g
.
ject
(no
t re
ver
xclu
enn
ork
vidu
as opposed to ‘‘fake” smiles. This effect was partially mediated by threats to ‘‘relational needs” (Williams,
responses that could facilitate reconnection with others.
Published by Elsevier Inc.
ial, and
ersive
hat hu
cting d
(e.g., L
duals
greater performer control, can conceal negative emotions or fake
the desired positivity associated with a real smile (Ekman, Friesen,
& O’Sullivan, 1988). Unlike non-Duchenne smiles, Duchenne smiles
are strong signals of the person’s cooperative intent (Brown &
Moore, 2002); people tend to exhibit more Duchenne smiles while
non-Duchenne smiles, it did not showwhether rejected individuals
‘‘use” this information in any way. The identification of such smiles
is only beneficial if such discrimination produces responses useful
for satiating current needs. Thus, excluded individuals should also
prefer to interact with persons expressing true approach displays
(e.g., Duchenne smiles) rather than those not expressing such dis-
plays, which is an untested hypothesis. Moreover, previous
researchers explicitly directed participants’ attention to the verac-
ity of the smiles. It remains unclear whether rejected individuals
* Corresponding author. Address: Department of Psychology, Psychology Build-
ing, Miami University, Oxford, OH 45056, United States.
Journal of Experimental Social Psychology 46 (2010) 196–199
Contents lists availab
t
.e
E-mail address: bernstmj@muohio.edu (M.J. Bernstein).
their need to belong better identify social information, like facial
expressions (Pickett, Gardner, & Knowles, 2004), and those who
experience exclusion engage in greater behavioral mimicry (Lakin,
Chartrand, & Arkin, 2008), an indication that individuals fearing or
experiencing rejection show increased attention to social cues use-
ful for securing reaffiliation.
One rather powerful cue of affiliation intent is positive affect,
which is often communicated by smiling (e.g., Brown, Palameta,
& Moore, 2003). Specifically, ‘‘Duchenne” (genuine) smiles occur
automatically in response to the experience of happiness (Ekman,
Davidson, & Friesen, 1990). Non-Duchenne smiles, which are under
previously found that those recalling an exclusion experience
could better discriminate between Duchenne and non-Duchenne
smiles than those recalling an acceptance or mundane experience
(Bernstein, Young, Brown, Sacco, & Claypool, 2008). Accurate iden-
tification of non-verbal signals of approach affords an avenue for
successful reaffiliation. Because excluded persons have a greater
reaffiliative need, it would be essential for them to focus on part-
ners most likely to meet these needs, which may be facilitated
by accurate perception of real and deceptive smiles.
Though this previous work showed that rejected individuals
have an acute ability to differentiate between Duchenne and
Introduction
Being included in groups is essent
social relationships feels highly av
1995). Inclusion is so important t
evolved mechanisms capable of dete
to facilitate reconnection with others
Downs, 1995). For example, indivi
0022-1031/$ - see front matter Published by Elsevier
doi:10.1016/j.jesp.2009.08.010
failing to secure stable
(Baumeister & Leary,
mans appear to have
eficits in belongingness
eary, Tambor, Terdal, &
dispositionally high in
engaging in pro-social behaviors than when not (Mehu, Grammer, &
Dunbar, 2007). Thus, ‘‘real” smilers are good candidates for poten-
tial affiliation whereas fake smilers are potentially deceptive.
Given the hazards facing the socially rejected, it may be useful
for such persons to quickly and accurately discriminate between
facial expressions of emotion (e.g., happy versus sad), and to accu-
rately distinguish between real and false signals of motivational in-
tent, especially affiliation intentions. Consistent with this logic, we
Face perception
Emotions 2007) and fully mediated by threats to self-esteem. These results suggest that exclusion yields adaptive
FlashReport
A preference for genuine smiles followin
Michael J. Bernstein a,*, Donald F. Sacco a, Christina M
aMiami University, Oxford, OH, United States
b St. Louis University, St. Louis, MO, United States
a r t i c l e i n f o
Article history:
Received 3 June 2009
Revised 29 July 2009
Available online 28 August 2009
Keywords:
Social exclusion
Ostracism
Smiles
a b s t r a c t
Research indicates that re
(Duchenne) and deceptive
nity). We hypothesized tha
uals displaying Duchenne
experiences of inclusion, e
individuals (10 with Duch
much they would like to w
participants, excluded indi
Journal of Experimen
journal homepage: www
Inc.
social exclusion
Brown b, Steven G. Young a, Heather M. Claypool a
ed individuals are better than others at discriminating between genuine
n-Duchenne) smiles (i.e., true versus false signals of affiliative opportu-
jected individuals would show a greater preference to work with individ-
sus non-Duchenne smiles. To test this, participants wrote essays about
sion, or mundane events. They then saw a series of 20 videos of smiling
e and 10 with non-Duchenne smiles). Participants then indicated how
with each target. Analyses revealed that compared to included and control
als showed a greater preference to work with individuals displaying ‘‘real”
le at ScienceDirect
al Social Psychology
lsevier .com/locate / jesp
will attend to and act on these differences in smile types spontane-
ously, without having their sincerity explicitly questioned.
Participants and design
with success (e.g., Maner, DeWall, Baumeister, & Schaller, 2007).
Participants then responded to 16 items assessing their levels of
belonging, control, self-esteem, and meaningful existence (four
items each) felt during the experience (adapted from Zadro, Wil-
liams, & Richardson, 2004).
Once completed, participants were told they would see videos
of individuals and that they were to imagine that the person in
each was a potential partner for a project on which they might
work. Participants were to indicate how much they would like to
work with each person on a Likert-scale (1 = not at all; 7 = very
much) for all 20 videos. Upon completion, participants responded
to demographic questions, were probed for suspicion, thanked,
was a main effect of smile. Participants preferred working with
individuals exhibiting Duchenne rather than non-Duchenne
enne (M = 4.57, SD = .74) versus non-Duchenne smiles (M = 4.15,
2
M.J. Bernstein et al. / Journal of Experi ental Social Psychology 46 (2010) 196–199 197
One hundred and twenty-five individuals (81 females) partici-
pated for course credit and were randomly assigned to a 3 (social
experience: exclusion, inclusion, or control) � 2 (smile: Duchenne,
non-Duchenne) mixed-model design with repeated measures on
the latter. There were no effects of target or participant sex which
are not discussed further.
Materials
The facial stimuli were those used in Bernstein et al. (2008) and
were obtained from the BBC science website (http://www.bbc.co.
uk/science/humanbody/mind/surveys/smiles).1 Participants watched
20 videos (approximately 4 s each) one at a time, each depicting an
individualwith an initially-neutral expression that shifted to a smiling
expression, that then returned to a neutral expression (10 Duchenne
and 10 non-Duchenne smiles). Thirteen men and seven women were
depicted in the videos.2 Presentation orderwas counterbalanced, such
that participants saw one of two possible stimuli orders.3
Procedure
Participants performed two ostensibly unrelated tasks. They
first completed an essay task constituting the manipulation of so-
cial experience. Participants wrote about a time they felt ‘‘rejected
or excluded,” ‘‘accepted or included,” or ‘‘their morning yesterday”
(control condition). This manipulation has been used previously
1 Pre-testing revealed that faces displaying Duchenne versus non-Duchenne smiles
did not differ in perceived attractiveness, trustworthiness, or positivity (p > .37).
The current research will investigate these issues. We randomly
assigned participants to an exclusion, inclusion, or control condi-
tion and asked them to rate their desire to work with targets dis-
playing both Duchenne and non-Duchenne smiles. Importantly,
we did not draw participants’ attention to the targets’ smiles, nor
did the instructions ever suggest that smile sincerity varied. We
predicted that all perceivers would show a preference for working
with targets exhibiting Duchenne rather than non-Duchenne
smiles, but that this pattern would be strongest for rejected
individuals.
Furthermore, we examined possible mechanisms driving this
outcome. Williams (2007) has shown that four basic needs are
thwarted following rejection which can be grouped into two cate-
gories: ‘‘relational” needs (belonging, self-esteem) and ‘‘efficacy/
existence” needs (control, meaningful existence). He has proposed
that individuals engage in reaffiliative (rather than antisocial) reac-
tions when ‘‘relational needs” are most impacted. Given that the
outcome under investigation is affiliative, these ‘‘relational” as op-
posed to ‘‘efficacy” needs may operate as the mediator of the Duch-
enne preference. Additionally, work on the sociometer model
(Leary et al., 1995) argues that self-esteem drops following
perceived inclusion threats and motivates humans to engage in
behaviors to re-establish their social ties. From this perspective,
self-esteem alone may be the key mediator of our proposed find-
ing. This work will investigate which of these need threats mediate
differential preferences for working with individuals displaying
real versus deceptive smiles.
Methods
2 The stimuli included three ethnic minorities. Removing these from the analyses
left the results unchanged. Thus, all analyses included all stimuli.
3 There were no counterbalancing effects on any results.
m
SD = .71; p < .001; g = .17).
3.6
3.8
4
4.2
4.4
4.6
4.8
Exclusion Control Inclusion
Social Experience
D
es
ire
to
W
or
k
w
ith
T
ar
ge
ts
Duchenne
Non-Duchenne
smiles, F(1, 122) = 24.46, p < .001, g2 = .17. As predicted, this effect
was qualified by an interaction between social experience and
smile, F(2, 122) = 3.26, p = .04, g2 = .05 (Fig. 1). Participants in the
control condition showed a marginal preference for individuals
exhibiting Duchenne (M = 4.43, SD = .76) versus non-Duchenne
smiles (M = 4.28, SD = .67; p = .096; g2 = .02). Participants in the
inclusion condition showed a similar marginal effect, Duchenne
(M = 4.60, SD = .88) versus non-Duchenne smiles (M = 4.43, SD =
.83; p = .055; g2 = .03). Excluded participants, however, showed a
significant and larger preference for workingwith those with Duch-
and debriefed.
Results
Basic needs
To examine if the social-experience manipulation was success-
ful, we calculated each of the four basic needs (belonging, control,
self-esteem, meaningful existence) separately for each participant.
In all cases, exclusion led to less basic-need satisfaction compared
to control and included participants (ps < .001), while the latter
two groups did not differ from one another (ps > .35).
Preference scores
Of primary interest was whether social exclusion influences the
desire to work with targets exhibiting Duchenne and non-Duch-
enne smiles. For each participant, we averaged (separately) their
preference scores for targets with real smiles and those with fake
smiles. These averages were subjected to a 3 (social experience:
exclusion, control, inclusion) � 2 (smile: Duchenne, non-Duch-
enne) mixed-model ANOVA, with repeated measures on the latter.
There was no main effect of social experience (p > .52), but there
Fig. 1. The effect of social experience on desire to work with targets exhibiting real
and fake smiles (error bars represent the standard error of the mean).
To further examine the differences between social-experience
p < .001). In both cases, excluded participants had less satisfaction
of those needs. Relational needs were further related to prefer- Thus, individuals with Duchenne smiles exhibit an affiliative
signal. Because rejected individuals should have the strongest
Social
Experience
0=Inclusion/Control
1=Exclusion
Preference-Difference
Score
Relational Needs
β = -1.00*
β=.180
(β= .262*)
β= -.079+
Fig. 2. The effect of social experience on desire to work with real- versus fake-
Social
Experience
0= Inclusion/Control
1= Exclusion
Preference-Difference
Score
Self-Esteem
β= -.88*
β= .166
(β= .262*)
β= -.108*
198 M.J. Bernstein et al. / Journal of Experimental Social Psychology 46 (2010) 196–199
ence–difference scores (b = �.107, p = .008), but efficacy was not
(p > .13) and thus cannot act as a mediator. Thus, as individuals felt
less satisfaction of their relational needs, they showed a greater
preference to work with Duchenne-smile (versus non-Duchenne-
smile) targets. Finally, when regressing social experience and rela-
tional needs simultaneously onto preference–difference scores, the
effect of relational needs remained marginally significant (b = �.08,
p = .068), whereas the effect of social experience became non-sig-
nificant, b = .18, p = .10. A Sobel test revealed the drop in the mag-
nitude of the relationship between social experience and the
preference–difference score was marginally significant, z = 1.71,
p = .087 (Fig. 2).
conditions, for each participant we subtracted his/her average pref-
erence score for fake-smile targets from his/her average preference
score for real-smile targets, thus creating a ‘‘preference–difference”
score where higher numbers indicated a greater preference to
work with real-smile targets. A one-way ANOVA on this was signif-
icant, F(2122) = 3.26, p = .04; post hoc analyses revealed that while
control (M = .15, SD = .55) and inclusion participants (M = .16,
SD = .59) did not differ (p > .85), both significantly differed from re-
jected participants (M = .42, SD = .51; ps < .035). These results are
consistent with the hypothesis that excluded individuals would
show an even greater preference for working with Duchenne ver-
sus non-Duchenne targets.
Mediational analyses
As mentioned earlier, Williams (2007) has shown that rejection
thwarts four psychological needs. These needs can be divided into
two categories which are purported to account for different behav-
ioral reactions to rejection. When belonging and self-esteem are
most impacted, individuals tend to engage in reaffiliative behav-
iors, whereas threats to control and meaningful existence lead to
antisocial responses (Williams, 2007). Because our outcome vari-
able related to reaffiliation, we suspected that threats to the
belonging/self-esteem needs, and not the control/meaningful exis-
tence needs, might act as a mediator of our primary effect. We
chose to begin our examination by looking at these subcategories
of the basic needs as possible mediators.
Belonging and self-esteem needs were highly correlated (r = .79,
p < .001), as were control and meaningful existence (r = .75,
p < .001). We thus averaged belonging and self-esteem needs into
one index (Relational Needs) and did the same with control and
meaningful existence (Efficacy Needs). In all cases, higher values
indicate more satisfaction of the need state.4
Given that the preference scores and basic needs did not differ
between inclusion and control conditions, we combined these con-
ditions and dummy coded the social-experience variable
(0 = inclusion/control, 1 = exclusion). We then used the ‘‘prefer-
ence–difference” score as our dependent measure.
Following Baron and Kenny (1986), we first examined the rela-
tion between social experience (dummy coded) and the prefer-
ence–difference score. This relation was significant, b = .26,
p = .012. Excluded individuals showed a greater preference for
working with Duchenne-smile targets than those in the control/
inclusion conditions. We then found significant relationships be-
tween social experience and both possible mediators, threats to
relational (b = �1.00, p < .001) and efficacy needs (b = �1.04,
4 The indices of the four basic needs were all significantly correlated with each
other (rs > .63, ps < .001), and the relational and efficacy indices were as well (r = .82,
p < .001).
Given the marginal findings above, we examined belonging and
self-esteem separately as possible mediators to compare William’s
(2007) and Leary and colleagues’ (1995) models. Both belonging
(b = �1.13, p < .001) and self-esteem (b = �.88, p < .001) were pre-
dicted by social experience, such that excluded individuals experi-
enced less of each than those in the inclusion/control conditions.
Further, both belonging (b = �.07, p = .05) and self-esteem
(b = �.132, p = .002) predicted the preference–difference score,
such that as the participants’ satisfaction of these needs decreased,
they showed a greater preference to work with people exhibiting
Duchenne versus non-Duchenne smiles. When regressing social
experience and belonging simultaneously onto the preference–dif-
ference scores the effect of social experience remained significant,
(b = .22, p = .05) whereas belonging did not (p > .25). Thus satisfac-
tion of belonging needs was not a mediator. However, when
regressing social experience and self-esteem simultaneously onto
the preference–difference scores, full mediation occurred (Fig. 3);
self-esteem remained significant (b = �.11, p = .01) whereas the ef-
fect of social experience dropped to below significance (b = .17,
p = .12). A Sobel test revealed that this drop in the magnitude of
the relationship between social experience and our preference–dif-
ference score was significant, z = 2.16, p = .03. Such results are
quite consistent with Leary and colleagues’ (1995) sociometer
hypothesis.
Discussion
Bernstein and colleagues (2008) showed that rejected individu-
als better discriminate real from deceptive smiles when directly
asked to do so. The current work extends this finding by showing
that excluded individuals show a greater desire to work with tar-
gets exhibiting real versus fake smiles, compared to included or
control participants. Duchenne smiles function to convey true feel-
ings of positive affect and signal a desire to cooperate while non-
Duchenne smiles function only to mask some unknown intent.
smiling targets as mediated by relational needs satisfaction (� indicates significant
at <05; + indicates marginal significance).
Fig. 3. The effect of social experience on desire to work with real- versus fake-
smiling targets as mediated by self-esteem needs satisfaction (� indicates signif-
icant at <05; + indicates marginal significance).
reaffiliative need, we predicted that they should show the greatest
desire to work with such targets, and our findings supported this
hypothesis. Importantly, participants were not told that smile sin-
cerity varied across targets. Thus, without drawing participants’
attention to smile veracity, we illustrated that perceivers, espe-
cially excluded ones, spontaneously made judgments favoring
Duchenne-smile targets.
This work also found that the primary finding was marginally
mediated by perceivers’ relational needs satisfaction (the combina-
tion of belonging and self-esteem needs). This meditational model
showed that exclusion decreased one’s sense of relational needs,
and that the more these needs were threatened, the bigger (mar-
ginally) one’s preference for Duchenne-smile targets. Interestingly,
efficacy needs showed no relationship with the preference mea-
sure. Examining these findings vis-à-vis Williams’ (2007) model,
which suggests that threats to relational needs are good predictors
of pro-social and affiliative responses whereas threats to efficacy
needs result in more antisocial and self-focused responses, reveals
that the current evidence is congruent with his theorizing.
However, the fact that self-esteem alone fully mediated the ef-
fect warrants further discussion. Self-esteem is the