Perspective-taking: Decreasing stereotype expression, stereotype accessibility, and in-group favoritism

Abstract

Using 3 experiments, the authors explored the role of perspective-taking in debiasing social thought. In the 1st 2 experiments, perspective-taking was contrasted with stereotype suppression as a possible strategy for achieving stereotype control. In Experiment 1, perspective-taking decreased stereotypic biases on both a conscious and a nonconscious task. In Experiment 2, perspective-taking led to both decreased stereotyping and increased overlap between representations of the self and representations of the elderly, suggesting activation and application of the self-concept in judgments of the elderly. In Experiment 3, perspective-taking reduced evidence of in-group bias in the minimal group paradigm by increasing evaluations of the out-group. The role of self-other overlap in producing prosocial outcomes and the separation of the conscious, explicit effects from the nonconscious, implicit effects of perspective-taking are discussed.

Full text

Journal of Personality and Social Psychology Copyright 2000 by the American Psychological Association, Inc.
2000, Vol. 78, No~ 4, 708-724 0022-3514/00/$5.00 DOI: 10.1037//0022-3514.78.4.708
Perspective-Taking: Decreasing Stereotype Expression, Stereotype
Accessibility, and In-Group Favoritism
Adam D. Galinsky
Northwestern University
Gordon B. Moskowitz
Princeton University
Using 3 experiments, the authors explored the role of perspective-taking in debiasing social thought. In
the 1st 2 experiments, perspective-taking was contrasted with stereotype suppression as a possible
strategy for achieving stereotype control. In Experiment 1, perspective-taking decreased stereotypic
biases on both a conscious and a nonconscious task. In Experiment 2, perspective-taking led to both
decreased stereotyping and increased overlap between representations of the self and representations of
the elderly, suggesting activation and application of the self-concept in judgments of the elderly. In
Experiment 3, perspective-taking reduced evidence of in-group bias in the minimal group paradigm by
increasing evaluations of the out-group. The role of self-other overlap in producing prosocial outcomes
and the separation of the conscious, explicit effects from the nonconscious, implicit effects of
perspective-taking are discussed.
Navigating the contemporary social world has become some-
thing of an obstacle course. Any interpersonal behavior, verbal or
nonverbal, suggesting discrimination or favoritism based on group
membership is subject to condemnation with the potential for
public censorship and legal sanction, as well as internal compunc-
tion. How do individuals navigate their social world without dis-
playing attitudes that could be the fodder for accusations? What
strategies are effective in debiasing social thought, of calibrating
one's actions with contemporary social mores? Take the situation
of an individual interacting with an ethnic minority. An intuitively
appealing strategy to prevent the group's stereotype from affecting
the interaction is to actively try to prevent any references to that
stereotype from entering into consciousness. Unfortunately, a re-
cent line of research has shown that the intentional suppression of
stereotypic thoughts ironically can produce the very thoughts one
is suppressing (Macrae, Bodenhausen, Milne, & Jetten, 1994;
Wegner, 1994). An alternative strategy for social maneuvering
Adam D. Galinsky, Department of Organizational Behavior, Northwest-
ern University; Gordon B. Moskowitz, Department of Psychology, Prince-
ton University.
This research is based in part on the doctoral dissertation submitted by
Adam D. Galinsky to Princeton University and has benefited from the
comments of the committee members, Joel Cooper, Marcia Johnson, and
Dale Miller. Portions of this research were presented at the annual confer-
ence of the American Psychological Society, Washington, DC, May 1998.
We thank Ian Skurnik, Beno~t Monin, and Sarah Igo for their helpful
comments and encouragement throughout the life of the project. We thank
Eliot Smith for his helpful advice regarding the procedures for Experi-
ment 2. The research was supported by a National Science Foundation
(NSF) predoctoral fellowship to Adam D. Galinsky and by an NSF facil-
itation award.
Correspondence concerning this article should be addressed to Adam D.
Galinsky, Department of Organizational Behavior, Leverone Hall, 2001
Sheridan Road, Kellogg School of Management, Northwestern Univer-
sity, Evanston, Illinois 60208. Electronic mail may be sent to
agalinsky@nwu.edu.
708
involves the active consideration of alternative viewpoints, fram-
ings, hypotheses, and perspectives. Using three experiments, we
explored the processes of perspective-taking in reducing the ex-
pression and accessibility of social stereotypes, in increasing the
positivity of group-based judgments, and in eliminating in-group
favoritism. The results suggest that perspective-taking can reduce
the accessibility and application of stereotypic knowledge and that
perspective-taking reduces stereotypic responding because of in-
creased overlap between representations of the self and represen-
tations of the out-group.
Perspective-Taking
The ability to entertain the perspective of another has long been
recognized as a critical ingredient in proper social functioning.
Davis (1983) found that perspective-taking, as measured by an
individual-difference measure, was positively correlated with both
social competence and self-esteem. Piaget (1932) marked the
ability to shift perspectives as a major developmental breakthrough
in cognitive functioning, and Kohiberg (1976)recognized its im-
portance in his classification of moral reasoning. The presence of
perspective-taking can inspire great gestures of altruism (Batson,
1991, 1998), and its absence can incite the devastations of social
aggression (Richardson, Hammock, Smith, Gardner, & Signo,
1994). Early studies of perspective-taking focused on the emo-
tional reactions of participants induced to take the perspective of
an individual in need; these experiments were concerned with
demonstrating the existence of empathy. One could conclude from
the early experiments that perspective-takers' emotional experi-
ence comes to resemble that of the targets (see Batson, 1991). The
active consideration of imagining how a target is affected by his or
her situation produces an empathic arousal that leads the
perspective-taker to offer greater assistance to the target.
Perspective-taking also affects attributional thinking and evalu-
ations of others. Jones and Nisbett (1971) noted that actors and
observers differ in the attributions they make. Whereas actors are
likely to recognize situational forces pulling and pushing behavior

PERSPECTIVE-TAKING 709
in systematic ways, observers rely on others' dispositions as the
explanation for behavior. Although recognizing actors possess
privileged information about how they have acted in similar situ-
ations in the past, Jones and Nisbett also suggested actors and
observers process information differently because of divergent
perspectives. For the observer, the behavior of the individual is
salient; for the actor, attention is focused outward toward the
environment. Storms (1973) reversed the normal perspective of
actors through the use of a videotape; actors who observed their
own part of the conversation produced causal attributions that were
relatively more dispositional than situational. Regan and Totten
(1975) extended the research by Storms to the psYchological
shifting of perspectives--they turned dispositional explanations
into situational ones by asking participants to actively take the
perspective of the person they were observing. Perspective-takers
made the same attributions for the target that they would have
made if they themselves had found themselves in that situation.
Perspective-Taking and the Egocentric Self
Perspective-taking has been shown to lead to a merging of the
self and the other, in which the perspective-taker's thoughts toward
the target become more "selflike" (Davis, Conklin, Smith, & Luce,
1996). After perspective-taking, there is a greater self-target over-
lap, such that a greater percentage of self-descriptive traits are
ascribed to the target. The representation of the target constructed
by the perspective-taker comes to resemble the perspective-taker's
own self-representation.
Davis et al. (1996) found that ascription of self-descriptive traits
to the target was not due to increased liking for the target but rather
to the cognitive accessibility of the self-concept. Although
perspective-takers felt the target was more similar to themselves
than control participants and they liked the target more, these
effects did not mediate the amount of self-target overlap. This
suggests that there are two separate processes involved in perspec-
tive-taking: a conscious, explicit effect and a nonconscious, im-
plicit effect. When perspective-takers are asked direct questions
about the target person, then they will presumably feel that the
perspective-taking manipulation is relevant to that judgment and
consciously give responses that are consistent with that manipu-
lation. However, during perspective-taking, the self-concept gets
activated and applied toward the target. The activation and appli-
cation of the self-concept, like other knowledge structures (see
Bargh, 1997, for a review), occurs implicitly (at a nonconscious
level). Further evidence for the nonconscious effect of perspective-
taking manipulations on self-concept activation comes from the
fact that self-target overlap is unaffected by dividing the attention
of participants (Davis et al., 1996). Cognitive load interfered with
the effortful act of perspective-taking (the general ascription of
traits) but left the more automatic process (the ascription of self-
relevant traits) intact. Davis et al. concluded that the processes of
perspective-taking can be divided into more controlled and more
automatic ones and that "the effect of perspective-taking instruc-
tions on the ascription of self traits results from differences in
cognitive accessibility that are created by priming the self-
concept" (p. 723).
In explaining the effects of perspective-taking, researchers have
attempted to tease apart the differences between imagining how
another person feels and imagining how you would feel if you
were in the target's position. Stotland (1969) and Batson, Early,
and Salvarani (1997) found that although both types of
perspective-taking are associated with increased empathic feelings,
only the latter kind is associated with increased feelings of and
physiological manifestations of distress. Batson, Early, et al. sug-
gest that increases in feelings of distress are a sign of egoistic
motivation, a motivation they claim is distinct from altruism (Cial-
dini et al., 1987).
Imagining the self in the target's perspective is more likely to
spontaneously occur than imagining how another person is
uniquely affected by the situation confronting that person. The
probability of perspective-taking increases when one has endured
the same slings and arrows as the target person. Clore and Jeffery
(1972) found that traveling around campus in a wheelchair in-
creased sensitivity to the plight of people with disabilities. Prior
experience with a difficult situation or the realization that one will
confront a similar situation in the future increases empathic re-
sponding (Batson et al., 1996). In addition, relationship closeness
predicts perspective-taking and altruism; as relationship closeness
increases, so too does empathic responding and willingness to help
an individual in need (Cialdini, Brown, Lewis, Luce, & Neuberg,
1997).
Increased self-target overlap occurs both when participants
imagine themselves in the target's place and when they imagine
what it would be like to be the target (Davis et al., 1996). Thus,
although the emotional response does appear to be affected by the
type of perspective-taking manipulation (Batson, Early, et al.,
1997; Stotland, 1969), the cognitive consequences of perspective-
taking appear to be independent of the type of experimental
manipulation.
Perspective-Taking and Stereotyping
Given that perspective-taking produces many positive interper-
sonal benefits--even if egoistically motivated--these benefits
may extend to intergroup judgments and interaction. Focusing on
situational constraints and influences rather than relying on dispo-
sitionalism to explain the behavior of a stereotyped target might
reduce the infiltration of stereotypes into judgments ranging from
courtroom verdicts (Bodenhausen & Wyer, 1985) to employee
promotion decisions (Fiske, Bersoff, Borgida, Deaux, & Heliman,
1991) to dispositional inference (Duncan, 1976).
The increased self-other overlap after perspective-taking has
implications for intergroup relations because recent research has
found that it is the in-group's association with the self that leads to
ethnocentric responses in favor of the in-group (Cadinu & Roth-
bart, 1996; Smith & Henry, 1996). As the merging of self and
in-group increases, so too does the favoritism toward the in-group
(Turner, 1987). Just as the positive evaluation of the self extends
to the in-group, the increased self-other overlap after perspective-
taking could lead to more positive evaluations of the target, which,
in turn, might then generalize to the group as a whole.
The increased accessibility of the self-concept after perspective-
taking (Davis et al., 1996) might result in the use of the self-
concept over the stereotypic construct when categorizing and
evaluating a member of a stereotyped group. This should occur
because categories in the mind compete with each other to win the
metaphorical race to capture stimuli (Allport, 1954; Bruner, 1957).
When two constructs are equally applicable for categorizing an
individual, the more recently activated construct will be used. For
example, seeing an Asian woman putting on makeup increases the

710 GALINSKY AND MOSKOWITZ
accessibility of the female stereotype while inhibiting the Asian
stereotype (Macrae, Bodenhausen, & Miln e, 1995). Thus, the
increased accessibility of the self-concept might diminish the
accessibility and application of the stereotype by becoming the
dominant construct used to categorize ambiguous stimuli.
There is work to suggest that perceived connections to targets
can affect the use of stereotypes (Brewer, 1988; Neuberg & Fiske,
1987). The more personalized the contact is to a target, the less
likely it is that stereotypes will be used in the categorization
process (Brewer, 1996). Research on outcome dependency, a sit-
uation in which mutual collaboration between two or more part-
ners is essential for completing a task, has found a reduction in the
reliance on stereotypic information when forming impressions of
one's partners. According to Neuberg and Fiske (1987), outcome
dependency promotes attribute-based processing of information
with less reliance on stereotyped judgments because the cost of
nonveridical perception increases when more precise prediction is
sought.
It is possible that the positive benefits from outcome-
dependency accrue by means of the perspective-taking process.l
When placed in an outcome-dependent situation and desiring
accurate assessment of their task partner, participants might at-
tempt to take their partner's perspective in order to fully appreciate
the characteristics that their partner possesses. There are differ-
ences, however, between perspective-taking and outcome depen-
dency. The close analysis of individuating information triggered
by outcome dependency is more resource dependent than is the
ascription of self-traits to a target individual during perspective-
taking (Davis et al., 1996). Because perspective-taking does not
require the construction of an interdependent setting or creation of
a superordinate goal (Neuberg & Fiske, 1987; Sherif, 1966) to
exert its influence, it may function in more diverse domains than
task interdependence. Even if the cognitive processes of outcome
dependency and perspective-taking share significant overlap, it is
important to demonstrate that perspective-taking can produce the
same benefits as outcome dependency.
Thought Suppression
In the introduction, we mentioned that suppression is an intu-
itively appealing strategy for trying to prevent the accessibility and
application of social stereotypes. Debiasing social thought gener-
ally involves attempts by the individual to exert control over the
content and nature of cognitive processes. Mental control allows
the individual to direct cognition in the service of currently held
goals. Attempts at mental control, however, do not always meet
with success (Macrae et al., 1994; Wegner, 1994). Because a
representation of the to-be-suppressed thoughts must be held up as
an object somewhere in awareness in order to deny such thoughts
entrance into consciousness, the decision to engage in suppression
ironically can increase the accessibility of the unwanted thoughts
(Macrae et al., 1994). The continual search for instances of the
suppressed thought also serves as a form of repetitive priming
(Higgins, 1989; Macrae et al., 1994).
This heightened accessibility of a stereotype can manifest itself
during suppression when the operating system is disabled by
insufficient cognitive resources. Weguer, Erber, and Bowman
(1993) demonstrated that when resource depletion is paired with
intentions to suppress stereotypic thoughts, those very thoughts
become more accessible. The ironic effects not only emerge when
cognitive resources are scarce because of performing multiple
cognitive tasks, but the act of suppression itself is an effortful,
resource-demanding mental process that can create the conditions
facilitating these effects (Macrae, Bodenhausen, Milne, &
Wheeler, 1996; Sherman, Stroessner, Loftus, & Deguzman, 1997).
Ironic effects also emerge once the intention to suppress is
removed--these effects are known as
rebound effects
(Macrae et
al., 1994). Like any opponent-process system that involves simul-
taneous excitation and inhibition, removing the restraining force
leads to the temporary hyperaccessibility of the inhibited process
or thought. Stereotype suppressors, despite showing reduction in
stereotypic thoughts on a first task, subsequently produced more
stereotypic thoughts, greater avoidance of stereotypic targets (mea-
sured through physical distancing), and faster response latencies in
recognizing stereotypic words than did participants who had fo-
cused on the stereotype rather than suppressed it.
Although suppression may intuitively seem to be a viable pro-
cess for preventing stereotypic thoughts from emerging, the above
review indicates three potential pitfalls in its effectiveness. First,
an abundance of cognitive resources is required, without which,
the monitoring system's scan across the mental landscape proceeds
without restraint. Second, suppression can serve as a form of
cognitive load, disabling processes of individuation (Neuberg &
Fiske, 1987). Third, suppressed thoughts often become more,
rather than less, salient after the termination of the suppression
goal.
Experiment 1
Perspective-taking and suppression as goals for affecting the
accessibility of constructs and interpersonal interactions differ in a
number of important ways. On the one hand, perspective-taking
leads to a more personalized approach to the target, which can be
effective in reducing the accessibility of stereotypes (Brewer,
1996). On the other hand, suppression maintains focus on group-
level characteristics and the target as an instantiation of the group.
Perspective-taking covaries with perceived similarity with targets
(Cialdini et al., 1997; Davis et al., 1996), whereas suppression
leads to both psychological (Galinsky & Moskowitz, 1999) and
physical distancing behaviors (Macrae et al., 1994). Perspective-
taking leads to the accessibility of the self-concept (Davis et al.,
1996), whereas suppression leads to the hyperaccessibility of the
stereotype (Macrae et al., 1994). The first experiment was de-
signed to investigate the processes associated with thought sup-
pression and perspective-taking in the context of stereotypes. To
explore the processes by which perspective-taking could affect
stereotyping and intergroup relations, the first two experi-
ments focused on both the explicit and implicit effects of
perspective-taking.
To test whether perspective-taking could have positive inter-
group, and not just interpersonal, consequences, we borrowed a
paradigm that Macrae et al. (1994) used to explore the conse-
quences of stereotype suppression. In the Macrae et al. experiment,
1 Tetlock, Skitka, and Boettger (1989) suggested that accountability, the
social pressure to justify one's views or decisions to others, leads to
perspective-taking; accountability-induced perspective taking shifts an in-
dividual's information-processing strategy from a heuristic and peripheral
one to a more central and systematic one (Chaiken, 1980; Petty & Ca-.
cioppo, 1986).

PERSPECTIVE-TAKING 711
participants were shown a photograph of a person who fit the
stereotype of a "skinhead" and asked to write a short narrative
essay about a typical day in the life of the individual. Half of the
participants were told to actively avoid using any stereotypical
preconceptions in their narrative essay, whereas the other half were
given no instructions. In a lexical decision task used to measure the
accessibility of the stereotype, the stereotype was hyperaccessible
for suppression participants relative to the control condition.
We posited that perspective-taking would produce the positive
consequences of stereotype suppression (i.e., limiting the expres-
sion of stereotypical content) without the ironic side effect of
hyperaccessibility. In our first experiment, participants were ex-
posed to a photograph of an elderly man and asked to write an
essay describing a day in his life. One third of the participants were
given no explicit instructions, one third were asked to suppress any
stereotypic preconceptions that might bias their narrative essays,
and one third were told to take the perspective of the individual in
the photograph when writing their narrative essay. After complet-
ing a lexical decision task, participants wrote a second narrative
essay about a different elderly man. Finally, participants were
shown a photograph of a young African American man and asked
to write a third narrative essay. The third photograph was included
to see whether the experimental instructions would generalize to a
different social group.
We predicted that control participants would write narrative
essays that contained more stereotypic content than both
perspective-taking and suppression participants, who would not
differ from each other. That is, on a conscious, explicit task,
perspective-takers would inhibit the expression of stereotypic con-
tent. Further, only suppression goals should lead to stereotype
hyperaccessibility on the lexical decision task. Perspective-taking
should prevent the hyperaccessibility of stereotypes because the
personalized approach toward the target reduces the focus on
group-level characteristics that aid in the activation of stereotypes
(Brewer, 1988, 1996) while simultaneously increasing trait overlap
between representations of the self and of the group represented in
the photograph. This increased accessibility of the self should
direct the manner in which the target is categorized, reducing the
impact of one's stereotype. The lexical decision task was used to
demonstrate the effect of perspective-taking on tasks that do not
allow for the operation of conscious goals.
Macrae et al. (1994) found that rebound effects emerged on a
second narrative essay, in which no instructions on how to write
the essay were provided. We were interested in whether
perspective-takers as well as thought suppressors would continue
to write less stereotypically based essays. We predicted that the
effect of the perspective-taking instructions given before the first
narrative essay would carry through to the later, related task. The
combination of the explicit instructions and the similarities in the
task should alert perspective-takers to continue to take the per-
spective when writing the narrative essays.
Because perspective-takers psychologically approach targets
and suppressors distance themselves from targets (see Galinsky &
Moskowitz, 1999), we expected that perspective-taking would
affect not only the content expressed toward targets but also the
evaluation expressed toward targets. Davis et al. (1996) found that
perspective-taking led participants to like the target more.
Perspective-takers were predicted to limit the expression of ste-
reotypic content as suppressors did, while also expressing more
favorable content about the target compared with suppressors.
Method
Participants and design. Participants were 37 undergraduates who
were tested individually and received credit for participation as part of a
course requirement. The study had a 3 (experimental condition: control vs.
stereotype suppression vs. perspective-taking) × 2 (word type: stereotype
consistent vs. stereotype irrelevant) mixed design with repeated measures
on the second factor.
Procedure. Participants arrived in the laboratory and were told that
they were going to participate in a number of unrelated tasks that all
involved language processing. Participants were then given instructions
very similar to those given by Macrae, et al. (1994). The experimenter
explained that he was interested in their ability to construct life-event
details from visual information alone. All participants were then shown a
black and white photograph (presented on a computer screen) of an older
man sitting on a chair near a newspaper stand. Participants were then asked
to write a short narrative essay about a typical day in the life of the
individual. Before constructing their narrative essay, one third of the
participants were randomly assigned to the control condition and were
given no additional instructions. One third were randomly assigned to the
suppression condition and were instructed that "previous research has
demonstrated that thoughts and impressions are consistently influenced by
stereotypic preconceptions, and therefore you should actively try to avoid
thinking about the photographed target in such a manner." The final third
of the participants were instructed to adopt the perspective of the individual
in the photograph and "imagine a day in the life of this individual as if you
were that person, looking at the world through his eyes and walking
through the world in his shoes." Participants were asked to construct their
narrative essay on a sheet of paper with 27 lines and told to take approx-
imately 5 rain to complete the task. With regard to the perspective-taking
manipulation, Galinsky (1999a) presented evidence that this manipulation
increased the type of perspective-taking that involves imagining how the
self would feel and act. This perspective-taking manipulation evoked an
increase in distress emotions, which have been shown to be associated with
manipulations that instructed perspective-takers to imagine how the self
would be affected by the target's situation (Batsun, Early et al., 1997;
Stotland, 1969).
To separate the lexicai decision task from the narrative essay task,
participants were given a series of math tasks, including circling numbers
divisible by 7 and counting backward by 3s and 6s from a specified
number. The experimenter explained that the math task was included
because one of the professors in the department was interested in whether
quantitative thinking and linguistic thinking composed one mental system
or two separate mental systems. They were told that this relationship could
be explored experimentally by having half of the participants, like them-
selves, do a math task before doing a language task, and the other half of
the participants do only a language task; the researchers could then look at
whether having done the math task facilitated or inhibited performance on
the language task. In reality, all participants completed the math task,
which took approximately 15 min to complete. 2
When finished with the math task, participants were placed in front of a
computer screen and told that the language task was called a lexical
decision task. They were informed that several strings of letters were going
2 The time between the narrative essay and the lexical decision time was
longer in our experiment than in the experiment by Macrae et al. (1994).
We increased this time for two reasons. First, pilot testing determined that
this prevented any participants from becoming suspicious. Second, we
Wanted to find further support for Macrae et al.'s (1994) contention that
rebound effects are the result of repetitive priming consistent with the
synapse model of construct accessibility. Because the action potential of a
repeatedly activated construct dissipates and decays more slowly over time
than a less frequently activated construct, then stereotype suppressors
should display evidence of stereotype activation even after a delay between
the narrative essay task and the lexical decision task.

712 GALINSKY AND MOSKOWITZ
to flash briefly on the computer screen and that it was their job to
determine, as quickly as possible, whether those letters composed a word
in the English language. Participants were told to maximize both speed and
accuracy (Fazio, 1990). Participants were given a short practice session
consisting of nine trials to acquaint them with the task. After the practice
session, participants were given the major experimental block, consisting
of 10 words (5 of which were stereotype consistent and 5 of which were
stereotype irrelevant) in a fixed, random order. Participants responded
using a standard keyboard; one of the keys had been labeled word and the
other key had been labeled non-word. For each trial, the string of letters
appeared on the screen for 180 ms. After participants responded by hitting
one of the two keys, a plus sign accompanied by a beep appeared on the
screen to focus the participants' attention and to let them know the next
trial was about to appear. After the last trial, participants were informed the
task had concluded.
All of the words were evahiatively negative in implication, consistent
with the procedures of Macrae et al. (1994). They were selected on the
basis of previous pretesting in which 20 participants, other than the ones
who participated in the experiment, rated 47 traits for how typical they
were of elderly men. We .pretested the typicality of the words only in
reference to elderly men because the individual in the photograph was a
man and because previous work by Brewer, Dull, and Lui (1981) suggested
that the elderly stereotype could be separated into meaningful subcatego-
ties. The stereotype-consistent and stereotype-irrelevant words were
matched on both valence and length. The stereotype-consistent words
chosen represent the 5 traits rated as the most stereotypical of elderly men:
lonely, dependent, traditional, stubborn, and forge~eul. Only five words
were chosen as stereotype-consistent words because only five words that
were not related to the health or physical state of the elderly or not
synonyms of already selected words were rated above the midpoint of the
scale, a The stereotype-irrelevant words chosen were the five words rated
the least stereotypical (adjusting to ensure equal valence and word length):
jumpy, scheming, cowardly, envious, and deceptive. There were an equal
number of nonwords to prevent response biases.
After participants completed the lexical decision task, they were shown
a photograph of a second elderly man and were asked to write a second
narrative essay; following Macrae et al. (1994) they were told to write the
narrative essay with no mention of the previous experimental instructions.
Finally, participants were presented with a third photograph of an African
American man who appeared to be in his late teens to early 20s and asked
to write one final narrative essay. Participants were thoroughly debriefed
and thanked for their participation.
Results
Stereotypicality and valence of first narrative essay. One rater
blind to both experimental conditions and predictions and one rater
blind only to experimental condition estimated both the overall
stereotypicality of the contents of each passage and its overall
valence. The raters used a 9-point scale for both judgments, with
one scale anchored at 1 (not at all stereotypical) and 9 (very
stereotypical) and the other anchored at 1 (very negative) and 9
(very positive). Macrae et al. (1994) measured only the overall
stereotypicality of the essays. The rating of overall stereotypicality
has been the standard rating method used in most articles on
stereotype suppression that used this paradigm (Macrae et al.,
1994; Wyer, Sherman, and Stroessner, 2000). The valence rating
was included to explore whether stereotype suppression and
perspective-taking instructions produce not only lower stereotypic
contents of the essays, but also more positive evaluations of the
target (Batson, 1991; Davis et al., 1996). Valence is important in
understanding the consequences and dynamics of stereotype sup-
pression with regard to the elderly because previous research has
found that college-age students automatically associate negative
traits with the elderly (Perdue & Gurtman, 1990).
For the ratings of stereotypicality, interrater reliability was high,
r(37) --- .84, and therefore the ratings were averaged. A one-way
analysis of variance (ANOVA) conducted on the averaged ratings
of stereotypicality of the essays revealed a significant effect of
condition, F(2, 34) = 5.9, p < .01. A planned contrast comparing
the perspective-takers and the stereotype suppressors against par-
ticipants from the control condition was significant, F(1,
34) = 10.2, p < .003. Both stereotype suppressors (M = 5.4) and
perspective-takers (M = 4.4) wrote less stereotypical essays of the
elderly than did participants in the control condition (M = 6.8).
For valence ratings, interrater reliability was high, r(37) = .79,
and therefore the ratings were averaged. A one-way ANOVA
conducted on the rated valence of the essays revealed a marginal
effect of condition, F(2, 34) = 3.1, p < .057. A planned contrast
comparing the perspective-taking condition against the other two
conditions was significant, F(1, 34) = 5.3, p = .03, Perspective-
takers (M = 6.8) expressed more positive evaluations of the target
individual than did stereotype suppressors (M = 5.8) and partici-
pants in the control condition (M = 5.2). These results suggest that
stereotype suppression can reduce the expression of stereotypical
content, but this does necessarily translate into the expression of
more positive content. Perspective-taking, on the other hand, both
reduced the expression of stereotypical content and increased the
expression of positive content relative to the control condition. The
more favorable evaluations produced by perspective-takers is sim-
ilar to the increased liking of targets found by Davis et al. (1996).
Lexical decision latencies. The principal dependent measure
was the mean time to respond to stereotype-consistent words
relative t.o stereotype-irrelevant words (see Figure I). Incorrect
classifications (i.e., calling a letter string that was a word a non-
word) were excluded from the statistical analyses; there was
an 8.9% error rate across the trials. Reaction times for each trait
were examined for outliers and those reaction times that were more
than three standard deviations away from that trait's mean were
eliminated; outliers were quite rare and accounted for less
than 0.1% of responses. To remove skewness, the raw responses
were transformed using a square-root transformation; this trans-
formation most closely approximated a normal distribution (Fazio,
1990). Participants' transformed lexical decision latencies were
submitted to a 3 (experimental condition: control vs. stereotype
suppression vs. perspective-taking) × 2 (word type: stereotype
consistent vs. stereotype irrelevant) mixed-model ANOVA with
repeated measures on the second factor. As expected, only an
Experimental Condition × Word Type interaction emerged from
the analysis, F(2, 34) = 4.7, p = .02. One-way ANOVAs were
conducted on each of the word types across experimental condition
to explicate the interaction. The one-way ANOVA conducted on
the transformed response latencies to stereotype consistent words
was marginal, F(2, 34) = 3.1, p = .058; a planned contrast
comparing the reaction times of stereotype suppressors to those of
perspective-takers and control participants was significant, F(1,
34) = 5.8, p = .02. Stereotype suppressors (M = 451 ms)were
faster to respond to stereotype-consistent words compared with the
3 Pilot testing found that the health-related and physically related words
increased suspicion over connections between the narrative essay task and
the lexical decision task. Therefore no health-related words were included.

PERSPECTIVE-TAKING 713
,-,54O
s20
0
"~
480
I--
=440
o
,i
~ 420
m
@
~'400
Figure 1.
condition.
[] Stereotype-consistent words
• Stereotype-irrelevant words
Control Suppre~ion Perspec"d ve-
Taking
Condition
Response time measure to stereotype-consistent and stereotype-irrelevant words by experimental
perspective-takers (M = 529 ms) and the control condition (M =
510 ms). The one-way ANOVA conducted on the transformed
response latencies to stereotype-irrelevant words was not signifi-
cant, F(2, 34) < 1. Stereotype suppressors (M = 523 ms) were not
faster to respond to stereotype-irrelevant words relative to
perspective-takers (M = 516 ms) and the control condition (M =
515 ms). In addition, within-conditiqn analyses found that only
stereotype suppressors demonstrated facilitation for stereotype-
consistent words relative to stereotype-irrelevant words,
fill) = 2.6, p < .03. Perspective-takers displayed the reverse
pattern, responding marginally slower to stereotype-consistent
words relative to stereotype-irrelevant words, t(11 ) = 1.6, p <. 13.
Control participants did not differ in their latencies to stereotype-
consistent and stereotype-irrelevant words, t(12) < 1. These re-
suits lend support to the hypothesis that perspective-taking would
not lead to the hyperaccessibility of the stereotype despite reduced
expression of stereotypic content in the narrative essay. The fact
that suppression led to evidence of hyperaccessibility even when
stereotype accessibility was measured after a delay supports Mac-
rae et al.'s (1994) assertion that increased accessibility of sup-
pressed stereotypes are due to repetitive priming during the nar-
rative essay task. The search for failures of suppression increases
attention to the suppressed stereotype and thus it gets more fre-
quently activated. This increase in the frequency of activation
leads to greater action potential of the construct, which dissipates
and decays more slowly over time (Higgins, 1989).
Stereotypicality and valence of second narrative essay. The
ratings of stereotypicality ratings, r(37) = .82, and the ratings
valence, r(37) = .74, were each averaged across coders. Although
a one-way ANOVA was marginal for the stereotypicality ratings,
F(2, 34) = 2.4, p < .10, a contrast comparing the perspective-
takers and the suppressors against the control condition was sig-
nificant, F(1, 34) = 4.4, p < .05. Both perspective-takers
(M = 4.3) and suppressors (M = 4.2) wrote less-stereotypically
based essays than did control participants (M = 6.1). Although the
one-way ANOVA for the valence ratings was marginal, F(2,
34) = 3.2, p = .055, a contrast comparing the perspective-takers
against the suppressors and participants in the control condition
was significant, F(1, 34) = 6.2, p < .02. Perspective-takers ex-
pressed more positive evaluations of the target (M = 5.8) than did
suppressors (M = 4.9) and control participants (M = 4.7). A1-
though Macrae et al. (1994) found suppression of the skin_head
stereotype was followed by increased expressions of stereotypic
content, the current experiment revealed suppression of the elderly
stereotype led to the continued inhibition of stereotypic expression
in a subsequent task. This inhibition of stereotypic expression (an
explicit measure of stereotype control), along with evidence of
stereotype hyperaccessibility (an implicit measure of stereotype
accessibility), supports theories that discuss a distinction between
implicit and explicit evidence for stereotyping and prejudice (De-
vine, 1989).
Stereotypicality and valence of third narrative essay. The third
photograph that participants wrote narrative essays about was of an
African American man in his late teens to early 20s. Interrater
reliabilities for stereotypicality and valence were acceptable,
r(36) = .72, and r(36) = .75, respectively. 4 Thus, the ratings for
each measure were averaged across coders. Unlike the previous
two narrative essays, no differences were found between experi-
mental conditions on the averaged ratings of essay stereotypicality,
F(2, 33) = 1.5, p > .20. All participants wrote essays that did not
rely on the stereotype for African Americans (GM = 2.6). Partic-
ipants commented during postexperimental debriefing that they
were careful to not stereotype this target. The conscious regulation
that control participants described suggests that the socially sen-
sitive nature of the stereotypes triggered spontaneous efforts at
suppression. For valence ratings, the one-way ANOVA was mar-
ginal, F(2, 33) = 3.1, p = .058. A planned contrast comparing the
perspective-taking condition against the suppression and control
conditions was significant, F(1, 33) = 5.8, p < .02. Perspective-
takers (M = 7.0) continued to express more positive evaluations of
the target than did either suppressors (M = 5.7) or control partic-
ipants (M = 6.1). Overall, participants expressed more positive
evaluations toward the African American target compared with the
elderly targets.
4 One participant failed to write the final narrative essay, and thus the
degrees of freedom for these analyses differ from the analyses of the ftrst
two narrative essays.

714 GALINSKY AND MOSKOWlTZ
Discussion
Although stereotype suppressors were able to restrict the ex-
pression of stereotypic content in their narrative essays, they
responded significantly faster to stereotype-consistent words in a
lexical decision task. Perspective-takers, however, restricted their
expression of stereotypic content without displaying facilitation
for stereotype-consistent words on the lexical decision task.
Perspective-takers also expressed more positive attitudes toward
the target in their essays than did either stereotype suppressors or
control participants across all three essays.
Our results indicated that suppressors did not show evidence of
rebound effects on an explicit, obtrusive measure, unlike the
results of Macrae et al. (1994). Recent research both by Monteith,
Spicer, and Tooman (1998) and by Wyer, Sherman, and Stroessner
(2000) failed to find rebound effects on explicit measures of
stereotype expression and application. In the Wyer et al. experi-
ment when race was kept constant across two tasks (a narrative
essay about an African American as Task 1 and impressions of an
African American as Task 2), no rebound effects were found.
When race was made ambiguous on the second task, suppressors
(vs. control participants) showed greater accessibility effects--
rebound effects--for the stereotype. In the Monteith et al. study,
no rebound effects occurred for either high- or low-prejudiced
participants when the second task was an obtrusive measure of
stereotype application and it resembled the first task (both were
narrative essays); when the second task was an unobtrusive mea-
sure of stereotype accessibility, high-prejudice suppressors did
show evidence of hyperaccessibility of the stereotype. Thus, it is
not surprising that stereotype suppressors in our Experiment 1
continued to exert control over their responses on the explicit
measure, even after having demonstrated hyperaccessibility in the
previous, unobtrusive lexical decision task.
Davis et al. (1996) presented evidence that the effects of
perspective-taking can be divided into consciously controlled ef-
fects and more implicit effects not under direct conscious control.
In the Davis et al. study, increased liking due to perspective-taking
did not mediate increases in self-target overlap. In addition, the
increases in self-target overlap were generally impervious to de-
pletions of cognitive resources. We found that perspective-taking
increased evaluations of the very targets of perspective-taking. In
addition, perspective-takers showed evidence of implicit stereo-
type control rather than stereotype hyperaccessibility. Like the
findings of Davis et al., however, we found that these conscious
effects of increased evaluations did not mediate the nonconscious
effect of decreased stereotype accessibility. Covarying out the
effect of the evaluations expressed in the first narrative essay
increased, rather than decreased, the effect of the experimental
conditions on the transformed response latencies, F(2, 33) = 4.5,
p = .02. We have suggested that the nonconscious effect of
decreased stereotype accessibility is mediated by activation of the
self-concept. However, we did not measure self-concept accessi-
bility in Experiment 1. The next experiment more directly tests this
assumption.
Experiment 2
In the second experiment, we sought to more clearly explicate
the process by which the positive benefits of perspective-taking
occur. Davis et al. (1996) found that perspective-taking led par-
ticipants to ascribe self-descriptive traits to a target person. In other
words, they saw more of themselves in that person. This suggests
that perspective-taking activates a stored representation of the self,
which then gets used in the process of categorization and exerts an
influence on the interpretation of later ambiguous events and
behaviors (Higgins, 1996). Thus, perspective-taking increases the
accessibility of the perspective-taker's self-knowledge. In addi-
tion, research on categorization suggests that knowledge structures
compete with each other to win the metaphorical race toward
categorization; only one category tends to be dominant at a time,
even when another category is equally applicable (Macrae et al.,
1995). If the self-concept gets activated and brought on line then
it, rather than the stereotype, could be used to categorize the target.
In the Davis et al. (1996) study, self-descriptive traits were
ascribed to a target person who did not belong to a stereotyped
group (male participants took the perspective of a male target and
female participants took the perspective of a female target). Would
the ascription of self-relevant traits occur not only for target
individuals but also for the group to which the target belonged?
And would it occur if the target were a member of an out-group?
To explore this question the next experiment followed the
methodology of Smith and Henry (1996) and Aron, Aron, Tudor,
and Nelson (1991). The Smith and Henry study had participants
describe themselves, their in-group, and an out-group according
to 90 heterogeneous traits. They found that participants responded
more quickly and with fewdr errors to words that both the self and
the in-group possessed. In our next experiment we had participants
rate themselves on the 90 traits that Smith and Henry used. After
completing that task, participants completed the same narrative
essay task that we used in Experiment 1; participants in one of
three experimental conditions (control, stereotype suppression, and
perspective-taking) wrote a narrative essay about an elderly man.
After a few filler tasks, participants were shown the same 90 traits
with which they had rated the self and were asked to provide their
personal opinions about the general characteristics of the elderly
using those traits. We predicted that perspective-takers would
show greater overlap between the representations of the self and
representations of the elderly compared with the other two condi-
tions. Because we wanted to show that the ascription of self-
relevant traits to the group was an implicit effect of perspective-
taking (Davis et al., 1996), great care was taken to separate the
narrative essay task from the task in which they assigned traits to
the group. Experiment 1 showed that when a subsequent task is
very similar to the task in which perspective-taking is manipulated,
then participants continue to pursue that goal. Thus, we attempted
to prevent participants from continuing to consciously pursue the
perspective-taking goal so that we could look at the nonconscious
consequences of that goal--activation and application of the
self-concept.
In addition, a few of the 90 traits are relevant to the stereotype
of the elderly. These traits were a measure of the stereotypicality
of the participants' representation of the elderly. We predicted that
perspective-takers would attribute fewer stereotypic traits to the
elderly.
Finally, Davis et al. (1996) found increased self-target overlap
on positive traits only--there was no effect on negative traits. Like
Smith and Henry (1996) and Smith, Coats, and Walling (1999), we
separated the words into positive and negative subgroups to test
whether valence interacted with the experimental conditions. Ap-

PERSPECTIVE-TAKING 715
proximately half of the traits used were positive and the other half,
negative.
Method
Participants and design.
Participants were 85 undergraduates who
were paid for their participation. They were tested in groups of 3 to 6. The
design of the study was a single-factor, between-participants design with
three levels of the manipulated variable (condition: control vs. stereotype
suppression vs. perspective-taking). Three participants were removed for
failing to write the narrative essay, leaving 82 participants in the final
analysis.
Procedure.
Participants arrived in the laboratory and were told that
they were going to participate in a number of unrelated tasks, some of
which involved language processing and others that involved perceptions
of the self and others. Participants were first given the list of 90 traits from
Smith and Henry (1996) and asked to rate how well each trait described
them using a scale anchored at l
(extremely unlike) and 7 (extremely like).
Next, participants completed the narrative essay task jn which they were
presented with the same photograph and instructions from Experiment I.
This task was emphasized as dealing with language processing and lin-
guistic expression. Participants were given a series of attitudes and per-
sonaiity questionnaires that took approximately 15 min. The filler tasks
were included to separate the narrative essay task from the trait assignment
to the elderly. After these filler tasks, participants were given the same list
of 90 traits and asked to
Please rate how well each trait describes the elderly using the follow-
ing scale. Although not all group members are exactly alike, group
members tend to be similar on many traits and you should provide
your personal opinions about the general characteristics of the elderly.
These were the same instructions that Smith and Henry (1996) used.
Great care was made to separate the trait assignment task from the
narrative essay task. We wanted to ensure that any increased overlap
between participants' own self-concept and representations of the elderly
was not the result of participants thinking that they were to continue
actively taking the perspective of the elderly. First, participants were told
that the two tasks were completely separate experiments. The narrative
essay task was described as part of the language processing study and the
trait ratings were described as part of a study on representations of groups.
Second, the social group to which the person in the photograph used in the
narrative essay belonged was never mentioned; by not providing a label,
participants would be less likely to connect the two tasks. Third, none of
the participants expressed an effect of the narrative essay on their judg-
ments of the elderly. Fourth, pilot testing demonstrated that participants do
not associate the applying of serf-relevant traits as an effect of perspective-
taking. Finally, because approximately half of the words were positive and
half of the words were negative, one might expect that the increased
overlap should occur only for positive traits if participants were strategi-
cally using perspective-taking. In the Davis et al. (1996) experiment,
participants assigned traits to the target of perspective-taking immediately
after taking the perspective. They found no effect of perspective-taking on
the ascription of negative words. This suggests some motivation to assign
positive traits when perspective-taking is being strategically used. For all
these reasons we are confident that any increase in self-group overlap for
perspective-takers is not an artifact or due m demand characteristics. After
rating the elderly, participants were debriefed.
Results and Discussion
stereotypicality and valence, interrater reliability was acceptable,
rs(82) > .67, and therefore the ratings were averaged across
coders. A one-way ANOVA conducted on the averaged ratings of
stereotypicality of the essays revealed a significant effect of con-
dition, F(2, 79) = 4.6, p = .01. A planned contrast comparing the
perspective-takers and the stereotype suppressors against partici-
pants from the control condition was significant, F(1, 79) = 9.0,
p < .01. Both stereotype suppressors (M = 4.8) and perspective-
takers (M = 5.0) wrote less stereotypical essays of the elderly than
did participants in the control condition (M = 6.0).
For valence ratings, the one-way ANOVA was significant, F(2,
79) = 5.9, p < .01. A contrast comparing the perspective-taking
condition against the suppression and control conditions was sig-
nificant, F(1, 79) = 7.3, p < .01. Perspective-takers (M = 6.0)
expressed more positive evaluations of the target compared with
participants in the control participants (M = 4.9) and the suppres-
sors (M = 5.6). These results replicate the pattern found in Ex-
periment 1 in which perspective-takers and suppressors both re-
stricted the expression of stereotypic content, but only perspective-
takers expressed more positive evaluations of the target.
Stereotypical trait attributions to the elderly.
Five traits were
selected that were stereotypical of the elderly:
weak, dependent,
self-reliant
(reverse coded),
worrier,
and
dull.
These traits were
chosen because they were in the upper quarter of the traits used in
pretesting for Experiment 1. A one-way ANOVA conducted on the
attributions to the elderly using the stereotypical traits was signif-
icant, F(2, 79) = 3.5, p < .04. A planned contrast comparing the
stereotypical ratings of the perspective-takers (M = 3.99) against
the control participants (M = 4.53) and suppressors (M = 4.35)
was significant, F(I, 79) = 6.3, p < .02. Perspective-takers rated
the elderly less stereotypically than did participants in the other
two conditions.
Degree of overlap between representations of the self and the
elderly. To
assess degree of overlap between representations of
the self and representations of the elderly, the absolute value was
taken of the difference between ratings for the elderly and ratings
for the self for each of the 85 nonstereotypic traits (a smaller
number means greater self-group overlap). 5 These 85 traits were
divided into negative and positive subgroups on the basis of the
valence ratings of 10 independent participants who judged whether
the trait was positively valenced or negatively valenced using a
7-point scale anchored at 1
(negative)
and 7
(positive).
Using a
median split, 43 of the words were coded as negative and 42 of the
words were coded as positive.
The absolute value of the difference between self-ratings and
ratings for the elderly were submitted to a 3 (experimental condi-
tion: control vs. stereotype suppression vs. perspective-taking) × 2
(word type: positive vs. negative) mixed-model ANOVA with
repeated measures on the second factor.
There was a significant main effect for experimental condition,
F(2, 79) = 3.6, p = .03. The interaction testing whether there were
any differences between the positive and negative traits across the
experimental conditions was not significant, F < 1. A planned
contrast comparing the degree of overlap between representations
Stereotypicality and valence of narrative essay.
Two raters
blind to both predictions and experimental conditions estimated
the overall stereotypicality of the contents of each passage and its
valence using the same scales as Experiment 1. For ratings of
5 The pattern of the data remain the same (and the effect is stronger)
when the stereotypical traits are included in the average of all the traits, but
to provide a more conservative test of our hypothesis we present the
average of the 85 traits that were not stereotypic of the elderly.

716 GALINSKY AND MOSKOWITZ
of the self and of the elderly of the perspective-takers against those
of the control participants and suppressors was significant, F(1,
79) = 6.8, p = .01. Perspective-takers (M = 1.66) displayed more
overlap between their representations of the self and their repre-
sentations of the elderly than both participants in the control
condition (M = 1.91) and suppressors (M = 1.86). Our results
indicate that not only do perspective-takers ascribe self-descriptive
traits to a target (see also Davis et al., 1996), but this ascription
extends to the social group the person represents.
We next tested whether it was the increased self-group overlap
that diminished the degree of stereotypical responding by
perspective-takers. An analysis of covariance (ANCOVA) was
conducted across the experimental conditions on the ratings of the
stereotypical traits with the degree of self-group overlap covaried
out. The subsequent ANCOVA revealed a significant effect of the
covariate, F(1, 78) = 4.6, p < .04. The introduction of the
covariate also eliminated the significant effect of experimental
condition on the ratings for the stereotypical traits, F(2, 78) = 1.9,
p = . 15. Perspective-takers were attributing self-descriptive traits
to the elderly, and this increased self-group overlap in represen-
tations resulted in less stereotypical ratings.
The ascription of self-descriptive traits to the elderly occurred
for both positive and negative words. Davis et al. (1996) did not
find reliable ascription of negative traits to the target of
perspective-taking. Finding the effect on both positive and nega-
tive words suggests that participants were not strategically using
the perspective-taking manipulation when ascribing traits to the
elderly. If this were true, it would have been predicted that the
ascription would have occurred only for positive traits as it did in
the Davis et al. experiments. These results suggest that participants
were not conscious of ascribing self-relevant traits in constructing
a representation of the elderly as a social group. The self-concept,
both its positive and negative elements, got activated and applied
(Higgins, 1996). In addition, we performed an ANCOVA similar
to the one we performed in Experiment 1. Although controlling for
the effect of the evaluations expressed in the narrative essay in the
degree of self-group overlap did reduce the effect to nonsignifi-
cance, F(2, 78) = 2.6, p = .078, the effect of experimental
conditions was still marginal. This finding replicates the pattern of
results from Davis et al. (1996) in which the majority of self-target
overlap effects remained at least marginal when controlling for
liking toward the target. Like in Experiment 1, the nonconscious
effects of perspective-taking, increased overlap between self and
group caused by activation of the self-concept, appear to be dis-,
tinct from the more conscious effects of expressing positive eval-
uations in the narrative essays.
One problem with the conclusion that perspective-taking is a
constructive alternative to suppression is that Experiments 1 and 2,
like most previous experiments on stereotype suppression, used a
stereotype that is not particularly socially sensitive ("the elderly")
and one that most participants do not feel the need to inhibit under
ordinary conditions (Macrae et al., 1996). Evidence from the third
narrative essay of Experiment 1 supports the contention that as the
socially sensitive nature of the stereotype increases so do sponta-
neous efforts at suppression. Wyer, Sherman, and Stroessner
(1998) led participants to spontaneously suppress stereotypes by
stating that an attitude survey concerning African Americans was
for a political group entitled "African Americans for Intellectual
Understanding" compared with a control condition who were not
told for which group the survey was intended and who were simply
told to be honest and accurate in their surveys. Macrae, Boden-
hausen, and Milne (1998) found that conditions that result in
heightened self-focus induce spontaneous efforts at suppression.
The Wyer et al. experiment and our Experiment 1 suggest that the
likelihood of spontaneous suppression increases not only when
self-focus is high, but also when social and political discussion
surrounding the group is particularly incendiary.
Galinsky and Moskowitz (1999) demonstrated both that re-
bound effects, as measured by an unobtrusive, lexical decision
task, occur even when the stereotype being suppressed is socially
sensitive and that the benefits of perspective-taking survive expo-
sure to a socially sensitive stereotype. They used the African
Americans stereotype and the methodology closely followed that
of Experiment 1 reported here. Participants wrote narrative essays
from a photograph of an African American and then completed a
lexical decision task. Because the third narrative essay from Ex-
periment 1 found that the socially sensitive nature of the stereotype
leads to spontaneous suppression, Galinsky and Moskowitz had
control participants rely on the stereotype in constructing their
narrative essays--participants were in a stereotype expression
condition. Forcing participants to express the stereotype is analo-
gous to the original manipulations on thought suppression (e.g., the
"think about white bears" vs. "suppress thoughts about white
bears" instructions used by Wegner, Schneider, Carter, & White,
1987) and provided a stronger test of the hypothesis that suppres-
sion makes the stereotype hyperaccessible. Results showed that
suppressors responded more quickly to stereotype-consistent
words even when compared with stereotype expressers, demon-
strafing hyperaccessibility as it is classically defined. In addition,
their experiment used a no-narrative essay control. The response
latencies of perspective-takers were almost identical to the laten-
cies of the no-essay control, suggesting that perspective-taking had
inhibited the activation of the stereotype.
Experiment 3
In the final experiment, we explored the role of perspective-
taking in affecting attitudes and evaluations of out-groups. The
experiment was also conducted to find further evidence that self-
concept activation and application are integral components of the
effectiveness of perspective-taking in affecting intergroup evalua-
tions. The experiments already presented have compared suppres-
sion and perspective-taking as two potential strategies used to
exert control over stereotyping. Suppression, however, may not be
a viable strategy when no content is known about a group, or when
there is not an integrated construct used to describe the group;
without known content there is nothing to suppress. Such a situ-
ation exists when groups are differentiated and created on the basis
of responses to a novel situation. The use of perspective-taking,
unlike suppression, does not depend on an integrated construct.
Perspective-taking is still a viable strategy for reducing intergroup
bias in this situation because the activation and application of the
self should occur regardless of whether an integrated construct
exists. Seeing the self in the other should lead to more positive
out-group evaluations, an essential component of decreasing in-
group favoritism. Using the next experiment, we examined
whether perspective-taking can decrease the in-group favoritism
that often results from heightened group distinctiveness.
For Experiment 3, we relied on the minimal group paradigm,
where the mere categorization of people into groups, even when

PERSPECTIVE-TAKING 717
the "group" is defined by a trivial distinction, has been shown to
be sufficient to create in-group biases. Such biases range from
favoring the in-group when allocating points or money (Tajfel,
Biliig, Bundy, & Flament, 1971), to differential evaluations of
in-group and out-group members on evaluative traits (Locksley,
Ortiz, & Hepburn, 1980), to both attributional and memorial biases
(Howard & Rothbart, 1980) such as the ultimate attribution error
(Pettigrew, 1979; Taylor & Jaggi, 1974).
Past research has examined forces that attenuate such biases.
Locksley etal. (1980) showed that by providing individuals with
information about allocation decisions and trait ratings by other
members of the in-groups and out-groups, in-group favoritism was
eliminated only when the out-group was more rewarding (e.g.,
offered more points) than the in-group. Thompson (1993) found
that an integrative negotiation that produces a mutually beneficial
agreement eliminated in-group favoritism as measured by trait
ratings; specifically the positivity of the out-group evaluations was
increased. Wilder (1986) noted that processes that diminish the
salience of the in-group/out-group distinction, or that lead to the
individuation of out-group members, reduce biases. Some of these
processes are disclosure of personal information about out-group
members (Wilder, 1978), removal of cues that highlight dissimi-
larity between the groups, and recognition of overlapping social
identities. Wilder (1986) explains that the effectiveness of these
manipulations results from the perceiver taking the perspective of
the out-group while focusing on contextual features other than
category membership. In the next experiment, we explored this
proposed mediational role of perspective taking in bias reduction.
Experiment 2 established that taking the perspective of an
individual member of a social group increases the overlap between
representations of the self and representations of the target's group.
Perspective-taking may be a useful strategy for reducing in-group
favoritism because inclusion of representations of the in-group
within representations of the self have been shown to mediate the
bias. That is, one explanation for the psychological favoring of a
minimal in-group is that participants extend their positive self-
representations to encompass their group (Cadinu & Rothbart,
1996; Smith & Henry, 1996). Recent research using the minimal
group paradigm suggests that the in-group favoritism effects are
produced by an automatic evaluation effect in which the in-group
label acquires its positive meaning because of its association with
the self (Otten & Moskowitz, 2000; Otten & Wentura, 1998).
Cadinu and Rothbart (1996) presented evidence consistent with the
hypothesis that in-group favoritism is a self-anchoring effect.
Participants in their study demonstrated a stronger correlation
between the self and in-group ratings when the self-ratings pre-
ceded the in-group ratings and they were more likely to generalize
from the self to the in-group than vice versa. The process of taking
the perspective of what it is like to be a member of the opposite
group should lead to a creation of a cognitive representation of the
other group that now overlaps with the participants' own self-
representation (Davis et al., 1996). As the level of overlap between
the self and out-group increases so should the positivity of out-
group evaluations (Thompson, 1993).
It remains possible that perspective taking may affect intergroup
evaluations in the minimal group paradigm not just through in-
creasing self/out-group overlap in representations, but also by
calling into question the group label. In minimal group settings,
participants are likely to selectively recruit memories consistent
with the feedback label (Ross, Lepper, & Hubbard, 1975). By
taking the perspective of an individual with the opposite estimation
tendency, perspective-takers might recognize the presence of their
own past behaviors that are consistent with the opposite estimation
tendency. To test whether perspective-taking simply works
through calling into question the label through recruitment of
memories that are inconsistent with the label rather than increasing
the overlap between representations of the self and representations
of the out-group, another condition, in addition to the perspective-
taking manipulation, was included. These participants were asked
to recall a recent experience where they estimated something in the
direction opposite to their estimation tendency.
One final way in which perspective taking might reduce in-
group favoritism could be through promoting thoughts about di-
mensions on which the in-group and out-group are similar to each
other. Some researchers have found that when separate groups are
recategorized as one group, recategorized participants decrease
bias by increasing their ratings of former out-group members
because former out-group members are now considered to be part
of the larger, more inclusive in-group; that is, they are considered
to be in-group members (Gaertner, Mann, Murrell, & Dovidio,
1989). In the Gaertner et al. experiment, participants were merged
into one group and worked on a problem-solving task. In the
current experiment, any merging of the groups through contem-
plating areas of similarity would be purely psychological. Any
perceptions of similarity also could be independent of the self, a
Ink we hypothesize to be crucial in increasing evaluations of the
out-group. In addition, Davis etal. (1996) found that increased
liking after perspective-taking was independent of self-other over-
lap; although thinking of similarities may increase liking, it would
not be necessarily expected to increase self-other overlap. There-
fore, in addition to the perspective-taking manipulation and the
recalling-of-past-behavior manipulation, in Experiment 3 we also
examined whether thinking about similarities between the groups
would eliminate in-group favoritism.
Participants not only rated how well each trait described both
the in-group and the out-group, but they also assigned a valence or
"favorability rating" to each trait. The favorability rating was used
because Esses and Zanna (1995) found that evaluative meaning of
traits can change when describing out-group members. For exam-
ple, the trait intelligence when describing Jews (when they are an
out-group) may be interpreted negatively as conniving. With re-
gard to group-based evaluations (Brewer, 1979), loyal may be
considered positively when describing the in-group, but take on
negative connotations, such as clannish or exclusionary, when
describing the out-group. No previous experiment has investigated
the use of favorability ratings in the context of the minimal-group
paradigm.
We predicted that only perspective-taking, which would involve
an increase in self/out-group overlap, would increase evaluations
of the out-group. Perspective-takers would be more likely to assign
favorable traits to the out-group and also maintain the positive
connotations of the group-relevant words in the context of the
out-group.
Method
Participants and design.
Participants were 40 undergraduates who
received credit for participation as part of a course requirement. The design
of the study was a 4 (experimental condition: control vs. perspective-taking
vs. behavioral recall vs. perceived similarity) X 2 (ratings: in-group vs.
out-group) mixed design with repeated measures on the second factor.

718 GALINSKY AND MOSKOWITZ
Procedure.
After arriving at the laboratory, participants were told that
they would be participating in two separate tasks. The first task was a dot
estimation task in which participants were asked to estimate the number of
dots presented on a computer. After this task, the computer provided
feedback informing participants that they consistently tended to overesti-
mate the number of dots presented. They were further told that each style
of estimation does not relate to the accuracy of judgments, but simply
represents different patterns of responding.
After the feedback, the primary experimental manipulation took place.
One fourth of the participants were told to write a short narrative essay
about a day in the life of an underestimator, to "go through the day as if you
were an underestimator, walking through the world in their shoes and
looking at the world through their eyes." One fourth of the participants
were asked to "recall a time in the past two weeks when you underesti-
mated something. Try to recall when and where this underestimation
occurred, concentrating on how you felt and reacted." If perspective-taking
simply involved providing access to a range of experiences that would call
into question the strength and direction of the estimation tendency label,
there would be no differences between the perspective-taking and behav-
ioral recall task. One fourth of the participants were asked to write a short
narrative essay about the ways in which overestimators and underestima-
tors are similar to each other. The final one fourth of the participants did
not write a narrative essay and went straight to the dependent measures.
After writing the narrative essays (or after receiving the feedback for
control participants), participants were told
We are interested in the intuitions lay individuals have about the
characteristics of the different estimation tendencies. The measures
will explore your intuitions and expectations about the psychological
construct of perceptual style and estimation tendency. Although psy-
chologists know a lot about what perceptual style relates to,
little research has been done on what individuals think this construct
relates to.
The judgments involved rating the two different estimation tendencies
on 10 different positive dimensions (considerate, cooperative, friendly,
generous, honest, kind, loyal, sincere, trustworthy, understanding) consid-
ered to be desirable in a valued group member (Brewer, 1979) using a
7-point scale anchored at 1
(never true) and 7 (always true).
Participants
were asked to indicate their expectations about the personality character-
istics of each group. They first rated the in-group (i.e., overestimators)
along all 10 dimensions and then rated the out-group (i.e., underestimators)
along all 10 dimensions. Participants not only rated how well each trait
describes both groups, but they also assigned a valence or "favorability
rating" to each trait. The favorability rating was used because Esses and
Zanna (1995) found that evaluative meaning of traits can change when
describing out-group members. Participants were asked to assign a valence
to each characteristic using the following scale: , , 0, +, ++.
Participants first rated the valence of the traits in the context of the in-group
(i.e., overestimators) along all 10 dimensions and then rated the valence of
the traits in the context of the out-group (i.e., underestimators) along all 10
dimensions. The measure used by Esses and Zanna combined (multiplica-
tively) both the valence rating with ratings of the percentage of the group
to which each trait applied. Participants were therefore asked to rate the
percentage of both in-group members and out-group members for which
each trait applied; participants first made percentage ratings for the in-
group for all 10 traits and then they made ratings for the out-group.
Next, participants answered a number of questions that examined how
much they perceived their estimation tendency influenced their lives. They
were first asked to rate what percentage of their daily activities was
affected by their estimation tendencies. After that question, the activities
were broken down into different types, which included academic tasks and
personal tasks (e.g., hobbies, friendships, athletics, and value of posses-
sions); specifically, participants were asked "what effect does your esti-
mation tendency have on the following activities" and they answered for
each activity using a 7-point scale anchored at 1
(no effect) and 7 (complete
effect).
Participants then rated the percentage of their friends that shared
their estimation tendency. Finally participants rated how satisfied they
were with their estimation tendency using another 7-point scale anchored
at 1
(very unsatisfied) and 7 (very satisfied).
Results and Discussion
Assignment of traits to the in-group and the out-group.
Rat-
ings of the traits for the in-group were combined into one sum-
mated index, as were the ratings for the out-group (yielding a range
of possible scores from 10 to 70). These indices were submitted to
a 4 (experimental condition: control vs. perspective-taking vs.
behavioral recall vs. similarity) × 2 (group: in-group vs. out-
group) mixed-model ANOVA with repeated measures on the
second factor (see Figure 2). A significant main effect for group,
F(1, 36) = 26.1, p < .001, was qualified by the predicted Condi-
tion x Group interaction, F(3, 36) = 4.3, p = .01. Level of
in-group favoritism was tested within each condition. Participants
in the control condition rated the in-group (M = 52.8) more
favorably than the out-group (M = 42.8), t(9) = 5.3, p < .001.
Participants in the similarity condition rated the in-group
(M = 45.6) more favorably than the out-group (M = 41.2),
t(9) = 2.6, p < .03. Participants in the behavioral recall condition
rated the in-group (M = 48.8) marginally more favorably than the
55
50
"0
_=
45
I-
"o
E 4o
E
~ as
30
I D In-group 1
• Out-group J
Control Similarity Behavioral Perspective-
Recall Taldng
CondiUon
Figure 2.
Trait ratings of the in-group and the out-group.

PERSPECTIVE-TAKING 719
out-group (M = 42.2), t(9) = 2.1, p = .06. However, participants
in the perspective-taking condition did not rate the in-group
(M = 47.8) any higher than the out-group (M = 47.0), t(9) < 1.
One-way ANOVAs were conducted separately for the ratings of
the in-group and the out-group across experimental condition in
order to address the question of the process by which bias was
reduced in the perspective-taking condition. The one-way
ANOVA for the in-group was significant, F(3, 36) = 2.8, p = .05.
Participants in each of the experimental conditions rated the in-
group less favorably compared with those in the control condition;
a contrast comparing the three experimental conditions against the
control condition was significant, F(1, 36) = 7.3, p = .01. Al-
though the one-way ANOVA was not significant for the out-group
evaluations, F(3, 36) = 1.3,p > .30, a planned contrast comparing
the perspective-taking condition against the other three conditions
was marginal, F(1, 36) = 3.6, p = .07. The combination of this
contrast along with the lack of difference between in-group and
out-group evaluations for perspective-takers, suggests that taking
the perspective of what it is like to be an out-group member
increased ratings of the out-group to a level comparable to that of
the in-group.
Assignment of trait valence to the in-group and the out-group.
Valence values were transformed into numbers ranging from
2(--) to 2 (++). Like Esses and Zanna (1995), the valence of
each trait was combined with the percentage of group members to
which each characteristic was attributed using the formula,
10
S, = ~ (Vi, × P's),
i=t
where S = sum, V = valence, P = percentage, and g = group.
(The same pattern emerges when only valence is investigated
independent of percentages.) The valence scores for the in-group
and out-group were submitted to a 4 (experimental condition:
control vs. perspective-taking vs. behavioral recall vs. similar-
ity) × 2 (group: in-group vs. out-group) mixed-model A.NOVA
with repeated measures on the second factor. A significant main
effect for group, F(1, 36) = 23.2, p < .001) was qualified by the
predicted Condition × Group interaction, F(3, 36) = 3.2, p < .04.
As with trait ratings, level of in-group favoritism was tested within
each condition. Participants in the control condition rated valence
of words in the context of tile in-grou p (M = 7.2) more positively
than in the context of the out-group (M = 2.7), t(9) = 4.0, p <
.003. Participants in the similarity condition rated the connotative
meaning of the words marginally more positively in relation to the
in-group (M = 4.8) than in relation to the out-group (M = 2.9),
t(9) = 2.1, p < .067. Participants in the behavioral-recall condition
rated the words in the context of the in-group (M = 5.6) marginally
more favorably than in the context of the out-group (M = 3.3),
t(9) = 2.1, p = .06. However, participants in the perspective-
taking condition did not differentiate the connotative meaning of
the words between the in-group (M = 6.1) and the out-group
(M = 5.6), t(9) < 1.
One-way ANOVAs were conducted separately for the ratings of
the in-group and the out-group across experimental condition.
Unlike the trait ratings, no significant difference emerged when
looking only at the valence scores for in-group ratings; the one-
way ANOVA was not significant, F(3, 36) < 1. Although the
one-way ANOVA did not reach significance for the out-group
evaluations, F(3, 36) = 1.9, p =. 15, a contrast testing our specific
hypothesis that perspective-taking would increase the valence and
connotative meaning of words in the context of the out-group
compared with the other three conditions was significant, F(1,
36) = 2.3, p = .02. For out-group ratings, perspective-takers had
higher valence scores compared with the other three conditions.
The assignment of both traits and the valence of those traits, their
connotative meaning, in the context of the in-group and the out-
group suggests that perspective-taking decreased bias in the min-
imal group paradigm by increasing the evaluation of the out-group.
Satisfaction with group membership and perceived influence of
estimation tendency.
No significant main effects emerged for the
items measuring perceived influence of estimation tendency on
dally life. Although perspective-takers did not show the same level
of in-group favoritism, they were just as satisfied with their esti-
mation tendency and they saw it as equally influential in their dally
activities as participants in the other conditions.
Given that perspective-takers did not show any reduction on the
valence ratings of traits when eValuating the in-group, the main
conclusion from this experiment is that the perspective-taking
reduced bias by increasing evaluations of the out-group. This
benefit appears to occur independently of the recall of autobio-
graphical memories that contradict the implications of group mem-
bership. The results demonstrate the success of perspective-taking
at alleviating intergroup bias, even when there is no known content
of the stereotype of the out-group or a specific target individual
whose perspective one has taken (as participants were told to take
the perspective of "an underestimator," not a specific individual).
The results, along with those of Experiment 2, suggest that
perspective-taking increased the evaluations of the out-group
through the creation of a cognitive representation of the out-group
that now overlaps with the participants' own self-representation.
This is consistent with the work of Otten and Wentura (1998),
Cadinu and Rothbart (1996), and Smith and Henry (1996), all of
whom presented evidence that in-group favoritism is created and
perpetuated by extending positive self-representations to encom-
pass the in-group. These results also extend those of Experiment 2,
which demonstrated that taking the perspective of a specific indi-
vidual can affect representations of the out-group.
In addition, the experiment demonstrated for the first time in the
minimal group paradigm that the evaluative meanings of group-
relevant traits (see Brewer, 1979) differ when placed in the context
of in-group evaluations than when placed in the context of out-
group evaluations. Participants in the non-perspective-taking con-
ditions rated the evaluative meaning of group relevant traits (e.g.,
loyal, cooperative, kind) more positively when rating the traits'
meanings in relation to the in-group than in relation to the out-
group. This evaluative shift has important implications for inter-
group relations. The in-group may be suspicious and have tacit
distrust of any positive behaviors by the out-group, regardless of
their genuineness. Research on minority influence has also dem-
onstrated such attfibutional shifts in connotative meaning when a
majority group member judges a fellow majority group member
versus a minority group member (see, e.g., Clark & Maass, 1988).
Perspective-taking eliminated this shift, maintaining the positive
implications of the words when rating valenced meaning in the
context of the out-group. For perspective-takers, kind behaviors by
the out-group or minority in-group might be taken at face value, as
a sign of authentic positive regard.

720 GALINSKY AND MOSKOWITZ
General Discussion
In the three experiments, we focused on the role of perspective-
taking in reducing a number of biases that have implications for
stereotyping and intergroup relations. The results support the con-
tention that perspective-taking is a successful strategy for debias-
ing social thought. Perspective-taking tended to increase the ex-
pression of positive evaluations of the target, reduced the
expression of stereotypic content, and prevented the hyperacces-
sibility of the stereotype construct. These positive consequences
occur regardless of whether the stereotype under consideration is
not very socially sensitive (e.g., the elderly) or, as Galinsky and
Moskowitz (1999) found, particularly socially sensitive (e.g., Af-
rican Americans). Group-based judgments became more positive
even when stereotypic content about an out-group was unknown.
Perspective-taking decreased differential treatment of in-groups
and out-groups in the minimal group paradigm by increasing the
applicability of positive traits and the evaluative meaning of those
traits when judging the out-group.
evaluations toward the target in their narrative essays. Perspective-
taking also prevented hyperaccessibility of the stereotype on a task
that prevented the conscious use of the perspective-taking goal--
participants responded more slowly to stereotype-consistent words
compared with suppressors in a lexical decision task. In addition,
the effects on the lexical decision latencies were independent of
the evaluations participants expressed in their essays. Davis et al.
(1996) suggested that increased self-other overlap was a noncon-
scious effect of perspective-taking that was independent of liking
or perceived similarity. Experiment 2 extended Davis et al.'s
finding of increased self-other overlap to the situation in which
the self now overlaps more with the out-group. Similar to Davis et
al.'s findings, the results of increased self-group overlap were
mostly independent of the evaluations expressed in the essays. The
combination of Experiments 1 and 2 suggest that perspective-
taking nonconsciously increases the accessibility of the self-
concept, which then diminishes the accessibility and application of
the stereotype because only one construct tends to be dominant at
any one time (Allport, 1954; Bruner, 1957; Macrae et al., 1995).
Increasing Self-Other Overlap Through
Perspective-Taking
Self-Other Overlap and Prosocial Behavior
The pattern of data across all the experiments support re-
cent theorizing that perspective-taking produces positive conse-
quences-from shifting attributions to sympathy to providing help
to those in need--by increasing the overlap between the self and
the target of perspective-taking (Cialdini et al., 1997; Davis et al.,
1996; Regan & Totten, 1975). Experiment 2 provided direct evi-
dence that the positive effects of perspective-taking occur through
increasing the overlap between representations of the self and the
target group. Representations of the group are assimilated to the
activated self-concept and this process decreased stereotypic re-
sponding. Previous research documenting self-other overlap after
perspective-taking (Cialdini et al., 1997; Davis et al., 1996; Neu-
berg et al., 1997) have looked only at the overlap between the self
and individual targets. Wright, Aron, McLaughlin-Volpe, and
Ropp (1997) found that knowledge of an in-group member's close
relationship with an out-group member improved attitudes toward
the out-group. Wright et al. (1997) proposed, but did not provide
evidence, that this effect is mediated by the processes of self-other
overlap:
In an observed in-group/out-group friendship, the in-group member is
part of the self, the out-group member is part of that in-group mem-
ber's self, and hence part of myself.., then to some extent the
out-group is part of myself. (p. 76)
Our research demonstrates that this increase in overlap extends to
include the group that the target represents. Perspective-taking
changes representations of the group to be more self-like. The self
was applied to representations of the elderly and of African Amer-
icans in Galinsky and Moskowitz (1999), and even underestima-
tors. Increasing the overlap between representations of self and
representations of the out-group may go a long way in alleviating
out-group hostilities and promoting minority influence.
The results also support the distinction between conscious,
explicit processes and nonconscious, implicit processes. In Exper-
iment 1, perspective-taking decreased bias on a conscious task
related to the perspective-taking manipulation; participants limited
the expression of stereotypic content and expressed more positive
The increased self-other overlap has important implications for
the on-going debate over whether increased helping after
perspective-taking is truly altruistic or egoistically motivated (Bat-
son, 1997; Batson, Sager, et al., 1997; Cialdini et al., 1997;
Neuberg et al., 1997). Cialdini et al. presented evidence that the
conditions that increase empathic concern also increase self-other
target overlap. In addition, the relationship between empathy and
willingness to help was eliminated when the degree of self-other
overlap was statistically controlled. Cialdini et al. claimed that
helping follows empathy because empathy serves as an emotional
signal of oneness. Thus, there is evidence that self-other merging
can reliably mediate the link between perspective-taking and
helping.
Cialdini et al. (1997) raised the question that if empathy-induced
helping is caused by increased self-other overlap, then seemingly
altruistic acts may be selfishly motivated. If perspective-taking
blurs the distinction between self and other, then it becomes harder
to distinguish the selflessness of altruism from a more selfish
motivation. Cialdini et al. discuss the possibility that it is not just
the seeing more of the other in oneself that increases helping, but
also seeing more of oneself in the other. Given that humans are
driven by inclusive fitness concerns (Hamilton, 1964), the desire to
ensure the welfare and survival of their genes above and beyond
self-survival concerns, then feelings of oneness increase the belief
that the self really is in the other and therefore activates the
motivation to help.
Batson, Sager, et al. (1997) suggested that the effects of
perspective-taking on helping are independent of self-other over-
lap. In their studies, participants took either the perspective of a
student from their university or the perspective of a student from
a rival university. The link between empathy and helping was
unqualified by the group membership of the target. They suggested
that this lack of difference, along with failures to find any medi-
ation for their measures of self-other merging, demonstrates that
perspective-taking is independent of perceived connections to
targets. The research presented here suggests that although
perspective-taking may not depend on shared group memberships,

PERSPECTIVE-TAKING 721
the degree of self-target overlap is a critical determinant of
whether perspective-taking will be effective.
Gestures of altruism, of heroic helping, may be selfishly moti-
vated. Recognizing the contemporary discrimination against dis-
advantaged groups after perspective-taking (Galinsky, 1999a) may
depend on seeing the self as a potential victim. If we see more of
ourselves in the elderly, then we will be kinder to them and less
likely to activate stereotypic representations of that group. The
paradox is that selfishly motivated acts can produce positive social
consequences.
Perspective-Taking and Stereotype Accessibility
The results of Experiment 1 and those of Galinsky and Mos-
kowitz (1999) suggest that perspective-taking can be a useful
strategy for decreasing the accessibility of stereotypes. Galinsky
and Moskowitz (1999) found no differences in stereotype acces-
sibility between a perspective-taking condition and a no-prime
control, suggesting that perspective-taking prevented the activation
of the stereotype. In the present article, we have argued that
increased self-outgroup overlap diminishes the accessibility of the
stereotype. Given that knowledge structures compete for the right
to categorize a stimulus (Bruner, 1957) and that only one category
tends to be dominant at a time (Macrae et al., 1995), the activation
of the self-concept wins the battle for categorization over the
stereotype.
With the accessibility of the stereotype curtailed, stereotypes are
less likely to color perceptions, memory, and judgment or to be
used to categorize ambiguous behaviors (for reviews of stereotype
effects, see Stangor & Lange, 1994; yon Hippel, Sekaquaptewa, &
Vargas, 1995). Because perceived differences and self-threats
(Fein & Spencer, 1997; Spencer, Fein, Wolf, Hodgson, & Dunn,
1998) can increase the accessibility of stereotypes, the self-target
(Davis et al., 1996) and self-group overlap (Experiment 2) asso-
ciated with perspective-taking can help diminish the tendency to
activate stereotypes. As Brewer (1988, 1996) noted, stereotypes
are less likely to be activated when a strong interpersonal orien-
tation characterizes the relationship between a perceiver and a
target; relationship-based and situationally relevant traits are more
likely to become activated rather than the stereotype.
Recent research suggests that the automatic activation of ste-
reotypes does not occur for all individuals. Fazio, Jackson, Dun-
ton, and Williams (1995), Lepore and Brown (1997) and Mosko-
witz, Gollwitzer, Wasel, and Schaal (1999) have all shown that a
select group of participants do not demonstrate automatic activa-
tion effects. The Lepore and Brown study showed that a group
representation is automatically activated by racial primes but that
there is variability in the amount of overlap between the group
representation and the stereotype. Those studies suggested that for
some individuals the negative stereotype for a group can be con-
trolled and replaced with a more positive representation. In the
studies presented here, the increase in self-outgroup overlap
changes the representation of the group to be more positive. There
is now more overlap between the representation of the group with
the self than with the stereotype.
Other research has looked at situations and temporary goals that
reduce the accessibility of stereotypes. Gilbert and Hixon (1991)
and Spencer et al. (1998) found that cognitive load disrupted
stereotype activation. Macrae et al. (1997) demonstrated that pro-
cessing goals focused on content irrelevant to social categorization
(e.g., looking for a dot on a picture of a woman) prevented the
activation of stereotypes. Kawakami et al. (1998) found the link
between the group representation and the stereotype can be sev-
ered by training participants to replace the stereotype with differ-
ent beliefs. Gollwitzer, Schaal, Moskowitz, Hammelbeck, and
Wasel (1999) demonstrated that the intention to be egalitarian,
supplemented by practicing an egalitarian response to stereotypic
targets, can control the activation of stereotypes.
Perspective-taking is perhaps another useful processing strategy
that temporarily dissipates the strength of the links between group
representations and the stereotype. In fact, perspective-taking may
be a particularly pragmatic strategy because attentional focus re-
mains on the individual. Perspective-taking allows for the individ-
ual to diminish the level of stereotype accessibility by remaining
present in an interracial or intergender interaction rather than
requiting psychological disengagement and withdrawal.
Perspective-Taking and Suppression: Approach and
Avoidance
Stereotype suppression is effective at controlling responses on
explicit, obtrusive measures of stereotyping but has deleterious,
unintended consequences on nonconscious, implicit stereotyping.
Consistent with previous research (Monteith et al., 1998; Wyer et
al., 2000), rebound effects did not emerge in our research on
measures (e.g., narrative essays) that alerted suppressors to keep
the goal of suppression active. When response assessment ap-
peared to have no relevance to stereotypes or suppression (lexical
decision task), rebouiad effects emerged (see also Monteith et al.,
1998).
Perspective-taking and thought suppression are strategies that
can be used to decrease the accessibility and application of ste-
reotypes. An important difference between these strategies is how
one navigates toward that goal. Stereotype suppression can be
conceived of as an avoidance- or prevention-oriented strategy and
perspective-taking as an approach- or promotion-oriented strategy.
Higgins (1998) pointed out that goals can be framed either in terms
of prevention or promotion; an individual taking a test may be
concerned with not getting a bad grade (prevention) or with
striving toward a good grade (promotion). Similarly, Wegner and
Wenzlaff (1996) discussed the distinction between strategies of
mental control geared toward approaching a particular mental state
and strategies aimed at avoiding a particular mental state. Mos-
kowitz (1996) and Devine (1998) used a similar distinction in
distinguishing between internally motivated control of prejudice,
in which an individual approaches the goal of being nonprejudiced,
and externally motivated control of prejudice, in which an indi-
vidual retreats from being or appearing prejudiced.
Suppression of stereotypic knowledge is a strategy designed to
avoid the state of prejudice. The activation of the avoidance or
prtvention regulatory system produces a general tendency to avoid
intergroup interactions (Devine, 1998) and to display distancing
behaviors in intergroup interactions (Macrae et al., 1994; Word,
Zanna, & Cooper, 1974). Suppression affects not only the acces-
sibility of stereotypes, but also the immediate and delayed moti-
vations of the suppressor.
Perspective-taking focuses attention outward toward the target
individual. The potential activation of an approach or promotion
focus by perspective-taking might have a number of beneficial
effects that could contribute to the reduction of stereotyping. An

722 GAL1NSKY AND MOSKOWITZ
approach or promotion focus can affect persistence at a task in the
face of potential failure and, if failure has occurred, can produce
feelings of guilt and recrimination (Devine, Monteith, Zuwerink,
& Elliot, 1991) that aid in the establishment of coping strategies
used to respond more effectively in the future (Monteith, 1993;
Monteith & Voils, in press; Moskowitz, in press). The approach
orientation of perspective-taking may lead to increased intergroup
contact; creation of more interdependent, ethnically mixed groups;
more positive interactions; and a reduction in mistrust (Aronson,
Blaney, Stephan, Sikes, & Snapp, 1978; Sherif, 1966).
Limits of and Biases Inherent in Perspective-Taking
Experiments presented by Galinsky (1999a, 1999b) also estab-
lished some boundary conditions for the effects of perspective-
taking to emerge. First, the effects appear to be group specific; in
Galinsky's (1999a) study, taking the perspective of an African
American target increased awareness of continued discrimination
directed toward African Americans but taking the perspective of
that target did not increase, but decreased, sensitivity to discrim-
ination directed toward women. The consequences of taking the
perspective of a member of one social group do not seem to
generalize beyond that group. Second, Galinsky (1999b) found
that the beneficial outcomes of perspective-taking required instruc-
tions that are especially vivid, process oriented, and descriptive.
Other research has found that perspective taking may not always
prove successful. Dovidio, Allen, and Schroeder (1990) found that,
after inducing empathy, assistance only increased for the problem-
atic situation for which empathy was induced and did not activate
a more general tendency to help. Thompson (1993) found that
when a mutually beneficial outcome to a negotiation was not
possible, no attenuation of in-group favoritism occurred.
Perspective-taking can also conflict with other beliefs and norms,
such as justice, fairness, and equity. Because the target of
perspective-taking gets accorded "favored" status, perspective-
taking can lead to preferential treatment of the targets of
perspective-taking (Batson, Klein, Highberger, & Shaw, 1995).
Batson, Klein, et al. argued that empathy-induced altruism (the
increased concern for the welfare of another) and morality (defined
as the upholding of a given moral principal) should be considered
independent social motives because their participants altered a
system of equal distribution to preferentially allocate resources to
the target of empathy. In addition, perspective-taking can reduce
overall contributions to the collective in a social dilemma para-
digm by preferentially allocating resources to the target of empa-
thy (Batson, Batson, et al., 1995).
Conclusion
Despite the progressive drive toward equality in the United
States at both the philosophical and constitutional level, negative
stereotypes persist. These stereotypes are known, if not applied, by
the vast majority of the populace (Devine, 1989). The entrench-
ment of stereotypes makes social relations potentially problematic
and can infuse interactions with palpable feelings of unease and
discomfort. How can one defend egalitarian principles against the
everyday activation of stereotypes? The current research explored
two different strategies for debiasing social thought--stereotype
suppression and perspecfive-taking. Stereotype suppression ap-
pears to be an effective strategy for reducing the expression of
stereotypes, but it not only fails to reduce but can exacerbate bias
where stereotyping is most insidious and invidious, at the implicit,
nonconscious level. Perspective-taking, however, appears to di-
minish not just the expression of stereotypes but their accessibility.
The constructive process of taking and realizing another per-
son's perspective furthers the egalitarian principles themselves;
perspective-taking is an effective reinforcement of contemporary
admonitions to consider previously ignored or submerged perspec-
fives as a routine part of social interchange and inquiry.
References
Allport, G. (1954). The nature of prejudice. Reading, MA: Addison-
Wesley.
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.
Aronson, E., Blaney, N., Stephan, L., Sikes, J., & Snapp, M. (1978). The
jigsaw classroom. Beverly Hills, CA: Sage.
Bargh, J. A. (1997). The automaticity of everyday life. In R. S. Wyer, Jr.
(Ed.), Advances in social cognition: Vol. 10. The automaticity of every-
day life (pp. 1-61). Mahwah, NJ: Erlbaum.
Batson, C. D. (1991). The altruism question: Toward a social-
psychological answer. Hillsdale, NJ: Erlbaum.
Batson, C. D. (1997). Self-other merging and the empathy-altruism hy-
pothesis: Reply to Neuberg et al. (1997). Journal of Personality and
Social Psychology, 73, 517-522.
Batson, C. D. (1998). Altruism and prosocial behavior. In D. T. Gilbert,
S. T. Fiske, & G. Lindsey (Eds.), The handbook of social psychology
(Vol. 2, pp. 282-316). Boston: McGraw Hill.
Batson, C. D., Batson, J. G., Todd, M. R., Brummet, B. H., Shaw, L. L., &
Aldeguer, M. R. (1995). Empathy and the collective good: Caring for
one of the others in a social dilemma. Journal of Personality and Social
Psychology, 68, 619-631.
Batson, C. D., Early, S., & Salvarani, G. (1997). Perspective taking:
Imagining how another feels versus imagining how you would feel.
Personality & Social Psychology Bulletin, 23, 751-758.
Batson, C. D., Klein, T. R., Highberger, L., & Shaw, L. L. (1995).
Immorality from empathy-induced altruism: When compassion and jus-
tice conflict. Journal of Personality and Social Psychology, 68, 1042-
1054.
Batson, C. D., Sager, K., Garst, E., Kang, M., Rubchinsky, K., & Dawson,
K. (1997). Is empathy-induced helping due to self-other merging?
Journal of Personality and Social Psychology, 73, 495-509.
Batson, C. D., Sympson, S. C., Hindman, J. L., De.cruz, P., Todd, R. M.,
Weeks, J. L., Jennings, J., & Burris, C. T. (1996). "I've been there, too":
Effect on empathy of prior experience with a need. Personality & Social
Psychology Bulletin, 22, 474-482.
Bodenhansen, G. V., & Wyer, R. S. (1985). Effects of stereotypes on
decision-making and information-processing strategies. Journal of Per-
sonality and Social Psychology, 48, 267-282.
Brewer, M. B. (1979). In-group bias in the minimal intergroup situation: A
cognitive-motivational analysis. Psychological Bulletin, 86, 307-324.
Brewer, M. B. (1988). A dual process model of impression formation. In
K. T. Srull & R. S. Wyer, Jr. (Eds.), Advances in social cognition
(Vol. 1, pp. 1-36). Hillsdale, NJ: Erlbanm.
Brewer, M. B. (1996). When stereotypes lead to stereotyping: The use of
stereotype in person perception. In C. N. Macrae, C. Stangor, & M.
Hewstone (Eds.), Stereotypes and stereotyping (pp. 254-275). New
York: Guilford Press.
Brewer, M. B., Dull, V., & Lui, L. (1981). Perceptions of the elderly:
Stereotypes as prototypes. Journal of Personality and Social Psychol-
ogy, 41, 656-670.
Bruner, J. S. (1957). On perceptual readiness. Psychological Review, 64,
123-152.

PERSPECTIVE-TAKING 723
Cadinu, M. R., & Rothbart, M. (1996). Self-anchoring and differentiation
process in the minimal group setting.
Journal of Personality and Social
Psychology, 70,
661-677.
Chaiken, S. (1980). Heuristic versus systematic information processing and
the use of source versus message cues in persuasion.
Journal of Per-
sonality and Social Psychology, 39,
752-766.
Cialdini, R. B., Brown, S. L., Lewis, B. P., Luce, C., & Neuberg, S. L.
(1997). Reinterpreting the empathy-altruism relationship: When one
into one equals oneness.
Journal of Personality and Social Psychol-
ogy, 73,
481-494.
Cialdini, R. B., Schaller, M., Houlihan, D., Arias, K., Fultz, J., & Beaman,
A. L. (1987). Empathy-based helping: Is it selflessly or selfishly moti-
vated?
Journal of Personality and Social Psychology, 52,
749-758.
Clark, R. D., III.,& Maass, A. (1988). Social categorization in minority
influence: The case of homosexuality.
European Journal of Social
Psychology, 18,
347-364.
Clore, G. L., & Jeffery, K. M. (1972). Emotional role playing, attitude
change, and attraction toward a disabled person.
Journal of Personality
and Social Psychology, 23,
105-111.
Davis, M. H. (1983). Measuring individual differences in empathy: Evi-
dence for a multidimensional approach.
Journal of Personality and
Social Psychology, 44,
113-126.
Davis, M. H., Corddin, L., Smith, A., & Loce, 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.
Devine, P. G. (1989). Stereotypes and prejudice: Their automatic and
controlled components.
Journal of Personality and Social Psychol-
ogy, 56,
5-18.
Devine, P. G. (1998). Beyond the isolated social perceiver: Why inhibit
stereotypes? In R. S. Wyer, Jr. (Ed.),
Advances in social cognition:
Vol. 11. Stereotype activation and inhibition
(pp. 69-81). Mahwah, NJ:
Erlbaum.
Devine, P. G., Monteith, M. J., Zuwerink, J. R., & Elliot, A. J. (1991).
Prejudice with and without compunction.
Journal of Personality and
Social Psychology, 60,
817-830.
Dovidio, J. F., Allen, J. L., & Schroeder, D. A. (1990). The specificity of
empathy-induced altruism: Evidence of altruistic motivation.
Journal of
Personality and Social Psychology, 59,
249-260.
Duncan, B. L. (1976). Differential social perception and attribution of
intergroup violence: Testing the lower limits of stereotyping of blacks.
Journal of Personality and Social Psychology, 34,
590-598.
Esses, V. M., & Zanna, M. P. (1995). Mood and the expression of ethnic
stereotypes.
Journal of Personality and Social Psychology, 69,
1052-
1068.
Fazio, R. H. (1990). A practical guide to the use of response latency in
social psychological research. In C. Hendrick & M. Clark (Ed.),
Re-
search methods in personality and social psychology. Review of person-
ality and social psychology
(Vol. ll, pp. 74-97). Newbury Park, CA:
Sage.
Fazio, R. H., Jackson, J. R., Dunton, B. C., & Williams, C. J. (1995).
Variability in automatic activation as an unobtrusive measure of racial
attitudes: A bona fide pipeline?
Journal of Personality and Social
Psychology, 69,
1013-1027.
Fein, S., & Spencer, S. J. (1997). Prejudice as self-image maintenance:
Affirming the self through negative evaluations of others.
Journal of
Personality and Social Psychology, 73,
31-44.
Fiske, S. T., Bersoff, D. N., Borgida, E., Deaux, K., & Heliman, M. E.
(1991). Social science research on trial: The use of sex stereotyping
research in
Price Waterhouse v. Hopkins. American Psychologist, 46,
1049-1060.
Gaertner, S. L., Mann, J., Murrell, A., & Dovidio, J. F. (1989). Reducing
intergroup bias: The benefits of recategorization.
Journal of Personality
and Social Psychology, 57,
239-249.
Galinsky, A. D. (1999a).
Perspective taking: Debiasing social thought.
Unpublished doctoral dissertation, Princeton University.
Galinsky, A. D. (1999b). The
benefits and limits of perspective taking in
decreasing the solicitation of expectancy-consistent information.
Manu-
script submitted for publication.
Galinsky, A. D., & Moskowitz, G. B. (1999).
Stereotype suppression and
perspective-taking in debiasing social thought: The role of stereotype
and coanterstereotype accessibility.
Manuscript submitted for publica-
tion.
Gilbert, D. T., & Hixon, J. G. (1991). The trouble of thinking: Activation
and application of stereotypic beliefs.
Journal of Personality and Social
Psychology, 60,
509-517.
Gollwitzer, P. M., Schaal, B., Moskowitz, G. B., Hammelbeck, H. J. P., &
Wasel, W. (1999).
Implementation effects on stereotype and prejudice
activation.
Manuscript submitted for publication.
Hamilton, W. D. (1964). The genetic evolution of social behavior.
Journal
of Theoretical Biology,
7, 1-52.
Higgins, E. T. (1989). Knowledge accessibility and activation: Subjectivity
and suffering from unconscious sources. In J. S. Uleman & J. A. Bargh
(Eds.),
Unintended thought
(pp. 75-123). New York: Guilford Press.
Higgins, E. T. (1996). Knowledge activation: Accessibility, applicability,
and salience. In E. T. Higgins & A. W. Kruglanski (Eds.),
Social
psychology: Handbook of basic principles
(pp. 133-168). New York:
Guilford Press.
Higgins, E. T. (1998). Promotion and prevention: Regulatory focus as a
motivational principle. In M. P. Zanna (Ed.),
Advances in experimental
socialpsychology
(Vol. 30, pp. 1-41). San Diego, CA: Academic Press.
Howard, J. W., & Rothbart, M. (1980). Social categorization and memory
for in-group and out-group behavior.
Journal of Personality and Social
Psychology, 38,
301-310.
Jones, E. E., & Nisbett, R. E. (1971). The
actor and the observer."
Divergent perceptions of the causes of behavior.
Morristown, NJ: Gen-
eral Learning Press.
Kawakami, K., Dion, K. L. & Dovidio, J. F. (1998).
Kicking the habit: The
effects of suppression training on stereotype activation.
Unpublished
manuscript, University of Nijmegen, the Netherlands.
Kohlberg, L. (1976). Moral stages and moralization: The cognitive-
developmental approach. In T. Lickona (E,d.),
Moral development and
behavior
(pp. 31-53). New York: Holt, Rinehart & Winston.
Lepore, L., & Brown, R. (1997). Category and stereotype activation: Is
prejudice inevitable?
Journal of Personality and Social Psychology, 72,
275-287.
Locksley, A., Ortiz, V., & Hepburn, C. (1980). Social categorization and
discriminatory behavior: Extinguishing the minimal intergroup discrim-
ination effect.
Journal of Personality and Social Psychology, 39,
773-
783.
Macrae, C. N., Bodenhansen, G. V., & Miine, A. B. (1995). The dissection
of selection in person perception: Inhibitory processes in social stereo-
typing.
Journal of Personality and Social Psychology, 69,
397-407.
Macrae, C. N., Bodenhansen, G. V., & Miine, A. B. (1998). Saying no to
unwanted thoughts: Self-focus and the regulation of mental life.
Journal
of Personality and Social Psychology, 74,
578-589.
Macrae, C. N., Bodenhansen, G. V., Miine, A. B., & Jetten, J. (1994). Out
of mind but back in sight: Stereotypes on the rebound.
Journal of
Personality and Social Psychology, 67,
808-817.
Macrae, C. N., Bodenhansen, G. V., Milne, A. B., Thorn, T. M. J., &
Castelli, L. (1997). On the activation Of social stereotypes: The moder-
ating role of processing objectives.
Journal of Experimental Social
Psychology, 33,
471-489.
Macrae, C. N., Bodenhausen, G. V., Milne, A. B., & Wheeler, V. (1996).
On resisting the temptation for simplification: Counterintentional effects
of stereotype suppression on social memory.
Social Cognition, 14,
1-20.
Monteith, M. J. (1993). Self-regulation of prejudiced responses: Implica-
tions for progress in prejudice-reduction efforts.
Journal of Personality
and Social Psychology, 65,
469-485.

724
GALINSKY AND MOSKOWITZ
Monteith, M. J., Spicer, C. V., & Tooman, G. D. (1998). Consequences of
stereotype suppression: Stereotypes on AND not on the rebound. Jour-
nal of Experimental Social Psychology, 34, 355-377.
Monteith, M. J., & Voils, C. I. (in press). Exerting control over prejudiced
responses. In G. B. Moskowitz (Ed.), Cognitive social psychology: On
the tenure and future of social cognition. Hillsdale, NJ: Erlbaum.
Moskowitz, G. B. (1996). The mediational effects of attributions and
information processing in minority social influence. British Journal of
Social Psychology, 35, 47-66.
Moskowitz, G. B. (in press). On preconscious control: Compensatory
cognition and the implicit regulation of goal pursuit, In G. B. Moskowitz
(Ed.), Cognitive social psychology: On the tenure and future of social
cognition. Hillsdale, NJ: Erlbaum.
Moskowitz, G. B., Gollwitzer, P. M., Wasel, W., & Schaal, B. (1999).
Preconscious control of stereotype activation through chronic egalitarian
goals. Journal of Personality and Social Psychology, 77, 167-184.
Neuberg, S. L., Cialdini, R. B., Brown, S. L., Luce, C., Sagarin, B. J., &
Lewis, B. P. (1997). Does empathy lead to anything more than super-
ficial helping? Comment on Batson et al. (1997). Journal of Personality
and Social Psychology, 73, 510-516.
Neuberg, S. L., & Fiske, S. T. (1987). Motivational influences on impres-
sion formation: Outcome dependency, accuracy-driven attention, and
individuating processes. Journal of Personality and Social Psychol-
ogy, 53, 431-444.
Otten, S., & Moskowitz, G. B. (2000). Evidence for implicit evaluative
in-group bias: Affect-biased spontaneous trait inference in a minimal
group paradigm. Journal of Experimental Social Psychology, 36, 77-89.
Otten, S., & Wentura, D. (1998). About the impact of automaticity in the
minimal group paradigm: Evidence from an affective priming task.
Unpublished manuscript.
Perdue, C. W., & Gurtman, M. B. (1990). Evidence for the automaticity of
ageism. Journal of Experimental Social Psychology, 26, 199-216.
Pettigrew, T. F. (1979). The ultimate attribution error: Extending Allport's
cognitive analysis of prejudice. Personality and Social Psychology Bul-
letin, 5, 461-476.
Petty, R. E., & Cacioppo, J. T. (1986). The elaboration likelihood model of
persuasion. In L. Berkowitz (Ed.), Advances in experimental social
psychology (Vol. 19, pp. 123-205). New York: Academic Press.
Piaget, J. (1932). The moral judgment of the child. London: Kegan, Paul,
Trench & Trubner.
Regan, D. T., & Totten, J. (1975). Empathy and attribution: Turning
observers into actors. Journal of Personality and Social Psychology, 32,
850-856.
Richardson, D. R., Hammock, G. S., Smith, S. M., Gardner, W., & Signo,
M. (1994). Empathy as a cognitive inhibitor of interpersonal aggression.
Aggressive Behavior, 20, 275-289.
Ross, L., Lepper, M. R., & Hubbard, M. (1975). Perseverance in self-
perception and social perception: Biased attributional processes in the
debriefing paradigm. Journal of Personality and Social Psychology, 32,
880-892.
Sherif, M. (1966). Group conflict and co-operation: Their social psychol-
ogy. London: Routledge & Kegan Paul.
Sherman, J. W., Stroessner, S. J., Loftus, S. T., & Degnzman, G. (1997).
Stereotype suppression and recognition memory for stereotypical and
nonstereotypical information. Social Cognition, I5, 205-215.
Smith, E. R., Coats, S., & Walling, D, (1999). Overlapping mental repre-
sentations of self, in-group, and parmer: Further response time evidence
and a counectionist model. Personality & Social Psychology Bulle-
tin, 25, 873-882.
Smith, E. R., & Henry, S. (1996). An in-group becomes part of the self:
Response time evidence. Personality & Social Psychology Bulletin, 22,
635-642.
Spencer, S. J., Fein, S,, Wolf, C. T., Hodgson, H. L., & Duun, M. A.
(1998). The automatic activation of stereotypes: The role of self-image
threat. Personality and Social Psychology Bulletin, 24, 1139-1152.
Stangor, C., & Lange, J. (1994). Mental representations of social groups:
Advances in understanding stereotypes and stereotyping. In M. P. Zarma
(Ed.), Advances in experimental social psychology (Vol. 26, 357-416).
San Diego, CA: Academic Press.
Storms, M. D. (1973). Videotape in the attribution process: Reversing
actors' and observers' points of view, Journal of Personality and Social
Psychology, 27, 165-175.
Stotland, E. (1969). Exploratory studies in empathy. In L. Berkowitz (Ed.),
Advances in experimental social psychology (Vol. 4, pp. 271-313). San
Diego, CA: Academic Press.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social
categorization and intergroup behavior. European Journal of Social
Psychology, 1, 149-178.
Taylor, D. M., & Jaggi, V. (1974). Ethnocentrism and causal attribution in
a South Indian context. Journal of Cross-Cultural Psychology, 5, 162-
171.
Tetlock, P. E., Skitka, L., & Boettger, R. (1989). Social and cognitive
strategies for coping with accountability: Conformity, complexity, and
bolstering. Journal of Personality and Social Psychology, 57, 632-640.
Thompson, L. (1993). The impact of negotiation on intergroup relations.
Journal of Experimental Social Psychology, 29, 304-325.
Turner, J. C. (1987). Rediscovering the social group: A self-categorization
theory. Oxford: Basil Blackwell.
yon Hippel, W,, Sekaquaptewa, D., & Vargas, P. (1995). On the role of
encoding processes in stereotype maintenance. In M. P. Zanna (Ed.),
Advances in experimental socialpsychology (Vol. 26, pp. 177-254). San
Diego, CA: Academic Press.
Wegner, D. M. (1994). Ironic processes of mental control. Psychological
Review, 101, 34-52.
Wegner, D. M., Erber, R., & Bowman, R. E. (1993). On trying not to be
sexist. Unpublished manuscript.
Wegner, D. M., Schneider, D. J., Carter, S., m., & White, L. (1987).
Paradoxical effects of thought suppression. Journal of Personality and
Social Psychology, 58, 409-418.
Wegner, D. M., & Wenzlaff, R. M. (1996). Mental control. In E. T. Higgins
& A. W. Kruglanski (Eds.), Social psychology: Handbook of basic
principles (pp. 466-492). New York: Guilford Press.
Wilder, D. A. (1978). Reduction of intergroup discrimination through
individuation of the out-group. Journal of Personality and Social Psy-
chology, 36, 1361-1374.
Wilder, D. A. (1986). Social categorization: Implications for creation and
reduction of intergroup bias. In L. Berkowitz (Ed.), Advances in exper-
imental social psychology (Vol. 19, pp. 293-355). San Diego, CA:
Academic Press.
Word, H. H., Zanna, M. P., & Cooper, J. (1974). The nonverbal mediation
of self-fulfilling prophecies in interracial interaction. Journal of Exper-
imental Social Psychology, 10, 109-120.
Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The
extended contact effect: Knowledge of cross-group friendships and
prejudice. Journal of Personality and Social Psychology, 73, 73-90.
Wyer, N. A., Sherman, J. W., & Stroessner, S. J. (1998). The spontaneous
suppression of racial stereotypes. Social Cognition, 16, 340-352.
Wyer, N. A., Sherman, J. W., & Stroessner, S. J. (2000). The roles of
motivation and ability in controlling the consequences of stereotype
suppression. Personality and Social Psychology Bulletin, 26, 13-25.
Received June 11, 1999
Revision received November 19, 1999
Accepted November 22, 1999 •