INGROUP COOPERATION AND OUTGROUP COMPETITION: DO RELATIONSHIPS
WITHIN THE INGROUP AFFECT REPRESENTATIONS OF THE OUTGROUP?
By
Victor Nahuel Felix de Souza Keller

A THESIS
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Psychology - Master of Arts
2017

ABSTRACT
INGROUP COOPERATION AND OUTGROUP COMPETITION: DO RELATIONSHIPS
WITHIN THE INGROUP AFFECT REPRESENTATIONS OF THE OUTGROUP?
By
Victor Nahuel Felix de Souza Keller
When deciding whether to engage in a conflict, individuals must assess how likely they
are to win. Research has shown that when individuals are unlikely to win a fight they see their
opponents as physically larger. This project tests whether cues of ingroup cooperativeness are
used as an indication that one will succeed in a conflict with an outgroup member. In two
experiments, ingroup cooperativeness is manipulated by having subjects play a public goods
game with either a cooperative or a selfish ingroup. This manipulation did not affect perceptions
of an outgroup member’s physical size, regardless of whether the groups were real (Study 1 –
university affiliation) or minimal (Study 2 – over- or under-estimators). In Study 1, the
manipulation appears to not have been an effective cue to ingroup cooperativeness. In Study 2,
the outgroup may not have been sufficiently threatening to elicit assessments of the likelihood of
winning a conflict. To test whether ingroup cooperativeness plays a role in decisions to engage in
conflict, future studies should address the limitations in the present studies.

TABLE OF CONTENTS

LIST OF TABLES

iv

LIST OF FIGURES

v

Introduction
Evolution of Conflict Decision-Making
Cognitive Mechanisms for Human Conflict Decision-Making
The Role of Groups in Conflict Decision-Making
The Present Research

1
1
3
6
8

Study 1
Method
Participants
Materials and Procedure
Results and Discussion

10
10
10
10
12

Study 2
Method
Part 1: Generating the classification images
Participants
Materials and procedure
Part 2: Rating the classification images
Results and Discussion
Part 1: Self-reported formidability of the outgroup
Part 2: Representations of the outgroup generated with the RCP

16
17
17
17
17
18
18
19
19

General Discussion

22

Conclusion

25

APPENDIX

26

REFERENCES

32

iii

LIST OF TABLES

Table 1: Word stimuli used in the attribute categories of the IATs

27

Table 2: Effect of the manipulation on perceived outgroup formidability (mean on
Fessler’s scales) for male, female, white, black, and Asian participants in
Study 1

28

Table 3: Effect of the manipulation on IAT D scores for male, female, white,
black, and Asian participants in Study 1

29

Table 4: Effect of the manipulation on perceived outgroup formidability (mean on
Fessler’s scales) for male, female, white, black, and Asian participants in
Study 2

30

iv

LIST OF FIGURES

Figure 1: Classification images of an outgroup member’s face generated with the
reverse correlation procedure for cooperative (A) and selfish ingroup
(B) conditions

v

31

Introduction
In group situations, such as sporting events or political marches, displays of commitment
to the ingroup seem to co-occur with displays of contempt towards an outgroup. Research
suggests that the former does not necessarily lead to the latter (Brewer, 1999). However, it is
possible that in specific circumstances ingroup cooperation could inform the individual’s
decision to engage in a conflict. By considering an adaptive function of ingroup relationships in
intergroup conflicts, this project seeks to test mechanisms by which displays of commitment to
the ingroup could lead to aggression towards an outgroup.
Evolution of Conflict Decision-Making
Animals from many different species use physical conflict as a means of attaining and
controlling resources. For instance, baboons use aggression as a way of imposing dominance
(Hausfater, 1974), male lions brawl amongst themselves over access to female lions (Packer &
Pusey, 1982), and chimpanzees raid neighboring chimpanzee groups to expand their territory
(Mitani, Watts, & Amsler, 2010). Despite being recognized as more prosocial than many other
species, humans also use physical conflicts to access material resources, increase status, and
become more desirable to mates (Buss & Shackelford, 1997).
Even though aggression can be beneficial, engaging in it at every opportunity is unlikely
to worthwhile in the long run. In addition to risking death, being aggressive may lead to serious
physical injury or damage to the individual’s reputation. Furthermore, physical conflicts demand
effort and time that could be used in more efficient ways, thereby imposing opportunity costs.
Thus, the decision to initiate a physical confrontation must consider the likely costs and benefits.
This raises the question: how do individuals decide whether engaging in a physical conflict will
be worthwhile?

1

To answer this question, evolutionary biologists have developed models describing which
decision-making strategies would be most likely to evolve. Maynard Smith and Parker (1976)
show that to determine the likelihood of success in a conflict, individuals must assess their ability
to attain and control resources and the value of possessing the resource both relative to the
opponent’s ability and the value of the resource to the opponent. One’s ability to attain and
control resources relative to the opponent’s ability is referred to as the resource holding potential
(RHP). For example, personal strength and number of individuals in one’s group are indicators
of RHP. If one is physically strong and the opponent is physically weak, then one has a higher
resource holding potential. Assuming that the resource is equally valuable to both individuals,
the individual that bears the highest RHP is more likely to reap benefits from a conflict.
However, a resource may not be equally valuable for both individuals – a situation referred to as
payoff asymmetry. For example, an individual who does not have food stocked for the winter
would receive a larger payoff for attaining more food than an individual who had stock for the
winter. Even if the one individual had a higher RHP, the benefit of attaining food may be so high
for the other individual that it offsets the potential cost of fighting for the latter individual. Thus,
according to Maynard and Smith (1976) decision to engage in a conflict over a resource must
consider the RHP and payoff asymmetry of both parties.
If this model accurately describes animal decision-making, indicators of RHP and payoff
asymmetry should influence animal aggression. This has been observed in many studies on
animal behavior. Payoff asymmetry, for example, is a well-documented precursor to conflict –
animals deprived of food tend to be more aggressive than those who are not deprived of food
(Houston & McNamara, 1999). Size – a reliable indicator of RHP – has also been found to
predict aggression (Schuett, 1997). There is also evidence in social species that the number of

2

individuals in a group may function as an indicator of RHP. For example, Benson-Amram,
Heinen, Dryer, and Holekamp (2011) found that hyenas were less likely to approach outgroup
hyenas if hyena audio recordings led them to believe they were outnumbered. This sensitivity to
the number of individuals in rival groups holds for species with different phylogenetic histories,
such as chimpanzees (Wilson, Hauser, & Wrangham, 2001) and lions (McComb, Packer, &
Pusey, 1994).
Humans likely faced the selective pressures of intergroup conflict. The fossil record
shows evidence of interpersonal violence among early humans (Thorpe, 2003) and other
hominids (Berger & Trinkaus, 1995). Furthermore, Bowles (2009) reviewed archeological and
ethnographic data and estimated that homicides may have accounted for 14% of deaths on
average. These data indicate that aggressive encounters affected human fitness for most, if not
all, of our evolutionary history. Therefore, modern humans should have cognitive mechanisms
selected to deal with interpersonal conflicts. If this is the case, how do these mechanisms carry
out their function?
Cognitive Mechanisms for Human Conflict Decision-Making
Maynard Smith and Parker’s (1976) evolutionary model describes the functions that
conflict-related decision-making mechanisms should fulfill. Such mechanisms would need to be
able to assess resource holding potential (RHP) and payoff asymmetries as well as generate
courses of action. With these functions in mind, psychologists can infer which design features
these cognitive mechanisms should have.
Neuberg, Kenrick, and Schaller (2011) propose that the adaptive problems faced by
humans resulted in a self-protection system that would carry out the decision-making processes
required by evolutionary-biological theory. They describe the system in two processes: (1)

3

detecting the threat of a conflict and (2) outputting cognitive, emotional, and/or behavioral
experiences that prepare the individual to make decisions – either escalate or de-escalate conflict.
To detect cues of impending conflicts, the first process must be sensitive to situational
characteristics correlated with RHP and payoff asymmetries. For example, displays of anger
frequently precede physical aggression which, in turn, is associated with larger RHP and payoff
asymmetries. Studies reviewed by Neuberg et al. (2011) show that individuals detect anger
expressions more rapidly and effectively than other expressions (Becker, Kenrick, Neuberg,
Blackwell, & Smith, 2007; Fox, Lester, Russo, Bowles, Pichler, & Dutton, 2000; Schupp et al.,
2004), suggesting a mechanism for detecting interpersonal threats.
The system’s second process should prepare the individual to follow through with the
course of action suggested by the threat-detection process: either conflict escalation (e.g.,
fighting, displaying threat signals) or conflict de-escalation (e.g., fleeing, surrendering
resources). To accomplish this, the second process can affect the individual’s behavior,
emotions, or cognitions so that the optimal action (according to the first processes’ calculations)
is most likely to be pursued. An example highlighted by Neuberg et al. (2011) is the emotion of
fear. They point out that fear – but not other negative emotions – is felt when encountering
members of groups stereotyped as aggressive (Schaller & Neuberg, 2008), and that this emotion
specifically encourages fleeing behavior. In sum, the second process seems to act as an output of
the system, preparing the individual for action.
Studies by Fessler and colleagues illustrate how the outputs of the self-protection system
are not only emotional and behavioral, but can even be perceptual, such as distortions of the
perceived size of the aggressor. They argue that perceiving an opponent as smaller – less
formidable – motivates the individual to escalate the conflict, and perceiving an opponent as

4

larger motivates the individual to de-escalate conflict. If such perceptual biases are to be
functional, they should operate according to the relevant RHP computations. This is consistent
with the evolutionary model – since formidability is a reliable indicator of RHP, a bias in its
perception can emphasize differences in RHP and lead to more adaptive responses. Some of their
studies show how cues pertaining to RHP of the individual and the opponent affect perceptions
of the opponent’s size.
Throughout their studies, Fessler et al. find that cues of RHP and payoff asymmetries
affect perceptions of physical formidability. Fessler and Holbrook (2013a), for example, had
participants rate the size of an outgroup male while participants were either bound to their seats
or not bound to their seats. Because constricting movement reduces RHP, the self-protection
system should induce the individual to de-escalate conflict. Fessler and Holbrook’s (2013a)
results support this rationale as participants who were bound to their chairs perceived the
outgroup male as physically larger.
In addition to detecting indices of RHP, the self-protection system must also assess
payoff imbalances. If the opponent in a confrontation has more to gain in escalating a conflict,
they may be willing to incur larger costs than one is prepared to bear. Fessler, Holbrook, Pollack,
and Hahn-Holbrook (2014) argue that parents would often have found themselves in such a
situation. For those who are rearing their children, sustaining wounds in battle would mean being
unable to provide resources for their offspring as well as for themselves – a large inclusive
fitness cost. Fessler et al. (2014) found that parents rated a threatening male as physically larger
than did non-parents, indicating that the possibility of bearing larger costs induces the system to
prepare for conflict de-escalation.

5

Insofar as cues indicate RHP or payoff asymmetries, they should affect outputs of the
self-protection system. For instance, Fessler and colleagues also found that lacking upper body
strength (Fessler, Holbrook, & Gervais, 2014) or encountering impulsive opponents (Fessler,
Tiokin, Holbrook, Gervais, & Snyder, 2014) make the individual perceive the opponent as more
formidable. Another source of information regarding RHP and payoffs are the people in one’s
surroundings.
The Role of Groups in Conflict Decision-Making
Much like when confronting a single aggressor, individuals in a group conflict must base
their decisions on an assessment of their RHP and payoff asymmetry of escalating the conflict
(Tooby & Cosmides, 1988). The group affords unique indicators of these elements and can
therefore sway individual decision-making by engaging the self-protection system.
Fessler and colleagues conducted studies demonstrating group-level influences on the
system’s processes. A notable characteristic that should inform the individual’s RHP is the
relative sizes of the groups. Being in the group with more members signals a greater likelihood
of success and should produce outputs that motivate conflict. Fessler and Holbrook (2013b)
demonstrated this in a study in which men who were accompanied by friends (vs. alone) rated a
threatening outgroup male as less formidable.
However, number of group members is not the only arbiter of intergroup conflicts. The
degree to which group members coordinate effectively is also important to the likelihood of
success. Groups that are able to time their actions so as to maximize damage while minimizing
costs (for example, by coordinating to target essential outgroup members) are more likely to
succeed in a conflict. This can be seen in occasions such as historic military confrontations as

6

well as current sporting events. Therefore, individuals who can base their decisions on indicators
of efficient coordination will more effectively distinguish wasteful from fruitful conflicts.
When an individual decides to engage in an intergroup conflict, they are implicitly
deciding to share the potential costs and benefits that may come from the conflict with other
group members. Because relationships within the group vary in how cooperative they are (e.g.,
some individuals may withhold resources, restrict mating, abuse their hierarchical standing, or
not reciprocate favors), it can be disadvantageous to an individual that all group members receive
their share of the profit. Furthermore, uncooperative relationships in the group add uncertainty as
to whether all members will split attained resources equally, will carry out their coordinated
efforts as agreed upon, or are withholding information as to whether all will benefit equally from
a conflict. Consequently, members of an uncooperative group1 risk failure in coordinating
actions and should have a lower RHP than those of a cooperative group. Before inducing the
individual to participate or not in the conflict, the self-protection system should then be sensitive
to how cooperative other group members are.
Fessler and Holbrook (2014) give initial evidence for this by showing that individuals
who walked in synchrony (vs. in their own pace) perceived an antagonist man to be less
formidable. They argue that synchronizing actions with others indicates that they are willing and
able to coordinate, which increases the potential for winning conflicts. However, to my
knowledge, no study has directly manipulated the cooperative relationships among group
members in addition to measuring how outgroup threats are seen.

1

The term cooperative will be used to refer to the likelihood of coordinating given the understanding that
coordination is a type of cooperation (Thomas, DeScioli, Haque, & Pinker, 2014). Moreover, the overall
cooperativeness of an individual may function as a cue of the likelihood of coordinating with other ingroup
members.

7

The Present Research
This project includes two studies aimed to test whether cooperative relationships within
the group affect how outgroup males are represented. To manipulate ingroup relationships,
participants played a public goods game (Marwell & Aimes, 1979) against ingroup members
who, unbeknownst to the participants, were computerized players that were either cooperative or
selfish (e.g., Park & Stone, 2010). In this economic game, participants could contribute amounts
of a personal endowment to a private or a public account. At the end of each round, they received
the amount they had invested in their private account in addition to a subset of 300% of what all
players had invested in the public account. The computerized players either contributed more
than 90% (cooperative) or less than 10% (selfish) of their endowments to the public account.
The public goods game mimics some features of coordination problems. For example, if
all participants contribute to the public account (i.e., coordinate), they receive the largest possible
payoff. Alternatively, if there are selfish group members, investment in the public account may
lead to losses. Thus, the public goods game is similar to a coordination problem in conflict
situations. As compared to other economic games (e.g., prisoners dilemma), this may render it
more informative of whether group members are likely to coordinate in an actual conflict.
While the public goods game was used to manipulate the cue fed into the self-protection
system, the two studies used different strategies to assess the system’s output. In the first study,
Michigan State University students reported how formidable they believed a threatening student
of a rival university (University of Michigan) was on the measures used by Fessler and
colleagues in their research program. Using real groups in the manipulation of ingroup
relationships and measurement of perceived outgroup formidability has the advantage of
providing a rich context (e.g., experience based on past interactions, strong identification).

8

The second study addresses potential limitations and attempts to extend the conclusions
of the first study. To begin with, minimal groups were used instead of university affiliation
(Tajfel, Billing, Bundy, & Flament, 1971; Ratner, Dotsch, Wigboldus, & van Knippenberg,
2014). This tests whether the effect generalizes to other groups which lack a rich context to
ground participants’ judgments. Second, participants completed the study in group sessions and
were led to believe that they would compete with the outgroup participants in the room on a
series of strength tasks. This confers a more consequential situation where participants may
suffer or enjoy the results of a competition. Finally, a reverse correlation image classification
procedure (RCP; described in more detail below) was used to gauge participants' mental
representations2 of the outgroup.

2

I remain agnostic as to whether the measures used by Fessler et al. assess perception or representations. Fessler et
al. (2013) concede that their measure does not guarantee that perception is being assessed purely (also, see Firestone
& Scholl, 2015, for a critique of measures of perception). Furthermore, distorted representations would likely have
the same function as distorted perceptions (Fessler & Holbrook, 2013).

9

Study 1
Method
Participants. 355 Michigan State University undergraduates participated in the
experiment (61% female; 67% white, 13% Asian, 11% black). The sample size varies for each
analysis depending on how many students completed the measure. The total sample size grants
the study approximately 80% power to detect an effect of d = 0.30 (two-sided independentsamples t-test; alpha = .05).
Materials and Procedure. Subjects were recruited to participate in a study about the
university community and person perception in which they would interact with other MSU
students and provide their opinions on students’ characteristics. They completed the procedure in
a computer lab in sessions of three to eight participants.
To manipulate the participants’ relationship with the ingroup, they played a rigged
public-goods game with computerized players purported to be the other MSU students in the lab.
Participants began each of four rounds with 50 tickets. In each round, participants invested their
tickets in a private or a public account. The total number of tickets invested in the public account
was then tripled and redistributed equally to all participants. Therefore, at the end of the round
they received what they invested in their private account and their share of what was
redistributed from the public account. The manipulation of ingroup cooperativeness was how
much the computerized players invested in the public account. In the cooperative-ingroup
condition, computerized players consistently invested most (about 90%) of their tickets in the
public account. In the selfish-ingroup condition, they invested very little (about 10%) in the
public account. At the end of the game, the average number of tickets each participant earned

10

throughout the rounds would be entered in a raffle for a 50 dollar gift-card to a local
supermarket.
After seeing the game’s results, participants read a description of an encounter with a
male undergraduate student from the University of Michigan (UM), which is known to rival
MSU in athletic competitions. The description instructed participants to imagine they were
leaving a football game MSU played against UM. They pictured themselves walking down a
poorly lit street when a male in his late twenties wearing a UM football jersey came toward them
angrily cursing MSU for having won the game. They were then asked to estimate the height,
size, and muscularity of the man on the items used by Fessler and Holbrook (2013). They also
indicated how likely on a 9-point scale (1 = not at all likely; 9 = extremely likely) they would
have been to feel fear, swear at the man, physically aggress the man, cross the street, turn around
and walk away, and tell the man to leave them alone.
For exploratory purposes, participants then completed two seven-block Implicit
Association Tests (IAT; Greenwald, McGhee, & Schwartz, 1998; McDonald, Asher, Kerr, &
Navarrete, 2011). First, they completed an evaluative IAT with MSU and UM as the target
categories, and positive and negative as the attribute categories. Then, they completed a
stereotype IAT with the same target categories, but physical and mental as the attribute
categories. Pictures of White males in MSU and UM t-shirts were used as the target category
stimuli, while words were used as the attribute category stimuli (Table 1). The D score for was
used as a measure of implicit attitudes and stereotypes (Greenwald, Nosek, & Banaji, 2003).
Next, they completed two versions of a 10-item measure of group identification (α = .91;
e.g., “I feel a bond with [group]”; 1 = strongly disagree, 7 = strongly agree; Leach et al., 2008).
In the first version, the items referred to MSU as the ingroup (α = .91); in the second, they

11

referred to UM as the ingroup (α = .85). As a manipulation check, participants indicated on a 7point scale (1 = not at all; 7 = extremely) how supportive, trustworthy, helpful, and cooperative
they thought MSU students were (α = .92). After answering standard demographic questions (in
addition to questions about their weight, height, physical strength), participants completed a
funneled debriefing to determine whether they were aware of the study’s goal or suspected the
players of the public goods game to be computerized.
Results and Discussion
To assess whether the manipulation of MSU student cooperativeness was effective, an
independent-samples t-test was performed comparing participants in the cooperative and selfish
conditions on the average ratings of student cooperativeness. Participants who played the public
goods game with cooperative MSU students (M = 5.31, SD = 0.96) did not think MSU students
were significantly more cooperative than did participants who played with selfish MSU students
(M = 5.22, SD = 1.05), t(333) = 0.81, p = .421, d = 0.08 95% CI [-0.12, 0.29]. Even though the
manipulation did not affect self-reports of ingroup cooperativeness, the main analyses were done
in case self-reports of ingroup cooperativeness are not the relevant variable to conflict decisionmaking. For example, if the self-reports are too vulnerable to social desirability effects or if
implicit knowledge of ingroup cooperativeness is more important than explicit knowledge in
swaying perceptions of outgroup formidability.
While there should be a main effect of ingroup cooperativeness on perceptions of
outgroup formidability, a more powerful test of the hypothesis would also consider identification
with the ingroup. If one does not identify with the ingroup, then ingroup cooperativeness would
have a much smaller effect on perceptions of outgroup formidability. Thus, the effect of ingroup
cooperativeness on outgroup formidability might be moderated by identification with the

12

ingroup. To test this, three exploratory multiple regressions were performed on the perceived
size, muscularity, and height of the UM man (Fessler & Holbrook, 2014). The predictors in the
regression were the participants’ conditions (effects-coded: 1 = cooperative, -1 = selfish), the
participants’ identification with MSU, and the interaction between the former and the latter.
After excluding two outliers, the regression on perceived size of the UM assailant revealed no
main effect of the ingroup cooperativeness manipulation (b = 0.34, SE = 0.31, p = .267), no main
effect of identification with MSU (b = 0.07, SE = 0.06, p = .233), and no interaction effect
between both predictors (b = -0.05, SE = 0.06, p = .394).3 The results were the same for
perceived muscularity (manipulation: b = 0.19, SE = 0.33, p = .564; identification: b = 0.09, SE =
0.06, p = .127; interaction: b = -0.04, SE = 0.06, p = .501) and height (manipulation: b = 0.08, SE
= 0.83, p = .921; identification: b = 0.15, SE = 0.16, p = .323; interaction: b = 0.01, SE = 0.16, p
= .967). To test whether ingroup cooperativeness is more relevant to individuals of a specific sex,
race, or ethnicity, the effects of the manipulation on overall perception of outgroup formidability
for each sex, race, or ethnicity within the sample was assessed using independent-samples t-tests.
These tests also revealed no effect of ingroup cooperativeness on perceptions of outgroup
formidability (Table 2). Thus, ingroup cooperativeness does not seem to affect perceptions of
outgroup formidability, regardless of identification with the ingroup, sex, race, or ethnicity.
To explore whether there was an effect on automatic evaluations of the outgroup instead
of self-reports, the IATs’ D scores were compared between conditions using independentsamples t-tests. These analyses suggest that the manipulation of ingroup cooperativeness does

3

When the outliers were included, there were significant main effects of the manipulation (participants in the
cooperative ingroup condition found the UM assailant to be larger than did those in the self ingroup condition, p =
.042) and identification (those who identified more with MSU found the UM male to be larger, p = .037), and a
marginally significant interaction effect (identification had a stronger positive association with perceived size for
those in the selfish ingroup condition, p = .083). Because these effects depended on the inclusion of only two cases
out of 335, I chose to focus on the effects when the outliers were removed.

13

not affect participants’ implicit stereotypes of UM student formidability (cooperative ingroup: M
= 0.08, SD = 0.30; selfish ingroup: M = 0.07, SD = 0.31), t(309) = 0.37, p = .714, d = 0.04 95%
CI [-0.16, 0.24], or their implicit attitudes toward UM students (cooperative ingroup: M = 0.39,
SD = 0.31; selfish ingroup: M = 0.40, SD = 0.33), t(309) = 0.31, p = .755, d = 0.03 95% CI [0.17, 0.24]. The same results were found for individuals of all sexes, races, and ethnicities within
the sample (with the exception of Asian participants; see Table 3).
While ingroup cooperativeness does not seem to greatly affect implicit stereotypes or
attitudes of outgroups, it could be the case that implicit attitudes are only affected by ingroup
cooperativeness when there are implicit stereotypes of the outgroup as formidable. Therefore, a
regression was performed with the strength IAT, manipulation, and their interaction predicting
implicit attitudes. There was a main effect of the strength IAT (those who had a stereotype of
UM students as stronger tended to have more negative attitudes toward UM4; b = 0.16, SE =
0.06, p = .005), but no main effect of the manipulation (b = -0.01, SE = 0.02, p = .739) or an
interaction effect (b = -0.01, SE = 0.06, p = .907).
This study offers a test of the hypothesis insofar as the manipulation check did not
provide useful information. For example, if implicit knowledge of ingroup cooperativeness is
more relevant than the explicit knowledge assessed by self-report or if the measure was affected
by extraneous variables such as social desirability. If this was the case and the manipulation was
in fact successful at manipulating the relevant aspects of perceived ingroup cooperativeness, it
would seem like ingroup cooperativeness does not affect explicit or implicit perceptions of
outgroup formidability.

4

Due to the relative nature of the D score, these results could also be interpreted as an association between seeing
MSU students as mental (i.e., intellectual) and having positive attitudes toward them.

14

A more likely explanation for the null effects in this study is that the participants may not
have found the short interaction in a computerized game to be diagnostic of their ingroup’s
cooperativeness. Some of the participants’ comments at the end of the sessions suggest that they
may have justified ingroup selfishness as a misunderstanding of the game’s instructions (by
themselves or their ingroup), or as coming from individuals who were not representative of their
ingroup. The short interaction provided by the public goods game may not be well suited to
change people’s opinions of ingroup members with which they interact frequently.
Another limitation of Study 1 is the conflict scenario. The cognitive mechanisms used to
deal with intergroup conflict were selected in situations of imminent conflict. Because the
conflict scenario they were exposed to was hypothetical, it may be the case that the relevant
cognitive mechanisms were not activated during the study.
Study 2 addresses the limitations of Study 1. Instead of using groups with which
participants have had frequent interactions, Study 2 will test the effect in a minimal-groups
context, which should be more amenable to the manipulation. Furthermore, testing the effect in a
minimal-groups context will indicate whether mere membership in a group is sufficient to bring
about the effect. The participants will be led to believe that the groups in the study will engage in
a physical confrontation. This should activate the relevant cognitive mechanisms to a greater
extent than when reading about a hypothetical scenario. Finally, in addition to Fessler et al.’s
self-report measures of formidability, Study 2 will use a reverse correlation procedure (RCP) to
assess representations of a typical outgroup member’s face. The RCP is a data-driven,
inconspicuous measure that produces visual renderings of faces that approximate peoples’ actual
representations. Thus, Study 2 will arguably be a stronger test of the effect of ingroup
cooperativeness on perceptions of outgroup formidability.

15

Study 2
Influencing people to think they will lose a confrontation affects how they represent their
opponents (Fessler et al., 2013). Because the RCP is data-driven, it offers a way of rendering
representations of outgroup males that are arguably more similar to individuals’ actual
representations. Furthermore, the procedure is less vulnerable to biases of self-reports (e.g.,
expectancy effects) since the measured traits are not mentioned to the participants generating the
classification images. Thus, the RCP may be an especially powerful method of detecting
differences in representations of outgroups members.
The procedure works by having participants judge whether blurry images of objects
belong to the category. To assess representations of the category dominant male, for example,
blurry face stimuli are generated by superimposing random visual noise on a base face with a
neutral expression (Dotsch & Todorov, 2012). The researchers then have participants judge
which of two face stimuli appearing side-by-side belong to a dominant male. Finally, the faces
participants deemed as belonging to a dominant man are superimposed to generate a
classification image. An independent sample can then rate the classification image on the traits
(e.g., formidability, trustworthiness) of the researchers’ choosing.
To implement the RCP, this study was comprised of two parts. In the first part,
participants were assigned to minimal groups, completed the ingroup relationship manipulation,
the formidability scales, as well as the RCP, which had participants select face stimuli that
appeared to belong to outgroup males. The stimuli selected by the participants were used to
generate, for each condition, an average face of all faces participants considered as that of an
outgroup member (i.e., classification image). In part two, a different sample of participants rated
the classification images on traits related to formidability.

16

Method
Part 1: Generating the classification images.
Participants. 272 MSU undergraduates were recruited from the subject pool (65%
female; 62% white, 18% Asian, 8% black). The sample size varies for each analysis depending
on how many students completed the measure. The total sample size grants the study
approximately 80% power to detect an effect of d = 0.35 (two-sided independent-samples t-test;
alpha = .05).
Materials and procedure. Participants completed the study in a room with eight cubicles
containing computers. They were told that the study was about perception and strength, and that
they would complete perception tasks and a strength competition. They were instructed not to
communicate with each other in any way before the strength competition.
First, they completed the minimal group assignment used by Ratner et al. (2014). This
consisted of a task in which participants estimated the number of dots that appeared on their
computer screens. They were told that people systematically vary in numerical estimation style
and that we wanted to see how they tended to perceive object quantities: if they either over- or
under-estimate the number of objects. After finishing the task, they received randomized
feedback indicating that they were under- or over-estimators. All participants were unaware of
the estimation styles of other participants. They were also told that the over- and underestimators would later compete in a series of strength tasks.
To manipulate the cooperativeness of ingroup members, participants took part in the
public goods game used in Study 1. They played against computerized players purported to be
the ingroup members in the room and that contributed either more than 90% or less than 10% of

17

their endowments to the public account. The participants did not know the actual investment
decisions of other participants, only of the computerized players.
Next, participants estimated the size and muscularity of the typical outgroup member
using Fessler and Holbrook’s (2013) rating scales. After estimating an outgroup member’s
formidability, they completed the reverse correlation procedure (RCP). Participants were told
that research indicates that people can tell whether other individuals are over- and underestimators even in poor visual conditions. Furthermore, they were told that in the task they would
look at images of past participants who completed the numerical estimation style task and
indicate who they believed were under- or over-estimators. The RCP consisted of 300 trials in
which participants saw two blurry faces. In each trial, they indicated which of the two faces they
believed belonged to an individual whose numerical estimation style is the opposite of theirs –
i.e., an outgroup member. As a manipulation check, they rated their ingroup’s supportiveness,
trustworthiness, and cooperativeness on a 7-point scale (1 = not at all; 7 = extremely). Finally,
they answered the demographic and debriefing questions used in Study 1.
Part 2: Rating the classification images. A separate sample of 128 students evaluated
the condition-level classification images in an online survey. They rated the faces on 9-point
scales (1 = not at all; 9 = extremely) of strength, dominance, and aggressiveness, as well as the
size and muscularity scales from Fessler and Holbrook (2013).
Results and Discussion
An independent-samples t-test revealed that those in the cooperative ingroup condition
(M = 4.99, SD = 1.09) judged their ingroup to be significantly more cooperative than those in the
selfish ingroup condition (M = 3.56, SD = 1.51), t(266) = 8.73, p < .001, d = 1.06 95% CI [0.80,
1.31]. Thus, it appears that participants’ opinions of the minimal groups were substantially

18

amenable to the short interactions of the public goods game. In the following sections, results
will be reported for self-reports of outgroup formidability and the ratings of faces generated with
the RCP.
Part 1: Self-reported formidability of the outgroup. Despite the effective
manipulation, there was no difference between those with a cooperative ingroup (size: M = 3.61,
SD = 1.42; muscularity: M = 3.14, SD = 1.58) and those with a selfish ingroup (size: M = 3.78,
SD = 1.28; muscularity: M = 3.21, SD = 1.47) in perceived size (t(267) = 1.01, p = .313, d = 0.12
95% CI [-0.12, 0.36]), or muscularity of the outgroup (t(267) = 0.41, p = .681, d = 0.05 95% CI
[-0.19, 0.29]). Nevertheless, there was a difference in perceived height (t(262) = 2.17, p = .031, d
= 0.26 95% CI [0.02, 0.50]) of the outgroup between those in the cooperative ingroup condition
(height: M = 70.07, SD = 3.49) and those in the selfish ingroup condition (height: M = 71.31, SD
= 5.40). These results give some indication that ingroup cooperativeness affects perceived
outgroup formidability. However, due to the multiple comparisons that were performed as well
as the weak effect only on perceived height, it is unclear whether this is more than just noise. The
effect of the manipulation on overall perceived outgroup formidability was tested separately for
individuals of different sexes and races/ethnicities. These analyses revealed that ingroup
cooperativeness is not more relevant for specific groups of individuals than others (Table 4).
Part 2: Representations of the outgroup generated with the RCP. While self-reports
of perceived formidability may not have been substantially influenced by the manipulation, the
representations generated with the RCP may be more sensitive to the interactions with ingroup
members. Thus, another test of the hypothesis can be done by comparing formidability ratings of
the RCP’s visual renderings from both conditions (Figure 1). A paired-samples t-test reveals that
the outgroup face images were not different on perceived size (t(126) = 0.85, p = .397),

19

muscularity (t(126) = 0.90, p = .371), height (t(125) = 0.39, p = .700), or aggression (t(126) =
1.85, p = .066) for those in the cooperative (size: M = 2.43, SD = 1.06; muscularity: M = 3.83,
SD = 0.73; height: M = 70.15, SD = 2.12; aggression: M = 3.82, SD = 0.97) and selfish ingroup
conditions (size: M = 2.33, SD = 1.00; muscularity: M = 3.76, SD = 0.96; height: M = 70.31, SD
= 2.88; aggression: M = 3.64, SD = 1.07).
The results seem to suggest that even though the manipulation was effective, it did not
influence perceptions of outgroup formidability. However, it may be the case the perceptions of
ingroup cooperativeness were associated with perceptions of ingroup strength in such a way that
the effect of cooperativeness on outgroup formidability was inhibited. To understand this,
consider what would be inferred about a person’s strength if they were cooperative or selfish.
Because nurturing interpersonal relationships is a strategy to access more resources, it may be
more likely to be used by those who have trouble accessing resources otherwise – i.e., less
formidable individuals. Thus, if ingroup members are cooperative, one may infer that they are
also physically weak. This would have opposing effects on one’s RHP – having cooperative
ingroup members increases RHP, but weak ingroup members decrease RHP – and perceptions of
outgroup formidability would remain unaffected. If this in fact is the case, those in the
cooperative ingroup condition should have rated their ingroup as weaker than did those in the
selfish ingroup condition. To test for this effect, an independent-samples t-test was performed on
ratings of ingroup formidability. Contrary to the rationale outlined, those in the cooperative
ingroup condition (M = 4.74, SD = 1.11) thought their ingroup was stronger than did those in the
selfish ingroup condition (M = 4.08, SD = 1.46), t(267) = 4.10, p < .001, d = 0.50 95% CI [0.26,
0.74]. Thus, despite the compounding effects that having ingroup members that are both
cooperative and strong should have on perceived outgroup formidability, the manipulation did

20

not affect how outgroup members were seen. In the next section, limitations of the present
studies, directions for future research, and the theory outlined in the introduction will be
discussed.

21

General Discussion
According to previous work, perceptions of outgroup formidability should be influenced
by how likely one is to reap benefits from a confrontation. In theory, this likelihood would be
greater if ingroup members signaled interest in coordination. Two experiments aimed to test the
hypothesis that having cooperative ingroup members leads to perceiving outgroup members as
less formidable. Both experiments did not confirm this hypothesis – perceptions of outgroup
formidability were independent from ingroup cooperativeness. The reasons for these null results
may be different for both studies.
The main limitation in Study 1 appears to be the manipulation of ingroup
cooperativeness. The ingroup in this study was the participants’ university, and because
interactions with ingroup members are so frequent, participants’ views of their ingroup’s
cooperativeness may not have been swayed by short interactions in the public goods game.
Furthermore, the conditions of the game – i.e., no face-to-face contact with ingroup members –
may have prevented participants from using the interactions as diagnostic information on their
ingroup’s cooperativeness.
To address these limitations of the first study, future studies could make the manipulation
of ingroup cooperativeness seem more diagnostic of their ingroup dispositions. This could be
achieved in a few ways. Some of them are: (1) increasing the clarity of the public good game’s
instructions, (2) adding more interactive features in the game, such as the option to comment on
other players’ behaviors, and (3) emphasizing that the players were selected at random from the
ingroup. These alterations may prevent participants from justifying the selfishness of their
ingroup as a misunderstanding of the game’s instructions or as the result of playing with
individuals who are not representative of their groups.

22

While using real groups has the advantage of increasing external validity, testing the
effect in a minimal-groups context provides useful information on the necessary conditions for
the effect to emerge. The ability of Study 2 to make such contributions may have been
compromised by its limitations. Perhaps the main issue with study concerns how much the
design provoked a threat of a physical conflict. Despite being told that a physical competition
between the groups would take place, participants may have been unmotivated to do well in the
competition leading them to not feel threatened. In the absence of feelings of threat, the cognitive
mechanisms involved in calculating the odds of winning a conflict might not have been
activated. To address this, a future study could increase the stakes of the competition either by
introducing prizes for winning or costs for losing. This might have the interactive effect of
increasing participants’ motivation to do well and their expectations of how motivated the
outgroup is. Under these conditions, the participants may take their ingroup’s cooperativeness
into consideration when judging the formidability of the outgroup.
While addressing the limitations discussed might be enough to bring about the effect, it
may be the case that there is no effect to be detected, at least under the conditions that could be
tested in a laboratory. A potential reason for this is that ingroup cooperativeness is probably not
the main variable the decision-making mechanism has to consider when calculating the odds of
winning an intergroup conflict. Number and strength of ingroup members, weapons or other
material means available to harm the outgroup, and a suitable terrain may take precedence when
considering the likelihood of winning a conflict. That is not to say that ingroup cooperativeness
does not play a role. Ability and willingness to strategize and coordinate may be signaled
through cooperative behaviors. However, those signals may play a more important role in
conditions under which other factors are less relevant. For example, in large-scale post-

23

agriculture-era battles when the number of individuals was too large to compare precisely, the
weapons were similar between groups, and the battle happened in a terrain previously scouted,
then signals of coordination and commitment to the ingroup (e.g., chants of war) may have
played an important role in individual decisions to keep fighting. Nevertheless, to determine the
role of ingroup cooperation in conflict situations, experiments such as the present ones must be
run, especially if possibilities are to be ruled out.

24

Conclusion
Despite an enthusiastic (and perhaps rightly so) endorsement of falsificationism in
psychological science, theories about human thought are not easily discarded. As noted by postPopperian philosophers, along with substantive theories come auxiliary hypotheses about the
workings of measurements and the adequacy of procedures in establishing the conditions
required by the theory for it to be relevant. If even one or two of these auxiliary hypotheses are
incorrect, then the attempt at testing the theory is uninformative as to whether the theory or the
auxiliary hypotheses (or both) are incorrect. When researching consequential topics such as
intergroup conflict, developing strong tests of the theories is crucial. And developing these tests
requires an iterative process of exploratory work determining which measures and procedures are
most adequate. While certainties are never achieved, and testing auxiliary hypotheses runs into
similar problems as testing the substantive theories, continuing to gather data on important
questions about the human mind is preferable to preemptively giving up. Hopefully, future
studies that better address the auxiliary hypotheses of the present studies will lend stronger tests
of the role of ingroup cooperativeness on how we see outgroup members.

25

APPENDIX

26

Table 1:
Word stimuli used in the attribute categories of the IATs
Positive

Negative

Physical

Mental

Joy

Agony

Athletic

Math

Love

Terrible

Strong

Brainy

Peace

Horrible

Basketball

Aptitude

Wonderful

Nasty

Run

Library

Pleasure

Evil

Agile

Scientist

Glorious

Awful

Jump

Idea

Laughter

Failure

Dance

Learn

Happy

War

Rhythm

Thinking

Muscular

Bookish

Football

Reading

27

Table 2:
Effect of the manipulation on perceived outgroup formidability (mean on Fessler’s scales) for
male, female, white, black, and Asian participants in Study 1
Cooperative

Selfish

ingroup

ingroup

M(SD)

M(SD)

t(df)

p

d [95 CI]

Male

-0.12(0.81)

-0.27(0.83)

1.04(128)

.299

0.16 [-0.14, 0.46]

Female

0.13(0.80)

0.12(0.78)

0.10(202)

.919

0.01 [-0.24, 0.26]

White

0.10(0.71)

0.06(0.80)

0.39(224)

.694

0.05 [-0.19, 0.29]

Black

-0.05(1.01)

-0.25(0.76)

0.67(35)

.506

0.16 [-0.30, 0.61]

Asian

0.02(1.06)

-0.08(0.94)

0.34(40)

.734

0.08 [-0.36, 0.51]

28

Table 3:
Effect of the manipulation on IAT D scores for male, female, white, black, and Asian participants
in Study 1

Male

Female

White

Black

Asian

Cooperative

Selfish

ingroup

ingroup

M(SD)

M(SD)

t(df)

p

d [95 CI]

Attitude IAT

0.38(0.32)

0.38(0.34)

0.03(117)

.977

0.004 [-0.30, 0.31]

Strength IAT

0.01(0.33)

-0.004(0.32)

0.31(117)

.760

0.05 [-0.25, 0.35]

Attitude IAT

0.39(0.30)

0.41(0.33)

-0.44(188)

.663

-0.06 [-0.31, 0.20]

Strength IAT

0.13(0.28)

0.12(0.31)

0.29(188)

.772

0.04 [-0.21, 0.29]

Attitude IAT

0.43(0.27)

0.42(0.33)

0.36(209)

.726

0.04 [-0.20, 0.29]

Strength IAT

0.10(0.30)

0.09(0.32)

0.05(209)

.961

0.01 [-0.24, 0.25]

Attitude IAT

0.33(0.34)

0.42(0.36)

-0.75(31)

.458

-0.17 [-0.63, 0.28]

Strength IAT

0.09(0.33)

-0.03(0.37)

0.95(31)

.351

0.22 [-0.24, 0.67]

Attitude IAT

0.13(0.32)

0.37(0.30)

-2.33(35)

.026

-0.52 [-0.96, -0.07]

Strength IAT

0.01(0.30)

0.06(0.31)

-0.43(35)

.672

-0.10 [-0.53, 0.34]

Note. Larger means indicate stronger associations between U of M and strength and U of M and
negativity.

29

Table 4:
Effect of the manipulation on perceived outgroup formidability (mean on Fessler’s scales) for
male, female, white, black, and Asian participants in Study 2
Cooperative

Selfish

Ingroup

Ingroup

M(SD)

M(SD)

t(df)

p

d [95 CI]

Male

-0.23(0.85)

0.06(0.93)

-1.55(89)

.125

-0.31 [-0.71, 0.09]

Female

-0.003(0.75)

0.06(0.71)

-0.56(169)

.578

-0.08 [-0.38, 0.21]

White

-0.15(0.74)

0.02(0.82)

-1.39(163)

.167

-0.21 [-0.51, 0.09]

Black

0.03(0.94)

0.31(0.78)

-0.77(20)

.450

-0.29 [-1.02, 0.45]

Asian

0.18(0.80)

0.02(0.74)

0.64(42)

.528

0.18 [-0.37, 0.72]

30

Figure 1:
Classification images of an outgroup member’s face generated with the reverse correlation
procedure for cooperative (A) and selfish ingroup (B) conditions

31

REFERENCES

32

REFERENCES

Becker, D. V., Kenrick, D. T., Neuberg, S. L., Blackwell, K. C., & Smith, D. M. (2007). The
confounded nature of angry men and happy women. Journal of Personality and Social
Psychology, 92(2), 179-190.
Benson-Amram, S., Heinen, V. K., Dryer, S. L., & Holekamp, K. E. (2011). Numerical
assessment and individual call discrimination by wild spotted hyaenas, Crocuta
crocuta. Animal Behaviour, 82(4), 743-752.
Berger, T. D., & Trinkaus, E. (1995). Patterns of trauma among the Neandertals. Journal of
Archaeological Science, 22(6), 841-852.
Bowles, S. (2009). Did warfare among ancestral hunter-gatherers affect the evolution of human
social behaviors?. Science, 324(5932), 1293-1298.
Brewer, M. B. (1999). The psychology of prejudice: Ingroup love and outgroup hate?. Journal of
Social Issues, 55(3), 429-444.
Buss, D. M., & Shackelford, T. K. (1997). Human aggression in evolutionary psychological
perspective. Clinical Psychology Review, 17(6), 605-619.
Crofoot, M. C., & Wrangham, R. W. (2010). Intergroup aggression in primates and humans: the
case for a unified theory. In Mind the Gap (pp. 171-195). Springer Berlin Heidelberg.
Dotsch, R., & Todorov, A. (2012). Reverse correlating social face perception. Social
Psychological and Personality Science, 3(5), 562-571.
Fessler, D. M., & Holbrook, C. (2013a). Bound to lose: Physical incapacitation increases the
conceptualized size of an antagonist in men. PloS One, 8(8), e71306.
Fessler, D. M., & Holbrook, C. (2013b). Friends shrink foes the presence of comrades decreases
the envisioned physical formidability of an opponent. Psychological Science, 24(5), 797802.
Fessler, D. M., & Holbrook, C. (2014). Marching into battle: Synchronized walking diminishes
the conceptualized formidability of an antagonist in men. Biology Letters, 10(8),
20140592.
Fessler, D. M., Holbrook, C., & Gervais, M. M. (2014). Men’s physical strength moderates
conceptualizations of prospective foes in two disparate societies. Human Nature, 25(3),
393-409.

33

Fessler, D. M., Holbrook, C., Pollack, J. S., & Hahn-Holbrook, J. (2014). Stranger danger:
Parenthood increases the envisioned bodily formidability of menacing men. Evolution
and Human Behavior, 35(2), 109-117.
Fessler, D. M., Tiokhin, L. B., Holbrook, C., Gervais, M. M., & Snyder, J. K. (2014).
Foundations of the Crazy Bastard Hypothesis: Nonviolent physical risk-taking enhances
conceptualized formidability. Evolution and Human Behavior, 35(1), 26-33.
Firestone, C., & Scholl, B. J. (2015). Cognition does not affect perception: Evaluating the
evidence for ‘top-down’ effects. Behavioral and Brain Sciences, 20, 1-77.
Fox, E., Lester, V., Russo, R., Bowles, R. J., Pichler, A., & Dutton, K. (2000). Facial expressions
of emotion: Are angry faces detected more efficiently? Cognition & Emotion, 14(1), 6192.
Greenwald, A. G., McGhee, D. E., & Schwartz, J. L. (1998). Measuring individual differences in
implicit cognition: The implicit association test. Journal of Personality and Social
Psychology, 74(6), 1464-1480.
Hausfater, G. (1974). Dominance and reproduction in Baboons (Papio cynocephalus).
Contributions to Primatology, 7, 1-150.
Houston, A. I. & McNamara, J. M. (1999). Models of Adaptive Behaviour: An Approach Based
on State. Cambridge: Cambridge University Press.
Leach, C. W., Van Zomeren, M., Zebel, S., Vliek, M. L., Pennekamp, S. F., Doosje, B., ... &
Spears, R. (2008). Group-level self-definition and self-investment: A hierarchical
(multicomponent) model of in-group identification. Journal of Personality and Social
Psychology, 95(1), 144-165.
Marwell, G., & Ames, R. E. (1979). Experiments on the provision of public goods. I. Resources,
interest, group size, and the free-rider problem. American Journal of Sociology, 13351360.
Maynard Smith, J., & Parker, G. A. (1976). The logic of asymmetric contests. Animal
Behaviour, 24(1), 159-175.
McDonald, M. M., Asher, B. D., Kerr, N. L., & Navarrete, C. D. (2011). Fertility and intergroup
bias in racial and minimal-group contexts: Evidence for shared
architecture. Psychological Science, 22(7), 860-865.
Mitani, J. C., Watts, D. P., & Amsler, S. J. (2010). Lethal intergroup aggression leads to
territorial expansion in wild chimpanzees. Current Biology, 20(12), R507-R508.

34

Neuberg, S. L., Kenrick, D. T., & Schaller, M. (2011). Human threat management systems: Selfprotection and disease avoidance. Neuroscience & Biobehavioral Reviews, 35(4), 10421051.
Packer, C., & Pusey, A. E. (1982). Cooperation and competition within coalitions of male lions:
kin selection or game theory? Nature, 296(5859), 740-742.
Ratner, K. G., Dotsch, R., Wigboldus, D. H., van Knippenberg, A., & Amodio, D. M. (2014).
Visualizing minimal ingroup and outgroup faces: Implications for impressions, attitudes,
and behavior. Journal of Personality and Social Psychology, 106(6), 897-911.
Schaller, M., & Neuberg, S. L. (2008). Intergroup prejudices and intergroup conflicts. In:
Crawford, C.; Krebs, DL. (Ed.), Foundations of Evolutionary Psychology (pp. 399-412).
Mahwah, NJ: Lawrence Erlbaum Associates.
Schuett, G. W. (1997). Body size and agonistic experience affect dominance and mating success
in male copperheads. Animal Behaviour, 54(1), 213-224.
Schupp, H. T., Öhman, A., Junghöfer, M., Weike, A. I., Stockburger, J., & Hamm, A. O. (2004).
The facilitated processing of threatening faces: An ERP analysis. Emotion, 4(2), 189-200.
Tajfel, H., Billig, M. G., Bundy, R. P., & Flament, C. (1971). Social categorization and
intergroup behaviour. European Journal of Social Psychology, 1(2), 149-178.
Thomas, K. A., DeScioli, P., Haque, O. S., & Pinker, S. (2014). The psychology of coordination
and common knowledge. Journal of Personality and Social Psychology, 107(4), 657-676.
Tooby, J., & Cosmides, L. (1988). The evolution of war and its cognitive foundations. Institute
for Evolutionary Studies Technical Report, 88(1), 1-15.
Thorpe, I. J. (2003). Anthropology, archaeology, and the origin of warfare. World
Archaeology, 35(1), 145-165.
Wilson, M. L., Hauser, M. D., & Wrangham, R. W. (2001). Does participation in intergroup
conflict depend on numerical assessment, range location, or rank for wild
chimpanzees?. Animal Behaviour, 61(6), 1203-1216.

35