ATTITUDES AND SOCIAL COGNITION
Social Exclusion Impairs Self-Regulation
Roy F. Baumeister and C. Nathan DeWall
Florida State University
Natalie J. Ciarocco
Florida Atlantic University
Jean M. Twenge
San Diego State University
Six experiments showed that being excluded or rejected caused decrements in self-regulation. In
Experiment 1, participants who were led to anticipate a lonely future life were less able to make
themselves consume a healthy but bad-tasting beverage. In Experiment 2, some participants were told
that no one else in their group wanted to work with them, and these participants later ate more cookies
than other participants. In Experiment 3, excluded participants quit sooner on a frustrating task. In
Experiments 4 – 6, exclusion led to impairment of attention regulation as measured with a dichotic
listening task. Experiments 5 and 6 further showed that decrements in self-regulation can be eliminated
by offering a cash incentive or increasing self-awareness. Thus, rejected people are capable of self-
regulation but are normally disinclined to make the effort.
Human beings rely on group life for their health, well-being,
comfort, and other positive outcomes. Being accepted into a social
group is therefore an almost indispensable goal of human striving.
Obtaining such acceptance is, however, a long and difficult task
that may entail years of learning how to behave in socially accept-
able ways, acquiring marketable skills, cultivating good relation-
ships, and building a favorable reputation. To succeed at those
endeavors, people must have an effective capacity for altering their
behavior so as to conform to externally (socially) defined stan-
dards. That capacity is often defined as self-regulation. It is there-
fore plausible that one of the overarching purposes of self-
regulation is to secure acceptance by others.
If self-regulation exists partly for the sake of securing and
maintaining social acceptance, then social rejection may affect it.
On theoretical grounds, people might respond to social exclusion
with either an improvement or an impairment of self-regulation.
The thrust of previous empirical findings led us to anticipate
impairments more than improvements, but in principle either out-
come would have been plausible and theoretically meaningful.
Why Social Exclusion Might Impair Self-Regulation
Multiple studies have shown that being accepted versus rejected
by social groups has a wide range of effects on individuals. Health,
happiness, and well-being are strongly tied to whether one is
accepted or rejected, such that people deprived of close social ties
suffer more negative physical and psychological consequences
than those with strong social networks (Cacioppo, Hawkley, &
Berntson, 2003; Lynch, 1979; Myers, 1992). Ostracized individu-
als exhibit a broad range of distress and pathology (Williams,
Laboratory studies of rejection have compared the behavior of
rejected participants with the behavior of those who experience
social acceptance, those who encounter some other misfortune that
does not entail rejection, or simple control condition participants
who do not have any particular experience. Rejected people are
more likely than the others to behave aggressively (Buckley,
Winkel, & Leary, 2004; Twenge, Baumeister, Tice, & Stucke,
2001). They are less likely to act in prosocial ways, such as
cooperating with someone or providing help (Twenge, Ciarocco,
Cuervo, & Baumeister, 2003). They exhibit an assortment of
cognitive deficits such as impaired logical reasoning (Baumeister,
Twenge, & Nuss, 2002). They show distorted time perception, an
emphasis on the present rather than the future, a seemingly lethar-
gic passivity, and an avoidance of self-awareness (Twenge, Ca-
tanese, & Baumeister, 2003). They also exhibit self-destructive
tendencies, as indicated by various increases in self-defeating
behaviors such as foolish risk taking and unhealthy choices
(Twenge, Catanese, & Baumeister, 2002).
Although many behavioral effects of rejection have been suc-
cessfully documented in these studies, the inner processes that
mediate them remain elusive. The simple initial hypothesis that
Roy F. Baumeister and C. Nathan DeWall, Department of Psychology,
Florida State University; Natalie J. Ciarocco, Department of Psychol-
ogy, Florida Atlantic University; Jean M. Twenge, Department of
Psychology, San Diego State University.
We gratefully acknowledge support from National Institutes of Health
Grants MH-57039, MH-65559, and MH-12329.
Correspondence concerning this article should be addressed to Roy F.
Baumeister, Department of Psychology, Florida State University, Talla-
hassee, FL 32306. E-mail: firstname.lastname@example.org
Journal of Personality and Social Psychology, 2005, Vol. 88, No. 4, 589 –604
Copyright 2005 by the American Psychological Association 0022-3514/05/$12.00 DOI: 10.1037/0022-3522.214.171.1249
emotional distress will mediate behavioral effects has not been
consistently supported, and if anything it has been more consis-
tently disconfirmed (Twenge, Catanese, & Baumeister, 2003;
Twenge et al., 2001, 2002). In many studies, rejected participants
do not report emotional states that differ significantly from the
states of accepted or control participants. Moreover, when differ-
ences in mood or emotion have been found, mediation analyses
have suggested that emotional reactions are essentially unrelated to
behavioral effects. Even laboratories that have reliably found sig-
nificant main effects of social exclusion on mood or emotion have
not found that it mediates behavioral outcomes (Buckley et al.,
If emotion is not the essential mediator between rejection and its
behavioral consequences, then what is? Other inner processes need
to be investigated. Cognitions would be one candidate, but the
cognitive effects of rejection documented thus far seem insuffi-
cient to explain the full range of behavioral effects. That is,
rejection may impair logical reasoning (Baumeister et al., 2002),
but it seems unlikely that the aggressive or antisocial behaviors of
rejected people stem from faulty logic.
Self-regulation seems a more promising candidate. Self-
regulation, defined as the capacity to control or alter one’s re-
sponses, is a vital mechanism for producing adaptive and socially
desirable behavior. If rejection could be shown to impair or un-
dermine self-regulation, then a broad range of socially undesirable
behaviors might ensue, consistent with what has already been
found regarding the behavior of rejected individuals.
The hypothesis that rejection impairs self-regulation is rendered
more plausible by three sets of empirical findings. First, it appears
that rejection impairs intellectual performance and cognitive pro-
cessing only when conscious, executive control is required. Auto-
matic information processing, such as in rote memory, is appar-
ently unaffected among rejected participants (Baumeister et al.,
2002). Self-regulation is closely related to executive control (and,
indeed, is probably one major form of it). Hence, the pattern of
cognitive impairments indirectly implicates self-regulatory deficits
as a possible cause.
Second, the behavioral findings contain a seeming contradic-
tion. Rejected participants exhibit more antisocial behavior and a
reduced willingness to perform altruistic, self-sacrificing actions
such as helping others. Such findings depict rejected individuals as
selfish. Yet other data show them also to be more self-defeating
than other people. For example, rejected people take more foolish
risks, make more unhealthy choices, and procrastinate more than
others (Twenge et al., 2002). Selfishness and self-defeating behav-
ior are seemingly at opposite ends of the spectrum of self-
interested behavior. They both tend to involve self-regulation
failures, however. Effective self-regulation is often needed to
make people overcome selfish impulses and do what is best for
others. Self-regulation is also needed for the pursuit of enlightened
self-interest, such as enabling the person to resist impulsive temp-
tations. The most common form of self-defeating behavior in-
volves seeking short-term benefits that are accompanied by long-
term costs, such as smoking cigarettes or even failing to delay
gratification (see Baumeister & Scher, 1988; Mischel, 1974,
1996). Self-regulation may have evolved to enable people to resist
such impulses and cultivate long-term gains.
Third, rejection appears to create a deconstructed mental state
that may be inimical or at least detrimental to self-regulation.
Twenge, Catanese, & Baumeister (2003) found that rejected peo-
ple had a distorted sense of the passage of time. Time perception
is linked to effective self-regulation, and as self-regulation deteri-
orates, time is perceived as moving more slowly (Vohs &
Schmeichel, 2003), just as it does among rejected persons. Like-
wise, rejection causes people to avoid self-awareness (Twenge,
Catanese, & Baumeister, 2003), and self-awareness is vital for
effective self-regulation (Carver & Scheier, 1981). Self-regulation
also benefits from meaningful thought, such as comparison with
standards (Carver & Scheier, 1981), whereas rejected people ex-
hibit decrements in meaningful thought (Baumeister et al., 2002).
Thus, again, the pattern of effects of rejection is broadly consistent
with the notion that self-regulation is impaired, though direct
evidence is lacking.
The broader context for this investigation is the view that human
social life contains an implicit bargain. People have a strong need
to belong (Baumeister & Leary, 1995), and they survive, flourish,
and reproduce by means of inclusion; however, they also have
naturally selfish impulses that create conflict and friction among
neighbors. If people are to live together, they must curb some of
these selfish impulses and make certain other sacrifices. For ex-
ample, they may not want to wait their turn, respect the property of
others, pay taxes, or let their children’s lives be put at risk in war,
but such sacrifices are often demanded by cultures. People are
often willing to make such sacrifices, however, because these risks
and costs are generally offset by the substantial rewards that come
from belonging to the group. Living alone in the forest is not, for
humans at least, a promising strategy for survival or reproduction.
Hence, people employ their capacity for self-regulation to override
certain selfish impulses and behave in ways that can secure and
maintain acceptance in the social group.
This bargain can break down on either side, however. People
who fail to control themselves properly are often rejected by others
and by society. Poor self-regulators are subject to abandonment by
peers, divorce, and even imprisonment. For example, people who
are poor at self-regulation are less successful in accommodating to
their relationship partners (Finkel & Campbell, 2001), and children
with poor self-control are less accepted and less popular with peers
(Maszk, Eisenberg, & Guthrie, 1999). Likewise, deficient self-
regulation has been implicated as a central cause of criminality
(Gottfredson & Hirschi, 1990; Longshore, 1998; McGuire &
Broomfield, 1994), and most modern societies forcibly exclude
criminals to the point of imprisoning them. Thus, if an individual
breaks the bargain by failing to self-regulate, society may break its
promise of social inclusion by rejecting that individual.
Conversely, rejection may signify that society has broken its
promise, and so the person may respond by abandoning self-
regulation. It is one thing to make altruistic sacrifices if they are to
be compensated by the multifaceted rewards of belonging to a
social group; it is quite something else to make those sacrifices if
one is not going to be thus compensated. Rejection may therefore
undermine the implicit social bargain on which individual self-
regulation is based. This could lead to self-regulation failure in
either of two ways. The impact of rejection could (at least tempo-
rarily) make the person disgruntled and therefore unwilling to
make the effort and sacrifices required for self-regulation. Alter-
natively, the rejection might directly cause the self-regulation
system to stop working, independent of any conscious or deliberate
reaction by the individual. The difference is essentially one of
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
whether the person becomes unable or merely unwilling to self-
regulate, and two experiments in the present investigation (Exper-
iments 5 and 6) were directly concerned with this question.
Might Rejection Facilitate Self-Regulation?
Although the present investigation was guided by the prediction
of self-regulation failure, on a priori grounds one might have made
the opposite prediction: Rejection might ideally serve as a stimulus
to improvement in self-regulation. The basis for this prediction
would be the assumption that people have a basic and strong need
to belong (see Baumeister & Leary, 1995). To the extent that this
is an overriding and unswerving goal, the main reaction to any
experience of social exclusion or rejection would be to redouble
one’s efforts to secure inclusion. Insofar as self-regulation might
improve one’s chances for eventual social acceptance, it would be
adaptive to respond to rejection with increased efforts at
Rejection might increase self-regulation especially in circum-
stances in which some aspect of the self contributed to the rejec-
tion. The procedures used in the present study and in our previous
investigations seemingly locate the basis for rejection in something
about the self. That is, in one procedure other group members
reject the individual after meeting and talking with him or her, and
in the other the exclusion forecast is based on an ostensible
inventory of the person’s personality. In contrast, if exclusion were
based on a random procedure or on something external to the
individual, self-regulation would seem useless.
Self-regulation is by definition a means of changing something
about the self, most commonly to remain in line with external
(social) standards. If something about the self has elicited rejection
or exclusion, then the optimal response would be to change that
something, which would involve self-regulation. Self-regulation
would in this case be central to an effort to alter the self so as to
make it more appealing to future potential partners. For these
reasons, it might well be predicted that manipulations of social
exclusion or rejection would stimulate and facilitate the capacity
for self-regulation in the service of making oneself more accept-
able to others and thereby preventing further rejection (and, pos-
sibly, of eventually reversing the most recent rejection).
To be sure, past findings have offered little encouragement to
the view that people respond to rejection in adaptive, self-
improving ways. Adaptive responses to rejection ought ostensibly
to include increases in prosocial behavior, decreased self-defeating
behavior, improved intellectual functioning, and decreased aggres-
siveness, but past work has yielded the opposite findings (for
example, Baumeister et al., 2002; Buckley et al., 2004; Kirk-
patrick, Waugh, Valencia, & Webster, 2002; Twenge et al., 2001,
2002; Warburton, Williams, & Cairns, 2003). It was mainly on the
basis of these past findings, rather than any a priori theoretical
grounds, that we began this series of studies looking for self-
regulation failure rather than facilitation.
The Present Research
In short, it is plausible that rejection may either impair or
improve self-regulation. The present studies tested the hypothesis
that social exclusion (rejection from social groups or relationships)
would cause a change in effective self-regulation and sought to
illuminate how this effect might come about.
To provide converging evidence and rule out alternative expla-
nations, we used different procedures across the first four studies.
There were two manipulations of social exclusion (not being
chosen by anyone in a group and getting bogus feedback that one
will end up alone in life). Self-regulation was measured with four
different procedures (drinking a healthy but bad-tasting beverage,
impulsively eating cookies, persistence in the face of failure, and
dichotic listening). Experiments 5 and 6 were follow-ups to Ex-
periment 4, and thus they involved quite similar procedures. The
goal of those final two studies was to ascertain whether manipu-
lations of self-serving incentives (Experiment 5) and self-
awareness (Experiment 6) would counteract the effect found in
We also measured emotion or mood in most of these studies. As
noted, our previous research has not generally shown that mood or
emotional reactions mediate the behavioral effects of rejection (for
example, Twenge et al., 2001). However, emotional distress has
been found in other work to lead to impaired self-regulation (for
example, Grilo, Shiffman, & Wing, 1989; Keinan, 1987; Rosenthal
& Marx, 1981; Sayette, 1993; Tice, Bratslavsky, & Baumeister,
2001; Wegener & Petty, 1994). Given the theoretical possibility
that social exclusion could produce emotional distress or bad
moods, we believed that it was important to check whether such
reactions might contribute to the hypothesized impairment of
Experiment 1 manipulated social exclusion by randomly assign-
ing people to receive bogus feedback about the future trajectory of
their social lives. Those in the crucial (future alone) condition were
told, ostensibly on the basis of their responses to a questionnaire,
that they would likely end up alone in life. Two control conditions
were used. Participants in the future belonging condition were told
that they would most likely spend the rest of their lives surrounded
by people who cared about them. In the misfortune control con-
dition, participants were told that their futures would probably be
marred by a tendency to be accident prone, which would entail
various injuries and other problems. The purpose of this design
was to permit two types of comparison, one between anticipated
social exclusion and anticipated inclusion and the other between
two anticipated unpleasant futures, only one of which involved
social exclusion. The latter comparison should help counteract the
possible alternative explanation that any differences obtained be-
tween the inclusion and exclusion conditions reflected simply the
anticipation of a pleasant versus unpleasant future.
The procedure used to measure self-regulatory failure was bor-
rowed from previous work (Muraven, 1998). Participants were
encouraged to drink a bad-tasting beverage. Because the drink
tastes bad, self-regulation is required to force oneself to consume
it. It is also noteworthy that the bad taste was created by vinegar,
which is actually health enhancing. The need to use self-regulation
to do something that is healthy but unpleasant thus resembles a
great many exercises of self-regulation, such as forcing oneself to
take one’s medicine, to perform physical exercise when disin-
clined, to drag oneself out of bed despite wishing for more sleep,
or to work or study instead of having fun. To increase the incentive
to drink the bad-tasting beverage, the experimenter also offered a
modest financial inducement that was to be proportional to the
amount drunk. By indicating that the research design was set up to
reward people for drinking more, we hoped to provide a subtle
legitimization of the desirability of drinking. Providing money also
was intended to convey the impression that drinking might not be
inherently pleasant (hence the need for extrinsic incentives).
If social exclusion impairs self-regulation, then participants in
the future alone condition should be less successful than partici-
pants in other conditions on the self-regulation task. The prediction
would therefore be that people would consume less of the bad-
tasting beverage in the future alone condition than in either of the
other two conditions. In contrast, if social exclusion facilitates
self-regulation, then people in the future alone condition should
consume more of the bad-tasting beverage than people in the other
Participants. Participants were 36 undergraduates (24 male) taking
part in connection with their introductory psychology class. Sixty-seven
percent of the participants were White, and 33% were of racial minority
backgrounds. Their average age was 18.7 years. Four additional partici-
pants were dropped because they rated the drink’s taste as good or neutral,
thereby violating the premise on which the procedure was based. One other
participant expressed suspicion about the feedback and was dropped.
Measures and procedure. On entering the laboratory, participants first
completed a short demographic questionnaire and the Eysenck Personality
Questionnaire (EPQ; Eysenck & Eysenck, 1975). After finishing the de-
mographic questionnaire and EPQ, participants were given accurate feed-
back regarding their extraversion score. This helped to bolster the extent to
which participants perceived the results of the personality test as truly
indicative of their personality. Participants were then given false feedback
about the implications that their extraversion would have for their future
belongingness. Following a procedure developed by Twenge et al. (2001),
participants were randomly assigned to one of three social feedback con-
ditions: future belongingness, misfortune control, or future alone. Depend-
ing on their extraversion score, future belongingness participants were told:
“Scoring high or fairly high in extraversion means that you like people and
people like you. Being an extravert is a really good thing for relationships.
I’ll read the personality description to you” or “Being an introvert can be
a good thing for relationships. There’s been some research showing that
introverts have an easier time keeping relationships together. Instead of
running around meeting new people all the time, they’re good at keeping
the relationships they have. I’ll read the personality description to you.”
Regardless of their extraversion score, participants assigned to the mis-
fortune control condition were told: “What I really notice on your ques-
tionnaire is that you score high on a scale that is correlated with being
accident prone later in life. I’ll read the personality description to you.”
Finally, depending on their extraversion score, participants assigned to the
future alone condition were told: “Scoring high or fairly high in extraver-
sion is a good thing for meeting people, especially when you are in college
. . . but there’s been some research that has shown that people who score
high on extraversion have trouble keeping their relationships stable later in
life” or “Being an introvert is not really a good thing for relationships.
Once you get out of college, it is harder to meet people, so it is easier if you
score really high on extraversion. If you do not it makes it more difficult
to meet people. I’ll read the personality description to you.”
Participants were then read a personality description. Those assigned to
the future belongingness condition were told: “You are the type who has
rewarding relationships throughout life. You are likely to have a long and
stable marriage and have friendships that will last into your later years. The
odds are that you’ll always have friends and people who care about you.”
In contrast, misfortune control participants were told: “You are likely to be
accident prone later in life—you might break an arm or a leg a few times,
or maybe be injured in car accidents. Even if you haven’t been accident
prone before, these things will show up later in life, and the odds are you
will have a lot of accidents.”
Finally, future alone participants were told: “You are the type who will
end up alone later in life. You may have friends and relationships now, but
by mid-20s most of these will have drifted away. You may even marry or
have several marriages, but these are likely to be short-lived and not
continue into your 30s. Relationships do not last, and when you are past the
age where people are constantly forming new relationships, the odds are
you’ll end up being alone more and more.” Participants subsequently rated
their mood on a scale ranging from 1 (very negative)to7(very positive).
Participants were then seated at a table with 20 small paper cups that
each contained 1 oz (29.6 ml) of vinegar drink (made with drink mix, 1 cup
of sugar, 4 cups of water, and 2 cups of vinegar). They were told “This is
a drink that does not taste good to most people. However, it is not harmful.
In fact, it is good for you. I will give you a nickel for every ounce you
drink. How much you drink is up to you.” The number of ounces drunk was
the primary measure of self-regulation.
After the participant drank, he or she was asked to fill out a final
questionnaire including rating the drink’s taste. Then followed a probe for
suspicion, starting with a general question about how the participant
understood the study and proceeding to more specific questions as to
whether the participant suspected any deception. The goal was to ascertain
whether the participant spontaneously expressed doubt or suspicion about
the veridicality of the feedback (as one did). A full debriefing followed,
whereafter participants were paid, thanked, and dismissed.
Validation of measure. The drink was meant to be distasteful; after
consuming it, participants were asked to rate the drink’s taste on a scale
ranging from 1 (tastes not so bad)to10(tastes really bad). The average
taste rating was 6.86 (SD ⫽ 1.40), with 67% of participants rating the drink
a 7 or higher. Thus, the drink was regarded as unpleasant by our sample.
Results and Discussion
The main question in Experiment 1 was whether self-regulation
would be impaired or facilitated among people whose belonging-
ness needs were thwarted. Specifically, participants in the crucial
condition were told they would probably end up alone in life, and
we measured self-regulation by asking these people to make them-
selves drink a substantial amount of a bitter vinegar drink. A
one-way analysis of variance (ANOVA) revealed significant vari-
ation among the three conditions as to how much people drank,
F(2, 33) ⫽ 3.33, p ⬍ .05. Planned comparisons using the ANOVA
mean square error confirmed the differences. The people in the
social exclusion (future alone) condition consumed significantly
fewer ounces of the drink than the people in the anticipated
acceptance condition, t(33) ⫽ 2.23, p ⬍ .03, d ⫽ 0.78. People who
anticipated being alone in life also consumed less than the people
who were told they would be accident prone, t(33) ⫽ 2.15, p ⬍
.04, d ⫽ 0.75. The accident prone (misfortune control) and future
belonging groups did not differ, t ⬍ 1, ns. Means are presented in
These results provide important confirmation that some loss of
self-control is one of the negative effects of social exclusion.
Indeed, the difference between the exclusion and inclusion
groups suggested a large effect size (d ⫽ 0.78; see Cohen, 1977).
Learning that they were likely to be alone in life apparently had a
strong and substantial impact on people’s ability to make them-
selves perform an aversive, if healthful, behavior. These results
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
contradict the rival prediction that social exclusion would facilitate
We also investigated the role of mood and emotional distress.
We found marginal trends toward differences in mood among the
three conditions, F(2, 33) ⫽ 2.79, p ⬍ .08. According to pairwise
comparisons (which, admittedly, lose some legitimacy insofar as
the overall ANOVA failed to reach significance), the future alone
and future belonging conditions differed significantly, t(33) ⫽
2.02, p ⬍ .05, as did the future belonging and misfortune control
conditions, t(33) ⫽ 2.11, p ⬍ .05. The future alone and misfortune
control groups did not differ in their mood ratings, F ⬍ 1, ns. Thus,
a weak pattern of differences suggested that people did feel worse
after hearing that they would be alone in life than after hearing
they would have good interpersonal relationships, but this slightly
negative mood following the social exclusion manipulation did not
differ from the mood of people who heard they would be accident
Given that our manipulation seemed to have produced at least
some effect on mood, we conducted the test for mediation by
mood, using the future alone and future belonging conditions and
following the procedures outlined by Baron and Kenny (1986).
The effect of the experimental manipulation of belongingness on
self-regulation (drinking) remained significant even after control-
ling for mood, r(21) ⫽ .46, p ⬍ .03. Furthermore, there was no
effect of mood on the dependent variable once we controlled for
experimental condition, r(21) ⫽⫺.10, ns. Either of these findings
alone would be sufficient to reject a mood mediation theory,
according to Baron and Kenny (1986), and together they speak
convincingly against mood mediation.
Another telling argument against the possible role of emotion
was revealed in the findings from the misfortune control condition,
in which mood and self-regulatory performance differed. That is,
participants in that condition reported feeling like those in the
future alone condition, but they acted like participants in the future
belonging condition. If mood was responsible for the impairment
of self-regulation, then the relatively bad moods of participants in
the misfortune control condition should have impaired their self-
regulatory performance too, but they did not.
Thus, Experiment 1 confirmed the prediction that social exclu-
sion would cause a drop in self-regulation. This drop was not
mediated by mood.
The main purpose of Experiment 2 was to provide converging
evidence using a different manipulation of social exclusion and a
different measure of self-regulation. Experiment 1 relied on the
bogus feedback procedure that manipulated social exclusion by
telling people that they would eventually find themselves alone in
life. In Experiment 2, we relied on a much more immediate form
of social exclusion, namely rejection by peers. For this, we adapted
a procedure used by Leary, Tambor, Terdal, and Downs (1995; see
also Nezlek, Kowalski, Leary, Blevins, & Holgate, 1997; Twenge
et al., 2001). A group of people performed a get-acquainted task,
after which members were asked to rate which people they would
like to work with individually. By random assignment, half of the
people were told that no one had expressed an interest in working
with them, which constitutes a palpable and seemingly unanimous
social rejection. The rest were told that everyone had chosen them
as desirable partners.
The dependent variable involved eating snack foods (cookies).
In previous work, a majority of students at the university where
this study was conducted rated eating cookies and other fattening
snacks as an undesirable, unhealthy behavior that they would
prefer to avoid (for example, Tice et al., 2001). Consumption of
fattening foods is widely recognized as a growing, worldwide
problem (contributing to the so-called “globesity” epidemic). Cur-
tailing this consumption qualifies as self-regulation, because peo-
ple are attracted to the pleasant taste of cookies and similar foods
and must therefore overcome their desire to eat these snacks. Use
of cookie abstention as a measure of self-regulation was especially
appealing in the present context as a complement to the measure
used in Experiment 1. If the results converged, they would show
that rejection impairs self-regulation regardless of whether self-
regulation is aimed at promoting or inhibiting oral consumption.
The widespread publicity devoted to dieting and slimness as
well as to healthy eating has elevated public recognition of the
importance of good eating habits. Binge eating, meanwhile, con-
stitutes an important and rising pattern of pathological self-
defeating behavior (Heatherton & Polivy, 1992; O’Neil & Jarrell,
1992; Stunkard, 1993). The relatively mild eating binges by which
many people violate their diets and jeopardize their physical fit-
ness are not considered pathological, but they are often recognized
as reflecting similar causal processes and as being a problematic,
self-defeating behavior that the individuals themselves wish they
could avoid (see Brownell, 1991; Heatherton & Baumeister, 1991;
Heatherton, Polivy, & Herman, 1991). In addition, the common
stereotype that social rejection leads to overeating seemed to be a
potentially valuable instance of the pattern we sought to demon-
strate in this article.
Self-Regulation, Experiments 1–3
Experiment and measure
Alone/rejected Belong/accepted Misfortune Other
Experiment 1: ounces drunk 2.31 2.81 7.91 7.69 7.50 6.83
Experiment 2: cookies eaten 8.94 6.46 4.40 3.02
Experiment 3: persistence 21.35 8.35 27.67 8.24 28.50 3.42 29.22 1.83
Note. High self-regulation is denoted by high numbers in Experiment 1 and low numbers in Experiments 2 and
3. Persistence scores for Experiment 3 are in minutes.
Participants. Participants were 38 undergraduate students (24 male)
enrolled in introductory psychology, taking part to fulfill a course require-
ment. The sample was 74% White and 26% racial minority, and partici-
pants’ average age was 18.3 years. Three additional participants expressed
suspicion about the feedback and were dropped from the analyses.
Materials and procedure. Participants arrived at the laboratory in
single-sex groups of 4 – 6. They were given nametags, after which they
were given both written and oral instructions to (a) learn each other’s
names and then (b) talk for about 20 min using a set of questions as a guide
(the questions were taken from the relationship closeness induction task
developed by Sedikides, Campbell, Reeder & Elliot, 1999). After 20 min,
the experimenter led the participants to separate rooms, where they com-
pleted a demographic form and stated how long it had been since their last
meal. They then completed a page with the following instructions: “We are
interested in forming groups in which the members like and respect each
other. Below, please name the two people (out of those you met today) you
would most like to work with.” The experimenter collected these sheets,
telling the participants she would return with their group assignments.
Instead, participants were randomly assigned to be accepted or rejected
by the group. The accepted participants were told: “I have good news for
you— everyone chose you as someone they’d like to work with. But we
cannot have a group of five (or four, or six) people, so I’ll have you do the
next task alone.” The rejected participants, on the other hand, were told: “I
hate to tell you this, but no one chose you as someone they wanted to work
with. So I’ll have you do the next task alone.”
Participants then rated their mood on a scale ranging from 1 (very
negative)to7(very positive). A bowl containing 35 bite-sized chocolate
chip cookies was placed in front of them; the experimenter then asked them
to taste-test the cookies, eating “as much as you need to judge the taste,
smell, and texture.” They were also given a taste-test form asking them to
rate the taste of the cookies and the likelihood that they would buy a box.
Such procedures are commonly used in eating research because they
deflect attention from the quantity that the participant consumes and offer
a legitimate rationale (that is, trying to do a good job) for eating a large
amount. Participants were then left alone with the bowl of cookies and the
form for 10 min.
A careful debriefing followed, with special emphasis to make it clear
that the acceptance versus rejection feedback had been randomly assigned.
It was explained that none of the participants in fact worked together;
instead, all worked alone, having received one of the two random mes-
sages. When the participant was satisfied and expressed clear understand-
ing that he or she had not actually been rejected, the participant was
thanked and dismissed. After all participants left, the remaining cookies
Results and Discussion
Does being rejected by a group lead to overeating? Participants
in the rejection condition ate an average of almost nine cookies, as
compared with about half that amount in the accepted condition
(see Table 1). The difference was significant, F(1, 37) ⫽ 7.99, p ⬍
.01, and of substantial magnitude (d ⫽ 0.98). In other words, social
rejection produced a large increase in unhealthy eating.
This effect remained strong when other variables were added to
the ANOVA. Controlling for group session, the rejected people
still ate more than the accepted ones, F(1, 37) ⫽ 10.87, p ⬍ .01.
They ate more after controlling for how much time had elapsed
since their last meal, F(1, 37) ⫽ 7.59, p ⬍ .01. And they even ate
more when we controlled for how good they said the cookies
tasted, F(1, 37) ⫽ 5.77, p ⬍ .025. With all three variables included
in the analysis, the effect of rejection on eating was still signifi-
cant, F(1, 37) ⫽ 8.12, p ⬍ .01, and indeed it was almost identical
to the result of the initial, one-way ANOVA.
Thus, the main hypothesis was supported. Rejected people ate
significantly more cookies. The contrary prediction, that self-
regulation would be facilitated by social rejection, was directly
Several additional findings shed light on the basis of the main
finding. There was a trend suggesting that people in the rejection
condition rated the cookies as better tasting than people in the
accepted condition rated them, t(37) ⫽ 1.49, p ⬍ .15. The im-
provement in taste does not seem to have been a crucial mediator
of the increased eating, however. As already reported, the increase
in eating remained significant even after controlling for taste. In
fact, the correlation between rated taste of cookies and amount
eaten failed to reach significance among the rejected participants,
r(18) ⫽ .24, ns. In contrast, taste did predict eating among the
people in the accepted condition, r(20) ⫽ .42, p ⬍ .05. This
suggests that accepted participants ate in relative moderation as a
response to the good taste, whereas the rejected participants ate in
excess regardless of taste.
Mood was significantly different between the two experimental
conditions, t(37) ⫽ 2.50, p ⬍ .05. The people who had been
accepted rated their moods as being slightly more favorable (M ⫽
4.80 on a 7-point scale) than people in the rejected condition (M ⫽
4.06, which is almost precisely at the midpoint of the 7-point scale,
thus literally expressing no genuine distress). A mediation analysis
following the procedures recommended by Baron and Kenny
(1986) failed to show that mood mediated between rejection and
In fact, the correlation between mood and number of cookies
eaten was negative in the rejection condition, r(18) ⫽⫺.38, p ⬍
.13, but positive in the acceptance condition, r(20) ⫽ .25, p ⫽ .30.
More precisely, rejected participants ate more when they felt
worse, whereas accepted participants ate more when they felt better.
Neither correlation was significant, but the difference between the
two correlations was significant at p ⬍ .05.
A particular benefit of Experiment 2 is that it involved a differ-
ent procedure for manipulating social exclusion, namely rejection
by a group. This design does suffer, however, from the fact that
everyone is either accepted or rejected, and thus there is no pure
control group. As a result, it is impossible to know whether the
observed differences were driven by the rejection or by the accep-
tance condition. However, all of the other studies in this investi-
gation showed that the rejected group differs from the pure control
group, whereas the acceptance manipulations usually do not. More
generally, bad events have a stronger impact than good events,
across a wide and diverse range of psychological phenomena (for
a review, see Baumeister, Bratslavsky, Finkenauer, & Vohs,
2001). Hence, we think the effects are mainly due to the rejection
Another interesting aspect of this procedure is that technically
everyone experiences some degree of social exclusion, because
everyone ends up having to work alone. Still, apparently people
are quite willing to work alone cheerfully if they have just heard
that everyone wanted to work with them. Working alone because
everyone rejected you is apparently something quite different.
Thus, apparently the social messages of acceptance and rejection
were sufficient to produce very different outcomes, even though
technically both led to working alone on the next task.
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
Our first two studies measured self-regulation in terms of con-
sumption. That is, effective self-regulation consisted of making
oneself either consume something or refrain from consuming
something, in both cases contrary to one’s normal and natural
inclinations. The main purpose of Experiment 3 was to measure
self-regulation in another sphere. For this, we used persistence on
unsolvable puzzles. This has been used as a measure of self-
regulation in many previous studies (for example, Muraven, Tice,
& Baumeister, 1998). Unsolvable puzzles are frustrating and dis-
couraging, and people may often be inclined to quit trying. How-
ever, persistence in the face of failure is admired in our culture and
is sometimes rewarded with success, and so effective self-
regulators may override the impulse to quit and instead force
themselves to keep trying for a longer period of time.
The manipulation of social exclusion was the same bogus feed-
back manipulation used in Experiment 1. The main prediction was
therefore that socially excluded participants, who were told that
they would likely end up alone in life, would quit faster on the
frustrating puzzles than would participants in the control
As noted, we have entertained the opposite prediction, namely
that social exclusion would cause improvements in self-regulation.
Given that the results of the first two studies were in the opposite
direction, we do not discuss this hypothesis in connection with
each subsequent experiment.
An additional feature of the design of this study was that it
provided a no-feedback control group. This ensures that obtained
differences are actually due to the messages of social exclusion
and rejection instead of to messages of acceptance (and misfor-
tune). It is somewhat plausible (though perhaps not parsimonious)
to suggest that differences in performance could arise because
feedback indicating future social acceptance energizes people to
perform well and because feedback about future injuries and
misfortunes also causes good performance out of a sense of ad-
venture. If so, these two conditions would differ from the no-
feedback control group, which would resemble the future alone
Another refinement in Experiment 3 involved our investigation
of potential mediators. Although mood failed to mediate the self-
regulation impairments in Experiments 1 and 2, this might con-
ceivably have been due to our reliance on a single-item measure of
mood. In this study, we used a reputable and well-validated mea-
sure of mood (Mayer & Gaschke, 1988). We also measured state
self-esteem as a possible mediator. It seemed possible, a priori, that
social rejection could cause reductions in self-esteem. For exam-
ple, the sociometer model depicts self-esteem as an internal mea-
sure of eligibility for belongingness (see Leary & Baumeister,
2000; Leary et al., 1995). A drop in self-esteem might conceivably
lead to a reduced sense of self-efficacy or in other ways an
impairment of self-regulatory performance. The sociometer model
could even be compatible with our central line of reasoning,
according to which rejection undermines the purpose of self-
regulation. That is, it might be suggested that low self-esteem
signifies the belief that one will ultimately fail to maintain social
acceptance, and so there is little reason to bother making the efforts
and sacrifices required for self-regulation.
A final refinement for Experiment 3 was the inclusion of ma-
nipulation checks. The manipulations of social exclusion and re-
jection used in this investigation have been used in previous
studies, but there has not been much attempt to establish that
participants actually receive and understand the message of rejec-
tion. In Experiment 3, participants were asked both to state the
content of the manipulation they had received and to indicate
whether they believed it.
Participants were 45 undergraduates (26 female) taking part for course
credit in introductory psychology. The sample was 71% White and 29%
racial minority. Five participants were excluded from all statistical analy-
ses. Two of these participants did not follow instructions, and the remain-
ing 3 knew that the puzzles were unsolvable.
Participants began by completing the EPQ (Eysenck & Eysenck, 1975).
They were then exposed to the same manipulation used in Experiment 1.
Therefore, they were assigned to one of four groups: future alone (they
would be alone later in life), future belonging (they would have good
relationships), misfortune control (they were prone to accidents), and
no-feedback control (no prediction given). Next, participants completed the
Brief Mood Introspection Scale (BMIS; Mayer & Gaschke, 1988) and the
State Self-Esteem Scale (Heatherton & Polivy, 1991). The BMIS includes
two subscales measuring mood valence and arousal.
The experimenter then introduced the task and explained that it was
closely linked to overall intelligence. This spatial task involved tracing
geometric figures, as used in former experimental designs (Baumeister,
Bratslavsky, Muraven, & Tice, 1998). The task required the participant to
fully trace a geometric figure without retracing any lines and without lifting
his or her pencil from the paper. Multiple slips of paper were provided for
each figure so that the person could try to trace the figure over and over.
The experimenter used one such figure to demonstrate how the task
worked. Each participant was then given a second figure to practice solving
in front of the experimenter. Any questions a participant had were an-
swered at that time. After this practice period, the experimenter presented
two additional figures for tracing. Participants were told to try their best to
complete both of the puzzles. In addition, participants were told that they
could take as much time and as many trials as they wanted. Participants
were informed that their performance on the task did not depend on the
number of trials or time taken for completion, but only on whether they
were able to trace the figures successfully. Although the two figures
presented during the explanation and trial periods were solvable, the final
two figures were not solvable. There was no way to successfully complete
the tracing of the geometric figures given.
The experimenter left the room and timed how long the participant
worked on the task before giving up (signified by the participant ringing a
bell). An a priori decision was made to stop any participants still working
after 30 min. (There was ample time left at the end of the session for the
participant to work the full 30 min and still not be feeling that it was time
to leave because the scheduled hour had elapsed.) After 30 min or on
hearing the bell, the experimenter reentered the room, collected a postex-
perimental questionnaire, and then debriefed, thanked, and dismissed
Results and Discussion
Participants told they would be alone later in life gave up much
faster on the frustrating puzzle-tracing task (see Table 1). An
ANOVA focusing on task persistence indicated significant varia-
tion among the four conditions, F(3, 44) ⫽ 6.13, p ⫽ .002. People
in the future belonging, misfortune, and no-feedback control
groups persisted longer on the unsolvable puzzle-tracing task than
people in the future alone group. A Tukey honestly significant
difference test revealed that the future alone group was signifi-
cantly different from the other groups at p ⬍ .03. No differences
were found among the remaining three groups. Comparing the
future alone group with the other three conditions yielded a large
estimated effect size (d) of 1.31.
Persistence times were more variable in the future alone condi-
tion than in the other three conditions, which violates the assump-
tion of homogeneity of variance. It is unlikely that the finding of
significance was seriously distorted or produced by this heteroge-
neity, however. In fact, the longest duration in the future alone
condition (26.95 min) was less than the mean persistence in any of
the other conditions. Moreover, various statistical works have
generally concluded that F and t tests are fairly robust with respect
to small violations of homogeneity of variance (Box, 1954; Glass
& Hopkins, 1996). Still, we ran a separate t test that did not assume
homogeneity of variance, and it confirmed that the future alone
group persisted less than the other three (combined) conditions,
t(10.8) ⫽ 2.78, p ⬍ .05.
Thus, people in the future alone condition were less likely to
persist on the frustrating task. They did worse than participants
who received positive feedback about acceptance, worse than
participants who received unpleasant feedback about future inju-
ries and misfortunes, and worse than participants who received no
feedback at all. The last result indicates that the obtained differ-
ences are probably not due to any positive effect of social accep-
tance feedback. Instead, the message of social rejection appears to
be the crucial factor.
A possible criticism of this procedure is that quitting early on
unsolvable problems is adaptive, and so quitting early might reflect
good rather than bad self-regulation. This has, however, been
addressed in previous work. Baumeister et al. (1998) replicated a
study on ego depletion using solvable rather than unsolvable
anagrams, and the same manipulation that caused shorter persis-
tence on unsolvable anagrams (as well as decrements in other
forms of self-regulation; see also Muraven et al., 1998) led to
poorer performance on solvable anagrams. In the present studies,
as in those previously mentioned, participants did not anticipate
that some problems would be unsolvable, and so there was no basis
for viewing early quitting as a product of good self-regulation.
Quitting in the face of failure can therefore reasonably be inter-
preted as an indication of poor or depleted self-regulation.
We then explored the possible role of mood. There were no
differences among the four groups in either mood valence or
arousal as measured by the BMIS mood measure. Mediation
analyses also disconfirmed any notion that mood mediated
The role of state self-esteem as a potential mediator was also
explored. There were no significant differences among conditions
in self-esteem as measured by the State Self-Esteem Scale; how-
ever, we still decided to run the mediation analyses. As noted
earlier, the simple bivariate correlation between time persistence
on the task and exclusion condition was r(43) ⫽⫺.55, p ⬍ .001.
If self-esteem mediated the path between social exclusion and task
persistence, the correlation between social exclusion and task
persistence should have lost significant power when controlling for
state self-esteem. Contrary to the state self-esteem mediation hy-
pothesis, the correlation between exclusion condition and task
persistence was still highly significant (and almost unchanged in
magnitude) after controlling for state self-esteem, r(42) ⫽⫺.54,
p ⬍ .001. In addition, state self-esteem was not correlated with
task persistence when controlling for exclusion condition, r(42) ⫽
⫺.10, ns. These findings indicate that state self-esteem did not
mediate the association between social exclusion and task
As a means of ensuring that participants were paying attention
to the false feedback, they were asked several questions on a
follow-up questionnaire. Participants were asked what their extra-
version score was, with the choices of high, medium, and low
extraversion. In addition, participants were asked the prediction for
their future based on that personality score. All of the members of
the future belonging and future alone groups responded correctly
to these two questions, and 91% of the members of the misfortune
group answered correctly. Because the members of the no-
feedback control group were not given any feedback about their
level of extraversion or future, they were not asked to respond to
these questions. Overall, results showed that participants were
paying attention to the feedback given by the experimenter and
retained what they were told during the experimental session.
These findings confirm that the manipulation was effective.
Three additional questions were asked on the follow-up ques-
tionnaire. Participants were asked how much they thought the
prediction might describe their future, how difficult the puzzle was
to do, and how easy it was for them to give up on the puzzle. There
were no significant differences on the ratings of the difficulty of
the puzzle or the ease of giving up on the puzzle, both Fs ⬍ 1, ns.
All participants rated the puzzle as quite difficult (the four cell
mean ratings of difficulty ranged between 5.80 and 6.50 out of a
maximum difficulty of 7).
The only significant difference on the follow-up questionnaire
involved the question of whether the prediction might describe the
participant’s future, F(2, 30) ⫽ 33.68, p ⬍ .001. Not surprisingly,
the future alone participants expressed greater skepticism than
participants in the other conditions about the accuracy of the
feedback they received. This is consistent with the general pattern
that people are relatively more skeptical and critical of feedback
that goes against what they want to believe (Lord, Ross, & Lepper,
We checked for whether differences in belief versus skepticism
might have mediated persistence on the puzzles. The correlation
between persistence and belief (vs. skepticism) in the feedback
failed to reach significance, r(31) ⫽ .29, ns. After controlling for
believability, the belongingness manipulation still significantly
predicted persistence, r(31) ⫽⫺.47, p ⬍ .01. Conversely, con-
trolling for experimental condition reduced the correlation be-
tween skepticism/belief and persistence to r ⫽⫺.18, ns. Thus, the
main finding that rejection impaired self-regulation was indepen-
dent of the rating of belief in the feedback.
In summary, Experiment 3 provided further evidence that self-
regulation is impaired by social exclusion. This effect was not
mediated by arousal, mood, belief in feedback, or state
Experiment 4 sought to provide converging evidence from yet
another domain of self-regulation, namely attention control. Ex-
periments 1 and 2 examined self-regulation in the context of
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
consumption behaviors, and Experiment 3 examined it in the
sphere of task persistence. Cognitive control is another important
sphere of self-regulation. Indeed, by some accounts (for example,
Baumeister, Heatherton, & Tice, 1994; Carver & Scheier, 1981),
controlling attention is a centrally important process that has
powerful links to self-control success versus failure in many other
spheres. For example, people seeking to regulate their sexual
impulses may avoid exposure to sexual stimuli; dieters may seek to
avoid cues that remind them of tempting, fattening foods; and
affect regulation often involves attempting to avoid stimuli (for
example, funny thoughts at a funeral) that may elicit inappropriate
In this experiment, a dichotic listening procedure was used to
measure self-regulation. Information was presented simulta-
neously to both ears, and the participant’s task required ignoring
the material spoken in one ear so as to be able to screen the list of
words presented to the other ear. We predicted that socially ex-
cluded participants would be less successful at this task.
Thirty right-handed undergraduates (24 women) participated in this
study in exchange for partial course credit.
One additional participant was
excluded as a result of expressed suspicion, and 2 participants were
excluded as a result of an inability to understand the dichotic listening
instructions. The sample was 61% White and 39% racial minority. Partic-
ipants’ average age was 18.4 years.
On entering the laboratory, participants were told that the purpose of the
study was to understand better how personality relates to verbal and
nonverbal performance. Participants first completed a short demographic
questionnaire and the EPQ (Eysenck & Eysenck, 1975). After finishing the
demographic questionnaire and the EPQ, participants were given accurate
feedback regarding their extraversion score, followed by the bogus feed-
back manipulation as in Experiment 1. Participants were randomly as-
signed to the future alone, future belonging, and misfortune control
After receiving their personality description, participants completed the
BMIS (Mayer & Gaschke, 1988). Participants were then told that the next
portion of the study involved them completing a brief listening game.
Participants were asked to sit at a separate desk in the same laboratory
room on which a tape recorder and set of headphones were located. The
experimenter gave the participant a recording sheet and explained (in part),
In the right ear, there will be a speech regarding a policy issue. In the
left ear, you will hear a female voice speaking a series of words. Your
job is to ignore the speech in your right ear and pay attention to what
is going on in the left ear [and] write down each word spoken in the
left ear that contains the letter “m” or the letter “p.” It is important that
you pay close attention to the left ear and try your best to ignore the
speech in your right ear.
In an initial version of the study (n ⫽ 15), participants were instructed
to write down words heard in the left ear that began with the letters m and
p instead of all words that contained the letter m or p. This was apparently
too easy, because the majority of participants in all social feedback con-
ditions completed the task at or near perfection when only identifying
words that began with each respective letter. The task instructions were
therefore revised to the current form to enable meaningful comparisons of
performance across conditions. However, the high success rate on the
initial, easy version of the task speaks against any alternative explanation
that participant performance could be explained by rebelling against the
experimenter or psychologically withdrawing from the experiment itself
(such as in the future alone condition).
The contents of the female voice recording were the 1,000 most fre-
quently used words in the English language, of which the first 255 were
included in the final recording. Of these 255 words, 38 contained the letter
m and 10 contained the letter p. The recording sheets were then scored
against a key to determine the number of words containing the letter m or
p the participant had correctly identified. After completing the dichotic
listening task, participants completed the Edinburgh Handedness Inventory
(Oldfeld, 1971), responded to a brief questionnaire designed to probe for
suspicion, and were fully debriefed.
Results and Discussion
Dichotic listening task. Social exclusion produced significant
decrements in self-regulated attention. An ANOVA indicated that
there was significant variation among the experimental conditions,
F(2, 27) ⫽ 5.70, p ⬍ .01 (see Table 2). Planned comparisons
revealed that future alone participants performed significantly
worse than future belongingness participants, t(27) ⫽ 2.56, p ⬍
.01, d ⫽ 1.23, and also significantly worse than misfortune control
participants, t(27) ⫽ 2.67, p ⬍ .01, d ⫽ 1.17.
Right-handed individuals are often recruited for dichotic listening
studies as a result of their relative uniformity in left-sided language
representation (Geffen & Caudrey, 1981), which is regularly associated
with right-ear dominance. This right-ear advantage among right-handed
individuals represents a robust effect that has been replicated in numerous
studies involving the use of varying procedures (Bryden, 1988). Because
the present study examined self-regulated attention after social exclusion,
it was important to be sure that all participants possessed identical aural
preferences in terms of their dominant and nondominant ears.
Performance on Dichotic Listening Task, Experiments 4 – 6
Experiment and measure
Experiment 4: basic 35.60 6.94 42.70 4.40 42.00 3.53
Experiment 5: basic (no cash) 39.30 3.71 43.50 2.00 40.00 3.94
Experiment 6: basic (no mirror) 36.50 3.40 41.60 1.81 43.17 1.50
Experiment 5: cash incentive 43.13 2.36 41.38 3.54 43.38 2.00
Experiment 6: mirror 42.60 2.15 40.60 3.50 40.50 6.14
Note. Data indicate mean tallies of stimulus words correctly identified (out of 48).
Although a diagnostic prediction of future social exclusion
produced significant decrements in self-regulated attention, partic-
ipants who received a negatively valenced future diagnostic fore-
cast (for example, that of being accident prone later in life)
remained essentially unaffected in their ability to regulate their
attention. In fact, the performance of misfortune control partici-
pants was statistically indistinguishable from that of future belong-
ingness participants, t(20) ⫽ 0.30, ns. Thus, the observed decre-
ments in self-regulation were unique to social exclusion and not
merely due to one receiving a negatively valenced diagnostic
prediction about one’s future.
Mood and emotion. To examine the alternative possibility that
the observed decrements in self-regulated attention were due to
emotional distress, we conducted two one-way ANOVAs using the
valence and arousal subscales of the BMIS (Mayer & Gaschke,
1988) as dependent measures. According to the results of these
analyses, there was no significant variation among the three social
feedback conditions in terms of self-reported arousal (for example,
arousal– calm), F(2, 27) ⬍ 1, ns, and there was marginal evidence
of variation in mood valence (for example, pleasant– unpleasant),
F(2, 27) ⫽ 2.70, p ⫽ .08. With regard to mood valence, planned
comparisons did reveal that future alone participants reported a
less positive mood (M ⫽ 12.40, SD ⫽ 8.98) than future belonging
participants (M ⫽ 21.00, SD ⫽ 6.80), t(27) ⫽ 2.15, p ⬍ .05. They
also showed a trend toward reporting a slightly less positive mood
than misfortune control participants (M ⫽ 14.80, SD ⫽ 8.90),
t(27) ⫽ 1.56, p ⫽ .11.
Multiple mediational analyses were conducted to determine
whether mood valence or arousal mediated between the social
exclusion manipulation and performance on the self-regulation
(dichotic listening) task. All analyses failed (in multiple ways) to
satisfy the requirements for mediation. In short, neither mood
valence nor mood arousal mediated the effect of manipulated
social exclusion on dichotic listening.
The first four experiments provided converging evidence that
social exclusion produces reductions in self-regulation. In the next
two, we turned our attention to the question of why it has that
effect. To be sure, we had initially focused on examining self-
regulation as a possible explanation of why some of the behavioral
effects of social exclusion occur, but it is always possible to seek
further explanation of an explanation. In particular, it seemed
essential to ascertain whether rejected people become unable, or
merely unwilling, to self-regulate.
We proposed that self-regulation, which essentially involves the
capacity to stifle one’s own self-serving impulses so as to engage
in socially desirable behaviors, serves the purpose of maintaining
membership in social groups. Therefore, a rejection experience
undermines the raison d’eˆtre of self-regulation, which in turn
produces self-regulation failure. Most plausibly, this reaction
would arise because the rejected person sees no need to restrain
selfish or impulsive behavior, given that the rewards of such
restraint (in the form of social acceptance) will not be forthcoming.
A variation on this would be that some aspects of self-regulation
become distasteful in the wake of rejection. In particular, self-
regulation depends on self-awareness (Carver & Scheier, 1981,
1982), and self-awareness might well be unpleasant after rejection.
In either case, this line of reasoning suggests that rejected individ-
uals could control themselves if they wanted, but they do not want
to do so.
A plausible alternative, however, would be that rejection makes
people incapable of self-regulation. As noted in the introduction,
most of our work has failed to find emotional distress as an
immediate response to rejection, even though being rejected would
seemingly be an important cause of distress and anxiety (for
example, Baumeister & Leary, 1995; Baumeister & Tice, 1990).
One possible reason for the apparent lack of emotion is that people
seek to stifle their emotional responses, either simply to avoid
feeling bad or to avoid embarrassing themselves by displaying
distress in front of the other people in the laboratory. Stifling
emotional distress is itself a form of self-regulation; insofar as all
acts of self-regulation draw on a limited resource (for example,
Muraven & Baumeister, 2000), the effort to stifle emotion could
deplete the resource, thereby leaving the person less capable of
further acts of self-regulation. Put another way, rejection may
cause people to use their limited capacity for self-regulation to
regulate their emotions, leaving less capacity for other acts of
The hypothesis that rejection makes people incapable of self-
regulation could conceivably arise in another manner. We have
proposed that the purpose of self-regulation is to make one behave
in a socially acceptable manner, and thus rejection undermines the
basic reason for self-regulating. It is possible that this connection
is hard-wired, and so an experience of rejection renders the person
incapable of self-regulating.
The procedure for Experiment 5 was based on that of Experi-
ment 4. We manipulated social exclusion using the bogus feedback
manipulation and measured subsequent self-regulation with the
same dichotic listening task. The main change in Experiment 5 was
to provide some of the participants with a motivating reason to
self-regulate, namely a cash incentive for accurate performance on
the dichotic listening task. If rejected people are incapable of
self-regulating, then the offer of money for effective self-
regulation should make no difference. In contrast, if they are
merely unwilling to self-regulate for the sake of fitting with others,
then offering them a self-serving incentive to perform well could
be enough to make them self-regulate effectively.
Participants. Fifty-one right-handed undergraduates (41 women) par-
ticipated in this study in exchange for partial course credit. Two additional
participants were excluded from the analyses, one who expressed suspicion
about the feedback and one who exhibited an inability to understand the
directions for the dichotic listening task. The sample was 82% White and
18% ethnic minority. Participants’ average age was 19.0 years.
Materials and procedure. Participants entered the laboratory and com-
pleted the EPQ (Eysenck & Eysenck, 1975). After completing the EPQ,
participants were exposed to the same social feedback manipulation used in
Experiment 1. After receiving their future diagnostic forecast of exclusion,
belongingness, or proneness to accidents, participants completed the BMIS
When participants had finished reporting their mood, the experimenter
instructed them to sit at a desk at the opposite end of the laboratory room.
The desk contained a tape recorder and a set of headphones that partici-
pants were told would allow them to complete a brief listening game.
Participants then received the same instructions and materials for the
dichotic listening task that were used in Experiment 4.
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
Participants were also randomly assigned to either a pay or a no-pay
condition. Participants assigned to the no-pay condition received the stan-
dard dichotic listening instructions used in Experiment 4. Participants
assigned to the pay condition, however, received the standard dichotic
listening instructions and the following additional instructions from the
We are also offering a monetary reward for performance on the
listening game. If you correctly identify 62% of the words in the left
ear that contain the letter “m” or “p,” we will pay you $5. If you
correctly identify 82% of the words in the left ear that contain the
letter “m” or “p,” we will pay you $10. Finally, if you correctly
identify 100% of the words spoken in the left ear that contain the letter
“m” or “p,” we will pay you $20.
To boost the credibility of the payment, the experimenter showed each
participant assigned to the pay condition an envelope filled with the money
that would be used to pay the participant for his or her successful perfor-
mance. The experimenter also put a $5 bill on the participant’s desk when
describing what would be required to earn the $5 reward, another $5 bill for
the $10 reward, and two additional $5 bills for the $20 reward. Participants
were not told the exact number of words they would have to identify to
receive each monetary reward sum, the reason being that they might lose
motivation and relax after reaching one goal or on realizing they would not
reach the next goal.
After completing the dichotic listening task, participants completed the
Edinburgh Handedness Inventory and a brief suspicion probe. Then they
were fully debriefed.
Results and Discussion
Dichotic listening performance. The main question of Exper-
iment 5 was whether offering a financial incentive for good per-
formance on the self-regulation task would offset the negative
effects of exclusion. Consistent with an answer of yes, a 2 ⫻ 3
ANOVA yielded a significant interaction between payment con-
dition and the exclusion manipulation, F(2, 45) ⫽ 4.72, p ⬍ .01.
The results are summarized in Table 2. Planned comparisons
confirmed the benefits of pay in overcoming the effects of social
exclusion. Future alone participants who were motivated to earn a
monetary reward performed significantly better than nonpaid fu-
ture alone participants, t(16) ⫽ 2.41, p ⬍ .02. The cash incentive
also improved the performance of misfortune control participants,
t(15) ⫽ 1.99, p ⫽ .05, but it had no effect on the performance of
participants in the future belonging condition, t ⬍ 1, ns.
To clarify the interaction, we conducted further analyses. A
one-way ANOVA focusing on the three nonpaid conditions indi-
cated significant variation among these conditions, F(2, 24) ⫽
3.76, p ⬍ .04. Planned comparisons demonstrated that nonpaid
future alone participants regulated their attention significantly
worse than nonpaid future belonging participants, t(24) ⫽ 2.47,
p ⬍ .02, d ⫽ 1.41, but no differently from the nonpaid misfortune
control participants, t ⬍ 1, ns. These findings provide a replication
of what we found in Experiment 4.
A parallel one-way ANOVA was then conducted on the paid
conditions, and it revealed no significant variation among condi-
tions, F(2, 21) ⫽ 1.29, ns. Pairwise comparisons likewise revealed
no differences between conditions. As Table 2 shows, the future
alone participants who had the cash incentive performed quite
These results suggest that self-regulation deficits following re-
jection indicate an unwillingness, rather than an inability, to reg-
ulate the self. When rejected people were offered a self-serving
incentive to self-regulate, they self-regulated effectively. Appar-
ently the capacity to self-regulate remained intact, but the normal
motivation to self-regulate was diminished. Providing a cash in-
centive offset the negative effect of social exclusion.
Precisely how the cash incentive counteracted the impact of
rejection cannot be gleaned from these data. It is possible that it
had attentional (distraction), emotional (good news reducing neg-
ative affect or increasing positive affect), or motivational (a new
goal toward which to strive) effects. Distraction alone seems an
unlikely explanation, insofar as many previous studies have moved
participants’ attention on to a new task but still revealed decre-
ments in self-regulation (for example, Baumeister et al., 1998;
Muraven et al., 1998). It is possible that the unwillingness to
self-regulate is rooted in negative affect, which was attenuated by
the offer of money, though as we saw in the earlier studies negative
affect has not mediated the effects of our rejection manipulations
generally. The money almost certainly served as a motivational
goal that provided a subjectively compelling justification for in-
creased effort. In any case, the most important conclusion from
these data is that rejected people remain capable of self-regulation,
though they need something (for example, money) to induce them
to put forth the requisite effort.
Mood and emotion. A one-way ANOVA revealed significant
variation among the three social feedback groups in terms of
self-reported mood valence, F(2, 48) ⫽ 4.32, p ⬍ .02. Planned
comparisons revealed that participants in the future alone condi-
tion rated their mood as significantly less pleasant (that is, low in
mood valence; M ⫽ 13.28, SD ⫽ 12.00) than future belonging
participants (M ⫽ 24.80, SD ⫽ 12.70), t(48) ⫽ 3.00, p ⬍ .004, d ⫽
0.93. Future alone participants also had a marginally more nega-
tive mood than the misfortune control participants (M ⫽ 19.80,
SD ⫽ 8.53), t(48) ⫽ 1.61, p ⫽ .09, d ⫽ 0.63. A one-way ANOVA
focusing on the arousal subscale yielded no significant variation
among conditions, F ⬍ 1, ns. Mediation analyses showed that
neither valence nor arousal mediated the link between social
exclusion and dichotic listening performance.
The results of Experiment 5 pointed toward the conclusion that
rejection makes people unwilling, rather than unable, to regulate
themselves effectively. Experiment 6 sought to replicate this con-
clusion using a different procedure and, assuming replication, to
shed possible light on the causal process by which rejection
Whereas Experiment 5 counteracted the effects of rejection by
offering a new motivation for effective self-regulation, Experiment
6 sought to counteract those effects by seating some participants in
front of a mirror during the dichotic listening task. Mirrors have
long been used in laboratory research to direct participants’ atten-
tion to themselves (for example, Carver & Scheier, 1981; Duval &
Wicklund, 1972). Thus, Experiment 6 investigated whether self-
focused attention might improve self-regulation among rejected
Self-focused attention has long been linked to self-regulation.
Carver and Scheier (1981, 1982) proposed that the purpose of
self-awareness is to serve self-regulation, insofar as self-awareness
involves a comparison of the self’s current state against relevant
standards such as social acceptability and personal ideals. In lab-
oratory work, increasing self-awareness appears to improve self-
regulation (for example, Carver & Scheier, 1981, 1982; Sentyrz &
Bushman, 1998). Self-awareness is also a state that involves think-
ing about how recent and current events reflect on the self (Hull &
Levy, 1979). When events reflect badly on the self, people may
seek to avoid or escape self-awareness (for example, Baumeister,
1990, 1991; Greenberg & Musham, 1981; Heatherton & Baumeis-
Being rejected or excluded is one event that should make
self-awareness aversive. People may enjoy thinking about what a
success experience means about the self, because those implica-
tions would be flattering to the self. In contrast, a failure or
rejection experience more likely signifies a possible fault, inade-
quacy, or misdeed, any of which would reflect badly on the self. At
least one previous finding suggested that an experience of social
exclusion causes people to want to avoid self-awareness, as re-
flected in an unusually strong avoidance of seats facing mirrors
(Twenge, Catanese, & Baumeister, 2003).
Thus, rejection may make people avoid self-awareness because
it is aversive to think about what possible faults in themselves may
have caused the rejection. Insofar as self-regulation depends on
self-awareness, however, a side effect of avoiding self-awareness
would be an impairment in self-regulation. If that is correct, then
stimulating self-awareness by placing participants in front of a
mirror should counteract the impact of rejection.
In contrast, if social exclusion renders people incapable of
self-regulation, then placing them in front of a mirror should have
little or no effect. If anything, it could cause a further impairment
on the dichotic listening task, because the sight of one’s own
reflection in the mirror should draw attention away from the
listening task, leaving less attention available for tracking the
Participants. Forty-five right-handed undergraduates (37 female) par-
ticipated in this study in exchange for partial course credit. Nine additional
participants were excluded from the analyses because they expressed
suspicion about the feedback, 1 participant was excluded as a result of an
inability to understand the instructions for the dichotic listening task, and
1 participant was excluded as a result of previous participation in a study
involving the same rejection manipulation. The sample was 60% White and
40% racial minority. Participants’ average age was 18.6 years.
Measures and procedure. Participants entered the laboratory individ-
ually to take part in a study ostensibly concerned with the relationship
between personality and verbal and nonverbal performance. After provid-
ing informed consent, participants completed a brief demographic ques-
tionnaire and the EPQ. By random assignment, they received the future
alone, future belonging, or misfortune control feedback, as in the preceding
studies. After this, they completed the BMIS mood measure.
When participants had completed the BMIS, they were told that the next
part of the study required that they sit at a desk at the opposite end of the
room for a listening game. Participants were randomly assigned to either a
high self-awareness or low self-awareness condition. Participants assigned
to the high self-awareness condition sat at a desk that faced a 12-in. ⫻
36-in. (30.5-cm ⫻ 91.5-cm) mirror. Low self-awareness participants, on
the other hand, sat at the same desk, but the mirror was turned around and
had a sign attached to the back that read “Being used for another study.
Please do not move.” The desk contained a tape recorder and a set of
headphones. Participants were given a recording sheet and instructions for
the listening game. The instructions and materials for the dichotic listening
task were the same as in Experiment 4. After participants had completed
the listening game, they completed the EHI and a suspicion probe and were
then thoroughly debriefed.
Results and Discussion
Dichotic listening performance. The main question of Exper-
iment 6 was whether increasing self-awareness by placing people
in front of a mirror would counteract the negative effect of social
exclusion on self-regulation. To test this, we conducted a 2 ⫻ 3
ANOVA on performance on the dichotic listening task. This
analysis revealed a significant interaction between mirror and
exclusion feedback, F(2, 39) ⫽ 5.61, p ⬍ .007. The results are
shown in Table 2.
Planned comparisons confirmed that increasing self-awareness
offset the impact of social exclusion. Among participants who
received the future alone feedback, those in the mirror condition
outperformed those in the control (low self-aware) condition sig-
nificantly, t(39) ⫽ 2.79, p ⬍ .01. In contrast, the mirror did not
significantly improve the performance of either the misfortune
control or the future belonging participants (both ts ⬍ 1, ns).
The low self-awareness condition replicated the results of Ex-
periment 4 (and the no-pay condition of Experiment 5). A one-way
ANOVA focusing on dichotic listening performance scores indi-
cated significant variation among the three feedback conditions,
F(2, 16) ⫽ 13.33, p ⬍ .0001. Planned comparisons revealed that
future alone participants low in self-awareness identified signifi-
cantly fewer words than did future belongingness participants low
in self-awareness, t(16) ⫽ 3.23, p ⬍ .0001, d ⫽ 1.90. In addition,
future alone participants low in self-awareness performed signifi-
cantly worse than misfortune control participants low in self-
awareness, t(16) ⫽ 2.68, p ⫽ .001, d ⫽ 2.54. The effect sizes for
these observed differences in dichotic listening performance
among low self-awareness participants were considerable, exceed-
ing the standard criteria for large effects (Cohen, 1977).
In contrast, a one-way ANOVA focusing on the mirror condi-
tion failed to show any variation among conditions as a function of
social feedback, F(2, 23) ⫽ 0.59, p ⬍ .56. Also, pairwise com-
parisons revealed no significant differences.
Thus, putting people in front of a mirror effectively counteracted
the impact of the social exclusion manipulation. Experiments 4, 5,
and 6 all showed that rejected people performed worse than
participants in the other conditions on the dichotic listening task.
However, both Experiments 5 and 6 also reversed this pattern,
either by providing a cash incentive for good performance (Ex-
periment 5) or by stimulating self-awareness (Experiment 6). Ap-
parently people are still capable of self-regulation after rejection,
but they are disinclined to self-regulate effectively in response to
external demands. Experiment 6 suggests that the particular reason
socially excluded people do not want to self-regulate is that they
avoid self-awareness. Making them self-aware counteracts this
Mood and emotion. To investigate whether mood and emo-
tional response influenced self-regulation, we conducted two one-
way ANOVAs using each subscale of the BMIS as a dependent
measure. Results from these analyses indicated that there was no
significant variation among the social feedback groups in terms of
either self-reported mood arousal, F(3, 42) ⬍ 1, ns, or mood
valence, F(3, 42) ⬍ 1, ns. Not surprisingly, mediation analyses
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
revealed no evidence that mood mediated the effect of the manip-
ulations on self-regulation.
Most societies and social groups occasionally reject individuals,
and one common cause of rejection is failure at self-regulation.
That is, people who are unreliable, who act on selfish impulses,
who break moral or legal rules, who fail to keep promises, who
subject others to emotional outbursts, who exhibit inappropriate
aggressive actions, and who in other ways fail at self-regulation are
often rejected or excluded by others. In such a context, the optimal
or adaptive response to social exclusion might be to try to increase
one’s self-regulatory performance. More generally, rejection often
is based on some undesirable trait or behavior, and so the most
adaptive response would be to seek to remedy that failing, toward
which self-regulation— defined as the capacity to change oneself
and one’s responses—would be instrumental. Hence, one might
predict and hope that social exclusion or rejection would stimulate
a broad tendency to improve self-regulation.
Against that line of reasoning, the present studies consistently
showed that self-regulation suffered in the wake of social exclu-
sion. Rejected or excluded participants performed worse on an
assortment of self-regulation tasks. Our investigation manipulated
social exclusion in the form of bogus feedback indicating that the
person would likely end up alone in life (Experiments 1, 3, and
4 – 6) and in the form of telling participants that no one else in their
group had wanted them as an interaction partner (Experiment 2).
These experiences of social exclusion rendered people less able to
make themselves drink a healthy but bad-tasting beverage (Exper-
iment 1); more prone to eat unhealthy snack foods (Experiment 2);
more prone to give up quickly at a frustrating, discouraging puzzle
task (Experiment 3); and less able to counteract distraction so as to
identify target words out of a stream of incoming verbiage (Ex-
periments 4 – 6).
Thus, instead of stimulating the seemingly adaptive and socially
desirable response of better self-regulation, rejection seems to
elicit the opposite. Our two final experiments were designed to
shed some light on why socially excluded people may be poor at
self-regulating. Experiment 5 suggested that they become unwill-
ing, rather than unable, to self-regulate. Given a compelling (cash)
incentive, the excluded participants were able to self-regulate
effectively after all. If the rejection experience had in some way
damaged or impaired their capacity for self-regulation, they would
not have been able to perform so well, even to earn money.
Apparently the capacity for self-regulation remains intact after
social exclusion, but the excluded person ordinarily does not want
to put forth the effort or make the sacrifices that self-regulation
Why might self-regulation become more distasteful in the wake
of rejection? Experiment 6 indicated that self-awareness may be a
contributing factor. Effective self-regulation requires a certain
degree of self-awareness to supervise the process of monitoring
and changing the self (Carver & Scheier, 1981): It is difficult to
alter the self to bring it into line with goals and standards if one
cannot be aware of where the self stands in relation to those
standards. But rejection makes self-awareness aversive, probably
because it would direct thought toward what deficiencies in the
self might have elicited the rejection (Hull & Levy, 1979; Twenge,
Catanese, & Baumeister, 2003). The implication is that excluded
people are disinclined to think about themselves, and the avoid-
ance of self-awareness undermines self-regulation. In Experiment
6, in which self-awareness was stimulated by placing participants
in front of a mirror, participants self-regulated effectively even
after social exclusion.
Self-regulation tends to be adaptive in most situations, and so it
is difficult to put a positive spin on the responses exhibited here.
The remarkable human capacity for self-regulation may have de-
veloped in part to help people curb selfish, impulsive, and other-
wise socially undesirable actions so as to be able to coexist and
cooperate with other human beings in a complex, interdependent
society. Undoubtedly, though, self-regulation has costs and can be
aversive, because it depends on focusing attention on the self’s
shortcomings to remedy them, because curbing selfish impulses
means not getting what one wants, and because altering the self
requires effort. One may be willing to bear these costs to the extent
that they accompany the substantial rewards of social acceptance
and belongingness. When those promised rewards are withheld,
however, one loses that willingness.
Before concluding that social exclusion leads to decrements in
self-regulation, it is necessary to consider several possible alter-
native explanations. A first possibility, which was anticipated in
the design of these studies, was that social exclusion or rejection
simply constitutes a form of bad news. If this view is correct, it
would indicate that receiving unpleasant feedback is the operative
cause of self-regulation failure, and so the present results (though
correct as far as they go) would not really indicate anything
specific about social exclusion. Theory should focus on the impact
of negative feedback rather than on a thwarted need to belong.
However, this explanation is contradicted by the results of our
misfortune control condition in these studies. Participants in that
condition received bad news (that is, they would be accident prone,
and so their future lives would be marred by frequent experiences
of pain and injury), but, unlike the participants who received the
social exclusion feedback, they did not respond with failed
There is ample evidence in previous work that emotional dis-
tress leads to self-regulation failure (for example, Grilo et al.,
1989; Keinan, 1987; Rosenthal & Marx, 1981; Sayette, 1993; Tice
et al., 2001; Wegener & Petty, 1994), and in fact we had originally
thought that many of the effects of social exclusion would be
mediated by emotional reactions. On an a priori basis, it was
highly plausible that social rejection could cause emotional dis-
tress, which in turn would lead to poor self-regulation. This would
not invalidate our results but would affect their interpretation.
However, multiple findings contradict the view that emotional
distress is an important factor. In the present investigation (and in
our previous studies), there was not much evidence that social
exclusion produced emotional distress. Even when emotion or
mood did differ significantly as a function of social exclusion, it
did not mediate self-regulation failures, and indeed the mediation
analyses generally failed on all counts, any one of which would
have been sufficient to reject a hypothesis of mood mediation.
Furthermore, participants in the misfortune control condition typ-
ically reported moods resembling those of participants in the social
exclusion (future alone) condition, but they behaved in the same
manner as people in the future belonging condition. If mood
mediated self-regulation failure, the bad moods in the misfortune
control condition should have led to poor self-regulation, but they
The finding that social exclusion did not produce emotional
distress is itself perhaps a reason for concern, insofar as one might
expect on a priori grounds that being excluded or rejected by
others would lead to emotional distress. We have four explanations
for the lack of apparent emotional response, all of which could be
true simultaneously. First, we generally measured emotion directly
after the exclusion or rejection feedback, whereas many emotions
take some time to build. We suspect that if we had measured
emotion an hour or two later (and not debriefed participants as to
the bogus nature of their exclusion feedback), we might well have
found more emotion. Second, participants in our studies receive
the exclusion feedback rather unexpectedly. Mostly they had no
inkling that they were going to receive such unwelcome news until
the experimenter actually gave it to them. In contrast, many
rejections in everyday life can be anticipated, such as when one is
waiting for decision letters from graduate or medical schools (or
journal editors, for that matter) or during fraternity or sorority rush,
in which everyone knows when the bids (and, by extension, the
rejections) will be communicated. The anticipation of feedback
about social acceptance and rejection allows suspense and arousal
to build and could prepare the person to have an emotional re-
Third, the slight majority of most of our research participants
have been men, and it is possible that men either do not know or
do not admit to the full extent of their emotions, whereas female
samples might report more extreme emotions and therefore furnish
enough variation to permit more sensitive tests of hypotheses
about emotion (LaFrance & Banaji, 1992). Consistent with this
view, we found at least some significant variation in mood valence
or positivity in Experiment 4, which had the highest proportion of
female participants. Fourth, it is also possible that the immediate
response to social rejection is a state of cognitive deconstruction
that includes emotional numbness (see Campbell, Baumeister,
Dhavale, & Tice, 2003; Twenge, Catanese, & Baumeister, 2003).
In any case, though, emotion and mood do not appear to play a
mediating role in the behavioral outcomes of social exclusion that
we have studied.
Another potential alternate explanation of our findings is that
the initial manipulations of rejection or exclusion caused partici-
pants to adopt a negative, hostile, or resentful attitude toward the
research project, and as a result they were less cooperative on the
second task than they otherwise would have been (and less coop-
erative than control condition participants). Such an alternate ex-
planation could be leveled at a great many manipulations in social
psychology and is difficult to rule out definitively. However, there
are some signs that it may be false. For one thing, there was no
sign of resentment or hostility displayed on the manipulation
checks, nor did such comments emerge in the debriefings of
participants. For another, the decrements in self-regulation were
not uniform enough to be compatible with a resentment explana-
tion. Experiment 4 was initially attempted with a somewhat easier
version of the dichotic listening task. Specifically, participants in
the first version were instructed simply to list words they heard
that began with m or p, as opposed to having those letters any-
where inside them. Resentful participants presumably would have
refused to list those words just as earnestly as they might have
refused to list the harder words, but in the easier version of the task
there was no difference in performance as a function of exclusion
feedback. Nearly all participants in the future alone condition
correctly identified all of the target words, which suggests that they
were not refusing to comply with instructions or detaching them-
selves from their assigned task.
Yet another possibility is that the feedback of social exclusion
was so impactful, puzzling, or threatening that participants devoted
much time and effort to thinking about it, such as mentally re-
viewing counterarguments and trying to refute it. In essence, the
impact of the exclusion manipulation was to distract participants,
thereby impairing subsequent information processing. This might
have interfered with their performance on subsequent tasks. The
present investigation did not directly address this possibility, but it
was in fact a central concern in a previous investigation conducted
by Baumeister et al. (2002). That investigation was focused on the
cognitive effects (including intelligent performance) of social ex-
clusion and involved methods similar to those used in the present
studies. The pattern of results there contradicted the distraction
hypothesis in multiple ways. If rejected participants were dis-
tracted, they should have shown decrements in noticing and en-
coding information, but they did not. Excluded and rejected people
were quite capable of processing information and attending to
external stimuli, and indeed the only intellectual decrements were
observed on tasks that required active, effortful management.
Another objection to this alternative explanation is that if excluded
participants were distracted by preoccupation with how to think
about the rejecting feedback, then focusing their attention on
themselves with a mirror (in Experiment 6) should have exacer-
bated the effect; in fact, we found the opposite, namely that the
mirror wiped out the effect.
A final alternate explanation might be that the primary effect of
rejection is to lower self-esteem. A reduction in self-esteem might
conceivably engender lower confidence in one’s ability to perform
well, which might in turn have produced some of the present
findings. Experiment 3 specifically investigated the possibility of
mediation by changes in state self-esteem. As that study revealed,
state self-esteem was not reliably altered by the exclusion manip-
ulation, nor did it have a clear effect on the dependent measure of
self-regulation. State self-esteem does not appear to have been a
Five years ago, a group of us set out on this line of work from
the basic assumption that the need to belong is a basic, pervasive,
and powerful motivation. We reasoned that thwarting that need
would lead to emotional distress, which in turn would produce a
variety of behavioral effects. In empirical tests, we repeatedly
found the behavioral effects but no sign of emotional distress or of
a mediating role for emotion. Hence, despite what were often very
large effects on behaviors, the inner processes following from
social exclusion or rejection remained unknown.
The present investigation moves toward a new and different
understanding of the relevant inner processes that ensue when the
need to belong is thwarted. Specifically, we found, across six
studies, that self-regulation is substantially impaired among people
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE
who have just received news of social rejection or future exclusion.
Self-regulation was measured in terms of consumption, persis-
tence, and attentional control, and the convergence across these
very different types of self-regulation points to a general impair-
ment of the capacity for self-regulation.
In our view, the broader implication of these results is that much
of self-regulation is a costly, effortful, and therefore fragile pro-
cess. The immense advantages and rewards that flow from social
acceptance may make people normally willing to tolerate the costs
and sacrifices that self-regulation requires. But people may lose
that willingness when social acceptance and its rewards are not
Past work has shown that socially excluded individuals exhibit
increased aggression, poorer intellectual performance, a loss of
prosocial behavior, and a susceptibility to self-defeating behavior
patterns. At the societal level, and at multiple points in history,
groups and categories of people who have felt excluded by the
dominant culture have shown sadly similar patterns as reflected in
high crime rates, underperformance in schools and intellectual life,
withdrawal from positive contributions to the general societal
good, and elevated rates of substance abuse, suicide, and other
self-destructive patterns. The present findings suggest what may be
a common underlying process. Effective self-regulation allows
individuals to control and alter their behavior so as to resist
temptations, stifle socially undesirable impulses, follow rules, pur-
sue enlightened self-interest despite short-term costs, and make
positive contributions to society. As such, individual self-
regulation is essential to one’s own well-being as well as that of
others. Messages of rejection or exclusion can cause people to be
less willing to make the exertions and sacrifices needed for effec-
tive self-regulation, with potentially tragic results for both them-
selves and others. Though our findings are hardly adequate for
prescribing social change, they do lend support to the view that
promoting a more widely inclusive society, such that fewer groups
or individuals feel left out, would reduce the extensive harm and
heartbreak that often follow from self-regulation failure.
Baron, R. M., & Kenny, D. A. (1986). The moderator-mediator variable
distinction in social psychological research: Conceptual, strategic, and
statistical considerations. Journal of Personality and Social Psychology,
Baumeister, R. F. (1990). Suicide as escape from self. Psychological
Review, 97, 90 –113.
Baumeister, R. F. (1991). Escaping the self: Alcoholism, spirituality,
masochism, and other flights from the burden of selfhood. New York:
Baumeister, R. F., Bratslavsky, E., Finkenauer, C., & Vohs, K. D. (2001).
Bad is stronger than good. Review of General Psychology, 5, 323–370.
Baumeister, R. F., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998).
Ego depletion: Is the active self a limited resource? Journal of Person-
ality and Social Psychology, 74, 1152–1165.
Baumeister, R. F., Heatherton, T. F., & Tice, D. M. (1994). Losing control:
How and why people fail at self-regulation. San Diego, CA: Academic
Baumeister, R. F., & Leary, M. R. (1995). The need to belong: Desire for
interpersonal attachments as a fundamental human motivation. Psycho-
logical Bulletin, 117, 497–529.
Baumeister, R. F., & Scher, S. J. (1988). Self-defeating behavior patterns
among normal individuals: Review and analysis of common self-
destructive tendencies. Psychological Bulletin, 104, 3–22.
Baumeister, R. F., & Tice, D. M. (1990). Anxiety and social exclusion.
Journal of Social and Clinical Psychology, 9, 165–195.
Baumeister, R. F., Twenge, J. M., & Nuss, C. K. (2002). Effects of social
exclusion on cognitive processes: Anticipated aloneness reduces intelli-
gent thought. Journal of Personality and Social Psychology, 83, 817–
Box, G. E. P. (1954). Some theorems on quadratic forms applied in the
study of analysis of variance problems. Annals of Statistics, 25, 290 –
Brownell, K. D. (1991). Dieting and the search for the perfect body: Where
physiology and culture collide. Behavior Therapy, 22, 1–12.
Bryden, M. P. (1988). An overview of the dichotic listening procedure and
its relation to cerebral organization. In K. Hugdahl (Ed.), Handbook of
dichotic listening: Theory, methods, and research (pp. 1– 43). Chiches-
ter, England: Wiley.
Buckley, K., Winkel, R., & Leary, M. (2004). Reactions to acceptance and
rejection: Effects of level and sequence of relational evaluation. Journal
of Experimental Social Psychology, 40, 14 –28.
Cacioppo, J. T., Hawkley, L. C., & Berntson, G. G. (2003). The anatomy
of loneliness. Current Directions in Psychological Science, 12, 71–74.
Campbell, W. K., Baumeister, R. F., Dhavale, D., & Tice, D. M. (2003).
Responding to major threats to self-esteem: A preliminary, narrative
study of ego-shock. Journal of Social and Clinical Psychology, 22,
Carver, C. S., & Scheier, M. F. (1981). Attention and self-regulation: A
control theory approach to human behavior. New York: Springer-
Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful conceptual
framework for personality-social, clinical and health psychology. Psy-
chological Bulletin, 92, 111–135.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences.
New York: Academic Press.
Duval, S., & Wicklund, R. A. (1972). A theory of objective self-awareness.
New York: Academic Press.
Eysenck, H. J., & Eysenck, S. B. G. (1975). Manual of the Eysenck
Personality Questionnaire. San Diego, CA: EDITS.
Finkel, E. J., & Campbell, W. K. (2001). Self-control and accommodation
in relationships: An interdependence analysis. Journal of Personality
and Social Psychology, 81, 263–277.
Geffen, G., & Caudrey, D. (1981). Reliability and validity of the dichotic
monitoring test for language laterality. Neuropsychologia, 19, 413– 423.
Glass, G. V., & Hopkins, K. D. (1996). Statistical methods in education
and psychology (3rd ed.). Boston: Allyn & Bacon.
Gottfredson, M. R., & Hirschi, T. (1990). A general theory of crime.
Stanford, CA: Stanford University Press.
Greenberg, J., & Musham, C. (1981). Avoiding and seeking self-focused
attention. Journal of Research in Personality, 15, 191–200.
Grilo, C. M., Shiffman, S., & Wing, R. R. (1989). Relapse crises and
coping among dieters. Journal of Consulting and Clinical Psychology,
57, 488 – 495.
Heatherton, T. F., & Baumeister, R. F. (1991). Binge eating as escape from
self-awareness. Psychological Bulletin, 110, 86–108.
Heatherton, T. F., & Polivy, J. (1991). Development and validation of a
scale for measuring state self-esteem. Journal of Personality and Social
Psychology, 60, 895–910.
Heatherton, T. F., & Polivy, J. (1992). Chronic dieting and eating disor-
ders: A spiral model. In J. Crowther, S. Hobfall, M. Stephens, & D.
Tennenbaum (Eds.), The etiology of bulimia: The individual and familial
context (pp. 133–155). Washington, DC: Hemisphere.
Heatherton, T. F., Polivy, J., & Herman, C. P. (1991). Restraint, weight
loss, and variability of body weight. Journal of Abnormal Psychology,
100, 78 – 83.
Hull, J. G., & Levy, A. S. (1979). The organizational functions of the self:
An alternative to the Duval and Wicklund model of self-awareness.
Journal of Personality and Social Psychology, 37, 756–768.
Keinan, G. (1987). Decision making under stress: Scanning of alternatives
under controllable and uncontrollable threats. Journal of Personality and
Social Psychology, 52, 639– 644.
Kirkpatrick, L. A., Waugh, C. E., Valencia, A., & Webster, G. D. (2002).
The functional domain specificity of self-esteem and the differential
prediction of aggression. Journal of Personality and Social Psychology,
82, 756 –767.
LaFrance, J., & Banaji, M. (1992). Toward a reconsideration of the
gender-emotion relationship. In M. Clark (Ed.), Emotion and social
behavior (pp. 178 –201). Newbury Park, CA: Sage.
Leary, M. R., & Baumeister, R. F. (2000). The nature and function of
self-esteem: Sociometer theory. In M. Zanna (Ed.), Advances in exper-
imental social psychology (Vol. 32, pp. 1– 62). San Diego, CA: Aca-
Leary, M. R., Tambor, E. S., Terdal, S. K., & Downs, D. L. (1995).
Self-esteem as an interpersonal monitor: The sociometer hypothesis.
Journal of Personality and Social Psychology, 68, 518–530.
Longshore, D. (1998). Self-control and criminal opportunity: A prospective
test of the general theory of crime. Social Problems, 45, 102–113.
Lord, C., Ross, L., & Lepper, M. (1979). Biased assimilation and attitude
polarization: The effects of prior theories on subsequently considered
evidence. Journal of Personality and Social Psychology, 37, 2098 –
Lynch, J. J. (1979). The broken heart: The medical consequences of
loneliness. New York: Basic Books.
Maszk, P., Eisenberg, N. G., & Guthrie, I. K. (1999). Relations of chil-
dren’s social status to their emotionality and regulation: A short-term
longitudinal study. Merrill-Palmer Quarterly, 45, 468 – 492.
Mayer, J. D., & Gaschke, Y. N. (1988). The experience and meta-
experience of mood. Journal of Personality and Social Psychology, 55,
McGuire, J., & Broomfield, D. (1994). Violent offences and capacity for
self-control. Psychology Crime & Law, 2, 117–123.
Mischel, W. (1974). Processes in delay of gratification. In L. Berkowitz
(Ed.), Advances in experimental social psychology (Vol. 7, pp. 249 –
292). San Diego, CA: Academic Press.
Mischel, W. (1996). From good intentions to willpower. In P. Gollwitzer
& J. Bargh (Eds.), The psychology of action (pp. 197–218). New York:
Muraven, M. (1998). Mechanisms of self-control failure: Motivation and
limited resources. Unpublished doctoral dissertation, Case Western Re-
serve University, Cleveland, OH.
Muraven, M., & Baumeister, R. F. (2000). Self-regulation and depletion of
limited resources: Does self-control resemble a muscle? Psychological
Bulletin, 126, 247–259.
Muraven, M., Tice, D. M., & Baumeister, R. F. (1998). Self-control as
limited resource: Regulatory depletion patterns. Journal of Personality
and Social Psychology, 74, 774–789.
Myers, D. (1992). The pursuit of happiness. New York: Morrow.
Nezlek, J. B., Kowalski, R. M., Leary, M. R., Blevins, T., & Holgate, S.
(1997). Personality moderators of reactions to interpersonal rejection:
Depression and trait self-esteem. Personality and Social Psychology
Bulletin, 23, 1235–1244.
Oldfeld, R. C. (1971). The assessment and analysis of handedness: The
Edinburgh Inventory. Neuropsychologia, 9, 97–113.
O’Neil, P. M., & Jarrell, M. P. (1992). Psychological aspects of obesity and
dieting. In T. A. Wadden & T. B. VanItallie (Eds.), Treatment of the
seriously obese patient (pp. 252–270). New York: Guilford Press.
Rosenthal, B. S., & Marx, R. D. (1981). Determinants of initial relapse
episodes among dieters. Obesity/Bariatric Medicine, 10, 94 –97.
Sayette, M. A. (1993). An appraisal-disruption model of alcohol’s effec-
tiveness on stress responses in social drinkers. Psychological Bulletin,
114, 459 – 476.
Sedikides, C., Campbell, W. K., Reeder, G. D., & Elliot, A. J. (1999). The
relationship closeness induction task. Representative Research in Social
Psychology, 23, 1– 4.
Sentyrz, S. M., & Bushman, B. J. (1998). Mirror, mirror on the wall, who’s
the thinnest one of all? Effects of self-awareness on consumption of
fatty, reduced-fat, and fat-free products. Journal of Applied Psychology,
83, 944 –949.
Stunkard, A. J. (1993). A history of binge eating. In C. G. Fairburn & G. T.
Terence (Eds.), Binge eating: Nature, assessment, and treatment (pp.
15–34). New York: Guilford Press.
Tice, D. M., Bratslavsky, E., & Baumeister, R. F. (2001). Emotional
distress regulation takes precedence over impulse control: If you feel
bad, do it! Journal of Personality and Social Psychology, 80, 53– 67.
Twenge, J. M., Baumeister, R. F., Tice, D. M., & Stucke, T. S. (2001). If
you can’t join them, beat them: Effects of social exclusion on aggressive
behavior. Journal of Personality and Social Psychology, 81, 1058 –1069.
Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2002). Social
exclusion causes self-defeating behavior. Journal of Personality and
Social Psychology, 83, 606– 615.
Twenge, J. M., Catanese, K. R., & Baumeister, R. F. (2003). Social
exclusion and the deconstructed state: Time perception, meaningless-
ness, lethargy, lack of emotion, and self-awareness. Journal of Person-
ality and Social Psychology, 85, 409 – 423.
Twenge, J. M., Ciarocco, N. J., Cuervo, D., & Baumeister, R. F. (2003).
Social exclusion reduces prosocial behavior. Unpublished manuscript.
Vohs, K. D., & Schmeichel, B. J. (2003). Self-regulation and the extended
now: Controlling the self alters the subjective experience of time. Jour-
nal of Personality and Social Psychology, 85, 217–230.
Warburton, W., Williams, K. D., & Cairns, D. (2003, April). Effects of
ostracism and loss of control on aggression. Paper presented at the 32nd
Annual Meeting of the Society of Australasian Social Psychology, Bondi
Wegener, D. T., & Petty, R. E. (1994). Mood management across affective
states: The hedonic contingency hypothesis. Journal of Personality and
Social Psychology, 66, 1034–1048.
Williams, K. D. (2001). Ostracism: The power of silence. New York:
Received November 23, 2003
Revision received September 7, 2004
Accepted September 8, 2004 䡲
Instructions to Authors
For Instructions to Authors, please consult the March issue of the volume or visit www.apa.org/
journals/psp and click on the “Instructions to Authors” link in the Journal Info box on the right.
BAUMEISTER, DEWALL, CIAROCCO, AND TWENGE