Jumping on the 'badwagon'? How group membership influences responses to the social exclusion of others

Abstract

In four studies, we addressed whether group membership influences behavioral and neural responses to the social exclusion of others. Participants played a modified three-player Cyberball game (Studies 1-3) or a team-selection task (Study 4) in the absence or presence of a minimal group setting. In the absence of a minimal group, when one player excluded another player, participants actively included the excluded target. When the excluder was from the in-group and the excluded player from the out-group, participants were less likely to intervene (Studies 1-3), and also more often went along with the exclusion (Study 4). fMRI results (Study 3) showed that greater exclusion in the minimal group setting concurred with increased activation in the dlPFC, a region associated with overriding cognitive conflict. Self-reports from Study 4 supported these results by showing that participants' responses to the target's exclusion were motivated by group membership as well as participants' general aversion to exclude others. Together, the findings suggest that when people witness social exclusion, group membership triggers a motivational conflict between favoring the in-group and including the out-group target. This underscores the importance of group composition for understanding the dynamics of social exclusion.

Full text

UNCORRECTED MANUSCRIPT
1
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Jumping on the “badwagon”? How group membership influences responses to the social
exclusion of others
Gert-Jan Lelieveld1,2, Lasana T. Harris3, Lotte F. van Dillen1,2
1Institute of Psychology, Leiden University, the Netherlands, 2Leiden Institute for Brain
and Cognition, Leiden, the Netherlands, 3Department of Experimental Psychology,
University College London
Word count: 11929
Author Note
Correspondence concerning this article should be addressed to Gert-Jan Lelieveld,
Institute of Psychology, Leiden University, P.O. Box 9555, 2300 RB Leiden, the
Netherlands. Phone: +31(0) 71 – 5276615; Fax: +31(0) 71 – 5273080. E-mail:
lelieveldgj@fsw.leidenuniv.nl
© The Author(s) 2020. Published by Oxford University Press.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
2
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Abstract
In four studies, we addressed whether group membership influences behavioral and
neural responses to the social exclusion of others. Participants played a modified three-
player Cyberball game (Studies 1-3) or a team-selection task (Study 4) in the absence or
presence of a minimal group setting. In the absence of a minimal group, when one player
excluded another player, participants actively included the excluded target. When the
excluder was from the in-group and the excluded player from the out-group, participants
were less likely to intervene (Studies 1-3), and also more often went along with the
exclusion (Study 4). fMRI results (Study 3) showed that greater exclusion in the minimal
group setting concurred with increased activation in the dlPFC, a region associated with
overriding cognitive conflict. Self-reports from Study 4 supported these results by
showing that participants’ responses to the target’s exclusion were motivated by group
membership as well as participants’ general aversion to exclude others. Together, the
findings suggest that when people witness social exclusion, group membership triggers a
motivational conflict between favoring the in-group and including the out-group target.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
3
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
This underscores the importance of group composition for understanding the dynamics of
social exclusion.
Keywords: Social exclusion; fMRI; Group membership; dlPFC; Cyberball
Jumping on the “badwagon”? How group membership influences responses to the
social exclusion of others
Previous research on social exclusion strongly focused on its detrimental effects for
victims (Williams, 2007), but the answer to the question why people exclude others and
under which circumstances remains inconclusive (e.g., Wesselmann, Wirth, Pryor,
Reeder & Williams, 2013). The scarce research on the decision to include versus exclude
has shown that inclusion is the norm in most social situations (Kerr, Seok, Poulsen,
Harris, & Messe, 2008), and that explicit instructions to ostracize others induce emotional
distress (Zadro, Williams, & Richardson, 2005). Still, social exclusion occurs frequently
among both children (Wang, Iannotti, Luk, & Nansel, 2010) and adults (Williams, 2007),
underscoring the need to better understand the factors driving social exclusion. In this
light it is important to consider that exclusion is typically a group effort. To understand
the dynamics of social exclusion it is thus important to incorporate this group context and
not only focus on the initiator of exclusion, but also examine how others within the group
react in turn. These other group members might intervene by trying to again include the
excluded target, observe the situation without addressing the exclusion, or actively go
along with the exclusion (Malti, Strohmeier, & Killen, 2015). Different motives may
underlie the decision to intervene or not, such as the motivation to include the excluded
target because of exclusion aversion and/or the motivation to favor the person who
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
4
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
initiated the exclusion. These motives are not mutually exclusive and can create a
dilemma for people when deciding how to respond to the exclusion of a target. The main
goal of the current research was to study whether social inclusion norms towards the
target, could emerge without group conformity norms to reciprocate the excluder. Thus,
group membership may play a central role in encouraging the decision to include the
target or favor the excluder.
There are many arguments for why group membership should affect social
exclusion. Prior research has shown that in-group members are seen as more similar in
attitudes and values than out-group members, and that this shapes our social interactions
(Tajfel & Turner, 1986). People have more affinity for in-group members and tend to
favor them over out-group members (Hewstone, Rubin, & Willis, 2002). Out-group
members, compared to ingroup members, elicit less trust (Voci, 2006). Moreover, we
grant fewer resources (Tajfel, Billig, Bundy, & Flament, 1971) and offer less help to out-
group members (Levine, Cassidy, Brazier, & Reicher, 2002). Group membership may
thus be a key determinant of social exclusion dynamics.
Surprisingly, research has so far mainly considered the effects of group
membership on victims of social exclusion. This research has shown that social exclusion
leads to pain and distress, regardless of whether it was initiated by an in-group or out-
group member (Smith & Williams; Williams, Cheung, & Choi, 2000), or even a strongly
disliked out-group member (Gonsalkorale & Williams, 2007, although see Bernstein,
Sacco, Young, Hugenberg, & Cook, 2010; Goodwin, Williams, & Carter-Sowell, 2010;
Wirth & Williams, 2009, for a more nuanced perspective). Little research, however, has
addressed the effects of group membership on the process of exclusion itself. In one
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
5
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
exception, Vrijhof et al. (2016) examined responses to the social exclusion of in-group
and out-group members among adolescents. They found that adolescents generally
applied a strong inclusion norm; they actively tried to include both in-group and out-
group members, even though adolescents’ empathic concern was associated more with
the inclusion of in-group members than with the inclusion of out-group members. This
study provided a first step in examining the effects of group membership on the process
of exclusion, but with mixed results for adolescents’ motives versus actual behavior. It is
moreover still an open question how group membership affects reactions to social
exclusion among adults, or what the underlying (brain) mechanisms are for different
reactions to social exclusion. To address this hiatus, in the first three studies we
investigated people’s behavioral and brain responses to the exclusion of another
individual using a modified three-player Cyberball game (a computerized ball-tossing
game; Williams, Cheung, & Choi, 2000). Exclusion was programmed such that one
player (the excluder) threw the ball consistently to the participant at the cost of another
player (the excluded target). To manipulate group membership, we used a minimal group
paradigm, which creates groups based on arbitrary dimensions, thereby reducing any bias
from existing knowledge about specific social groups (Tajfel, 1970). In this minimal
group the excluder was always the in-group and the excluded target the out-group. By
manipulating group membership through self-selection (Study 1) as well as random
assignment (Studies 2 and 3), we moreover tested the robustness of its effects.
We chose the perspective of a group member that does not initiate the exclusion
but responds to the exclusion initiated by another group member, because bullying
research has shown that such facilitatory actions substantially reinforce the negative
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
6
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
experience of bullying (Espelage, Green, & Wasserman, 2007). Also, people who
observe bullying of out-group victims, compared to in-group victims, hold less negative
attitudes towards in-group aggressors (Nesdale, Killen, & Duffy, 2013), and less often
intervene (Palmer, Rutland, & Cameron, 2015). In the current research, we addressed
whether similar dynamics apply to social exclusion. We reasoned that group members
could react to social exclusion in three ways: 1) by going along with the exclusion of the
target (by reducing the number of throws to the excluded player), 2) by compensating and
actively including the target (increasing throws to the excluded player), or 3) by doing
neither (dividing tosses equally among the two players). Whereas the latter (on-the-fence)
option does not involve active exclusion, it could still be considered facilitatory since the
excluded player ends up receiving fewer balls than equal distribution norms propose.
We moreover predicted participants’ responses to the social exclusion of another
individual to be determined by the salience of the players’ group membership, something
we addressed by means of our minimal group manipulation. In the absence of such
information, we expected participants to compensate because of the strong inclusion
norm (Zadro & Gonsalkorale, 2014). However, when we make group membership
salient, and when an in-group member initiates the exclusion of an out-group member,
we expected that this could create a dilemma in participants between favoring the in-
group and avoiding the exclusion of the out-group target. That is, whereas participants
may feel it is normative to not exclude others (Wesselmann et al., 2013), people may at
the same time wish to reciprocate and favor the exclusionary behavior of the in-group
member (Gaertner & Insko, 2000). As a consequence, we expected participants to remain
unbiased and divide tosses equally among the two players.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
7
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
This motivational conflict is likely moderated by one’s identification with ones
in-group (Akerlof & Kranton, 2000), such that the stronger this identification, the more
participants are inclined to reciprocate an in-group excluder as opposed to compensating
an excluded out-group target. To examine this, we additionally assessed participants’ felt
connection with the in-group versus out-group player in the minimal group setting and
examined their relation with participants’ tossing behavior, with greater relative in-group
identification predicted to concur with reduced compensation.
In addition, in Study 3, we assessed to what extent cognitive conflict concurred
with the activation of the two opposing motives (i.e., to include the out-group versus to
favor the in-group) when participants witnessed social exclusion in a minimal group
setting. Because this conflict is not necessarily expressed in people’s ultimate choices for
exclusion versus inclusion, and because people are not always able to report on the
conflict they experience while making a decision, we used functional neuroimaging
(fMRI) to assess participants’ real-time brain indices of cognitive conflict during the
Cyberball game. A large body of neuroimaging literature points to the central role of the
dorsal anterior cingulate cortex (dACC) and the dorsolateral prefrontal cortex (dlPFC;
Botvinick, 2007; Van Veen & Carter, 2006) across tasks involving simple stimulus-
response rules (Van Veen & Carter, 2005), as well as more complex social dynamics, like
moral dilemmas (Greene et al., 2004), unethical behavior (Lelieveld, Shalvi, & Crone,
2016), and social rejection (Somerville, Heatherton, & Kelley, 2006). Accordingly, these
were our regions of interest to test the assertion that a minimal group setting induces
greater conflict resulting from the inclusion norm towards the out-group member
competing with the norm to favor the in-group member.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
8
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
In the first three studies, we used the Cyberball paradigm to examine people’s
responses to the social exclusion of others. In a final study (Study 4), we investigated
people’s responses to social exclusion with another paradigm, to see whether the findings
of the first three studies translate to a different group setting that involved a team
selection task. In this paradigm (adjusted from Doolaard, Lelieveld, Noordewier, Van
Beest, & Van Dijk, 2020) participants are instructed to perform an estimation task in a
team of four members. Following a practice round but before the actual game begins
participants are given the opportunity to adjust the composition of their team by
excluding one of the other three players. This paradigm enabled us to study social
exclusion in a larger group (i.e., a group of four). Moreover, in addition to the gradual act
of social exclusion in the first three studies, operationalized as the relative number of
throws to each other player, this paradigm allowed us to study how group membership
affects the binary decision to include or exclude a member of the team. Finally, in Study
4 we assessed different motives underlying participants’ responses to social exclusion
using self-report measures, to further examine whether people experienced conflict
between the norm for inclusion and ingroup-favoritism.
In four studies, involving behavioral as well as brain measures, we thus
investigated the effect of group membership on compensating versus facilitating social
exclusion of others, and the dilemma people may experience when deciding between
these two options. The first two studies involved a between-participants manipulation of
inclusionary status (inclusion vs. exclusion) and group membership (minimal group vs.
control) within an adjusted version of the Cyberball game. We thus examined how people
respond to the exclusion of an individual in the absence versus presence of a minimal
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
9
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
group setting. In the third neuroimaging study, we extended this set-up and used an
fMRI-compatible experimental design with inclusionary status as within-subjects factor
and group membership as between-subjects factor. This allowed us to study the brain
mechanisms underlying people’s reactions to social exclusion in the presence and
absence of a minimal group. In a final study, we examined participants’ responses to
social exclusion in a different group setting where people could adjust the composition of
a team, using a between-participant manipulation of the order of the exclusion decision
(initiating vs. responding to the decision) and group membership (minimal group vs.
control).
Study 1
Method
Design and participants. The study used a 2 (inclusionary status: inclusion vs.
exclusion) × 2 (group membership: minimal group vs. control) between-participants
design. Our sample size was determined based on a power analysis revealing that 128
participants were required to detect a medium effect size (Cohen’s d = .50) at the 5%
level with a power of .80. One hundred twenty-six undergraduates from Leiden
University (75 women, 51 men, Mage = 20.88, SDage = 2.31) eventually participated. They
were recruited from the faculties of humanities, medicine, law, and science, but not from
the faculty of social sciences, to ensure unfamiliarity with Cyberball. Participants were
randomly assigned to the four conditions. All materials and datasets used in our studies
are publicly available on the Open Science Framework, using the following link:
https://osf.io/39gxr/?view_only=57d747f50cad4dc7bc0c1caf770d1f67
Procedure
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
10
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Upon arrival at the laboratory, participants were led to a cubicle and received
further instructions via the computer screen. Participants played a three-player Cyberball
game (Williams et al., 2000); a computerized ball tossing game. Originally, the game was
programmed such that the participant was the exclusion target and would at some point
stop receiving the ball. We modified this set up, such that now one of the virtual players
was excluded by having the other player throw all balls to the participant (for a similar
version, see Riem, Bakermans-Kranenburg, Huffmeijer, & Van Ijzendoorn, 2013).
Participants were told they were to play a game of Cyberball with two others over
the Internet. They learned that they were Player C, and the others were Player A and B.
Participants were unaware that the behavior of the other players was preprogrammed. In
the inclusion condition, Player A was programmed to throw the ball 50% of the times to
the participant (Player C) and 50% of the times to Player B. In the exclusion condition,
Player A threw the ball 100% of the times to the participant, thereby fully excluding
Player B. In all conditions, Player B was programmed to at all times throw the ball 50%
of the time to the participant and 50% of the time to Player A, thus displaying no biased
behavior towards any of the other two players. The percentage of ball tosses from the
participant to Player B (the exclusion target) comprised our main dependent variable. We
calculated this as 100(NCB / NTotal), where NCB is the number of throws from the
participant to Player B, and NTotal is the total number of throws. Both games proceeded
for 45 throws in total (from all three players).
Before the game started, participants chose their avatar to be blue, yellow, or
green as a basis for our minimal group manipulation. In the minimal group condition,
Player A’s color was matched to the participant’s choice, whereas Player B was given a
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
11
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
different color. For instance, if the participant chose the color blue, Player A would also
be blue, but player B would be yellow (see Figure 1A). In the control condition where
group membership was absent, all players were assigned a different color. For instance, if
a participant chose the color blue, Player A would be green and Player B yellow (see
Figure 1B).
Following Cyberball, participants received a manipulation check of our minimal
group manipulation. Participants indicated for Players A and B separately how connected
they felt to them (i.e., “To what extent did you feel connected to Player A/B? ”, “I had a
lot in common with Player A/B”, and “Player A/B was a member of my group”), all on
seven point Likert scales (1 = not at all; 7 = very strongly). We averaged responses into a
single index of perceived group membership (αPlayer A = .80; αPlayer B = .78). See
supplemental material for other questions we asked, including results.
Participants next estimated the number of tosses Player B had received in both the
inclusion and exclusion games, as a manipulation check of our exclusion manipulation.
At the end of the session, participants were debriefed, were paid €1,50, and thanked for
their participation. All procedures were approved by the ethical committee of the Leiden
University Institute of Psychology.
Results
Manipulation checks.
Perceived group membership. A 2 (inclusionary status) × 2 (group membership)
Analysis of Variance (ANOVA) on the perceived group membership ratings of Player A
yielded only a main effect of group membership, F(1, 122) = 9.01, p = .003, partial η2 =
.07, indicating that independent of whether participants took part in an inclusion game or
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
12
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
an exclusion game, participants identified more with Player A when they both had the
same color (M = 4.36, SD = 1.50), rather than a different color (M = 3.54, SD = 1.53).
The 2 × 2 ANOVA of the perceived group membership ratings of Player B also only
revealed a main effect of group membership, F(1, 122) = 6.05, p = .015, partial η2 = .05,
indicating that when only Player B had a different color, participants identified less with
Player B (M = 3.25, SD = 1.28) than when all players had a different color (M = 3.86, SD
= 1.47). Together, these results confirm that our minimal group manipulation was
effective.
Perceived exclusion of target. A 2 × 2 ANOVA only showed a main effect of
inclusionary status, F(1, 122) = 46.60, p < .001, partial η2 = .28. Participants thought
Player B received fewer balls in the exclusion game (M = 12.19, SD = 3.07) than in the
inclusion game (M = 15.52, SD = 2.42), confirming the effectiveness of the exclusion
manipulation.
Ball tosses to target. A 2 × 2 ANOVA showed a significant main effect of
inclusionary status, F(1, 122) = 12.53, p = .001, partial η2 = .09, qualified by a significant
interaction, F(1, 122) = 6.35, p = .013, partial η2 = .05 (see Table 1 for means and
standard deviations). In the inclusion games, the number of tosses in the group
membership conditions did not differ, t(122) = -.68, p = .500, Cohen’s d = .20, 95% CI [-
.07 – .03], suggesting that group membership alone did not affect the number of throws to
the two other players. But in the exclusion games, group membership did influence
participants’ ball tossing behavior; they less frequently tossed the ball to the excluded
target in the minimal group condition than in the control condition, t(122) = 2.90, p =
.004, Cohen’s d = .64, 95% CI [.02 – .12], as predicted. Moreover, participants actively
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
13
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
compensated for the target’s exclusion in the control condition, with their percentage of
tosses towards the target significantly exceeding the even distribution of 50%, t(33) =
6.51, p < .001, Cohen’s d = 1.22, 95% CI [.07 – .14]. Such compensation was not
observed in the minimal group condition, t(29) = 1.42, p = .165, Cohen’s d = .23, 95% CI
[-.01 – .08].
Group identification and tosses to the target. To examine whether the strength
of identification with in-group Player A relative to out-group Player B was associated
with participants tossing behavior in the minimal group condition, we correlated
difference scores between the perceived group membership of Player A and Player B
with the percentage of ball tosses to Player B. For our hypotheses it was important to use
the difference score, instead of the identification with Player A and B separately, as it
allowed us to investigate the throwing behavior of participants who were the most
affected by our manipulation (and thus identified more with Player A, while at the same
time less with Player B). This correlation was significantly negative in the minimal group
condition (r = -.18, p = .049), indicating that the stronger the felt connection with the in-
group excluder relative to the out-group target, the fewer balls participants threw to the
excluded target. This correlation was non-significant in the control condition (r = -.09, p
= .463).
Discussion
Study 1 provided initial evidence that a simple minimal group manipulation made
participants compensate less for the exclusion of a target in Cyberball, whereas they
actively compensated for the exclusion in the absence of a minimal group setting. The
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
14
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
more participants identified with the excluder than the excluded target in a minimal group
setting, the less likely they were to compensate.
To replicate the results of Study 1, and to rule out that shared preferences drove
group membership perceptions (Festinger, 1957) due to the overlap of their color
preferences with those of another player, in a second study, the Cyberball players were
assigned a color, instead of choosing this themselves (see Dunham, Baron, & Carey,
2011), providing an even more stringent test of the notion that a minimal group setting
affects reactions to the social exclusion of others.
Study 2
Method
Design and participants. The study again involved a 2 (inclusionary status:
inclusion vs. exclusion) × 2 (group membership: minimal group vs. control) between-
participants design. Using similar selection criteria as in Study 1, we aimed for the
inclusion of 128 participants. One hundred twenty-two undergraduates from Leiden
University (84 women, 38 men, Mage = 20.63, SDage = 2.19) comprised the final sample
and were randomly assigned to the four conditions.
Procedure. The procedure was similar to Study 1, except that the players’ colors
were now preprogrammed by the experimenter, without any accompanying information.
As depicted in Figure 1, the participant (Player C) was always assigned the color blue and
Player B the color yellow. In the minimal group condition, Player A was assigned the
same color as the participant (i.e., blue) and a different color than both other players (i.e.,
green) in the control condition.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
15
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
After playing the Cyberball game participants responded to the same three items
from Study 1 measuring perceived group membership of Player A (α = .83) and player B
(α = .76), and their estimated number of throws to Player A and B.
Results
Manipulation checks.
Perceived group membership. A 2 (inclusionary status) × 2 (group membership)
ANOVA of perceived group membership ratings of Player A yielded only a main effect
of group membership, F(1, 118) = 6.31, p = .013, partial η2 = .05, indicating that
participants identified more with Player A when they had the same color (M = 4.13, SD =
1.42) than a different color (M = 3.48, SD = 1.42). A main effect of group membership on
the ratings of Player B, F(1, 118) = 8.03, p = .005, partial η2 = .06, further indicated that
participants identified less with Player B in the minimal group condition (M = 3.39, SD =
1.03) than in the control condition (M = 3.99, SD = 1.31), confirming the effectiveness of
our group membership manipulation.
Perceived exclusion of target. As expected, the 2 × 2 ANOVA showed a main
effect of inclusionary status, F(1, 118) = 11.64, p = .001, partial η2 = .09. Participants
estimated Player B to have received fewer balls in the exclusion game (M = 12.78, SD =
5.45) than in the inclusion game (M = 15.37, SD = 2.84). The ANOVA also yielded a
main effect of group membership, F(1, 118) = 6.20, p = .014, partial η2 = .05, indicating
that participants thought Player B received fewer balls in the minimal group condition (M
= 13.15, SD = 3.00) than in the control condition (M = 15.02, SD = 5.45) irrespective of
whether this was the in -or exclusion condition. Because this main effect was unexpected,
we conducted follow-up planned comparisons to examine the differences between the
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
16
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
specific conditions. These analyses showed that the difference between the minimal
group (M = 11.59, SD = 2.43) and control conditions (M = 13.90, SD = 7.09) was
significant in the exclusion condition, t(122) = 2.11, p = .037, Cohen’s d = .44, 95% CI
[.07 – 2.25], but not significant in the inclusion condition (Mminimal group = 14.61, SDminimal
group = 2.76 vs. Mcontrol = 16.13, SDcontrol = 2.75; t(122) = 1.40, p = .163, Cohen’s d = .44,
95% CI [-.31 – 1.83]). It is not surprising that in the exclusion condition participants
indicated that they thought Player B received fewer balls in the minimal group condition,
because this was what actually happened (see results for the percentage of ball tosses to
player B). Although this pattern was not observed in Study 1, these follow-up
comparisons thus confirmed that the manipulation of inclusionary status was successful.
Ball tosses to target. The 2 × 2 ANOVA showed a significant main effect of
inclusionary status, F(1, 118) = 11.05, p = .001, partial η2 = .09, qualified by a significant
interaction, F(1, 118) = 4.51, p = .036, partial η2 = .04 (see Table 2 for means and
standard deviations). In the inclusion condition, group membership had no effect on
tossing behavior, t(122) = -.65, p = .519, Cohen’s d = .17, 95% CI [-.03 – .01]. In the
exclusion condition, however, the percentage of participants’ ball tosses to Player B was
lower in the minimal group setting than in the control condition, t(122) = 2.34, p = .021,
Cohen’s d = .57, 95% CI [.004 – .05]. Mimicking Study 1’s findings, participants
actively compensated for Player B’s exclusion in the control condition, with their
percentage of tosses exceeding 50% significantly, t(30) = 12.20, p < .001, Cohen’s d =
2.50, 95% CI [.08 – .11], whereas such compensation was only marginally observed in
the minimal group condition, t(28) = 1.98, p = .058, Cohen’s d = .42, 95% CI [-.002 –
.092].
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
17
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Group identification and tosses to the target. We again correlated difference
scores between the perceived group membership of Player A compared to Player B with
the ball tosses to Player B. This correlation was significantly negative in the minimal
group condition, (r = -.21, p = .038), but non-significant in the control condition (r = -.15,
p = .223), thereby replicating Study 1.
Discussion
In Study 2 a minimal group setting was created by automatically assigning
participants a color without any accompanying information rather than having
participants choose the color themselves, as was done in Study 1. In line with the results
from Study 1, this minimal group setting again caused participants to throw fewer balls to
an out-group player to the benefit of the in-group player. The correlational results,
mimicking those of Study 1, revealed how greater identification with the in-group
excluder was again associated with a decrease in throws to the excluded out-group target.
Together, this pattern of results across two studies point to the possibility of a
motivational conflict that participants might have experienced between including the out-
group member on the one hand and favoring the in-group member on the other hand.
Still, the reliance on self-report measures and the fact that these assessments were made
only after the Cyberball game preclude strong conclusions about the occurrence of such
conflict. In a third neuroimaging study we therefore further investigated whether
cognitive conflict arose during participants’ decisions to go along with or compensate for
social exclusion in a minimal group setting.
Study 3
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
18
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Using a similar set-up as in the previous studies, we measured people’s brain
activity using fMRI while they played the Cyberball game involving or not the social
exclusion of another player when group membership was salient or not. Because our aim
was to establish whether motivational conflict would occur when participants had to
respond to social exclusion in an intergroup setting, we focused our fMRI analyses
specifically on the roles of the dACC and dlPFC, as these regions have reliably been
shown to be associated with cognitive conflict (e.g., Van Veen & Carter, 2006).
Method
Participants. Our sample was determined at a minimum of N = 40, based on
recent neuroimaging studies investigating the neural mechanisms underlying social
exclusion with a similar experimental set-up (Van der Meulen et al., 2016, 2017). The
final sample consisted of 45 healthy right-handed paid volunteers, who were all students
from Leiden University. Due to a technical error during scanning the data from two
participants were lost. We therefore analyzed the data from 43 participants (25 female, 18
male; Mage = 20.95, SDage = 1.86; age range 18-25). None reported to have any history of
neurological or psychiatric disorder and all were medication-free. All participants gave
written informed consent for the study, and all procedures were approved by the medical
ethical committee of the Leiden University Medical Center (LUMC).
Design. We used an fMRI-compatible experimental design with one within-
subjects factor with two levels (inclusionary status: inclusion vs. exclusion) and one
between-subjects factor with two levels (group membership: minimal group vs. control).
Participants were randomly assigned to the two group membership conditions. We only
manipulated group membership in the exclusion game, not in the inclusion game, to
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
19
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
avoid habituation to the minimal group manipulation throughout the two subsequent
games. In the inclusion game, the three players did not have a color. Our design was
therefore not fully factorial. It however still allowed us to compare the effects of group
membership (minimal group vs. control) during the exclusion game, and the effects of
inclusionary status (inclusion vs. exclusion) within the minimal group and control
conditions separately.
Procedure. After participants were welcomed and placed in the fMRI scanner,
they received instructions about Cyberball. Participants next played two consecutive
Cyberball games. In the first game, both other players equally included the other player
and the participant. This game was similar to the inclusion game of Studies 1 and 2, but
the three players were not colored as to avoid habituation to our minimal group
manipulation. As in Studies 1 and 2, we assessed participants’ ball tosses to the target
during the game. After the game ended, participants reported the perceived exclusion of
the target, by indicating with their left and right index finger whether they thought the
target received more or fewer than 15 ball tosses (i.e., one third of the total number of
ball tosses) from the other player.
Next, participants played the second Cyberball game, which was explained to be
with different people from the first game. Now, one player was consistently excluded by
the other player (as in the exclusion games of Study 1 and 2) in either a minimal group
setting or control condition, similar to Studies 1 and 2 (see Figure 1A and 1B). We
counterbalanced whether the excluded player was Player A or B, to make sure that
inclusion vs. exclusion behavior was not restricted to the left vs. right visual field. We
again measured participants’ tossing behavior. Following the game, participants again
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
20
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
indicated whether they thought the target received more or fewer than 15 ball tosses from
the other players. At the end, all participants were asked what they thought the study was
about. None of the participants guessed the true purpose of the study, or reported any
doubts about the cover story.
fMRI Data Acquisition. Scanning was performed on a 3.0T Philips Achieva
scanner at the Leiden University Medical Center. Functional data were acquired using a
T2*-weighted echo-planar imaging (EPI) sequence (echo time/TE = 30 ms, repetition
time/TR = 2200 ms, slice-matrix = 80 × 80, slice-thickness = 2.75 mm, slice gap =
0.28mm gap, field of view [FOV] = 220 mm), during two fMRI runs which lasted for
approximately 5 minutes each. At the end of the scan session, a high-resolution T2-
weighted high-resolution anatomical scan (same slice prescription as EPI) was collected.
fMRI Data Analysis. Data pre-processing and analyses were conducted with
SPM8 software (http://www.fil.ion.ucl.ac.uk/spm/software/spm8) implemented in
MATLAB (Mathworks, Sherborn, MA). All functional images were realigned and slice-
time corrected using the middle slice as reference. They were spatially normalized to T1
templates and spatially smoothed with a Gaussian kernel (8mm, full-width at half-
maximum). For motion, we used a cutoff point of 3mm. None of the participants
exceeded this threshold. A canonical haemodynamic response function (HRF) was
convolved at the onset of the ball tosses.
Analyses were carried out using the general linear model in SPM8. Whereas
previous fMRI research on targets of exclusion mostly focused on receiving vs. not
receiving the ball, we focused on the brain mechanisms underlying throwing behavior.
We focused on throwing behavior, regardless of whether this was to the excluder or the
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
21
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
excluded target, because we were interested in brain mechanisms underlying the decision
to throw to either one of players. We compared brain activity during these events in the
exclusion game (i.e., ExclusionThrow) to the inclusion game (i.e., InclusionThrow),
resulting in the ExclusionThrow > InclusionThrow contrast and vice versa. For these
contrasts, we subsequently examined the moderating role of group membership by
comparing the minimal group condition to the control condition. Although we were less
interested in the brain mechanisms involved in the traditionally investigated participant
perspective of receiving vs. not receiving the ball, we nonetheless also compared the
brain regions involved in these events separately during the inclusion game (i.e.,
InclusionGet versus InclusionOut) and exclusion game (i.e., ExclusionGet versus
ExclusionOut), as depicted in Table 3.
We computed contrast parameter images for each participant and submitted them
to second-level group analyses. At the group level, we computed whole-brain contrasts
between conditions by performing one-sample t-tests, treating participants as a random
effect. We further performed two-sample t-tests to investigate the moderating role of
group membership. Results were considered significant at an uncorrected threshold p <
.001 with an extent threshold of ten continuous voxels. Thresholds were based on
recommendations from Lieberman & Cunningham (2009), to produce a desirable balance
between Type I and Type II errors. Table 3 reports which results remained significant
with an FDR p < .05 or FWE p < .05, > 10 contiguous voxels threshold.
We extracted parameter estimates from the regions that were identified in the
whole-brain analyses using the MARSBAR toolbox for SPM8 (Brett, Anton, Valabreque,
& Poline, 2002), to further visualize patterns of activity.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
22
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Results
Behavioral results
Perceived exclusion of target. The logging of one participant’s estimations failed
due to a technical error, leaving 42 participants for this analysis. A Chi-square test of
their estimations showed a significant effect of inclusionary status, χ2 (1, N = 84) = 14.42,
p < .001, φ = .41. Participants more often estimated targets to have received fewer than
15 ball tosses in the exclusion game (34 out of 42: 81.0%) than in the inclusion game (17
out of 42: 40.5%). Within the exclusion games, group membership did not further affect
these estimations, χ2 (1, N = 42) = .21, p = .706, φ = .07.
Participants’ ball tosses to target. Planned comparisons of the number of ball
tosses during the two exclusion games showed a significant difference between the
minimal group and control condition, t(86) = 2.46, p = .016, Cohen’s d = .68, 95% CI
[.55 – 14.67]. Participants’ tosses to the exclusion target were more frequent in the
control condition than in the minimal group condition, similar to Studies 1 and 2. In the
control condition, participants actively compensated for the target’s exclusion, as their
percentage of tosses towards the excluded target significantly exceeded an even 50%
distribution, t(18) = 7.50, p < .001, Cohen’s d = 1.67, 95% CI [.11 – .19]. Unlike in
Studies 1 and 2, participants in the minimal group condition also tossed the ball to the
exclusion target significantly more than 50%, t(23) = 2.66, p = .014, Cohen’s d = .54,
95% CI [.02 – .13].
fMRI results
Responses to exclusion. The ExclusionThrow > InclusionThrow contrast
revealed activation in the dlPFC, which is depicted in Figure 2A and Table 3. We also
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
23
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
displayed ROI patterns for this dlPFC activation across different conditions, as depicted
in Figure 2B. A paired t-test of the parameter estimates revealed that in the exclusion
condition (where we manipulated group membership) dlPFC activation was greater for
participants in the minimal group condition compared to the control condition, t(86) =
2.04, p = .045, Cohen’s d = .50, 95% CI [.002 – .20]. To further examine the effects of
group membership, we conducted a two-sample t-test comparing ExclusionThrow >
InclusionThrow for the Minimal Group > Control contrast. This revealed significantly
greater activation in the dlPFC in the minimal group condition compared to the control
condition (see Figure 3, Table 3 for all relevant statistics). The reverse contrasts (Control
> Minimal group and InclusionThrow > ExclusionThrow) did not reveal any significant
activation. None of the whole brain contrasts revealed increased activation in the dACC.
Brain-behavior correlations. To investigate whether the activation in the dlPFC
was correlated to participants’ throwing behavior in the Cyberball game, we extracted
parameter estimates of our dlPFC region in the ExclusionThrow-InclusionThrow contrast
and correlated these with participant’s ball tosses to the target across all conditions. As
Figure 2C shows, this correlation was significantly positive, (r = .21, p = .048), indicating
that the stronger the dlPFC activity, the more frequently participants threw the ball to the
excluded target.
Discussion
Consistent with those of Studies 1 and 2, the findings of Study 3 showed that,
participants actively included an excluded target in the absence of a minimal group
setting. They more often chose not to compensate in a minimal group setting, where an
in-group member excluded an out-group target. The fMRI results revealed increased
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
24
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
activation in the dlPFC during participants’ tossing behavior in the exclusion game
compared to the inclusion game, suggesting that overall, they experienced greater conflict
when a fellow player was being excluded. Importantly, however, this activation was
stronger in the presence than in the absence of a minimal group setting. This suggests that
participants’ throwing decisions while witnessing social exclusion employed greater
cognitive control when the exclusion was initiated by an in-group member and the target
was an outgroup member than when group membership was not made salient. This
occurred perhaps to resolve the conflict between two opposing motives, namely, to
include others, and to favor the ingroup. The results further showed that dlPFC activation
correlated positively to inclusion of the target across exclusion conditions. The stronger
the dlPFC activity, the more frequently participants threw the ball to the excluded target,
suggesting that cognitive control occurred primarily when participants decided to include
the target (rather than reciprocate the excluder), further strengthening our conflicting
motives account.
Taken together, the results of these three studies provided converging evidence
that differences in group membership influence responses to social exclusion of another
individual. Still, the use of the modified three-player Cyberball game also created some
limitations. In all three studies, the exclusion of the target was directly dependent on the
inclusion of the excluder (and vice versa). That is, participants excluded one player by
throwing the ball to the other player, who was therefore automatically included in the
game. With such a design, it is thus impossible to dissociate people’s exclusion from their
inclusion decisions. To address this, we therefore conducted a final study, where we
employed a team selection paradigm without such a direct relation between the inclusion
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
25
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
of one person and the exclusion of another person. Moreover, to further examine to what
extent our observed pattern of findings could be explained by perceived group
membership versus reciprocity norms, in Study 4 we directly measured both constructs as
potential motivations for participants’ responses to the social exclusion of others.
Whereas the reciprocity norm did not seem to motivate participants’ throwing decisions
in the absence of a minimal group setting in the previous three studies, these measures
would allow us to obtain more direct evidence that participants’ responses to social
exclusion within a minimal group setting are primarily motivated by their concerns over
group membership. In addition, testing our ideas further with a new task allowed us to
extend our findings to a different group context, namely team selection.
Study 4
In Study 4, we used a task where participants could adjust the composition of a
team by excluding another team member. We adjusted the paradigm from Doolaard et al.
(2020). In their paradigm participants completed a competitive group task as a group in
which the goal was to estimate which of two dot clouds contain the most dots. Based on
the performance of the fellow team members, participants could decide to exclude team
members. In our version of the task, we did not give participants feedback on how they
performed on the task. Participants first played a practice round and before the actual
game started participants could choose to adjust the composition of the team by
excluding a potential player. We manipulated the group membership of the players in a
similar way as in the first three studies, by assigning different colors to the different
players to create minimal groups.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
26
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
In the first three studies, we showed that differences in group membership led
participants to throw fewer balls to the excluded target, but only when another player
initiated the exclusion. When the other player did not initiate the exclusion, differences in
group membership did not lead participants to throw fewer balls to the target. In the
current study, we aimed to extend these findings. To do so, we varied the decision order
in which participants could choose to exclude another player or not. One half of the
participants were the first in the team to make this decision (the initiate condition).
Participants in this condition could thus initiate the exclusion of another player. Based on
the findings from the first three studies, we expected that differences in group
membership would not lead participants to initiate the exclusion of an out-group player
more than an in-group player. The other half of the participants only made the decision to
exclude another player after two other team members had already made their decision
(the respond condition). In this condition, these other two players always initiated the
exclusion of a fourth player, who was to make their choice following the participant.
Participants in this condition thus responded to the exclusion of one player that was
initiated by two other players. Based on the first three studies, we expected that group
membership would influence participants’ exclusion decisions in this condition, such that
participants would more often decide to go along with the exclusion initiated by the two
other players of an out-group rather than an in-group target. We preregistered the study’s
experimental set-up and main hypotheses at https://osf.io/529zf.
Method
Design and participants. The study used a 2 (decision order: initiate vs. respond)
× 2 (group membership: minimal group vs. control) between-participants design.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
27
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Previous work using the same paradigm included 40 participants per cell, based on a power
analysis that indicated a significant difference with an alpha level of .05, a power of β =
.80, and an effect size of φ = 0.31 (Doolaard et al., 2020). Because we incorporated an
additional manipulation we decided to increase the sample size and aimed to collect 50
participants per cell. Due to a logging error the data from three participants were lost,
leaving the data of one hundred ninety-seven participants (130 females, Mage = 37.71, SD
= 12.46) for our analyses. Participants were recruited through the online research
platform Prolific Academic (https://www.prolific.ac/). To make sure the participants
understood our task, we selected only people from the United Kingdom and who were
native English speakers. All procedures were approved by the ethical committee of the
Institute of Psychology of Leiden University.
Procedure. After giving informed consent, participants were explained that the
experiment consisted of a computerized group task, in which participants allegedly
formed a team with three other participants. In reality the participants completed the task
alone, and the responses of their team members were programmed beforehand. Before
starting the actual task, participants learned that each player would be represented by an
avatar of a specific color. Depending on the position of their first initial in the alphabet,
this would be one out of five colors. After filling out their first initial, participants were
presented with their avatar and learned that their avatar was assigned the color orange.
Participants were then informed that together with their team members they were
going to perform a task in which each participant had to indicate as fast and accurately as
possible which of two pictures (see Figure 4) contained the most dots (a procedure
similar to the dot-estimation task, see Gerard & Hoyt, 1974). They learned that they
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
28
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
played this game against another team and that the team with the highest average team
score would win. Participants learned that all team members would first play a practice
round of 10 trials. In each trial participants had ten seconds to make their decision, after
which the next trial was presented. In between trials participants did not receive any
feedback about whether they correctly answered the trial or not. Even though participants
could not interact with the other members of the group (to avoid that differences in
content and valence of the interaction would influence their decisions), we emphasized
that like them the other members of their group also completed the practice round. This
was done to create the feeling that participants were really part of a team with which they
competed against another group. After the practice round, participants did not receive any
feedback on their team members’ or own performance, to make sure performance did not
influence their decisions.
Participants were then told that each team member was asked to indicate with
whom they wanted to be in the team in the next round. They were informed that they
could choose to be in a team with three or four players. Participants were informed that
not the absolute, but the average team score achieved in the game would determine
whether they would win, and so there was no advantage of choosing to play with four
over three team members. Participants then saw a picture with four avatars depicting
themselves and their three team members (Figure 5A-D). Similar to the first three studies,
in the minimal group condition two team members were assigned the same color as the
participant, and one team member was assigned a different color (see Figure 5A and C).
In the control condition, all four team members had a different color (see Figure 5B and
D). Depending on the decision order, participants were then told when they could choose
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
29
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
with which players they wanted to be in the team. Participants could either initiate the
exclusion of one of the other players (initiate condition) or they could respond to the
exclusion of a player initiated by the others (respond condition). In the initiate condition
(see Figure 5A and B), participants learned they were Player 1 and were the first to
choose their team members, after which Players 2, 3, and 4 would take their turns. In the
respond condition (see Figure 5C and D), participants learned they were Player 3 and that
they could choose their team members after (seeing the choice of) Players 1 and 2, and
before Player 4. Players 1 and 2 would thus make their selection first, and always
excluded Player 4.
Participants in all conditions were instructed to click once on a player’s icon if
they wanted to select that player for the team, and twice on the icon of a player if they did
not want to have that player in the team for the game. They learned that players who were
excluded would receive the message that they had not been chosen to be part of the team,
and that these players would continue with a different task. This information was added,
so that being excluded would not be perceived as an advantage (i.e. finish the experiment
early). After making their selection to in/exclude, participants answered a questionnaire
where they had to indicate on a 7-point scale (1 = absolutely not, 7 = absolutely) to what
extent they agreed with statements about 1) the conflict they experienced, 2) the extent to
which reciprocity and group membership motivated their choice, and 3) how aversive
they were to exclusion.
We measured conflict with two statements (“I felt torn when deciding on the team
composition” and “I experienced conflict when selecting the team players”). Responses
were averaged into a single index of conflict (α = .91). We measured group membership
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
30
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
as a motive for participants’ choice with two statements (“My decision to select a player
for the team was based on whether or not the player belonged to my group” and “My
decision to exclude a player from the team was based on whether or not the player
belonged to my group”). Responses were averaged into a single index of group
membership (α = .81).
Reciprocity was measured only in the respond conditions, because there was no
behavior to reciprocate in the initiate condition. We measured reciprocity by asking
participants to indicate to what extent they agreed with two statements (“I did what I did
because I was thankful the players before me chose me in their team” and “I based my
decision on whether or not the players before me chose me in their team”). Responses
were averaged into a single index of reciprocity (α = .81).
We measured exclusion aversion with two statements (“I did not like excluding
one of the players from the team” and “I found it difficult to exclude one of the players
from the team”). Responses were averaged into a single index of exclusion aversion (α =
.79). Finally, to check the manipulation of group membership we asked participants to
what extent they agreed with the statement “Player X was a member of my group”, with
X being Player 2, 3, or 4 in turn in the initiate condition and Player 1, 2, or 4 in the
respond condition. After answering this question about each of the other players,
participants were thoroughly debriefed and were paid £0.88.
Results
Manipulation check. To establish that our minimal group manipulation had been
successful, the manipulation check question to what extent “Player X was a member of
my group”, had to show differences between the group membership conditions, in
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
31
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
particular for Player 4. A 2 × 2 ANOVA on the ratings of Player 4 yielded only a main
effect of group membership, F(1, 193) = 23.57, p < .001, partial η2 = .11, but not of
decision order, F(1, 193) = 1.41, p = .237, partial η2 = .01, and no interaction, F(1, 193) =
.65, p = .420, partial η2 = .00. This indicates that irrespective of whether participants were
first or followed in their team composition decision, Player 4 was considered to be less of
a group member in the minimal group condition (M = 3.52, SD = 2.35) than in the control
condition (M = 5.10, SD = 2.24), thus confirming that our group membership
manipulation was successful.
We also examined participants’ responses to the manipulation check for the other
two players. Note that in the minimal group condition and the control condition
participants were informed that the other three players – which besides Player 4 also
included Player 2 and 3 (in the initiate condition) or Player 1 and 2 (in the respond
condition) - were part of their team. It would therefore make sense that participants
would overall give high ratings to this question. Indeed, although the means were higher
in the minimal group condition (M = 6.05, SD = 1.61 and M = 6.00, SD = 1.51) than in
the control condition (M = 5.65, SD = 1.97 and M = 5.82, SD = 1.56), 2 (group
membership) × 2 (decision order) ANOVAs did not yield any main or interactions effects
on these ratings, Fs < 2.49, ps > .12.
Exclusion behavior. To examine exclusion behavior across conditions, we first
conducted a logistic regression analysis with group membership (minimal group vs.
control) and decision order (initiate vs. respond) as independent variables and
participants’ exclusion (yes/no) of the target (Player 4) as the dependent variable. This
analysis yielded main effects of group membership, Wald’s χ2 (1, N = 197) = 13.67, p <
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
32
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
.001, and of decision order, Wald’s χ2 (1, N = 197) = 13.67, p < .001. The interaction was
not significant, Wald’s χ2 (1, N = 197) = .44, p = .51. We also analyzed our results with
another widely used method to study interaction effects with a dichotomous dependent
variable: the linear probability model (Wooldridge, 2013). A linear probability model is a
special case of a binomial regression model, where the probability of observing an event
or not (in this case whether participants excluded or not) is treated as depending on one or
more explanatory variables. For a detailed analysis of the difference between the linear
probability model and binary logistic models see Hellevik (2009). When analyzing our
results with the linear probability model, we do find a significant interaction effect.
Results show a significant main effect of decision order (β = -.27, p < .001) and of group
membership (β = -.27, p < .001), as well as a significant interaction effect (β = .17, p =
.01).
We then performed follow-up Chi-square tests to investigate differences between
specific conditions. Because the logistic regression interaction effect was not significant,
we used a Bonferroni correction and divided p = .05 by the number of Chi-square tests
we performed (i.e., 6). The follow-up Chi-square tests were thus considered significant
when p < .008. In line with our hypotheses, these results showed that participants who
responded to the exclusion more often chose to exclude the target in the minimal group
condition (22 out of 48: 46.8%) than participants in the control condition (6 out of 50:
12.0%), χ2 (1, N = 197) = 13.74, p < .001, φ = .-37, as well as compared to participants in
the minimal group condition who initiated the choice (6 out of 50: 12.0%), χ2 (1, N =
197) = 13.74, p < .001, φ = -.37, and participants in the control condition who initiated
the choice (2 out of 49: 4.1%), χ2 (1, N = 197) = 22.70, p < .001, φ = -.48. Chi-square
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
33
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
tests between the other three conditions did not yield any significant differences (ps >
.27).2
Decision conflict. A 2 × 2 ANOVA of the conflict ratings yielded a main effect of
group membership, F(1, 193) = 12.70, p < .001, partial η2 = .06, and of decision order,
F(1, 193) = 63.42, p < .001, partial η2 = .25. These main effects were qualified by a
significant interaction, F(1, 193) = 9.27, p = .003, partial η2 = .05. Planned comparisons
showed that participants in the minimal group condition who responded to the exclusion
experienced more conflict (M = 4.68, SD = 1.44) than participants in the control
condition who responded to the exclusion (M = 3.16, SD = 1.99, t(193) = 4.66, p < .001,
d = .88, 95% CI [-2.16 – -.88]), who in turn experienced more conflict than participants
who initiated the choice in the minimal group condition (M = 2.15, SD = 1.37, t(193) =
3.14, p = .002, d = 1.80, 95% CI [-1.67 – -.35]), and in the control condition (M = 2.03,
SD = 1.57, t(193) = 3.49, p = .001, d = .63, 95% CI [-1.79 – -.47]). Participants in the
minimal group condition who initiated the choice and participants in the control condition
who initiated the choice did not differ significantly in the level of experienced conflict,
t(193) = .37, p = .713, d = .08, 95% CI [-.52 – .76]).
Group membership motive. A 2 × 2 ANOVA of the group membership ratings
yielded a main effect of group membership, F(1, 193) = 4.43, p = .037, partial η2 = .02,
and of decision order, F(1, 193) = 13.38, p < .001, partial η2 = .07. These main effects
were qualified by a significant interaction, F(1, 193) = 9.62, p = .002, partial η2 = .05.
Planned comparisons confirmed our predictions that participants who responded to the
exclusion indicated that the group membership of the players motivated their team
selection decision more in the minimal group condition (M = 3.74, SD = 1.87) than
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
34
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
participants in the control condition (M = 2.50, SD = 1.51, t(193) = 3.67, p < .001, d =
.73, 95% CI [-1.91 – -.57]), or participants who initiated the choice in the minimal group
condition (M = 2.13, SD = 1.61, t(193) = 4.77, p < .001, d = .92, 95% CI [-2.28 – -.94]),
or the control condition (M = 2.37, SD = 1.68, t(193) = 4.04, p < .001, d = .77, 95% CI [-
2.04 – -.70]). The other conditions again did not differ from one another (ps > .69).
Reciprocity motive. Planned comparisons of the reciprocity ratings yielded no
significant effect of group membership, t(96) = .15, p = .882, d = .03, 95% CI [-.53 –
.46], confirming that the motive to reciprocate the other players did not vary across the
minimal group (M = 2.57, SD = 1.27) and control conditions (M = 2.61, SD = 1.20).
Exclusion aversion. A 2 × 2 ANOVA of participants’ exclusion aversion ratings
yielded no main effects of group membership, F(1, 193) = .06, p = .814, partial η2 = .00,
or decision order, F(1, 193) = 1.29, p = .257, partial η2 = .01, and no interaction, F(1,
193) = .36, p = .550, partial η2 = .00. Across conditions participants indicated to be
relatively exclusion averse, with an overall mean score that was above the midpoint of
the seven-point scale (M = 4.69, SD = 1.94).
Mediated moderation. We explored whether the interaction effect of group
membership and decision order on participants’ exclusion behavior would be mediated
by experienced conflict and/or by group membership motives. To examine this, we
performed a mediated moderation analysis (Muller, Judd, & Yzerbyt, 2005). To test this,
we used Hayes’s (2018) PROCESS bootstrapping command with 10,000 iterations
(model 8) to test the indirect effect (Preacher, Rucker, & Hayes, 2007) of the interaction
term of group membership and decision order on exclusion behavior through experienced
conflict and/or group membership motives (controlling for the unique effects of group
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
35
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
membership and decision order). The analysis revealed a significant indirect effect of
group membership motives on exclusion behavior (the 95% CI did not contain zero, a × b
= .66, SE = .32, 95% CI [.20 – 1.46]), but not of experienced conflict (the 95% CI did
contain zero, a × b = -.13, SE = .27, 95% CI [-.73 – .37]). The findings thus suggest that
group membership motives can explain the effects of group membership on participants’
decisions in response to the exclusion by the other players. However, because of the
exploratory nature of this analysis, these findings should be interpreted with caution.
Discussion
Employing a different paradigm, within a different group setting, the findings of
Study 4 further supported the results of our first three studies that a simple minimal group
manipulation made participants compensate less for the exclusion of an out-group target.
When deciding on whom to select for a team, participants more often decided to go along
with the exclusion by another player in the presence than in the absence of a minimal
group. When participants initiated the decision, they excluded less often, regardless of the
presence or absence of a minimal group. We investigated several motives for
participant’s choices. Self-report ratings of experienced conflict showed that responding
to the exclusion of an out-group member that was initiated by an in-group member
increased the experience of conflict, converging with the fMRI results of Study 3. In line
with these findings, the results showed that even when participants decided to go along
with the exclusion of an out-group target, they still indicated (across all conditions) to be
exclusion averse. Moreover, when participants responded to the exclusion of an out-
group player, they also indicated that their decision to exclude or not was based on
whether or not the player belonged to their group. Finally, reciprocity motives did not
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
36
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
play a role in participants’ responses to social exclusion. Mediated moderation analyses
showed that although experienced conflict was higher when participants excluded more,
conflict did not predict exclusion behavior significantly. Therefore, though conflict
occurs, it may not be the essential mechanism that drives the exclusion behavior. Instead,
the analysis showed that participants’ exclusion decisions were motivated by the group
membership of the players.
General Discussion
In three behavioral studies and one behavioral fMRI study using different
experimental paradigms we investigated the effect of group membership on participants’
responses to the social exclusion of others, by varying the absence or presence of a
minimal group setting. In the first three studies, we employed a modified version of the
three-player Cyberball game and examined participants’ ball tosses to an excluded target
in the absence or presence of a minimal group setting. In these studies participants
actively included an excluded target in the absence of a minimal group setting (i.e.
increased the number of tosses towards), but chose not to intervene when an in-group
member excluded an out-group target (i.e. distributing their tosses more or less evenly).
Although participants did not fully exclude the out-group target in this case, the result of
their indecisive behavior was that, compared to the other players, the target received
significantly fewer balls. Correlation results from Studies 1 and 2 moreover showed that
the more participants identified with the excluder than the excluded target in a minimal
group setting, the less frequently they compensated by again throwing the ball to the
excluded target, which suggests that participants experienced a motivational conflict
between favoring the in-group and avoiding the exclusion of the out-group target.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
37
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Note though that in the three-player Cyberball setting of our first three studies,
throwing the ball to the excluder automatically ruled out a throw to the excluded player,
making it difficult to disentangle inclusion of the in-group member from exclusion of the
out-group member. In a fourth study, we therefore employed a different paradigm that
allowed us to dissociate these responses. In this paradigm, participants could adjust the
composition of a team by in- versus excluding players from an initial group of four, while
we again manipulated group membership through the absence or presence of a minimal
group setting. The results of this study showed that whereas participants were exclusion
averse in the absence of a minimal group setting, they decided to actively exclude out-
group targets when this was initiated by an in-group member. Mediated moderation
analyses showed that group membership motives accounted for this decision to go along
with the exclusion. In addition, Study 4 allowed us to replicate the Cyberball findings of
the first three studies in a different group setting, namely team selection.
In addition to self-reports and behavioral measures, our third study also employed
neuroimaging that allowed us to assess through more implicit measures to what extent
participants experienced conflicting motives while deciding to exclude or not. These
neuroimaging findings revealed that during the exclusion game compared to the inclusion
game activation increased in the dlPFC, a brain region widely associated with the
resolution of cognitive conflict (Van Veen & Carter, 2006). Importantly, this relative
activation was even stronger in the presence than in the absence of a minimal group
setting, suggesting that participants’ throwing decisions following exclusion concurred
with greater cognitive control in response to conflict when the exclusion was initiated by
an in-group member than in the absence of a minimal group setting. Further analyses
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
38
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
moreover revealed that dlPFC activation was positively correlated to compensation. The
stronger the dlPFC activity, the more frequently the participants threw the ball to the
excluded target, suggesting that cognitive conflict was primarily present when
participants decided to override the tendency to reciprocate the excluder and instead
again include the exclusion target. Finally, the self-report measures in Study 4 supported
the fMRI findings, showing that participants experienced more conflict when they
responded to the exclusion of a target in the presence than in the absence of a minimal
group manipulation. Moreover, when participants decided to go along with the exclusion
of an out-group target, they still indicated to be exclusion averse, suggesting that different
motives have affected participant’s decisions.
Although together these findings suggest that participants experienced conflict
when they responded to the exclusion of an out-group target initiated by an in-group
member, the mediation analysis in Study 4 showed that this self-reported conflict was not
associated with their exclusion decisions. Note though that participants experienced self-
reported conflict only after the team-selection task had already been completed. Perhaps
then, through their decision, they had already resolved this conflict, irrespective of
whether this involved going along with the exclusion of the out-group target, or not.
Experienced conflict may therefore not have affected participants’ exclusion decisions.
Instead, our mediation analysis showed that whether or not participants went along with
the exclusion of an out-group target could be explained by group membership motives.
That is, participants’ decision to exclude a player or not were based on whether or not the
player belonged to their group.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
39
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Our neuroimaging findings from Study 3 showed no association of participants’
responses to social exclusion with the ACC, even though this was one of our regions of
interest. Previous work on the involvement of the ACC and dlPFC in conflict adaptation
suggests that the ACC is primarily associated with conflict monitoring, whereas the
dlPFC is more involved with conflict resolution (Kerns, 2006; Kerns et al., 2004; Smith
et al., 2019). More recent work using lesion patients showed that while the dlPFC plays a
fundamental role in behavioral adaptation in response to conflict, the ACC is sensitive to
the level of conflict, but is not crucial for handling conflict (Boschin, Brkic, Simons, &
Buckley, 2017). Along those lines, the dlPFC may have guided participants’ ball tosses
more than the ACC. At this point, the above interpretation is still speculative in nature
and future research is required to further establish the role of the dlPFC in participant’s
responses to exclusion in our studies.
Limitations and Implications
A limitation of our first three studies was that participants were faced with a
dilemma where the inclusion of one player was directly linked to the exclusion of the
other player. That is, when participants decided to throw the ball to the out-group target,
they at the same time excluded their in-group member, and vice versa. In daily life, many
decisions involve such dilemmas (e.g., whom to work with on an assignment, whom to
pass the ball to in a game of soccer, whom to talk to at a party). We were therefore
specifically interested in how people deal with this tension between the inclusion of one
person and at the same time the exclusion of another person, and how group membership
affects this decision-making process. However, to disentangle these decisions, we also
replicated our findings with a different paradigm, where participants responded to social
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
40
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
exclusion in a situation without a direct relation between the exclusion of one player and
the inclusion of another.
Another limitation is the focus on social exclusion dynamics in relatively small
groups. In the current studies we measured people’s responses to the social exclusion of
others in a three-person (Studies 1-3) or a four-person (Study 4) interaction. We,
however, have no reason to expect that our findings are restricted to these smaller groups.
Previous research has shown that even in larger groups participants still notice and
respond to the exclusion of a fellow group member (Jones, Wirth, Ramsey, & Wynsma,
2019). Future research could investigate to what extent group membership also plays a
moderating role in the responses to social exclusion in larger groups.
Our mediation analyses in Study 4 showed that participants’ decision to exclude
a player or not were based on whether or not the player belonged to their group.
However, in the absence of a direct comparison of the participants’ attitudes towards in-
group versus out-group players, it is difficult to conclude whether in-group favoritism or
out-group derogation explain participant’s exclusion behavior in our studies. It is,
however, important to note that whereas group membership affected participants’
reactions to social exclusion, it did not affect participants’ ball tosses in the inclusion
game (in Studies 1 and 2). If participants by default had the intention to exclude an out-
group member, our minimal group manipulation would have resulted in participants
throwing the ball more to the in-group player, independent of inclusionary status (i.e., in
both the inclusion and exclusion condition). Moreover, in Study 4 we showed that when
participants were the first to decide, they rarely decided to exclude the out-group target.
Only when the other players initiated the exclusion, participants decided to exclude the
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
41
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
target. Together, these findings thus more likely reflect in-group favoritism (including the
in-group member; Greenwald & Pettigrew, 2014) rather than out-group derogation
(excluding the out-group target; Halevy, Bornstein, & Sagiv, 2008; Yamagishi &
Kiyonari, 2000).
Participants in our four studies compensated less for the exclusion of a target when
the target was an out-group member than when group membership was not made salient.
Although these findings fit with literature demonstrating that people are less cooperative
with out-group members than with in-group members (Balliet, Wu, & De Dreu, 2014;
Goette, Huffman, & Meier, 2006), and share fewer resources with out-group members
(Baldassarri & Grossman, 2013; Chen & Li, 2009), they diverge from a recent four-player
Cyberball study involving adolescents (Vrijhof et al., 2016). In this study it was found that,
regardless of whether the excluded target was an in- or out-group member, adolescents
actively included the excluded target. One explanation for the differences between their
findings and our own results could be the difference in age groups. Perhaps our adult
participants were more affected by the minimal group manipulation than the adolescent
participants. Another explanation could be that variations in design and procedure explain
the differences with Vrijhof et al. (2016). Because all participants in their study also played
four other versions of the Cyberball game, where group membership was manipulated
differently or not at all, participants’ behavior in the fifth game (which resembled our
current studies) may have been influenced by the norm that had already emerged during the
four previous games. Future research could examine whether our findings among adults
would hold up in a design that more closely resembles Vrijhof et al. (2016), or reversely, in
adolescents when the design more closely resembles that of our Studies 1-3. Indeed, their
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
42
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
findings did reveal that adolescents were showing greater empathic concern for in-group
compared to out-group players, suggesting that the minimal group manipulation did affect
their affective responses.
Somewhat unexpected, and unlike in Cyberball Studies 1 and 2, the participants in
Cyberball Study 3 tossed the ball to the exclusion target significantly more than 50% in
the minimal group condition, thus compensating the exclusion by the other player.
Because our neuroimaging-compatible set-up required that all participants first played an
inclusion game (without a group membership manipulation) before they proceeded with
the exclusion game, they may have been more inclined to transfer this inclusion norm to
the secondary exclusion game (much like the adolescents in the study by Vrijhof et al.,
2016). The within-subjects design of Study 3 may thus have weakened the effect of our
minimal group manipulation. Future research could investigate whether the differences
between the third and the previous two Cyberball studies can be explained by the transfer
of an inclusion norm across games.
Conclusion
Most research on social exclusion has focused strongly on its detrimental effects
on victims (Williams, 2007). In the last decade, however, research has begun to also
examine the sources of exclusion (for a special issue see for example Volume 155, Issue
5 of the Journal of Social Psychology), and more specifically, the actors involved. As a
result, new paradigms have been developed to investigate actors of social exclusion. The
current studies add to this emerging research perspective by focusing on individuals who
respond to social exclusion that is initiated by other group members (see also Riem et al.
2013; Van der Meulen et al., 2016, 2017; Vrijhof et al. 2016) and further underline the
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
43
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
importance of taking the group dynamics into account when examining the emergence of
social exclusion (Nezlek, Wesselmann, Wheeler, & Williams, 2012; Poulsen & Carmon,
2015).
The current research is the first to show that when in-group members initiate the
exclusion of an out-group member, people either choose not to intervene and
consequently fail to adequately compensate for this exclusion (Studies 1-3), or, people
choose to go along with the exclusion as to favor their in-group but experience increased
levels of conflict (Study 4). Irrespective of whether people act in a more passive manner
or choose to actively jump on the “badwagon”, the consequence of their behavior is
relative exclusion of the target. Our findings thus stress the importance of involving all
members of a group when studying social exclusion behavior, according with social-
ecological theories highlighting the role of peers, colleagues, teachers, and families
(Swearer & Espelage, 2004; Williams, 2007). Viewing social exclusion as a group
dynamic, rather than a social interaction between an actor-victim dyad, allows educators
and researchers to think about prevention and intervention efforts that include all
individuals within a group, as minimal as this group may be.
Footnotes
1 In all four studies there were no differences in relevant demographics between the
different groups. Gender and age were equally distributed across conditions.
2 We also examined whether participants chose to exclude the other two players who
were not the target (Players 2 and 3 in the initiate condition and Players 1 and 2 in the
respond condition). In none of the conditions, more than 3 participants chose to exclude
one of these players. There were no significant differences between conditions.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
44
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
45
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Conflict of Interest Statement
The authors declare that there is no conflict of interest.
References
Akerlof, G. A., & Kranton, R. E. (2000). Economics and identity. The Quarterly Journal
of Economics, 115, 715-753.
Baldassarri, D., & Grossman, G. (2013). The effect of group attachment and social
position on prosocial behavior. Evidence from lab-in-the-field experiments. Plos
One, 8.
Balliet, D., Wu, J., & De Dreu, C. K. W. (2014). Ingroup favoritism in cooperation: A
meta-analysis. Psychological Bulletin, 140, 1556-1581.
Boschin, E. A., Brkic, M. M., Simons, J. S., & Buckley, M. J. (2017). Distinct roles for
the anterior cingulate and dorsolateral prefrontal cortices during conflict between
abstract rules. Cerebral Cortex, 27, 34-45.
Botvinick, M. M. (2007). Conflict monitoring and decision making: Reconciling two
perspectives on anterior cingulate function. Cognitive, Affective, and Behavioral
Neuroscience, 7, 356-366.
Brett, M., Anton, J. L., Valabregue, R., & Poline, J. B. (2002). Region of interest analysis
using an SPM toolbox. NeuroImage, 16, 497.
Chen, Y., & Li, S. X. (2009). Group identity and social preferences. American Economic
Review, 99, 431-457.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
46
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Doolaard, F. T., Lelieveld, G. -J., Noordewier, M. K., van Beest, I., & van Dijk, E.
(2020). Get out or stay out: How the social exclusion process affects actors, but
not targets. Journal of Experimental Social Psychology, 88, 103946.
Dunham, Y., Baron, A. S., & Carey, S. (2011). Consequences of “minimal” group
affiliations in children. Child Development, 82, 793-811.
Espelage, D., Green H. D., & Wasserman, S. (2007). Statistical analysis of friendship
patterns and bullying behaviors among youth. New Directions for Child &
Adolescent Development, 118, 61-75.
Festinger, L. (1957). A theory of cognitive dissonance. Stanford, CA: Stanford University
Press.
Gaertner, L., & Insko, C. A. (2000). Intergroup discrimination in the minimal group
paradigm: Categorization, reciprocation, or fear? Journal of Personality and
Social Psychology, 79, 77-94.
Gerard, H. B., & Hoyt, M. F. (1974). Distinctiveness of social categorization and attitude
toward ingroup members. Journal of Personality and Social Psychology, 29, 836–
842.
Goette, L., Huffman, D., & Meier, S. (2006). The impact of group membership on
cooperation and norm enforcement: Evidence using random assignment to real
social groups. American Economic Review, 96, 212-216.
Greenwald, A. G., & Pettigrew, T. F. (2014). With malice toward none and charity for
some: Ingroup favoritism enables discrimination. American Psychologist, 69, 669.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
47
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Greene, J. D., Nystrom, L. E., Engell, A. D., Darley, J. M., & Cohen, J. D. (2004). The
neural bases of cognitive conflict and control in moral judgment. Neuron, 44, 389-
400.
Halevy, N., Bornstein, G., & Sagiv, L. (2008). “In-group love” and “out-group hate” as
motives for individual participation in intergroup conflict: A new game paradigm.
Psychological science, 19, 405-411.
Hellevik, O. (2009). Linear versus logistic regression when the dependent variable is a
dichotomy. Quality and Quantity, 43, 59–74.
Hewstone, M., Rubin, M., & Willis, H. (2002). Intergroup bias. Annual Review of
Psychology, 53, 575-604.
Jones, E. E., Wirth, J. H., Ramsey, A. T., & Wynsma, R. L. (2019). Who is less likely to
ostracize? Higher trait mindfulness predicts more inclusionary behavior.
Personality and Social Psychology Bulletin, 45, 105-119.
Kerns, J. G. (2006). Anterior cingulate and prefrontal cortex activity in an FMRI study of
trial-to-trial adjustments on the Simon task. NeuroImage, 33, 399-405.
Kerns, J. G., Cohen, J. D., MacDonald, A. W., Cho, R. Y., Stenger, V. A., & Carter, C. S.
Anterior cingulate conflict monitoring and adjustments in control. Science, 303,
1023-1026.
Lelieveld, G. -J., Shalvi, S., & Crone, E. A. (2016). Lies that feel honest: Dissociating
between incentive and deviance processing when evaluating dishonesty.
Biological Psychology, 117, 100-107.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
48
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Levine, M., Cassidy, C., Brazier, G., & Reicher, S. (2002). Self-categorization and
bystander non-intervention: Two experimental studies. Journal of Applied Social
Psychology, 32, 1452-1463.
Lieberman, M. D., & Cunningham, W. A. (2009). Type I and Type II error concerns in
fMRI research: re-balancing the scale. Social Cognitive and Affective
Neuroscience, 4, 423-428.
Malti, T., Strohmeier, D., & Killen, M. (2015). The impact of onlooking and including
bystander behaviour on judgments and emotions regarding peer exclusion. British
Journal of Developmental Psychology, 33, 295-311.
Muller, D., Judd, C. M., & Yzerbyt, V. Y. (2005). When moderation is mediated and
mediation is moderated. Journal of Personality and Social Psychology, 89, 852-
863.
Nesdale, D., Killen, M., & Duffy, A. (2013). Children’s social cognition about proactive
aggression. Journal of Experimental Child Psychology, 116, 674-692.
Nezlek, J. B., Wesselmann, E. D., Wheeler, L., & Williams, K. D. (2012). Ostracism in
everyday life. Group Dynamics: Theory, Research, and Practice, 16, 91-104.
Palmer, S. B., Rutland, A., & Cameron, L. (2015). The development of bystander
intentions and social-moral reasoning about intergroup verbal aggression. British
Journal of Developmental Psychology, 33, 419-433.
Poulsen, J. R., & Carmon, A. F. (2015). Who would do that ? A theory-based analysis of
narratives of sources of family ostracism. The Journal of Social Psychology, 155,
452-470.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
49
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Preacher, K. J., Rucker, D. D., & Hayes, A. F. (2007). Addressing moderated mediation
hypotheses: Theory, methods, and prescriptions. Multivariate Behavioral
Research, 42, 185-227.
Riem, M. M., Bakermans-Kranenburg, M. J., Huffmeijer, R., & Van Ijzendoorn, M. H.
(2013). Does intranasal oxytocin promote prosocial behavior to an excluded
fellow player? A randomized-controlled trial with Cyberball.
Psychoneuroendocrinology, 38, 1418-1425.
Smith, A., & Williams, K. D. (2004). R U there? Effects of ostracism by cell phone
messages. Group Dynamics: Theory Research, and Practice, 8, 291-301.
Smith, E. H. et al. (2019). Widespread temporal coding of cognitive control in the human
prefrontal cortex. Nature Neuroscience, 22, 1883-1891.
Somerville, L. H., Heatherton, T. F.., & Kelley, W. M. (2006). Anterior cingulate cortex
responds differentially to expectancy violation and social rejection. Nature
Neuroscience, 9, 1007-1008.
Swearer, S. M., & Espelage, D. L. (2004). A social-ecological framework of bullying
among youth. In D. L. Espelage & S. M. Swearer (Eds.), Bullying in American
schools: A social-ecological perspective on prevention and intervention (pp. 1–
12). Mahwah, NJ: Erlbaum.
Tajfel, H. (1970). Experiments in intergroup discrimination. Scientific American, 223, 96-
102.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and
intergroup behaviour. European Journal of Social Psychology, 1, 149-178.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
50
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Tajfel, H., & Turner, J. C. (1986). The social identity theory of intergroup behavior. In S.
Worchek & W. Austin (Eds.) Psychology of intergroup relations (pp. 7-24).
Chicago, IL: Nelson-Hall.
Van der Meulen, M., Van Ijzendoorn, M. H., & Crone, E. A. (2016). Neural correlates of
prosocial behavior: Compensating social exclusion in a four-player cyberball
game. Plos One, 7, 11.
Van der Meulen, M., Steinbeis, N. A. J., Achterberg, M., Bilo, E., Van den Bulk, B. G.,
Van Ijzendoorn, M. H., & Crone, E. A. (2017). The neural correlates of dealing
with social exclusion in childhood. Neuropsychologia, 103, 29-37.
Van Veen, V. & Carter, C. S. (2005). Separating semantic conflict and response conflict
in the Stroop task: a functional MRI study. NeuroImage, 27, 497-504.
Van Veen, V. & Carter, C. S. (2006). Conflict and cognitive control in the brain. Current
Directions in Psychological Science, 15, 237–240.
Voci, A. (2006). The link between identification and in-group favourtism: Effects of
threat to social identity and trust-related emotions. British Journal of Social
Psychology, 45, 265-284.
Vrijhof, C. I., Van den Bulk, B. G., Overgaauw, S., Lelieveld, G. -J., Engels, R. C. M. E.,
& Van IJzendoorn, M. H. (2016). The Prosocial Cyberball Game: Compensating
for social exclusion and its associations with empathic concern and bullying in
adolescents. Journal of Adolescence, 52, 27-36.
Wang, J., Iannotti, R. J., Luk, J. W., & Nansel, T. R. (2010). Co-occurrence of
victimization from five subtypes of bullying: physical, verbal, social exclusion,
spreading rumors, and cyber. Journal of Pediatric Psychology, 35, 1103-1112.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
51
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Wesselmann, E. D., Wirth, J. H., Pryor, J. B., Reeder, G. D., & Williams, K. D. (2013).
When do we ostracize? Social Psychological and Personality Science, 4, 108-115.
Williams, K. D. (2007). Ostracism. Annual Review of Psychology (Vol. 58). New York:
Annual Reviews.
Williams, K. D., Cheung, C. K. T., & Choi, W. (2000). CyberOstracism: Effects of being
ignored over the Internet. Journal of Personality and Social Psychology, 79, 748-
762.
Williams, K. D., & Sommer, K. L. (1997). Social ostracism by one’s coworkers: Does
rejection lead to loafing or compensation? Personality and Social Psychology
Bulletin, 23, 693-706.
Wooldridge, J. M. (2013). A binary dependent variable: The Linear Probability Model.
Introductory Econometrics: A Modern Approach (5th international ed.). Mason,
OH: South-Western. pp. 238–243.
Yamagishi, T., & Kiyonari, T. (2000). The group as the container of generalized
reciprocity. Social Psychology Quaterly, 63, 116-132.
Zadro, L., & Gonsalkorale, K. (2014). Sources of ostracism: The nature and
consequences of excluding and ignoring others. Current Directions in
Psychological Science, 23, 93-97.
Zadro, L., Williams, K. D., & Richardson, R. (2005). Riding the O’ Train: Comparing the
effects of ostracism and verbal dispute on targets and sources. Group Processes
and Intergroup Relations, 8, 125-143.
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
52
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Table 1
Percentage of Ball Tosses to the Excluded Player as a Function of Inclusionary Status
and Group Membership (Study 1)
Inclusion
Exclusion
M
SD
M
SD
Minimal group
51.54% a
9.36
53.35% a
12.90
Control
49.84% a
7.35
60.57% b
9.46
Note. Means with different superscripts differ significantly across all cells (ps <.05,
analyzed with simple-effect analyses).
Table 2
Percentage of Ball Tosses to the Excluded Player as a Function of Inclusionary Status
and Group Membership (Study 2)
Inclusion
Exclusion
M
SD
M
SD
Minimal group
52.62% a
7.90
54.52% a
12.29
Control
51.18% a
8.75
59.80% b
4.47
Note. Means with different superscripts differ significantly across all cells (ps <.05,
analyzed with simple-effect analyses).
Table 3
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
53
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Brain regions revealed by whole-brain contrasts, including MNI coordinates. Peak
voxels reported at p < .001 uncorrected, at least 10 contiguous voxels (voxels size was
3.0 × 3.0 × 3.0 mm).
Anatomical Region
L/R
voxels
Z
MNI coordinates
x
y
z
ExclusionThrow > InclusionThrow
Dorsolateral Prefrontal Cortex
L
239
4.66
-27
41
31
*
4.59
-24
53
25
*
4.34
-18
20
34
*
InclusionThrow > ExclusionThrow
Visual Cortex
L/R
4874
4.94
39
-85
-8
**
4.92
48
-76
4
**
4.73
45
-46
-23
**
[ExclusionThrow > InclusionThrow] Minimal Group >
[ExclusionThrow > InclusionThrow] Control
Dorsolateral Prefrontal Cortex
L
21
3.70
-24
41
28
3.57
-15
50
25
ExclusionGet-ExclusionOut
Dorsal Anterior Cingulate Cortex
L/R
580
6.48
-3
-4
52
**
5.37
-6
11
40
**
4.96
-6
17
34
**
Motor Cortex
L
661
6.95
-42
-22
58
**
6.31
-30
-16
64
**
5.19
-54
-19
46
**
ExclusionOut-ExclusionGet
Visual Cortex
L
18
3.89
-39
-16
37
*
R
90
5.74
15
-85
4
**
3.43
6
-82
25
**
Motor Cortex
L/R
539
6.01
27
-25
64
*
5.61
36
-22
49
*
5.38
-12
-28
67
*
InclusionGet > InclusionOut
Anterior Cingulate Cortex
L/R
5282
7.12
-6
2
49
**
Temporoparietal Junction
6.59
-54
-22
34
**
Motor Cortex
6.91
-33
-10
54
**
Insula
L/R
1146
6.70
39
14
7
**
6.56
45
11
7
**
6.51
36
17
10
**
InclusionOut > InclusionGet
Visual Cortex
L
53
6.79
-12
-88
1
**
R
187
6.79
15
-85
4
**
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
54
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
6.05
9
-82
22
**
5.80
12
-82
31
**
* The results remained significant with an FDR-corrected threshold of p < .05, with an
extent threshold of 10 contiguous voxels.
** The results remained significant with an FWE-corrected threshold of p < .05, with an
extent threshold of 10 contiguous voxels.
Figure 1 The group membership manipulation used in Studies 1-3. Participants in the minimal group
condition (Figure 1A) played a game of Cyberball where they (Player C) had the same color as the
person initiating the exclusion (Player A), but the target (Player B) had a different color. Participants
in the control condition (Figure 1B) played a game where all three players had different colors.
Figure 2 (A) Whole-brain results for regions active in the ExclusionThrow > InclusionThrow contrast
(threshold at p <.05, FDR corrected). Activation was detected in the dlPFC (MNI coordinates: x = -
27, y = 41, z = 31). (B) Parameter estimates plotted for the minimal group and control conditions
of the exclusion game, and for all participants of the inclusion game (we did not manipulate group
membership in the inclusion condition). (C) Activation in the dlPFC correlated positively with the
percentage of throws to the excluded player, across all
conditions.
Figure 3 Whole-brain results of the two sample t-test for regions active in the ExclusionThrow >
InclusionThrow contrast for Minimal Group > Control (threshold at p < .001, uncorrected).
Activation was detected in the dlPFC (MNI coordinates: x = -24, y = 47, z = 28).
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020

UNCORRECTED MANUSCRIPT
55
Running head: RESPONSES TO SOCIAL EXCLUSION OF OTHERS
Figure 4. Screenshot of the dot estimation task. Participants selected the picture with most dots.
Figure 5 The manipulation of exclusion decision and group membership used in Study 4. Participants
were asked whether or not they wanted to adjust the group composition in the presence of a
minimal group manipulation (A and C) or in the absence of one (B and D). Moreover, they either
had to initiate the decision (initiate condition; A and B) or to respond to the decision made by other
players (respond condition; C and D).
Downloaded from https://academic.oup.com/scan/advance-article-abstract/doi/10.1093/scan/nsaa070/5841602 by University College London user on 29 May 2020