words, it is not merely a cognitive expectation that higher-status group
members will make superior contributions to those of lower-status group
members, but group members agree that the higher status members have the
right to contribute more and to receive more deference from others (Wagner,
1985 : 113) . Research in the status characteristics tradition suggests that
contributions exceeding expectations or attempts at leadership on the part of
lower-status members are experienced as morally offensive . As a result, the
status of the low-status contributor tends to decline rather than increase as a
result (Ridgeway, 1989 : 142) .
Low-status actors are similarly constrained in their use of "task cues" to
enhance their status in the group . Research in the expectation states program
has shown that nonverbal demeanor (e .g . confident tone of voice, relaxed
posture, maintenance of direct eye contact) affects the perception of a person's
competence (Dovidio & Ellyson, 1985 ; Ridgeway, 1987 ; Ridgeway, Berger &
Smith, 1985) . Because of their effect on perceptions of task competence, these
nonverbal demeanors have been called "task cues" (Berger et al ., 1986) . Yet, in
the same way that active demonstration of task competence will be perceived
as illegitimate attempts to gain status on the part of low-status actors, so too is
the use of confident task cues . Tuzlak (1988) uses the term "boomerang effect"
to refer to the "loss of influence experienced by those who possess a low state
of status but appear confident" (Tuzlak, 1988 : 273) . Research demonstrates that
boomerang effects occur when actors low in status with respect to race and
gender exhibit confident nonverbal demeanor (Henry & Ginzberg, 1985 ;
Ridgeway, 1982) . In short, successful task contributions and confident task
cues tend to backfire on low-status actors when a power and prestige order has
been legitimated .
Goff nan . Even to a greater extent than expectation states theory, Goffman
presents a portrait of social interaction as a very conservative force . Like
expectation states theory, Goffman identifies several interaction dynamics that
contribute to the maintenance of the status quo . Some of these correspond to
dynamics identified within the expectation states tradition . Unlike expectation
states theory, however, performance expectations do not play a central role in
Goffman's account. In part, this difference can be explained by the fact that
Goffman's primary focus of analysis is not task-oriented interaction . Yet, I
believe that Goffman would argue that the interaction dynamics he describes
occur in most social settings, maybe even especially task-oriented settings in
which a person's identity is attached to their role performance . Thus, I argue
44
ANN BRANAMAN
that Goffman's analysis is appropriately applicable to task-oriented groups,

particularly in real-life groups where individuals are more likely to identify
with their task performances than they would in the context of an experimental
setting .
While expectation states theory demonstrates the ways in which performance expectations are often biased in favor of persons who hold high status in
the outside world, nonetheless the core logic of the theory holds that allocation
of prestige and influence in task groups reflects expected differences in
competence . For Goffman, by contrast, interaction dynamics may or may not
express what actors honestly believe about one another . Instead, often
interaction dynamics affirm or defend normative hierarchies that cannot justify
their existence with any substantial evidence of differences in ability or
worthiness between actors .
In general, Goffman suggests that the basic norms of social interaction favor
the maintenance of the social status quo . Goffman emphasizes the importance
of "self-presentation" in social settings for defining oneself in relation to
others . Goffman defines self-presentation as any attempt, either intentional or
unintentional, of the individual to control the images that he/she projects of
him/herself in social interaction . In the context of task-oriented groups, selfpresentation would include both task contributions and task cues . Goffman
goes so far as to say that the self is a product of performance in social settings
(Goffman, 1959) . Rarely, however, are individuals in command of the everyday
performances that are so crucial to defining them. Goffman emphasizes that
interaction norms constrain the performances of individuals in such a way that
favors the maintenance of an established social hierarchy .
The norms of social interaction, according to Goffman's analysis, prescribe
that individuals present themselves in accordance with their own social status
and accord others the respect and deference that is warranted by their social
status . In "The Nature of Deference and Demeanor" (1967), Goffman says that
people convey their social desirability through demeanor, that is through
attributes such as discretion, sincerity, modesty, self-control, and poise (1967 :
77) . One key way in which people express these positive attributes is by
according deference in congruence with their own and others' social positions,
thus maintaining the ritual order (1967 : 81) . Furthermore, Goffman points out
that unless the "lines" (i .e . patterns of verbal and nonverbal acts by which a
person expresses his view of the situation, his evaluation of others and himself)
taken by social participants in social interaction are obviously inconsistent with
their social status, the basic rules of social interaction require honoring the lines
that people take in social interaction (Goffman, 1967 : 5) . Applied to a taskoriented situation in which the ability of the low-status actor outstrips the task
45
competence of a high-status actor, Goffman's analysis suggests that the lowstatus actor could only gain status within the group if the high-status person
willingly adopted a relatively subordinate line . As long as the high-status
person adopted a more dominant line, and possessed diffuse status characteristics that made such a line appear to be basically reasonable, other group
members would likely feel compelled to affirm it .
It is more likely that embarrassment rather than status gain will result from
failure to abide by any of these interactional norms . It is for this reason that
what Goffman believes to be the basic human desire to receive affirmation of
self in social interaction produces a normative pressure towards maintaining the
established social order. Because "maintaining face" requires knowing one's
place, Goffman proposes that interactional norms generally favor the
maintenance of the status quo . Taking advantage of opportunities for gaining
face is risky. Individuals generally determine that more is to be lost than to be
gained by challenging the status quo and, consequently, most affirm the
established order .
Social life is an uncluttered, orderly thing because the person voluntarily stays away from
the places and topics and times where he is not wanted and where he might be disparaged
for going . He cooperates to save his face, finding that there is much to be gained from
venturing nothing (1967 : 43) .
One of Goffman's central points is that through our attachment to face we are
attached to society. We maintain face by following social norms, showing
deference for and affirming the faces of others, and presenting ourselves in
accordance with our own places in the status hierarchy . Although Goffman
suggests that individuals enjoy some liberties in choosing the images they will
present to others, he also emphasizes the socially-determined limitations on
this power of self-determination . The individual is socially constrained to
express a "workable definition of himself'- in other words, a definition which
is closely attuned to the one that others can accord him (1971 : 366) . Generally,
the self-definition that others will support is one that is consistent with the
person's status characteristics .
The effect of "facework" - interactional work oriented towards affirming the
identities and protecting the dignity of social participants - is typically to
maintain the ritual order of social life . Goffman distinguishes between three
basic kinds of facework : the "avoidance process," the "corrective process," and
"making points ." The avoidance process refers to the avoidance of situations in
which threats to face are likely to occur . The corrective process refers to the
attempts to re-establish the faces of the participants in the interaction when
threats to face occur. Making points involves introducing favorable facts about
onself and unfavorable facts about others (Goffman, 1967) . All of these types
46
ANN BRANAMAN
of facework, Goffman suggests, are used by all people in a wide variety of

situations . But in settings in which social hierarchies have been established,
these types of facework tend to be employed to the advantage of high-status
actors and to the disadvantage of low-status actors .
Once the "faces" of individuals have been established in social situations,
Goffman argues, actors have both a "defensive" orientation towards maintaining their own face and a "protective" orientation towards those of others .
Applied to a task-group setting, when evidence of task abilities emerges that
threatens to disrupt a legitimated social hierarchy, actors may employ the
avoidance process to protect the established order. For example, faced with a
task in which he is incompetent relative to a low-status actor, the high-status
actor may feign disdain or lack of interest in this aspect of the group's task and
"let" the low-status actor take the lead here "if it's something that she really
cares about ." In extreme situations, the vast majority of the group's tasks may
be defined in such a manner such that low-status actors accomplish most of the
group's tasks without disrupting the hierarchies of face in the group .
Alternatively, if tasks cannot be ceded to low-status actors without disrupting
the group's hierarchy, the tasks themselves may be avoided . Though avoidance
of the tasks themselves will lead to sub-optimal outcomes for the group, in
many actual group settings standards of success may be manipulated in such a
way as to deny this reality . Other group members, even low-status persons, may
often contribute in protecting the face of the high-status actor(s) by
participating in devaluing or avoiding task concerns that might threaten the face
of high-status actors . If a low-status actor were to challenge the avoidance
techniques of the high-status actor(s) and his team, this would likely be
regarded as an inappropriate act of aggression .
Even when they do not inwardly believe that hierarchies of competence
correspond to the hierarchies of face in a particular group setting, they
nonetheless may feel normative pressure to participate in the protection of the
faces of high-status actors . In situations when a low-status actor fails to support
the established hierarchy of face in a group situation or acts in such a way as
to threaten it, other groups members may feel moral indignation or at least a
strong sense of interactional awkwardness that motivates them to engage in
corrective face-work to restore the established hierarchy . For example, they
might make humor out of the breach with a smirk or a tone of condescension
in accepting the low-status person's contribution .
Similarly, the third type of facework - "making points" - generally works to
the advantage of high-status actors . Social norms typically permit high-status
actors to present their positive qualities and even obligate others to open the
curtains and provide applause for their presentations . When challenged by
47
lower-status actors, furthermore, refusal to enter the fray of status competition

is viewed as honorable and taken as further evidence of superiority . Though
low-status actors may equally desire favorable recognition from others and
might wish to challenge the status quo if they could, their awareness that the
cards are stacked against them often motivates them to contribute to
maintaining the status quo . A low-status actor who attempts to introduce
favorable information about himself and unfavorable information about a
higher-status "opponent" is perceived to be self-seeking and out-of-line .
Obviously, this insight is in accord with research in the expectation states
tradition that shows that task contributions and confident task cues on the part
of low-status actors are perceived to be illegitimate attempts to gain status in
the group .
A low-status actor's best bet to receive a modicum of favorable regard from
others in the group is to accept one's inferior status . In Stigma (1969), Goffman
discusses a basic norm of social life that he believes perpetuates hierarchies of
status in society : the "good adjustment" norm . A "good adjustment," he argues,
is expected of individuals in society who possess stigmatizing attributes . And,
according to his analysis, anyone who does not fit the following description
possesses stigmatizing attributes : "young, married, white, urban, northern,
heterosexual Protestant father of college education, fully employed, of good
complexion, weight, and height, and a recent record in sports" (Goffman, 1963 :
128) . Though the characteristics that are stigmatized in any particular place and
time may differ from Goffman's description, the important point is that Stigma
was not only an analysis of the consequences of dramatic forms of stigma, but
was also an analysis of the effects of more ordinary diffuse status
characteristics .
For a person who possesses a devalued status characteristic, the "good
adjustment" entails accepting the standards of the broader society according to
which one is defined inferior . It requires underplaying the significance of one's
devalued status characteristics, while at the same time not presenting oneself as
equal to a person possessing status characteristics that are more highly valued .
Very importantly, it entails avoiding any communication that suggests a
perception that one is an object of discrimination . The "good adjustment"
means judging yourself as inferior according to the dominant standards,
judging these standards as fair, and avoiding calling attention to discrimination
or exclusion . Violation of this norm, either by presenting oneself as equal or by
calling attention to discrimination or unfairness of standards, results in further
devaluation. To receive favorable regard from others, as limited as this may be,
depends on acceptance of this norm .
48
ANN BRANAMAN
Another mechanism that tends to support existing social hierarchies is

asymmetrical access to "territories of the self." A central point linking
Goffman's analysis to the analysis of interaction hierarchies is the link he drew
between social power, command of territory, and the dignity of the self . Central
to the maintenance of the dignity of the self, according to Goffman, is the
command of "territories of the self." Goffman identifies eight kinds : personal
space, the stall, use space, the turn, the sheath, possessional territory,
information preserve, and conversation preserve (Goffman, 1971 : 29-40) . The
level of self-worth one is able to sustain in the eyes of others, according to
Goffman, depends very much on access to these various territories of the self
- e.g . controlling information about oneself, controlling resources that attest to
one's worth and/or that can be used to exercise influence, control of boundaries
around one's body, being able to speak without being ignored or persistently
interrupted, being able to summon others into conversation and command
responses from them, being able to make others wait while demanding
immediate access to their time, etc .
While he suggests that most people can make at least some of these claims
some of the time, Goffman emphasizes that one's place in various stratification
orders determines when, where, and to what degree one can claim these
territorial preserves necessary for sustaining a sense of worth . "In general, the
higher the rank, the greater the size of all territories of the self and the greater
the control across the boundaries" (Goffman, 1871 : 40-41) . While not all of the
"territories" analyzed by Goffman are applicable in a task-oriented group
setting, several are. Particularly important for understanding the conservative
tendencies of social interaction are the many asymmetries in conversational
privilege . For example, the ability to speak at will and thereby control the
issues of group discussion and the ability to ignore or dismiss topics that might
upset the status quo work to the advantage of high-status actors in maintaining
their position in the group's power and prestige order . Clearly, territories of the
self are not merely privileges of status . They are also markers of status and
instruments for maintaining status .
Goffman's analysis of territories of the self was a major influence on Nancy
Henley's classic work Body Politics (1977), a book that showed that gender
differences in nonverbal communication - e .g. men's tendency to command
more space, to maintain more eye contact while speaking, to smile less, and to
inititate touch - paralleled the differences that have consistently been observed
between dominants and subordinates in all kinds of hierarchical relationships .
Henley argued that these asymmetries in nonverbal communication simultaneously express and maintain established hierarchy . While such communication
patterns may typically express and maintain hierarchy at the same time, it is
49
important to recognize that the expressive and defensive functions of

asymmetries in nonverbal communication and other hierarchical interaction
patterns may sometimes diverge .
In particular, to reiterate a central point, hierarchical interaction patterns may
express actual or expected inequalities of competence or worth or be used to
affirm or defend normative hierarchies that otherwise might rest on tenuous
ground . According to Goffman's portrayal, social interaction may often serve
the latter function more than the former . For instance, a scene similar to many
Goffman describes is one in which a person keeps another waiting extra
minutes outside his office not because his time is in heavy demand but because
he wants to convey an image of superior importance . Another example would
be when one person controls the group discussion, interrupts others, and
dismisses the contributions of low-status actors not because he believes that
what he has to say is much more important than what others are saying, but
rather because doing so affirms his sense of superiority and keeps lower-status
actors in their place .
A third way that social hierarchies are perpetuated is through what Goffman
called mortification processes . Mortification is, in Goffman's analysis, a central
set of processes by which the self-determination, status, and worth of
individuals are controlled, minimized, and denigrated . Asylums (Goffman,
1961) examines the impact of total institutions on the self experiences of
inmates . He wrote about the ways in which the inmate is subjected to a range
of mortifying conditions - denial of autonomy, restriction of free movement,
denial of private belongings and tools for everyday living, lack of privacy,
subjection to the authority of the staff, imposition of degrading postures and
stances, ridicule and denigration, etc . In Goffman's observations of staffpatient interactions in psychiatric hospitals, he found that patients were
denigrated by the staff not only for behaviors caused by the hospital
environment - such as begging for cigarettes or waiting eagerly outside the
dining room for meals - but they were also ridiculed when they behaved
"normally" (e .g . trying to engage a staff member in an everyday conversation ;
writing or sitting alone to think ; forming relationships with other patients ;
discussing future goals or plans for life in the outside world .) At best, these
behaviors were considered unserious, but more usually staff would convey that
such behaviors reflected the patient's delusional grandiosity or failure to face
up to realistic personal limitations . The mortifying conditions and acts of
mortification to which the staff subjected the patients were taken by the staff as
appropriate treatment for such incompetent, unworthy, and undignified selves
as the inmates were thought to be . Yet, Goffman emphasizes that the mortifying
50
ANN BRANAMAN
conditions affirm and product the incompetence, unworthiness, and indignity

much more than they respond to it.
Clearly, the experience of the psychiatric patient and other inmates of total
institutions is far from the experience of even the most low-status member of
a task-oriented group . There is, however, an important element of Goffman's
analysis of mortification processes that may have more general applicability :
the ceremonial deprivation and denigration involved in mortification . Some
mortification practices (such as denying a psychiatric patient use of potentially
dangerous personal care items or restricting them from leaving the building)
may follow from the staff's beliefs about what is appropriate given the
condition of the patient, just as some practices on the part of task-group
members (such as interrupting or dismissing the contribution of a low-status
person) may follow from their beliefs in the person's incompetence . Many
aspects of mortification, however, particularly those involving active denigration, cannot be accounted for by beliefs about what the patient is or needs, but
rather must be seen as a ceremonial denial of the person's worthiness intended
to affirm the hierarchical relationship entailed in the setting . In everyday life
group situations, denigrating treatments of some by others may be common in
some situations but quite rare in others . What is particularly significant about
them, however, is that such behaviors cannot be explained in terms of beliefs
about the relative competence and worthiness of actors, but rather seem to be
oriented to affirming and defending a particular set of arrangements .
Low-status people, Goffman suggests, are easily subject to what he calls the
looping effect . Self-respecting and socially respected individuals are generally
able to separate self from disrespectful treatments by others (1961 : 35-36) .
Through face-saving reactive expressions - i .e. rolling the eyes or offering a
contemptuous smile - the individual is able to establish a distance between the
denigrating treatment and the self (1961 : 36) . In status-asymmetrical relations,
however, the low-status person may not be allowed to express separateness
from the degradation. Any attempt to separate one's identity from the degrading
treatment received from a higher-status other will be taken as evidence of the
deficiency of the low-status person, will be disallowed as an expression of
separateness from degrading treatment, and will be taken as the object of the
next attack.
Applying the looping effect to a task-oriented situation, a high-status person
disagrees with a low-status person and vehemently expresses this disagreement . Believing that the higher status person is being defensive and
unreasonable, the low-status person rolls her eyes and lets out a loud sigh . In
doing so, she indicates her disdain for the disagreement and the disagreer and
separates herself from the dismissive response her suggestion received . In the
51
eyes of the rest of the group, however, this move is not only ineffective but is
experienced as offensive . The high-status actor and some or all of the other
group members react with hostility to this eye-rolling sigher and seek to put her
back in her place (only this time a little lower) .
This is what Goffman means when he refers to the looping effect : "an agency
that creates a defensive response . . . takes this very response at the target of its
next attack" (1961 : 35-36) . The higher-status party is able to impose this
looping effect, thus denying the lesser person's expression of a separate self . A
defensive response on the part of a higher-status person, on the other hand,
establishes not only the superiority of the self relative to the contaminating
treatment but also de-faces the offender . The looping effect is a particular
hazard the low-status actor faces if she challenge an established power and
prestige hierarchy. Goffman's analysis of the looping effect is consistent with
Ridgeway and Johnson's (1990) finding that negative emotional reactions on
the part of low-status actors tend to be suppressed, while high-status actors are
freer to express negative emotion in response to disagreement .
A related point is that low-status persons may easily be caught in what
Goffman calls a "frame trap ." People with low social status and little social
power have little power to frame events or to combat interpretive frameworks
applied to them. A person judged to be incompetent, for example, carries no
weight in combating the judgment. The protests can be discounted, taken as
further evidence of incompetence (1974 : 445) . The tendency to deny the power
to contribute to the framing of reality makes it possible for the individual to be
contained in what Goffman calls a "frame trap" - an arrangement of the world
in such a way that every bit of evidence and every attempt by the framed
individual to correct the misinterpretation is taken to confirm the original
interpretation (1974 : 480) . In general, people with higher social status are
accorded more authority to frame events, themselves, and others ; people with
lower social status are vulnerable to being contained in a frame trap .
III. COGNITIVE, NORMATIVE, AND AFFECTIVE

BASES OF COMMITMENT TO ESTABLISHED SOCIAL
HIERARCHIES
Though the language and style of analysis differs greatly, parallels can easily
be drawn between the conservative interaction dynamics described by Goffman
and those described by the expectation states theoretical research program .
There are particular affinities between the barriers produced by interaction
norms in Goffman's analysis and the barriers produced by legitimation as
analyzed by the expectation states program . Both demonstrate that low-status
52
ANN BRANAMAN
actors frequently get caught in status traps . Upon observation, the dynamics
described by each appear very similar . This is especially the case if analysis is
restricted to behavior and does not consider the meanings actors attach to their
behavior. Arguably, the status characteristics and expectation states program
systematizes the insights in Goffman's work that pertain to hierarchy in group
interaction .
Yet, their explanations of the conservative tendencies of social interaction
are different. According to the core logic of expectation states theory, the
conservative tendencies of social interaction are based primarily in cognition .
For Goffman, hierarchies are rooted more in norms and emotion . For
expectation states theory, the tendency for a low-status actor to become
"trapped" in a low-status position is a byproduct of the necessity for human
beings to use social categories to process information . That diffuse status
characteristics such as race and gender are unjustly employed as predictors of
competence and that group hierarchies that mirror societal inequalities are
more likely to receive legitimacy speaks to the continuing significance of race
and gender bias in our culture . Nonetheless, according to the core logic of the
theory, the status trap is based in beliefs about competence and commitments
to meritocracy (insofar as performance expectations mediate the relationship
between diffuse status characteristics and power and prestige within the
group).
In Goffman's analysis, however, social behavior appears to have less to do
with cognition and merit than it does with "looking good", or, at the very least,
not looking bad . Actors are much less concerned with assessing the relative
competence of actors than they are with sizing up the norms (especially the
normative hierarchy) of the social situation . Driven by their own desire for
positive regard in social situations, even if this is the more limited regard
accorded to the "well-adjusted" low-status actor, they are less motivated to
respond to participants on the basis of merit than they are to pay homage to
existing hierarchies . In Goffman's analysis, people make ceremonial affirmation of the existing order and engage in a variety of defensive practices to
maintain it that cannot be accounted for by performance expectations alone .
Whereas the basic logic of expectation states theory suggests that even
legitimated hierarchies should become delegitimated when contradictory
evidence about the relative competence of actors emerges, Goffman's analysis
provides a better account of situations in which this does not occur .
Expectation states theory suggests that the hierarchies of social life, at least
the task-oriented ones, are best characterized as flawed meritocracies . They
may even be deeply flawed and hence not very meritocratic at all . The bias
introduced by non-relevant diffuse status characteristics to predict competence
53
and allocate opportunities not only causes a systematic underestimation of the

merit of low-status members but also presumably produces sub-optimal group
productivity. Expectation states theory does not assume that status generalization, or the "importing" of status distinctions that exist in the outside world
into task-group interaction, is necessarily a conscious process or a logical one .
In their overview of the theory of expectation states and status characteristics,
Webster and Foschi (1988) make a number of points quite to the contrary . First,
they point out that "status generalization often occurs in the absence of logical
or evidential bases" (1988 : 2) . Second, even though success in completion of
the group task may be the ultimate goal of all group members, the use of status
characteristics to form performance expectations for group members often
produces inefficient and undesirable results when the good ideas of low-status
members are not used and the bad ideas of high-status members are given
undue weight (1988 : 2) . Third, "most instances of status generalization occur
outside the realm of conscious thought" (Webster & Foschi, 1988 : 4) . Yet,
despite their emphasis on the sub-rationality of status generalization processes,
they nonetheless assume that the process is primarily a cognitive one . In fact,
they explicitly distinguish status generalization from "sexism" or "racism"
(Webster & Foschi, 1988 : 4) .
Goffman, on the other hand, suggests that the hierarchies of group
interaction are better characterized as orders to which the identifies of members
becomes so attached that they are compelled to sustain them even if it means
performing a show in which they do not honestly believe . Though Goffman
does not focus on task-oriented interaction like expectation states theory does,
his analysis suggests that actors in any kind of social situation are much more
driven by normative and affective attachment to images and orders than they
are motivated to achieve instrumental goals .
In actual situations, both logics apply to varying degrees . Most task-oriented
groups are likely driven by commitment to both merit and order and serve both
instrumental and emotional ends for their participants . In favorable circumstances, such commitments pose no contradiction : the order to which the
identifies of members have become attached allocates respect and influence in
a reasonably meritocratic manner and functions in such a way as to most
effectively fulfill the goals of the group ; similarly, instrumental ends can be
achieved as emotional needs of group members are satisfied . In less favorable
circumstances, however, meritocractic allocation of respect and influence may
violate the established order of the social group and the emotional needs of
certain group members whose identities are most attached to this order. In such
situations, members may experienced pressure to outwardly reevaluate the
competence of participants in such a way as to avoid the violation . In such
54
ANN BRANAMAN
circumstances, the affective commitment to established hierarchies and

identities may override and/or determine the cognitive evaluations of merit.
Despite the fact that the basic logic of expectation states theory privileges
cognition, specifically performance expectations, in accounting for the
emergence and stability of hierarchies in group interaction, theory and research
within this program has also explored the role played by norms and emotions.
Research on legitimation suggests that norms play a powerful role in
determining the stability of a group hierarchy, and research exploring the
relationship between affect and status structures suggests that emotion is also
an important stabilizing force . Neither of these areas of research, however,
suggest that norms or emotion outweigh performance expectations as
determinants of the stability of a group hierarchy .
When power and prestige orders have become legitimated, according to
developments in the expectation states program, it is no longer clearly the case
that performance expectations directly mediate the relationship between status
characteristics and the allocation of respect and influence . In fact, when
hierarchies in group interaction have become legitimated, the relationship
between competence or even beliefs about competence and the allocation of
respect and influence may become quite tenuous . To the extent that the
legitimacy of a power and prestige order rests on something other than
performance expectations, a departure from the basic logic of expectation states
theory seems to be indicated .
Research on legitimation has shown that performance expectations are not
the most important determinant of whether or not a social order receives
legitimacy in the first place (Berger et al ., 1998 : 392) . A social order in which
diffuse status characteristics and performance expectations are consistent
among group members is more likely to attain legitimacy than a social order in
which there is a disjuncture between performance expectations and diffuse
status characteristics . For example, a group in which middle-age males perform
slightly better than young females is more likely to attain legitimacy than a
group in which young females perform dramatically better than middle-age
males (Berger et al ., 1998 : 393) . Thus, this research indicates that the
legitimacy of a hierarchy in group interaction is partially based in performance
expectations and partially based on a moral sense that some categories of actors
are more worthy of respect and influence than others irrespective of their
performance .
According to research in the expectation states program, however, inequalities in task success and failure between group members do have an impact on
the legitimation process . When task successes and failures produce performance expectations that reverse the .order of expectations of a legitimated social
55
order, the legitimacy of that order is likely to be undermined (Berger et al .,

1998 : 395) . So even though the legitimacy of a social hierarchy does not
ultimately depend on performance expectations, the emergence of inconsistent
performance expectations is likely to upset the legitimacy of a social order .
Similarly, though research in the expectation states program has shown that
status-related emotions play an important role in stabilizing group hierarchies,
it does not suggest that emotions override performance expectations . Shelly's
research (1988, 1993) shows that we tend to give more performance
opportunities, more positive evaluations, and more influence to those we like
than to those we dislike, thus allowing feelings about others to shape the
group's power and prestige order (Shelly, 1993 : 118-119) . But he also finds
that sentiments have a weaker effect than task ability or formal authority in
determining the distribution of power and prestige in a task-group . He finds that
effects of sentiments are no stronger than the effects of diffuse status
characteristics and concludes that they are likely weaker (Shelly, 1993 : 123) .
Ridgeway and Johnson's research (1990) show that socioemotional behavior
in task groups tends to contribute to the perpetuation of status hierarchies, yet
they believe that differences between high-status and low-status persons in
socioemotional behavior derive from their status within the group and do not
determine it . Whether a group member feels anger in response to a task
disagreement from another group member, according to Ridgeway and
Johnson, depends on the performance expectations that he holds for himself
relative to the other (Ridgeway & Johnson, 1990 : 1199) . When the person
holds higher expectations for himself relative to the other, he is most likely to
feel anger towards the disagreer ; when the person holds lower expectations for
himself relative to the other, he is most likely to attribute the disagreement to
himself (1200) . Further, the expression of negative socioemotional expressions
towards the disagreer is more accepted by other group members from a highstatus member than from a low-status member . When low-status members
express negative emotion towards what is perceived by the group to be a
legitimate disagreement on the part of a higher-status member, the low-status
person will be subject to negative sanctions from other members (Ridgeway &
Johnson, 1990 : 1204) .
Similarly, Lovaglia and Hauser's (1996 ; Lovaglia, 1997) research focuses on
how emotional reactions stabilize status hierarchies, yet they too view
emotional responses as largely derivative of status positions and not
determinative of them . Drawing on Kemper's (1984) theory of emotions, they
argue that positive emotions such as satisfaction, happiness, or love are
integrating emotions ; conversely, negative emotions such as anger, resentment,
or fear are differentiating emotions . As a result of the privileges of their status,
56
ANN BRANAMAN
Lovaglia and Hauser propose that high status persons are more likely to feel
integrating emotions and thus act in ways to promote the group involvement of
low-status actors . Conversely, as a result of the lesser respect and influence they
receive, they argue that low-status persons are more likely to feel differentiating
emotions and will resist the influence of high-status actors . Thus, they argue
that integrating emotions are compatible with high status, while differentiating
emotions are compatible with low status . They propose that compatible
emotions experienced by both high and low-status actors tends to decrease
status differences in task groups, while incompatible emotions tends to increase
them (Lovaglia & Hauser, 1996 : 869) . However, when the integrating emotions
of the high-status actor leads her to give low-status group members more
influence, she may consequently feel that her status has declined relative to
them and may in turn experience negative emotion . On the basis of her negative
emotion, she may then act in ways to reduce the influence of low-status
members and hence solidify the status difference between them . This in turn
create negative emotions on the part of low-status actors, generating a selfperpetuating cycle . Lovaglia and Hauser argue that this reciprocal process
recurs in such a way as to produce a self-stabilizing status structure (Lovaglia
& Hauser, 1996 : 870) .
Though they view emotional reactions as largely derivative of social status,
Lovaglia and Hauser's analysis illustrates an important point that I have
attempted to highlight in my analysis of Goffman - i .e. that status insecurity,
not demonstrable superiority in competence, on the part of high-status actors
tends to generate defensive efforts to affirm an established hierarchy . Though
Goffman is similar to researchers in the expectation states program in viewing
status differences between actors as determining the patterns of their
interaction, the crucial difference between the two perspectives is the
insignificance that performance expectations play in Goffman's analysis .
Goffman's world is one in which there are few stable realities and few
opportunities to demonstrate who one "really" is . Hence, the construction of a
particular reality depends on self-presentation and a great deal of interactional
work. In at least some task-oriented groups, on the other hand, presumably
there are some more-or-less concrete ways to demonstrate one's competence
and to justify one's status within the group . In such situations, I would argue,
heavy use of mechanisms such as those described in Goffman's analysis
suggests an insecurity of status and a need to defend and affirm a reality that
is not self-evident .
Thus, I believe a crucial distinction suggested by the comparison of
expectation states theory and Goffman is the distinction between group
57
hierarchies that express and reflect actual or expected inequalities, and those
that seek to affirm and defend otherwise tenuous ones .
IV. CONCLUSION
Though commitment to group hierarchies will typically be maintained by some
combination of cognition, norms, and emotion, the question of the relative
strength of these motives may have important implications on a practical level .
Do hierarchical patterns in social interaction primarily express actual or
expected inequalities of competence, or are they motivated more by the desire
(conscious or not) to protect the established social order and identities of
actors? Are individuals essentially reasonable beings who are rationally, albeit
imperfectly so, oriented to achievement of instrumental goals (at least in the
context of task-oriented and collective group settings)? Or are they more
suitably characterized as affective beings driven by pride, fear, envy, or any
number of other emotions and who often care more about maintaining their
desired identity and status than they do about accomplishing any instrumental
goal? To the extent that the latter is true, does the play of such emotions thwart
the emergence of competence, impede the achievement of instrumental goals,
and/or produce any other form of dysfunction in the group? These are the basic
questions by which my comparison of expectation states theory and Goffman
has been motivated .
This comparison of expectation states theory and Goffman suggests some
propositions to be explored in further research .
1 . When a person lacks competence at a task or is not clearly superior to
other group members in task competence, status insecurity is more likely to be
experienced by group members with culturally valued diffuse status characteristics than it is to be experienced by group members with devalued diffuse
status characteristics . This derives from the greater pressure experienced by the
high-status actor to "maintain face" consistent with what may be attributed to
him on the basis of his diffuse status characteristics . This applies particularly
to group situations in which the high-status member significantly identifies
with his or her position within the group .
2 . As a result of status insecurity, a group member who possesses culturally
valued diffuse status characteristics is likely, independently of the person's
relative competence, to express negative emotion to group members with
devalued diffuse status characteristics when the latter demonstrate task
competence or otherwise assume influential positions in group interaction .
Again, this applies to situations in which the individual's position within the
group is a significant part of his or her identity.
58
ANN BRANAMAN
3 . Even when other group members believe that a group member with
culturally devalued diffuse status characteristics is more competent than group
members with more highly valued diffuse status characteristics, they may
nonetheless feel normative pressure to accord more deference towards the
group member with more highly valued diffuse status characteristics . They will
be particularly likely to feel such pressure when the person with valued diffuse
status characteristics adopts a "line" in social interaction consistent with his
diffuse status characteristics .
4 . Despite the official norms in task groups that prescribe that power and
prestige positions should be allocated according to the respective task
competencies of group members, face claims made by group members with
culturally valued diffuse status characteristics (if they are made) will outweigh
competence-based claims by members with devalued diffuse status characteristics in generating behavioral support from other group members .
5 . Contradictions between evident hierarchies of competence within a group
and hierarchies of diffuse status characteristics provoke emotionally-driven
defenses of hierarchy on the part of group members with highly valued diffuse
status characteristics, particularly when the diffuse status characteristic and
competence at the task are central to the person's identity .
6. Such emotionally-driven defenses of hierarchy will undermine group
solidarity, will divert attention from the task, and will thereby undermine the
group's ability to successfully accomplish group tasks . Goffman (1959, 1967)
argues that social interaction breaks down when individuals in social
interaction are unable to affirm the images of self and the interactional lines
taken by others . When faces have been damaged and reparations are not made,
he points out, the cost is the social relationship itself . In task groups,
accordingly, the cost includes the task as well . The task and the solidarity of the
group may be "saved" by the person with devalued diffuse status characteristics
hiding competence and not requiring the group to confront the contradiction
between hierarchies of competence and hierarchies of diffuse status characteristics . And, indeed, in groups where less competent but higher status others can
perform the task at a minimal level of adequacy, this may seem to be the
optimal solution to all involved . But in situations where the low-status person
is more than "twice as good" or where the high-status person is egregiously
incompetent but will not willingly cede his power and prestige position, the
contradiction may pose an insoluble problem for the group .
7 . Because of norms against "irrational" bases of commitment to established
orders, irrational bases of commitment will rarely be self-evident . Though
subjective judgement will inevitably be required in determining the degree of
59
rationality of commitment to an established group order, more irrationallydetermined hierarchies may be identified by a preponderance of defensive
facework and/or negative socioemotional behavior relative to reasoned focus on
task concerns . Ridgeway and Johnson (1990) argue that negative socioemotional behavior is rare in most task groups . This is the case, they argue,
because mostly group members agree on the appropriate allocation of status
within the group. High-status persons receive few disagreements and thus feel
no need to react with negative socioemotional behavior . Low-status persons, on
the other hand, receive more disagreements, but because of their lower status
tend to attribute the disagreement to themselves and thus do not react with
negative socioemotional behavior. In situations where evidence of competence
contradicts the established status order, however, agreement on the appropriate
allocation of status may be less likely. Accordingly, low-status persons may
more often disagree with high-status actors, thus provoking negative emotional
responses . Even when the normative hierarchy is not directly challenged by
disagreements or "excessive" task contributions by low-status actors, the need
for high-status actors to mask their incompetence may lead to a preponderance
of facework relative to task-work .
8 . Earlier I argued that assessment of the degree to which commitment to
hierarchy is based on beliefs about competence or on norms, emotions, and
identities is important because it will affect the strategies of intervention that
would be most successful in challenging the legitimacy of a status hierarchy in
which actual competence of actors contradicts the power and prestige order of
the group . Research in the expectation states program has shown that using
direct authority to tell group members that a diffuse status characteristic is not
relevant to the task and publishing information on actual task capacities are
relatively effective ways of subverting the process of status generalization
(Webster & Driskell, 1985 : 133-134) . I want to suggest here that similar
strategies may be effective for confronting emotionally-driven defenses of
hierarchy, but for different reasons than those implied by expectation states
theory . Informal demonstration of task competence on the part of low-status
actors, i .e . direct attempts to contribute to the group's task, will likely provoke
backlash in situations where hierarchies are emotionally-driven . Yet, because
emotions are not supposed to determine the power and prestige order in modern
task-oriented settings, some form of official evidence (recognized by all to be
legitimate) of actual task capacities may remove the smokescreen covering
more irrational motives . Whether or not unmasking irrational bases of
commitment to established social orders is sufficient to undermine them
remains to be seen .
60
ANN BRANAMAN
REFERENCES
Bales, R. F. (1950) . Interaction Process Analysis : A Method for the Study of Small Groups .
Cambridge, MA : Addison-Wesley.
Bales, R . (1953) . The equilibrium problem in small groups. In: T. Parsons, R. F. Bales & E. A .
Shils (Eds), Working Papers in the Theory of Action, (pp . 111-161) . Glencoe, IL : Free
Press .
Berger, J ., Cohen, B . P., & Zelditch, M ., Jr. (1966) . Status Characteristics and Expectation States:
A Process Model. In: J . Berger, M. Zelditch, Jr. & B . Anderson (Eds), Sociological
Theories in Progress, Vol. 1 . (pp . 47-73) . Boston : Houghton-Mifflin .
Berger, J., Cohen, B . P., & Zelditch, M. Jr. (1972). Status Characteristics and Social Interaction .
American Sociological Review, 37, 241-255 .
Berger, J ., & Conner, T. L. (1974) . Performance expectations and behavior in small groups : A
revised formulation. In : Expectation States Theory: A Theoretical Research Program (pp.
85-110) . Cambridge, MA: Winthrop .
Berger, J ., Conner, T . L ., & McKeown, W. L . (1974). Evaluations and the formation and
maintenance of performance expectations. In : Expectation States Theory: A Theoretical
Research Program (pp. 27-51) . Cambridge, MA : Winthrop .
Berger, J ., Webster, M ., Ridgeway, C ., & Rosenholtz, S . J. (1986) . Status cues, expectations, and
behavior. In: E . Lawler (Ed .), Advances in Group Processes, Vol . 3, (pp. 1-22). Greenwich,
CT: JAI Press .
Berger, J., Conner, T. L ., & Fisek, M . H . (1974) . Expectation States Theory : A Theoretical
Research Program . Cambridge, MA : Winthrop .
Berger, J ., Fisek, M . H., Norman, R . Z., & Wagner, D . G. (1985) . The Formation of Reward
Expectations in Status Situations . In : J . Berger & M . Zelditch (Eds), Status, Rewards, and
Influence (pp . 215-261) . San Francisco : Jossey-Bass .
Berger, J., Fisek, M . H., Norman, R . Z., & Wagner, D . G. (1985). Introduction : Expectation States
Theory : Review and Assessment . In : J . Berger & M. Zelditch (Eds), Status, Rewards, and
Influence. San Francisco: Jossey-Bass .
Berger, J., Fisek, M . H ., Norman, R. Z ., & Zelditch, M. Jr. (1977) . Status Characteristics in Social
Interaction : An Expectation States Approach. New York : Elsevier.
Berger, J ., Fisek, M. H ., Norman, R. Z ., & Wagner, D . G. (1985) . Formation of Reward
Expectations in Status Situations . In : J . Berger & M . Zelditch (Eds), Status, Rewards, and
Influence. San Francisco: Jossey-Bass .
Berger, J., Fisek, M. H., & Norman, R. Z . (1989) . The Evolution of Status Expectations : a
Theoretical Extension . In : J. Berger, M . Zelditch & B . Anderson (Eds), Sociological
Theories in Progress : New Formulations . Newbury Park, CA: Sage .
Berger, J. et al . (1998) . The Legitimation and Delegitimation of Power and Prestige Orders .
American Sociological Review, 63(3), 379-405 .
Berger, J ., Wagner, D. G ., & Zelditch, M. (1985) . Expectation states theory : Review and
assessment . In: J. Berger & M. Zelditch (Eds), Status, Rewards, and Influence (pp . 1-72) .
San Francisco : Jossey-Bass .
Cohen, E . G., & Roper, S . S . (1985). Modification of Interracial Interaction Disability . In: J. Berger
& M . Zelditch (Eds), Status, Rewards, and Influence . San Francisco: Jossey-Bass .
Cohen, E. G., Lotan, R ., & Cantanzarite, L. (1988). Can Expectations for Competence be Altered
in the Classroom? In : M . Webster & M . Foschi (Eds), Status Generalization: New Theory
and Research. Stanford, CA : Stanford University Press .
61
Collins, R. (1990) . Stratification, Emotional Energy, and the Transient Emotions . In: T. D . Kemper
(Ed .), Research Agendas in the Sociology of Emotions (pp . 27-57) . Albany: State
University of New York Press .
Conner, T. L . (1985) . Response Latencies, Performance Expectations, and Interaction Patterns . In :
J . Berger & M . Zelditch (Eds), Status, Rewards, and Influence. San Francisco : JosseyBass .
Derber, C. (2000) . The Pursuit of Attention: Power and Ego in Everyday Life, (2nd ed .). Oxford :
Oxford University Press .
Dovidio, J . F., & Ellyson, S . L. (1985) . Patterns of visual dominance behavior in humans . In : S .
Ellyson & J . Dovidio (Eds), Power, Dominance, and Nonverbal Behavior (pp . 129-150) .
New York : Springer-Verlag .
Driskell, J. E ., & Webster, M . Jr. (1997). Status and Sentiment in Task Groups . In : J . Szmatka, J .
Skvoretz & J . Berger (Eds), Status, Network, & Structure : Theory Development in Group
Process (pp . 179-200) . Stanford University Press .
Fineman, S . (1993) . Organizations as Emotional Arenas . In : S . Fineman (Ed .), Emotions in
Organizations (pp . 9-35). London: Sage Publications.
Fisek, M . H ., Berger, J ., & Norman, R . Z . (1991) . Participation in heterogenous and homogenous
groups: A theoretical integration . American Journal of Sociology, 97, 114-142 .
Foddy, M. (1988) . Paths of Relevance and Evaluative Competence . In : M . Webster & M . Foschi
(Eds), Status Generalization: New Theory and Research. Stanford, CA : Stanford
University Press .
Foddy, M ., & Smithson, M . (1989) . Fuzzy Sets and Double Standards : Modeling the Process of
Ability Inference . In: J . Berger, M . Zelditch & B . Anderson (Eds), Sociological Theories
in Progress : New Formulations . Newbury Park, CA: Sage .
Foschi, M ., & Foddy, M . (1988) . Standards, Performances and the Formation of Self-other
Expectations . In : M . Webster & M . Foschi (Eds), Status Generalization : New Theory and
Research. Stanford, CA : Stanford University Press .
Foschi, M . (1989) . Status Characteristics, Standards, and Attributions . In : J . Berger, M . Zelditch
& B . Anderson (Eds), Sociological Theories in Progress : New Formulations (pp . 58-72) .
Newbury Park, CA : Sage.
Foschi, M . (1991) . Gender and Double standards for Competence. In : C . Ridgeway (Ed.), Gender,
Interaction, and Inequality. New York : Springer-Verlag .
Foschi, M . (1996) . Double Standards in the Evaluation of Men and Women . Social Psychology
Quarterly, 59, 237-254 .
Foschi, M . (1997) . Status, Affect, and Multiple Standards for Competence . In : J . Szmatka, J .
Skvoretz & J . Berger (Eds), Status, Network, & Structure : Theory Development in Group
Process (pp . 201-221). Stanford University Press .
Freese, L . (1974) . Conditions for status equality in formal task groups . Sociometry, 37, 174-188 .
Freese, L . (1976) . The Generalization of specific performance characteristics . Sociometry, 39,
194-200.
Gardner, C . B . (1995) . Passing By: Gender and Public Harassment . Berkeley : University of
California Press.
Goffman, E. (1959) . The Presentation of Self in Everyday Life . Garden City, NY: Doubleday,
Anchor Books .
Goffman, E. (1961) . Asylums . Garden City, NY: Doubleday, Anchor Books .
Goffman, E. (1963) . Stigma: Notes on the Management of Spoiled Identity. Englewood Cliffs, NJ :
Prentice Hall .
62
ANN BRANAMAN
Goffman, E. (1967). Interaction Ritual: Essays on Face-to-Face Behavior. Garden City, NY:
Doubleday, Anchor Books.
Goffman, E. (1974). Frame Analysis : An Essay on the Organization of Experience . NY: Harper
and Row.
Heise, D . R . (1977) . Social action as the control of affect . Behavioral Science, 22, 163-177 .
Heise, D . R . (1979) . Understanding Events . New York: Cambridge University Press.
Henley, N . (1977). Body Politics: Power, Sex, and Non-verbal Communication . Englewood Cliffs,
NJ : Prentice-Hall .
Henley, N . J . (1995). Body Politics Revisited : What do we know Today? In : P. J . Kalbfleisch &
M . J. Cody (Eds), Gender Power, and Communication in Human Relationships (pp .
27-61) . Hillsdale, New Jersey: Lawrence Erlbaum Associates, Publishers.
Henry, F, & Ginzberg, E. (1985) . Who Gets the Work : A Test of Racial Discrimination in
Employment . Toronto : Urban Alliance on Race Relations .
Howard, J . (1995) . Social Cognition . In : K. Cook, G. Fine & J . House (Eds), Sociological
Perspectives on Social Psychology (pp . 90-117) . Boston: Allyn and Bacon .
Howard, J . A ., Blumstein, P., & Schwartz, P. (1986) . Sex, Power, and Influence Tactics in Intimate
Relationships. Journal of Personality and Social Psychology, 51(1), 102-109.
Johnston, J . R . (1985) . How Personality Attributes Structure Interpersonal Relations . In : J. Berger
& M . Zelditch (Eds), Status, Rewards, and Influence. San Francisco: Jossey-Bass.
Kemper, T. D. (1990) . Social Relations and Emotions : A Structural Approach . In : Kemper, (Eds),
Research Agendas in the Sociology of Emotions (pp . 207-237) . Albany: State University of
New York Press .
Lawler, E . J ., & Thye, S . R . (1999) . Bring emotions into social exchange theory. Annual Review
of Sociology, 25, 217-244.
Lovaglia, M . J ., & Hauser, J. A. (1996) . Emotional reactions and status in groups . American
Sociological Review, 61, 867-883 .
Lovaglia, M . J . (1997) . Status, Emotions, and Structural Power. In: J. Szmatka, J . Skvoretz & J .
Berger (Eds), Status, Network, & Structure : Theory Development in Group Process (pp.
159-178) . Stanford University Press .
Lockheed, M. E ., & Hall, K. P (1976) . Conceptualizing sex as a status characteristic . Journal of
Social Issues, 32, 111-124 .
Meeker, B . F., & Weitzel O-Neill, P. A . (1985) . Sex Roles and Interpersonal Behavior in Taskoriented Groups . In : J . Berger & M . Zelditch (Eds), Status, Rewards, and Influence . San
Francisco : Jossey-Bass .
Moore, J. C ., Jr. (1968). Status and influence in small group interaction . Sociometry, 31, 47-63 .
Pugh, M . D., & Wahrman, R . (1985) . Inequality of Influence in Mixed-sex Groups . In : J. Berger
& M . Zelditch (Eds), Status, Rewards, and Influence . San Francisco : Jossey-Bass .
Ridgeway, C . (1982) . Status in groups : The importance of motivation . American Sociological
Review, 47, 76-88 .
Ridgeway, C . (1987) . Nonverbal behavior, dominance, and the basis of status in task groups .
American Sociological Review, 52, 683-694.
Ridgeway, C . L. (1988) . Gender Differences in Task Groups : a Status and Legitimacy Account . In :
M . Webster & M. Foschi (Eds), Status Generalization: New Theory and Research .
Stanford, CA : Stanford University Press .
Ridgeway, C . L. (1989) . Understanding Legitimation in Informal Status Orders . In: Berger,
Zelditch & Anderson (Eds), Sociological Theories in Progress: New Formulations .
Newbury Park, CA : Sage.
63
Ridgeway, C . L. (1991). The Social Construction of Status Value : Gender and Other Nominal
Characteristics. Social Forces, 70, 367-386.
Ridgeway, C . L. (1994) . Affect. In : M . Foschi & E. J. Lawler (Eds), Group Processes : Sociological
Analyses (pp . 205-230) . Chicago : Nelson Hall.
Ridgeway, C . L., & Berger, J . (1986) . Expectations, legitimation, and dominance behavior in task
groups . American Sociological Review, 51, 603-617 .
Ridgeway, C . L., & Berger, J . (1988) . The Legitimation of Power and Prestige Orders in Task
Groups . In : M. Webster & M . Foschi (Eds), Status Generalization : New Theory and
Research . Stanford, CA: Stanford University Press .
Ridgeway, C ., Berger, J ., & Smith, L. (1985) . Non-verbal cues and status : An expectation states
approach . American Journal of Sociology, 90, 955-978 .
Ridgeway, C ., & Johnson, C . (1990) . What is the relationship between socioemotional behavior
and status in task groups? American Journal of Sociology, 95, 1189-1212.
Ridgeway, C . L., Johnson, C., & Diekema, D . (1994) . External Status, Legitimacy, and
Compliance in Male and Female Groups . Social Forces, 72 .
Ridgeway, C . L ., & Walker, H . A . (1995) . Status Structures . In : K. Cook, G . Fine & J. House (Eds),
Sociological Perspectives on Social Psychology . Boston, MA : Allyn and Bacon .
Riecken, H . W. (1958) . The effect of talkativeness on ability to influence group solutions of
problems . Sociometry, 21, 309-321 .
Rogers, M. (1980) . Goffman on Power, Hierarchy, and Status . In : J . Ditton (Eds), The View from
Goffinan . London : Macmillan .
Scheff, T. J . (1990). Socialization of Emotions : Pride and Shame as Causal Agents . In : Kemper
(Eds), Research Agendas in the Sociology of Emotions (pp . 281-304) . Albany : State
University of New York Press .
Scher, S . J ., & Heise, D . R . (1993) . Affect and the perception of injustice . Advances in Group
Processes, 10, 223-252 .
Shelly, R . K . (1988) . Social differentiation and social integration . In : M . Webster Jr. & M . Foschi
(Eds), Status Generalization: New Theory and Research (pp . 366-376) . Stanford, CA :
Stanford University Press .
Shelly, R . K. (1993). How sentiments organize interaction . Advances in Group Processes, 10,
113-132 .
Sherif, M ., White, B . J ., & Harvey, O. J . (1955) . Status in experimentally produced groups .
American Journal of Sociology, 60, 370-379 .
Smith-Lovin, L., & Robinson, D . T. (1992). Gender and Conversation Dynamics . In : Ridgeway
(Ed.), Gender, Interaction, and Inequality. New York, NY: Springer-Verlag .
Smith-Lovin, L. (1990) . Emotion as the Confirmation and Disconfirmation of Identity : An AffectControl Model. In : T. Kemper (Ed .), Research Agendas in the Sociology of Emotions (pp .
238-270) . Albany : State University of New York Press .
Smith-Lovin, L . (1995) . The Sociology of Affect and Emotion . In : K. S . Cook, G. A. Fine & J . S .
House (Eds), Sociological Perspectives on Social Psychology (pp . 118-148). Boston : Allyn
and Bacon.
Stewart, P. (1988) . Women and Men in Groups : a Status Characteristics Approach to Interaction .
In: M. Webster & M . Foschi (Eds), Status Generalization: New Theory and Research .
Stanford, CA: Stanford University Press .
Strodtbeck, F. L ., James, R . M ., & Hawkins, C . (1957) . Social status in jury deliberations .
Strodtbeck, F. L ., & Mann, R. D . (1957) . Sex role differentiation injury deliberations . Sociometry,
19,3-11 .
64
ANN BRANAMAN
Thoits, P. A . (1989) . The sociology of emotions . Annual Review of Sociology, 15, 317-342 .
Torrance, E. P. (1954) . Some consequences of power differences in decision making in permanent
and temporary three-man groups . In : A . P. Hare, E. F. Borgatta & R . F. Bales (Eds), Small
Groups (pp. 600-609) . New York: Knopf.
Tuzlak, A . (1988) . Boomerang Effects : Status and Demeanor Over Time . In : M . Webster & M .
Foschi (Eds), Status Generalization : New theory and Research . Stanford, CA: Stanford
University Press .
Wagner, D . G . (1988). Gender Inequalities in Groups : a Situational Approach. In : M . Webster &
M . Foschi (Eds), Status Generalization : New Theory and Research. Stanford, CA: Stanford
University Press .
Wagner, D . G. (1988) . Status Violations: Toward an Expectations States Theory of the Social
Control of Deviance . In : M . Webster & M . Foschi (Eds), Status Generalization : New
Theory and Research . Stanford, CA : Stanford University Press .
Wagner, D . G., R . S . Ford, & T. W. Ford . (1986) . Can gender inequalities be reduced? American
Webster, M ., Jr., & J . E. Driskell, Jr . (1985) . Status generalization . In: J . Berger & M . Zelditch
(Eds), Status, Rewards, and Influence. San Francisco : Jossey-Bass .
Webster, M., Jr., & Foschi, M . (1988) . Status Generalization : New theory and Research . Stanford,
CA : Stanford University Press .
Whyte, W. F. (1943) . Street Corner Society. Chicago : University of Chicago Press .
Zelditch, M. Jr., & Walker, H . A . (1984) . Legitimacy and the Stability of Authority . In: E . Lawler
(Ed .), Advances in Group Processes, Vol . 3 . Greenwich, CT: JAI Press .
Zelditch, M . Jr. (1985) . Three Questions about Status . In : J . Berger & M. Zelditch (Eds), Status,
Rewards, and Influence . San Francisco : Jossey-Bass .
Zelditch, M . Jr., Lauderdale, P., & Stublarec, S . (1980) . How are inconsistencies between status
and ability resolved? Social Forces, 58, 1025-1043 .
Zimmerman, D . H ., & West, C . (1975) . Sex Roles, Interruptions, and Silences in Conversation. In:
B . Thorne & N . Henley (Eds), Language and Sex: Difference and Dominance . Rowley,
MA : Newbury House Publishers .
EGOCENTRIC EMPATHY GAPS IN

SOCIAL INTERACTION AND
EXCHANGE
David Dunning, Leaf Van Boven and
George F. Loewenstein
ABSTRACT
Social exchange theories posit that people engage in diverse forms of
exchange to enhance their own interests . Knowing whom to exchange with
and what to exchange, however, requires an understanding of other
people's wants and needs. Gaining such an understanding requires skill at
perspective taking : assessing what other people's preferences are and how
they differ from one's own . We discuss a systematic bias in interpersonal
perspective taking that can limit people's ability to reap the benefits of
social and economic exchange. People systematically overestimate the
similarity between their own perspective and that of other people who are
in different psychological situations from their own . We show that such
"egocentric empathy gaps" occur in transactions between buyers and
sellers. Owners are subject to the endowment effect, valuing their
possessions more simply because they own them. Non-owners fail to
appreciate the psychological impact of endowment and thus make
imperfect choices when interacting with owners . We describe how difficult
it is for people to learn about the psychology of endowment and explain
how misunderstanding that psychology can lead to enmity and perceptions
Advances in Group Processes, Volume 18, pages 65-97.

Ali rights of reproduction in any form reserved .
ISBN : 0-7623-0767-6
65
66
DAVID DUNNING, LEAF VAN BOVEN AND GEORGE F . LOEWENSTEIN

of unfairness. We discuss the broader relevance of egocentric empathy
gaps for social policy and pluralistic ignorance .
INTRODUCTION
To thrive, or even survive, people need to engage in exchanges with other
people . Economists were the first to draw attention to the importance of
exchange (e .g. Smith, 1776/1976), but social psychologists, both on the
sociological and psychological side of the field, have also devoted considerable
attention to the problem of how people fulfill their personal goals through their
transactions with others . Whereas economists have been especially focused on
the exchange of money for goods and services, psychologists have hypothesized about and studied a much wider range of exchanges involving, for
example, compliments, favors, and even love (Blau, 1964 ; Homans, 1958) .
For an exchange to be mutually beneficial, each individual must receive
something of greater value than what he or she gives up . From the perspective
of any individual, then, information about the other person's values constitutes
an essential input into the decision of who to exchange with, what to exchange,
and the terms on which exchange should take place . Whether the objects of
exchange are tangible or intangible, social exchange is facilitated when the
parties to the exchange have an understanding of one another's preferences
(e .g . Neale & Bazerman, 1983) .
Much of bargaining theory in economics is devoted to the question of how
people infer other people's preferences from their bargaining behavior, as well
as how they communicate their own preferences through their own bargaining
behavior (Fudenberg & Tirole, 1983 ; Rubinstein, 1985) . In naturalistic
contexts, however, there is a much wider array of information that can be
brought to bear on such interpersonal "perspective taking ." People can rely on
personal information they have about the other person, stereotypes about
people with the other person's characteristics, or intuitions about the effect of
the other person's situation on his or her preferences . Successful academic
collaboration, for example, requires a complex exchange of ideas, expertise,
levels of effort, ordering of authorship, and so on . The decision of who to work
with, what level of effort to put into the collaboration and how to divide up the
sub-tasks, are all facilitated by the ability to see the project from one's
collaborators' perspectives .
Anticipating the preferences of another, however, is not a trivial matter, and
there are many pitfalls that can cause errors in taking the perspective of another
person . Successful perspective-taking often requires bridging gaps caused both
by stable differences between individuals and by transient situational factors .
Egocentric Empathy Gaps
67
Unfortunately, people rarely, if ever, possess all the information they need to
take another person's perspective, and the information they do have may be in
error. Furthermore, even when they do have the information required to take
another person's perspective accurately, they are unlikely to possess the time,
energy, or resources to use that information to arrive at an optimal judgment .
Such limitations of time, energy, or processing power constitute sources of
"bounded rationality" (March & Simon, 1958) which inevitably lead to errors
in perspective taking . Some of these errors are undoubtedly non-systematic .
For example, constraints on information or information processing, while
substantial, are likely to cause random errors in the inferences people make
about those they exchange with .
The focus of this chapter is on how difficulties in perspective-taking can
cause systematic biases in anticipating the preferences of others . In particular,
people suffer from egocentric empathy gaps : they perceive that the preferences
and reactions of other people are more similar to their own than those
preferences and reactions actually are . We trace this difficulty to the problems
people have in anticipating how they themselves would react if placed in the
other person's shoes . We also show that this difficulty leads people to make
suboptimal decisions in their dealings with others .
We argue that these difficulties in perspective taking, and the egocentric
empathy gaps they engender arise in two differing circumstances . The first
circumstance is when people possess private information - information not
shared by other persons . In these situations, people often act as though others
possess that same information (or can anticipate it) even though they
understand explicitly that the information is theirs alone . We will touch on how
this informational empathy gap can influence and disrupt social exchange . The
second circumstance arises when people are in a different emotional or drive
state, or have different preferences, than those around them . People often act as
if others feel more like they do at the moment, or share their preferences more,
than is actually the case .
We explore the specific implications of these difficulties, in particular, for the
dynamics of buying and selling . The roles of owner and buyer evoke different
psychologies, in that the informational sets, tastes, and emotional attachments
to objects differ in significant ways between owners and buyers . Successful
transactions between buyers and owners would be facilitated by understanding
these differences, but owners and buyers often fail to anticipate how much the
tastes and preferences of people in the other role differ from their own .' In this
chapter, we describe experiments that reveal egocentric empathy gaps between
owners and buyers, and discuss the consequences of these gaps for people
engaged in social exchange. We also discuss how difficult it is for people to
68
DAVID DUNNING, LEAF VAN BOVEN AND GEORGE F. LOEWENSTEIN
learn the psychology producing these gaps, and outline some implications of
that lack of learning for emotions that influence exchange relationships, as well
as for perceptions of fairness in those relationships . We address how general
these empathy gaps are, and note that people suffer these egocentric gaps
because they often cannot anticipate how they themselves would respond if
placed in the situation or role of the other person . We end by touching on the
implications of this inability for social understanding, social policy, and
pluralistic ignorance .
EGOCENTRIC EMPATHY GAPS IN PERSPECTIVE

TAKING
Both sociologists and psychologists have asserted that the ability to see the
other person's point of view is a crucial one to succeed in social interaction . For
example, George Herbert Mead (1934) asserted that :
It is generally recognized that the specifically social expressions of intelligence, or the
exercise of what is often called "social intelligence," depend upon the given individual's
ability to take the roles of, or "put himself in the place of," the other individuals implicated
with him in given social situations ; and upon his consequent sensitivity to their attitudes
toward himself and toward one another. . . . [T]his putting of one's self in the places of
others, this taking by one's self of their roles or attitudes, is not merely one of the various
aspects or expressions of intelligence or of intelligent behavior, but is the very essence of
its character (p . 141) .
Work in psychology echoes this point . The ability to empathize with others has
been linked to many benefits, such as self-reports of social competence (Davis,
1983), reduced use of stereotypes (Galinsky & Moskowitz, 2000), reduced
aggression (Richardson, Hammock, Smith, Gardener & Signo, 1994),
increased satisfaction with (and longevity of) romantic relationships (Arriaga
& Rusbult, 1998 ; Franzoi, Davis & Young, 1985 ; Long & Andrews, 1991 ;
Simpson, Ickes & Blackstone, 1995), cooperativeness in children (Johnson,
1975), and increased prosocial or altruistic behavior (Batson, 1991), to name a
few. Other benefits have been observed in settings involving exchange
relationships . Salespeople who can accurately discern the thoughts and feelings
of customers tend to make more sales (Comer & Drollinger, 1999) . Negotiators
who characteristically try to see things from the other person's point of view
are more likely to succeed, resolving more issues and reaching more optimal
agreements (Neale & Bazerman, 1983) . All this research suggests that Dale
Carnegie, the pioneering "self-help" guru, was right in his dictum that a
"formula that will work wonders for you is [to try] honestly to see things from
the other person's point of view" (Carnegie, 1936, p . 175) .
69
Perspective taking is difficult, however, and people often fall prey to

egocentric empathy gaps . A central component of our arguments is that people
fall prey to these empathy gaps because they often use themselves as a starting
point for assessing others' perceptions (Holyoak & Gordon, 1983 ; Markus,
Smith & Moreland, 1985 ; Sherif & Hovland, 1961) . For example, people tend
to base their estimates of how other people act in general on their own behavior,
in a phenomenon known as the false consensus effect, and thus overestimate
how many of their peers will behave similarly (Ross, Greene & House, 1977) .
In addition, when deciding whether another person's performance is a "good"
or "bad" one, people spontaneously refer to their own performances as a
standard of comparison, and judge the other person accordingly (Dunning &
Cohen, 1992 ; Dunning & Hayes, 1996 ; Dunning & McElwee, 1995) .
To be sure, there is a good deal of scholarship that suggests that people
should use themselves as a reference point and assume similarity between
themselves and other people . How one assesses a situation often is a good
indicator of how others will evaluate it (Dawes, 1989 ; Hoch, 1987 ; Ross &
Ward, 1995) . For example, if Jerry thinks that an automobile is a terrific buy,
then he would do well to assume that Fred thinks similarly (Dunning, Griffin,
Milojkovic & Ross, 1990) .
However, this useful rule breaks down in subtle yet important circumstances,
including those that involve exchange . It breaks down because people are not
entirely successful in anticipating how they themselves will react when placed
in different psychological situations where their responses will be influenced
by different informational, emotional, or physiological factors . For example,
consider the recent case of New York Mayor Rudolph Giuliani, as he
discovered he had prostate cancer just as he prepared for a run at the Senate
seat from New York. Speaking about arriving at a course of treatment about a
life-threatening condition that had taken the life of his father :
I mean, I didn't understand the impact of this . It was, what I thought at first was, O .K I'll
take a week or two, or three, whatever it takes, to figure out what the treatment is, I will
then decide on the treatment, and I'll figure out does that mean I can run or I can't run . . .
I know it sounds silly, but I thought of it like a budget decision or a legal decision
(Bulmiller, 2000, A14).
As a consequence, when people use their own anticipated reactions as a basis

to predict and evaluate the reactions of others in different psychological
situations, they make errors . Importantly, the situation need not change all that
dramatically for people to make errors . Such gaps in empathy can occur even
for people taking on different roles in the same situation, such as the seller and
potential buyer in an economic transaction . Buyers and sellers stand in different
psychological situations because of their roles . They differ in the emotional
70
attachments they assign to the object between them . And, as described below,
it is surprisingly difficult for them to understand and anticipate the thoughts and
feelings of the person in the other role .
INFORMATIONAL EMPATHY GAPS : HINDSIGHT AND

THE CURSE OF KNOWLEDGE
Experiments in social psychology have repeatedly documented one circumstance in which people fall prey to empathy gaps, failing to appreciate how
another person in a different psychological circumstance will think and react.
Specifically, when a person is in possession of "privileged information" not
shared by another individual, and knows full well that the second person is not
privy to it, he or she still responds as though the second person had possession
of that privileged knowledge . This bias arises because people cannot mentally
recreate their own original state of ignorance, so they fail to understand and
anticipate the behavior of others successfully .
Intrapersonal Empathy Gaps : Hindsight Bias
Once "in the know," it is difficult to recreate the world in which one did not
know. That is, once people are in the possession of a piece of information, such
as the outcome of an event or the correct answer to a question, they suffer
intrapersonal empathy gaps . They fail to anticipate how they would have
reacted if they did not have the information .
The clearest example of the bias comes from work on hindsight bias, also
known as the knew-it-all-along effect. The hindsight bias refers to people's
tendency to overestimate how much they could have anticipated that something
was true after being informed of its truthfulness (Fischhoff, 1975, 1977) . The
hindsight bias has been observed for answers to trivia questions : After being
told the correct answer, people overestimate the likelihood that they would have
gotten the question correct if asked . The bias has also been documented for
predictions of future events : Once those events take place, people overestimate
how likely they would have predicted that those events would occur (for
reviews, see Christian-Szalanski & Fobin Millham, 1991 ; Hawkins & Hastie,
1990) . Importantly, this bias arises even when people are informed of its
existence and admonished not to fail prey to it (Fischhoff, 1977 ; Kamin &
Rachlinski, 1995) . It even produces errors in memory. In experiments in which
people make predictions of future events, and then are recontacted after some
of those events occur, people consistently misremember how well they had
anticipated the events that had since transpired (e .g. Wood, 1978) .
71
Why does this hindsight bias occur? It arises because receiving information
about an event causes people to revise their representation of the information
they have about that event in three crucial ways . First, once people know the
outcome of an event, they more closely associate that outcome with the event
(e .g . Hertwig, Gigerenzer & Hoffrage, 1997) . Thus, when they think of the
event, the outcome just comes to mind more automatically . For example, if told
that Hawaii was admitted to statehood soon after Alaska, people will likely
remember that fact reflexively if asked which was the 50th state admitted to the
Union . Second, knowing the outcome prompts people to re-evaluate the
importance of various pieces of information surrounding the event . For
example, if told that the British easily won their 19th century war against the
Gurkas of Nepal, participants revise the importance they place on the fact that
the British had superior arms (Fischhoff, 1975) . Third, knowing the outcome
causes people to spin causal scenarios that would produce it (e .g . the
tremendous discipline of the British army meant they could persevere in a
difficult war), and these causal scenarios are inevitably brought to mind when
people reconsider the event . Thus, these changes in the cognitive representation
of an event, once formed, are automatically brought to mind and bias how
much the person thinks he or she could have anticipated the supposed outcome
(Fischhoff, 1975) .
Hindsight bias has implications for judgments and evaluations of others, for
it can lead good decisions to seem obvious and bad decisions to seem
foreseeable . Baron and Hershey (1988), for example, demonstrated such an
effect in a study in which participants evaluated decisions made by doctors .
Participants were given the information that the doctors had at their disposal,
and some participants were told whether the doctor's actions had led to success
or failure . The outcome information had a significant impact on appraisals of
the doctor. Doctors whose decisions led to failure were judged more harshly
than those whose decisions led to success, even though participants in both
conditions read about the same decision and were briefed on what exact
information the doctor had at his or her disposal .
Interpersonal Empathy Gaps: The Curse of Knowledge
The same cognitive processes that produce hindsight bias also contaminate
people's impressions of the information currently possessed by others . That is,
people suffer from a curse of knowledge (Camerer, Loewenstein & Weber,
1989; Nickerson, Baddeley & Freeman, 1987), behaving as if others possess
privileged information that only they themselves possess, even though they are
made explicitly aware that other people are uninformed . Camerer, Loewenstein
72
and Weber (1989) demonstrated the curse of knowledge in an experiment in

which participants played the role of brokers in a stock market, and could earn
actual money if they played their role well . Participants earned reduced profits
because they often acted as though other participants were somehow aware of
information they privately and exclusively knew - even though they explicitly
knew that others did not have this information . The curse of knowledge has also
been demonstrated in more naturalistic contexts. For example, real-life
marketers who have studied a product intensely often have a difficult time
predicting the tastes and beliefs of consumers who are not so informed about
the product (Hoch, 1980) . As a consequence, they sometimes create flawed
advertising campaigns .
People also act as though others can easily understand their feelings about
social interactions, even though it is often difficult for others to surmise a
person's true emotion, an effect termed the illusion of transparency (Gilovich,
Savitsky & Medvec, 1998). For example, in one study, students were asked to
drink a number of cups of Kool-aid, one of which had been spiked with a badtasting ingredient. They were told to disguise their emotions during the
exercise, so that others watching them taste the drinks could not tell which cup
had held the noxious sample of Kool-aid . After completing the exercise, and
although motivated to hide their true feelings, students still overestimated the
proportion of their peers who could spot when they tasted the noxious sample .
In essence, they thought their disgust had "leaked out" when it, in fact, was not
so apparent to others .
This illusion of transparency has immediate consequences for negotiation
situations. People often feel that their emotions and motives are obvious to
others when they, in fact, are not . Vorauer and Claude (1998) demonstrated this
in a study in which students negotiated with a presumed partner over a series
of social dilemma problems. Negotiators thought their motives, such as being
firm or gaining the liking of the other person, were transparent . However, they
were not . Observers watching the negotiation could only guess the negotiator's
most important goal at chance levels . One might think that this bias would be
reduced or eliminated if negotiators were given ample opportunities to explain
their actions, but that turned out not to be the case . Indeed, giving such chances
only increased negotiators' perceptions that their motives were transparent, but
did not help observers divine which goals the negotiators were pursuing .
The implications of these illusions of transparency are clear for social
interaction and bargaining . To the extent that people think their moves toward
conciliation are obvious, they will be doubly angered when those moves seem
to be ignored by other individuals . To the extent that people think their resulting
73
anger and frustration is obvious, the spiral toward more anger - and possible
breakdown of the negotiation - might be made inevitable .
ON THE PSYCHOLOGY OF BUYING AND SELLING :

THE ENDOWMENT EFFECT
The research above focuses on egocentric empathy gaps that are produced by
differences in information . People with more information often mispredict and
misunderstand the beliefs and behaviors of others not possessing that
information . But what about emotion? When a person is in one affective state,
can they understand and anticipate the behavior of another person in a different
affective state? By affective states, we refer to a number of differing
phenomena, such as anger, or hunger, or (importantly) one's emotional
response to an object - such as one's car, house, or even a coffee mug . In work
in our labs, we have begun a research program examining whether such
egocentric empathy gaps occur for situations involving more affective or
emotional phenomena . When someone attempts to understand or predict how
another person "feels" when that second person is in a different situation or
role, how successful will the first person be in his or her impression?
We have investigated this question by focusing on a phenomenon of obvious
relevance to exchange relationships . That phenomenon centers on the
psychology of ownership . Several psychological and behavioral economic
experiments have demonstrated that owners tend to place a higher value on
their possessions than do people who do not own them, a phenomenon known
as the endowment effect. The acquisition of an object causes a virtually
instantaneous change in how much the owner values it . For example,
Kahneman, Knetsch and Thaler (1990) gave coffee mugs to half the students
gathered for a classroom lecture . They were then asked what was the lowest
price at which they would sell their mug to another individual . Students not
given mugs were asked the highest price they would pay to obtain a mug . The
prices cited by owners were typically over twice the amount buyers were
willing to pay . The median price suggested by owners was $5 .25 and was only
$2 .25 for the buyers . Similar results were obtained for such objects as
binoculars and pens .
Similar phenomena have been observed outside of the laboratory or
classroom . For example, Viscusi, Magati and Huber (1987) approached people
at hardware stores and shopping malls and described an insecticide that cost
$10. They were told that using the insecticide carried some health risks, with
an injury rate of roughly . 15 out of 10,000 who used the product . Asked how
much they would pay to gain complete protection from those risks, people
74
responded on average that they would pay $3 .78 . Another group of people had
the same insecticide described, but asked how much lower the price would have
to be to accept two health risks that inflicted injuries in 2 out of every 10,000
users . People were not very enthusiastic about losing their protection against
these risks . Indeed, 77% refused to buy any product with the increased risk, no
matter how much the price was reduced .
The act of giving people an object causes them to not only be reluctant to
part with the object, but also to assign more positive characteristics to it than
they would if they did not possess it . Beggan (1992) conducted a study in
which he gave half of his participants a "cold cup insulator" to take home with
them . These participants rated the object more favorably than did groups not
given the insulators, . Indeed, they rated the insulators more favorably than a
group allowed to handle the insulators and to think about them, but just did not
get to keep them. Similarly, Samuelson and Schauser (1988) asked participants
to rate a number of potential social policies for their value and worthiness .
They discovered that respondents rated a policy more favorably when told that
it was the one already in place, an effect they termed the status quo bias .
Why would ownership cause people to value an object more - indeed, often
causing them to double the price they would assign to the object, if not more?
According to accepted theory, the endowment effect arises from how the
human "pleasure machine" is set up to deal with prospects of losses versus
prospects of benefits . According to prospect theory (Kahneman & Tversky,
1979), losses loom larger in human decision making than gains . For example,
if one walked up to other people and asked them whether they would agree to
a friendly wager to flip a single coin, with a 50% chance of winning $20 and
a 50% chance of losing the same amount, one would not find many people
willing to agree to the wager . Indeed, only a third of individuals given an
opportunity to wager on a 50% chance of winning $200 versus a 50% chance
of losing $100 choose to take up the bet, even though taking the bet has an
expected value of $50 (Tversky & Shafir, 1992) . This reluctance to gamble
arises because the disutility or displeasure caused by a loss of size X is of
greater magnitude than the utility and pleasure associated with an equivalent
gain of size X. One can readily see how the relative impact of gains and losses
would produce the endowment effect . When a potential buyer examines the
prospect of obtaining a mug, the utility associated with that gain is not as great
as the potential disutility experienced by an owner contemplating giving up the
same mug .
Other studies have ruled out alternative explanations for the endowment
effect . The phenomenon does not appear to arise out of a general reluctance to
trade, in that participants are quite willing to buy and sell tokens that can be
75
redeemed for a specified amount of money, and show no difference in the value
they place on those tokens . Also, the phenomenon does not appear to arise from
strategic pricing . Even when told that they can have multiple chances to buy or
sell mugs, buyers and owners fail to change their prices in strategic ways, first
asking for advantageous prices and then moderating their demands once those
prices are unsuccessful (Kahneman et al ., 1990) . The endowment phenomenon
also arises in situations in which strategic considerations are precluded . In one
such study, students were given a choice between two different rewards for
participating in an experiment . One group was given a choice between
receiving a coffee mug and a bar of Swiss chocolate, and 56% chose the mug .
In another group, students were initially given a coffee mug and then asked if
they would like to exchange their mug for the exact same chocolate bar . A full
89% opted to keep the mug . In a third group, students were initially given a
chocolate bar, and then asked if they would like to trade for a mug . In this
condition, only 10% exchanged the candy for the mug (Knetsch, 1989) .
In sum, several studies have revealed that people value an item more once
they are endowed with it. The implications of this phenomenon for exchange
relations are clear : To obtain a good, buyers must often offer a price that they
initially think is too high . To sell an item, owners must realize that potential
buyers may not imbue the item with the same value that they assign to it, and
so must lower their demands beyond what they might think is reasonable . To
the extent that neither side understands that they must make the adjustment,
exchanges will not occur . Indeed, it has been noted by behavioral economists
that the endowment effect explains why rates of trading between buyers and
sellers are often surprisingly low (Kahneman et al ., 1990) . Buyers and owners
diverge in the value they place on objects, and often those differences preclude
satisfactory trades.
EMPATHY GAPS BETWEEN BUYERS AND OWNERS :

MISPERCEPTIONS OF THE ENDOWMENT EFFECT
Although the tastes and preferences of owners and buyers diverge, the
consequences of those differences for social exchange would be alleviated if
people possessed at least an adequate understanding of the psychology of
ownership . If they could "go outside themselves" and imagine how someone
in the other role feels about the good or commodity in question, they could
have a better shot at making trades that could be mutually satisfying . For
example, buyers could make higher bids for another person's possessions,
confident in the knowledge that he or she would (instantaneously) value those
possessions more once he or she had them, with the gain in utility outstripping
76 DAVID DUNNING, LEAF VAN BOVEN AND GEORGE F . LOEWENSTEIN

the pain of the higher price . However, do people understand this principle about
the psychology of ownership? Do they understand how ownership (or the lack
thereof) changes the value assigned to goods and commodities? In a series of
laboratory experiments, we examined whether people possess an adequate
understanding of the impact of this endowment effect on the tastes and actions
of their peers .
Intrapersonal Empathy Gaps : Biased Predictions of Oneself
At first blush, it would appear obvious that people should have some intuition
that owners value objects more than non-owners . After all, people often
experience the task of buying and selling objects, and occasionally have a
chance to negotiate with people in the other role . People should therefore be
intimately familiar with the psychology of ownership . However, there is
growing evidence that people may not be perfectly "in tune" with the impact
that ownership has on their own valuation of goods and commodities .
Consistent with our thinking about the genesis of egocentric empathy gaps,
people fail to realize how much their own tastes and preferences will change
once they are placed in the role of ownership . This failure was demonstrated
by two experiments performed by Loewenstein and Adler (1995). They showed
students from Carnegie-Mellon University and the University of Pittsburgh
some mugs with the appropriate college logo . Students were told that they
would be given one of the mugs and would be given the opportunity to sell it
back to the experimenter . Their task was to estimate the price at which they
would sell the mug . Students were presented with a list of prices that began
with $ .25 and continued in 25-cent increments to $10 . At each price, students
were asked whether they thought that they would sell or keep the mug . On
average, students predicted that they would sell the mug from around $3 .27 to
$3 .73, depending on the specific class they were in . However, when actually
given a mug and given a real chance to sell it back to the experimenters,
students cited significantly higher prices for their mugs, from around $4 .56 to
$5 .40.
In a second experiment, Loewenstein and Adler (1995) discovered similar
mispredictions by non-owners in a situation in which they had incentives to be
accurate. Namely, students were told that they might receive a mug, and that
they should report the lowest price at which they would sell that mug for in case
they did get one . The authors presumed that people with an uncertain prospect
of receiving a mug would not yet feel endowed with it, and thus would
underestimate how much they would value the mug once they owned it . This
77
prediction was borne out, with non-owners significantly underestimating, by

$1 .80, how much they would ask for the mug, once they owned it .
Interpersonal Empathy Gaps : Biased Predictions of Others
The studies conducted by Loewenstein and Adler (1995) showed that people
experience intrapersonal empathy gaps when they predict how they would
respond in a different role, with people mispredicting how much owning a mug
would influence their preferences . But would these intrapersonal gaps produce
interpersonal ones? When predicting the tastes and preferences of another
person, would people again underestimate the impact of the endowment
effect?
In a series of studies, we examined the extent to which people could
accurately predict the behavior of owners and non-owners when they were in
the other role . In one such study (Van Boven, Dunning & Loewenstein, 2000,
Study 1), in a classroom setting, we gave roughly half the students a Cornell
University travel mug, worth about six dollars at the local bookstore . We then
offered everyone in the room a deal . Owners were asked whether they would
sell the mug at a number of prices that began at $0 .50 and rose in 50-cent
increments to $10 . They were told that the experimenter would name a
randomly selected price at the end of the session, and that if the student had
said that he or she would sell the mug at that price, then he or she must do so .
Students not owning mugs were given a questionnaire listing the same series of
prices, and were asked at each price whether they would buy a mug at that
price . As with the owners, the experimenter told non-owners that he would
name a randomly selected price at the end of the experimental session, and that
if the student had mentioned that he or she would buy a mug at that price, than
he or she would have to do so . The experimenter explained that he took checks
and IOUs . Not surprisingly, the choices of students revealed a strong
endowment effect - the lowest selling price cited by mug owners on average
($5 .40) was over three times higher than the highest buying price reported by
non-owners ($1 .56) .
Of key importance, however, was an additional set of questions in which
students were asked to estimate the preferences and wishes of people
occupying the other role . Non-owners were asked to forecast the lowest price
at which the average owner would sell the mug . Owners were asked to predict
the highest price the average non-owner was willing to pay to buy a mug .
Owners and non-owners answered these questions by filling out the exact
questionnaire that had been completed by students in the other role .
78
Did students successfully anticipate the preferences of people in the other

role? The answer was a glass half empty and half full . The estimates of owners
and non-owners revealed that they understood that ownership would make a
difference in people's tastes and preferences, but they failed to anticipate fully
the magnitude of those differences . Non-owners realized that owners would ask
for more money to part with their mugs than non-owners would pay to acquire
one, but they still underestimated the owners' lowest selling prices by $1 .44, or
by roughly 25%, a significant difference . Owners, similarly, realized that nonowners would be willing to pay less money to buy the mug than they demanded
to part with one themselves, but owners significantly overestimated how much
non-owners would pay by $1 .37 - an 88% difference . A second study
replicated these findings, and found that a reward for providing correct
estimates of the other side did not reduce the amount of error in those
estimates .
Empathy Gaps and Market Behavior
Two additional studies indicated that non-owners underestimate owners' lowest
selling price, and that this underestimation can produce tangible costs in an
experimental market . Specifically, we asked students to attempt to complete a
one-shot exchange that could earn them some money (Van Boven, Dunning et
al ., 2000 Study 3) . Students in psychology and economics classes took part and
we gave mugs to a random selection of half the students . These students were
asked to report to us the lowest price at which they would sell their mug . The
other half of students were given instructions detailing that they would be given
a chance to buy the mugs back for us ; they would play the role of buyer's agent,
not actually acquiring the object for themselves but acting for a third party (us)
who wanted the object . Such buyer's agents are common in many different real
world markets . In some cities, buyers' agents commonly inspect homes and
makes bids for their customers . In many states, antique dealers scour antique
stores and fairs for items their clients wish to acquire .
In our experiment, students in the role of buyer's agent were given $10 and
paired with one of the mug owners . Their task was to name a buying price at
which they thought the mug owner would sell . If the mug owner did sell, we
would get the mug back and the agent would get whatever was left over of the
$10 . However, if their bid failed, they would receive nothing and the owner
would keep the mug . There would be no opportunity to make a second bid .
Thus, it was in the agent's best interest to intuit the lowest price at which the
owner would sell the mug, and bid that amount or slightly more .
79
Buyers' agents' made bids that were too low, thus revealing an inadequate
understanding of the power of the endowment effect. On average, the agents
bid $4 .92, but mug owners typically wanted at least $6 .83 for their mugs, so
only 25% of the agents succeeded making a bid high enough to be successful
and earn some money. A supplementary analysis revealed that buyer's agents
had made bids that were not "rational," in the sense that they had produced
outcomes that were suboptimal . An ancillary analysis of owners' selling prices
indicated that a bid of $6 .50 would have been optimal for buyers' agents . If all
agents bid $6 .50, then, given the distribution of owners' selling prices, 62% of
bids would have been accepted each with a profit of $3 .50, and thus the average
profit - that is, the expected value - would have been $2 .15 . Higher bids would
have been more frequently accepted, but at a lower profit ; lower bids would
have been more often rejected, even if an accepted bid would have produced
greater profit . Buyers' agents' low bids produced an average profit across all
transactions (successful and unsuccessful) of only $ .75, or only 35% of what
they could have earned if they had bid optimally .
In short, in a one-shot exchange situation, buyer's agents showed an
imperfect understanding of the endowment effect, thinking that owners would
value commodities in much the same way as they did themselves . This flaw in
perspective taking produced tangible costs, prompting them to bid too low for
the owner's possessions and thus miss out on chances to earn money . However,
although our studies (so far) only focused on the behavior of buyer's agents in
one-shot exchanges, we have reason to believe that the misunderstandings
about the psychology of ownership occur more generally . We will describe in
the next section how misunderstandings of the endowment effect can lead to
difficulties in situations involving multiple chances to exchange .
We can also speculate that people other than buyers and their agents - other
non-owners, that is - will similarly misperceive the psychology of the owner .
Although we have no formal data, we have no reason to believe, for example,
that agents acting on the behalf of owners will understand their client's
valuation any better than anyone else . Because seller's agents would not own
the object themselves, they, too, will underestimate the impact of endowment
of an owner's preferences . Indeed, in one best-selling book on buying a home,
co-written by a real estate agent who has represented many sellers over the
years, comes the following statement :
We think the average buyer is brighter than the average seller . How else can you explain
why buyers are generally so much more realistic about property prices? It's not as though
there are two different real estate markets : an expensive one for sellers and a cheap one for
buyers. Sellers have access to exactly the same comps [comparison prices] buyers do . Yet
buyers' initial offering prices tend to be far more realistic than sellers' initial asking
80

prices . . . . [Thus, the] better you understand the warped thought processes of these sellers,
the better you can handle their unreasonable objections to your eminently fair offer (Tyson
& Brown, 1996, pp . 210-211).
PROBLEMS IN LEARNING ABOUT THE ENDOWMENT

EFFECT
Taken together, the studies described thus far reveal that people suffer from
egocentric empathy gaps in situations involving buying and selling . Nonowners underestimate how much owners value their possessions ; owners
overestimate how much non-owners will pay for the same possessions .
Importantly, this gap arises even in situations in which people could profit from
an adequate understanding of the psychology of ownership . Participants
playing the role of buyer's agent did not anticipate how much owners would
want for their possessions, and so made bids that left money on the table.
In short, participants in these studies had shown an inadequate understanding
of the dynamics of the endowment effect . They had not realized how impactful
the role of ownership could be . But what about after the fact? Once buyer's
agents saw how much owners wanted for their possessions, would they come
to some understanding of the endowment effect in social transactions?
Misunderstanding the Endowment Effect
In subsequent studies we addressed the question whether people could learn to
recognize fully the endowment effect . One might imagine that people could
learn about the dynamics of endowment through experience and that, through
observing the actions of others (and themselves), they could come to appreciate
how ownership causes people to value a commodity more . However, our initial
studies on learning produced some pessimistic conclusions . After witnessing
the operation of the endowment effect first-hand, participants did not naturally
recognize the effect for what it is . Rather, they attributed the surprising
behavior of others to anything but the operation of the endowment effect .
This lack of learning was revealed by a second buyer's agent study in which
buyers' agents again made bids for a mug owned by another person . In this
study, agents made bids that were too low, bidding on average $6 .19 on mugs
that owners, on average, wanted $7 .38 for. Only 19% of the agents in this
replication study succeeded in producing bids that made them money . When
buyer's agents bids had failed, we asked them to explain the reasons for that
failure . These spontaneous explanations revealed little awareness of the
endowment effect . Indeed, they tended to put the blame squarely on the
shoulders of the owners, offering such comments as "The owner wanted to get
81
more money than the mug was really worth," "The owner didn't want me to
make any money," and "The owner must be from the city [New York City]
where things are crazy expensive!"
More formal analyses revealed that both owners and buyer's agents tended
to lay blame for failure at the feet of the other person as opposed to the
endowment effect . In close-ended questions, owners and agents were given a
number of plausible explanations for why the bid had failed and asked to rate
how reasonable they thought the explanations were . Both owners and agents
rated personal greed on the part of the person in the other role as a more
plausible explanation than they did the endowment effect, even when the
endowment effect was explicitly described to them .
In short, participants revealed a tendency known to all social psychologists
on the psychological side of the ledger . They had underestimated the power of
the situation - or, in this case, the power of the role of being an owner - and
instead attributed the failure of the bid to the personality of the other person .
That tendency is referred to, in different quarters, as either the fundamental
attribution error (Nisbett & Ross, 1980) or correspondence bias (Jones, 1990) .
In general, this tendency refers to the underweighing of situational forces and
overweighing of personal characteristics in the explanation, prediction, and
judgment of the behavior of other people .
The presence of the correspondence bias may explain why it would be
difficult for people to learn about the presence and magnitude of the
endowment effect in social transactions . As people buy, sell, haggle, and argue
with others over the exchange of commodities, the tendency of people to
attribute transaction failures to something about the other individual will
prevent them from learning about the general dynamic of the endowment
effect, no matter how much experience they gain . For example, if John
repeatedly bargains with Marsha, he may learn that he will often have to offer
more than he would initially expect. However, although he may get the
behavior right, this will not necessarily mean that he will gain insight into why
offering more is the correct behavior to bring about successful exchanges . As
suggested in our studies on attribution, John may learn what to offer, but he
may come to attribute the need to offer more to Marsha's greed rather than to
the general psychology of endowment that operates on everybody .
Attributing failed transactions to other people may hinder individuals from
reaching an understanding of endowment even if they haggle with several
individuals over a wide variety of objects . This is because such experiences
tend to be piecemeal . Each person must attempt to consummate a transaction
with Person A over Object X, then negotiate with Person B about Object Y, and
so on . As they do, they may miss the general pattern that people want more for
82
their possessions because they readily attribute their failure with Person A to
his or her personality (greed), their failure with Person B to his or her
intelligence (misinformed about the true value of the object), and so on . That
is, once failures and surprises are attributed to the other person, the individual
is unlikely to seek out a general situational dynamic, such as the endowment
effect, that could explain the common pattern in others' behavior.
The (Non)Impact of Formal Instruction
Given the difficulty in learning about the endowment effect in a piecemeal
fashion, one might place more faith in formal instruction . Perhaps if people
were lectured about the endowment effect and its impact in the classroom, they
would have a better appreciation of it in their own lives . However, data again
gives cause for pessimism about the potential of formal education to produce
an adequate understanding of the endowment effect . Recall our description of
the first study of Van Boven, Dunning et al . (2000) in which we strolled into a
series of classrooms, gave half the students mugs, and asked owners and nonowners to estimate the value that people in the other role assigned to those
mugs . As we debriefed students after their participation in the study was over,
we became aware that some students had just had a lecture on the endowment
effect a few weeks before in a social psychology class . Indeed, the prime
experimental example used in the lecture had been the mug studies we had
based our procedure on . Thus, we decided to look at the estimates of students
who were enrolled in that class versus those who were not . Interestingly, the
estimates of students in that course were no more accurate than those of
students not so "educated ."
The (Partial) Impact of Experience with the Endowment Effect
But perhaps we are too pessimistic . Perhaps, to gain insight into the dynamics
of endowment, the best experience would be to negotiate with a number of
different individuals . In that way, people may learn the general principle that
people value their possessions to an unexpected degree . We have explored this
possibility in a series of follow-up buyer's agent experiments . In these
experiments, we examined whether the lack of understanding about the
endowment effect survived repeated experiences with the preferences and
tastes of owners . We brought participants into the laboratory and gave mugs to
half of them. The other half of the participants were asked to play the role of
buyer's agent, but had a chance to make bids on the mugs of five different
owners . After each bid, they received feedback about whether their bids had
83
succeeded . Buyers' agents raised their bid on each successive trial . On the first
trial, they made a bid that, on average, was only 75% of what owners typically
asked for. By the fifth try, their bid was 88% of what owners asked for,
indicating that they had learned that they must change their behavior to make
a successful bid (Van Boven, Loewenstein & Dunning, 2000) .
However, that learning was rather narrow . After the fifth bid, the
experimenter distributed pens to the mug owners, and asked buyer's agents to
make bids on those pens .' Buyer's agents made initial bids that were only 75%
of what the owners typically asked for, and the increase in their bids on
subsequent trials was no steeper than it had been during the trials with the
mugs . In short, giving non-owners experience with owners in which they
gathered information about how much owners valued their mugs did not afford
any generalized understanding of the psychology of ownership . Buyers' agents
started on "square one" when the commodity in question was switched from
mugs to pens, and they showed no "savings" in their learning curve . Buyers'
agents seem to have learned that sellers, for some reason, valued the mug more
than they expected them to, but they did not attribute the unexpectedly high
prices to the endowment effect . The experience with mugs did not teach them
a lesson to be generalized to another (even slightly different) situation (Van
Boven, Loewenstein & Dunning, 2000) .
MISUNDERSTANDING THE ENDOWMENT EFFECT

AND RATIONALITY
Our argument is that it is intrinsically difficult for people to anticipate and
understand the beliefs, preferences, and actions of others who are in different
psychological situations or roles . People predict the thoughts, feelings, and
behavior of others in a different situation by first imagining themselves in that
situation and then adjusting their estimate for known differences between
themselves and the other person . Such perspective-taking goes awry in the first
stage of the process because people have difficulty projecting themselves into
a situation or role different from the one they are currently in .
Rational Choices Misapplied?
However, one can look at our experiments, particularly the buyer's agent
studies described above (Van Boven, Dunning et al ., 2000), and disagree with
our assessment that participants are acting irrationally . To be sure, participants
in our buyer's agent studies made suboptimal choices, bidding too low for the
coffee mugs owned by other students, but perhaps this was simply because they
84
were applying a rational script, based on everyday experience, that leads to

optimal outcomes in their day-to-day dealings but fails to do so in the
laboratory situation we set up . Psychologists often observe that the
"heuristics" that people use to make judgments and reach decisions can be
rational in a number of common situations though they lead to mistakes in
some specialized situations (Tversky & Kahneman, 1974) .
If one looks closely at the social structure of exchange that we set up in our
buyer's agent studies, one sees that it differs in two ways from the typical
situation buyers and sellers confront in their everyday worlds . First, we allowed
buyer's agents only one bid for the owner's coffee cup . We did not let the
owner and agent haggle, nor did we allow subsequent bids . In the everyday
world, both bargaining and multiple bids are allowed . Because of this, people
start off the negotiation with an advantageous, but unrealistic bid . The owners
ask for too much ; the agents offer too little . In our experiment, buyer's agents
may have applied this rational script . (Note, however, that because we allowed
only one bid, the application of this otherwise rational script is irrational .)
Second, in the everyday world, people often are not constrained to dealing
with just one owner . If one owner sets a high price, the buyer is allowed to seek
out other owners who might ask for a lower price, perhaps even an
unreasonably low price. In our laboratory situation, however, buyers' agents
were not allowed to try again with different owners . Thus, they may have
applied a script in which one bids a low price in the hopes that the specific
owner one is dealing with just happens to want very little for his or her mug .
Because buyer's agents could not pursue this strategy with multiple owners,
they were left with suboptimal outcomes .
Providing Buyers' Agents with Insight
However, evidence from another study we conducted weighs against these
interpretations of our findings . This study showed that buyer's agents bid too
little money for the owners' mugs exactly because they could not successfully
project themselves into the psychology of being an owner . Giving them a
chance to "walk" in the owner's shoes for a while led them to make more
optimal bids for the coffee mugs . Van Boven, Dunning and Loewenstein (2000,
Study 5) brought students into the laboratory and assigned them to play the role
of buyer's agents . Before they made a bid, however, they were asked to
estimate what price they would set if they were owners. This introspection
exercise did not help agents to gain insight into the owner's behavior or to make
more optimal bids . Students in this condition underestimated how much they
85
would ask for. Echoing previous studies, they also made bids that were too
low .
However, in a second group, we were able to induce agents to gain insight
into their own potential behavior and to make more optimal bids . We did this
by asking them, like we did for the first group, to guess how much they would
demand if they were an owner - but only after giving them their own mug . Now
placed in the exact same situation as the owners, with their own mug to think
about and consider selling, agents made significantly more accurate predictions
about what their demands would be if asked to sell the mug . In addition, when
bidding for another person's mug, they made significantly higher bids .
Supplemental analyses revealed that these higher bids occurred because of
the insight these agents had gained about their own potential behavior as an
owner. Because they could more successfully project themselves into the
psychology of ownership, and could realize that they would make high
demands themselves, they were significantly more likely to make high bids for
the mug of another individual. That is, awareness that one would ask for a high
price one's self statistically mediated the relationship between having a mug
and bidding higher for another person's mug .
IMPLICATIONS FOR THE SOCIAL PSYCHOLOGY OF

EXCHANGE
Taken together, our findings on egocentric empathy gaps between buyers and
owners have diverse implications for social psychological theorizing about
social exchange . More specifically, the findings heighten and extend recent
thought about emotion in social exchange, as well as the issue of fairness .
Emotion
Recent work in the sociology of exchange has repeatedly demonstrated the
importance of emotion . Emotions come into play at various points of the
exchange process . People display or hide their emotions according to social
norms surrounding exchange interactions . People use their emotions as
information concerning what they think about the exchange and the person they
are exchanging with . Emotions can also bias interpretations of events, with
negative emotions leading people to make more unfavorable interpretations of
the exchanges than they would otherwise make . Emotions such as guilt and
sympathy can serve to curb the excesses of self-interest in exchange . Favorable
emotions toward another person can also lead to cohesion and cooperation (see
86
Lawler & Yoon, 1996, 1998 ; for a review of how emotions may come into play
at various stages of social exchange, see Lawler & Thye, 1999) .
Our studies reveal how emotion, particularly negative emotion, can be
produced during social exchange . After one of our studies, conducted in a
classroom, owners were asked to reveal how much they wanted for their coffee
mugs to the set of buyer's agents collected there . After the owners began to
report surprisingly high prices - at least to the buyer's agents - one agent was
heard to exclaim, "Those f-king owners are the most f-king greediest people
I've ever seen!" The angry tone in this young woman's voice was a surprise to
us, as was the number of buyer's agents who nodded their heads in
agreement.
Many buyer's agents in our studies shared this woman's belief that the
owners were greedy . When given a chance to attribute the failure of a
transaction to the endowment effect or to greed, both owners and buyer's agents
endorsed greed with more enthusiasm (Van Boven, Dunning et al ., 2000 ; Study
4) . We do not know if they shared the intensity of this belief expressed by this
one outspoken participant. However, one can see how misunderstanding the
endowment effect can lead to attributions that promote negative affect . To the
extent that buyers attribute the failure or difficulty of an exchange to something
"wrong" about the other person, they will naturally become angry (see Smith
& Ellsworth, 1985 ; Weiner, 1985, 1986) .
This anger can lead to intransigence in a negotiation . It can even lead to
breaking off the negotiation . Several studies have shown that people's
impressions of the intentions of other people color the decisions they make in
negotiations or exchange . For example, they may be willing to accept a
settlement if a computer randomly generates it, but they may reject the exact
same settlement if they think another person proposed it, one possibly
harboring self-interested or malevolent motives (e .g . Blount, 1995 ; Larrick &
Blount, 1997) . Beyond this, emotions can color the interpretation of the other
person's subsequent actions . The simple presence of certain emotions, such as
happiness, sadness, and anger, tend to bring to mind different cognitions and
beliefs more readily, which can alter the interpretations of actions and events
(Bower, 1991) . Thus, misunderstandings of the dynamics of endowment may
be the source of emotions that make easy exchanges more difficult to
consummate .
Fairness
When people enter into exchanges, they do so not so much to promote their
self-interest as they do to promote it fairly. That is, people are sensitive to how
87
well their benefits compare with those gained by others (Loewenstein,

Thompson & Bazerman, 1989) . They will reject outcomes that are beneficial if
they believe the other person is unfairly benefiting more . Nowhere is this
clearer than in experiments on the ultimatum game . In the game, two people are
asked to play two different roles . An amount of money, say $10, is given to
Person 1, who must propose a way to split the money with the other person .
Only one proposal can be given . Person 2 is given the task of either accepting
or rejecting the split. If the split is accepted, than the money is distributed
according to the proposal . If the split is rejected, the money is given back to the
experimenter and neither person benefits. Studies show that the second person
will reject splits even when they benefit that person in an absolute sense (e .g .
Guth, Schmittberger & Schwarze, 1982) . For example, if Person 1 suggests a
split of $9 .95 to himself or herself and a nickel for Person 2, the split is almost
always rejected - even though Person 2 does benefit from the split . (As any
economist can show, a person is better off with a nickel than with nothing,
Q .E.D .)
The problem with fairness is that people often have different impressions of
what is fair, and usually believe that the fair resolution is one that will benefit
themselves relative to the other person (Thompson & Loewenstein, 1992 ;
Babcock & Loewenstein, 1997) . The psychology of endowment contributes to
this state of affairs . If people give more weight to their losses than to potential
gains in any exchange, it is easy for both sides to feel they have been unduly
sacrificing and that the other side has not been sacrificing enough . Our work
shows that people may not understand this dynamic in negotiation and
exchange. When the other side claims that their sacrifice is a great one, or when
they begin to complain about the unfairness of a proposed settlement, one
might find such statements to be inexplicable . Or, one might decide that these
statements reveal either calculated strategy (e .g . clearly they know that's not
fair, so they're saying that just to take advantage of me), malevolent avarice
(e .g . they really are greedy, after all), or, at best, evidence that the other side is
misinformed.
This problem has been pointed out before (Thompson & Loewenstein,
1992) . However, our studies suggest that the problem may be more severe than
previously expected . In past thinking about this issue, people have noted how
bad feelings can arise between two people because of different ideas about
fairness . Our work suggests, however, that people might have some insight into
the fact that others will think differently from them . They will understand, for
example, that owners want more for their goods than buyers initially want to
offer, and then they try to change their behavior or their beliefs accordingly .
However, our work suggests that the adjustments they make will be insufficient.
88
For example, they will realize that owners ask for higher prices, but will fail to
realize just how much more owners will need to part with their mugs . As a
consequence, the behavior of people in the other role will be doubly surprising .
People will have thought that they have already corrected for the impact of the
other person's role . Thus, there will be no further attribution to make other than
the other person is being strategic, selfish, or misinformed . Partial success in
perspective taking, therefore, could potentially be worse than no success at
all .
UNDERESTIMATING THE EFFECTS OF AFFECT

At its most general, our analysis is that people fail to anticipate or understand
the impact of affective states on their own behavior and, as a result,
misunderstand the impact of these same states on the behavior of others .
Loewenstein (1999 ; Loewenstein & Schkade, 2000) refer to this problem as a
hot/cold empathy gap . One example is the endowment effect, in which
ownership causes greater attachment to an object than a potential buyer might
foresee, but most likely there are myriad similar examples in which people
cannot envision the impact of other emotional states .
We have begun to investigate some of those examples in our research . In one
such project, we are investigating whether people can anticipate the impact that
fear of social embarrassment has on the behavior of oneself and others (Van
Boven, Loewenstein, Welch & Dunning, 2000) . In one such study, a guest
instructor made a proposal to half the class, randomly chosen . They were asked
if they would volunteer to come up to the front of the class to dance to the song
"Superfreak" by R&B legend Rick James for 5 minutes . They would be paid
$5 for their performance . The other half of the class was not given this
proposal, but were rather asked if they would have volunteered if given the
chance . All students were asked whether a randomly selected student (other
than themselves) who was given the opportunity to dance would agree to do
so.
Students not facing the actual choice underestimated the impact of the
potential social embarrassment on their own behavior and that of their peers .
Whereas only 8% of the students actually given the choice volunteered to
dance, a full 33% of the students not given the offer stated that they would have
volunteered, and that 36% of their peers would have done likewise . Students
given the offer were wiser about the potential actions of their peers . They
predicted that only 16% of their peers would volunteer - a total that was much
closer to the proportion of volunteers actually observed .
89
On Social Policy
If people commonly underestimate the impact of affective states on the

behavior of their peers - because they underestimate how those states would
influence their own behavior - there are many implications for social policy.
Successful social policy often depends on understanding and anticipating how
people will feel in relevant situations . For example, anti-drug policies must
anticipate the intensity of the cravings that addicts feel for their drug of choice .
To prevent date rape, one must anticipate the likely behavior of potential
perpetrators when sexually aroused . However, policy decisions on how to
confront these issues are often made by government officials sitting in
comfortable conference rooms, a psychological situation that is far away from
one facing, for example, the drug addict or sexual offender . On a more personal
level, people regulate their own behavior based on how they think they, or
acquaintances, will feel about certain situations (e .g. will I be devastated if she
turns me down for a date?) when they have yet to confront those situations . To
the extent that people are not calibrated about their own or another person's
emotional responses, the decisions they reach may be imperfect .
Although preliminary, recent research suggests that people are not good at
anticipating the impact of affective states on human behavior. More
specifically, they do not do a perfect job understanding how such states will
influence their own behavior. For example, people's current level of hunger
influences their preferences for food a week hence : When people are hungry,
they choose an unhealthy snack (e .g . a Snickers bar) more than when they are
not hungry. When not hungry, people are more likely to choose a healthy snack
(e.g . an apple) than when hungry . These effects occur even when the
respondent knows when the chosen items will be delivered a week later - i .e .
at a time when he or she is likely to be hungry or satiated (Read & van
Leeuwen, 1998) . Similarly, college-aged men do not seem to anticipate
accurately how they will respond to situations when they are sexually aroused
(Loewenstein, Nagin & Paternoster, 1997) . When in an aroused state, they are
significantly more likely to predict that they would act in a sexually aggressive
manner than they do when they are not in such a state .
The inability to project one's self into emotionally-laden situations can also
cause people to overestimate the impact of positive and negative events on their
emotional well-being . A growing body of research suggests that people
overpredict how long they will be devastated by aversive events and elated over
positive ones . For example, Gilbert and colleagues (Gilbert, Pinel, Wilson,
Blumberg & Wheatley, 1998) asked assistant professors coming up for tenure
to predict how happy they would be if they obtained tenure, and also how
90
saddened they would be if their tenure was denied . A few months after tenure
decisions had been handed down, these same individuals were approached
again and asked about their emotional state . Those receiving tenure were not as
happy as they thought they would be ; and those denied tenure were not as
devastated as they had anticipated. A similar pattern of findings arose for
people obtaining HIV tests at a local health clinic . Participants overpredicted
how anxious and distressed they would be several weeks after receiving a
positive result suggesting they had the virus . They also overestimated how
relieved they would be to get back a negative result indicating perfect health
(Sieff, Dawes & Loewenstein, 1999) .
Misunderstanding the power of affective states might lead to two different
types of errors in the conduct of social policy . First, it may lead people to
propose interventions that miss the mark . If people cannot anticipate the power
of drug addiction, for example, they may propose treatment strategies that are
too weak to be successful . Second, misunderstanding the power of emotional
or physiological states might lead people to make unwarranted attributions for
why social policies fail . For instance, if a drug intervention fails, people might
attribute the failure to weaknesses in the drug addict's character as opposed to
the strength of the addiction. Thus, people may abandon social policies too
quickly. They might decide that the subjects of the intervention are
uncooperative, not trying hard enough, or are somehow unworthy . However,
the true source of the intervention's failure might be the weakness of the
intervention in the face of powerful viseral or emotional states, and thus more
powerful interventions might be called for at exactly the time that the general
public might be tempted to abandon hope .
On Pluralistic Ignorance
We have described a tendency for people to overestimate the similarity between
their own perceptions and the perceptions of other people who are in a different
psychological state . Sometimes, however, people appear to underestimate the
similarity between their own perceptions and the perceptions of other people
who are in the same psychological state . Such is the case with pluralistic
ignorance, or the tendency for people to estimate that a policy or norm is
widely supported by a group of people when the policy or norm, in fact, enjoys
little support (O'Gorman, 1975, 1976) . One reason for pluralistic ignorance is
that people fail to recognize that other people live "internal lives" filled with
anxiety, dread, and doubt similar to people themselves . For example, college
students fail to realize that their peers are often wracked with the same feelings
of bashfulness, self-criticism, indecisiveness, and self-consciousness as they
91
are themselves (Miller & McFarland, 1987 ; Van Boven, 2000) . If placed in a
classroom with other students, and an instructor walks to the front of the room
to deliver an incomprehensible and incoherent lecture, they fail to realize that
others are just as bewildered and anxious as they are (Miller & McFarland,
1987) .
Why might people overestimate the similarity between themselves and
others in some situations, experiencing egocentric empathy gaps, but
underestimate the similarity between themselves and others in other situations,
experiencing pluralistic ignorance? We believe that the discovery of an answer
- or answers as will likely be the case - will constitute significant advance of
social psychological theory. We do not pretend to have that answer (yet) . But
one possibility concerns people's beliefs about the observability of different
emotional states . The instances of pluralistic ignorance described above all
involve self-conscious emotions . If people believe that such emotions are
observable then they need not use themselves as a basis for predicting others
because they think (erroneously) they can simply look at others and infer their
internal states . Note that (at least) two things are necessary for people to
observe others' emotions and avoid using themselves as information : First, they
must believe that other people's internal states are observable ; second, they
must be able to observe other people . In our studies, at least one of these was
never true . In the studies of buyers and sellers, for example, people did not see
the particular person they were paired with . Even if they did, they may not have
believed that feelings associated with ownership are readily observable . In the
"Superfreak" study, which involved the self-conscious emotion of embarrassment, people did not know exactly who was assigned to which condition, so
they may have looked to themselves instead of looking to others (Van Boven,
Loewenstein et al ., 2000) . Only time will tell if this speculation withstands
empirical examination .
CONCLUDING REMARKS
We began this chapter by noting that such diverse commentators as George
Herbert Meade and Dale Carnegie have extolled the importance of accurate
perspective taking . To deal with others successfully, one must have an adequate
understanding of their thoughts, beliefs, and desires . However, we also noted
that perspective taking is an intrinsically difficult task, so that often people do
not know how people in other roles will perceive situations and react . In
particular, we noted that the role of ownership produces a different attitude
toward an object, and non-owners do not necessarily anticipate that change in
attitude .
92
Along the way, we noted how this lack of understanding between owners and
non-owners might be a specific instance of a more general issue in exchange
relationships - as well as in many social interactions. People may have a
difficult time anticipating how they, and thus others, will react when confronted
with any situation that produces an emotional overtone. Without feeling that
emotion themselves, people may fail to anticipate what impact a situation
might have on the behavior and preferences of others, respond to them in a
suboptimal way, and draw incorrect lessons about the actions they subsequently
observe .
We have no idea how pervasive these empathy gaps are or how far-reaching
their consequences are for interpersonal interactions . We focused here on the
consequences of egocentric empathy gaps for buyers and sellers engaged in
economic exchanges. However, the situations we produced in the lab are
hardly the most emotional ones people might confront as they go about their
lives. Emotions in the real world can run high, and thus we can only speculate
that such empathy gaps are likely to be even more pervasive and consequential
in the types of emotionally charged interactions that people might encounter
outside of the laboratory .
NOTES
1 . In essence, buyers and sellers suffer a version of a problem that has plagued
theoretical economists and sociological exchange theorists for years : It is often difficult
to make an interpersonal comparison between how much two different people would
attach to an object or event (Heckathorn, 1983 ; Homans, 1961 ; Luce & Raiffa, 1957) .
Often, economists throw up their hands in trying to measure the extent to which two
people would gain equally or unequally . The layperson has the same difficulty, but with
a twist. He or she will often assume that the comparison between his or her utility and
that of another person can be determined, and egocentrically concludes that the person
will assign the same or similar utility to an object or event as he or she would . As will
become obvious below, for buyers estimating the utility for sellers and seller estimating
the utility of buyers, that egocentric assumption will often be erroneous and can lead to
suboptimal outcomes .
2 . We replicated this procedure eight times, for a total of nine markets . Across
markets, we varied the commodities that were traded, but all were similarly priced at the
campus store.
ACKNOWLEDGMENTS
David Dunning's contributions were supported by National Institute of Mental
Health Grant RO1 56072 . George Loewenstein's contributions were supported
by the Center for the Study of Human Dimensions of Global Change at
93
Carnegie Mellon University (National Science Foundation Grant SBR9521914) .
REFERENCES
Arriaga, X . B ., & Rusbult, C. B . (1998). Standing in My Partner's Shoes : Partner Perspective
Taking and Reactions to Accommodative Dilemmas . Personality and Social Psychology
Bulletin, 24, 927-948 .
Babcock, L., & Loewenstein, G. (1997) . Explaining Bargaining Impasse : The Role of Self-serving
Biases. Journal of Economic Perspectives, 11, 109-126 .
Baron, J ., & Hershey, J . C . (1988) . Outcome Bias in Decision Evaluation . Journal of Personality
and Social Psychology, 54, 569-579 .
Batson, C . D . (1991). The Altruism Question : Toward a Social-Psychological Answer. Hillsdale,
NJ: Erlbaum.
Becker, G . S . (1991) . A Treatise on the Family (2nd ed.) . Cambridge, Mass . : Harvard University
Press.
Beggan, J. K. (1992) . On the Social Nature of Non-social Perception : The Mere Ownership Effect .
Blau, P. M. (1964) . Exchange and Power in Social Life . New York : Wiley.
Blount, S . (1995) . When Social Outcomes Aren't Fair: The Effect of Causal Attributions on
Preferences. Organizational Behavior and Human Decision Processes, 63, 131-144 .
Bower, G . H . (1981). Mood and Memory. American Psychologist, 36, 129-148 .
Bulmiller, S . (2000, May 20) . Guiliani Quits Senate Race ; G . O . P Congressman Gathers Support
to Face First Lady. New York Times, 149, A1, A14 .
Camerer, C ., Loewenstein, G ., & Weber, M . (1989) . The Curse of Knowledge in Economic
Settings : An Experimental Analysis . Journal of Political Economy, 97, 1232-1254 .
Carnegie, D . (1936) . How to Win Friends and Influence People . New York : Pocket Books .
Christiansen-Szalanski, J . J . J ., & Willham, C . F. (1991) . The Hindsight Bias: A Meta-analysis .
Organizational Behavior and Human Decision Processes, 48, 147-158 .
Coleman, J . S . (1990) . Foundations of Social Theory. Cambridge, Mass . : Harvard University
Press .
Comer, L . B., & Drollinger, T. (1999). Active Empathetic Listening and Selling Success : A
Conceptual Framework. Journal of Personal Selling and Sales Management, 19, 15-29 .
Cook, K. S . (1987) . Social Exchange Theory. Newbury Park, Calif : Sage Publications .
Davis, M. H . (1983) . Measuring Individual Differences in Empathy : Evidence for a Multidimensional Approach. Journal of Personality and Social Psychology, 44, 113-126 .
Dawes, R. M . (1989) . Statistical Criteria for Establishing a Truly False Consensus Effect. Journal
of Experimental Social Psychology, 25, 1-17 .
Dunning, D ., & Cohen, G . L . (1992). Egocentric Definitions of Traits and Abilities in Social
Judgment . Journal of Personality and Social Psychology, 63, 341-355 .
Dunning, D ., & Hayes, A . F. (1996) . Evidence for Egocentric Comparison in Social Judgment .
Dunning, D ., & McElwee, R . O . (1995) . Idiosyncratic Trait Definitions : Implications for SelfDescription and Social Judgment . Journal of Personality and Social Psychology, 68,
936-946.
Dunning, D ., Griffin, D . W., Milojkovic, J . H., & Ross, L . (1990) . The Overconfidence Effect in
Social Prediction. Journal of Personality and Social Psychology, 58, 568-592 .
94
Fischhoff, B . (1975) . Hindsight Is Not Equal to Foresight: The Effect of Outcome Knowledge on
Judgment Under Uncertainty . Journal ofExperimental Psychology : Human Perception and
Performance, 1, 288-299 .
Fischhoff, B . (1977) . Perceived Informativeness of Facts . Journal of Experimental Psychology :
Human Perception and Performance, 3, 349-358.
Franzoi, S . L., Davis, M. H ., & Young, R . D . (1985). The Effects of Private Self-Consciousness
and Perspective Taking on Satisfaction in Close Relationships . Journal of Personality and
Fudenberg, D ., & Tirole, J . (1983) . Sequential Bargaining with Incomplete Information . Review of
Economic Studies, 50, 221-247 .
Galinsky, A . D., & Moskowitz, G . B . (2000). Perspective-taking : Decreasing Stereotype
Expression, Stereotype Accessibility, and In-Group Favoritism . Journal of Personality and
Gilbert, D . T., Pinel, E . C ., Wilson, T. D ., Blumberg, S. J ., & Wheatley, T. P. (1998) . Immune
Neglect : A Source of Durability Bias in Affective Forecasting . Journal of Personality and
Social Psychology, 75, 617-638.
Gilovich, T., Saavitsky, K ., & Medvec, V. H . (1998) . The Illusion of Transparency: Biased
Assessments of Others' Ability to Read One's Emotional States . Journal of Personality and
Social Psychology, 75, 332-346.
Guth, W., Schmittberger, R ., & Schwarze, B . (1982) . An Experimental Analysis of Ultimatum
Bargaining. Journal of Economic Behavior and Organization, 3, 367-388 .
Hawkins, S. A ., & Hastie, R. (1990) . Hindsight: Biased Judgments of Past Events After the
Outcomes are Known . Psychological Bulletin, 107, 311-327 .
Heckathorn, D . D . (1983) . Extensions of Power-Dependence Theory : The Concept of Resistance .
Social Forces, 61, 1206-1231 .
Hertwig, R ., Gigerenzer, G., & Hoffrage, U . (1997). The Reiteration Effect in Hindsight Bias .
Psychological Review, 104, 194-202 .
Hoch, S . J . (1987). Perceived Consensus and Predictive Accuracy : The Pros and Cons of
Projection . Journal of Personality and Social Psychology, 53, 221-234.
Hoch, S. J . (1988) . Who Do We Know : Predicting the Interests and Opinions of the American
Consumer. Journal of Consumer Research, 5, 315-324 .
Holyoak, K. J., & Gordon, P C . (1983) . Social Reference Points . Journal ofPersonality and Social
Humans, G . C . (1958) . Social Behavior as Exchange . American Journal of Sociology, 62,
597-606 .
Johnson, D. W. (1975) . Cooperativeness and Social Perspective Taking. Journal of Personality and
Jones, E . E. (1990). Interpersonal Perception . New York: Macmillan.
Kahneman, D ., Knetsch, J . L., & Thaler, R. H. (1990) . Experimental Tests of the Endowment
Effect and the Coase Theorem . Journal of Political Economy, 98, 1325-1348.
Kahneman, D ., Knetsch, J . L ., & Thaler, R . H . (1991) . The Endowment Effect, Loss Aversion, and
Status Quo Bias . Journal of Economic Perspectives, 5, 193-200 .
Kahneman, D ., & Tversky, A. (1979) . Prospect Theory: An Analysis of Decision Under Risk .
Econometrica, 47, 263-291 .
Kamin, K . A ., & Rachlinski, J . J. (1995) . Ex Post Does Not Equal Ex Ante : Determining Liability
in Hindsight . Law and Human Behavior, 19, 89-104 .
Knetsch, J . L . (1989) . The Endowment Effect and Evidence of Nonreversible Indifference Curves .
American Economic Review, 79, 1277-1284 .
95
Larrick, R . P., & Blount, S . (1997) . The Claiming Effect : Why Players Are More Generous in
Social Dilemmas than in Ultimatum Games . Journal of Personality and Social Psychology,
72,810-825 .
Lawler, E . J ., & Yoon, J . (1996) . Commitment in Exchange Relations: Test of a Theory of
Relational Cohesion . American Sociological Review, 61, 89-108.
Lawler, E . J ., & Yoon, J . (1998) . Network Structure and Emotion in Exchange Relations . American
Sociological Review, 63, 871-894.
Lawler, E. J ., & Thye, S . R . (1999) . Bringing Emotions into Social Exchange Theory . Annual
Review of Sociology, 25, 217-244.
Loewenstein, G . (1999) . A Visceral Account of Addiction. In: J . Elster & O .-J . Skog (Eds), Getting
Hooked: Rationality and Addiction (pp . 235-264) . Cambridge, England: Cambridge
University Press .
Loewenstein, G., & Adler, D . (1995) . A Bias in the Prediction of Tastes . The Economic Journal,
105,929-937 .
Loewenstein, G ., Nagin, D ., & Paternoster, R. (1997) . The Effect of Sexual Arousal on
Expectations of Sexual Forcefulness . Journal of Research in Crime and Delinquency, 34,
443-473 .
Loewenstein, G ., & Schkade, D . (1999) . Wouldn't it be Nice? Predicting Future feelings . In : D .
Kahneman, E . Diener & N. Schwarz (Eds), Well-Being : The Foundations of Hedonic
Psychology (pp . 85-105). New York : Russell Sage Foundation Press .
Loewenstein, G., Thompson, L., & Bazerman, M. H. (1989) . Social Utility and Decision Making
in Interpersonal Contexts . Journal of Personality and Social Psychology, 57, 426-441 .
Long, E . C ., & Andrews, D . W. (1991) . Perspective Taking as a Predictor of Marital Adjustment .
Luce, R ., & Raiffa, H . (1975) . Games and Decisions . New York : Wiley.
March, J. H ., & Simon, H. A . (1958) . Organizations . New York : Wiley.
Markus, H ., Smith, J., & Moreland, R. L . (1985). Role of the Self-Concept in the Perception of
Others . Journal of Personality and Social Psychology, 49, 1494-1512 .
Mead, G. H . (1934) . Mind, Self, and Society . Chicago : University of Chicago Press .
Miller, D . T., & McFarland, C . (1987) . Pluralistic Ignorance : When Similarity is Interpreted as
Dissimilarity . Journal of Personality and Social Psychology, 53, 298-305 .
Neale, M. A., & Bazerman, M. H . (1983) . The Role of Perspective-Taking Ability in Negotiating
Under Different Forms of Arbitration . Industrial and Labor Relations Review, 36,
378-388 .
Nickerson, R . S ., Baddeley, A ., & Freeman, B . (1987) . Are People's Estimates of What Other
People Know Influenced by What They Themselves Know? Acta Psychologica, 64,
245-259 .
Nisbett, R. E., & Ross, L. (1980) . Human Inference: Strategies and Shortcomings of Social
Judgment. Englewood Cliffs, NJ : Prentice-Hall.
O'Gorman, H . J . (1975) . Pluralistic Ignorance and White Estimates of White Support for Racial
Segregation . Public Opinion Quarterly, 39, 313-330 .
O'Gorman, H . J. (1976) . Pluralistic Ignorance: A Replication and Extension . Public Opinion
Quarterly, 40, 449-458.
Prentice, D . A ., & Miller, D . T. (1993). Pluralistic Ignorance and Alcohol Use on Campus : Some
Consequences of Misperceiving the Social Norm. Journal of Personality and Social
Read, D ., & van Leeuwen, B. (1998) . Predicting Hunger : The Effects of Appetite and Delay on
Choice . Organizational Behavior and Human Decision Processes, 76, 189-205 .
96
Richardson, D. R., Hammock, G . S ., Smith, S. M., Gardner, W., & Singo, M . Empathy as a
Cognitive Inhibitor of Interpersonal Aggression . Aggressive Behavior, 20, 275-289.
Ross, L. D ., Greene, D ., & House, P. (1977) . The `False Consensus Effect' : An Egocentric Bias
in Social Perception and Attribution Processes . Journal of Experimental Social Psychology,
13,279-301 .
Ross, L. D ., & Ward, A. (1995) . Psychological Barriers to Dispute Resolution . Advances in

Experimental Social Psychology, 27, 255-304 .
Rubinstein, A. (1985) . A Bargaining Model with Incomplete Information about Time Preferences .
Econometrica, 53, 1151-1172 .
Samuelson, W., & Zackhauser, R . (1988) . Status Quo Bias in Decision Making . Journal of Risk
and Uncertainty, 1, 7-59 .
Sherif, M., & Hovland, C . I. (1961) . Social Judgment: Assimilation and Contrast Effects in
Communication and Attitude Change. New Haven, Conn . : Yale University Press .
Sieff, E . M., Dawes, R . M ., & Loewenstein, G . (1999) . Anticipated Versus Actual Reactions to
HIV Test Results . American Journal of Psychology, 112, 297-311 .
Simpson, J. A ., Ickes, W., & Blackstone, T. (1996) . When the Head Protects the Heart : Empathetic
Accuracy in Dating Relationships . Journal of Personality and Social Psychology, 69,
629-641 .
Smith, A. (1776/1976) . The Wealth of Nations. Chicago : University of Chicago Press .

Smith, C . A ., & Ellsworth, P. C . (1985) . Patterns of Cognitive Appraisal in Emotion . Journal of
Thompson, L., & Loewenstein, G . (1992) . Egocentric Interpretations of Fairness and Interpersonal
Conflict . Organizational Behavior and Human Decision Processes, 51, 176-197 .
Tversky, A., & Kahneman, D. (1974). Judgment Under Uncertainty : Heuristics and Biases .
Science, 185, 117-142 .
Tversky, A., & Shafir, E. (1992) . The Disjunction Effect in Choice Under Uncertainty .
Psychological Science, 3, 305-309 .
Tyson, E., & Brown, R. (1996) . Home Buying for Dummies . Foster City, CA : IDG Books .
Van Boven, L., Dunning, D ., & Loewenstein, G . (2000) . Egocentric Empathy Gaps Between
Owners and Buyers : Misperceptions of the Endowment Effect . Journal of Personality and
Van Boven, L. (2000) . Pluralistic Ignorance and Political Correctness : The Case of Affirmative
Action. Political Psychology, 21, 267-276 .
Van Boven, L ., Loewenstein, G .,& Dunning, D . (2000) . Biased Perceptions of Others' Tastes :
Underestimation of Selling Prices by Buyer's Agents . Unpublished manuscript, University
of British Columbia.
Van Boven, L ., Loewenstein, G., Welch, E ., & Dunning, D . (2000) . Empathy Gaps in Social-Risk
Aversion . Unpublished Manuscript, University of British Columbia .
Viscusi, W. K., Magati, W. A ., & Huber, J. (1987) . An investigation of the Rationality of Consumer
Valuation of Multiple Health Risks. RAND Journal of Economic Science, 18, 465-479 .
Vorauer, J . D ., & Claude, S-D . (1998) . Perceived Versus Actual Transparency of Goals in
Negotiation . Personality and Social Psychology Bulletin, 24, 371-385 .
Vorauer, J . D ., & Ratner, R . K . (1996) . Who's Going to Make the First Move? Pluralistic Ignorance
as an Impediment to Relationship Formation . Journal of Social and Personal Relationships, 13, 483-506 .
Weiner, B . (1985) . An Attributional Theory of Achievement Motivation and Emotion .

97
Weiner, B . (1986) . An Attributional Theory of Motivation and Emotion. New York: SpringerVerlag .
Wood, G . (1978) . The Knew-It-All-Along Effect . Journal of Experimental Psychology : Human
Perception and Performance, 4, 345-353 .
INTO THE LOOKING GLASS :

DISCERNING THE SOCIAL MIND
THROUGH THE MINDBLIND
David Sally
ABSTRACT
Autism results in an asocial mind, and hence, is a disorder worthy of the
attention of social psychologists, sociologists and economists . Work in
developmental psychology and neuroscience reveals that autistic individuals have a faulty theory of mind and cannot take the attitude of the other
The "mindblind" have a disablement of the neurally-based sympathetic
system: they manifest deficits in imitation, in emotional contagion, in
understanding the thoughts, desires and plans of other people, in
pretending and in conversation . The same intertwined set of deficits is
present when, on occasion, non-autists lose their mindseeing abilities . A
theory of mind and sympathy are shown to be critical to a variety offorms
of social interaction.
INTRODUCTION
It is a traditional story line that most of us, but not all, can appreciate : out of
the tragedy of one comes gain for the many . Most of us, but not all, can
sympathize with the plight and emotions of the victim, while yet also seeing the
broader context in which the greater good is served - knowledge gained, lives
protected, honor restored . If there was a silver lining to the iron tamping rod
ISBN: 0-7623-0767-6
99
100
DAVID SALLY
that mistakenly exploded out of the drilled-rock, dynamite-filled cocoon and

clean through the forebrain of Phineas Gage, it was not apparent to him as his
pleasant personality was lost, his ambition waned and his life fell apart
(Damasio 1994) . However, the specificity and breadth of the changes in Mr.
Gage created by the accident, and the particular bundle of deficits suffered by
those who, like him, have had targeted cerebral injuries, or have had brain
lesions or localized strokes, allow modern neuroscientists to map and
understand the components and capacities of the brain.
Rather than neural mappings, many social scientists are concerned with the
social mind (e .g. Dennett, 1987 ; Gilbert, 1998 ; Goffman, 1983 ; Nisbett &
Ross, 1980), perceptions, emotions, cognitions and decisions resulting in
manifest social behaviors . Most disciplines would agree that the social
environment, directly or indirectly, concurrently or prospectively, guides the
development of the social mind within each individual . The social mind may
grow and evolve in response to pressure from culture, market, group, game or
natural selection, but sociologists, economists and evolutionary psychologists
might all agree that there is a sequence of social interactions, past and present,
that undergirds or constructs the social mind . Mead (1934) writes,
I want to be sure that we see that the content put into the mind is only a development and
product of social interaction . It is a development which is of enormous importance . . . . but
originally it is nothing but the taking over of the attitude of the other. To the extent that the
animal can take the attitude of the other and utilize that attitude for the control of his own
conduct, we have what is termed mind; and that is the only apparatus involved in the
appearance of the mind (p . 191) .
The "taking the attitude of the other" may be demonstrated in the adherence to
conventions, the learning of norms, the identification of the other's strategy set,
the recognition of the other's identity or interests, the acknowledgement of the
other's power, the deciphering of the other's meaning, etc . That the social mind
is present is never in doubt, but where it comes from and what it consists of are
relatively unknown .
Virtually unacknowledged by social scientists is the existence of a natural,
Phineas-Gage-like experiment in the social mind . Autism is a condition that
disrupts much of the social mind, largely because autistic individuals seem to
lack a theory of mind and therefore, have difficulty taking the attitude of
another. Finding another's mind to be fundamentally uninterpretable has been
labeled by Baron-Cohen (1995) as mindblindness . The decision making and
behavioral deficits of the mindblind reveal, by contrast, what the capability of
mindseeing does for the majority of individuals who are not autistic .
Into the Looking Glass
1 01
The first section of this paper will attempt to see through the eyes of the
mindblind . We will review the significant advances developmental psychologists have made in the last few decades in characterizing the causes and effects
of autism, including the gross behavioral symptoms in the areas of social
interaction, communication and pretense, the core psychological deficits,
especially theory of mind, and the connections among the specific cognitions
and behaviors of the mindblind . The looking-glass reflection of the picture we
discern through the eyes of the autistic person will be shown, in the second
section, to be consistent with the social relations theory advanced by Sally
(2000a) for those who are mindseeing . This theory of sympathy is a balance
model that builds on the work of Hume, Smith, Cooley, Mead and Heider,
among others, and examines the relationship of affect, evaluation, physical and
psychological distance, mindreading,' and taking the role and interests of
another. The last section will examine the possibility of occasional mindblindness and the specific implications for social psychologists, sociologists and
economists .
The chief claim of this study is that autism proves that a theory of mind and
the sympathetic process are, simultaneously and inextricably, essential to
language, play, interaction, cohesion, imagination and strategy . Hence, the
paper encourages synthesis across the social sciences, for it suggests that the
choice of an optimal strategy is linked to the sharing of a smile or a touch, and
to speaking through hints, metaphors and jokes . The long-term promise of this
line of inquiry is that, as the understanding of the neural basis of autism, theory
of mind and sympathy advances over the next decade, social scientific
researchers examining situations of occasional mindblindness or of exceptional
sympathy and identification should be able to connect a portfolio of social
cognitions and behaviors to specific brain capacities .
CONTINUALLY MINDBLIND
A Looking-Glass World?
Suppose you could not suppose. Suppose every interaction you had left the
other person with the sense of queerness and of something amiss . Suppose you
could not hint or joke or keep a conversation on track . Impairments in
imagination, social functioning, and communication are central to the diagnosis
of autism . Other common symptoms include sensory and perceptual sensitivity,
ritualistic and obsessional compulsions, mental retardation, and self-injury
(Frith, 1989) . Approximately 10% of autists have savant abilities and are
brilliant artists, musicians or number theorists (Sacks, 1995), a critical
1 02
DAVID SALLY
reminder that autistic individuals have talents and capabilities in certain areas
that may meet or exceed those of the mindseeing . In addition, there is a
spectrum of disability as some "high-functioning" autists are able to construct
their own cognitive bridges to span certain social gaps . However, as a leading
researcher says, "Autism . . . does not go away. . . . Nevertheless, autistic
people can, and often do, compensate for their handicap to a remarkable
degree . [But] there remains a persistent deficit . . . something that cannot be
corrected or substituted" (Uta Frith, quoted in Sacks, 1995, p . 252) .
The etiology of this condition was addressed in the only social scientific
(outside of developmental psychology) article on autism . In hindsight,
Bettelheim's (1959) argument is a Freudian nightmare : feral children were not
really raised by wolves, rather they were autistic, a condition brought about by
parental rejection and emotional deprivation. He elaborated on this viewpoint
in The Empty Fortress, evoking images of babies suffocating at the breast,
sadistic maternal teasing, withdrawal and rejection, and prisoners in German
concentration camps (Bettelheim, 1967) . 2
This harsh theory is now discredited : the evidence is overwhelming that
autism is an organic brain disorder with multiple causes - "genetic
predisposition, pre- or post-natal viral infection . . . , chromosomal damage,
biological agents still unknown" (Park, 1998, pp . 30-31) .
Although it may be difficult to read the minds of the mindblind, we can
certainly apply our sympathetic capabilities to understand the feelings and
thoughts of those who are close to autistic individuals . One neurologist
describes an autistic child's impairment in interaction :
Meeting him at intervals of several months, one is welcomed and bid goodbye with the
same impersonal kindness as if contact were only real as long as it lasted during concrete
presence . . . it is a presence without emotional content (Kurt Goldstein, quoted in Sacks,
1995, p . 216) .
As Frith said above, some autistic people can compensate, but not entirely,
often there is still a measure of oddness :
What is it like to have to learn the myriad rules of human interaction one by one, by rote?
By rote, because the criterion of `how would I feel if' is unavailable, since so much of what
pleases (or distresses) her does not please others, and so little of what pleases (or distresses)
others pleases her. . . . What's it like? It can be funny ; that's the best of it . Five years ago
Jessy gave me an unexpected Mother's Day gift - a can of cat litter deodorizer, beautifully
wrapped, topped with a tiny restaurant packet of strawberry jam . The attribution of feelings
and desires to others still hag a way to go (Park, 1998, p . 37) .
Jessy, who must try so hard to control her crying, often asks, Ìs it a good reason to cry?'
(Park, 1998, p . 41) .
The speech of another very high-functioning autist, Temple Grandin, is overly

scripted and monological :
103
She spoke well and clearly, but with a certain unstoppable impetus and fixity . A sentence,
a paragraph, once started, had to be completed ; nothing was left implicit, hanging in the air
(Sacks, 1995, p. 257) .
Images of transparency recur :

People around me were transparent . . . . Even a sudden loud noise didn't startle me from
my world (Temple Grandin, quoted in Sacks 1995, p . 255) .
The wisps were tiny creatures, almost transparent . They hung in the air directly above me,
and looked something like wisps of hair . . . . My bed was also surrounded and totally
encased by tiny spots that I called stars, so that it seemed to me I lay in some kind of
mystical glass coffin. . . . By looking through the stars and not at them, I could see them
. . . (Williams, 1992, p. 10) .
Autism is when your two-year-old looks straight through you to the wall behind - through
you, her father, her sister, her brother, or anybody else . You are a pane of glass . Or you are
her own personal extension, your hand a tool she uses to get the cookie she will not reach
for herself (Park, 1998, p . 30).
The metaphor here is an exact reversal of Cooley's (1902) looking glass self "Each to each a looking glass/Reflects the other that doth pass" (p . 184) . There
is limited reflection in autism .
One is also reminded of Alice's speculation about life in the Looking-glass
House : from the outside, you cannot discern directly if the fire is lit, the words
in the books go the wrong way, "you can just see a little peep of the passage
. . . . if you leave the door of our drawing-room wide open : and it's very like
our passage as far as you can see, only you know it may be quite different on
beyond" ; and, from the other side if you were to pass through as Alice does,
none of the animated objects seem to be able to see you or hear you, poetry is
Jabberwocky, and if you don't write your feelings down, as the White King
attempts to do, you will never remember them (Carroll, 1871) .
Like the White King, Temple Grandin possesses a mental archive of social
situations and emotions :
[Her experiences] were like a library of videotapes, which she could play in her mind and
inspect at any time -'videos' of how people behaved in different circumstances . She would
play these over and over again and learn, by degrees, to correlate what she saw, so that she
could then predict how people in similar circumstances might act (Sacks, 1995, p . 260) .
The Theory of Mind

Cognitive and developmental psychologists have identified three core cognitive
deficits in autism - theory of mind, central coherence, and executive function .
Theory of mind (ToM) is equivalent to the common-sense psychology studied
by Heider (1958) : 3 it is the framework of tested hypotheses and accumulated
1 04
DAVID SALLY
evidence through which we deduce and anticipate the desires, beliefs, feelings
and thoughts of other people in order to predict their behavior. As Bem (1967)
shows, we may use the same theory to make sense of ourselves and guide our
own behavior. Central coherence represents the ability to integrate parts into a
whole, to unite content and context to find higher levels of meaning (Frith,
1989 ; Happe, 2000) . Finally, executive function stands beside other brain
functions and allows the individual to disengage from present stimuli, inhibit
responses, and plan sequences of actions (Hughes, 1996) .
There is a dispute within the literature about the relative importance of these
three deficits, but almost every researcher agrees that an autistic person has
some impairment of his ToM . 4 Two frequently replicated tests establish the lack
of a ToM in the mindblind ; both have to do with the deduction of beliefs (see
Yirmiya, et al . (1998) for a meta-analysis of these experiments) . In the Sally/
Anne task (Baron-Cohen, 1995), two individuals are dramatized through
puppetry as co-present in a certain setting, making the setting common
knowledge (to the characters, the tested child, and the experimenter), in
particular, that Sally has just placed her ball in a covered basket . Sally skips out
of the room, and Anne moves the ball to a nearby box and closes the top . Now,
the audience is asked, where will Sally look for her ball when she returns?
Normal four-year-old children, and those with Down's Syndrome and with a
mental age of four years, answer, "The basket!" Autistic children of the
matched mental age answer, "The box ." This task is obviously trivial for the
mindseeing who can easily keep track of where the ball really is versus where
Sally believes it to be .
One objection might be that answering "the box" just reveals a confused
mind, not a mindblind one . A second analogous experiment refutes this
interpretation (Leekam & Perrier, 1991 ; Leslie & Thaiss, 1992) . This time, a
child is shown a scene and given a Polaroid camera with which to photograph
the venue. He looks through the lens and snaps . A prominent item in the scene,
such as a ball, is moved to a new location, making the photograph "false," and
then the child is asked, "Where will the ball be in the photo?" Autistic children
have no difficulty reading the mind of the camera, for they indicate the original
position . Their accuracy here suggests that autists suffer from a blindness
specific to the minds of people.
That the weakness in ToM causes inaccuracies in self-insight is demonstrated by another "false belief' test (Perner et al ., 1989) . A brightly labeled
candy canister is shown to a child, who, after being queried about its contents,
replies "candy." The top is taken off, only to reveal an inedible, non-sugary set
of pencils . The top is replaced and two questions are posed - "What did you
1 05
think was in here?" and "What will the next kid think is in here?" In reply to
both, normal children say, "candy" and those with autism, "pencils ."
Of course, as they get older, many autistic children can successfully pass
these simple ToM tests . The developmental lag continues throughout
childhood, however, and most autists remain blind to such second-order
thoughts as "what does he think she wants?" or "why doesn't he believe me?"
(Baron-Cohen, 2000) . High-functioning autistic people such as Temple
Grandin may continue to develop a logical ToM, but they can never close the
gap on their intuitive mindseeing peers .
Neuroscientists are in the early stages of locating the ToM within the brain .
For example, Sabbagh and Taylor (2000) document increased electrical brain
activity in the left frontal area for normal subjects doing a false belief task
versus those doing a false photograph task . Modem Phineas Gage counterparts,
suffering from lesions due to stroke in the right hemisphere, had much more
difficulty with advanced ToM tests than did non brain-damaged controls
(Brownell et al ., 2000) . Extrapolating from a variety of early studies, Frith and
Frith (1999, 2000) suggest that the ToM is found in a distributed neural system
incorporating the medial prefrontal cortex, which includes areas activated in
monitoring the self's inner states, and the superior temporal sulcus, which is
associated with the detection of the movement of animate objects, especially
eyes, hands, and mouth .
Hence, autistic individuals have an impaired ToM and are mindblind . As a
result, they are non-social, exceptional members of society :
[T]he line of our visual regard, the intensity of our involvement, and the shape of our initial
actions, allow others to glean our immediate intent and purpose . . . . At the very center of
interaction life is the cognitive relation we have with those present before us, without which
relationship our activity, behavioral and verbal, could not be meaningfully organized
(Goffman, 1983, pp . 3-4) .
Society is an interweaving and interworking of mental selves . I imagine your mind, and
especially what your mind thinks about my mind, and what your mind thinks about what
my mind thinks about your mind . I dress my mind before yours and expect that you will
dress yours before mine . Whoever cannot or will not perform these feats is not properly in
the game (Cooley, 1927, pp . 200-201).
Their absence from the game is manifest in violations of protocol in social

interaction - not saying hello, standing too close or askew, talking too loud, etc .
(Hobson & Lee, 1998) . It is also apparent in relationship difficulties, for the
mindblind do not generally have friends or spouses : even highly functional
autistic children are unable to describe what a friend is or to distinguish a friend
from an acquaintance (Lord & Magill, 1989) . What other problems are
attendant upon their absence from the social game? The next two sections will
examine the precursors and consequences of mindblindness .
1 06
DAVID SALLY
Precursors of Mindblindness
In normal children, the ToM does not emerge until four years of age : most
three-year-olds will fail the simple false belief tests . On what basis, then, is a
toddler interacting with her parents? Does she lack a social mind? The answer
is clear given that a normal toddler will exhaust you after a single hour with so
much interaction, communication, and imagination . Gopnik and Meltzoff
(1997) suggest most persuasively that ToM is a natural outgrowth of the
infant's innate talent for imitation :
Babies automatically seem to assume that their own internal feelings and the actions of
others can be represented in the same way . Imitation is nature's way of solving the other
minds problem and the mind/body problem in one fell swoop (Gopnik, Capps & Meltzoff,
2000, p. 54) .
The evidence is startling : newborns, having emerged less than one hour earlier,
"differentially imitated tongue protrusion, mouth opening, and lip protrusion"
(Gopnik & Meltzoff, 1997, p . 130) .
This same correspondent, mimetic approach continues throughout early
childhood:
In each case, desires [age 2], perceptions [3], beliefs [4], and interpretations [6 or 7], babies
seem to begin by assuming that they and other people have the same internal states and
gradually develop an understanding of differences. . . . They assume that they are like
others and that their internal states map on to the behaviour of others (Gopnik, Capps &
Meltzoff, 2000, p . 58) .
To carry out this process, the baby must be able to : 1 . Differentiate human
beings from other objects ; 2 . Be "involuntarily" drawn to attend to humans ; 3 .
Share joint attention with another person ; 4 . Mime an other's actions, either
automatically or intentionally ; 5 . Translate feeling to action and action to
feeling, either consciously or non-consciously . Somewhere between mirroring
a tongue-protruding razz and predicting the composition of another's thoughts,
an autistic child will have problems in at least one of these five areas .
A face may be not clearly differentiated from an inanimate object . Using
imaging techniques, Schultz et al . (2000) find that the brain activity caused by
the mindblind's perceptual processing of faces is equivalent to that created by
the mindseeing's perceptual processing of objects . In a similar result, when
autistic subjects process upside-down faces, they have less difficulty than nonautistic subjects (Hobson, Ouston & Lee, 1988 ; Tantam et al ., 1989) . Experts
perceive their regularly scrutinized objects holistically (for example, birders
and a swallowtail), while other objects are seen through an analysis of distinct
segments (Schultz et al ., 2000) . That the autist appraises faces objectively may
bespeak a lack of accumulated expertise in visages . He may have more trouble
1 07
recognizing a familiar face (Boucher, Lewis & Collis, 1998) . A mindreader

becomes an authority on the countenance, but the autist remains ever the
amateur.
Autists are not involuntarily drawn to human "stimuli ." While playing with
a toy at a table, an autistic child is much less likely to turn his head and look
at you clapping hands or calling his name, than he is when you crank the handle
of a jack-in-the box or shake a rattle (Dawson et al ., 1998) . This indifference
to the social also ruins joint attention : this same young autist is unlikely to
follow your eyes if you break your mutual gaze and turn your attention to a
picture on the wall (Dawson et al ., 1998 ; Roeyers, van Oost & Bothuyne,
1998) . Sigman and Kasari (1995) review other naturalistic studies in which
autistic children would bring a troublesome toy to their parents for assistance,
but rarely brought a plaything to them so they could admire it together. Look,
over there - the mindblind point not to establish mutual observation and
consideration, but rather to obtain the object (turning the interactor's hand into
a tool, as Park alluded to above) (Baron-Cohen, 1995, 2000) .
Moreover, autism seems to directly impair a person's mimetic abilities .
Many of the classic games of childhood - peek-a-boo, sing-along, follow the
leader, Simon says - require and develop coordinated imitation . Normal
children take to these games naturally and easily (note that Simon says is
difficult because of the contrast between the correct move and the strong
impulse to imitate the leader - hence, a quarter of the normal fourth to eighth
graders lost the game in Hartshorne and May's (1929) test) . Autistic children
are less likely to mimic others spontaneously (Whiten & Brown, 1998), and,
even when directly asked to do so, may not be able to imitate certain actions .
The young mindblind child has a much easier time replicating an adult who
pushes a toy truck than one who bends an index finger up and down (Stone,
Ousley & Littleford, 1997) . The older autistic child may be able to reproduce
the basics of a demonstrated action, e .g . wiping one's own brow, but not the
style, e.g . abruptly and roughly or smoothly and lightly (Hobson & Lee, 1999) .
Hobson and Lee (1999) conclude that their subjects could not "identify with the
experimenter as acting in relation to himself' (p . 657) . Because of the ToM and
imitative deficits in the autistic individual, the other minds and mind/body
problems remain unresolved .
This all means that the interaction order, as Goffman (1983) termed the
sphere of face-to-face interaction, is drastically altered, early on, for the
mindblind . Propinquity does not have the psychological effects that it has in
normal children . While the autistic child prefers to be more proximal to a
parent than to a stranger, he is less likely than a normal child or a child with
108
DAVID SALLY
Down's syndrome to utilize the co-presence to gaze at the parent, smile at her,
or give her things (Dissanayake & Crossley, 1996) .
Other Consequences of Mindblindness
This alteration of the interaction order is widespread and long-lived . A list of
commonplace behaviors that are absent in autism would include the following :
interpreting facial expressions ; choosing appropriate gifts (cat litter deodorizer
and strawberry jam) ; responding to hints in conversation ; improvising chitchat ; supplying missing information to the audience ; engaging in spontaneous
pretend play ; keeping secrets ; enjoying fiction; reacting actively to another's
distress or joy ; forming reciprocal friendships (Frith, Happe & Siddons, 1994 ;
Frith & Frith, 2000) .
It is worthwhile to explore several of these deficits in depth . Reading
another's eyes and face can be troublesome for the mindblind . Young autists
can look at a cartoon face glancing at one of four candies and correctly answer,
"Which one is he looking at?" while being puzzled by "Which one does he
want?" (Baron-Cohen, 1995) . Autistic adults, seeing a photo of a person's eyes,
can identify the gender of the person as accurately as a normal subject can, but
fail to do much better than chance when selecting which of two mental states
the eyes reflect (Baron-Cohen et al ., 1997) .
One of the abnormalities diagnostic of autism is lack of spontaneous pretend
play. For a blanket-covered stack of sofa cushions to become a happy,
productive domicile for a "mother" and a "father" requires that each child be
capable of representing the real elements as something else, e .g . the gap in the
pillows is a door, and the shoebox is a stove, and that each find "playing house"
rewarding . The mindblind child is capable of make believe: if requested
directly to answer the banana, he is likely to lift it to his ear . But, as Carruthers
(1996) argues, the internal motivation for such play is extinguished by the lack
of an internal ToM :
You cannot enjoy supposing or imagining without being conscious of your (mental)
activity. In general, enjoying Xing presupposes awareness of Xing - which is why you
cannot enjoy digestion, sleepwalking, or subliminal perception (p. 265, emphasis in
original) .
Temple Grandin, despite her high level of intelligence, does not read novels or
plays both because it takes a huge effort, due to her faulty ToM, to figure out
what Romeo and Juliet are up to, and because the activity is not rewarding or
interesting enough (Sacks, 1995) .
The trouble Ms . Grandin has in empathizing with fictional characters is
reflective of a more general problem with helping in autism. The link to ToM
109
is clear : if someone cries out sharply and suddenly, squeezes her eyes shut, and
shakes her hand vigorously back and forth, but none of these actions is gripping
or explicable, you would neither look nor reach out willingly to help . In the
psychology laboratory, this drama commences when the experimenter appears
to slam his thumb with a toy hammer and hops around in pain . Far fewer
autistic than normal or developmentally delayed children stare at the face of the
injured man or even look up from what they are doing (Sigman & Kasari, 1995 ;
Charman et al ., 1997) .
Lastly, mindblindness affects the language faculty deeply . The autist's
problems with joint attention and pointing suggest that Augustine's method of
learning language is not applicable :
When they (my elders) named some object, and accordingly moved towards something, I
saw this and I grasped that the thing was called by the sound they uttered when they meant
to point it out . Their intention was shewn by their bodily movements . . . : the expression
of the face, the play of the eyes (cited in Wittgenstein, 1958, p . 2) .
Indeed, the language facility of autistic children followed over several years by
Sigman and Ruskin (1999, cited in Tager-Flusberg, 2000) was correlated with
their readiness to respond to offers for joint attention as toddlers .
However, as Wittgenstein (1958) demonstrates, learning and using language
is more than just heaping one name on top of another ; in his famous phrase,
"Uttering a word is like striking a note on the keyboard of the imagination" (p .
4) . Without a sense of how the strings hit by the key's hammer will resonate,
due to nescience of their tautness, thickness, length and composition, and to the
absence of an internally heard tone, the mindblind speaker is unable to play the
normal chords of language . Pragmatic uses of language create dissonance for
the autistic individual : metaphor, poetry, irony, sarcasm, joking, all ring false to
his ears and leave him unmoved (Happe, 1993 ; Tager-Flusberg, 2000) .
Moreover, since "it is natural that we speak by hints" (Vygotsky, 1986, p . 238),
the mindblind listener will overlook any subtlety or implication, and the
speaker's intended meaning will be ignored in favor of a literal interpretation
of his words (Mitchell, Saltmarsh & Russell, 1997) .
Poignantly and symbolically, many autistic speakers experience confusion
with their pronouns (Lee, Hobson & Chiat, 1994) . Even the simple pronoun
requires an ability to change perspective : when I say "I," you must translate the
word into "you," and when you say "you," you must know that I will think "L"
As this pronominal inversion and conversion reveal, without a theory of mind,
without the sympathetic ability to take the other's perspective, life in the
looking-glass house becomes impossibly complicated - the animate and
inanimate are confounded, words are backwards and jabberwocky, joint
attention is fleeting, and feelings need to be recorded and catalogued .
110
DAVID SALLY
CONTINUALLY MINDSEEING
Sympathy and Theory of Mind
How queer the mindseeing are! Show them a film depicting simple geometric
shapes moving independently along various non-linear vectors, and when two
figures randomly happen to move in parallel (as any two particles in Brownian
motion will sometimes do), these people start seeing relationships, emotions,
and stories (Heider & Simmel, 1944 ; Heider, 1958)! By contrast, show the
mindblind the same film and they see exactly what's there - "The small circle
went inside the rectangle. The big triangle was in the box with the circle"
(Subject quoted in Klin, Schultz & Cohen, 2000, pp . 363-364) . 5 Mindreaders
have a hair-trigger on their ToM's : they can find intention, emotion, and belief
in the casual, chaotic heap of animal entrails (Zeitlyn, 1995), the stochastic
stress-induced sliding of the plates in the Earth's crust, a blinking shadow cast
diagonally through the trees by a passing cloud on a moonlit night, or an ant
dragging an oversized crumb up and over, up and over, up and over a series of
little twigs and stones .
The looking glass reflection (and Phineas-Gage-like inference) of our
conclusions about mindblindness is that in the mindseeing, all the following
elements are interwoven : synchronicity in interaction, joint attention, mutual
gaze, imitation, the sharing of thoughts, the sharing of emotions, empathy and
helping, figurative speech, and all non-literal uses of language . Throughout the
history of the social sciences, various authors have examined (under different
names) the individual's theory of mind and taking the attitude of an other : the
Scottish moral philosophers, Hume and Smith ; the (first) evolutionary
psychologist, Darwin (1936) ; the pragmatists, Cooley and Mead ; the folk
psychologist, Heider. Also, the last twenty years have witnessed the creation
and vibrant growth of the field of social cognition and the incorporation of
emotions into previously "cold" approaches (Lawler & Thye, 1999 ; Schwarz,
1998 ; Zajonc, 1998) . One might ask, is there a modem theory of the social
mind that combines all or many of these elements? It will not surprise the
mindperceiving reader to find that the present author has developed such a
framework .
To express continuity with earlier writers, especially those named just above,
I have called our innate theory of mind and mindreading ability, sympathy .
Cooley (1902) writes,
[S]ympathy [is] the sense of primary communication or an entering into and sharing the
mind of someone else . When I converse with a man, through words, looks, or other
symbols, I have more or less intelligence or communion with him, we get on common
11 1
ground and have similar ideas and sentiments . If one uses sympathy in this connection and it is perhaps the most available word - one has to bear in mind that it denotes the
sharing of any mental state that can be communicated, and has not the special implication
of pity. . . . (p . 102) .
"Sympathy" is a risky term to use, because it has been attached to different

phenomena with some attendant confusion : 'sympathy is pity and helping,
cognitive sharing, emotional contagion, imitation, projection of the self .
However, the point is this : sympathy is all of these things, because
mindreading, though distributed broadly, is all of a piece . Hence, the variety
and complexity of the symptoms of autism, and hence, the need for modem
researchers to consider the synthesis of the results in their domain of expertise
with those in other areas of social life .
In earlier work (Sally, 2000a), I have stated that sympathy is the
interpersonal process of identification that incorporates the motivation and
ability to read another's mind, and the belief that we can do so . 6 Sympathy is
developmental in nature, advancing from motor mimicry to emotional
contagion to the correspondence of mental states (Ribot, 1897 ; Gopnik &
Meltzoff, 1997) . The process may occur non-consciously and involuntarily : the
sensitive trigger to ToM is often pulled by some unexpected element of the
situation - the teetering of a bowling ball on the lip of the gutter (Stewart,
1854), the awkward way a surprised enemy soldier holds up his unbuttoned
pants while fleeing (Orwell, 1952) . Sympathy results in an enlarged selfinterest, creating one basis for altruistic actions (Smith, 1790) .' Sympathy is the
foundation of social capital, making personal relationships, strong ties and
weak ties, and "social power" possible :
It is well understood by men of the world that effectiveness depends at least as much upon
address, savoir-faire, tact, and the like, involving sympathetic insight into the minds of
other people, as upon any more particular faculties . There is nothing more practical than
social imagination ; to lack it is to lack everything (Cooley, 1902, p . 107) .
Sympathy is one basis of group solidarity . If positive emotions are

unrecognized and not shared because one member is mindblind, then frequent
successful exchange will not lead to relational cohesion (Lawler & Yoon, 1996 ;
Lawler & Thye, 1999) . Also, Collins (1989) suggests that feelings of group
membership are produced by co-participation in interaction rituals, these
feelings being by-products of a mutual focus of attention and a second-order,
shared awareness of this first-order joint attention . Clearly, given the difficulty
an autistic person has with first- and second-order beliefs and joint attention, a
ToM is necessary to create a group identity through interaction .
On the cognitive side of group behavior, a ToM is also indispensable for a
transactive memory . Wegner (1995) writes,
1 12
DAVID SALLY
A complete transactive memory in a group occurs when each member keeps current on
who knows what, passes information on a topic to the group's expert on the topic, and
develops a relative sense of who is expert on what among all group members (p . 326,
quoted in Nickerson, 1999) .
Without the ability to understand, at a general level, what other people know,
such a distributed system would unravel .
Finally, identifying with another, as mentioned earlier, is critical to language
and conversation . Every utterance has an infinity of meanings, and so,
communication is of necessity a complex coordination game (Sally, 2000b).
Meaning is a puzzle . As the puzzle-maker, the speaker must create a sentence
that he knows the listener will find soluble, and to do so, he must bring to mind
key characteristics of the audience . The listener, in turn, must activate her
model of the speaker to help solve the puzzle of what he means . To speak, I
must take the role of the audience ; to listen, I must take the role of the utterer ;
to communicate, our projections must converge (Mead, 1934) . When the
mindseeing speak to another, we seize control of that person's mind for a set
of instants . Ironically, though, in preparing to speak, we have already let the
other into our own mind . The harder puzzles of meaning created by metaphors,
irony, jokes, allusions, hints, indirectness, and implication can only be solved
through a strong sense of sympathy (Sally, 2000c) .
A Balanced Model of Sympathy
We know from our review of the development literature that the child's ToM
varies according to the circumstances of the social situation . A baby will
imitate its mother more readily ; a toddler will stay closer to her than to a
stranger, and send her more gazes and smiles . Older children (and adults) can
resolve ambiguous verbal statements and social situations more readily, since
such events are more familiar . The factors affecting the extent of sympathy
from one person to another are captured in the balance model shown in Fig . 1 .8
Physical and psychological distance are represented in the boxes labelled
physical proximity and similarity and familiarity, respectively . Affection refers
to "liking vs . disliking," and so is a summary measure of the emotional attitude
toward the other . Evaluation is the valence, the overall goodness, of a stimulus .
I will have more to say about each of these factors below, but the reader should
refer to Sally (2000a) for a thorough discussion .
As with Heider's (1958) balance theory, two types of equilibria are assumed
to exist among the four underlying factors :
113
Fig. 1 . A Balanced Model of Sympathy .
(closeness, (similarity, familiarity), liking, goodness)

and
(distance, (difference, strangeness), disliking, badness) .
For example, we seek to avoid those things that are different or strange, or that
we dislike or believe are bad; we come to like those people with whom we
interact frequently, who have similar tastes and beliefs, and who we believe are
admirable . At the first equilibrium, the other's thoughts are read ; her actions,
believed; her decisions, foretold ; her metaphors, understood, her hints,
followed, and her jokes, laughed at ; and, finally, her interests, strongly
acknowledged . At the latter equilibrium, there is relative mindblindness .
1 14
DAVID SALLY
Closeness
There were two uses of glass imagery in the reports about autism - glass as an
invisible, yet real barrier, to contact and friendship, and other people as
transparent and non-reflecting . For the mindseeing, co-presence, similarity and
familiarity can lead to a strong sense of closeness, allowing us to array various
other individuals at different distances from our self (Smith, 1790 ; Lewin,
1948 ; Little, 1965 ; Aron, Aron & Smollan, 1992 ; Davis et al ., 1996) .
Nearness translates into sympathy, both an enlarged self and ToM efficiency .
Closeness can create a kind of cognitive confusion anecdotally present in the
intermingling of spouses' stories and experiences - "you went to that party? . . .
I thought it was me ." In a laboratory, it takes subjects a longer time to recognize
traits that are unique to themselves as opposed to traits shared with their
spouses (Aron et al ., 1991) . Stinson and Ickes (1992) explicitly tested the link
between closeness and mindreading . Pairs of friends were secretly videotaped
while they waited for a promised experiment to begin . Each friend then
watched the tape and recorded any thoughts or feelings he or she had at various
moments during the interaction . During a second and final viewing of the tape,
each friend was asked to record what the other was feeling or thinking . In the
control condition pairs of strangers followed the same procedure . The results
were clear: pairs of friends were much more accurate in their mindseeing than
strangers were .
Those researchers who have studied the non-verbal behaviors and positionings of humans have demonstrated that physical proximity is not just the
Euclidean distance from one body's center of gravity to another's, but is also
a function of angle, tilt, gesture, and especially, eye contact (Argyle & Dean,
1965) . As the unseeing gaze of the autist shows, a ToM is necessary for the
mutual glance to have the power that Simmel (1908) suggests : "The eye of a
person discloses his own soul when he seeks to uncover that of another"
(p . 308) .
The autistic person, to overstate the case a little, is filled with the cold
realization that every other person is fundamentally distinct from himself . Just
as a moving triangle is a moving triangle, a different person is a different
person . For the non-autist, however, the initial hypothesis is that an unknown
other is akin, alike to me . The automatic imitation of neonates, Gopnik and
Meltzoff (1997) propose, is based on the inborn idea that others are "just like
me ." The acquaintanceship process is usually smooth among the mindseeing,
because in the absence of any strong negative information, we are programmed
to look for similarities with others (Hoch, 1987) ; "our first movement is to
believe in an undivided being between us" (Merleau-Ponty, 1969, p . 52) . If I
115
assume that we are alike, I am more willing to expend the sympathetic effort
and more confident that I can figure you out .
Affection
The correlations between liking someone and being close, both physically and
psychologically, has been well documented . Common hallways (Newcomb,
1956 ; Priest & Sawyer, 1967 ; Nahemow & Lawton, 1975) and shared traits
(Lazarsfeld & Merton, 1954 ; Tuma & Hallinan, 1979 ; Feld, 1982), experiences
(Latane, Eckman & Joy, 1966), and attitudes (Byrne & Nelson, 1965 ; Lydon et
al ., 1988) lead to friendship, and vice versa (Darley & Berscheid, 1967) .
As Hume (1740) realized long ago, familiarity leads to liking and
preference :
The mind finds a satisfaction and ease in the view of objects, to which it is accustom'd, and
naturally prefers them to others, which, tho', perhaps, in themselves more valuable, are less
known to it (p . 355) .
The connection of repeated proximity, familiarity, and affection can occur

automatically and involuntarily ; Zajonc (1968 ; 1980) has labeled this nonconscious connection the "mere exposure" effect . That the mind prefers those
things which multiple exposures make it "accustom'd" to, has been
demonstrated for objects as varied as random polygons, Chinese ideographs,
numbers and letters, flavors and colors, etc . (Zajonc, 1998 ; for a thorough
review, see Bornstein, 1989) . The mindblind may not become accustomed to
those they interact with, but Brockner and Swap (1976) demonstrate that if we
have seen the other person repeatedly even without any mutual acknowledgement, the laboratory equivalent of standing apart on the same train platform day
after day, we would prefer to interact with that person rather than some
"stranger."
The non-autist, even when his infancy is far behind, spontaneously imitates
facial expressions, postures, bodily mannerisms, and speech patterns and
elements (Bavelas et al ., 1986 ; Hatfield et al ., 1994) . Imitation leads to the
perception of similarity and a sense of affection . For therapists and patients
(Trout & Rosenfeld, 1980) and fellow classmates (La France, 1979), postural
mirroring was positively correlated with relationship strength ; Cappella (1997)
found that synchronous smiling was correlated with attraction between brand
new acquaintances .
The most direct evidence comes from Chartrand and Bargh (1999) . Through
a careful protocol and the use of twitchy accomplices, these researchers show,
initially, that subjects would imitate at a significant rate the confederates'
smiling or face rubbing or foot shaking during an interaction . The first
1 16
DAVID SALLY
experiment having demonstrated the innateness of mimicry, the second was

designed to determine its effects on affection . In this protocol, half of the
confederates mirrored the behavioral mannerisms of the unwitting subjects
during a discussion, and half refrained from imitation. Those participants who
had been mirrored expressed much more liking for their partners than did those
participants who interacted with unsympathetic partners . When they were
debriefed after the experiment, subjects had no idea that imitation had occurred
or that it had altered their feelings . So, such mirroring plays an essential role
in creating rapport, and its absence is another form of transparency in the
mindblind .
Evaluation
Because of the pattern of brain activity they observed in the normal film
viewers who saw a human drama enacted by the moving geometric shapes,
Klin, Schultz and Cohen (2000) conjecture that, neurologically, social
perception
is by necessity affectively mediated, given that the appreciation of interaction between
people may involve an evaluative component (good/bad, helpful/hostile) that is intrinsically
associated with a predisposition to respond in certain ways (to approach or to distance
oneself (p . 378) .
Evaluation (good and bad) is a basic component in the perception and

categorization of word meanings, facial expressions, photographic portraits,
emotions, and smells and tastes (Osgood, Suci & Tannenbaum, 1967 ; Zajonc,
1980, 1998 ; Berntson, Boysen & Cacioppo, 1993 ; Lang, Bradley & Cuthbert,
1990) . Based on a robust set of experimental data, Bargh (1997) states boldly,
All stimuli are evaluated immediately as good or bad, without the participant intending to
evaluate, having recently thought in terms of evaluation, or being aware of doing so .
Therefore, everything one encounters is preconsciously screened and classified as either
good or bad, within a fraction of a second after encountering it (pp . 22-23) .
One gets a sense from their memoirs that the mindblind must preconsciously
screen and classify another person, except the most familiar, as neutral on the
"good" scale and negative on the "bad" scale .' Because evaluation automatically invokes behavioral tendencies - good - approach, bad
withdraw
(Lewin, 1935 ; Berntson et al ., 1993) - the autist's negative evaluation of others
would partially account for his distancing and lack of eye contact .
The non-autist tends to bring evaluation, affection, familiarity, and proximity
to a positive equilibrium with respect to his acquaintances and friends . "An act
is good when it is performed by a friend" (Heider 1958, p . 54) . Having been
asked about our spouse, we readily admit that they are accommodating and
1 17
non-confrontational, qualities that seem to be "good ." After we read a (fake)

article that proclaims that engaging in disagreement is essential to a healthy
marriage, when asked again, we say that our spouse "disagrees with me
constantly" (Murray & Holmes, 1993) .
In the end, our ability both to actually read the mind of another and our
motivation to do so are dependent on a balance of distance, liking, and
goodness . We, the mindseeing, have fellow feeling for the other person if he is
present, if he copies our gestures, if his approach did not make us pull away,
if he shares our characteristics, if he is attractive, if we have seen him around,
if we like him, if we have heard good things about him, if we believe him to
be admirable . In these instances, we can shake his hand, look him in the eye,
help him out, give him a gift, joke with him, ask him a favor, or include him
in our club .
OCCASIONALLY MINDBLIND
While the previous sections showed that passing into the glass of autism is
valuable when thinking about the broad social relations theory of sympathy, it
is also true that other theories may benefit from occasional exposure to the
limited reflection of autism." Affect control theory, for example, suggests that
emotions arise from situational cognitions :
A person's definition of the situation provides a self identity and an identity for other, and
those identifications set sentiments that should be reflected in experiences . The person
builds events to push experiences in a direction that affirms the sentiments associated with
situational identities. If something happens to deflect experiences away from identity
confirmation, then a person builds corrective events to put social interaction back on track
(Heise, 1999, p . 7) .
The step-by-step processing of this passage seems to apply equally well to a

high-functioning autist, such as Temple Grandin with her videotape library of
social identities and vignettes, and yet, she struggles with social interaction and
emotional information . Do the mindblind recognize only an impoverished set
of identities? Do they fail to categorize situations based on the theoretical triad
of status, power and agency? Do they not seek to make events consistent with
situational identities? What role should ToM and sympathy play in affect
control theory?
We have seen that autism involves a severe disablement of the neurally based
sympathetic system: the mindblind often show deficits in imitation, in
emotional contagion, in understanding the thoughts, desires and plans of other
people, in pretending and in conversation . Accordingly, in interactive game
settings, autistic individuals should have less difficulty playing when their
1 18
DAVID SALLY
partner is not a person, should be prone to miscoordination and random

choices, should have problems cooperating with another player, and should be
less affected by the co-presence of another player . Colleagues from cognitive
neurosciences and I are currently conducting experiments contrasting the
decisions of the mindblind and mindseeing in a variety of oft-studied social
science games - prisoners' dilemma, pure coordination, dictator and ultimatum. We hope that these studies illuminate both the role of a ToM in game
playing, and the complexity of strategic situations for those with autism.
These last two examples should encourage all social scientists researching
social cognition, emotions, relationships, etc . to occasionally ponder the cloudy
looking glass of the autist . Moreover, the negative equilibrium of the balanced
model of sympathy - [distance, (difference, strangeness), disliking, badness] demonstrates that for the non-autist, mindblindness can nevertheless be an
occasional, situation-determined condition . In fact, our case in hand proves it :
the mindseeing are blind to the mind of the autistic person. If the autist is "an
anthropologist on Mars" (Sacks, 1995) among the mindseeing, he, in turn, is an
"alien" we approach with great caution and misunderstanding ."
The looking glass can be opaque when we naturally disapprove of and feel
apart from those who are evil, distant, different, and disliked . We would also
disregard their interests and refuse to lend them help. We may, in fact, simply
find them incomprehensible . How could a murderer or a thief commit such a
crime? Those people eat what?! There'll always be an England! Unless some
exogenous factor pertaining to this other person intervenes, e .g . meeting by
chance with no option to exit, glancing at a headline announcing the other's
receipt of a prize or a medal, or overhearing a strong positive opinion from a
third party, the lack of sympathy and mindblindness are reinforcing .
One solution to this darkened looking glass is to have an interpreter who
spans the gap between the mindblind person and the distant other . On the one
hand, the interpreter may have sympathetic capabilities because of a shared set
of experiences with the other individual . Having a diverse set of members is
one way in which a group can assure that it has some access to a relevant ToM .
On the other hand, the interpreter may simply have more advanced sympathetic
capabilities . For example, the normal septuagenarians in Happe, Winner and
Brownell's (1998) experiment were more accurate in interpreting stories
involving "double bluffs, mistakes, persuasions, and white lies" than a group of
callow twenty-something mindreaders . 12 The wisdom of elders may lie in their
highly developed ToMs! And so, diversity and grey hairs may combat
occasional mindblindness .
Righteousness can be blinding : first, there is the typical negative characterization of the opponent . In addition, the fanatic's sympathetic capacity is
1 19
constantly engaged with his idealized self or the approving generalized other
(Mead, 1934) in a noisy reaffirmation of correctness . This solipsistic busyness
makes it easy to crowd out the smiles, deeds, looks, and words of the
contending side .
Passing emotion can cloud the mind even of a close other. When we are
temporarily aggrieved at a spouse or a close friend, we find him inarticulate, his
voice inaudible, and his utterances garbled . Conversely, if someone noticeably
shuts his mind to us through curt or irrelevant answers, the sympathetic balance
moves toward the negative pole (Garfinkel, 1967 ; Wyer, Swan & Gruenfeld,
1995) .
Those at the top of an organization may not be able to read the minds of
those below accurately, because the natural signals are controlled and distorted .
Those below may have independent, deceitful reasons or more submissive,
mindguarding (Janis, 1982) motivations . This distortion is easier to achieve if
the boss is isolated, an outsider, unfamiliar, and disliked . In any case, he is the
recipient of only winning smiles, sycophantic gestures, and positive, literal
statements, and accordingly, is made blind .
Just as the non-autist reads too much into the film of geometric shapes, he
may also assume too much about the transparency of his own mind . In one
experiment, participants felt that despite their intention to stay neutral and not
give off any signs, observers would be able to guess when they told a little lie,
drank a foul tasting liquid, or were concerned about someone's bad behavior
(Gilovich, Medvec & Savitsky, 1998) . As a result, desires and beliefs that
should be made explicit are left unsaid : a spouse expects the other to "surprise"
him with just what he wants on his birthday ; a friend stews while waiting for
the other to apologize .
The assumption of similarity that initiates imitation and attribution in
acquaintanceship can distort the reading of the other's mind : sometimes we
assume that we see the contents of our own mind in the looking glass . This selfreflection is a weaker version of the inner conversation of righteousness :
We experience our own point of view more or less directly, whereas we must always attain
the other person's in a more indirect manner . Furthermore, we are usually unable to turn
our own viewpoint off completely when trying to infer the other's, and it usually continues
to ring in our ears while we try to decode the other's (Flavell, 1977, p . 124, quoted in
Nickerson, 1999) .
The vividness of the content in our mind seems to be related to the chance that
we will automatically project it into the mind of the other - a catchy tune, a
long-practiced skill, a money-making bit of inside information (Nickerson,
1999) .
120
DAVID SALLY
An intrapersonal example is the way in which we simply identify our future

self with our present self and work to extend and preserve our present
preferences (Sally, 2000d) . We imagine ourselves as fundamentally consistent .
So, the sober will tie themselves to the mast and plan against addiction, while,
before hitting bottom, even the drowned alcoholic will rationalize his continued
drinking (Sally, 2000e) . So, a young Pete Townshend will pen a lyric hopeful
of his death before middle age, yet his elder version will still be rocking and
rolling . So, a healthy person will fill out an advance directive asking for no life
support in a variety of circumstances, confident that life will not be worth living
in a compromised state, but having fallen ill, the same person will ask for a
ventilator, CPR, etc . (Dams et al ., 1996 ; Lynn et al ., 1997 ; for a review, see
Sally, 2000d) .
For a home office intelligence agent, this solipsistic bias should be an
ever-present concern . One front-line analyst advises, "Avoid mirror imaging
[- projecting your thought process or value system onto someone else -] at all
costs . . . . Not everyone is alike and cultural, ethnic, religious, and political
differences do matter" (Watanabe, 1997) . The field agent is less vulnerable to
mirror imaging for the same reason that this bias is neither widespread or
serious - the interaction order transforms the assumed consensus into a real
consensus ." This convergence is most clear in dialogue : assuming too much,
you say, "Rachel and Bob came back from their honeymoon," and I learn not
only about the concluded travel but also, instantly, about a recent wedding that
I apparently had not been invited to . Outside of the interaction order, the
assumption of similarity becomes the false consensus effect ; inside of the
interaction order, this assumption is usually verified ." Lastly, because of the
presumption of similarity, more people may choose to interact than otherwise
would - a world in which we realized accurately and coldly how different we
all are from one another is a world of unshaken hands, unmet glances,
unmimicked gestures, unachieved coordination, unread thoughts, unheeded
cries, unoffered cooperation, untold jokes, unused metaphors, unperformed
roles, unplayed games, and unseized friendships, in short, a world in which the
looking glass becomes a glass coffin .
NOTES
1 . The phrase, mindreading, is not meant to be discomfiting or to invoke images of
crystal balls, but rather, to serve as shorthand for the non-random anticipation and
prediction of another's thoughts, feelings and intentions .
2 . Park (1998) discusses the pain that this etiological theory caused her as a mother
of an autistic child.
121
3 . Other equivalent labelings are Dennett's (1987) "intentional stance," U . Frith's

"mentalizing" (Frith & Frith, 1999), and Gilbert's (1998) "ordinary personology ."
"Theory of mind" seems to be nearing the wire in the coinage race .
4. Some have claimed that the ToM deficit vastly overshadows the other two .
Interested readers should see Carruthers (1996) who claims that weak central coherence
is largely an effect of a ToM deficit ; Perner and Lang (2000) who argue that ToM is a
prerequisite of the executive function ; and Jarrold et al. (2000) and Plaisted (2000) for
counter-arguments .
5 . Autistic adult subjects detected only one-quarter of the usually salient social
interpretations, e.g. the rectangle is a room, the big triangle is a bully .
6. It is also true that in many static situations, we can manifest significant biases in
reading the minds of others . I will deal with these biases in the last section of the paper,
but for now, we should note that many dynamic settings improve the accuracy of the
social perceptions (Ickes, 1997 ; Swann, 1984) .
7 . Hence, sympathy can "solve" the prisoners' dilemma (Sally, 2000a, 2001) .
8 . Riecken and Homans (1954) advance the sentimental equation S op = f(iap Ap)suggesting that the liking of person o for person p was an increasing function of the
frequency of their interaction and the "goodness" of p's activities . The work of Lott and
Lott (1972), Clark and Mills (1979), Frank (1988), Baumeister and Leary (1995),
Gilbert (1998), and, especially, Bargh (1997) contain many of the elements here,
although none is focused on issues of sympathy and mindreading .
9 . Cacioppo and Berntson (1994) prove that evaluation occurs in a two-dimensional
space with a "positive" axis and a "negative" axis . Hence, an object may be both good
and bad, or some mixture, and an overall assessment results from a balancing between
the measures .
10 . Thanks to a referee for suggesting this line of thought .
11 . Accordingly, autism proves to be a fatal paradox for those who wish to push the
social mind argument to its endpoint . For example, Mead (1934) writes, "Mind arises
in the social process only when that process as a whole enters into, or is present in the
experience of any one of the given individuals involved in that process" (p . 134) . The
social process is not present in the mindblind, and yet his mind is present (at least,
partially) in the mind of the non-autist, and he is part of society .
12 . The ToM performance of the elders contrasted with other cognitive capacities
that lagged those of the youngsters .
13 . The bigger danger for the spy is that he will forget his original mission as he
gains sympathy and a theory of the enemy's mind .
14 . This distinction is one way to reconcile the large gap between the "bias" school
of person perception (e .g . Nisbett & Ross, 1980 ; Keysar, 1994 ; Gilovich, Savitsky &
Medvec, 1998) and the "accuracy" school (e .g . Swann, 1984; Funder, 1995 ; Ickes,
1997) .
REFERENCES
Argyle, M ., & Dean, J . (1965) . Eye-Contact, Distance and Affiliation . Sociometry, 28, 289-304.
Aron, A ., Aron, E . N ., & Smollan, D . (1992) . Inclusion of the Other in the Self Scale and the
Structure of Interpersonal Closeness . Journal of Personality and Social Psychology, 63,
596-612 .
122
DAVID SALLY
Aron, A ., Aron, E. N ., Tudor, M., & Nelson, G. (1991) . Close Relationships as Including the Other
in the Self. Journal of Personality and Social Psychology, 60, 241-253 .
Bargh, J . A . (1997) . The Automaticity of Everyday Life . Advances in Social Cognition, 10, 1-61 .
Baron-Cohen, S. (1995) . Mindblindness . Cambridge : MIT Press .
Baron-Cohen, S . (2000) . Theory of Mind and Autism : A Fifteen Year Review . In: S . Baron-Cohen,
H. Tager-Flusberg & D. J. Cohen (Eds), Understanding Other Minds : Perspectives from
Developmental Cognitive Neuroscience (2nd ed.) . Oxford : Oxford University Press .
Baron-Cohen, S ., Jolliffe, T., Mortimore, C.,& Robertson, M . (1997) . Another Advanced Test of
Theory of Mind : Evidence from Very High Functioning Adults with Autism or Asperger
Syndrome . Journal of Child Psychology and Psychiatry, 38, 813-822 .
Baumeister, R . F., & Leary, M . R. (1995) . The Need to Belong : Desire for Interpersonal
Attachments as a Fundamental Human Motivation . Psychological Bulletin, 117, 497-529 .
Bavelas, J. B ., Black, A ., Lemery, C . R., & Mullett, J . (1986) . Ì Show How You Feel' : Motor
Mimicry as a Communicative Act . Journal of Personality and Social Psychology, 50,
322-329 .
Bernston, G. G ., Boysen, S . T., & Cacioppo, J . T. (1993) . Neurobehavioral Organization and the
Cardinal Principle of Evaluative Bivalence . Annals of the New York Academy of Sciences,
702,75-102.
Bettelheim, B . (1959) . Feral Children and Autistic Children . American Journal of Sociology, 64,
455-467 .
Bettetheim, B . (1967) . The Empty Fortress. New York : The Free Press .
Bornstein, R . F. (1989) . Exposure and Affect : Overview and Meta-Analysis of Research,
1968-1987 . Psychological Bulletin, 106, 265-289 .
Boucher, J ., Lewis, V., & Collis, G . (1998) . Familiar Face and Voice Matching and Recognition in
Children with Autism . Journal of Child Psychology and Psychiatry, 39, 171-181 .
Brockner, J., & Swap, W. C . (1976) . Effects of Repeated Exposure and Attitudinal Similarity on
Self-Disclosure and Interpersonal Attraction . Journal of Personality and Social Psychology,
33,531-540 .
Brownell, H., Griffin, R., Winner, E ., Friedman, 0 ., & Happ6, F. (2000) . Cerebral Lateralization
and Theory of Mind. In : S . Baron-Cohen, H . Tager-Flusberg & D . J. Cohen (Eds),
Understanding Other Minds : Perspectives from Developmental Cognitive Neuroscience
(2nd ed.) . Oxford: Oxford University Press .
Byrne, D ., & Nelson, D . (1965) . Attraction as a Linear Function of Proportion of Positive
Reinforcements . Journal of Personality and Social Psychology, 1, 659-663 .
Cacioppo, J. T., & Berntson, G . G . (1994) . Relationship Between Attitudes and Evaluative Space :
A Critical Review, With Emphasis on the Separability of Positive and Negative Substrates .
Psychological Bulletin, 115, 401-423 .
Cappella, J . N . (1997) . Behavioral and Judged Coordination in Adult Informal Social Interactions :
Vocal and Kinesic Indicators . Journal of Personality and Social Psychology, 72, 119-131 .
Carroll, L. (1871) . Through the Looking Glass and What Alice Found There . New York: Dell
Publishing, published 1992 .
Carruthers, P. (1996) . Autism as Mind-Blindness : An Elaboration and Partial Defence . In : P.
Carruthers & P. K . Smith (Eds), Theories of Theories of Mind. Cambridge : Cambridge
University Press.
Charman, T., Swettenham, J ., Baron-Cohen, S ., Cox, A., Baird, G ., & Drew, A . (1997) . Infants
with Autism : An Investigation of Empathy, Pretend Play, Joint Attention, and Imitation.
Developmental Psychology, 33 781-789 .
123
Chartrand, T. L ., & Bargh, J. A . (1998) . The Chameleon Effect: The Perception-Behavior Link and
Social Interaction . Journal of Personality and Social Psychology, 76, 893-910 .
Clark, M . S ., & Mills, J . (1979) . Interpersonal Attraction in Exchange and Communal
Relationships . Journal of Personality and Social Psychology, 37, 12-24 .
Collins, R . (1989) . Toward a Neo-Meadian Sociology of Mind . Symbolic Interaction, 12, 1-32 .
Cooley, C . H . (1902) . Human Nature and the Social Order. New York: Charles Scribner's Sons.
Cooley, C . H. (1927) . Life and the Student. New York : Knopf.
Damasio, A . R. (1994). Descartes' Error: Emotion, Reason, and the Human Brain . New York :
Avon .
Danis, M ., Mutran, E ., Garrett, J . M ., Steams, S . C ., Slifkin, R. T., Hanson, L ., Williams, J. F., &
Churchill, L . R. (1996). A Prospective Study of the Impact of Patient Preferences on LifeSustaining Treatment and Hospital Cost . Critical Care Medicine, 24, 1811-1817 .
Darley, J. M ., & Berscheid, E. (1967) . Increased Liking as a Result of the Anticipation of Personal
Contact . Human Relations, 20, 29-40 .
Darwin, C . (1936) . The Origin of the Species and The Descent of Man . New York : The Modem
Library .
Davis, M . H., Conklin, L ., Smith, A ., & Luce, C . (1996) . Effect of Perspective Taking on the
Cognitive Representation of Persons : A Merging of Self and Other. Journal of Personality
and Social Psychology, 70 713-726 .
Dawson, G ., Meltzoff, A . N., Osterling, J., Rinaldi, J ., & Brown, E. (1998). Children with Autism
Fail to Orient to Naturally Occurring Social Stimuli . Journal of Autism and Developmental
Disorders, 28, 479-485 .
Dennett, D . C . (1987) . The Intentional Stance. Cambridge : MIT Press .
Dissanayake, C ., & Crossley, S . A . (1996) . Proximity and Sociable Behaviours in Autism :
Evidence for Attachment . Journal of Child Psychology and Psychiatry, 37, 149-156 .
Feld, S . L . (1982). Structural Determinants of Similarity Among Associates . American
Flavell, J . H . (1977) . Cognitive Development . Englewood Cliffs : Prentice Hall .
Frank, R . H . (1988) . Passions Within Reason . New York: W. W. Norton .
Frith, C . D ., & Frith, U . (1999) . Interacting Minds - A Biological Basis . Science, 286,
1692-1695 .
Frith, C . D ., & Frith, U . (2000) . The Physiological Basis of Theory of Mind : Functional
Neuroimaging Studies . In : S . Baron-Cohen, H . Tager-Flusberg & D . J . Cohen (Eds),
(2nd ed .) . Oxford: Oxford University Press.
Frith, U . (1989) . Autism : Explaining the Enigma . New York: Blackwell .
Frith, U ., Happe, F., & Siddons, F. (1994) . Autism and Theory of Mind in Everyday Life . Social
Development, 3, 108-124 .
Funder, D . C . (1995) . On the Accuracy of Personality Judgment : A Realistic Approach .
Psychological Review, 102, 652-670.
Garfinkel, H . (1967) . Studies in Ethnomethodology. Englewood Cliffs : Prentice-Hall .
Gilbert, D . T. (1998) . Ordinary Personology. In : D . T. Gilbert, S . T. Fiske & G. Lindzey (Eds), The
Handbook of Social Psychology (4th ed .) . Boston: McGraw-Hill .
Gilovich, T., Savitsky, K ., & Medvec, V. H . (1998) . The Illusion of Transparency : Biased
Assessments of Others' Ability to Read One's Emotional States . Journal of Personality and
Goffman, E . (1983) . The Interaction Order. American Sociological Review, 48, 1-17 .
124
DAVID SALLY
Gopnik, A ., Capps, L ., & Meltzoff, A. N . (2000) . Early Theories of Mind : What the Theory Theory
Can Tell Us About Autism. In: S . Baron-Cohen, H. Tager-Flusberg & D. J. Cohen (Eds),
Understanding Other Minds: Perspectives from Developmental Cognitive Neuroscience
(2nd ed.) . Oxford : Oxford University Press .
Gopnik, A., & Meltzoff, A . N. (1997). Words, Thoughts, and Theories . Cambridge: MIT Press.
Happ6, F (1993) . Communicative Competence and Theory of Mind in Autism : A Test of
Relevance Theory. Cognition, 48, 101-119 .
Happ6, F (2000). Parts and Wholes, Meaning and Minds : Central Coherence and its Relation to
Theory of Mind . In: S . Baron-Cohen, H. Tager-Flusberg & D . J. Cohen (Eds),
(2nd ed .). Oxford : Oxford University Press .
Happ6, F ., Winner, E ., & Brownell, H. (1998) . The Getting of Wisdom : Theory of Mind in Old
Age. Developmental Psychology, 34, 358-362 .
Hartshorne, H., & May, M . A . (1929) . Studies in Service and Self-Control. New York :
MacMillan .
Hatfield, E., Cacioppo, J . T., & Rapson ., R . L . (1994) . Emotional Contagion . Paris : Cambridge
University Press .
Heider, F (1958) . The Psychology of Interpersonal Relationships . Hillsdale: Lawrence Erlbaum
Associates .
Heider, F., & Simmel, M. (1944). An Experimental Study of Apparent Behavior . American Journal
of Psychology, 57, 243-259 .
Heise, D. R. (1999) . Controlling Affective Experience Interpersonally . Social Psychology
Quarterly, 62, 4-16 .
Hobson, R . P., & Lee, A . (1998) . Hello and Goodbye : A Study of Social Engagement in Autism.
Journal of Autism and Developmental Disorders, 28, 117-127.
Hobson, R . E, & Lee, A . (1999) . Imitation and Identification in Autism . Journal of Child
Psychology and Psychiatry, 40, 649-659.
Hobson, R. P, Ouston, J ., & Lee, A . (1988) . What's in a Face? The Case of Autism . British Journal
of Psychology, 79, 441-453 .
Hoch, S . J . (1987). Perceived consensus and predictive accuracy : The pros and cons of projection.
Hughes, C. (1996) . Control of Action and Thought : Normal Development and Dysfunction in
Autism: A Research Note. Journal of Child Psychology and Psychiatry, 37, 229-236.
Hume, D . (1740) . A Treatise of Human Nature . Oxford : Oxford University Press, published
1978 .
Ickes, W. (Ed .) (1997) . Empathic Accuracy. New York : Guilford .
Janis, I . (1982) . Groupthink. Boston : Houghton Mifflin .
Jarrold, C., Butler, D . W., Cottington, E. M., & Jimenez, F. (2000) . Linking Theory of Mind and
Central Coherence Bias in Autism and in the General Population . Developmental
Keysar, B . (1994) . The Illusory Transparency of Intention : Linguistic Perspective Taking in Text .
Cognitive Psychology, 26, 165-208.
Klin, A ., Schultz, R ., & Cohen, D. J. (2000) . Theory of Mind in Action : Developmental
Perspectives on Social Neuroscience . In: S . Baron-Cohen, H. Tager-Flusberg & D . J.
Cohen (Eds), Understanding Other Minds : Perspectives from Developmental Cognitive
Neuroscience (2nd ed.) . Oxford: Oxford University Press
La France, M . (1979). Nonverbal Synchrony and Rapport : Analysis by the Cross-Lag Panel
Technique. Social Psychology Quarterly, 42, 66-70 .
125
Lang, P. J., Bradley, M. M ., & Cuthbert, B. N . (1990) . Emotion, Attention, and the Startle Reflex .
Latane, B ., Eckman, J ., & Joy, V. (1966) . Shared stress and interpersonal attraction . Journal of
Experimental Social Psychology, 1(Supp .), 80-94 .
Lawler, E . J ., & Thye, S . R . (1999) . Bringing Emotions into Social Exchange Theory . Annual
Review of Sociology, 25, 217-244.
Lawler, E . J ., & Yoon, J . (1996) . Commitment in Exchange Relations : Test of a Theory of
Relational Cohesion . American Sociological Review, 61, 89-108.
Lazarsfeld, P. F., & Merton, R . K. (1954) . Friendship as a Social Process . In: M. Berger, T. Abel
& C . H. Page (Eds), Freedom and Control in Modern Society . Princeton : Van Nostrand.
Lee, A., Hobson, R . P., & Chiat, S . (1994) . I, You, Me and Autism : An Experimental Study .
Journal of Autism and Developmental Disorders, 24, 155-176 .
Leekham, S ., & Perner, J . (1991) . Does the Autistic Child Have a Metarepresentational Deficit?
Cognition, 40, 203-218 .
Leslie, A . M., & Thaiss, L . (1992) . Domain Specificity in Conceptual Development : Evidence
from Autism . Cognition, 43, 225-251 .
Lewin, K. (1935) . A Dynamic Theory of Personality. New York : McGraw-Hill .
Lewin, K. (1948). The Background of Conflict in Marriage . In: G . Lewin (Ed.), Resolving Social
Conflicts : Selected Papers on Group Dynamics . New York : Harper.
Little, K . B . (1965) . Personal Space. Journal of Experimental Social Psychology, 1, 237-247 .
Lord, C ., & Magill, J . (1989). Methodological and Theoretical Issues in Studying Peer Directed
Behaviour and Autism. In : G . Dawson (Ed .), Autism : Nature, Diagnosis and Treatment .
New York : Guilford Press .
Lott, A . J ., & Lott, B . E . (1972) . The Power of Liking: Consequences of Interpersonal Attitudes
Derived from a Liberalized View of Secondary Reinforcement . Advances in Experimental
Lydon, J . E ., Jamieson, D . W., & Zanna, M . P. (1988). Interpersonal Similarity and the Social and
Intellectual Dimensions of First Impressions . Social Cognition, 6, 269-288 .
Lynn, J ., Teno, J . M ., Phillips, R . S ., Wu, A . W., Desbiens, N., Harrold, J.,Claessens, M . T.,
Wenger, N ., Kreling, B ., & Connors, A . F., Jr. (1997) . Perceptions by Family Members of
the Dying Experience of Older and Seriously Ill Patients . Annals of Internal Medicine, 126,
97-106.
Mead, G . H. (1934) . Mind, Self and Society, C . W. Morris (Ed .) . Chicago : University of Chicago
Press .
Merleau-Ponty, M . (1969) . The Primacy of Perception and its Philosophical Consequences . In : A .
L . Fisher (Ed .), The Essential Writings . New York: Harcourt, Brace .
Mitchell, P., Saltmarsh, R., & Russell, H . (1997) . Overly Literal Interpretations of Speech in
Autism : Understanding that Messages Arise from Minds . Journal of Child Psychology and
Psychiatry, 38, 685-691 .
Murray, S . L., & Holmes, J . G . (1993) . Seeing Virtues in Faults : Negativity and the Transformation
of Interpersonal Narratives in Close Relationships . Journal of Personality and Social
Nahemow, L., & Lawton, M . P. (1975) . Similarity and Propinquity in Friendship Formation .
Newcomb, T. M . (1956) . The Prediction of Interpersonal Attraction . American Psychologist, 11,
575-586 .
Nickerson, R. S . (1999). How We Know - and Sometimes Misjudge - What Others Know :
Imputing One's Own Knowledge to Others. Psychological Bulletin, 125, 737-759 .
126
DAVID SALLY
Nisbett, R . E., & Ross, L . (1980) . Human Inference : Strategies and Shortcomings of Social
Judgment. New York: Prentice Hall.
Orwell, G. (1952) . Homage to Catalonia . New York : Harcourt, Brace, Jovanovich .
Osgood, C . E., Suci, G. J., & Tannenbaum, P. H. (1967) . The Measurement of Meaning . Urbana:
University of Illinois Press .
Park, C . C . (1998) . Exiting Nirvana. The American Scholar, 67 (Spring), 28-42.
Perner, J ., Frith, U., Leslie, A . M ., & Leekam, S . (1989) . Exploration of the Autistic Child's
Theory of Mind : Knowledge, Belief, and Communication. Child Development, 60,
689-700 .
Perner, J ., & Lang, B . (2000) . Theory of Mind and Executive Function : Is There a Developmental
Relationship? In : S. Baron-Cohen, H. Tager-Flusberg & D . J. Cohen (Eds), Understanding
Other Minds : Perspectives from Developmental Cognitive Neuroscience (2nd ed.) . Oxford :
Oxford University Press .
Plaisted, K . C . (2000). Aspects of Autism that Theory of Mind Cannot Explain . In : S . BaronCohen, H . Tager-Flusberg & D . J. Cohen (Eds), Understanding Other Minds : Perspectives
from Developmental Cognitive Neuroscience (2nd ed .). Oxford : Oxford University Press .
Priest, R . F., & Sawyer, J. (1967) . Proximity and Peership : Bases of Balance in Interpersonal
Attraction. American Journal of Sociology, 72, 633-649 .
Ribot, T. (1897) . The Psychology of Emotions . London : Walter Scott .
Riecken, H . W., & Homans, G. C . (1954). Psychological Aspects of Social Structure . In : G.
Lindzey (Ed.), Handbook of Social Psychology, Vol . II . Cambridge : Addison-Wesley.
Roeyers, H., van Oost, P, & Bothuyne, S . (1998). Immediate Imitation and Joint Attention in
Young Children with Autism . Development and Psychopathology, 10, 441-450 .
Sabbagh, M . A., & Taylor, M . (2000) . Neural Correlates of Theory-of-Mind Reasoning : An EventRelated Potential Study . Psychological Science, 11, 46-50 .
Sacks, O . (1995) . An Anthropologist on Mars. New York : Knopf.
Sally, D . (2000a) . A General Theory of Sympathy, Mind-reading, and Social Interaction, With an
Application to the Prisoners' Dilemma . Social Science Information, 39, 567-634 .
Sally, D . (2000b) . Can I Say "Bobobo" and Mean "There's No Such Thing as Cheap Talk"?
Manuscript, Cornell University .
Sally, D . (2000c) . What an Ugly Baby! Risk Dominance, Sympathy, and the Coordination of
Meaning. Cornell University .
Sally, D . (2000d) . Confronting the Sirens : Rational Behavior in the Face of Changing Preferences .
Journal of Institutional and Theoretical Economics .
Sally, D. (2000e). I, Too, Sail Past : Odysseus and the Logic of Self Control . Kyklos, 53,
173-200 .
Sally, D. (2001). On Sympathy and Games . Journal of Economic Behavior and Organization, 44,
1-30 .
Schultz, R. T., Gauthier, I ., Klin, A., Fulbright, R . K ., Anderson, A. W., Volkmar, F., Skudlarski,
P, Lacadie, C., Cohen, D . J ., & Gore, J. C . (2000). Abnormal Ventral Temporal Cortical
Activity During Face Discrimination Among Individuals With Autism and Asperger
Syndrome. Archives of General Psychiatry, 57, 331-340.
Schwarz, N . (1998). Warmer and More Social : Recent Developments in Cognitive Social
Psychology. Annual Review of Sociology, 24, 239-264.
Sigman, M., & Kasari, C . (1995) . Joint Attention Across Contexts in Normal and Autistic
Children . In : C. Moore & P. J . Dunham (Eds), Joint Attention : Its Origins and Role in
Development. Hillsdale : Lawrence Erlbaum .
127
Sigman, M ., & Ruskin, E . (1999) . Social Competence in Children with Autism, Down Syndrome
and Developmental Delays : A Longitudinal Study . Monographs of the Society for Research
in Child Development, 64, Serial No . 256 .
Simmel, G. (1908) . Sociology of the Senses : Visual Interaction . In : R. E . Park & E . W. Burgess
(Eds), Introduction to the Science of Sociology. Chicago : University of Chicago Press,
published 1921 .
Smith, A. (1790) . The Theory of Moral Sentiments . Indianapolis : Liberty Classics, published
1976 .
Stewart, D . (1854). Elements of the Philosophy of the Human Mind, vol . 3, Edinburgh : Thomas
Constable .
Stinson, L ., & Ickes, W. (1992) . Empathic Accuracy in the Interactions of Male Friends Versus
Male Strangers. Journal of Personality and Social Psychology, 62, 787-797 .
Stone, W. L ., Ousley, O . Y., & Littleford, C . D . (1997) . Motor Imitation in Young Children with
Autism: What's the Object? Journal of Abnormal Child Psychology, 25, 475-485 .
Swann, W. B ., Jr. (1984) . Quest for Accuracy in Person Perception : A Matter of Pragmatics .
Tager-Flusberg, H . (2000) . Language and Understanding Minds : Connections in Autism . In : S .
Baron-Cohen, H. Tager-Flusberg and D . J . Cohen (Eds), Understanding Other Minds :
Perspectives from Developmental Cognitive Neuroscience (2nd ed .) . Oxford : Oxford
University Press
Tantam, D ., Monaghan, L ., Nicholson, H ., & Stirling, J . (1989) . Autistic Children's Ability to
Interpret Faces : A Research Note. Journal of Child Psychology and Psychiatry, 30,
623-630 .
Trout, D . L ., & Rosenfeld, H. M. (1980) . The Effect of Postural Lean and Body Congruence on
the Judgment of Psychotherapeutic Rapport . Journal of Non-verbal Behavior, 4, 176-190.
Tuma, N . B ., & Hallinan, M. T. (1979) . The Effects of Sex, Race and Achievement on
Schoolchildren's Friendships . Social Forces, 56, 576-597 .
Vygotsky, L . (1986) . Thought and Language . Cambridge: MIT Press.
Watanabe, F. (1997) . Fifteen Axioms for Intelligence Analysts . Studies in Intelligence, 1, xxx .
Wegner, D . M . (1995) . A Computer Network Model of Human Transactive Memory . Social
Cognition, 13, 319-339.
Whiten, A ., & Brown, J. (1998). Imitation and the Reading of Other Minds : Perspectives from the
Study of Autism, Normal Children, and Non-Human Primates . In : S . Braten (Ed),
Intersubjective Communication and Emotion in Ontogeny : A Sourcebook, Cambridge :
Cambridge University Press .
Williams, D . (1992) . Nobody Nowhere . New York: Avon Books .
Wittgenstein, L . (1963) . Philosophical Investigations . Tr. G . E . M . Anscombe. Oxford : Basil
Blackwell .
Wyer, R. S ., Swan, S ., & Gruenfeld, D . H. (1995) . Impression Formation in Informal
Conversations. Social Cognition, 13, 243 .
Yirmiya, N ., O. Erel, M . S ., & Solomonica-Levi, D . (1998). Meta-Analyses Comparing Theory of
Mind Abilities of Individuals with Autism, Individuals with Mental Retardation, and
Normally Developing Individuals. Psychological Bulletin, 124, 283-307.
Zajonc, R. B . (1968) . Attitudinal Effects of Mere Exposure . Journal of Personality and Social
Zajonc, R. B . (1980) . Feeling and Thinking : Preferences Need No Inferences . American
Psychologist, 35, 151-175 .
128
DAVID SALLY
Zajonc, R. B . (1998). Emotions. In : D . T. Gilbert, S . T. Fiske & G. Lindzey (Eds), The Handbook
of Social Psychology, 4th Edition . Boston : McGraw-Hill .
Zeitlyn, D . (1995) . Divination as Dialogue : Negotiation of Meaning with Random Responses. In :
E . N . Goody (Ed .), Social Intelligence and Interaction . Cambridge : Cambridge Univ.
Press .
DO YOU TRUST? WHOM DO YOU

TRUST? WHEN DO YOU TRUST?
Chris Snijders and Gideon Keren
ABSTRACT
We try to provide a broader view on the factors that influence the decision
to trust and honor trust . Using the "Trust Game" as our experimental
paradigm, we consider three classes offactors that may be related to trust
issues. The first one considers individual differences with regard to the
probability to trust others (and honor trust of others), or disposition
factors . Which kinds of people are more likely to trust? Second, we
examine who is more likely to be trusted (anticipation factors), focusing
on the appearance of the person who is to be trusted. And third, we
analyze the circumstances under which trust is more likely to evolve
(situation factors). Trusting is easy if there is not much at stake, but if the
stakes and the risk increase, then how does that affect the willingness to
trust? In short, we consider the decision to trust to be dependent on who
you are, on who it is that has to be trusted, and on the specific situation.
Moreover, we analyze which of these three classes is more important,
using a set of experiments designed to measure the impact of disposition,
anticipation, and situation factors . The data suggest that disposition
factors play a minor role; the differences between people with regard to
their likelihood to trust are relatively small . Anticipation factors,
operationalized by varying alter's appearance, had a larger but somewhat
paradoxical effect. Those with a trustworthy appearance are indeed
trusted more easily, but they do not actually behave more trustworthy . By

All rights of reproduction in any form reserved.
ISBN : 0-7623-0767-6
129
130
CHRIS SNIJDERS AND GIDEON KEREN
far the strongest influences were found among the situation factors . Both
the payoffs and the structure of the game have a large impact on trust and
honoring trust .
INTRODUCTION
Suppose you are searching for a '69 Ford Mustang and come across a newspaper
advertisement offering exactly the car you were searching for . Arriving at the address
mentioned in the ad, you are confronted with a gray-haired guy, about 75 years old, called
Sal, who tells you he is an ex-mechanic who patches up old cars . The polish on the shiny
Mustang in the garage has the same roguish twinkle as the guy's eyes . The price he asks
sounds reasonable assuming the car is in as good a shape as he claims ("It's just a hobby,
my friend, I don't want to make a profit .") . Actually, if you were sure the car was in a good
shape as he claimed, you would certainly buy it . You realize, however, that Sal could just
as easily be an ex-salesman as an ex-mechanic, and that the twinkle in his eye might not
reflect his honest nature but the prospect of a large sum of money coming his way for a car
that isn't worth half the price . Would you buy the car? Would you buy it if the price were
lower? Would you buy the car if the guy were not a twinkling-eyed, gray-haired, 75-yearold, but a flashy 30-year-old yuppie? On what factors would you base your decision?
The basic issue of interpersonal trust is this : Ego has to decide whether or not
to hand over control over the situation to Alter. Alter can choose to honor or
betray the trust of Ego, has a certain incentive to betray, and Ego knows it .
Many daily life situations, including the one mentioned above, have a similar
structure that is well captured by the "Trust Game" (Dasgupta, 1988 ; Kreps,
1990) in Fig. 1 .
In the Trust Game, player 1 moves first and has to choose between two
possible acts : moving left (i .e . no trust) thus ending the game (resulting in each
player getting $20), or moving right (i .e . trust) thus letting player 2 decide .
Player 2, if getting his turn, must choose between moving left (i .e . abuse trust)
and keeping $100 for himself, or moving right (i .e . honor trust) entailing an
even split . If player 1 would be sure that player 2 would split the $100, he
would certainly let player 2 decide . Player 1, however, is uncertain about player
2's choice, and if sufficiently unsure may want to settle for the 20 dollars .
Figure 1 shows both this Trust Game, and the more general format in which P
represents the payments both players receive when player 1 decides not to trust,
R stands for payments to both players when player 1 trusts player 2 and player
2 reciprocates, and, finally, S represents what player 1 ends up with when
player 1 has trusted player 2 and player 2 subsequently decides to abuse trust
and receive T. The chosen symbols emphasize the resemblance with the
Prisoner's Dilemma . As in the Prisoner's Dilemma, the relation S < P < R < T
has to be satisfied .
131
Trust in single-shot encounters
i gets US 0
2 gets US 100
1 gets S
2 gets T
1 gets US 50
2 gets US 50
1 gets R
2 gets R
S<P<R<T
Fig. 1 . Trust Game (Example and General Format) .
The approach presented here is based on observing individuals' behavior in

Trust Games, and in that respect differs from many studies that appraise trust
by using one or another measurement scale. We measure a behavioral response
instead of an attitude or a mental disposition . This implies that our pragmatic
operational definition of "trusting" is "choosing to let player 2 decide in a Trust
132
Game", and similarly "being trustworthy" is taken to be identical with

"choosing the (RR) payoff in a Trust Game" . This definition fits much of the
recent literature on trust (e .g . Mayer, Davis & Schoorman, 1995 ; Gambetta,
1988 ; Rousseau, Sitkin, Burt & Camerer, 1998) . In a cross-discipline review on
the meaning of trust, Rousseau et al . (1998) conclude that there seems to be a
consensus that trust is "a psychological state comprising the intention to accept
vulnerability based upon positive expectations of the intentions or behavior of
another" . In the experimental approach presented here, this intention is
assessed with responses in Trust Games . This implies that we consider trust as
a decision under risk, where the risk stems from the possibility of opportunistic
behavior by alter. The trust is in the intention : ego trusts alter, who is capable
of honoring trust but may choose not to . You trust in the intention of your friend
to return the book he has borrowed . Another kind of trust is trust in capability :
ego trusts alter to be able to perform a certain task and alter tries to deliver but
may not be able to. You trust in the capability of the pilot to fly the airplane and
you can in all likelihood assume that a breach of trust is not related to the pilot's
intentions (cf. Misztal, 1996, Chapter 1) . The current paper is solely devoted to
the former interpretation of trust .
By combining the findings of several experiments employing the Trust
Game, we try to provide a broader view on the factors that influence the
decision to trust and honor trust . To this end, the experiments reported here are
categorized into three classes : We first describe experiments that examine
individual differences trying to identify which people are more likely to trust
(disposition factors) . We try to identify individuals or groups of individuals
who are, in general, more likely to trust (or, more likely to honor trust) .
Moreover, we want to find out what the size of the difference between (groups
of) individuals is. Do males trust more often than females? Or, do older people
honor trust more often? Can these differences be quantified? Second, we
examine who is more likely to be trusted (anticipation factors), focusing on the
appearance of the person who is to be trusted . Do people differ with regard to
their trustworthy appearance in the sense that people agree about who appears
trustworthy and who doesn't? How veridical are these perceptions on which
judgments of trust are formed? Finally we analyze the circumstances under
which trust is more likely to evolve (situation factors) . Trusting is easy if there
is not much at stake, but if the stakes and the risk increase, then how does that
affect the willingness to trust? In short, whether you trust someone depends on
who you are, on the one to be trusted, and on the specific situation . In terms of
the Trust Game, whether you trust someone in a Trust Game depends on who
you are, on with whom you are playing the Trust Game, and on what that Trust
Game looks like . In the present chapter we examine the relative impact of these
133
three classes of factors . As is apparent from the structure of the Trust Game, our
investigations are limited to static two-person settings .
Naturally, interactions between elements from different categories are
possible . For instance, some people may trust more under some conditions but
not under others . Or similarly, under some circumstances some people may be
trusted more than others . It is easy to come up with ways in which all three
kinds of factors interact. Some people may have different perceptions of the
same situation (disposition), which influences the way in which they look at the
other person (anticipation) in certain Trust Games (situation) . Though such
interactions are not only possible but also probable, they are beyond the realm
of this study ; only main effects are considered .
An important advantage of using the Trust Game is that it explicitly
disentangles two concepts that are sometimes treated as identical : trustfulness
(willingness to trust) and trustworthiness (willingness to honor given trust) .
Trust is sometimes considered to be a lubricant of social and economic life .
Underlying this assertion is the assumption that trustfulness and trustworthiness come hand in hand, which need not be true . In fact, most scales measuring
trust, like Rotter's (1967) Interpersonal Trust Scale, or the Dyadic Trust Scale
by Larzelere and Houston (1980), actually measure only trustfulness . Finally,
using the Trust Game enables us to manipulate the variables that form the basis
for decisions on trust simultaneously .
The current literature suggests that all the three factors mentioned above play
a role in the decision to trust (or mistrust) . Regarding disposition, there is
evidence that some individuals seem to be more prone to trust than others
(Yamagishi, 1986) . For instance, research on social dilemmas suggests that
economists tend to be less cooperative (Frank, Gilovich & Regan, 1993) and
therefore perhaps less trustful and trustworthy . Trust also seems to depend on
beliefs regarding the others' intentions and inclinations (Parks, Henager &
Scamahom, 1996) . Indeed, predicting and anticipating others' behavior is
based on previous knowledge and experience with those others, or by physical
appearance such as facial outlook and other behavioral signals (van Lange &
Kuhlman, 1994) . Finally, situational variables may also influence considerations of trust . For instance, there is substantial evidence showing that trustful
behavior in interdependent situations, like public good games, strongly
depends on the size of monetary incentives (Ledyard, 1995) .
In essence, we address the explanatory power of the "dispositionalist view"
as opposed to the "situationalist" view (Mischel, 1968), applied to the question
of trust . Mischel has convincingly shown that cross-situation consistencies are
low for several traits, like honesty and aggressiveness . Other research (e .g . Bern
& Allen, 1974) challenged these findings and argued, among other things, that
134
it may very well be that consistencies are stronger for specific domains . Trust
could be one of them .
In what follows, we describe a series of experiments on trust, conducted in
the past few years, and interpret the results in light of a broader literature . The
first three sections address the dispositional, the anticipation, and the situational
questions, respectively . The final section is an attempt to combine the different
results in order to obtain an overall assessment of the determinants of trust .
DO YOU TRUST?
The first approach to the question of trust is to examine whether some people
are more likely to trust than others . Several studies examined which
characteristics of people correlate with trustful behavior yet the empirical
evidence is quite mixed . Some studies investigated sex differences and trust,
but - possibly because trust is not consistently operationalized - findings are
not unequivocal (compare, for instance, Aranoff & Tedeschi, 1968 ; Rapoport et
al ., 1976 ; Dawes et al ., 1977 ; Terrell & Barrett, 1979 ; Wright & Sharp, 1979 ;
Heretick, 1984 ; Lagace & Gassenheimer, 1989 ; Steel, 1991) . Given the
inconsistent evidence, one may doubt whether any universal sex differences, as
far as trust in others is concerned, do exist. Such doubts have been echoed by
Orbell et al .'s (1994) review of sex (gender) differences in social dilemmas, and
by Kagel and Roth's (1995) review of public goods experiments .
Other characteristics that correlate with trust have been less extensively
studied . For instance, Larzelere (1984) found that religious (Christian) students
were less trustful than secular students . Yamagishi and Yamagishi (1994)
provide empirical evidence for the claim that Americans, in general, are more
likely to trust other people compared with Japanese . There is apparently a
positive relation between trust and marital status (Larzelere & Huston, 1980),
a negative relation between trust and suspicion (Misra & Kalro, 1979), and
some complicated relations between trust, race, and socio-economic status
(Switkin & Gynther, 1974 ; Terrell & Barrett, 1979 ; Steel, 1991) . Wheeless and
Grotz (1977) report a relation between trust and self-disclosure .
Liebrand et al . (1986) introduced the "Might versus Morality" effect : some
people evaluate behavior in terms of strong versus weak ("might"), while
others evaluate behavior in terms of good and bad ("morality") . Those who
emphasize might are more likely to behave non-cooperatively in trust issues
(i .e . they are less trustful and less trustworthy), because they tend to behave in
a way that maximizes their own profits without regard for the profit or loss of
others . Likewise, those that emphasize morality are more likely to behave
cooperatively in trust issues, because they tend to behave in a way that
135
maximizes profit from the point of view of the collectivity . Persons who have
been frequently exposed to the evaluation of behavioral alternatives in terms of
might, like those who have studied game theory or economics, may therefore
be less likely to trust others and less likely to honor the trust of others . The
latter factor in particular appears to have a strong effect . Economists tend to
trust significantly less often, as has been reported by Marwell and Ames (1981),
Carter and Irons (1991), and Frank et al . (1993) .
In the first set of experiments we report on (Snijders, 1996), participants
(university students) played several Trust Games and some of the individual
characteristics mentioned above were examined . Participants in the other study
were respondents of a survey among households who played a Trust Game,
controlling for various socio-economic background characteristics . The main
results of these experiments are briefly described below.
Experiment 1 : Effects of Disposition in a Student Population
Participants (466, mostly students) in the first set of experiments played several
(3, 5 or 6) one-shot Trust Games . Sessions were run under laboratory
conditions with about 20 participants per session . In total, 36 different Trust
Games were employed (average values of S, P, R, T were 0, $10, $20, $40
respectively) . For each Trust Game, participants were asked to indicate what
they would choose in the role of player 1 and what they would do in the role
of player 2 . Participants were told in advance that at the end of each session,
two participants would be randomly determined as winners : a player 1 and a
player 2 . Subsequently, one Trust Game would be randomly chosen and the
winners would be paid in accordance with their choices in that particular Trust
Game . This procedure implies that participants knew that "the other player" in
a Trust Game would in fact be a randomly determined other in the same
session . In addition, participants filled out a questionnaire measuring several
characteristics that we considered likely to correlate with trustful or trustworthy
behavior. All participants received a flat fee (about 5 U .S .$) . Further details are
provided in Snijders (1996) and Snijders and Keren (1999) .
The demographic information obtained from participants consisted of sex,
age, field of study, whether they carried a donor codicil, were blood donors, had
any knowledge about game theory, and a factor score of their high school
graduation package . As a last measure of disposition, the "Ring" measure of
social orientations (Liebrand, 1984) was included . This standard test measures
the extent to which a person values payoffs to others (Liebrand & McClintock,
1988) . Usually, the test results are used to divide subjects in "pro-socials" (or
"cooperators"), "individualists", and "competitors" . Competitors are relatively
136
Table 1 .

Net Effects of Disposition on (the probability of) Trustfulness and
Trustworthiness .
Hypothesis
General
Male
Age
Indicators of pro-social behavior
Donor Codicil
Blood donor
Might versus morality
Being a student of economics
Knowledge about game theory
Beta-graduation (math etc .)
Ring measure of social orientation
Pro-social (reference: others)
Trustfulness
Trustworthiness
+0.07
0
-0.09
0
+
+
+0 .10
0
0
0
0
0
0
0
0
0
+0 .20
+0 .19
scarce (about 15% of the subjects), so that individualists and competitors are
often clustered and labeled "pro-self' . In general, the prediction is that prosocials are more cooperative, more trusting, less cynical, more optimistic, and
more caring about other than pro-selfs . Table 1 summarizes the results of our
analyses (Snijders, 1996) . Results are based on a repeated measures Probit
analysis . The first column indicates the direction of the hypothesized effects .
The numeric entries in the following two columns represent the effect on the
probability to trust (honor trust) . For instance, there is a mean difference of
0 .10 in the probability to trust between subjects carrying a donor codicil and
those who do not .
Inspection of Table 1 suggests that males tend to be more trustful and less
trustworthy, and participants who carry a donor codicil seem to be more
trustful . Pro-socials are both more trustful and more trustworthy. Most of the
other indicators were not statistically significant . These weak results regarding
disposition emphasize the danger of assessing other persons' trustworthiness
through stereotyping or categorization (that is, by basing assessments on the
group to which the person belongs to) . Apparently, it is difficult to find a
category of people who trust (or honor trust) significantly more than others, at
least in a student population . Measuring a person's social orientation provides
the most useful information (hence, look for pro-socials, beware of proselves) .
The above experiment should not be considered as conclusive evidence that
differences in disposition to trust are scarce and small . Nevertheless, a literature
137
scan suggests that only few individual-differences indicators have been

unveiled, and even those usually show a weak effect .
A possible objection regarding this experiment is that for some indicators
there is probably not enough variation in the subject population to be able to
find relevant differences . For instance, if people become less trustful as they get
older, we are not likely to find such an effect in a population of students with
ages between 18 and 25 years . This lack of variation was the impetus for
conducting a second experiment .
Experiment 2 : Effects of Socio-Economic Characteristics in a Population at
Broad
The data for this experiment were obtained as part of a large survey of Dutch
households, with 1533 two-person households and 288 singles (Kalmijn,
Bernasco & Weesie, 1995) . Every interview ended with presenting respondents
with a booklet consisting of several socio- and psychometric scales, one or
another version of the Prisoner's Dilemma, and one Trust Game (Bruins &
Weesie, 1996) . About 70% (2283 out of 3354) booklets were returned, of which
half are of interest for our analysis here depending on which characteristics are
taken into account (for instance, not all respondents were willing to state their
incomes, or their political preference) . Behavior of the trustor in a Trust Game
was examined and correlated with several socio-economic characteristics of the
respondents . Admittedly, the choice of the socio-economic variables is
somewhat arbitrary, and the analyses should be considered explorative . Four
clusters were constructed labeled as `General characteristics', `View on life',
`Control over short term gratification', and `Previous breach of trust' .
Included under `General characteristics' were sex, education, household
income, and age . These characteristics are quite broad and permit different
sorts of reasoning regarding their relation with trust . For instance, one could
conjecture that as people grow older, they will trust more (because they care
less about trust being abused), less (because their trust has been betrayed more
often) or use better judgment (because they have learned when to trust and
when not) . Hence, these variables were simply added as controls .
The cluster `View on life' was assessed by kind of job, religion, political
affiliation, and whether one lives in a large city or small village . The third
group of indicators ('Control over short term gratification') was supposed to
examine susceptibility to short-term incentives . Underlying this cluster is the
supposition that the more likely one opts for immediate gratification (like
smoking or drinking), the less trustful she will be because the inability to resist
the (usually short-term) temptation to abuse trust . Finally, the category
138
`Previous breach of trust' consisted of indicators that could be conceived as

having experienced major breaches of trust : having divorced parents, having a
working mother (measured at age 14), the number of previous relationships, the
number of previous cohabiting relations, and the number of prior marriages .
Naturally, the effects of the different payoffs in the different Trust Games were
controlled . The results of this study, based on a Probit analysis that takes
clustering within households into account, are summarized in Table 2 . The
numeric entries represent the effect on the probability to trust . For instance, on
average the difference in the probability of trust between Roman Catholics and
others is 0.06 . Roman Catholics trust less.
Inspection of Table 2 suggests that only few socio-economic characteristics
are correlated with trust . Regarding trust, no evidence was found that responses
obtained from highly educated youngsters differ from responses from a
representative sample of inhabitants of the Netherlands . Two significant effects
stand out . People who donate to charity are also the ones high on trust, and
Table 2.
Net Effects of Socio-economic Characteristics on (the probability

of) Trustfulness .
Trustfulness
General characteristics
Male
Education
Income
Age
View on life
Job is related to economics
Education related to economics
Religion
Roman Catholics
Political pref. (higher= more rightwing)
Donates to charity
Living in city or in village
Control over short-term gratification
Smokes
Drinks
Previous breach of trust
Parents divorced
Mother worked when respondent 14
Number of previous relationships
Number of previous marriages
0
0
0
0
0
0
-0 .06
-0 .02
+0.13
0
0
0
0
+0.11
0
0
1 39
people whose mother worked when they themselves were fourteen years old
are also more likely to trust. The latter is inconsistent with our original
intuitions : The variable `mother worked when respondent 14' was meant to
serve as an admittedly extremely crude proxy for, say, "growing up in a cold
nest" .
Overall, the two experiments display a similar pattern . There is little
evidence to suggest that some groups, or people with certain traits, tend to trust
more often than others . Certainly our sample size warrants the expectation that
even small differences between individuals would have tended to materialize .
A related concern is that perhaps the Trust Game is not a very reliable
measurement instrument, and that we therefore look at trust only through a
haze of measurement error (cf. Thye, 2000) . A conclusive reply to this
argument can only be given based on analyses taking the measurement error
explicitly into account. What we can say is that in the remainder of this chapter,
other effects do materialize, in spite of possible measurement error, which
provides some support for the assertion that the Trust Game has reasonable
reliability as long as one assumes that measurement error does not vary
systematically with the experimental conditions . The effects of disposition we
find here, though relatively weak, seem to relate to pro-social behavior. Those
who score high on a scale measuring pro-social behavior and those who donate
to charity, trust more often . These results are compatible with Uslaner's (in
press) claim that trust is related to a person's general world-view and not so
much to a person's experience with issues regarding trust .
WHOM DO YOU TRUST?

How important are the characteristics of a person you consider to trust?
Certainly, the extent to which one is willing to trust someone else would
depend on characteristics of that other . Trusting a brother or other family
member is different from trusting a stranger . There is also evidence suggesting
that one is more likely to trust someone who is in some sense similar to oneself .
For instance, the Dutch trust the Dutch more than they trust other nationalities
(Snijders, 1998) . Similar results have been reported on cooperation between ingroup members, where in fact "belonging to the in-group" can be manipulated
quite easily (Tajfel & Turner, 1979, 1986), for instance by dividing a group of
people in two separate groups, with each group wearing different hats .
Kinship relation (e.g . family, friends, in-group) may naturally affect the level
of trust . Notwithstanding, in the current section we focus on trust based on
140
appearance . Even without any information about a person, trust in daily life is
often based on guts feelings and the belief (justified or not) that we are able to
accurately "read" a person's face or appearance ("first impressions are often
truest") . There are (at least) two reasons why appearance of others could guide
our assessments regarding their trustworthiness .
First, it is known that people are particularly susceptible to what has been
coined the fundamental attribution error or correspondence bias . There is an
inclination to attribute observed behavior to an actor's disposition more than to
situational circumstances, even when obvious situational incentives exist that
are sufficient to explain that behavior (Ross, 1977) . In other words, we tend to
interpret behavior as a strong sign of character . Moreover, not only a person's
behavior but also his or her appearance is often taken as displaying that
person's character. The shape and expression of the face, body language, and
clothing can be conceived as providing information about a person's nature,
and people are more than willing to use such cues .
A second, and related, argument relies on what has been termed
(dispositional) category based expectancies (Jones, 1990, page 79) : the belief
that members of a particular group (e .g . rock stars, people over 65, men with
thick black eyebrows, etc) share similar dispositions. A glance at someone's
appearance is often enough to relate that person to a group . A category-based
expectancy is frequently formed, especially when no other information about
the other person is available . Whether these conclusions about others, based on
whatever clue, are warranted is of course an entirely different question .
Evidence in the literature suggesting that people's behavior towards other
persons is affected by physical appearance is, to the best of our knowledge,
mainly related to physical attractiveness . Examples abound in diverse areas of
interest . For instance, physically attractive students are judged more favorably
by their teachers on a number of dimensions including intelligence and various
social skills (Ritts, Patterson & Miles, 1992) ; physical attractiveness of children
is associated with their teacher's expectations about them (Clifford & Walster,
1973) ; physically attractive persons with HIV are considered more responsible
for their condition than others (Agnew, Thompson & Vaida, 1994) ; physically
attractive female-offenders are treated differently by male intake personnel
(Rosenbaum & Chesney, 1994) ; men are more likely to help physically
attractive females (Harrel, 1978), and personnel professionals are reported to
have a (slight) bias in favor of physically attractive job candidates (Morrow et
al ., 1990) . The effect of appearance is not limited to physical attractiveness .
Brownlow (1992) suggests that credibility is affected by facial appearance, and
Lennon (1986) demonstrated an effect of clothing on first impressions of
14 1
strangers . Given the empirical evidence, the purpose of experiment 3 was to

directly examine possible relationships between trust and appearance .
Experiment 3 : Do People Base Their Decision to Trust on Appearance?
Participants, in one of our recent experiments, were shown pictures of four
different persons and were asked to assume that they had to play a Trust Game
with one of these persons . They were asked to assume that they would play the
role of player 1 and the persons displayed in the picture would play as player
2 (the trustee) . Their task was to rank order the pictures such that the person
with whom they would most like to play a Trust Game would be ranked first,
and the person they would least like to play with would be ranked last . Lacking
any other cues, it was assumed that participants would judge the pictures based
on how trustful the persons' pictures look like .
The pictures (taken from an old student year-book) were chosen by the
researchers and two assistants . Using our own subjective judgments, two
pictures (T1 and T2) were chosen that seemed to reflect trustworthy
appearances, and another two pictures (U1 and U2) that supposedly
represented untrustworthy appearances . The pictures were displayed in several
sequences so as to avert possible order effects.
The results were unequivocal : Subjects displayed a strong preference for the
two pictures that were originally judged as reflecting trustworthy appearance
(mean rank ordering for pictures T1 and T2 was 1 .54 and 1 .94 respectively,
compared with 3 .23 and 3 .26 for pictures U1 and U2) . More than 75% of the
subjects ranked T1 and T2 over U1 and U2, and the difference in mean ranking
between the former and the last pair was so large that it did not require any
statistical test . These results are compatible with previous research showing
that people tend to trust those who appear more friendly (Mulford et al .,
1998) .
Admittedly, although the results demonstrate the possible role of appearance
in judgments of trust, it does not disclose which facets make the appearance of
a person to be seen as more or less trustworthy . To answer this question, a
research program is required that is beyond the scope of the present paper .
We consider two natural follow-up questions . First, will the level of
agreement exhibited by our participants be maintained with pictures that are
not chosen selectively, as was the case in the previous experiment? Our
experiment only demonstrated that it is possible to select faces on which people
will exhibit high level of agreement regarding their trustworthiness, yet real life
142
is messier. A second, perhaps more important, question is whether trustworthiness judgments, based on physical appearance, are accurate . The evidence is
mixed, though empirical evidence seems to suggest that judging trustworthiness by physical appearance is often unwarranted .
Notwithstanding, there are some indications that appearance is not entirely
an unreliable cue . For instance, some studies show that physically attractive
people actually behave differently than others . They are reported to be more
assertive (Campbell, Olson & Kleim, 1990), and it seems that their
attractiveness has some impact on achievement and psychological well-being
(Umberson & Hughes, 1987) . Adams (1977), in a review article on physical
attractiveness, concluded that a relation exists between appearance and internal
behavioral processes. Note, however, that the fact that physically attractive
persons behave differently than others does not necessarily entail that other
people can correctly predict the behavior of such persons .
Another argument in favor of relying on cues based on appearance is that
people are reasonably accurate in judging another person's (self-reported)
degree of extraversion (Levesque & Kenny, 1993) . As noted already by
Cronbach (1955, 1958), research in interpersonal perception and target
accuracy is problematic . Kenny (1994) presented an extensive discussion on
methodological issues associated with this type of research and concluded that
"social science is beginning to see that the view of the person perceiver as a
biased and faulty perceiver is not complete" (1994, page 142) .
Other empirical evidence casts doubts about the predictive power of
expectations about the behavior of others . Dispositional category based
expectancies may have some value in the sense that correlations exist between
being a member of a certain group and the score on a trait or actual behavior.
However, these correlations are often small and people are poor assessors of
correlations . They tend to overestimate the degree of covariation, especially
when causal explanations along the lines of stereotypes can be made (Nisbett
& Ross, 1980, Chapter 5) . Moreover, their use of correlational information is
often inadequate in that they produce estimates that are too far away from the
population base-rate, particularly when the relevant correlation is weak
(Nisbett & Ross, 1980, Chapter 7) .
Given the inconclusive evidence from the existing literature, it was our aim
to examine whether the perceived trustworthiness of other persons is a
predictor for the probability of trustful behavior of the perceiver, and whether
this perception is correlated with the actual behavior of the perceived target. As
in the other studies, the Trust game was employed as a measure of actual
behavior.
143
Experiment 4 : Effect of Appearance When Perceived Trustworthiness is Not

Manipulated
In a laboratory experiment, 134 participants (mostly students) played a oneshot Trust Game with each of four other participants separately . The
experiment consisted of 17 sessions with 8 participants each (occasionally 7) .
Participants were seated in two opposite rows of four. Each participant was
asked to play a single (and the same) Trust Game with each of the four other
participants he or she was facing . For each dyad, a participant was told to
consider the Trust Game and choose what he or she would do as player 1 given
that the other participant was player 2, and what he or she would do as player
2 given that the other participant was player 1 (the so-called "strategymethod") . For each session, all participants were asked to decide on the basis
of the same Trust Game . For purposes of generality, four different Trust Games
were used across different sessions . After playing the Trust Games, participants
filled out a questionnaire in which they were asked, among other things, to rate
each of the 4 persons in the other row on some personality traits, including
trustworthiness . Communication between participants was not permitted (other
than just looking at the other person) . To prevent participants from thinking that
their ratings served as a consistency test for their responses in the Trust Games,
they were told that the task was meant to examine the degree of consistency in
the first impressions people have about other people (see Snijders & Keren,
1995 for details) .
To asses whether people are indeed more likely to trust those of whom they
say that they appear trustworthy, we compared the choices made by a
participant in the role of player 1 with that participant's rating of the
trustworthiness of the other . Answers to other items of the questionnaire (age,
sex, being a blood donor, donor codicil ownership, knowledge of game theory)
were used as control variables . Values of the payoffs in the Trust Game were
used as additional controls (employing dummy-variables) . The data were
submitted to a logistic regression analysis of subjects' decisions to trust, and a
summary of the results is portrayed in Table 3 . The numeric entries represent
the average effect on the probability to trust (honor trust) .
The analysis is intricate because of the dependencies in the data .
Complications arise not only because each subject made eight decisions (four
times on trustfulness, four times trustworthiness), but also because four
subjects in a single session played with the same players 2 . The difficulty of
dealing with these interdependencies is precisely why, according to Kenny
(1994), a lot of previous results should be interpreted with care . We refer the
reader to Snijders (1996) for the details of the analysis that was used here .
144
Table 3 . Net Effects of Perceived Trustworthiness on (the probability to)

Trust, and of Perceived Trustworthiness on (the probability of) Actual
Trustworthiness .
Anticipation
Player 2 appears trustworthy
Player 2 appears trustworthy
Hypothesis
Trustfulness
+
?
+0 .12
Player 2's
Trustworthiness
Table 3 shows that the mean effect of perceived trustworthiness on the

probability to trust equals 0 .12, which needs some clarification . Trustworthiness was measured on a 20-point scale. The estimated effect of perceived
trustworthiness equals 0 .02 per unit on that scale . About 90% of the subjects
assessed the difference between the most and the least trustworthy person they
were facing as less than 13 units on the scale, with a mean of about a 6 unit
difference . One could say that for the most extreme subjects, the effect of
perceived trustworthiness equals 0 .26 (13 times 0 .02) . On average, it is about
0 .12 (6 times 0 .02) . The effect of perceived trustworthiness on actual
trustworthiness cannot be statistically distinguished from zero . First impressions regarding the trustworthiness of alter are not an adequate predictor of the
actual behavior of alter. This can be understood loosely as providing empirical
evidence against Frank's (1988) idea that tell-tale signs like appearance ease
trust and cooperation between strangers . More precisely, we should say that
appearance apparently does not tell the right tale .
Additionally, subjects did not agree at all about who appears trustworthy
(inter-rater agreement kappa of 0 .04) . Real life is indeed messier than our
Experiment 3 in the sense that in this standard student population there is no
consensus about who looks trustworthy at all . It should be emphasized that the
experiment does not disclose which cues are used to assess trustworthiness . As
mentioned earlier, there is evidence in the literature for a mixture of potential
cues . Clothing (Lennon, 1986), physical beauty (Eagly, 1991) and facial
features (Berry & McArthur, 1985), are probably the most prominent potential
cues in the experimental design . Potential cues could also be based on
(perceived) similarity rather than on generally acknowledged properties of the
other individual . It is a well established research finding that people tend to like
those people whom they perceive to be similar to themselves, and likewise that
people assume that similar others will also like them (e .g . Newcomb, 1961 ;
145
Aronson & Worchel, 1966; Byrne, 1971 ; Condon & Crano, 1988) . Similarity
could possibly account for the fact that subjects disagreed about who looked
trustworthy : they only assessed those subjects as trustworthy who, in some
way, resembled themselves . One could hypothesize on the basis of similarity
that males perceive other males to be more likely to honor their trust ("he is one
of us guys"), and females perceive other females as more likely to honor their
trust ("she is one of us girls") . Alternatively, if a male judges another male's
trustworthiness, he has the advantage of being similar to the one he judges . The
other person is a male just like him, and therefore may better be able to judge
how that other person is going to behave in a Trust Game . This suggests that
player 1 may be more likely to rely on what he or she would do if he or she
were in player 2's shoes, if player 2 happens to be of the same sex . Additional
analyses revealed that no such effects are supported by the data.
WHEN DO YOU TRUST?

Besides possible disposition and physical appearance effects, it is quite likely
that characteristics of the specific situation under which trust is called for, may
also play a role . Here, a wide range of "situational variables" can be thought of .
For instance, trust is more easily established when one expects a long-term
relationship with the partner (Axelrod's, 1984 shadow of the future) . It is also
more readily established given a history of mutually trustful and trustworthy
behavior with the same partner. The presence of a third party may also facilitate
the formation of trust because the potential trust abuser fears for his reputation .
One straightforward aspect of the situation, and one that we will focus on, is
what is at stake . Lending someone $5 requires a different level of trust than
lending $5,000 . Indeed, the experimental literature provides ample support
regarding the sensitivity of people to the size of what is at stake . Even in
situations where a narrow interpretation of economic theory (people behave
like rational egoists) predicts non-cooperative behavior independent of the
payoffs, people are sensitive to changes in the payoffs (Rapoport, Guyer &
Gordon, 1976 ; Steele & Tedeschi, 1967 ; Ledyard, 1995 ; Sally, 1995) .
In the present section we investigate the claim that the degree of trustfulness
and trustworthiness depend on what is at stake. We thus predict that the level
of trust in the Trust Game would depend, among other things, on the values S,
P, R, and T The question is, of course, how differences in these parameters can
account for differences in trust . To this end we adopt what Kelly and Thibaut
(1978) called the "effective decision matrix" according to which people behave
as if they transform the given monetary payoffs into "effective payoffs", where
all attributes relevant to the decision are included . Several ways of creating an
146
"effective Trust Game" are possible. For instance, one could argue that what is
lacking from the model is the feeling of guilt that player 2 will have when he
or she abuses the trust of player 1 . This can be modeled as an effective Trust
Game by assuming that if player 2 in a Trust Game abuses the trust given by
player 1, player 2 values the payoff he then gets (7) somewhat less . Figure 2
displays the effective Trust Game .
In the guilt model, player 2 honors trust if and only if that player's "guilt
parameter" y2 is larger than (T-R)/(T-S) . Following the model, player 2 will not
abuse player l's trust due to her feelings of guilt for pocketing more cash than
player 1 (T for player 2, as opposed to S for player 1) . Specifically, this guilt
feeling is amplified by the fact that it is exactly player l's behavior that enables
player 2 to receive more money, in the first place . Other researchers have
proposed the same ratio as being related to the percentage of cooperative
choices in a Prisoner's Dilemma (Harris, 1969 ; Rapoport & Chammah, 1965),
though these ratios were inspired by arguments based on repeated rather than
one-shot games . Here, the ratio is a natural consequence of the assumptions of
the guilt model . Now suppose that indeed all persons in the role of player 2 bear
their personal guilt parameter and behave accordingly . Then, as the index
1 gets P
2 gets P
1 gets S
2 gets TFig. 2.
)~ (T-S)
1 gets R
2 gets R
An Implementation of the "Effective Form" of the Trust Game When Including

Guilt.
147
increases, the percentage of players 2 who possess a large enough guilt

parameter to be willing to honor trust decreases . Players 1, in the guilt model,
will trust only if the probability that they assign to player 2 honoring their trust
is larger than (P-S)I(R-S) . The probability that player 2 honors trust depends on
(T-R)l(T-S), so the behavior of player 1 depends on both indices . For
convenience, we label these ratios risk and temptation respectively .
The effective Trust Game in Fig . 2 is just one out of many possible effective
Trust Games, and it is tempting to debate which is the most appropriate way to
model guilt (or regret, or altruism) . However, that is not so much the issue here .
What can be shown is that the same indices (risk and temptation) tend to pop
up under different reasonable effective Trust Games . More precisely, under
different "reasonable" effective Trust Games, risk is a major determinant of
player l's behavior, whereas temptation affects the behavior of both player 1
and player 2 . Though other indices can be derived, these are often highly
correlated with risk and temptation (Snijders, 1996 ; Snijders & Keren, 1999) .
Note that this way of modeling Trust Games uses "psychological parameters,"
like y2 introduced in Fig . 2, but generates hypotheses about the relation
between the indices and behavior that do not necessitate actual measurement of
these parameters . The main conjecture of the present section is that trustfulness
(the behavior of player 1) varies with risk and temptation, and trustworthiness
(the behavior of player 2) varies with temptation .
Reanalysis of Experiment 1 : The Effect of the Monetary Incentives in Trust
Games
The data obtained in a set of experiments described earlier (Experiment 1) was
reanalyzed with a focus on the effects of the monetary incentives . Recall that
participants in this set of experiments played several one-shot Trust Games . In
total, 36 different Trust Games were employed (average values of S, P, R, T
were 0, $10, $20, $40 respectively) . The values of S, P, R, and T where chosen
such that there was an adequate variation in the values of risk ((P-S)/(R-S)) and
temptation ((T-R)/(T-S)) . The value of risk varied between 0 .08 to 0 .71, and
temptation between 0.09 and 0 .83 (both risk and temptation are contained
between 0 and 1) . For each Trust Game, participants were asked to indicate
their choice in the role of player 1 and their choice in the role of player 2 .
Table 4 summarizes the results of our analyses . Besides the payoff indices
(risk and temptation), we also include the results of a separate analysis using
the net effects of the separate payoffs . The numeric entries represent the effect
on the probability to trust (honor trust) of an increase of 1 in the independent
variable . For instance, on average the difference in the probability of
148
Table 4.

Net Effects of Monetary Payoffs and of Payoff Indices (risk and
temptation) on (the probability to) Trust .
Hypothesis
Model 1 : Monetary payoffs

S-payoff
P-payoff
R-payoff
T-payoff
Model 2: Indices of monetary payoff
Risk ((P-S)/(R-S))
Temptation ((T-R)l(T-S))
+
+
Trustfulness Hypothesis Trustworthiness
+0 .01
-0.02
+0 .01
0
+
+
0
0
+0.03
-0.02
-0.85
-0.18
0
-
0
-0.94
trustfulness between a risk index of 0 .2 and a risk index of 0 .7 equals (0 .7-0.2)

* -0 .85 =-0 .425 .
Table 4 indicates unequivocally, that the strongest predictors are risk for
trustfulness and temptation for trustworthiness. Among games with low risk
and high risk, we find an estimated difference in probability of trustfulness of
0.85 . That is, risk can vary between 0 and 1, and the percentage of trustfulness
then roughly varies between 0 .95 for low risk and 0 .10 for high risk. An even
larger difference between games associated with low or high temptation is
observed with regard to trustworthiness (0 .94) . These effects are substantially
larger than the effects of disposition and anticipation found in the previous
experiments . The data thus suggest that what matters most, given the factors we
considered, is the situation (defined by the payoffs) .
In some respect, the results of the above analysis are at odds with the
standard dictates of game theory. Following a strict game theoretical
interpretation, player l's behavior should depend entirely on what player 2 will
do . Since player 2's behavior is determined by the temptation to abuse trust,
player l's behavior should therefore heavily depend on player 2's temptation .
Our empirical results, however, suggest that the extent to which player 1 will
decide to trust is mainly determined by the risk involved . Specifically, it
depends on what player 1 can gain as compared to the no trust situation . The
temptation of the other seems to be of relatively minor importance (though still
of the same order of magnitude as the strongest effects of disposition and
anticipation in the previous experiments) . Apparently, people are somewhat
myopic with regard to trust : In considering whether to trust, they underscore
the potential consequences to themselves and tend to put less emphasis on the
likelihood that their trust will indeed be honored.
149
The separate effects of the payoffs S, P, R",, and T are harder to interpret
because, for instance, P is constrained between S and R so that the potential
increase in the probability to trust, depends on the particular Trust Game being
played. It should be noted that, contrary to our initial expectations, the
probability of trustworthy behavior does not depend on the payoff P (and also
not on risk) . The data show that not all people abuse trust, even if there is a
strong monetary temptation to do so . The question is what kind of motives may
induce player 2 not to give in to the temptation? One such motive could arise
from the fact that player 2 appreciates that player 1 has actually gone out on a
limb by trusting, and thus feels a sense of obligation to reciprocate . The larger
the potential sacrifice of player 1, the larger the motivation for reciprocity (i .e .
players 2 are more likely not to give in to temptation for larger values of P, or
for larger values of risk) . This, in turn, triggers the question what it is that
people are reciprocating when they decide to honor the trust given to them . To
answer this question we turn to another situational aspect besides the monetary
payoffs, namely the structure of the game .
Why is trust being honored? One reason, as suggested above, is reciprocity .
An alternative explanation is based on plain altruism. If a trustor is lucky
enough to be confronted with a trustee who has genuine feelings for the trustor
(or for trustors in general), trust will be honored . There is, however, an essential
difference between genuine altruism and reciprocity . Reciprocity implies a
certain obligation or debt, but this obligation is conditional on previous
behavior of the trustor. It is this conditional aspect that discerns it from
altruism . This is also why the judgment of the extent to which reciprocity has
taken place in an experimental setting is complicated by a methodological
problem . Just looking at what subjects do when confronted with one or several
Trust Games does not suffice . In fact, several researchers (e .g, McCabe,
Rassenti & Smith, 1996) have employed the concept of reciprocity to account
for cooperative behavior in different types of games, without taking into
account rival explanations based on altruism. Indeed, there is ample empirical
evidence that considerations of fairness and decency cause people to act
cooperatively even in games (like "dictator" or "ultimatum") in which there is
no place for reciprocity (e .g . Kagel & Roth, 1995, p . 270 ; Dufwenberg &
Gneezy, 1996) . Therefore, to infer reciprocity, it is essential to employ a control
group where the possibility of reciprocity is precluded .
We propose two ways in which characteristics of the situation may influence
the triggering of reciprocity, which we label Bookkeeping (of utility associated
with gains and losses) and Perceived Intentionality (Keren & Snijders, 1999) .
The first is based on calculations for achieving a balance in terms of costs and
150
benefits, the second is a subjective appraisal of the other's willingness to arrive

at mutual cooperation .
In the context of the Trust Game, the bookkeeping conjecture, as the name
suggests, is based on sole considerations of "give and take" . It relates to three
different factors : The amount of utility gained by the trustee (due the trust she
was endowed with), the possible loss conceded by the trustor, and the initial
relative advantage the trustor has given up . For further exposition, consider the
Trust Games depicted in Fig . 3 . The two Trust Games differ only with respect
to the payoffs associated with player 1 not trusting player 2 . According to the
bookkeeping conjecture there are three possible components that determine
whether given trust will be reciprocated . We use subscripts to denote the
payoffs to the two players, respectively . First, the difference (P, - S) embodies
the possible loss player 1 incurs by trusting player 2, and in this respect reflects
a potential sacrifice . In the second Trust Game in Fig . 3 this difference is larger
(40 - 0 versus 20 - 0), so the potential sacrifice is larger . Second, the difference
(R2 - P2), or (T - P 2 ) if player 2 does not honor trust, represents the increment
or possible gain in player 2's utility as a consequence of player 1 trusting .
Again, this difference is larger in the second Trust Game : player 2 gains more
if player 1 trusts in the second Trust Game . Finally, the difference (P, - P2) may
also play a role . Specifically, if P, > P 2, which implies a relative advantage for
player 1, then giving up this advantage can also be considered as part of the
sacrifice of player 1 . In other words, the condition P, > P 2 may magnify the
sacrifice of player 1 and trigger reciprocal behavior of player 2 . Once more,
player 1 sacrifices more when trusting in the second Trust Game . The
bookkeeping conjecture is supported if players 2 would tend to cooperate more
in games where (P, - S) is large, (R 2 - P2 ) is large, and P, > P2 . In other words,
one would expect more honoring of trust in the second Trust Game in Fig . 3 .
This conjecture is tested by examining the difference in behavior of participants
for different pairs of Trust Games, like the pair depicted in Fig . 3 .
In contrast to the bookkeeping hypothesis, the perceived intentionality
conjecture is independent of the specific payoffs or utilities at hand . For trust
to be honored, the trustor has to exhibit a genuine intention for mutual
cooperation . Perceived intentionality then refers to the extent to which the
trustor provides the reciprocator with adequate cues of her cooperative
intentions. Consider, for instance, the Trust Game depicted in the upper part of
Fig . 4 (portraying exactly the same game as in Fig . 1) .
Now imagine that the role of player 1 is eliminated and the game is reduced
to player 2's decision between the right and the left option, as depicted in the
lower part of Fig . 4 . Obviously, in this `restricted' game there is no place for
reciprocity since player 1 did not act at all and hence there is no action that can
151
1 gets US 0
2 gets US 100
1 gets US 50
2 gets US 50
1 gets US 0
2 gets US 100
1 gets US 50
2 gets US 50
Fig. 3. Trust Games with Different (P P) Outcomes : Testing the Bookkeeping

Hypothesis .
be reciprocated . A decision to choose the right option can only be motivated by

considerations of altruism or fairness . Evidence in favor of the intentionality
conjecture would be a difference in the proportion of players 2 who would
choose cooperatively across the two games mentioned above (thus, more
cooperation by player 2 in the Trust Game than in the restricted game) . This
conjecture was tested by comparing behavior of trustees in different Trust
Games with behavior of trustees in the control condition .
152
1 gets US 0
2 gets US 100
1 gets US 50
2 gets US 50
1 gets US 0
2 gets US 100
1 gets US 50
2 gets US 50
1 gets US
0
2 gets US 100
1 gets US 50
2 gets US 50
Fig. 4 . Trust Game, Coin-Flip Game, and Control Condition. Testing the Intentionality
Hypothesis .
153
In addition, in order to examine whether different levels of intentionality

play a role, the extent to which player 1 exhibited cooperative intentions was
varied . For instance, suppose that the Trust Game is being played, but the rules
of the game are such that player 1 has to make her choice by flipping a coin .
In that case, to choose right entails that player 1 has made a sacrifice, but it was
clearly not by free will . Since player l's decision is not intentional, there is less
room for expecting reciprocating behavior on player 2's part . We vary the
intentionality by varying the rules regarding the basis on which player 1 has to
decide (free will versus coin-flip) .
Experiment 5 : Empirical Test of the Bookkeeping and Intentionality
Hypotheses
In total, 879 participants, mostly students, played a single Trust Game, a single
control condition, or a single "coin-flip Trust Game" (thus, all comparisons are
based on between-subjects designs) . As independent factors, we considered
whether participants were paid in accordance with their choices or received a
flat fee, and whether they participated in the experimental lab or were
interviewed on campus . Five different base Trust Games were used . All
participants were assigned the role of player 2 . See Keren and Snijders (1999)
for further details . A summary of the results is portrayed in Table 5 . We focus
on the results of testing the bookkeeping and intentionality hypothesis . The
numeric entries in this table represent the average effect on the probability to
honor trust. For instance, on average there is a difference of 0 .25 in the
probability of trustworthiness between players 2 in a Trust Game and players
2 in the control condition.
Table 5.
Net Effects of Testing the Bookkeeping and Intentionality

Hypotheses .
Hypothesis
Trustworthiness
Bookkeeping hypothesis
P,-S payoff difference
R2_P2 payoff difference
P,-P2 payoff difference
+
+
+
0
0
0
Intentionality hypothesis
Game versus control condition
Active player 1 versus coin-flip
+
+
+0 .25
+0 .18
154
The results are clearly in favor of the intentionality hypothesis . The

bookkeeping hypothesis, which should lead to differences in trustworthiness
because of the payoff differences, received no support for any of the three
hypothesized effects (although combining all conditions does show a
significant but small effect). In games where all payoff differences are
relatively large simultaneously, trustworthy behavior is about 10% more likely .
The strongest result is where we compare different kinds of Trust Games with
their control condition . Honoring of trust is on average about 25% more likely
in the Trust Game than in its control condition . These results can be interpreted
as showing a main effect of intentionality, but not an interaction effect based on
bookkeeping . What is important as a trustor is that you convey intentionality
(as opposed to not conveying intentionality at all) . The degree to which you
show intentionality, as operationalized by payoff differences, is of minor
importance . This result can be conceived as rendering support for the notion
that interdependencies lead to trust and commitment, as argued by Lawler and
Yoon (1993) . Likewise, it provides some empirical support for Frank's (1988)
idea that signaling cooperative intent triggers cooperative behavior .
CONCLUSION AND DISCUSSION

Some people are more likely to trust and to honor trust . The strongest evidence
we found pointed to people who score high on the social orientation scale, and
to those who donate to charity . In general, this supports the view that people
with a more altruistic nature tend to trust and honor trust more often . The effect
size for these variables is about 10 percentage points . Surprisingly, there was
no evidence for trust differences with respect to sex, age, education, training in
economics, and many other variables that one could expect to have an
influence . As an interesting aside, especially the lack of a difference with
regard to age and education may serve as an argument against the objections
raised against the frequent use of a student population for experiments .
There are also trust differences connected to the appearance of the one to be
trusted. It is possible to pick out photographs of people who, according to the
majority of the population, appear less trustworthy than others . When we tried
to extrapolate these findings to interactions that resemble real life more
adequately, the results were weaker. There is certainly a substantial correlation
between what people say about the appearance of another player and how they
themselves behave in a Trust Game with that player, thus supporting the view
that appearance matters (with an effect size in the same order of magnitude as
the effects of disposition : about 10 to 15 percentage points) . However, whereas
participants did agree about pre-selected photos, they did not agree about real
155
life others at all . Moreover, their assessments of the other person's

trustworthiness had no relation to the actual behavior of the other person in a
Trust Game .
The most prominent effects on trust were found when we analyzed
situational characteristics : the payoffs and, to a lesser extent, the structure of
the game . Trustfulness correlates highly with the risk involved ((P-S)/(R-S) in
the Trust Game) for the trustor and trustworthiness correlates even higher with
the monetary temptation ((T-R)I(T-S) in the Trust Game) . Effect sizes are large :
about an 80 percentage point difference can be expected between situations
with the most extreme risk and temptation values . The structure of the game
also shows a relatively large effect . The plain fact that trust can be honored only
when it is given, induces trustworthy behavior by the trustee (an effect size of
about 25 percentage points) . It appears that subjects reciprocate the
intentionality that is displayed by the trustor. The strength of the intentions,
however, is unrelated to differences in payoffs .
Though situational characteristics seem to exert the strongest influences on
trust, it is worthwhile to note explicitly that this situationalist view need not be
at odds with possible mechanisms on the dispositional level guiding these
situational differences with regard to trust . That is, although we find strong
effects of monetary payoffs, the differences we documented may still be caused
by people's dispositions . For instance, people may honor trust more easily
because of a feeling of guilt . Given a certain distribution of the degree to which
people are sensitive to feelings of guilt when they abuse trust, one can find
differences in the percentage of trustworthy behavior between Trust Games .
For games with a larger value of temptation, a smaller proportion of the people
will experience feelings of guilt that are sufficiently strong to withstand the
temptation, and the percentage of trustworthy behavior therefore decreases .
Even though one finds a difference in temptation to be accountable for this
decrease, the effect of the difference in trust then runs through differences with
regard to, for instance, feelings of guilt .
It should be emphasized that the comparison between the effects associated
with the three proposed categories is far from being completed . Effects of
kinship, homophily, in-group favoritism, reputation, and culture are likely or
sometimes even shown to exist, but were not part of the present study .
A most important omission of the research reported here is the role of time,
an aspect that has been neglected in much of the research on trust . We
considered single shot Trust Games between strangers, whereas one could
rightfully argue that trust is typically something that grows slowly over time . In
fact, one of the more puzzling issues is why it grows slowly, but seems to
decrease rapidly once abused (Dasgupta, 1988 ; Gautschi, 2000) . Our single
156
shot games are not suited to deal with such issues, though it seems logical that
setups using repeated Trust Games could shed some light on the matter . There
is even a more compelling question at the heart of our approach, namely how
well the intricate topic of trust is captured by the abstract games we employ to
measure it . Some critical comments are in order .
Perhaps the most central issue is whether trust behavior can be conceived
(and studied) in an exclusively rational framework. Does the construct of trust
necessarily require consistency, given the structure of the situation and the
preferences of the actors? We know that predictions based on a narrow
economic interpretation of game theory (nobody trusts and nobody honors
trust) are plainly wrong . In response, one may claim the need for more
sophisticated models, for instance models that would more accurately capture
actors' preferences or would better represent the affective aspects associated
with trust. Alternatively, one may argue that an exclusive rational model cannot
(in principle) capture all the subtleties underlying trust considerations on the
simple grounds that people are not always rational, especially when trust is
concerned . The Trust Game undoubtedly captures some primary aspects of
trust yet it may not lend itself to some ("non-rational") particulars associated
with messy real life situations . In addition, the specific structure of the trust
game and the manner in which it is presented in controlled experiments may
implicitly guide participants to behave consistently. The situation presented to
participants is well structured, creating a "conversational logic" (Schwarz,
1996) that triggers neat and orderly behavior from subjects . Asking participants
to play several Trust Games could be interpreted as what is referred to as
`demand characteristics' . Specifically, participants might conclude (consciously or unconsciously) that the experimenter's aim is to test whether their
behavior is consistent across the games . By responding to such demands,
participants may exhibit behavior that is consistent across payoffs, whereas in
real-life this consistency would have been absent .
A related worry is that we cannot exclude that real-life trust is to a large
extent based on "minor considerations" typically unsuited for treatment in
systematic research : the rise of a brow, the bad taste of your coffee, or the
weather being sunny. Statistical models can predict behavior in a single shot
Trust Game reasonably well, not perfectly (Snijders & Keren, 1999) . The
unexplained part is still large, but what is worse is that this unexplained part
might be, for the larger part, random noise. What does it mean when we find
out that we will never explain more than 5% of the variance?
We feel we have only scratched the surface of the intricate subject of trust .
Our story in the introduction about the mechanic selling the car paints a picture
("Can he be trusted? Should I do it? What is he saying between the lines?") that
157
can only be partially captured in the abstract world of Trust Games, and even
in that abstract world we only looked at part of the main effects . Still, were we
forced to take a stand on the issue raised in the title of our contribution, then,
based on the empirical evidence available to us at this moment, we would claim
that the question is not so much who will trust, and also not whom will be
trusted, but whether the situation induces trustful and trustworthy behavior.
ACKNOWLEDGMENTS
This research was supported by a grant from the Royal Netherlands Academy
of Arts and Sciences (KNAW) .
REFERENCES
Adams, G . R. (1977) . Physical attractiveness research : Toward a developmental social psychology
of beauty . Human Development, 20(4), 217-239 .
Agnew, C . R ., & Thompson, V. D. (1994). Causal inferences and responsibility attributions
concerning an HIV positive Target: The double-edged sword of physical attractiveness .
Journal of Social Behavior and Personality, 9(1), 181-190 .
Aranoff, D ., & Tedeschi, J . (1968) . Original stakes and behavior in the prisoner's dilemma game .
Psychonomic Science, 12, 79-80 .
Aronson, E., & Worchel, S . (1966). Similarity versus liking as determinants of interpersonal
attractiveness. Psychonomic Science, 5, 157-158 .
Axelrod, R . (1984) . The Evolution of Cooperation . New York : Basic Books .
Bern, D . J ., & Allen, A . On predicting some of the people some of the time : the search for crosssituational consistencies in behavior . Psychological Review, 81, 506-520 .
Berry, D . S ., & McArthur, L . Z . (1985) . Some components and consequences of a babyface .
Brownlow, S . (1992) . Seeing is believing : Facial appearance, credibility, and attitude change .
Journal of Non-verbal Behavior, 16(2), 101-115 .
Bruins, J ., & Weesie, J . (1996). HIN95Exp. A set of experiments conducted in connection with
HIN95 . Codebook of HIN95Exp . ISCORE paper, no . 77 . Utrecht University .
Byrne, D . (1971) . The Attraction Paradigm . New York : Academic Press .
Campbell, K . E., Olson, K . R ., & Kleim, D . M . (1990) . Physical attractiveness, locus of control,
sex role, and conversational assertiveness . Journal of Social Psychology, 130(2), 263-265 .
Carter, J ., & Irons, M . (1991) . Are economists different, and if so, why? Journal of Economic
Perspectives, 5(2), 171-177 .
Clifford, M . M ., & Walster, E . (1973) . The effect of physical attractiveness on teacher
expectations . Sociology of Education, 46(2), 248-258 .
Condon, J . W., & Crano, W. D . (1988) . Inferred evaluation and the relation between attitude
similarity and interpersonal attraction . Journal of Personality and Social Psychology, 58,
6M 663 .
Cronbach, L. J. (1955) . Processes affecting scores on "understanding of others" and "assumed
similarity ." Psychological Bulletin, 52, 177-193 .
158
Cronbach, L . J. (1958) . Proposals leading to analytic treatment of social perception scores . In : R.

Tagiuri & L . Petrullo (Eds), Person Perception and Interpersonal Behavior (pp . 353-379).
Dasgupta, P (1988) . Trust as a commodity . In: D. Gambetta (Ed.), Trust. Making and Breaking
Cooperative Relations (pp. 49-72) . Oxford: Basil Blackwell.
Dawes, D ., McTavish, J ., & Shaklee, H. (1977). Behavior, communication, and assumptions about
other people's behavior in a commons dilemma situation . Journal of Personality and Social
Dufwenberg, M ., & Gneezy, U. (1996) . Efficiency and expectations in an experimental game .
Working paper, CENTER for Economic Research, Tilburg University, The Netherlands .
Eagly, A. H., Ashmore, R . D ., Makhijani, M . G ., & Longo, L . C . (1991) . What is beautiful is good,
but . . . : A meta-analytic review of research on the physical attractiveness stereotype.
Psychological Bulletin, 110, 109-128 .
Frank, R. H . (1988) . Passions Within Reason. The Strategic Role of the Emotions . New York:
Norton.
Frank, R . H., Gilovich, & Regan, D . T. (1993) . Does studying economics inhibit cooperation?
Journal of Economic Perspectives, 7(2), 159-171 .
Gambetta, D. (1988) . Can we trust trust? In: D . Gambetta (Ed.), Trust: Making and Breaking
Cooperative Relations (pp. 213-237) . Oxford: Basil Blackwell.
Gautschi, T . (2000) . History effects in social dilemma situations . Rationality and Society, 12,
131-162 .
Harrel, W. A . (1978). Physical attractiveness, self-disclosure, and helping behavior. Journal of
Social Psychology, 104(1), 15-17.
Harris, R. J. (1969). A geometrical classification for 2x2 interval-symmetric games . Behavioral
Science, 14, 138-146 .
Heretick, D. M . L . (1984) . Trust-Suspicion and gender differences in interpersonal functioning.
Journal of Research in Personality, 18, 27-40 .
Jones, E . E . (1990). Interpersonal Perception . New York: W.H. Freeman and Company.
Kagel, J. H., & Roth, A . E . (Eds) (1995) . The Handbook Of Experimental Economics . Princeton,
NJ: Princeton University Press .
Kalmijn, M. (1996) . Kalmijn, M., Bernasco, W., & Weesie, J. Households in the Netherlands 1995 .
Codebook of HIN95 . ISCORE paper no . 67 . Utrecht University .
Kelly, H . H ., & Thibaut, J. W. (1978). Interpersonal Relations : A Theory of Interdependence . New
York : Wiley.
Kenny, D. A . (1994) . Interpersonal Perception . A Social Relations Analysis . New York : The
Guilford Press .
Keren, G ., & Snijders, C . (1999) . Trust and reciprocity. ISCORE paper no . 168 . Utrecht
University.
Kreps, D . M . (1990) . Corporate culture and economic theory . In : J . E. Alt & K. A. Shepsle (Eds),
Perspectives on Positive Political Economy (pp. 90-143) . Cambridge : Cambridge
University Press .
Lagace, R . R., & Gassenheimer, J . B . (1989). A measure of global trust and suspicion : replication.
Psychological Reports, 65, 473-474.
van Lange, P. A. M ., & Kuhlman, D . M, (1994) . Social value orientations and impressions of
partner's honesty and intelligence : a test of the might versus morality effect . Journal of
Larzelere, R. L . (1984) . Dyadic trust and generalized trust of secular versus christian college
students. Journal of Psychology and Theology, 12(2), 119-124 .
159
Larzelere, R. L ., & Huston, T. L. (1980). The dyadic trust scale : toward understanding
interpersonal trust in close relationships . Journal of Marriage and the Family, 595-604 .
Lawler, E . J., & Yoon, J. (1993) . Power and the emergence of commitment behavior in negotiated
exchange . American Sociological Review, 58, 465-481 .
Ledyard, J . O . (1995). Public goods: a survey of experimental research . In: J. H . Kagel & A . E.
Roth (Eds), The Handbook Of Experimental Economics (Chapter 2) . Princeton, NJ :
Princeton University Press .
Lennon, S . J . (1986) . Additivity of clothing clues in first impressions . Social Behavior and
Personality, 14(1), 15-21 .
Levesque, M . J ., & Kenny, D . A. (1993) . Accuracy of behavioral predictions at zero acquaintance:
A social relations analysis. Journal of Personality and Social Psychology, 65, 1178-1187 .
Liebrand, W. B . G . (1984) . The effect of social motives, communication and group size on
behavior in an n-person multi-stage mixed-motive game . European Journal of Social
Psychology, 14, 239-264.
Liebrand, W. B . G ., & McClintock, C . G . (1988) . The Ring measure of social values: A
computerized procedure for assessing individual differences in information processing and
social value orientation . European Journal of Personality, 2, 217-230 .
Liebrand, W. B . G., Jansen, R . W. T. L., Rijken, V. M ., & Suhre, C . J . M . (1986) . Might over
morality : social values and the perception of other players in experimental games . Journal
of Experimental Social Psychology, 22, 203-215 .
Marwell, G ., & Ames, R . E. (1981) . Experiments on the provision of public goods . II . Provision
points, stakes, experience, and the free-rider problem . American Journal of Sociology,
85(4), 926-937 .
Mayer, R . C ., Davis, J . H ., & Schoorman, F. D. (1995) . An integrative model of organizational
trust . Academy of Management Review, 20, 709-734 .
McCabe, K . A., Rassenti, S . J ., & Smith, V. (1996). Game theory and reciprocity in some extensive
form experimental games . Proceedings of the National Academy of Sciences U.S.A.
(Economic Sciences), 93, 13421-13428 .
Mischel, W. Personality and assessment. New York: Wiley .
Misra, S ., & Kalro, A . (1979) . Triangle effect and the connotative meaning of trust in prisoner's
dilemma : a cross cultural study. International Journal of Psychology, 14(4), 255-263 .
Misztal, B . A . (1996) . Trust in Modern Societies. The Search for the Bases of Social Order.
Oxford : Basil Blackwell .
Morrow, P. C ., McElroy, J. C ., Stamper, B . G ., & Wilson, M . A . (1990) . The effects of physical
attractiveness and other demographic characteristics on promotion decisions . Journal of
Management, 16(4), 723-736 .
Mulford, M ., Orbell, J . et al. (1998) . Physical attractiveness, opportunity, and success in everyday
exchange. American Journal of Sociology, 103(6), 1565-1592 .
Newcomb, T. M . (1961) The acquiantance process . New York: Holt, Rinehart and Wilson.
Nisbett, R., & Ross, L . (1980) . Human Inference: Strategies and Shortcomings of Social
Judgment . Englewood Cliffs, NJ : Prentice-Hall .
Orbell, J., Dawes, R. M ., & Schwartz-Shea, P. (1994) . Trust, social categories, and individuals ; the
case of gender. Motivation and Emotion (special Issue : Trust and Distrust: Psychological
and Social Dimensions), 18(2), 109-128 .
Parks, C . T., Henager, R. E., & Scamahorn, S . D. (1996) . Trust and reactions to messages of intent
in social dilemmas . Journal of Conflict Resolution, 40(1), 134-151 .
Rapoport, A ., & Channnah, A . M . (1965) . Prisoner's Dilemma . Ann Arbor : University of
Michigan Press .
160
Rapoport, A., Guyer, M . J ., & Gordon, D . G . (1976) . The 2 x 2 game. Ann Arbor : University of
Michigan Press .
Ritts, V., Patterson, M . L., & Tubbs, M. E. (1992) . Expectations, Impressions, and judgments of
physically attractive students : a review. Review of Educational Research, 62(4), 413-426 .
Rosenbaum, J. L ., & Chesney-Lind, M . (1994) . Appearance and delinquency : A research note.
Crime and delinquency, 40(2), 250-261 .
Ross, L . (1977). The intuitive psychologist and his shortcomings . In : Berkowitz, L. (Ed .),
Advances in Experimental Social Psychology (Vol . 10) (pp. 174-221) . New York:
Academic Press .
Rotter, J . B . (1967) . A new scale for the measurement of interpersonal trust . Journal of Personality,
35,651-665 .
Rousseau, D . M ., Sitkin, S . B ., Burt, R . S ., & Camerer, C . (1998) . Not so different after all : A
cross-discipline view of trust . Academy of Management Review, 23(3), 393-404 .
Sally, D . (1995). Conversation and cooperation in social dilemmas . Rationality and Society, 7(1),
58-92 .
Schwarz, N. (1996) . Cognition and communication . Judgmental Biases, Research Methods, and
the Logic of Conversation . Mahwah, NJ : Lawrence Erlbaum Associates .
Snijders, C . (1996) . Trust and Commitments. Amsterdam : Thela Thesis.
Snijders, C. (1998) . Inter-national trust and organizational cooperation . ISCORE paper, no . 121,
Utrecht University.
Snijders, C ., & Keren, G. (1995) . Trust is in the eye of the beholder : The effect of physical
appearance on trust. ISCORE paper, no . 32 . Utrecht University.
Snijders, C., & Keren, G . (1999) . Determinants of Trust . In : D . Budescu, I . Erev & R . Zwick (Eds),
Games and Human Behavior. Mahwah, NJ : Lawrence Erlbaum Associates .
Steel, J . L. (1991). Interpersonal correlates of trust and self-disclosure . Psychological Reports,
68(3), 2, 1319-1320 .
Steele, M . W., & Tedeschi, J . T. (1967) . Matrix indices and strategy choices in mixed-motive
games . Journal of Experimental Psychology: General, 115, 131-154 .
Switkin, L. R ., & Gynther, M . D . (1974) . Trust, activism, and interpersonal perception in black and
white college students . Journal of Social Psychology, 94(1), 153-154.
Terrel, F., & Barret, R . K . (1979) . Interpersonal trust among college students as a function of race,
sex, and socioeconomic class . Perceptual and Motor Skills, 48, 1194.
Thye, S . (2000) . Reliability in experimental sociology . Social Forces, 78(4), 1277-1309 .
Umberson, D ., & Hughes, M . (1987). The impact of physical attractiveness on achievement and
psychological well-being . Social Psychology Quarterly, 50(3), 227-236.
Uslaner, E . M . (in press). The moral foundations of trust . Draft downloadable from http :/
/w ww.bsos .umd .edu/gvpt/uslaner.
Wheeless, L . R., & Grotz, J. (1977) . The measurement of trust and its relationship to disclosure.
Human Communication Research, 3(3), 250-257 .
Wright, T. L., & Sharp, E . G. (1979) . Content and grammatical sex bias on the interpersonal trust
scale and differential trust toward women and men . Journal of Consulting and Clinical
Psychology, 47(1), 72-85 .
Yamagishi, T. (1986) . The provision of a sanctioning system as a public good . Journal of
Personality and Social Psychology, 51(1), 110-116 .
Yamagishi, T., & Yamagishi, M . (1994) . Trust and commitment in the United States and Japan.
Motivation and Emotion, 18(2), 129-166 .
WHERE DO SOCIAL STRUCTURES

COME FROM?
Satoshi Kanazawa
ABSTRACT
While structuralism and network theory have been enormously successful
empirically, they have not been able to explain the origins of social
structures and networks . I contend that the emerging field of evolutionary
psychology can help us explain how some social structures and networks
emerge. I illustrate my point with a persistent empirical puzzle in the
social networks literature (why women have more kin in their personal
networks than men do), and provide an evolutionary psychological
explanation for this phenomenon. I test two implications of this
explanation with the 1985 Social Networks module of the General Social
Survey. The data provide support for the evolutionary psychological
explanation of women's kincentric networks.
INTRODUCTION
Structuralism, and its most successful versions, social network analysis
(Wasserman & Faust, 1994) and network exchange theory (Willer, 1999), are
among the dominant theoretical perspectives in sociology . Structuralist theories
explain individual behavior and interpersonal relations in terms of the actors'
locations in the social structure, in particular, their ties (or lack thereof) to other
actors and the latter's ties (or lack thereof) to still other actors . Network
Advances in Group Processes, Volume 18, pages 161-183.

ISBN: 0-7623-0767-6
161
162
SATOSHI KANAZAWA
analysis explains individual behavior in terms of the properties and configurations of the networks (the presence or absence of ties between nodes), not in
terms of the attributes of individual actors (Mayhew, 1980, 1981) .
While network theories have been enormously successful in explaining
individual behavior in terms of the properties of social structure, they leave
social structures themselves exogenous . Just as the microfoundations of
rational choice theory, which explain individual behavior partly in terms of
individual values and preferences, leave these values and preferences
exogenous (Stigler & Becker, 1977), network theories leave the networks
exogenous . Structural theories cannot answer the question : Where do social
structures come from? (just as rational choice theory cannot answer the
question : Where do individual values and preferences come from?) .
In this paper I will argue that the emerging field of evolutionary psychology
can explain the origins of some (albeit not all) social structures, just as I have
elsewhere (Kanazawa, 2001) argued that evolutionary psychology can explain
the origins of some (albeit not all) values and preferences in the microfoundations of rational choice theory. I will first provide critiques of
structuralism and network theory, especially their attempt to explain social
networks in terms of homophily. I will then sketch out the foundational
principles of evolutionary psychology and how it can potentially explain the
origins of social structures . I will illustrate how evolutionary psychology can
explain social networks by providing an evolutionary psychological explanation of a persistent puzzle in network theory : Why women have more kin in
their personal networks than men do . I will present empirical evidence from the
1985 Social Networks module of the General Social Survey that supports my
explanation for women's kincentric networks .
PROBLEMS WITH STRUCTURALISM

Apart from its inability to explain the origins of social structures, structuralism
has a few theoretical problems, despite its tremendous empirical success . First,
while structuralism and network theory purport to explain individual behavior
purely in terms of properties and configurations of the social structure, all
structural theories must nonetheless make some assumptions about the internal
states of individual actors .' Take, for instance, Blau's (1977a) macrostructural
theory, one of the most prolific and successful sociological theories from any
perspective . While the theory aims to explain the patterns of intergroup
relations from the properties of groups in society (mainly, the extent and
patterns of heterogeneity and inequality among individuals within and between
groups), the theory must nevertheless assume certain "sociopsychological
Where Do Social Structures Come From?
163
tendencies" (Blau, 1977b, p . 46) on the part of individual actors . For example,
individuals in Blau's theory must want to marry and they must want to
associate with others . In fact, Blau must assume a uniform level of desire to
marry and associate with others across all individuals ; otherwise, his theory
does not work . If individuals do not have a constant level of desire to marry or
associate with others, then heterogeneity, in the face of ingroup preferences,
will not lead to greater levels of intermarriage and intergroup relations because
individuals could simply choose not to marry or associate with anyone at all
when there are no ingroup members to marry or associate with . Then his
theorem (T- 11 : Increasing heterogeneity increases the probability of intergroup
relations (Blau, 1977a, pp . 78-83)) will be logically false .
Some of Blau's assumptions explicitly refer to internal states of individual
actors . For instance, the very first, and therefore the most fundamental,
assumption of the theory assumes that individuals are homophilous (A- 1 :
Social associations are more prevalent among persons in proximate than
between those in distant social positions (Blau, 1977a, pp . 36-41)) . Since no
prior assumptions are made about the structural constraints on associations, this
homophilous tendency must necessarily come from individual preferences and
desires to associate with others in similar social positions .' In other words, A-1
posits choice homophily, not induced homophily (McPherson & Smith-Lovin,
1987, pp . 371-372) . In fact, Blau (1977a, p . 36 ; emphases added) explicitly
states : "People in similar social positions share social experiences and roles,
and have similar attributes and attitudes, which promote social intercourse
among them . This is the reasoning underlying the first axiom, on which
numerous theorems rest."
While Blau's macrostructural theory, and other structural theories, must
assume certain sociopsychological tendencies, they cannot explain why
individuals have these tendencies . Why do individuals want to marry? Why do
individuals want to associate with others? Why are individuals homophilous?
Another theoretical problem of structuralism is that it treats all actors as
equivalent and interchangeable nodes in a social network (Blau, 1989, p . 53 ;
Smith-Lovin & McPherson, 1993, p . 223) . "Structuralists do not attribute
social or psychological characteristics to individual humans . . . . Social
phenomena are properties of social networks" (Mayhew, 1980, p . 346) . In
network theory, actors (egos) who have ties to identical or similar others
(alters) are called "structurally equivalent" (Lorrain & White, 1971) or
"regularly equivalent" (Sailer, 1978), and the theory predicts their behavior will
be similar (since they share all structural characteristics) . Alters in turn are
defined by to which other actors they have ties . For instance, network theory
does not posit that men and women are inherently different . It explains all sex
164
SATOSHI KANAZAWA
differences in behavior purely in terms of the differences in network ties (their

strengths, numbers, and densities) between men and women (McPherson &
Smith-Lovin, 1982 ; Smith-Lovin & McPherson, 1993) .
However, it is obvious that actors and their behavior are not entirely
reducible to their network ties, and there can be vast individual differences even
between actors who are structurally or regularly equivalent . And actors do
possess inherent attributes and characteristics . Take Mark's (1998b) theory of
musical taste acquisition, for example . He argues that individuals acquire their
tastes in music (what types of music they like) from others with whom they
associate . If one has many ties to others who listen to rock, one acquires a
preference for rock ; if one has many ties to others who listen to country, one
acquires a preference for country.
A moment's reflection will reveal, however, that this is not entirely true .
While we often acquire our musical tastes from our friends when we are young,
we do not do so from our parents, even though we may have equally strong ties
to both our friends and parents . In fact, we often develop a strong distaste for
a certain type of music precisely because our parents like it, or develop a strong
taste for it precisely because our parents hate it.' We developed a taste for Elvis
and the Beatles precisely because our parents listened to Lawrence Welk and
hated Elvis' gyrating hips and the Beatles' long hair . Of course, we liked Elvis
and the Beatles because our friends liked them . If we acquire our musical tastes
from those to whom we have close ties, why do we like the music our parents
hate and our friends like?Why do we rebel against our parents (and not our
friends), but then only when we are young? Why are our parents different from
our friends?
PROBLEMS WITH HOMOPHILY

Apart from the two theoretical problems identified above, the most significant
problem with structuralism and network theory is its inability to account for the
origins of social structures and networks . While all theories must leave some
factors exogenous, and no theories can explain everything, I believe that social
structures, which are the primary causal factors in structuralism, are too
important for it to leave exogenous, just as values and preferences, which are
among the primary causal factors in the microfoundations of rational choice
theory, are too important for it to leave exogenous .
One of the very few factors that structuralists use to explain the origin of
networks is homophily (McPherson & Smith-Lovin, 1987) . The principle of
homophily (Mark, 1998b, pp . 454-455) states that people who are similar in
sociodemographic characteristics are more likely to interact with each other
165
than are people who are dissimilar . A large number of empirical studies
conclusively demonstrate that personal networks are highly homophilous
(Fischer, 1982, pp . 179-190 ; Marsden, 1987 ; McPherson & Smith-Lovin,
1987, footnote 1) . Homophily in principle can explain the emergence of social
networks from the state of nature . Given a collection of atomized individuals
with no ties, a man is more likely to develop a tie with another man than with
a woman . A white is more likely to develop a tie with another white than with
a black. Eventually, a social network of a given type will emerge from the
collection of individuals with homophilous tendencies .
However, homophily as an explanation of the origins of social structures and
networks runs into at least four specific problems, all of which ultimately
derive from the fact that the concept of homophily is atheoretical. First, this
explanation of the emergence of social structures, in fact, the very concept of
homophily itself, violates one of the fundamental assumptions of structuralism
that actors do not have inherent attributes or characteristics (Mayhew, 1980 ;
Smith-Lovin & McPherson, 1993) . It is strictly with inherent individual
characteristics such as sex, race, ethnicity, education, and income that actors
can be more or less homophilous on these attributes (although Smith-Lovin &
McPherson (1993, footnote 2) deny that any of these attributes is truly
individual in nature) .
Second, given that individuals have multiple sociodemographic characteristics, and given that the multiple correlation among these dimensions R < 1 .0
(or, to use Blau's (1977a) language, given less than perfect consolidation of
multiple parameters), when individuals are homophilous on one dimension,
they are necessarily less homophilous on others . Individuals cannot be
maximally or equally homophilous on two or more dimensions simultaneously .
Conversely, once again, given R < 1 .0, one can always identify one dimension
on which individuals are necessarily more homophilous than other dimensions .
Given R < 1 .0, personal networks are by definition simultaneously homophilous on some dimensions and heterophilous on others (Blau, 1977b, pp .
44-46 ; Merton, 1972, pp . 21-29) . Chance (random pairing) is the only criterion
against which one can evaluate homophily .
Third, the very important distinction that McPherson and Smith-Lovin
(1987, pp . 371-372) make between choice homophily and induced homophily
turns out not to be a distinction . Choice homophily happens when individuals
have the opportunity to associate with either similar or dissimilar others and
they selectively choose to associate only or mostly with similar others to the
exclusion of dissimilar others . Induced homophily happens when individuals
have the opportunity to associate only or mostly with similar others because the
groups to which they belong are already homogeneous . In homogeneous
166
SATOSHI KANAZAWA
groups, most or all of individuals' associates are already similar to them even
when they choose their associates randomly from other group members and do
not make any conscious effort to associate only with similar others . McPherson
and Smith-Lovin's (1987) study of voluntary organizations in Nebraska shows
that most personal networks are homophilous because of induced homophily,
not choice homophily. Induced homophily, however, can take place only in the
context of homogeneous groups . How do groups get to be homogeneous in the
first place? Why do individuals join groups whose members are already similar
to them? It is obvious that what produces group homogeneity (a necessary
condition for induced homophily) is prior choice homophily (Feld 1982, p .
798) . Induced homophily at time t is the result of choice homophily at time t-i
(i=1, 2, . . oc) .
Finally, the most significant problem with the concept of homophily and its
atheoretical nature is that nobody knows where homophily comes from . Why
are individuals homophilous? Why are they more homophilous on some
dimensions than others? For instance, why are they more homophilous on age,
sex and race than on education and occupation (Lazarsfeld & Merton, 1954 ;
McPherson & Smith-Lovin, 1987, Table 1 ; Verbrugge, 1977)? If homophily is
so important and pervasive, then why isn't everybody gay? Marriage is one of
the most important social relations in anyone's life, and sex is one of the most
salient sociodemographic dimensions . Why then are most people decidedly not
homophilous in this very important social relation on this very salient
dimension? Obviously, the answer is that most people are biologically
heterosexual and therefore heterophilous on sex in marriage (just like a few are
homosexual and homophilous for the same reason) .' If biological and
evolutionary predispositions underlie whether or not one is homophilous in this
particular social relation on this particular dimension, is it unlikely that similar
predispositions also underlie other choices individuals make in their networks?
EVOLUTIONARY PSYCHOLOGICAL ORIGINS OF

SOCIAL STRUCTURES
I contend that one can solve these and other theoretical problems with
structuralism and network theory by introducing two assumptions . First, actors
are inherently different in ways other than their structural positions and their
network ties to others . The primary ways that actors can be different are in their
preferences and values, in what they want to do (regardless of what they are
structurally constrained to do) . Actors are therefore not interchangeable nodes .
Second, these inherent differences between actors influence their behavior, in
167
addition to the structural effects of their network positions on it .6 These two

assumptions combined lead us to predict that different actors (such as men and
women) would behave differently even if they are structurally or regularly
equivalent .' Of course, since the actors' current network positions are largely
the result of their past choices to associate with some and not others, these
assumptions also lead us to predict that different actors (such as men and
women) will occupy different structural locations . In other words, these two
assumptions help us figure out where social structures come from .
I believe that the emerging field of evolutionary psychology (Barkow,
Cosmides & Tooby, 1992 ; Buss, 1999) helps us understand how different actors
may hold different values and preferences, and how these individual
differences between actors produce different social structures and networks
through their choice of affiliative ties . Evolutionary psychology seeks to
discover universal human nature, which is a collection of domain-specific
psychological mechanisms . A psychological mechanism is an informationprocessing procedure or decision rule that natural and sexual selections have
equipped humans to possess in order to solve a particular adaptive problem (a
problem of survival or reproduction) . Unlike decision rules in microeconomic
subjective expected utility maximization theory or game theory, however,
evolved psychological mechanisms mostly operate behind our conscious
thinking .
Evolutionary psychology is premised on two broad generalizations . The first
generalization, to put it bluntly, is that there is nothing special about humans .
To put it more precisely, "certainly we are unique, but we are not unique in
being unique . Every species is unique and evolved its uniqueness in adaptation
to its environment. Culture is the uniquely human way of adapting, but culture,
too, evolved biologically" (van den Berghe, 1990, p . 428) . Human beings are
just like other animal species, and all the laws of nature, in particular, the laws
of evolution by natural and sexual selection, apply equally to humans as they
do to other species . The second broad generalization is that there is nothing
special about the brain as a human body part ; it is just like the hand or the
pancreas or any other body part. Just as a long history of human evolution has
shaped the hand or the pancreas to perform a specific function, so has the
evolution shaped the human brain to perform certain tasks (solving adaptive
problems) .
The second generalization leads to a very important implication of
evolutionary psychology . Just as the basic shape and functions of the hand and
the pancreas have not changed since the end of the Pleistocene epoch about
10,000 years ago, the basic functioning of the brain has not changed very much
in the last 10,000 years . The human body (including the brain) evolved over
168
SATOSHI KANAZAWA
millions of years during the Pleistocene epoch in the African savanna where
humans lived during most of this time (Maryanski & Turner, 1992, pp . 69-90) .
This environment - African savanna where humans lived in small bands of fifty
or so related individuals as hunter-gatherers - is called the environment of
evolutionary adaptedness (EEA) (Bowlby, 1969) or ancestral environment, and
it is to the EEA or the ancestral environment that our body (including the brain)
is adapted .
Figure 1 presents the basic theoretical structure of evolutionary psychology .
It argues that an adaptive problem leads to an evolved psychological
mechanism, which then usually leads to adaptive (fitness-maximizing)
behavior in the EEA . Evolutionary psychology assumes that most behavior in
the EEA maximizes inclusive fitness of the actor . However, it recognizes that
our current environment may be radically different from the EEA, yet our
psychological mechanisms (just like our hands and our pancreas) are still the
same as they were in the EEA and produce the same behavior as they did in the
EEA. This leads to the distinct possibility that our behavior in our current
environment might be completely maladaptive . To the extent that our current
environment is different from the EEA (to which all psychological mechanisms
are adapted), evolutionary psychology would predict that our current behavior
is maladaptive .
Relying as it does on universal human nature and its sex differences (distinct
male and female human natures) for its explanations of human behavior,
evolutionary psychology is particularly suited for explaining social phenomena
that are culturally universal, such as why it is that young men commit an
overwhelming majority of violent and property crimes in every human society
(Kanazawa & Still, 2000). However, it can also explain culturally variable
phenomena. By specifying how universal human nature interacts with varied
local environments, evolutionary psychology can also explain, for instance,
why women in some societies choose to many polygynously while those in
others choose to marry monogamously in the absence of the institution of
marriage (Kanazawa & Still, 1999) . However, due to its reliance on universal
human nature, evolutionary psychology cannot explain idiosyncratic differences in individual behavior. Behavior genetic and developmental
psychological theories can better explain such unique individual behavior.
Evolutionary psychology tends to explain the behavior of individuals in rough
categories, such as men and women (sex differences), rich and poor (class
differences), or young and old (age differences) .
While evolutionary psychology proper is a microlevel theory of individual
behavior, it can also explain some emergent phenomena . A key assumption in
the macrolevel application of evolutionary psychology is the methodologically
\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\\
a00
169
170
SATOSHI KANAZAWA
individualist one that the emergent properties at the macro level reflect the
aggregation of individual behavior at the micro level . Evolutionary psychology
can thus explain the emergence of some norms (Kanazawa & Still, 2001) . It is
my contention here that evolutionary psychology can also explain the
emergence of another aggregate phenomenon : Social structures and networks .
I believe that it can address and begin to solve some of the problems of
structuralism and network theory discussed above .
How can evolutionary psychology solve some of the theoretical problems of
structuralism and network theory? First, it can explain why most personal
networks are homophilous on such dimensions as sex, age, and race . One very
important implication of evolutionary psychology is that the human brain is
biased to perceive the environment as if it were still the EEA . Since the basic
architecture of the human brain has not changed since the end of the
Pleistocene epoch about 10,000 years ago, it has difficulty comprehending
elements that emerged in the meantime . This is why most people have innate
phobias of spiders and snakes, many species of which are poisonous and
therefore represented genuine threats to survival in the EEA, but they do not
have phobias of such evolutionarily novel dangers like cars and guns, even
though far more people in contemporary societies die of automobile accidents
and gunshot wounds than of spider or snake bites . The human brain, adapted
to the EEA, functions as if spiders and snakes represent some of the greatest
threats to human survival (Buss, 1999, pp . 62-63) .
Human society in the EEA was more or less egalitarian and there were few
clear differentiations among individuals . Major exceptions to this, however,
were sex and age (Maryanski & Turner, 1992, pp. 78-89) . Human society, just
like primate societies, has always had clear divisions of labor based on age and
sex, and has always been gerontocratic. These features of human society in the
ancestral environment put people into age and sex categories, and our ancestors
mostly associated with others of the same sex and similar ages . And, of course,
all human societies in the ancestral environment were racially and ethnically
homogeneous . Humans also possess a psychological mechanism that makes
them ethnocentric and preferentially associate with others in their deme (a local
population within which people marry endogamously) (Whitmeyer, 1997) . I
contend that individuals have homophilous tendencies today because we have
evolved psychological mechanisms that compel us to associate with others of
the same sex and race and similar age, and these evolved psychological
mechanisms reflect the social organization of human society in the EEA .
Evolutionary psychology, and its assumption of the human brain being
biased to perceive the environment as if it were the EEA, can also explain why
individuals are more homophilous on sex, age, and race than on education and
171
occupation (Lazarsfeld & Merton, 1954 ; McPherson & Smith-Lovin, 1987,

Table 1 ; Verbrugge, 1977) . This is because differentiations among humans
based on the former set of dimensions were meaningful in the EEA, while
similar differentiations based on the latter set were not . The human brain has
more difficulty differentiating others on the basis of their occupation or
educational attainment than on sex, age and race, just like the human brain has
more difficulty comprehending cars and guns as dangers to survival than
comprehending spiders and snakes as such . Thus individuals are unconsciously
more homophilous on primordial dimensions of sex, age and race than on
evolutionarily novel dimensions of occupation and education .
While sex, age, and race are more visible characteristics of individuals than
education or occupation, this is not why we are more homophilous on the
former than on the latter . Many markers of ethnic membership are not visible .
(The interminable civil wars in Somalia during 1990s were fought between
subclans within the same clan within the same tribe within the same race
(Geekie, 1993, p . 11 ; Gregory, 1992, p. 34 ; Sheehan, 1993, p . 41) .) Yet we are
always homophilous on ethnicity . In contrast, many other highly visible
characteristics (such as height or weight) do not as readily form the basis of
homophily . This is because ethnicity (no matter how visible or invisible the
markers) was always an important basis for defining group membership and a
deme in the EEA, while height and weight were not (although the latter
characteristics were important for mate selection and the human brain does
respond to them when selecting a mate) .
AN ILLUSTRATION : WOMEN'S KINCENTRIC

NETWORKS
As a concrete illustration of how evolutionary psychology can explain the
origins of social structures, I will turn to one persistent empirical puzzle in
network theory : Women's kincentric networks . Empirical studies on personal
networks repeatedly demonstrate that otherwise comparable men and women
have similar personal networks . The only exception to this rule is that women
have more kin and fewer coworkers in their personal networks than men do
(Campbell, 1988 ; Fischer & Oliker, 1983 ; Marsden, 1987) . While there appears
little doubt that this sex difference in personal networks exists, few in network
theory seem to know why . Why do women have more kin in their personal
networks than men do?
To my knowledge, Smith-Lovin and McPherson (1993, pp . 233-237 ;
Munch, McPherson & Smith-Lovin, 1997) are the only ones to offer an
explanation of this phenomenon . Using fictitious characters named Jim and
172
SATOSHI KANAZAWA
Jane, they explain how the compositions of their personal networks remain
more or less the same through adult years because "Jim is serious about his
career as an engineer [and] Jane is equally serious about her nursing".
However, the change begins when they become parents . "When their first child
is born, however, Jane's mother comes to visit for two weeks ; Jane begins to
use her sister as a babysitter for daytime care while she is working . . . . Because
more of her time is taken up with the baby, Jane's networks become more
centered on neighborhood and kin, to some extent at the expense of her work
and voluntary association friends . Jim's work and group ties are less altered"
(pp . 234-235) .
Their explanation, however, simply begs the questions : Why is it Jane's
mother who comes to visit after the baby is born, not Jim's (when Jim's mother
is presumably equally related to the baby as Jane's mother)? Or is she? Why
is it Jane's sister who becomes their babysitter, not Jim's sister (when both
sisters are presumably equally related to the baby)? Or are they? Smith-Lovin
and McPherson assume that it is Jane, not Jim, who is the primary caretaker of
the baby. Why is this so?
Evolutionary psychology can answer all of these questions . The fact that the
female gamete (egg) is greater in size and fewer in number than the male
gamete (sperm) (which is the biological definition of male and female), and the
fact that gestation takes place within the female body, together lead, directly or
indirectly, to almost all of the sex differences in preferences and behavior . One
of these differences is parental investment . Across all species for which these
two conditions hold, the female makes greater parental investment than the
male . In fact, for most species, the male parental investment is limited to the
sperm. The sex differences in parental investment occurs because males under
these conditions have far greater fitness ceiling than the females do ; males can
produce a far larger number of offspring in their lifetime than females can .
This is true of humans as well . Thus, while successful reproduction is
equally important to men and women, each child is far more valuable to a
woman than to a man because it represents a greater share of a woman's
lifetime reproductive potential than a man's . Men are exceptional in nature in
that they make a large amount of parental investment into their offspring
(compared to males of other species) . Nonetheless, women (just like females of
most other species) still make far greater parental investment into their children
than men do, because women's evolved psychological mechanisms compel
them to do so .
For these evolutionary reasons, women are more motivated to make parental
investment than men are . However, women cannot always do it alone;
sometimes, they need help from others, especially in the EEA where resources
173
were scarce and life was precarious . When mothers need help in their effort to
raise their children, nobody is more likely or willing to deliver it than their kin .
Women's kin are sometimes even more motivated to invest in the children than
the putative fathers are, because, due to paternity uncertainty (created by the
possibility of cuckoldry), the fathers may or may not be genetically related to
the children, whereas the maternal kin are always genetically related to the
children . For the same reason, paternal kin are not as motivated to invest in the
children as maternal kin are . I contend that this is why women, even today, have
a larger number of kin in their personal networks than men do . Women's
evolved psychological mechanism compels them to make greater parental
investment into the children, and women need to rely on their kin in case they
need help, materially or otherwise .
EMPIRICAL TESTS
I derive two specific empirical hypotheses from this evolutionary psychological
explanation of women's kincentric networks . First, if women maintain their ties
to their kin in case they need help with their parental investment, then women
who are materially better off should need less help from their kin, and therefore
less need to maintain their ties with them. Second, women who are currently
married should need less help from their kin than women without husbands,
because, even with residual paternity uncertainty, the fathers should be
motivated to make some parental investment into the offspring and thereby
lessen the mothers' burden . Women can make less parental investment into
their children if they have their mates present than if they didn't . I emphasize
that women need not make the decisions to have more or less kin in their
networks consciously. When they have more resources or are married, women
may just feel like not keeping in touch with their relatives, without really
knowing why . Women's evolved psychological mechanisms may respond to
external conditions beneath their conscious thinking .
At any rate, if my explanation is correct, then both family income and being
currently married should decrease the extent to which women have kin in their
personal networks . Further, these two variables should not have any effect on
the extent to which men have kin in their personal networks . It seems to me that
there are no other plausible explanations for the negative effects of resources
and marriage on kin network only among women but not among men . I
therefore test these two hypotheses with the 1985 Social Networks module of
the General Social Survey. Note that my precise prediction is that the
independent variables have significant negative effects on the dependent
variable among women, but not among men . My prediction is not the more
174
SATOSHI KANAZAWA
common one of an interaction effect between sex and the independent variables
of interest . A significant sex interaction effect would only demonstrate that the
coefficients for men and women are significantly different from each other. It
does not tell us if the coefficient is significant for women and not significant for
men, as I predict . (The sex interaction effect could be significant, and the
coefficients for men and women could both be significant or both be nonsignificant .) I therefore estimate the equations separately for men and women,
rather than include sex interaction terms .
Dependent variable . I use the measure of kin density as the dependent
variable . This is the proportion of kin among their (up to) five closest
associates, and thus varies from 0 to 1 .0 . Unlike a similar measure used by
Marsden (1987) and others, however, I exclude the spouse from the category of
kin, for two reasons . First, the respondent's current marital status is one of the
predictors in the following tests, so I need a measure of kin density that is
independent of whether or not they have a spouse . Second, and more
importantly, from the evolutionary psychological perspective, and particularly
for my explanation, the spouses do not count as kin because they are not
genetically related to the respondents .
Marsden (1987, p . 129) reports, that, when spouses are included among kin,
women's kin density is 0 .066 higher than men's (0 .580 vs . 514, p < 0 .01) .
When I exclude spouses from kin, the difference increases to 0.1064 (0 .3881
vs . 0 .2817, p < 0 .0001) . Thus the sex differences in kin density of personal
networks is even greater than previously thought .
Independent variables . I use the measures of total family income and
current marital status (1 if currently married) as main predictors of kin density.
I predict significant negative effects of both variables on women's kin density,
but not men's .
Control variables . Since the number of kin that the GSS respondents can
have in their personal networks is in reverse proportion to the number of
coworkers (given that the GSS limited the total number of associates to five),
I need to control for the respondent's work status . I include variables that
measure respondent's status which make it more likely that they have
coworkers in their personal networks : Full-time employment (1 if the
respondent has a full-time job), and occupational prestige (the Hodge-SiegelRossi Prestige Scores) . Since blacks, especially black males, have fewer ties to
their families than others (Marsden, 1987, Table 3), I also control for race (1 if
black) . Finally, in order specifically to test the explanation offered by SmithLovin and McPherson (1993 ; Munch et al ., 1997), I include a measure of
parenthood in the equations (1 if the respondent has had one or more children,
0 if the respondent is childless) . If Smith-Lovin and McPherson are correct,
175
then parenthood measured by this variable should have a significant positive

effect on the kin density of women's personal networks .
Results . Table 1, Columns (1) and (3), indicate that total family income has
a significantly negative zero-order correlation with kin density for both men
and women (p < 0.001 for men, p < 0.0001 for women) . Once I control for
other variables, however, family income no longer has a significant effect on
men's kin density. For women, family income continues to have a significantly
(p < 0 .01) negative effect on kin density, even after I control for their relevant
occupational status and race . Most importantly, a measure of parenthood does
not have a significant effect on kin density of women's personal network (even
though it has a significantly negative effect on men's kin density) . The results
in Table 1 therefore confirm my evolutionary psychological explanation of
Table 1 .
The Effect of Family Income on Kin Density .

Women
Predictor :
Family income
Men
(1)
(2)
(3)
-0 .0199****
(0 .0041)
-0.0130**
(0.0048)
-0 .0178***
(0 .0050)
(4)
-0.0080
(0.0058)
Controls :
Full-time job
-0 .0701**
(0 .0270)
-0.0674*
(0.0301)
Occupational
prestige
-0.0013
(0.0010)
-0.0024
(8 .7194-4)
Race
0 .0155
(0 .0451)
-0 .1776***
(0 .0461)
Parenthood
0 .0330
(0.0306)
-0.1069***
(0.0276)
Constant
Number of cases
R2
0 .5760
(0.0396)
0 .4653
(0.0522)
0.5995
(0 .0599)
770
706
648
634
0.0304
0 .0411
0.0195
0 .0771
Note : Standard errors are in parentheses .
*p < 0 .05
**p < 0 .01
***p < 0 .001
****p < 0 .0001
0.5612
(0 .0586)
176
SATOSHI KANAZAWA
women's kincentric networks and disconfirms Smith-Lovin and McPherson's

(1993) .
The pattern is virtually identical in Table 2 . Once again, being currently
married has a significantly negative zero-order correlation with kin density for
both men and women (p < 0 . 01 for women, p < 0 .00 1 for men) . Once again,
however, being currently married has no significant effect on men's kin density
once I include other variables in the equation . For women, being currently
married continues to have a significantly (p < 0 .01) negative effect on kin
density of personal networks even after I include all the control variables .
Parenthood once again has no significant effect on women's kin density (and a
significantly negative effect on men's) . The results in Tables 1 and 2 taken
together seem to suggest that women have more kin in their personal networks
Table 2. The Effect of Being Currently Married on Kin Density .

Women
Predictor :
Currently married
(2)
(3)
-0 .0758**
(0 .0242)
-0 .0784**
(0 .0258)
-0 .1026***
(0 .0262)
-0.0495
(0.0300)
0 .4273
(0 .0174)
844
0 .0115
-0 .0789**
(0 .0259)
-0 .0016
(9 .2555)
-0 .0352
(0 .0421)
0 .0538
(0 .0299)
0 .4806
(0 .0474)
774
0 .0361
0 .3464
(0 .0208)
687
0 .0220
-0.0810**
(0.0269)
-0.0027**
(8.3522)
-0.1649***
(0.0439)
-0.0864**
(0.0304)
0 .5512
(0.0427)
672
0.0761
Controls :
Full-time job
Occupational
prestige
Race
Parenthood
Constant
Number of cases
R2
Men
(1)
Note : Standard errors are in parentheses .

*p < 0.05
**p < 0.01
***p < 0 .001
****p < 0 .0001
(4)
177
because they need their kin's help in their effort to make parental investment
into their children .
Critics might argue, however, that the non-significant effect of parenthood on
women's kin density disconfirms my evolutionary psychological explanation
(as well as Smith-Lovin and McPherson's) . If, as I argue, women need their kin
to help raise their children, why doesn't being a parent have an effect on their
kin density?
This is because the human brain has difficulty making facultative choices in
response to situations that did not exist in the EEA . Take the example of our
preference for sweets and fats (Barash, 1982, pp . 144-147) . We have an
evolved psychological mechanism that compels us to consume sweet and fatty
foods because, in the nutritionally deficient EEA, those who had a taste for and
consumed more such foods (which contain higher calories) survived better and
had greater reproductive success than those who didn't have such a taste . This
psychological mechanism, however, does not respond facultatively to different
nutritional conditions . It does not say "If you are malnourished and can use
extra calories, then consume as many sweet and fatty foods as you can get your
hands on. If, however, you are not malnourished, then do not consume such
foods" . It does not say so because the second contingency ("if you are not
malnourished") never existed in the EEA ; our ancestors were always on the
verge of malnutrition . This psychological mechanism does not allow us to
make facultative choices in response to different conditions, and that is why we
have a constant craving for sweet and fatty foods regardless of our current
nutritional condition . We have a taste for such foods even though few of us are
malnourished today, and we get obese as a result.
Similarly, parenthood was a constant in the EEA . Given that humans in the
EEA were mildly polygynous (Alexander et al ., 1979 ; Leutenegger & Kelly,
1977), there were many men who did not have any mates and therefore did not
reproduce at all, but we are not descended from these men . And almost all
women had mates and reproduced in the EEA . Given the absence of reliable
means of birth control, parenthood was inevitable for anyone with mates
(which included most adult women) . Thus our evolved psychological
mechanisms do not allow us to make facultative choices on the basis of our
parental status ("If you have children, do X ; if you don't have children, do Y"),
because, once again, the second contingency ("if you don't have children")
never held true for our ancestors from whom we are descended and inherited
our psychological mechanisms . This is why all women, regardless of their
current parental status, are compelled to maintain ties with their kin in
preparation for making parental investment into their offspring . In contrast,
even in the EEA, some women and families were materially better off than
178
SATOSHI KANAZAWA
others, and some women were "married" while the husbands of others have
either died or left them. Thus the human brain, adapted to the EEA, can make
the facultative decisions such as "if you're poor, rely on your kin ; if you are
rich, do not rely on your kin" or "if you do not have a mate present, rely on
your kin ; if you have a mate present, do not rely on your kin ."
DISCUSSION
In this paper I have argued that evolutionary psychology can provide an
explanation for the origins of social structures and networks . I have constructed
one evolutionary psychological explanation for why women have more kin in
their personal networks than men, and have tested two hypotheses drawn from
the explanation with the 1985 Social Networks module of the General Social
Survey . The data provide support for the view that women maintain strong ties
to their kin because they may need help in their parental investment into the
offspring . Both total family income and being currently married have
significantly negative effects on kin density of women's personal networks,
while they have no significant effects in men's personal networks .
The case for evolutionary psychology's utility for structuralism and network
theory is far from solid, however, and I will need to subject more evolutionary
psychological theories of the origins of social structures to rigorous empirical
tests . Unfortunately, it is impossible to demonstrate the empirical validity of
such theories with the current networks data . They simply do not make fine
enough distinctions among kin to test further evolutionary psychological
hypotheses . All existing networks data recognize very rough categories of kin .
For instance, the 1985 Social Networks module of the General Social Survey,
which I use in this paper, only recognizes parents, siblings, children, and other
family members as categories of kin . The 1986 ISSP module on Social Support
and Networks does slightly better and recognizes mothers, fathers, sisters,
brothers, daughters, sons, grandparents, grandchildren, aunts and uncles (one
category), and other relatives . This is understandable since, without being
informed by modern evolutionary psychology, it is natural for social scientists
to assume that all grandparents are the same and aunts and uncles are the
same .
From the perspective of evolutionary psychology, however, even the fine
distinctions among kin that the 1986 ISSP module on Social Support and
Networks makes are not sufficient to test its hypotheses . For instance, one
would need to know whether the grandparents are maternal grandmother,
179
maternal grandfather, paternal grandmother, or paternal grandfather . We would

also need to know whether the aunts and uncles are maternal or paternal . These
minute distinctions make a difference for an evolutionary psychological theory
of kin networks .
For example, the four grandparents are very different from the evolutionary
psychological perspective. Maternal grandmothers, being the mother of the
mother of the grandchildren, are certain to be genetically related to them,
because there are no men (and thus paternity uncertainty) involved in that
branch of the family tree . Both maternal grandfathers and paternal grandmothers have one male link, and therefore one possibility of cuckoldry and of
not being related to the grandchildren . Paternal grandfathers, on the other hand,
have two male links and therefore two possibilities of cuckoldry and of not
being related to the grandchildren . Therefore, in terms of the possibility of
being related to the grandchildren, the following relationship holds : maternal
grandmother > maternal grandfather= paternal grandmother > paternal
grandfather. There is evidence that the amount of grief that grandparents
experience after the death of a child follows this precise pattern : Maternal
grandmothers mourn more than either maternal grandfathers or paternal
grandmothers, who in turn mourn more than paternal grandfathers (Littlefield
& Rushton, 1986) . I would therefore predict that, of the four grandparents,
women are most likely to have maternal grandmothers in their personal
networks, followed by maternal grandfather and paternal grandmother, and
women are least likely to have paternal grandfather in their personal networks .
What is significant about this prediction is that one would be able to adjudicate
between evolutionary psychological explanation of women's kincentric
networks and that based on pure homophily, because the latter would predict
that women are more likely to have either maternal or paternal grandmothers in
their personal networks than either maternal or paternal grandfathers .
A similar empirical test is possible with respect to aunts and uncles, or nieces
and nephews . An evolutionary psychological explanation would predict that
women are more likely to have maternal aunts and uncles, and maternal nieces
and nephews, in their personal networks than their paternal counterparts . An
explanation based on homophily would predict that women are more likely to
have aunts and nieces on either side in their personal networks than uncles and
nephews on either side . Of course, these empirical tests are currently
impossible to conduct because there exist no networks data that make such fine
distinctions among kin . I call for network theorists and researchers to take
evolutionary psychology seriously, and collect personal network data that make
finer distinctions among kin .
180
SATOSHI KANAZAWA
NOTES
1 . I owe this insight to Heather A . Haveman .
2. Carley (1991) argues that individuals who share the same information are more
likely to interact with each other than people who do not share the same information .
"Individuals may be more "comfortable" interacting with someone with whom they
have much in common, individuals may avoid "costs" because information exchanges
may be more efficient between similar individuals, or individuals may acquire
"rewards" because common knowledge may produce more opportunities for interaction" (p . 334) . Given the empirical fact of strong homophily on sex and ethnicity,
however, Carley must explain how and why individuals of the same sex or ethnicity
come to possess the same information in the first place, prior to and in the absence of
homophilous interaction.
3 . I owe this insight to Eliana Friedman Hechter .
4 . Of course, a more damning criticism of Mark's (1998b) study is that it really does
not test a structural theory of musical taste acquisition. All he demonstrates is that
different types of music have different niches, defined by the modal characteristics of
their fans, and those who are in the niche are more likely to develop a taste for the music
than those who are outside it (the niche hypothesis), and those who are inside the niche,
even when they are not fans, are more likely to know about the music than those who
are outside the niche (the familiarity hypothesis) . (Mark does derive and test other
hypotheses from his theory, however.) From these empirical patterns, Mark concludes
that individuals acquire their musical tastes from others through homophilous network
ties . However, this conclusion is not at all necessary . Perhaps a deliberately absurd
example will help make my point.
If I plot the age distribution of women who are diagnosed with breast cancer, I will
probably get a unimodal distribution, with the modal age around 50 . I will thus define
"the niche" for breast cancer as women between the ages of 45 and 55 . I will then
demonstrate that actors who are in the niche (for instance, a 47-year-old woman) are
statistically more likely to have breast cancer than those who are outside the niche (a
21-year-old woman), supporting the niche hypothesis of breast cancer acquisition .
Because personal networks are homophilous, middle-aged women are likely to have
other middle-aged women as friends, and young women are likely to have other young
women as friends . Then those who are in the niche, even when they don't have breast
cancer themselves, are more likely to know someone who does than those who are
outside the niche, supporting the familiarity hypothesis of breast cancer acquisition . I
have therefore demonstrated that women acquire breast cancer through their
homophilous ties to others in their networks .
I concur with Hofstadter (Forthcoming), Orians and Heerwagen (1992), Thornhill
(1998), and Wilson (1998, Chapter 10) that what we find esthetically pleasing in art,
literature, and music has evolutionary psychological origins and can be explained by
neuroscience . It would therefore be not at all be surprising if our tastes in music, and
how they change over time, have an evolutionary psychological basis. If most of us
become politically less liberal and more conservative as we age, and if this lifecourse
pattern has evolutionary causes (Kanazawa, 2001), it would be equally plausible that we
like loud music (or music that our parents hate) when we are young and quiet music (or
music that our parents like) when we are old for the same reasons .
181
5 . A gene that predisposes men toward homosexuality has been located at Xq28 (X
chromosome, arm q, location 28), even though the gene itself has not been identified
(Hamer et al ., 1993) . The same location, however, does not seem to influence women's
sexual orientation (Hu et al ., 1995) . Given that women's sexual orientation tends to be
more flexible than men's, it is entirely possible that no such genes exist for women
(Blum, 1997, pp . 127-157) .
6. I agree with Blau (1977b, p . 28, emphases added) wholeheartedly : "Of course,
there can be no doubt that technological and economic conditions, cultural values, and
psychological motives influence human behavior and hence social relations . This is not
at issue . Granted the existence of these influences, the question raised is what
independent influences the structure of social positions in a society or community exerts
on social relations" .
7 . For instance, Mark's (1998a) theory of information and social structure can
explain the emergence of social structures from undifferentiated systems . One of the
integral assumptions of his model, however, is the similarity assumption : "Individuals
with similar information are more likely to interact than are individuals with dissimilar
information" (Mark, 1998a, p . 312) . In other words, individuals in his model must : a)
be different in the information they possess ; b) be able to differentiate others on the
basis of the information they possess ; and c) act on this differentiation .
ACKNOWLEDGMENTS
I thank Paula England, Michael Hechter, Peter J . Katzenstein, Noah Mark,
Alexandra R . Maryanski, Alan S . Miller, David Strang, colloquium participants
at the Department of Sociology, Cornell University, and the Population Studies
Center, University of Pennsylvania, and Editors for their comments on earlier
drafts . I dedicate this paper to Phyllis Sinclair, Mary C . Still, Cara mia Kelly,
Michele E . Pinkow, and all the other women who over the years bugged me
about my own kin network (or lack thereof) . Their remarkably consistent and
persistent nagging provided me with the insight behind the hypothesis
regarding women and kin networks proposed herein . Direct all correspondence
to : Satoshi Kanazawa, Department of Sociology, Indiana University of
Pennsylvania, Indiana, PA 15705-1087 . Email : Kanazawa@grove .iup .edu .
REFERENCES
Alexander, R. D., Hoogland, J. L ., Howard, R . D ., Noonan, K. M ., & Sherman, P. W. (1979).
Sexual Dimorphisms and Breeding Systems in Pinnipeds, Ungulates, Primates and
Humans . In : N . A. Chagnon & W. Irons (Eds), Evolutionary Biology and Human Social
Behavior: An Anthropological Perspective (pp. 402-435) . North Scituate : Duxbury Press .
Barash, D . P. (1982) . Sociobiology and Behavior, Second Edition. New York : Elsevier.
Barkow, J . H., Cosmides, L ., & Tooby, J . (Eds) (1992). The Adapted Mind: Evolutionary
Psychology and the Generation of Culture. New York: Oxford University Press .
Blau, P. M . (1977a) . Inequality and Heterogeneity: A Primitive Theory of Social Structure . New
York : Free Press .
182
SATOSHI KANAZAWA
Blau, P. M. (1977b) . A Macrosociological Theory of Social Structure . American Journal of

Sociology, 83, 26-54 .
Blau, P. M. (1989). Structures of Social Positions and Structures of Social Relations . In : J. H.
Turner (Ed.), Theory Building in Sociology (pp . 43-59) . Newbury Park : Sage .
Blum, D. (1997). Sex on the Brain : The Biological Differences Between Men and Women . New
York: Penguin .
Bowlby, J . (1969). Attachment and Loss . Volume 1 : Attachment. New York: Basic .
Buss, D . M . (1999). Evolutionary Psychology : The New Science of the Mind . Boston : Allyn and
Bacon .
Campbell, K. E. (1988) . Gender Differences in Job-Related Networks . Work and Occupations, 15,
179-200 .
Carley, K. (1991) . A Theory of Group Stability . American Sociological Review, 56, 331-354 .
Feld, S . L. (1982) . Social Structural Determinants of Similarity Among Associates . American
Fischer, C . S . (1982) . To Dwell Among Friends: Personal Networks in Town and City. Chicago :
University of Chicago Press.
Fischer, C . S ., & Oliker, S . J. (1983) . A Research Note on Friendship, Gender, and the Life Cycle .
Social Forces, 62, 124-133 .
Geekie, R. (1993). U .S . Commits Force to Somalia, But For How Long? Africa Report, 38(1),
5-11 .
Gregory, S . S . (1992) . How Somalia Crumbled : Clan Warfare and a Glut of Weapons Have
Plunged the Country into Anarchy . Time, 140(24), 34 .
Hamer, D . H ., Hu, S ., Magnuson, V. L., & Pattatucci, A . M. L . (1993) . A Linkage Between DNA
Markers on the X Chromosome and Male Sexual Orientation . Science, 261, 321-327 .
Hofstadter, D. R . (Forthcoming) . Staring EMI Straight in the Eye - and Doing My Best Not to
Flinch. In : D . Cope (Ed .), Virtual Music: Computer Synthesis of Music Style . Cambridge :
MIT Press .
Hu, S ., Pattatucci, A . M. L ., Patterson, C ., Li, L ., Fulker, D. W., Cherny, S . S ., Kruglyak, L ., &
Hamer, D. H. (1995) . Linkage Between Sexual Orientation and Chromosome Xq28 in
Males but not in Females . Nature Genetics, 11, 248-256 .
Kanazawa, S . (2001) . De Gustibus Est Disputandum . Social Forces, 79, Forthcoming .
Kanazawa, S ., & Still, M. C . (1999). Why Monogamy? Social Forces, 78, 25-50.
Kanazawa, S ., & Still, M . C . (2000) . Why Men Commit Crimes (and Why They Desist) .
Sociological Theory, 18, 434-447.
Kanazawa, S ., & Still, M . C . (2001). The Emergence of Marriage Norms : An Evolutionary
Psychological Perspective . In : M . Hechter & K .-D . Opp (Eds), Social Norms . New York :
Russell Sage Foundation .
Lazarsfeld, P. F., & Merton, R . K . (1954) . Friendship as Social Process: A Substantive and
Methodological Analysis . In : M . Berger, T . Abel & C. H. Page (Eds), Freedom and Control
in Modern Society (pp . 18-66) . New York : Octagon .
Leutenegger, W., & Kelly, J . (1977) . Relationship of Sexual Dimorphism in Canine Size and Body
Size to Social, Behavioral and Ecological Correlates in Anthropoid Primates . Primates, 18,
117-136 .
Littlefield, C . H ., & Rushton, J . P. (1986) . When a Child Dies : The Sociobiology of Bereavement .
Journal of Personality and Social Psychology, 51, 797-802.
Lorrain, F. P & White, H. C . (1971). Structural Equivalence of Individuals in Social Networks .
Journal of Mathematical Sociology, 1, 49-80 .
183
Mark, N. (1998a) . Beyond Individual Differences : Social Differentiation from First Principles .
Mark, N. (1998b) . Birds of a Feather Sing Together. Social Forces, 77, 453-485 .
Marsden, P. V. (1987). Core Discussion Networks of Americans . American Sociological Review,
52,122-131 .
Maryanski, A ., & Turner, J . H. (1992). The Social Cage : Human Nature and the Evolution of
Society . Stanford: Stanford University Press .
Mayhew, B . H. (1980). Structuralism Versus Individualism : Part I, Shadowboxing in the Dark .
Social Forces, 59, 335-375, 461 .
Mayhew, B . H . (1981) . Structuralism Versus Individualism: Part II, Ideological and Other
Obfuscations . Social Forces, 59, 627-648, 461 .
McPherson, J . M ., & Smith-Lovin, L . (1982) . Women and Weak Ties : Differences by Sex in the
Size of Voluntary Organizations . American Journal of Sociology, 87, 883-904 .
McPherson, J . M ., & Smith-Lovin, L. (1987) . Homophily in Voluntary Organizations : Status
Distance and the Composition of Face-to-Face Groups . American Sociological Review, 52,
370-379 .
Merton, R . K . (1972) . Insiders and Outsiders : A Chapter in the Sociology of Knowledge . American
Journal of Sociology, 78, 9-47 .
Munch, A ., McPherson, J . M., & Smith-Lovin, L . (1997) . Gender, Children, and Social Contact :
The Effects of Childrearing for Men and Women . American Sociological Review, 62,
509-520 .
Orians, G . H ., & Heerwagen, J. H . (1992) . Evolved Responses to Landscapes . In : J. H . Barkow,
L. Cosmides & J. Tooby (Eds), The Adapted Min & Evolutionary Psychology and the
Generation of Culture (pp . 555-579) . New York : Oxford University Press .
Sailer, L . D . (1978) . Structural Equivalence : Meaning and Definition, Computation and
Application . Social Networks, 1, 73-90 .
Sheehan, E . R . F. (1993) . In the Heart of Somalia . New York Review of Books . January 14 :
38-43 .
Smith-Lovin, L ., & McPherson, J . M . (1993). You Are Who You Know : A Network Approach to
Gender. In : P. England (Ed.), Theory on Gender/Feminism on Theory (pp . 223-251) . New
York : Aldine .
Stigler, G . J ., & Becker, G. S . (1977) . De Gustibus Non Est Disputandum . American Economic
Review, 67,76-90 .
Thornhill, R. (1998) . Darwinian Aesthetics . In : C . Crawford & D . L . Krebs (Eds), Handbook of
Evolutionary Psychology : Ideas, Issues, and Applications (pp . 543-572) . Mahwah:
Lawrence Erlbaum Associates .
Verbrugge, L . M . (1977). The Structure of Adult Friendship Choices . Social Forces, 56,
576-597 .
van den Berghe, P . L. (1990). From the Popocatepetl to the Limpopo . In : B . M. Berger (Ed .),
Authors of Their Own Lives : Intellectual Autobiographies by Twenty American Sociologists
(pp . 410-431) . Berkeley : University of California Press .
Wasserman, S ., & Faust, K. (1994) . Social Network Analysis : Methods and Applications .
Cambridge : Cambridge University Press .
Whitmeyer, J . M . (1997) . Endogamy as a Basis for Ethnic Behavior. Sociological Theory, 15,
162-178 .
Willer, D . (Ed.) (1999) . Network Exchange Theory . Westport : Praeger.
Wilson, E. O . (1998) . Consilience: The Unity of Knowledge . New York : Knopf.
PROCESSING PERFORMANCE
EVALUATIONS IN HOMOGENEOUS
TASK GROUPS : FEEDBACK AND
GENDER EFFECTS
Martha Foschi, Sandra Enns and Vanessa Lapointe
ABSTRACT
This experiment investigates the development of performance expectations
in status-homogeneous task groups. The issue of central interest is
whether or not gender is a factor in expectation formation when group
members are of the same sex and work on a gender-neutral task. Male and
female undergraduates, participating in same-sex dyads, worked first
individually and then as a team on a novel, visual perception task . Apart
from the feedback they received from the experimenter at the end of the
individual performance phase, participants were given no information
with which to form differentiated expectations about self and partner for
this task We investigate effects from both sex of dyad and level of feedback
on: (a) influence behavior during the team phase, and (b) selected
variables obtained through self-reports. Rejection of influence data show
statistically significant effects from feedback only . Self-reports, on the
other hand, reveal significant results from both feedback and sex of subject
across several assessments of self's and partner's competence, and from
sex of subject in some of the items concerning perceptions of the
experimenter's status . Different conditions for the emergence of such
Copyright 2001 by Elsevier Science Ltd .
ISBN: 0-7623-0767-6
185
186
MARTHA FOSCHI, SANDRA ENNS AND VANESSA LAPOINTE
effects in behaviors as opposed to self-reports are identified and

discussed.
THEORETICAL BACKGROUND AND OBJECTIVES

Many everyday interactions involve participation in groups where members are
charged with completing a collective task . Some examples are: a jury ; a school
board making budget allocations ; students assigned to work as a team on an
academic project ; a wilderness rescue unit . A key decision that members of
such groups must arrive at (explicitly or not) concerns how much competence
each person has for the task at hand . If group members have previous
experience working together on this task or similar ones, they will use that
experience to guide their assessments of each other's competence. However,
their situation will be quite different if they have no previous acquaintance with
either each other or the task . In both of these cases, how individuals form ideas
about their relative levels of task competence will depend on whether they
perceive the group to be either homogeneous or heterogeneous on key
background variables (such as gender, skin color, and level of education) .
Expectation states theory (Berger et al ., 1977, 1985 ; Wagner & Berger, 1993 ;
Webster & Foschi, 1988) presents a comprehensive account of how task ability
is assigned in both types of groups . The theory has a long-standing tradition
and has received strong empirical support ; it is also the background for the
present study.
Two core concepts in the expectation states program are those of "status
characteristic" and "performance expectations" . The former is defined as any
valued attribute implying task competence . Such characteristics consist of at
least two levels or states (e .g. either high or low mechanical ability), one of
which is seen as worthier than the other. "Performance expectations" are beliefs
about the likely quality of group members' future performances on the task at
hand, and reflect levels of assigned competence . Status characteristics are
classified as ranging from specific to diffuse, depending on their perceived
applicability. A specific characteristic is associated with performance expectations in a limited area ; a diffuse characteristic carries expectations about
performance on a wide, indeterminate set of tasks . In many societies, gender,
ethnicity, and socio-economic class, for instance, constitute diffuse status
characteristics for large numbers of individuals . Expectation states are said to
develop for "self' (the focal actor) relative to each other member of the group ;
all propositions are formulated from self's point of view. The theory specifies
how and under what conditions expectations are formed ; for example, these
Processing Performance Evaluations In Homogeneous Task Groups
1 87
may be based on either the group members' status characteristics, or the

evaluations they receive on their actual task performances, or both .
In this article we are interested in those situations where group members
have no previous acquaintance with either each other or the task, are
homogeneous with respect to variables defining their status outside the group
(i .e . "external" factors such as gender and age), and perceive no linkage
between those factors and the task . For instance, consider the case where all
group members are women who are meeting for the first time . Their
impressions indicate that they are similar in age, ethnicity, and level of
education, they have no prior experience with their collective task, and they
perceive it to be unrelated to any of those four factors that make the group
homogeneous . Under those conditions, how do they form beliefs about each
other's ability? According to expectation states theory, group members in such
settings develop performance expectations on the basis of the feedback
received, namely, from either : (i) the evaluations they make of each other's
task-related acts (i .e . comments, suggestions, interruptions, silences, offers of
performance opportunities) as their interaction proceeds, or (ii) the evaluations
made by a source outside the group (see, for example, Wagner & Berger, 1993) .
A "source" is defined as a person (or persons) who is (are) accepted by a group
member as more capable of evaluating performances than is the group member
him or herself . The research presented here concerns evaluations from a source
and addresses the question of whether there are still effects from external status
under the conditions specified above - conditions which severely limit the
likelihood of such effects . More specifically, we focus on whether gender is a
factor in expectation formation in same-sex groups working on a genderneutral task, and investigate this issue through both behaviors and self-reports .
Homogeneous Groups, Gender, and Evaluations from a Source
According to the theory, once performance expectations have been developed,

they have a strong effect on the interaction that follows : they affect the granting
of opportunities to act, the rate at which these are accepted, the type of
evaluations made, and the amount of influence exerted . These four variables,
which are related to each other in that order, comprise the "power and prestige
order of the group" (Berger et al ., 1977) .
Researchers working in this tradition have developed a standard experimental setting through which the relationship between performance
expectations and influence can be investigated under different conditions (see
Webster & Sobieszek, 1974 : App . 1) . The setting allows for a wide variety of
hypotheses to be tested while it facilitates comparability, and thus increased
188
cumulativeness, across experiments. Next we describe the version of this

setting that has been most commonly used to study homogeneous groups
receiving evaluations from a source ; we also outline the scope conditions most
often included to specify the boundaries of the hypotheses under investigation
(see also Berger et al., 1977 : Ch. 3) .
The task setting is as follows . Two persons, self and a partner (or "other")
work, first individually and then as a team, to solve a series of visual perception
problems ; they have no previous acquaintance with either each other or the
task . Further, they have no differentiating information about each other except
the evaluations they receive from the source at the end of the individual phase .
Instructions describe the ability required for the task as valuable, motivate the
participants to do the task well (i .e . to be "task-oriented"), and encourage them
to take each other's ideas into account during the team phase (i .e . to be
"collectively oriented") . Most trials in that phase involve experimentally
created disagreements between the two persons regarding the correct answer,
and each participant must resolve every disagreement by either accepting or
rejecting the partner's choice . This series of decisions constitutes the measure
of rejection of influence, a variable which, as stated above, is directly
associated with performance expectations and, in turn, with assigned
competence .
Research conducted using this setting has investigated properties of the
group members, the source, and the evaluations, and has tested for the effects
of these variables on expectations. Thus, the work has examined the source's
task expertise and diffuse status characteristics relative to the subject's, and the
extent of the source's legitimation . The evaluations, in turn, have been either
equal or different for self and partner, and have either met or not met the
standards for ability and lack of ability set by the source . This work is reported
in the following articles which comprise a total of nineteen experiments : Berger
and Conner (1969) ; Berger and Fisek (1970) ; Berger et al . (1976) ; Camilleri
and Berger (1967 - 2 studies) ; Camilleri et al . (1972 - equal control
conditions) ; Conner (1977) ; Foschi and Freeman (1991) ; Foschi et al . (1985) ;
Ilardi and McMahon (1988 - Study 2 only) ; Martin and Sell (1980) ; Norman
et al. (1988) ; Parcel and Cook (1977 - 2 studies) ; Sobieszek (1972) ; Sobieszek
and Webster (1973) ; and Webster (1969 ; 1977 - 2 studies) . Results support the
authors' various predictions ; notably, in all cases substantial effects due to
feedback level were found .'
In all of these experiments, subjects participated in dyads undifferentiated by
external status . In particular, subjects were either explicitly or implicitly led to
believe that the partner was of the same sex as themselves . (When referring to
same-sex dyads we use, throughout this article, the expressions "sex of subject"
1 89
and "sex of dyad" interchangeably.) The experimental task, in turn, was either
left undefined in relation to gender, or explicitly dissociated from it . Seven of
these experiments included only male dyads, six had only female dyads, and
six included both types - this latter group is the one of special interest here . Of
those six, Camilleri and Berger (1967) (Study 2), Camilleri et al . (1972), and
Conner (1977) show no significant sex-of-dyad differences in rejection of
influence, Foschi et al . (1985) and Ilardi and McMahon (1988) report some
marked (although not statistically significant) such differences, and Foschi and
Freeman (1991) report significant sex-of-dyad differences . We note, however,
that only Foschi and Freeman (1991) and Ilardi and McMahon (1988) were
specifically designed to investigate sex-of-dyad effects . Thus, these are the only
studies that both analyze the data by this factor and include a sufficiently large
number of subjects per cell in this respect (i .e . at least fifteen, as it is generally
the case in this type of research) . Note also that neither study has a complete
design regarding level of feedback .
Bases for Differentiation Between Same-Sex Groups
Expectation states theory specifies that status effects on beliefs about
competence will occur only when a status factor has become "activated" (i .e.
salient) . Activation occurs either when there is a status difference between self
and other (e .g . one is a man and the other is a woman), or the task is linked to
status (e .g. it is perceived to be masculine), or both (see Berger et al ., 1977 ;
Shelly & Munroe, 1999 ; Wagner & Berger, 1997). Additional conditions for the
activation of gender have also been advanced . These involve characteristics of
the task groups and of the larger social context within which they function
(Fennell et al ., 1978 ; Foschi & Freeman, 1991 ; Ridgeway, 1988) . For example,
Ridgeway (1988) proposes that gender differences between participants and the
organizational authority that charges the group with its task could result in the
activation of gender.
We propose that there are other bases for sex differences in the expectations
formed by members of male and female dyads working on a gender-neutral
task and receiving evaluations from a source, as follows : (i) Such differences
could result if one of these groups meets a given scope condition of the theory
while the other does not (e .g . if the women are collectively oriented but the men
are not), or if the two sexes meet that condition to substantially different
extents . None of these possibilities is anticipated by expectation states theory .
(ii) Men and women in same-sex dyads could also differ in task-related factors
other than those included in the scope conditions, and that could, in turn, affect
rejection of influence . By "task-related factors" we refer to any variable that
190
provides either direct or indirect information about self's and other's task
competence in that setting .' For example, the women's standards for ability
could be more lenient than the men's .
Foschi and Freeman (1991) deal more extensively with such bases than any
of the other five studies identified in this group . Thus, that experiment includes
subjects' reports on task importance, seriousness about doing it well,
motivation, and control over performances, and these data were used to
investigate a possible association between each of these factors and the
obtained sex differences in influence behavior. No such association was found,
although, in each condition, results on reported ability in self and other
matched rejection of influence rates . More specifically : the women differentiated, as instructed, between feedback indicating either clearly or
ambiguously that the partner had more ability than themselves ; the men, on the
other hand, treated the information in both conditions as equally ambiguous
indications of the partner's superiority . The authors propose that these results
may be explained by sex differences in the status assigned to the source and/or
in the number of performances required for a convincing inference of ability .
The study, however, was not designed to provide data on either factor .
In this article we focus on three sets of key task-related variables . These
concern self's : (a) own task and performance requirements for the inference of
ability, (b) assessment of the partner's task choices and overall competence
relative to self, and (c) assessment of the experimenter's competence . Sex
differences in these factors, as well as in meeting scope conditions, could be
due to characteristics of the particular men and women in the sample studied
(see Related Issues below), or could be the result of additional conditions for
the activation of gender . In either case, whether men and women are similar or
different in these respects needs to be examined for a fuller understanding of
how they form expectations .
Research Questions
As discussed above, expectation states theory specifies that gender will be
activated if either the group members are differentiated by sex, or the task is
gender-linked, or both. The theory, however, makes no predictions regarding
activation of gender in the case of same-sex dyads working on a gender-neutral
task (see, for example, Ridgeway, 1988 ; Wagner & Berger, 1997) . Moreover,
there is only a limited number of studies examining how such dyads form
expectations from evaluations from a source, and the evidence from these
studies is not conclusive regarding sex effects, neither in terms of results nor in
the number of subjects included . Finally, the additional activation conditions
191
mentioned earlier have not been studied in detail, and the same is true of
additional factors that could be associated with gender in that setting . Thus,
further investigation of this topic is necessary .
We carry out a study systematically varying only the two independent
variables of main interest in this area, namely, sex of dyad and level of
feedback . The setting under consideration is designed to meet the following
scope conditions : self (i) values the ability required for the task ; (ii) is taskoriented ; (iii) is collectively oriented ; (iv) accepts the evaluations received from
the source; (v) receives no other information with which to form either higher
or lower expectations relative to the partner ; and (vi) believes that no
association has been established between any external status characteristic and
the task.
We also identify three sets of factors that we consider of central importance
in expectation formation in such a setting . We do not vary them experimentally
but rather, leave room for subjects to report their own values . We then assess
whether or not these factors vary by sex of subject (as well as level of feedback)
and, if so, whether the variation is sufficient to affect expectations .
(a) Task and performance requirements. These involve self's own standards
for ability and lack of ability (both for self and the partner) . Research has
shown that competence standards affect how performance evaluations are
processed (Foschi, 1996 ; Foschi & Freeman, 1991 ; Foschi et al ., 1985) . For
example, a person setting a demanding standard for ability (e .g . 85% or more
correct responses) will not infer as much competence from a performance of
75% correct responses as a person setting a more lenient standard (e .g . 65% or
more correct) . We propose that, for a more thorough understanding of the role
of standards in ability assignment, they should be considered in relation to a
specified number of performances - rather than in absolute terms. For example,
we expect that a standard of 85% applied to 100 performances will result in a
more definite inference regarding a person's ability than the same standard
applied to only 20. The perceived validity of the test used to assess competence,
as well as self's perceptions of task difficulty, are also of related interest here .
(b) Partner's task choices and overall competence relative to self. We
investigate several variables concerning self's views of the partner's choices,
namely, confidence in them, their perceived correctness and importance to self,
and impressions of the partner's seriousness in making them . In addition, we
ask directly about impressions of the two persons' overall competence, as
several expectation states studies show this variable to be correlated with
rejection of influence (Foddy & Smithson, 1999 ; Foschi, 1996, Foschi &
Freeman, 1991 ; Foschi et al ., 1985 ; but see also Lockheed et al ., 1983 ; Riches
& Foddy, 1989 for studies reporting no such correlation) . Note that self-reports
192
on assignment of competence are different from expectations . In line with most

of the literature in this area, we view expectation states as a not-directlyobservable, theoretical construct . Self-reports on competence, on the other
hand, are a person's account of those expectations, and are observable by
definition (see also Related Issues below) .
(c) Experimenter's competence . Evaluations may have a different impact on
expectations depending on the competence status accorded to the source . Thus,
evaluations from a high-status source are a clearer determinant of expectations
than those from a low-status source (Crundall & Foddy, 1981 ; Sobieszek, 1972 ;
Webster, 1969 ; Webster & Sobieszek, 1974) . In the present study, the
experimenter communicates test results but does not generate them ; the source
of evaluations is then the entire team behind the project . However, since the
experimenter is the most prominent member of that team, we focus on
perceptions of this person . We investigate the status accorded the experimenter
in terms of perceived knowledge, qualifications, and level of education . We also
ask about perceptions of the experimenter's overall standing relative to self .
Although other expectation-states research has investigated standards for
both ability and lack of ability, perceptions of task difficulty, and impressions
of self's and other's competence, it has not examined these variables in
connection with male and female dyads receiving different levels of evaluations
from a source. Moreover, to our knowledge, none of the other variables in the
three groups we identify has been studied before in this research tradition .
In sum, our approach is as follows : (1) Let us assume that, at every level of
feedback, male and female participants meet the specified scope conditions to
similar extents, and that they are also similar with respect to the other factors
we study. This situation, we propose, constitutes a set of sufficient conditions
to predict that expectations will be organized by the feedback from the source,
and that there will be no sex-of-dyad effects . Specifically, we predict that the
larger the difference in favor of self in the evaluations received, the higher
self's level of rejection of influence will be . We thus extend expectation states
theory by offering a prediction where the theory makes none . (2) On the other
hand, at each feedback level, men and women could differ in the extent to
which they meet scope conditions and/or in their values on the task-related
variables. We then explore whether participants do in fact vary in any of these
respects and, if so, whether or not these differences have an effect on rejection
of influence .
At this stage of the research, we focus on whether or not there is an
individual impact on influence from each of the factors we examine . We do not
specify relationships among them (for example, we do not hypothesize whether
confidence in partner's choices affects their perceived correctness, or vice-
193
versa) . Similarly, we do not assign weights to these factors relative to each

other, either separately or as groups of variables . We thus leave for the next task
to construct and empirically assess a model specifying, step-by-step, the role of
these variables in the relationship between sex of dyad and level of feedback,
on the one hand, and rejection of influence, on the other .
Related Issues
The present research also relates directly to several important theoretical and
methodological issues in the experimental study of status processes, as
follows .
A . Behavioral and self-report measures . In all studies using the standard
setting, the dependent variable of central interest has been rejection of
influence . This is assessed through the proportion of times a person rejects the
partner's choices over the series of trials that makes up their collective task . The
advantages of this measure are clear : the fact that it consists of relatively
spontaneous behavioral responses makes it less likely to be reactive than are
paper-and-pencil measures ; because it involves several responses, it is more
reliable than measures of a single act . Numerous studies (see, for example,
Wagner & Berger, 1993 for a review) testify to its usefulness . Additional
measures, collected through written questionnaires either before, during, or
after the experimental session have also been used in this setting, particularly
in the more recent studies (e .g . Foschi, 1996 ; Lovaglia, 1995 ; Riches & Foddy,
1989 ; Troyer & Younts, 1997) . These self-reports provide auxiliary data on
other factors of interest ; that is, they are manipulation checks on independent
variables and scope conditions, and measures of intervening and additional
dependent variables . Self-reports are also of value in verifying the comparability of studies using this research setting . Here we collect and report in detail on
a variety of such measures and on their relationship to rejection of influence .
It should be clear, however, that behaviors and written responses to
questionnaires provide data of a different nature : the former assess what people
do ; the latter, what they say - and the two are not always consistent with each
other. Thus, although both are useful, they often provide information on
reactions to different situations . This is also the case within studies using the
standard setting . Note in particular that the core of expectation states theory is
about the relationships between status characteristics, competence beliefs,
expectations, and behaviors, and that none of these links is assumed to be
conscious . Self-reports, on the other hand, are likely to make a person aware of
his or her responses . Furthermore, scope conditions such as task- and collective
orientation may not always be present when the objective is to complete a
194
questionnaire rather than to work on a task with a partner. Also, as we mention

earlier, findings show that reports on self's and other's competence do not
always correspond with rejection of influence . It is therefore important to
identify the conditions under which the two types of responses occur, and to
investigate the relationships between them (for discussions of behaviors and
self-reports in expectation states research, see Berger et al ., 1977 : Chap . 4 ;
Driskell & Mullen, 1988 ; Wesbter & Driskell, 1978 ; Wood & Karten, 1986 ;
Zeller & Wamecke, 1973 .) Through the use of both measures we thus aim to
provide a more thorough understanding of how expectations are formed in the
standard setting .
B . Gender as a status variable . The experimental study of gender poses
some interesting challenges . To begin with, sex of subject is a quasiexperimental (organismic) variable, as researchers cannot assign subjects at
random to its values . Ethnicity and physical attractiveness are other examples
of such variables . Inferences about them do not carry, therefore, the same
weight as those that arise from a properly executed random assignment . (On the
other hand, variables such as sex of partner can be, and often are, treated as
truly experimental variables.) Furthermore, note that gender is a particularly
holistic factor, i .e . it defines men and women in many important respects in
addition to the beliefs regarding competence that it generates (on this issue see,
for example, Wagner & Berger, 1997 : 22) . It is, however, possible to reduce the
shortcomings of gender as an experimental factor, and in the present study we
implement some of these possibilities . Thus, similar to other expectation states
experiments, we : (a) construct a situation where the emphasis is on solving a
collective, valuable task and where it is therefore more likely that only the
competence aspects of gender would play a part, and (b) take several measures
to increase homogeneity among subjects, including informing each dyad that
both persons are (as in fact they are) alike in several background variables .
Two additional points are worth discussing here: (1) Given that random
assignment is not possible in the case of sex of subject, one could argue in favor
of routinely studying either men or women instead . That, however, would not
be a solution . Even if no sex-of-subject effects are predicted, it is important to
ascertain empirically whether or not they have occurred, given the key role
gender plays in most social interactions (see Ilardi & McMahon, 1988 ; Wiley,
1995 for discussions of this point) . The record of expectation states research on
same-sex dyads receiving evaluations from a source shows a good balance in
this regard: of the studies reviewed earlier, several include either male or
female groups, while the rest include both types . For a more comprehensive
understanding of the topic under investigation, however, it is preferable to carry
out research of the latter kind . (2) From the perspective of expectation states
195
theory, the matter of interest is not the study of same-sex groups per se but,
rather, that of groups representing equal status . Thus, one could propose to
form such groups through a "new" (i .e . created by the experimenter) attribute
rather than by using an existing, organismic one . It would then be possible to
assign participants at random to, for example, groups of either equally high or
equally low status . Note, however, that the sex-of-subject issue does not go
away, as the participants who would be thus assigned would, in turn, be either
men or women . This is, then, an additional reason for including both male and
female subjects even when studying same-sex groups .
METHOD
Subjects and Experimenters
Subjects were 78 men and 78 women, all either first- or second-year
undergraduates in the faculties (schools) of Arts and Science at the University
of British Columbia . Average ages (with standard deviations in brackets) were
18 .78 [0 .88] for the men and 19 .03 [0.94] for the women . Participation was
voluntary and each person was paid $10 for the session . The chance of winning
one of nine lottery prizes in the $30-$50 range was used as an additional
incentive for participation . A pool of subjects was obtained by recruiting in
large introductory classes ; those students who indicated on the recruitment
form that they had taken courses beyond an introductory level in any of
psychology, sociology or computer science, and/or who had participated in
social psychology experiments were excluded as prospective subjects.
Participants were scheduled in same-sex pairs . A random order was followed
in scheduling male and female teams as well as in assigning them to one of
three feedback conditions : higher score for self than for other ; lower score for
self than for other ; or no score for either person . The study was thus a 2 (sex
of dyad) x 3 (level of feedback) between-subjects factorial design, with 26
subjects per cell . Each session was conducted by one of two female graduate
research assistants of similar age . Special attention was paid to maintaining
uniformity of appearance and delivery of instructions across sessions . Each
experimenter completed approximately half of the sessions per condition, and
this division of labor was evenly maintained across the duration of the study .
Procedures and Materials
For comparability, procedures were a variant of those used in the research
setting outlined earlier in this article . Instructions and questionnaires were
196
either adapted from reliable instruments used in related research, or were

pretested for this study ; the task was a previously developed, computer version
of the standard instrument (Foschi, 1996) . The two participants were seated
individually at adjacent stations separated by a wall and equipped with personal
computers said to be linked to each other. Subjects did not see or hear each
other before or during the session . The experimenter introduced herself as a
graduate student member of the research team . She stated that the purpose of
the study was to investigate performance on a "contrast sensitivity" task in both
individual and team work environments ; instructions were worded so as to
motivate participants to be task-oriented in the two phases . They were also told
that : their names and responses would be kept confidential ; at no time would
they be meeting the partner; they would not be required to justify their
responses ; and the completed questionnaires would be seen only by members
of the research team. These statements were included to encourage spontaneous responses to the various measures . Each dyad was informed that its
members were of the same sex, year and faculty at the university, and of
approximately the same age . Contrast sensitivity was described as a valuable,
recently discovered ability with important practical applications . Reliable
research was said to have shown it to be mainly intuitive and relatively specific,
and not significantly related to gender, age, education, mathematical abilities or
artistic skills . (In order to maintain task orientation, the instructions have to
strike a balance between the ability's specificity and its importance .)
The contrast sensitivity task consists of several trials, all designed to be
perceived as equally difficult. On each trial of the first part of the study, subjects
viewed a rectangular area covered to about the same extent by smaller
rectangles of two different colors (blue and white) . Subjects must decide which
of the two colors is predominant in the overall pattern . The task is actually
ambiguous to allow for both the manipulation of feedback and the
measurement of influence rejection ; subjects, however, were assured that,
although the task appeared difficult, in all cases one answer was right and the
other wrong . Brief instructions for the task as well as the visual stimuli
themselves were presented on the two computer screens .
During the first part of the study, all participants worked individually and
made decisions on 20 patterns. The computer program gave the subjects 10
seconds to look at each pattern, prompted them for a response and, after five
seconds, showed the next pattern . At the end of this series, those subjects in the
experimental conditions ((1), (3), (4) and (6)) saw the scores obtained by self
and partner (trial by trial as well as overall) displayed on the computer screen .
Scores on this "test" were either 15 correct answers by self and 9 by the partner,
or the reverse. Next, each person received a printout of these scores . We chose

Table 1 .
Condition
(1) Self Better
(2) No Scores
(3) Self Worse
(4) Self Better
(5) No Scores
(6) Self Worse
197
Overview of the Design.
Sex of Dyad
Score Received by Self
Score Received by Partner
F
F
F
M
M
M
15
9
15
9
15
9
15
9
the figures of 9 and 15 to signal one person's superiority over the other without
definitely establishing that one had ability and the other did not ; this allows for
the possibility of subjects activating and using various standards to interpret
these evaluations .' Subjects in the control conditions ((2) and (5)) received no
scores . An overview of the design appears in Table 1 . Next, all participants
completed a written questionnaire consisting of a variety of items : some were
checks on independent variables and scope conditions, others provided data on
the additional variables described earlier, and the remaining were fillers
included to maintain the realism of the cover story and help conceal possibly
reactive items .
During the second series of trials, the two participants were instructed to
work as a team and try to arrive at a correct choice in each trial . Subjects were
told that a team would be awarded two points whenever both persons were
correct, and that the three teams with the most points would each win a $30
prize (in addition to the lottery prizes for which all subjects were eligible) . For
consistency with other experiments with a similar design, the task varied
slightly during this phase . It now involved two patterns per trial, and subjects
had ten seconds to decide which of the two contained more white . The same
ability as in the first phase was said to be involved . After a subject made a
decision, the partner's "choice" was relayed. This message was manipulated to
result in 20 disagreements and 5 agreements on the initial choices . Each person
then had five seconds to make his or her own final selection . The decision to
remain with one's initial choice after a disagreement is referred to as a "selfresponse ;" the proportion of such responses operationalizes influence
rejection .
At the conclusion of this series, subjects completed a second questionnaire
that included further manipulation checks and measures of additional variables,
as well as fillers as described earlier. The instrument also served to assess any
misunderstandings and/or suspicions regarding the procedures . Next, subjects
198
were interviewed individually to check further on these issues, and then were
fully debriefed about the true nature of the study.
RESULTS
On the basis of the information obtained through the written questionnaires and
the post-experimental interviews, 17 subjects (10 men and 7 women) were
excluded from the analysis . This figure represents 10.90% of the total number
of participants . Rejection rules were conservative (i .e . they required extreme
values over several items) and formulated beforehand, and this percentage
compares favorably with the exclusion rates of similar experiments . The
excluded subjects can be classified as follows : one volunteered well articulated
suspicions about the true nature of the study, five misunderstood crucial
sections of the instructions and/or the written questions, and fifteen showed a
clear lack of task orientation and/or collective orientation . (Four subjects
belong in more than one of these categories .) There is no evident pattern in the
distribution of exclusions across conditions . Our analyses include only the 139
retained subjects .
We analyze all results through ANOVAs (either two-way or three-way) .
Except for the analysis of the number of correct responses estimated by
control-group subjects, all two-way ANOVAs involve sex of subject and level
of feedback . We use three-way ANOVAs in the following cases : since several
items appeared in both questionnaires, for those items we also check for
possible effects from phase of the study (namely, either individual or team) .' A
significant effect from this variable occurred only for task importance, and this
is discussed later in this section ; otherwise, the data have been collapsed across
questionnaires . More generally, we discuss and present details of all analyses
yielding significant results (and three borderline-significant results closest to
p = 0.05) . If effects from a variable do not meet these criteria, they are not
mentioned and the data from all its values are collapsed . Exceptions to this rule
are some of the results in Tables 2 and 4 : although there are no statistically
significant sex-of-subject effects in rejection of influence and confidence in
partner's choices, we nevertheless present these data by condition because of
their centrality to our discussion . No significant interaction effects were found
regarding any of the variables studied .
We do not analyze data by experimenter because we consider the number of
observations available for that analysis to be insufficient (10-12 per cell) ; in
any case, no substantial differences by experimenter are apparent regarding any
variable, as intended . In discussing our results, we group conditions (as per
Table 1) by their shared level of one of the independent variables . Thus, for
199
example, "subjects in Conditions (1) and (4)" is equivalent to "better

performers," and "subjects in Conditions (1), (2) and (3)" is equivalent to
"women" . Finally, we present results in the following order : manipulation
checks of independent variables and scope conditions ; the three sets of
additional factors ; and rejection of influence .
Manipulation Checks
All subjects reported believing, as informed, that they and the partner were of
the same sex and highly similar in age and level of education . Furthermore, all
subjects in the experimental conditions recalled exactly the scores received by
self and partner in the first phase, while those in the control conditions
estimated that, on average, they themselves had made 11 .26 [2 .91] correct
responses in that phase while the partner had made 12 .36 [2 .45] . Results from
a two-way ANOVA with sex of subject and recipient of estimated score (either
self or partner) as the factors indicate the difference between those two figures
to be significant (F (1, 90)=3 .87, p = 0 .052) . As another manipulation check,
subjects rated self's and other's performances from Poor (1) to Excellent (6) .
These ratings were affected by feedback as follows . The overall self-rating by
the better performers was 4 .73 [0.54] ; by those receiving no scores, 3 .45 [0 .83] ;
and by the worse performers, 2 .34 [0 .99] (F (2, 133)=100 .73, p=0 .000) . The
corresponding figures for the ratings of the partner's performance were : 3 .10
[0 .63], 3 .91 [0 .62], and 5 .16 [0 .48] ; F (2, 133) =142.97, p = 0 .000 .
We used five bipolar scales to assess perceptions of contrast sensitivity in
terms of its value and lack of association with other attributes . Overall means
from three of these items were : Creative (1) - Routine (6) : 3 .42 [1 .42] ; Intuitive
(1) - Learned (6) : 2 .27 [1 .06] ; and Specific (1) - Related to a Wide Variety of
Abilities (6) : 3 .91 [1 .37] . Responses to Masculine (1) - Feminine (6) showed
a borderline sex-of-subject effect (M (women) : 3 .80 [0 .77] ; M (men) : 3 .64
[0 .69] ; F (1, 266)=3 .52, p=0 .062), while responses to Important (1) Unimportant (6) were affected by both feedback and phase of the study : the
better performers rated the ability as being about as important (M : 3 .14 [1 .00])
as did those receiving no scores (M : 3 .07 [1 .06]), but significantly more
important than did the worse performers (M : 3 .52 [1 .10] ; F (2, 266)=4 .77,
p = 0 .009) ; on the other hand, the ability's importance was rated more highly
after the individual phase (M : 3 .37 [1 .17]) than after the team phase (M : 3 .10
[0 .94] ; F (1, 266)=4 .76, p=0 .030) .
The questionnaires also included three similar scales to assess the
participants' task orientation . Overall mean for Motivated (1) - Unmotivated
(6) was 3 .04 [1 .13] ; for Involved (1) - Uninvolved (6), 2 .71 [1 .15] . Results for
200
Interested (1) - Uninterested (6) show a significant sex-of-subject difference,

with men reporting more interest in the task (M : 2.68 [1 .12]) than do women
(M : 2 .98 [1 .36] ; F (1, 266)=4.01, p=0 .046) .
In addition, four Likert scales with values ranging from Strongly Agree (1)
to Strongly Disagree (6) were used, two on task orientation and two on
collective orientation . These scales yielded the following results . The overall
average for "I was very serious about doing the task well" was 2 .19 [0.92] ; for
"I lost interest in working hard at the task," 3 .99 [1 .30] . The overall mean
response to "Agreeing as a team regarding the correct decision was more
important to me than my own choice" was 2 .72 [1 .38], while responses to
"While working as a team on the contrast sensitivity task, it is best to consider
the other person's choices carefully" show women agreeing significantly more
with this item (M : 1 .92 [0.77]) than did men (M : 2 .34 [0 .94] ; F (1, 133)=8 .29,
p = 0 .005) .
Additional Factors
(a) Task and performance requirements . The better performers were asked to
indicate their standards for ability in self and lack of ability in other ; the worse
performers were asked for their standards for ability in the partner and lack of
ability in self; and those in the control groups were asked all four questions .
(Pre-tests had shown this to be the most logical set of questions to ask, given
the feedback received in each case ; for details on these questions see Foschi,
1996 : 242) . Standards were to be expressed in terms of the percentage of
correct responses required to definitely infer either ability or lack of ability . The
overall mean for the ability standard was 70 .94 [9 .91] ; corresponding results
for lack of ability show a borderline sex-of-subject effect as follows : M (men) :
38 .58 [11 .92] ; M (women) : 41 .98 [12 .19] ; F (1, 178)=3 .59,p=0 .060 . Subjects
were, in addition, asked to state the number of trials between 1 and 100 that
they would require for a convincing assessment of ability level (regardless of
whether high or low) . Results show a borderline effect from feedback : the
better performers required about the same number of trials (M : 54 .06 [28.26])
as did those receiving no scores (M : 54 .17 [28 .33]) but substantially more than
did the worse performers (M : 42 .50 [27 .14] ; F (1, 133)=2 .71, p=0.071).
We also used one scale to obtain subjects' perceptions of the validity of the
contrast sensitivity test (Valid (1) - Not Valid (6)), and another to assess
perceived task difficulty (Easy (1) - Difficult (6)) . The overall mean response to
the former was 2 .75 [1 .00], while responses to the latter varied with feedback :
the better performers considered the task to be easier (3 .53 [1 .26]) than did
201
those receiving no scores (4 .06 [1 .23]) who, in turn, rated it as easier than did
the worse performers (4 .44 [1 .31] ; F (2, 266) = 11 .91, p = 0 .000) .
(b) Partner's task choices and overall competence relative to self's . In this
section we analyze results from five scales concerning perceptions of self's and
partner's competence . Findings on two of these factors are shown in Table 2 .
Subjects indicated their confidence in the partner's choices on a scale ranging
from Very Confident (1) to Very Unconfident (6) . ANOVA shows a main effect
from feedback (M (better performers) : 3 .94 [0 .56] ; M (no scores) : 3 .49 [0 .72] ;
M (worse performers) : 2.59 [0 .73] ; F (2, 133)=48 .13, p=0 .000) . As to
importance to self of partner's choices (ranging from Important (1) to
Unimportant (6)), we found main effects from both feedback (M (better
performers) : 2 .71 [0 .99] ; M (no scores) : 2 .53 [1 .08] ; M (worse performers) :
1 .95 [0 .77] ; F (2, 133)=7 .71, p=0 .001) and sex of subject (M (men) : 2 .62
[1 .16] ; M (women) : 2 .21 [0 .79] ; F (1, 133)=5 .92, p=0 .016) . (Note that these
two factors are different from collective orientation . The latter concerns
attitudes towards the team and requires only that self take the partner's choices
into account, while the former refer to the perceived worth of those choices .)
Results on two other factors in this group appear in Table 3 . On perceived
correctness of partner's choices (Mostly Correct (1) - Mostly Incorrect (6)),
ANOVA yielded main effects from both feedback (M (better performers) : 3 .88
[0 .67] ; M (no scores) : 3 .34 [0 .60] ; M (worse performers) : 2 .44 [0 .86] ; F (2,
133)=47 .01,p=0 .000) and sex of subject (M (men) : 3 .39 [0 .99] ; M (women) :
3 .10 [0 .83] ; F (1, 133)=5 .13, p=0 .025) . Main effects from both independent
variables were also obtained for "My partner was very serious about doing the
task well" (with Strongly Agree (1) and Strongly Disagree (6) as the anchors) .
Results by feedback were : M (better performers) : 2 .99 [0 .87] ; M (no scores) :
2 .83 [1 .05] ; M (worse performers) : 2 .30 [0.70] ; F (2, 133)=8 .03, p=0 .001)
while by sex of subject, the following were found : M (men) : 2 .90 [1 .04] ; M
(women) : 2 .54 [0 .77] ; F (1, 133)=5 .67,p=0 .019 .
For the fifth factor in this set, relative ability, the anchors were Self Much
Worse than Partner (1) and Self Much Better than Partner (7) . Results appear
in Table 4 . ANOVA shows main effects from feedback (M (better performers) :
4 .98 [0 .70] ; M (no scores) : 3 .40 [1 .08] ; M (worse performers) : 1 .91 [0 .77] ; F
(2, 133) = 153 .14, p = 0 .000), and from sex of subject (M (men) : 3 .75 [1 .58], M
(women) : 3 .21 [1 .42] ; F (1, 133) = 11 .96, p=0 .001 .5
(c) Experimenter's competence . We assessed the subjects' perceptions of the
experimenter in this respect by means of four scales . ANOVA results on views
of her qualifications (Qualified (1) - Unqualified (6)) show women giving a
significantly better rating (M : 1 .27 [0 .51]) than men (M : 1 .53 [0 .74] ; F (1,
. Similarly, on Knowledgeable (1) -Not Knowledgeable
.017)
.86,p=0
133)=5
202
A
En
w d
y U
o U
U
N V1 00 00 --~ N
0
c -~ -~ o
cr"
O C~ M N 00
V1 N 00 ON r N N- N N N
O
U
a)
O
0)
A h
U
d
A
r4
N N N 'O \O v1
0 0 0 0 0 0
N 0
O O N
N b O N v
M M N M M N
N
N
O O
C
n
U
C
U
O
U
z
U
Cn a ~
M N M
N N 'tN Ni
.b
0
O
03~
ff
Uo
U
U
w w w
O
O
V) Z 00 0 Z 0
N m7
\O
(14 M ~O ~O N et
N CN v') O% O 00
O O O O O
rnoCD o NQI
l~ V1 M --N
N N N M M N
M 7 N ~O Q,
~O V~ v1 l~ ~ O
0 0 0 0 0
M I'*
N N N O f
M M N- M N
IT It
N N
N
M
N
N
V. W. V. ;
0
w
d
a
0
U
g, ;~
3~
w
O
N
M
o
s
Z
203
204
A
tn
IX - r,
000000
r- ~o It
O h M \O V1 M
0 0 0 0 0 0
G W)
.O N N -~
OO V) N
O C C C
0
~7 C cn - cn --~
O N O ,~-i
Vl M N
7 en -
N N N N N N
d N y d N
0
00 D _
.
w w
V
O
O
O
zc
0O chzv)on
- N en ~i
U
205
(6) women reveal a better evaluation (M : 1 .21 [0 .48] than men (M : 1 .56 [0 .76] ;
F (1, 133) = 10 .30, p = 0 .002) . On the other hand, no such effects were observed
in the other two measures : the overall mean for the research assistant's
perceived level of education (Bachelor's Degree in Progress (1) - Doctorate
Completed (6)) was 2 .88 [0 .85] ; the corresponding value for her standing
relative to self's (Inferior (1) - Superior (6)) was 4 .65 [0 .72] .
Rejection of Influence
ANOVA on the rejection of influence data yields a significant effect from
feedback, as follows : the better performers have a higher rate of self-responses
(M : 0 .63 [0.13]) than those receiving no scores (M : 0 .52 [0.14]) who, in turn,
have a higher rate than the worse performers (M : 0 .37 [0 .12] ; F (2,
133)=48 .69, p = 0 .000) . Table 4 shows results by condition .
DISCUSSION
Overall Review of Results
Findings show that feedback generates substantial differences in rejection of
influence whereas sex of dyad does not . A review of the manipulation checks
on feedback and sex of partner reveals that these factors were created as
intended. Similar checks on scope conditions had analogous results : in
particular, subjects show, overall, a high degree of similarity in perceiving
contrast sensitivity as gender-neutral and fairly intuitive, specific and valuable,
and in meeting the requirements of task- and collective orientation . (All these
findings correspond closely to those reported in Foschi, 1996, a study where
these same checks were used .) Although we found some statistically significant
differences in the manipulation checks, such effects occur only in a few cases
- in general, subjects show quite similar responses in these checks across the
six conditions and two phases . The differences appear in the following areas :
(1) Subjects receiving no scores estimated that the partner had a higher score
than themselves . This finding replicates the results from comparable groups in
Foschi et al ., 1985 and, in our view, indicates self's reluctance, in the face of
a difficult task and lack of control over the outcome, to risk stating in precise
numerical terms that he or she is better than the partner . (2) The effects of both
feedback and phase of the study on perceptions of the ability's importance
reflect the study's variations in these two factors and are not unexpected - these
effects may, in fact, be interpreted as providing indirect evidence of the validity
of the measure . (3) Finally, sex-of-subject differences were found in two of the
206
seven task- and collective-orientation questions : while men have higher values
than women in level of interest in the task, women agree more than men with
the statement that the partner's choices should be considered carefully .
More important to this analysis, the statistically significant differences found
in these checks are still within acceptable ranges and, as a result, do not affect
rejection of influence . Note that a scope condition specifies either one value or
a range of values of a given factor that is (are) predicted not to have an effect
on the outcome variable (see, for example, Berger et al ., 1977 : Ch . 3 ; Foschi,
1997 ; Walker & Cohen, 1985) . The wider that range, the more general the
theory is with respect to that factor . The results obtained here do not involve,
in any of the cases, values outside that acceptable range - that is, the scope
conditions have not become independent variables . For example, while men
and women report different values in level of interest in the task (i .e . 2 .68 and
2 .98, respectively), both figures are within the range of "at least average
interest" - which we would define as any value between 1 and 3 . In other
words, the statistically significant differences are not large enough to have
theoretical implications for rejection of influence .
Next we discuss the results from the additional factors, beginning with (a)
task and performance requirements. Responses on the validity of the test and
standards for ability show subjects to be highly similar in these respects . On the
other hand, standards for lack of ability indicate a borderline sex-of-subject
difference, and the number of trials required to be convinced of a person's
ability level is marginally affected by feedback . Overall, the values for the two
standards and the number of trials required reveal that subjects perceived the
task to be fairly (although not extremely) difficult, as follows . Standards for
both ability and lack of ability are relatively lenient (i .e . neither is very high),
even when the sex difference is noted . In addition, the number of trials required
for a decision on ability level is, overall, substantially higher than the twenty
trials that the subjects had completed in the "test" but, not unexpectedly, is
lowest for the worse performers . (These results also confirm our view that, for
a fuller understanding of their role in expectation formation, standards should
be considered in relation to a required number of performances .) The large
standard deviations associated with this number in all three feedback
conditions highlight the absence of guidelines to the subjects on what
constituted levels of this ability and also, to some extent, the difficulty and
novelty of the task . Finally, responses to a direct question on task difficulty
show the effects of feedback : again not surprisingly, the better one's
performance, the less difficult the task was perceived to be . The level of task
difficulty shown by these various measures is consistent with the one found in
comparable studies where such data are reported (see, e.g . Foddy & Smithson,
207
1996, 1999 ; Foschi, 1996 on both measures of standards and ratings of task
difficulty) . 6
Results from the other two sets of factors are as follows . Five items provide
(b) perceptions of partner's choices and overall competence, revealing main
effects from feedback in all five and sex-of-subject effects in four. All feedback
effects clearly correspond with the manipulations of that variable . In addition,
sex-of-subject differences indicate that, relative to men and at every level of
feedback, women portrayed the partner as having more ability, and as both
more likely to be correct and more serious about doing the task well . Women
also assigned more importance to the partner's choices than did men . It is worth
noting that such sex effects appeared in four of the five more general questions
concerning the worth of the partner's choices and the two participants' relative
competence (i.e . in those questions that leave room for interpretation), but not
in the more specific ratings of self's and other's performances that we used as
a manipulation check (see Results section) . Finally, the questions about (c) the
experimenter's competence show across-conditions subject similarity regarding
her higher level of education and standing relative to the respondent, but sex
effects regarding her superior qualifications and knowledge, with women
giving better ratings than men in both respects . (The sex effects in these two
factors are consistent with the women's more generous view of the partner's
competence observed in (b) ; see also footnote 5) .
Let us now consider all of the results taken together, and examine them in
terms of the two situations we posed under Research Questions . While the data
do not meet the conditions of either situation exactly, they are nevertheless
closer to the first . Thus, feedback levels and scope conditions were
implemented as intended, and notably to highly similar extents by men and
women . Subjects also show overall gender similarity in their task and
performance requirements, and the sex-of-subject effects on perceptions of the
experimenter's competence, although worth keeping in mind, are not a clearly
differentiating factor, as they appear in two of four variables . On the other
hand, there were highly consistent sex differences regarding views of the
partner's choices and overall competence relative to self's : while one item in
this group shows feedback effects only, the other four are affected by both
feedback and sex of subject . Effects from the latter, however, are in all four
cases smaller than those from feedback . Furthermore, men's as well as
women's values are always within the range for the corresponding feedback
level - as confirmed by the lack of interaction effects . Lastly, our outcome
variable, rejection of influence behavior, shows effects from feedback only.
This, we conclude, results from the combination of the following : men and
208
women were similar with respect to most factors, while the sex differences that
occurred in the rest of these factors were not sufficiently pronounced.
Our results also correspond with those obtained in five of the six expectation
states studies reviewed earlier on same-sex task dyads and evaluations from a
source . Foschi and Freeman (1991) thus remains the only study in this group
where significant sex-of-subject effects on influence were found . We propose
that it is likely that the men and women in that experiment varied substantially
in ways the study did not detect, and therefore that they instantiate the
conditions of the second of the two situations we pose. Although subjects in
Foschi and Freeman (1991) and in the present study were from highly similar
populations, it is probable that gender was a stronger diffuse status
characteristic for the participants of the earlier study, which was carried out
approximately a decade ago . Ensuing gender-based differences in : (i) the status
assigned to the experimenter and (ii) the degree of acceptance of the
combination of evaluations and standards that she was providing (neither of
which was measured in that study), could have affected formation of
expectations .
Gender, Evaluations, and Type of Context
Although the significant sex differences observed here do not affect influence
rates, they are nevertheless of interest in themselves and thus deserve a special
note on why they occurred . As discussed above, behavior and self-reports are
responses of a different nature . In addition, they often occur in different
contexts . In this study there were two such contexts : one involved working on
a collective task ; the other, completing questionnaires individually. We propose
that, because of their different norms, these two contexts constituted different
conditions for the activation of gender.
In this discussion we focus on reports concerning self's and other's
competence, both because of their close relationship to expectations and
because it is in these reports that sex effects appear most noticeably and
consistently. The act of either accepting or rejecting influence from the other
person reveals how a subject addresses the question "How much ability do I
have relative to my partner?" On the other hand, when a subject is completing
questionnaire items about competence, he or she is responding to "What do I
say about how much ability I have relative to my partner?" The two questions
are related but still different ; also, as noted earlier, subjects are more likely to
be aware of the question they are answering in the latter than in the former case .
We assume that in both cases they would be primarily concerned with
presenting themselves in line with the scores received, given the focus on task
209
competence that the two contexts have . That is, we assume that subjects would
be in compliance with the generally accepted norm in task settings of
respecting the available evidence about participants' relevant qualifications particularly when, as in the present case, that evidence is convincing . Results
from both behaviors and self-reports show clear feedback effects, in
accordance with this interpretation .
We propose that, in addition, each of these contexts included different
norms . When subjects were working with the partner, they were partially
responsible for a joint outcome . Thus, collective orientation was an important
norm in this situation . This norm implies costs, to the team, associated with
making a wrong choice (see Camilleri & Berger, 1967) . Manipulation checks
show that the norm was accepted by the participants, as intended . In that
context, only their earlier, individual-performance scores provided clearly
differentiating information that could be of use in deciding whether or not to
agree with the partner . Gender, as well as age and education, did not have such
a use, as all three equated the subjects and had been defined as not significantly
related to the task. Accordingly, none of these factors were activated .
On the other hand, when completing the questionnaires, the focus shifted
away from self and other . Whereas participants knew that the partner would
learn immediately about their task choices, the recipient of their written
answers was not that clearly specified - subjects were told that these would be
seen by "the research team ." Unlike the collective setting, there were no
apparent costs associated with giving inaccurate reports . Relative to the
unknown research-team members, subjects had only four items of information :
they themselves were either men or women, young, undergraduates, and had
received one of three types of feedback . We propose that the four items were
activated in trying to define how to relate to these unspecified others otherwise it would have been rather disconcerting to be in a situation where
information about neither self nor these others was highlighted . Of the four
factors, sex of subject and feedback differed across participants, while age and
education equated them . Taking all of this information into account, the result
was the activation, at each feedback level, of different norms concerning
gender and self-presentation of ability relative to the research team . (One could
also argue that the subjects viewed the research team as including, most
prominently, the female experimenter - as we have noted earlier. In that case,
the same argument about the activation of gender and self-presentation norms
would apply. However, we propose that it was not the experimenter alone who
activated these norms, as results regarding the competence status assigned to
her do not show a distinct sex-of-subject difference .)
210
In both Canada and the United States, particularly over the last two decades,
the status value of gender has diminished or even disappeared for many .
However, it is still highly common for men and women to accept different
norms regarding self-presentation of abilities . Thus, while men are expected to
show high confidence, the rule for women is to be modest . These norms reflect,
of course, gender as a status factor . (For reviews of research in this area, see
Bartol, 1999 and Wiley, 1995 . Work on different causal attributions made for
the same performance, either successful or unsuccessful by either a man or a
woman, is particularly relevant to this point . See also Carli, 1991 ; Ridgeway &
Berger, 1986 ; Meeker & Weitzel-O'Neill, 1977, on gender and the legitimation
of status-enhancement .) Our results on self-reports on perceptions of
competence are in line with this interpretation : findings from four measures
show, overall, men portraying a higher status for themselves relative to the
partner, than do women . This is also the case, to some extent, regarding the
experimenter . The interpretation is also consistent with the fact that most of the
sex differences occurred where such norms were most likely to be activated that is, in reports of one's status relative to others . For example, although
reporting the task as being either more or less difficult could be a way of
manipulating self-presentation, it is a rather indirect method of doing so, and
the results do not suggest that subjects were engaged in such a practice . More
generally, views on self's and other's competence reflect a close correspondence with the feedback received, and the rest of the self-reports also indicate
a high level of truthfulness (see, for example, those on subjects' own interest
and motivation) .
In sum, both rejection of influence data and self-reports provide valuable
information on how expectations are formed . Self-reports serve as manipulation checks, auxiliary measures, and evidence on additional variables . In the
case of the perceptions of competence examined here, we propose that
behaviors and self-reports were obtained in two contexts that included different
norms, and that this allowed for differences in external status to emerge . We
also interpret our results as indicative that gender was still a diffuse status
characteristic for our subjects, but that one setting blocked its effects while the
other did not . Furthermore, results show that even when sex effects emerge, in
no case was gender as strong a determinant of responses as was feedback - in
line with expectation states research on the effects of degree of relevance
between a given attribute and the task (see, for example, Berger et al ., 1977 :
58-61 ; Wood & Karten, 1986) .
It would be worthwhile to investigate further, and in different situations, our
explanation regarding gender effects . Of the six experiments reviewed earlier
on male and female dyads and evaluations from a source, only Foschi et al .
211
(1985) and Foschi and Freeman (1991) present data on both influence behavior
and self-reports . The latter, however, are analyzed in a way that makes them not
directly comparable to the present findings . One interesting way of testing the
explanation we propose would be, in a future study, to assign subjects at
random to either same-sex or opposite-sex dyads, and to obtain data from the
two types of measures . Since opposite-sex dyads should show effects from both
sex of partner and self-presentation rules, differences in self-reports between
the two types of such dyads should be larger than differences between male and
female dyads . In addition, for a more comprehensive understanding of status
effects in these two contexts it would be useful to carry out studies similar to
the present one but operationalizing status through other attributes .
SUMMARY AND CONCLUSIONS

We investigate the formation of performance expectations in status-homogeneous task groups, and address the question of whether status level of the group
(namely, high or low) has an effect in such settings . In particular, we examine
whether men and women, participating as members of same-sex dyads, differ
in the expectations they develop from the same level of performance on a
gender-neutral task. We carry out an experiment where two persons who
initially perceive each other as status equals in several key respects, first
perform a valued task individually, next are assigned at random to receive, from
a third party, different levels of feedback on these performances, and later work
on a similar task as a team. We investigate the relative effects of feedback and
sex of dyad on rejection of influence during the team interactions .
The situation we examine severely limits the likelihood of gender effects,
and expectations states theory provides no predictions as to whether or not they
will occur. Moreover, the topic itself has so far received limited attention, and
the evidence regarding gender effects in this setting shows some inconsistent
results . It is therefore important to investigate the issue more thoroughly, as the
presence of gender effects in this situation would point to the resilience of this
factor. We contribute to this topic by : (1) specifying bases (including
participants' views of the task's difficulty, the partner's choices, and the
experimenter's competence) for predictions about sex of dyad and influence,
(2) obtaining empirical evidence on these factors and systematically examining
their role on expectations, and (3) identifying, within this experimental setting,
two contexts for the activation of status even in same-status groups - depending
on whether subjects work on a task with a partner or complete questionnaires
individually .
212
Results from the study show a high level of both factual reporting and
internal consistency over a wide array of measures . Manipulation checks reveal
that levels of feedback, perceptions of sex of partner, and task- and collective
orientation had been created as intended - and in particular, that men and
women were highly similar in these respects . Male and female participants
were equally similar in their views of the task. Rejection of influence results
show, as predicted, strong effects from feedback but no effects from sex of
subject. On the other hand, self-reports reflect sex differences as well as
feedback effects in perceptions of self's and other's competence . Although
level of feedback is the primary organizer of responses for both men and
women, men's reports portray higher status for themselves relative to the
partner (and also, to some extent, the experimenter) than do women's . We
propose and discuss the following interpretation of these findings . The
completion of a joint task by male and female dyads who are alike in key taskrelated variables blocks gender effects on expectations and thus results in
similar rejection of influence behaviors by men and women . On the other hand,
sex differences still emerge in the participants' accounts of those expectations
because of the different conditions for the activation of gender that apply to the
two contexts .
NOTES
1 . There are also other expectation states studies examining how members of samesex groups respond to different levels of task evaluations from a source . These studies
(Camilleri & Conner, 1976 ; Foddy & Smithson, 1996; Foschi, 1971, 1986 ; Shelly &
Munroe, 1999) are not reviewed here because they do not use the two-person standard
setting and/or the rejection of influence measure . In any case, of these five, only Shelly
and Munroe (1999) provide data comparing male and female groups and, with the
exception of the granting of action opportunities, no significant differences in taskrelated interactions between these two groups were found . We also exclude from our
review those studies that, although investigating dyads in the standard setting, are not
of direct relevance here because their focus is to compare the effects of external status
with those of evaluations received by either (but not both) men or women working with
a same-sex partner . Two examples of this research are Freese and Cohen (1973) and
Knottnerus and Greenstein (1981) .
2 . Differences in the expectations formed by male and female dyads could still occur
due to other, non task-related factors . For example, it could be that male and female
participants vary in their personality characteristics (such as shy, outgoing, rigid) or in
the norms they hold regarding socio-emotional behaviors (e .g . in norms allowing less
expressiveness in men than in women) (Wagner & Berger, 1997), and that these
differences affect expectations even though they do not contain information on task
competence . Nevertheless, we propose that if men and women accept to similar extents
the specified focus on collective performance, the effects of these other factors on
expectations will be negligible relative to those of the task-related factors .
213
3 . Note that, because of our interest in the subjects' own standards, the present study
differs from those "source" experiments where participants receive both evaluations and
standards through which to interpret them (see, for example, Webster & Sobieszek,
1974 : 175) . The source of evaluations also differs : in our case it is the research team ;
in those other studies, a specific individual .
4 . More specifically, the various items (which are discussed in detail next in this
section) were distributed as follows . Manipulation checks of self's and other's
performance level, measures of standards, and number of trials required for an ability
inference were included only in the first questionnaire. Manipulation checks of partner's
sex, age and level of education, Likert scales on task- and collective orientation, and
indicators of perceptions of self's and other's competence, the validity of the test, and
the experimenter's qualifications were included only in the second questionnaire . The
rest of the questions appeared in both instruments .
5 . The second questionnaire also included an item on partner's likeability (Partner
Likeable (1) - Partner Not Likeable (6)), and ANOVA results show women giving
significantly more positive ratings of the partner than do men : M (men) : 3 .23 [0 .92] ; M
(women) : 2 .82 [0 .76] ; F (1, 133)=8 .40, p=0 .004 . Although the question was intended
only as a filler, its results are nevertheless worth reporting because of their consistency
with those from all but one of the other variables in this group, and because of the
interesting connections they reveal between gender status and affect processes (on this
topic, see Fisek & Berger, 1998 ; Lovaglia & Houser, 1996) .
6 . In this study we are concerned primarily with whether or not there are main effects
from either independent variable . It is thus beyond the scope of this discussion to
include internal analyses of the results . Nevertheless, we should note that, relative to the
control-group subjects, the worse performers reacted, overall, more markedly to the
received scores than did the better performers . This trend can be observed with respect
to the manipulation checks of self's and other's performance level and most of the
results shown in Tables 2 to 4 . Similar findings have also been reported in comparable
studies (see, for example, Berger & Conner, 1969; Foschi et al ., 1985) . In our view, this
asymmetry is directly related to the fairly high level of task difficulty. For a more
thorough understanding of how expectations are formed, it would be worthwhile to vary
this factor in future studies .
ACKNOWLEDGMENTS
The study reported here was carried out under Research Grant No .
410-97-0101 from the Social Sciences and Humanities Research Council of
Canada to the senior author. We gratefully acknowledge this support . We also
thank Kelly Giesbrecht, Joanna Kim and Susan Lindquist for their efficiency
and good judgment in conducting the experimental sessions, and Joseph
Berger, Thelma S . Cook, Gwendolyn L . Gerber, Fiona Kay, Carol Stoppel and
Henry A . Walker for their helpful suggestions on an earlier version of this
article . Direct correspondence to Martha Foschi, Department of Anthropology
and Sociology, University of British Columbia, Vancouver, B . C ., Canada V6T
1Z1 ( e-mail : mfoschi@interchange .ubc .ca) .
214
REFERENCES
Bartol, K. M . (1999) . Gender influences on performance evaluations . In: G . N . Powell (Ed.),
Handbook of Gender and Work (pp . 165-178). Thousand Oaks, CA : Sage .
Berger, J., & Conner, T. L. (1969) . Performance expectations and behavior in small groups . Acta
Sociologica, 12, 186-198 .
Berger, J ., & Fisek, M. H. (1970) . Consistent and inconsistent status characteristics and the
determination of power and prestige orders . Sociometry, 33, 287-304 .
Berger, J., Fisek, M . H., & Freese, L. (1976) . Paths of relevance and the determination of power
and prestige orders . Pacific Sociological Review, 19, 45-62 .
Berger, J ., Fisek, M . H ., Norman, R. Z ., & Zelditch, M. Jr. (1977) . Status Characteristics and
Social Interaction : An Expectation-States Approach . New York, NY: Elsevier.
Berger, J ., Wagner, D. G., & Zelditch, M. Jr. (1985) . Introduction : Expectation states theory Review and assessment . In : J . Berger & M . Zelditch, Jr. (Eds), Status, Rewards and
Influence: How Expectations Organize Behavior (pp . 1-72) . San Francisco, CA: JosseyBass .
Camilleri, S . F., & Berger, J . (1967) . Decision-making and social influence: A model and an
experimental test . Sociometry, 30, 365-378 .
Camilleri, S . F., & Conner, T. L . (1976) . Decision-making and social influence : An empirical test
of two models . Sociometry, 39, 30-38 .
Camilleri, S . F., Berger, J., & Conner, T. L . (1972) . A formal theory of decision-making . In : J.
Berger, M. Zelditch, Jr. & B . Anderson (Eds), Sociological Theories in Progress, Vol . 2 (pp.
21-37) . Boston, MA : Houghton Mifflin .
Carli, L . L. (1991) . Gender, status, and influence . Advances in Group Processes, 8, 89-113 .
Conner, T. L . (1977) . Performance expectations and the initiation of problem solving attempts .
Journal of Mathematical Sociology, 5, 187-198 .
Crundall, I ., & Foddy, M . (1981) . Vicarious exposure to a task as a basis of evaluative competence.
Social Psychology Quarterly, 44, 331-338 .
Driskell, J . E . Jr., & Mullen, B . (1988). Expectations and actions . In : M . Webster, Jr. and M. Foschi
(Eds), Status Generalization : New Theory and Research (pp . 399-412 and 516-519).
Stanford, CA: Stanford University Press .
Fennell, M. L ., Barchas, P R ., Cohen, E. G ., McMahon, A . M ., & Hildebrand, P. (1978) . An
alternative perspective on sex differences in organizational settings : The process of
legitimation. Sex Roles, 4, 589-604.
Fisek, M. H ., & Berger, J. (1998) . Sentiment and task performance expectations . Advances in
Group Processes, 15, 23-39.
Foddy, M ., & Smithson, M. (1996) . Relative ability, paths of relevance, and influence in taskoriented groups. Social Psychology Quarterly, 59, 140-153 .
Foddy, M ., & Smithson, M. (1999) . Can gender inequalities be eliminated? Social Psychology
Quarterly, 62, 307-324.
Foschi, M . (1971) . Contradiction and change of performance expectations . Canadian Review of
Sociology and Anthropology, 8, 205-222 .
Foschi, M . (1986) . Actors, observers, and performance expectations : A Bayesian model and an
experimental test. Advances in Group Processes, 3, 181-208 .
Foschi, M. (1996) . Double standards in the evaluation of men and women . Social Psychology
Quarterly, 59, 237-254.
Foschi, M. (1997) . On scope conditions . Small Group Research, 28, 535-555 .
215
Foschi, M ., & Freeman, S . K . (1991) . Inferior performance, standards, and influence in same-sex
dyads . Canadian Journal of Behavioural Science, 23, 99-113 .
Foschi, M., Warriner, G . K., & Hart, S . D . (1985) . Standards, expectations, and interpersonal
influence . Social Psychology Quarterly, 48, 108-117 .
Freese, L ., & Cohen, B . P (1973) . Eliminating status generalization . Sociometry, 36, 177-193 .
Ilardi, B . C ., & McMahon, A . M. (1988). Organizational legitimacy and performance evaluation .
Advances in Group Processes, 5, 217-244.
Knottnerus, J . D ., & Greenstein, T. N . (1981) . Status and performance characteristics in social
interaction : A theory of status validation . Social Psychology Quarterly, 44, 338-349 .
Lockheed, M . E., Harris, A. M ., & Nemceff, W. P. (1983) . Sex and social influence : Does sex
function as a status characteristic in mixed-sex groups of children? Journal of Educational
Lovaglia, M. (1995). Power and status : Exchange, attribution, and expectation states . Small Group
Research, 26, 400-426 .
Lovaglia, M . J ., & Houser, J. A. (1996) . Emotional reactions and status in groups . American
Martin, M . W., & Sell, J . (1980) . The marginal utility of information : Its effects upon decisionmaking . The Sociological Quarterly, 21, 233-242 .
Meeker, B . F., & Weitzel-O'Neill, P . A . (1977). Sex roles and interpersonal behavior in taskoriented groups . American Sociological Review, 42, 91-105 .
Norman, R. Z ., Smith, R., & Berger, J . (1988) . The processing of inconsistent status information .
In : M . Webster, Jr . & M. Foschi (Eds), Status Generalization : New Theory and Research
(pp . 169-187 and 493-495) . Stanford, CA : Stanford University Press .
Parcel, T. L., & Cook, K. S. (1977) . Status characteristics, reward allocation, and equity.
Sociometry, 40, 311-324 .
Riches, P., & Foddy, M . (1989) . Ethnic accent as a status cue . Social Psychology Quarterly, 52,
197-206 .
Ridgeway, C . L . (1988) . Gender differences in task groups : A status and legitimacy account. In :
M . Webster, Jr. & M. Foschi (Eds), Status Generalization: New Theory and Research (pp .
188-206 and 495-497) . Stanford, CA : Stanford University Press.
Ridgeway, C . L ., & Berger, J. (1986). Expectations, legitimation, and dominance behavior in task
groups . American Sociological Review, 51, 603-617 .
Shelly, R. K., & Munroe, P. T. (1999) . Do women engage in less task behavior than men?
Sociological Perspectives, 42, 49-67 .
Sobieszek, B . (1972) . Multiple sources and the formation of performance expectations . Pacific
Sobieszek, B ., & Webster, M . Jr . (1973) . Conflicting sources of evaluations . Sociometry, 4,
550-560 .
Troyer, L ., & Younts, C. W. (1997) . Whose expectations matter? The relative power of first-and
second-order expectations in determining social influence . American Journal of Sociology,
103,692-732.
Wagner, D . G ., & Berger, J. (1993). Status characteristics theory : The growth of a program . In : J .
Berger & M . Zelditch, Jr. (Eds), Theoretical Research Programs: Studies in the Growth of
Theory (pp . 23-63 and 454-463) . Stanford, CA: Stanford University Press .
Wagner, D . G ., & Berger, J. (1997) . Gender and interpersonal task behaviors : status expectation
accounts . Sociological Perspectives, 40, 1-32 .
Walker, H . A ., & Cohen, B . P. (1985) . Scope statements : Imperatives for evaluating theory .
216
Webster, M . Jr. (1969) . Source of evaluations and expectations for performance . Sociometry, 32,
243-258 .
Webster, M . Jr. (1977) . Equating characteristics and social interaction : Two experiments .
Sociometry, 40, 41-50 .
Webster, M . Jr., & Driskell, J. E . Jr. (1978) . Status generalization : A review and some new data .
Webster, M . Jr., & Foschi, M . (1988) . Overview of status generalization. In: M. Webster, Jr. & M .
Foschi (Eds), Status Generalization : New Theory and Research (pp. 1-20 and 477-478) .
Webster, M . Jr., & Sobieszek, B. (1974) . Sources of Self-Evaluation : A Formal Theory of
Significant Others and Social Influence . New York, NY: John Wiley.
Wiley, M . G. (1995) . Sex category and gender in social psychology. In : K . S . Cook, G. A . Fine &
J. S . House (Eds), Social Psychology: Sociological Perspectives (pp . 362-386) . Boston,
MA : Allyn and Bacon .
Wood, W., & Karten, S . J. (1986) . Sex differences in interaction style as a product of perceived sex
differences in competence . Journal of Personality and Social Psychology, 50, 341-347 .
Zeller, R. A ., & Warnecke, R . B . (1973) . The utility of intervening constructs in experiments .
Sociological Methods and Research, 2, 85-110 .
THE
CAMILLERI-BER
ER
MODEL
REVISITED
James W. Balkwell
ABSTRACT
More than three decades ago, Santo Camilleri and Joseph Berger
carried out a set of experiments on decision making and social influence.
In their experimental setting, two subjects worked together on a task . The
dependent variable was whether a subject would accept or resist the
other's influence, given a disagreement between them. One independent
variable involved a subject's ability compared with that of her or his
partner, a second involved the subject's responsibility for the team's final
decisions. Then, researchers did not have access to the statistical and
computational technology available today, so Camilleri and Berger
(1967) did not analyze their experimental data rigorously in terms of their
model. Doing so reveals surprisingly supportive results, especially after
some fine tuning based on more recent work. Perhaps most importantly,
this suggests what may be a promising approach to contemporary
questions about sentiment and task-group processes .
INTRODUCTION
Recently there has been much discussion about how to integrate such important
phenomena as sentiment and formal control into expectation states theory and
research (Shelley, 1993 ; Lovaglia & Houser, 1996 ; Wagner, 1998) . In other
Advances in roup Processes, Volume 18, pages 217-232 .

2001 by Elsevier Science Ltd .
ISBN: 0-7623-0767-6
217
218
JAMES W. BALKWELL
theoretical traditions - notably, group dynamics (Cartwright & Zander, 1968)

and social exchange theory (Homans, 1958) - these phenomena were handled
quite straightforwardly. At one time, many researchers considered expectation
states theory and these other approaches to be compatible . If this is less true
today, that may be due to a widely shared sense that models combining these
frameworks are unattractively eclectic and/or empirically inadequate [but see
Willer, Lovaglia, and Markovsky (1997) ; Thye (2000)] . Reevaluating these
beliefs was the motivation for this paper .
In the middle 1960s, Santo . Camilleri and Joseph Berger carried out a set
of experiments on decision making and social influence . Their experimental
setting involved two subjects working together on a task, their dependent
variable being whether a subject would accept or resist the other's influence,
given a disagreement between them . One independent variable involved a
subject's ability compared with that of her or his partner ; a second involved the
subject's responsibility for the team's final decisions . In a 1967 journal article
in what is now Social Psychology Quarterly, these researchers set forth a
theoretical model for the rejection of social influence that incorporated relative
ability and decision-making responsibility as its two predictors .
Why is it useful to reexamine a study carried out more than 30 years ago?
One reason is that this study is an impressive exemplar of how one might create
a model for a situation, based on pre-existing general theory . Camilleri and
Berger's work integrates social exchange ideas and the then relatively new
expectation states theory at least as agreeably as has any subsequent effort .
They derived not only their variables but also the functional form of their
model from prior theoretical considerations, providing an instructive example
of theoretically-based model construction . On the less exemplary side,
Camilleri and Berger carried out a weak data analysis whose results lent the
impression that their model was empirically questionable and that the
discrepancies between their observations and their predictions were patterned,
suggesting systematic shortcomings in expectation states theory and/or in the
other theoretical ideas embodied in their model . The weakest part of their
contribution has overshadowed the strongest parts, which illustrates the
damage that may result from weak or insufficient statistical analyses .
In fairness, these writers did not have available to them in the middle 1960s
the statistical and computational technologies that contemporary researchers
may take for granted . Below I shall report on my reanalysis of Camilleri and
Berger's data, using their own theoretical assumptions and approach in
conjunction with a more up-to-date and defensible methodology . To set the
stage for this presentation, I now describe Camilleri and Berger's research .
The Camilleri-Berger Model Revisited
2 19
CAMILLERI AND BER ER'S RESEARCH

At the beginning of each experimental session, the researchers gave each of two
subjects a test of "spatial judgment ability ." At the conclusion of this test,
depending on the experimental condition, they informed the subjects that one
of them, Person A, had scored either high (17 correct out of 20) or low (8
correct out of 20) . The other, Person B, also purportedly had scored either high
or low. There are four logically possible ability combinations, which they
designated (+ + ), (+ - ), (- + ), and (- - ), the first component referring to
the focal actor's ability, the second referring to her or his partner's ability . After
this ability manipulation, the researchers asked the participants to engage in a
cooperative task involving a sequence of two-step binary decisions . On each of
20 trials in this second part of the session, each participant was : (1) presented
with a "spatial judgment" problem, (2) given five seconds to select an initial
answer, (3) provided with her or his partner's initial answer, and (4) then given
five more seconds to select a final answer. Only the participants' final answers
were to count toward their team score, which the experimenter asked them to
try to maximize . The subjects' feedback on their initial choices was
manipulated, making it appear to them that, on each critical trial, they had
disagreed.
The most distinctive part of Camilleri and Berger's study was their second
explanatory factor : responsibility for making the team's final decisions . One
level of this factor involved the focal actor having no responsibility ; a second
involved the two actors having equal responsibility (meaning that each person's
final answer potentially added 1/2 point to their final team score) ; a third
involved the focal actor having full responsibility for making the team's final
decisions. This responsibility factor was crossed with the relative ability
variable I just described, producing 12 distinct experimental conditions . They
employed a fully factorial 3 x 4 experimental design . Let us now turn to
Camilleri and Berger's theoretically-based model .
CAMILLERI AND BER ER'S THEORETICAL MODEL

In Camilleri and Berger's (1967) paper, the probability that the focal actor
makes a stay response, as opposed to a change response, is conceptualized as
the gain (i .e . subjective expected utility to the actor in question) from staying,
divided by the sum of: (1) the gain from staying and (2) the gain from
changing . ains are defined so that they are always positive, the gain from a
particular alternative equalling the benefits from choosing that alternative, each
multiplied by its probability of occurrence, minus the costs from choosing
22 0
JAMES W. BALKWELL
other alternatives, each of these multiplied by its probability of occurrence . In

this setting, the costs (or potential costs) are disapproval from the focal actor's
partner, also from the experimenter, if the focal actor makes an incorrect final
decision, and self disapproval if the focal actor changes from her or his initial
choice in making her or his final choice . Regarding the latter, the hypothesis is
that self-consistency is rewarding and inconsistency is aversive (Homans, 1961 :
97) . Moreover, Camilleri and Berger explicitly assumed that a negative
outcome avoided is equivalent to a positive outcome, an idea also contained in
Homan's explication of his social exchange theory (1961 :58 ; see also
estinger, 1957 : 40) . After stating their theoretical assumptions and formalizing these algebraically, Camilleri and Berger (1967 : 130) derived the
following expression for P(S), the probability of staying (for the details of this
derivation, see the Appendix) :
P(S) _
u, +au2 +u)
u,+u2 +u 3
(1)
In this representation, u, denotes the utility to the focal actor of self-consistency

(not flip-flopping between initial and final judgments) ; u 2 denotes the utility to
the focal actor of approval from her or his partner ; and u 3 denotes the utility to
the focal actor of approval from the experimenter. inally, a denotes the focal
actor's subjective probability that he or she is correct, given a disagreement
between the two partners on their initial judgments . In a particular experimental
condition, these quantities are fixed in value ; however, Camilleri and Berger
assumed on theoretical grounds that a and u2 each would vary across their
experimental conditions . Using Eq. 1, it is easily shown that, if we restrict the
utilities to be non-negative, P(S) falls in the range required of a probability,
namely, between 0 and 1 .
An Actor's Expectation Advantage
The contemporary theoretical construct of "expectation advantage" did not
appear in the literature until 10 years after Camilleri and Berger's work was
published, but the quantity a in Eq . 1 clearly is a precursor of this subsequent
development . In the 1977 graph-theoretic formulation (Berger, isek, Norman
& Zelditch, 1977 : 91-134), the critical explanatory variable is essentially this
quantity, called the focal actor's expectation advantage over her or his partner.
Berger et al . (1977) understood this as the focal actor's performance
expectation for self minus that he or she holds for the other, denoted es - e0 .
There can be little question that Camilleri and Berger (1967) conceived of
expectations in the same way as Berger et al . (1977) . They saw these as
221
deriving from actors' status characteristics, including specific abilities, and

they saw these as being relational in nature . They assumed a (the subjective
probability in Eq . 1) to be a direct function of the focal actor's expectations for
self and for the other . The language employed in describing a in 1967 and in
describing e, - e o in 1977 is so consistent that we must assume there is a simple
functional relationship between them.
There are established procedures for calculating the numerical values of
es - e0 ; for the purposes of this study, I shall simply note that e s - e o equals zero
in the (+ +) and (- -) conditions, 1 .774 in the (+ -) conditions, and
- 1 .774 in the ( - +) conditions (for details, see Berger et al ., 1977 : 122-134 ;
Balkwell, 1991b : 165-169 ; Balkwell, 1999) .
Now Let's Put the Pieces Together
Equation 1 requires ratio-scale measurement of a, u,, u 2 , and u 3 . The numerical

values of the u's (utilities) are not unique, however, for we can multiply each
utility by the same constant without changing the model's predictions . Since
Camilleri and Berger assumed that u3 had the same value across all 12 of their
experimental conditions, we can assign any convenient value to u 3 , which only
fixes the unit of measurement . Let this measurement unit be chosen such that
u3 =- 1 in each condition.'
To parameterize Eq . 1 for statistical estimation, let the focal actor's
proportion of responsibility for the team's final decisions be denoted x . In the
no responsibility conditions, x equals zero ; in the equal responsibility
conditions, x equals one-half; and in the full responsibility conditions, x equals
one . Camilleri and Berger's (1967) verbal assumptions can be algebraically
rendered as follows :
(2)
The quantities (3,, (3 2, and R3 are fixed parameters whose values are unknown ;
we must estimate these from the experimental data . Substituting Eqs . 2 into Eq .
1 and rearranging the resulting expression yields the following (for details of
the derivation, please see the Appendix) :
22 2
JAMES W. BALKWELL
1
+(R2x+ 1)R3(es - e0)

1 2R'
P(S) = I + (3)
2
(3, + 02X+ 1
When researchers refer to "the Camilleri-Berger model," they typically mean
Eq . 1 . Nevertheless, using assumptions that are quite explicit in Camilleri and
Berger (1967), we can derive Eq . 3 . rom the latter we can see that the
Camilleri-Berger model is straightforwardly interpretable as an expectation
states model . Indeed, if decision-making responsibility were experimentally
held constant, Eq . 3 would be merely a linear function of the focal actor's
expectation advantage, empirically indistinguishable from the function set
forth by Berger et al . (1977 : 131) . More generally, the Camilleri-Berger model
includes the Berger et al . (1977) model as a special case . Not only is a a
theoretical precursor of es - e0 , but u, is a precursor of the Berger et al . (1977)
formulation's m. 2 Also of interest, Eq . 3 asserts a contingency involving
expectations and decision-making responsibility : the effect of expectations on
P(S) depends on the focal actor's responsibility for the team's final decisions .
Now let's consider the empirical adequacy of the Camilleri-Berger model,
using the most statistically tractable form of this model, that is, Eq . 3 .
AN EMPIRICAL REEVALUATION
or purposes of statistical estimation, Eq . 3 can be taken as the systematic
component of a statistical model, expressing the focal actor's probability of
staying as a function of her or his (1) expectation advantage and (2) decisionmaking responsibility. The random component of the model, representing
deviations from this systematic component, can be taken as the binomial
distribution . It is well known that the assumption of binomial random variation
is a weak one, leading to the same parameter estimates as various
alternatives .'
Results
rom the Camilleri-Berger data, we find ~, = 0 .529, R2 = 2 .65, and R3 = 0 .142 .
Table 1 presents the empirical proportions of stay-responses, the corresponding
parametric predictions calculated from Eq . 3, and the discrepancies between
these for each of Camilleri and Berger's (1967) 12 experimental conditions .
Although there are some discrepancies between the observations and the
corresponding predictions that cannot reasonably be attributed to chance alone
(X2 = 114 .246, d .f. = 9, p < 0 .001), the model nonetheless accounts for most of
223
N-00 ~O
r'0
N 7M SV SM N hN N
7 O\ 00
~O '0
r
coo coo 0 0 . 0 0
qqqoqooq
g
O
C
w
0
w
0
U
.b
U y
C- -~MN ~O a, N \O 0 0 00 N
00
N ~O
5 5C 00~OV10 ~O
N 0h rM 00
M O
0 0 0 0 0 0 0 0 0 0 0 0
a
ou
.d
A
pa
O
w
a>
on
s~
a>
P
rn a, -moo rn a S S o 0 C 0
N r 000 In S W) z N N en en
0 0 0 0 0 0 0 0 0 0 0 0
oa
0 0
00
0 ONO
0
~
o o 0
~0 N
\CCg
w
c
a
00
o n o o n o o n o o n o
-oo-~oo oo-moo
b
y
O
0
a
-O
N
00
C
I
5
O O N r N
N N N 0
0 0
0 0
o O
O C
o N N h
---~oooooo I I I
C
V
C
N
V
II
Cl
0
00
U
a
O = 7
z0 w z w u~ z
iii+++Ii7+++
OQ
w w
z 2 z
U
f
.a
224
JAMES W. BALKWELL
the variation in numbers of stay-responses across the experimental conditions

(R 2 = 0 .867) . It seems likely that there are some social processes operating in
the Camilleri-Berger experimental setting that are not captured by this model,
yet overall the model's adequacy is impressive .
INCORPORATIN MORE RECENT EST

DEVELOPMENTS
While the predictive power of Eq . 3 is considerable, I anticipate that it could be
further improved if we were to incorporate into the model more recent
hypotheses, applying part of the logic behind Camilleri and Berger's work with
ideas from more contemporary work . The approach I favor is to model rates or
amounts of behavior (as Camilleri and Berger essentially did), then to derive
quantities such as P(S) from these and from the usual definition of a
proportion .'
ollowing Balkwell (1991a), let B(t) denote the expected behavioral
production of some specified kind in a period of length t . Assuming no
externally induced changes in the focal actor's definition of the situation, this
would be proportional to t, although in general it might be a more complicated
monotonic function of t. Based on a theoretical argument, Balkwell (1991a :
357-358) derived the following function :
B(t)=texp[g1e,+g2e2+ . . . +gkek]
(4)
In this equation, denotes a baseline level of behavioral output . This is a

hypothetical level that would characterize the behavior in question if the period
were of length one and the members of the group were not differentiated by
status characteristics .
In situations such as that of Camilleri and Berger (1967), the behaviors of
interest are staying and changing. Under some simplifying assumptions,
Balkwell (1991a : 359) obtained the following expression for P(S) :
P(S)=
exp[m+q(e,
- e )]
1 + exp[m + q(es - e0)]
( 5)
This is equivalent to a logit function . The quantity m == log(,/ o) is the

logarithm of the ratio of the baseline rates of staying and changing . or the
experiments Balkwell discussed, , and o both would have been constant
across experimental conditions, making m simply a parameter to be estimated
from the data. However, for an experiment where the focal actor's control over
her or his team's final decisions varies across conditions, the following
specification is more appropriate :

Rs = exp(as + b x)
R o = exp(a o + box)
m =1og(p J0) = 'YO
225
(6)
+ -Y
ix
In the last line, yo =- as - a,, and y, _= bs - b0 .

In Camilleri and Berger's (1967 : 372) terms, as reflects the utility to the focal
actor of self-approval stemming initial-final choice consistency, a o reflecting
the value of self-disapproval stemming from inconsistency . Assuming the two
choices are equally attractive aside from this consistency consideration, as
should be greater than a o and thus yo should be positive . Also based on
Camilleri and Berger's (1967 : 372) theoretical discussion, we would expect b s
to be negative and b o to be positive, making y, negative . Substantively, y,
indexes the value of approval from the subject's partner, which in part would
reflect the subject's sentiment toward her or his partner . Approval from
someone who is liked or respected is more valuable than that from someone
who is not. While liking or other sentiment for the partner would not have
varied systematically across Camilleri and Berger's experimental conditions,
we can envision future experiments in which sentiment will be a manipulated
experimental variable . Building sentiment-effects into this model would be
straightforward.'
We can combine Eq . 5 and the last line in Eqs . 6 to obtain a model for P(S)
in the Camilleri-Berger experiment that reflects both certain aspects of
Camilleri and Berger's logic and more recent developments in expectation
states theory (i .e . Berger et al ., 1977 ; Balkwell, 1991a) .
P(S) = exp[-Yo+'Y,x+y2(es - eo)7
(7)
1 +exp[yo+'Ylx+'Y2(es - eo)]
or the sake of a simpler notation, I have renamed Eq . (5)'s q as y 2. Based on
expectation states theory (Berger et al ., 1977 : 130), y2 should be positive in
sign . Like Eq . 3, this specification entails three fixed parameters whose values
are unknown and therefore must be estimated from the experimental data .
Empirical Evaluation
rom the Camilleri-Berger data, we find yo = 0.848, y, _ - 0 .698, and
'Y2 =0 .530 . Table 2 presents the empirical proportions of stay-responses, the
corresponding parametric predictions calculated from Eq . 7, and the discrepancies between these for each of Camilleri and Berger's 12 experimental
conditions .
As with the original model, there are some discrepancies between the
observations and the corresponding predictions that cannot reasonably be
22 6
JAMES W. BALKWELL
N
O C ('1
pp~ N N N C N N
O O O O N
qqqOqOgOq
00
O O O O
k.
w
O
U 7
~t V N h M N V1 M 01 M
--~
00 00 ~O N N 0 N N N .Nr
N
r 0D Dip N
N M Np
M M 7
0 0 0 0 0 0 0 0 0 0 0 0
x
n
b
A
r
Q"
N
C O N V1 O ~ M M
N r
N N
00 O
lD ID N n lD r N ~
O O O O O O O O O O
I
N
N
N
O O g 0 0 0 g 0 0 0 0
~v~'~b~ CN C
N
H
4K
U
N
L
4 .1
4-I
O
QO
U
O v1 0 0 kn O O W) O O vn O
0 0 O O --~ C O -~ O O
b
N
L
M
T
O
fS'
b
b
000000-I i
d
U
a
W
O
V
L"
n
o
I
M
a
i
N
C
0
O
U
7 0 7 0
a 0
7 0
w zz u . ~ z w Zz w zz
I I I+++ I I I+++
++++++ I I I I I I
7
C!'
U
.a
227
attributed to chance alone (X'= 59 .435, d .f. = 9, p < 0 .001) . Nevertheless, the
model accounts for the great preponderance of the systematic variation in P(S)
from one experimental condition to another (R2 = 0 .931) ; moreover, the signs of
the parameter estimates all are in the predicted directions . inally, the
explained variation is notably higher for this model than for the original one
(93 .1% vs . 86 .7%) .
CONCLUSIONS
What can we conclude from these results that differ from what Camilleri and
Berger concluded? One thing is that the fit revealed in Tables 1 and 2,
especially the latter, is much better than that revealed by Camilleri and Berger's
analogous table . Much of their discussion about how to improve their model
may be irrelevant in the light of these results . Second, their analysis seemed to
suggest that expectation states theory may apply more adequately to high status
actors than to low status actors, a finding that I have seen referred to in several
other published papers . When we compare the observed and predicted P(S)
values for (+ #) and (- #) actors, we see little evidence that the model
predicts more adequately for the former than for the latter . It seems to me that
the model gives satisfactory predictions for both, there being no obvious
systematic difference . The earlier sense that EST does not apply as well to low
status actors as to high status actors appears to have rested on a defect in
Camilleri and Berger's statistical methodology .
lawed methodology or not, the strengths of Camilleri and Berger's paper
still merit attention 30 years after its publication . It provides a smooth and
pleasing theoretical integration of social exchange principles and status
processing principles . The authors derived not only their variables but also the
functional form of their model from prior theoretical ideas, something that we
too rarely find in contemporary work . In current expectation states theorizing,
there may be an inclination to throw out the social exchange ideas, replacing
these with special-purpose ideas that are not part of cumulative knowledge .
Related to this, there is an inclination to simply posit statistical interactions and
other effects, rather than deriving these from more basic theoretical
considerations . Even when such "posited" effects fit empirical data tolerably
well, the results have minimal implications for cumulative knowledge, because
we do not know why they occur . Camilleri and Berger's work from the 1960s
goes well beyond this and simultaneously does a much better job than once
seemed to be the case of accounting for empirical data .
Epitomizing the basic approach these remarks imply, there recently have
appeared several kindred bridge-building efforts that seek to combine
22 8
JAMES W. BALKWELL
expectation states theory with social exchange formulations (e .g . Willer,

Lovaglia & Markovsky, 1997 ; Thye, 2000) . Because of their focus on deriving
interesting consequences from more basic and general assumptions - as
Camilleri and Berger did - these recent endeavors may be particularly
promising .
The efforts reported in this paper may be directly relevant to contemporary
work on sentiment, performance expectations, and social behavior (Shelly,
1993 ; Lovaglia & Houser, 1996 ; Shelly & Troyer, 1998 ; isek & Berger, 1998) .
In his excellent presentation at the 11th Annual roup Processes Conference,
David . Wagner (1998) distinguished between a mediating conception and a
constituent conception of sentiment's role in the relationship between
performance expectations and status-related behavior. Conceptualizing sentiment as a component of performance expectations, as if it were like a status
characteristic, epitomizes the constituent approach . The Camilleri-Berger
model, on the other hand, represents a mediating approach . Sentiment and
formal control are easily handled within a social exchange framework . In
theories dating back at least as far as Ross (1921 : 131) and estinger,
Schachter, and Back (1950 : 72-100), the principle has existed that the greater
the positive sentiment a focal actor has for another, the more the focal actor's
behavior will be shaped by a concern for the other's approval . As Camilleri and
Berger (and this paper) have shown, a social exchange framework and an
expectation states framework can be complementary ; using them together in
constructing models for specified kinds of situations can produce appealing
formulations . Camilleri and Berger's seemingly forgotten paper from the 1960s
remains an exemplar for addressing the cutting-edge theoretical issues of
today .
APPENDIX
Derivation of Equations
eatures of the Situation . On each critical trial in Camilleri and Berger's (1967)
experiment, the subject made an initial choice between two alternatives,
received feedback that her or his partner had made the other choice, then made
a final choice that counted towards the "team score ." In this setting, taskorientation implies that the focal actor is concentrating on getting correct final
choices ; Camilleri and Berger's experimental instructions were designed to
insure this .
Let B, denote the benefit to the focal actor of consistency between her or his
initial and final choices, B 2 the benefit of approval from her or his partner, and
229
B3 the benefit of approval from the experimenter . Let C,, C 2 , and C3 denote the
costs of self-inconsistency, partner disapproval, and experimenter disapproval,
respectively. Because a cost is negative in its behavioral significance, this
formulation expresses the C's as negative numbers . inally, let a denote the
focal actor's subjective probability that he or she is correct, given a
disagreement on initial choices . Below is a matrix describing the focal actor's
potential benefits and costs from staying or changing on an experimental trial,
given that her or his initial choice is correct (which it is with probability (X) or
incorrect (which it is with probability 1 - a) .
Initial Choice Is :
Correct
(prob = a)
Incorrect
(prob =1 - a)
Stay
B,, B 2, B 3
B,, C 2 , C 3
Change
C,, C2 , C3
C,, B 2, B 3
inal choice
Equation 1 . The principle is that an actor's expected gain from an alternative

equals the benefits from that alternative, each weighted by its associated
probability, minus the cost from the other alternative(s), each weighted by its
associated probability . Applying this principle in this situation yields :
S = [aB, + aB2 + aB 3 + (1 - a)B,]
-[aC,+aC2+aC3+(1-a)C,]
= (B1 - C I ) +a[(B2 - C2)+(B3 - C3)]
= u,+a(u 2 +u 3 )
,= [1 -a)B2 +(1 -a)B3]
- [(1 - a)C2 +(1 - a)C3]
= 1 - a)[(B2 - C2) + (B3 - C3)]

= (1 - a)(U2 + u 3 )
Because each B is greater than or equal to zero, and each C is less than or equal
to zero, the substitutions, u; =- R; - C; (i = 1, 2, 3), denote quantities that are
greater than or equal to zero in each case . rom Camilleri and Berger's
experiment, the relative significance of benefits and costs cannot be
distinguished ; u 1 , u2, and u 3 denote the combined utility of the benefits and
costs from their respective sources .
Using Camilleri and Berger's theoretical assumption that P(S) = S/( S+ ),
Eq . 1 follows immediately :
23 0
JAMES W. BALKWELL
P(S) = u 1 + WU2+ u3)

u 1 +u 2 +u 3
( 1)
Equation 3. Note that we can write P(S) = X + all - X), where X equals u,/
1
(u 1 + u2 + u3 ) . rom Equations 2, we have a = 2 + 13 3 (e, - e0 ), and we can deduce
that X =13,/(13 1 + 132x + 1) . Making these substitutions :
P(S) =X+a(1 - X)
1
= +[2+133(e,-eo)][1-X]
1
=X +2 (1 - X) + 133(e, - eo)(l - X)
(3)
1 1
= 2 + 2X + 133(e, - eo)(1 - A)
1
2
1
2131 + (132x + 003(e, - eo )
_-+
13 1 +132x+1
The last line is Eq . 3 as presented in the text of this chapter .
Berger et al. unction. One unrecognized feature of the Camilleri-Berger
model is that it includes Berger, isek, Norman and Zelditch's (1977 : 131)
function for P(S) as a special case . To see this, we rearrange Eq . 3 and then
make some substitutions based on Eq. 2 .
1
131 + (132x + 1)133(e, - eo)
1 2
P(S)2+
131+132x+ 1
1
1 + 2131
+
(2 131+132x+1)
(2
u1
1
-u
2
u 2 1 + u3) +
133(132x+ 1)
es - e
131+[32x+1)(
~)
0
(13 u2+u3)
3+u
(es - eo)
= m + q(es - e 0)
The quantities m and q of the Berger et al . (1977 : 131) specification typically
have been taken as fixed parameters to be estimated from the research data .
231
Empirically based estimates of m and q, however, typically vary from one

study to another, suggesting that these parameters are not constant over a wide
range of situations . Based on this discomposition, we might expect both m and
q to vary with any features of a setting that affect the values of u,, u 2, and/or
U3.
NOTES
l . If sentiment toward the experimenter were a research variable, as it might be in
future studies, we would expect the value of u 3 to vary across the experimental
conditions, in which case this way of fixing the measurement unit would not be
appropriate .
2 . Berger et al .'s (1977) model is P(S) =m+q(e, - e0 ) . Expressed in the notation of
Eq. I'M= {1 + [u,/(u, + u2 + u3)1 } . Whenever u, = 0, m = 2. If u, > 0, m >
2
2. Variations
in estimates of m, which we often find across studies, may be due primarily to variations
in whatever experimental features affect the value to the focal actor of self-consistency.
One possibility is that emotional arousal enhances this (cf. Lovaglia & Houser, 1996) .
3 . In practice, subject heterogeneity and non-independent trials may produce overdispersion (response variation exceeding expected binomial variation) . This does not
affect the coefficient estimates or (after appropriate adjustment) their standard errors . It
does invalidate a strict interpretation of the X2 goodness-of-fit test.
4. An alternative is simply to posit a linear relationship between the focal actor's
expectation advantage (e s - e0) and the dependent variable of interest [e .g . P(S)J . Such
linear specifications may be inconsistent among themselves (Balkwell, 1991 : 356) and
they clearly fail to take account of the logical constraints on proportions, which are
bounded by zero and one .
5 . Research in the group dynamics tradition has shown that pressures toward
uniformity of opinion within a group vary with group cohesiveness, one important
measure of the latter being the extent to which the group's members like one another
(e .g. Back, 1951 : 12) . rom research in this tradition, we may reasonably infer that the
value of others' approval varies directly with liking for those others (also, see Homans,
1958 : 600-601) . In Camilleri and Berger's setting, this suggests that u 2 and u3 may vary
directly with the focal actor's liking for her or his partner and for the experimenter,
respectively.
ACKNOWLED MENTS
An earlier version of this paper was presented at the 93rd Annual Meeting of
the American Sociological Association, San rancisco, August 1998 . I want to
thank a Coeditor and one reviewer of Advances in roup Processes for
suggestions that strengthened my presentation .
232
JAMES W. BALKWELL
RE ERENCES
Back, K . W. (1951) . Influence Through Social Communication . Journal of Abnormal and Social
Balkwell, J . W. (1991a) . rom Expectations to Behavior : An Improved Postulate for Expectation
States Theory . American Sociological Review, 56, 355-369 .
Balkwell, J . W. (1991b) . Status Characteristics and Social Interaction : Assessment of Theoretical
Variants. In : E. J . Lawler, B . Markovsky, C . Ridgeway & H . A . Walker (Eds), Advances in
roup Processes, Vol . 8 (pp. 135-176) . reenwich, CT: JAI Press .
Balkwell, J . W. (1999) . Expectation States: Calculating An Actor's Value of ex . Unpublished
manuscript accessible at the Internet address http ://www.geocities.com/jwbalkwell/
expect .html (includes a utility for obtaining numerical es and eo scores) .
Berger, J ., isek, M. H ., Norman, R. Z ., & Zelditch, M. Jr. (1977) . Status Characteristics and
Social Interaction : An Expectation States Approach . New York : Elsevier.
Camilleri, S . ., & Berger, J. (1967) . Decision-Making and Social Influence: A Model and an
Experimental Test. Sociometry, 30, 367-378 .
estinger, L . (1957) . A Theory of Cognitive Dissonance . Stanford, CA: Stanford University
Press .
estinger, L ., Schachter, S ., & Back, K . W. (1950) . Social Pressure in Informal roups . New York:
Harper.
isek, M . H., & Berger, J. (1998) . Sentiment and Task Performance Expectations . In: J . Skvoretz
& J . Szmatka (Eds), Advances in roup Processes, Vol . 15 (pp . 23-40) . reenwich, CT:
JAI Press .
Homans, . C . (1958) . Social Behavior as Exchange . American Journal of Sociology, 62,
597-606 .
Homans, . C . (1961). Social Behavior: Its Elementary orms . New York: Harcourt, Brace, and
World .
Lovaglia, M . J ., & Houser, J . A. (1996). Emotional Reactions and Status in roups . American
Ross, E . A . (1921) . Principles of Sociology. New York: Century .
Shelly, R. K. (1993) . How Sentiments Organize Interaction . In : E . J. Lawler, B . Markovsky, C.
Ridgeway & H . A . Walker (Eds), Advances in roup Processes, Vol. 10 (pp . 113-132).
reenwich, CT: JAI Press.
Shelly, R . K., & Troyer, L. (1998) . Speech Duration in Emergent Structures in Task roups.
Presented at the 93rd Annual Meeting of the American Sociological Association, San
rancisco.
Thye, S . R. (2000) . A Status Value Theory of Power in Exchange Relations . American Sociological
Review, 65, 407-432 .
Wagner, D . . (1998) . Expectation States Theory : Developments, Challenges, and Opportunities .
Presented at the 11th Annual roup Processes Conference, San rancisco .
Willer, D ., Lovaglia, M. J ., & Markovsky, B . (1977). Power and Influence : A Theoretical Bridge.
Social orces, 76, 571-603 .

Advances in Group Processes Vol 18

Încărcat de

Informații document

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Advances in Group Processes Vol 18

Încărcat de

Drepturi de autor:

Formate disponibile

LIST OF CONTRIBUTORS

Department of Sociology, University of

Department of Sociology, Florida Atlantic

Department of Psychology, Cornell

Department of Anthropology and

Department of Anthropology and

School of Psychology, University of

Department of Sociology, Indiana

Faculty of Technology Management,

Department of Anthropology and

Department of Social and Decision

Department of Economics and

Department of Sociology/ICS, Utrecht

Leaf Van Boven

Faculty of Commerce,University of British

FROM PROTOTYPICALITY TO POWER :

Social Identity And Leadership

documented decline in interest in groups (e .g . Steiner, 1974, 1986), that carried

Social Identity And Leadership

transformational objectives of leadership - leadership is a matter of

SOCIAL IDENTITY AND SELF-CATEGORIZATION

Social Identity And Leadership

differences between people from different groups - a process that perceptually

SOCIAL IDENTITY MODEL OF LEADERSHIP

Influence of the Prototype

Social Identity And Leadership

and thus most influential . However, under conditions of enduring contextual

Social Identity And Leadership

EMPIRICAL TESTS OF THE ROLE OF

Social Identity And Leadership

processes associated with self-categorization as a group member transform the

regression format . Leadership schemas, group membership variables, and

Social Identity And Leadership

LEADERSHIP AND POWER

Social Identity And Leadership

reconciliation of these contrasting situations hinges on the relationship between

From Leadership to Power

This intergroup relationship is grounded in a status differential that is

Social Identity And Leadership

unilaterally exploitative intergroup behavior. This is because the overarching

Social Identity And Leadership

A DARK SIDE TO PROTOTYPE-BASED LEADERSHIP

Social Identity And Leadership

actively innovative . Finally, there is an attribution through which followers,

Social Identity And Leadership

Fiedler, F. E . (1965) . A contingency model of leadership effectiveness . In : L. Berkowitz (Ed .),

Social Identity And Leadership

Social Identity And Leadership

RATIONAL AND IRRATIONAL BASES

members are believed by most group members to be the most competent ;

Rational and Irrational Bases of Commitment to Group Hierarchies

an orderly flow of social interaction, are the primary motivational bases of

Rational and Irrational Bases of Commitment to Group Hierarchies

Rational and Irrational Bases of Commitment to Group Hierarchies

II. TWO ANALYSES OF THE STABILITY OF

Rational and Irrational Bases of Commitment to Group Hierarchies

are a number of factors that affect performance expectations . These include

in motion self-fulfilling interaction processes, demonstration of non-relevance

Rational and Irrational Bases of Commitment to Group Hierarchies

status characteristics exert more influence than diffuse status characteristics

Rational and Irrational Bases of Commitment to Group Hierarchies

A legitimated hierarchy rests not merely on beliefs about competence but on

that Goffman's analysis is appropriately applicable to task-oriented groups,

Rational and Irrational Bases of Commitment to Group Hierarchies