Than Intergroup Conflict
Simon Jonas Hadlich
University of Amsterdam
Much of human life takes place in groups, and membership in strong and successful groups allows us to achieve economic, political, and social goals. It is likely that humans are evolu-tionarily adapted for life in cooperative groups, and to sustain their in-group against threats. Such adaptations may occur in the context of group augmentation and bounded generalised reciprocity. While it has frequently been proposed that intergroup conflict promotes in-group favouritism and group-bounded cooperation, we posit that parochial cooperation is instead provoked by threats to the in-group, and may be lower during intergroup conflict when one’s group is motivated by predation. We test this hypothesis in the laboratory using a modified intergroup Predator-Prey Game in which groups of subjects face a threat or opportunity of social or non-social nature. We find that parochial cooperation is greater in defensive than in predatory groups, and that this gap widens during intergroup conflict. This finding supports the primacy of threat over conflict as an antecedent of parochial cooperation, though it does highlight the particular rule of intergroup conflict.
Late in the autumn of 2012, after a destructive path through the Caribbean, the hurricane Sandy hit the East Coast of the United States, devastating New York City and surrounding areas. The storm killed at least 43 people in the city and left hundreds of thousands temporarily home-less (Associated Press,2012). Soon, however, news of the havoc wreaked on the city by the forces of nature were ac-companied by stories of kindness between its residents (e.g. Conradt,2012). When many people were left without power to charge their phones, those who still had access to electric-ity led extension cords onto the street to offer help(Klimas, 2012). Later, a 78 year old pensioner, Barbara Joyce, had her badly damaged house rebuilt by strangers “who simply wanted to offer a helping hand” (Trepasso & Lemire,2013). 11 years earlier, New York was hit by another, even more deadly force when terrorists steered two passenger planes into the towers of the World Trade Center. 2,996 people were killed, and the strike left a gaping hole in the city’s skyline (Kean et al.,2004). Although the attack would lead to two wars, it was also followed by a spike of altruism towards
I would like to thank Daniel Balliet for his patient guidance and many insightful discussions, and Carsten de Dreu for his inspiration and challenges—I have learned much from both. Further thanks go to Femke ten Velden and Astrid Homan for their helpful comments on the proposal for this study. With Florian Wanders and Bobby den Bezemer I have spent many an evening discussing science, statis-tics, and life, and their input is visible throughout this thesis. I gratefully acknowledge a scholarship from the Studienstiftung des deutschen Volkes. Contact: simon@simoncolumbus.com.
fellow New Yorkers (Azarow et al.,2003). Blood donations increased sharply, and people made more charitable contribu-tions following the attacks (Azarow et al.,2003). Throughout the country, Americans responded with pro-social behaviour and a persistently heightened sense of patriotism to the terror (Landau et al.,2004;Yum & Schenck-Hamlin,2005;Huddy & Feldman,2011).
The responses of New Yorkers to the devastation by hurri-cane Sandy and the terror of the 9/11 attacks shows the power of threats and attacks to forge groups together, to heighten identification with one’s in-group and give rise to parochial altruism (Stein,1976;Baumeister & Leary,1995;de Dreu, 2010a; de Dreu, Balliet, & Halevy,n.d.). It has long been thought that mere conflict between groups gives rise to these feelings and behaviours (Sherif,1956; Pettigrew,1998; de Dreu,2010a). In contrast, we posit that is threat—whether social or non-social in its cause—rather than intergroup con-flict more generally that gives rise to parochial cooperation; and that intergroup conflict may be associated with decreased parochial cooperation when the aim of the group is predatory rather than defensive. We test this hypothesis using a modi-fied intergroup Predator-Prey Game, in which groups of sub-jects face a threat or opportunity of either social or non-social nature.
The structure of this paper is as follows: first, we describe the evolutionary mechanisms which may have given rise to parochial cooperation in humans. Then, we survey the liter-ature on in-favouritism and discuss how threat and conflict may motivate parochial cooperation; as well as the role of empathy and moral hypocrisy. Subsequently, we introduce the notion of asymmetric intergroup conflict and the
inter-group Predator-Prey Game. This is followed by the methods, results, a discussion of the findings, and a short conclusion.
The Benefits of Group Life
Membership in strong, cooperative groups is individu-ally beneficial. People join consumer and producer inter-est groups to attain political influence and financial bene-fits (Hansen,1985) and create and maintain institutions for collective action to sustain resources and livelihoods (E. Os-trom, 1990). Members of cooperative groups face better prospects in the work place (Jehn & Mannix,2001), are more likely to survive warfare (Bowles,2009), and are more able to attain their political goals (Hansen,1985).
Just as today, during our evolutionary ancestry belonging to a strong group would have benefitted an individual: mem-bers of a strong group may have an advantage in foraging or a greater likelihood of prevailing in intergroup conflict (West, Griffin, & Gardner,2007a;Choi & Bowles,2007; Bowles, 2009;de Dreu et al.,n.d.). The mere presence of other group members may be beneficial if their selfish behaviour, such as territorial defense, provides a positive by-product; such ‘passive’ group augmentation constitutes a by-product mutu-alism (Kokko, Johnstone, & Clutton-Brock,2001; Clutton-Brock, 2002; Bergmüller, Johnstone, Russell, & Bshary, 2007). However, individuals may also benefit from help-ing unrelated group members because a larger group in-creases the likelihood of survival, foraging success, and vic-tory in intergroup conflict (Clutton-Brock,2002;West et al., 2007a). Such benefits may arise from pseudo-reciprocity if the selfish behaviour of those helped benefits the helper (for example, their investment in territorial defense); or they may arise from indirect reciprocity if helpers later become the recipients of help themselves (Bergmüller et al.,2007). Reputation-building is not necessary for this to occur: merely increasing the size of the group increases the number of po-tential helpers and thus the likelihood to receive help oneself (Bergmüller et al.,2007).
However, although group membership provides benefits to the individual, members of a group face a social dilemma: although all members of a group are better off if all cooper-ate, each individual can improve their lot relative to the oth-ers by defecting (Dawes,1980). Cooperation in social dilem-mas poses an evolutionary puzzle, because individuals who defect would seem to have greater reproductive fitness than individuals who cooperate, thus preventing cooperation from spreading in a population (West et al.,2007a). Here, it may be helpful to take a step backwards. Fundamentally, cooper-ation may have evolved through direct fitness benefits (which accrue directly to the cooperative individual) or through in-direct fitness benefits (which accrue to individuals who are genetically related to the cooperative individual) (West et al., 2007a). In any case, the combined effect of direct and indi-rect fitness benefits must outweigh the costs for cooperation
to evolve (Hamilton,1964a,1964b;West et al.,2007a;West, El Mouden, & Gardner,2011;Birch & Okasha,2014). Var-ious mechanisms can sustain cooperation by changing the parameters of this formula, i.e. the relatedness between in-teraction partners and the costs and benefits of cooperation (Gardner & West,2004;West et al.,2007a).
Inclusive fitness can give rise to cooperation among rela-tives (kin selectionHamilton,1964a; Abbot et al., 2011).1
However, humans frequently cooperate with non-relatives, which cannot be explained by kin selection. Direct reci-procity enables cooperation among non-relatives in repeated interactions (Trivers,1971;Axelrod & Hamilton,1981); and punishment (i.e., negative direct reciprocity) can deter defec-tion (Yamagishi,1986;E. Ostrom,1990;E. Ostrom, Walker, & Gardner,1992;Gardner & West,2004). Neither of these mechanisms require groups; nor do they explain why human cooperation is often group-bounded. In contrast, group selec-tion accounts argue that cultural (Boyd & Richerson,1982; Zefferman & Mathew,2015;Richerson et al.,n.d.) or genetic (Choi & Bowles,2007;Bowles,2009) differences between groups have been the target of group-level selection, which has given rise to a predisposition to behave cooperatively. Alternatively, reputation-based indirect reciprocity explains parochial cooperation without direct reference to group sur-vival (Yamagishi, Jin, & Kiyonari,1999;Nowak & Sigmund, 1998; Leimar & Hammerstein, 2001; Nowak & Sigmund, 2005; Balliet, Wu, & de Dreu, 2014), though benefits de-rived from group survival may partly explain why individuals bestow positive reputations upon their group-serving peers (Alexander,1974).
Group Selection Accounts of the Evolution of Human Co-operation
Bowles and co-authors (Choi & Bowles, 2007; Bowles, 2009) have proposed genetic group selection driven by in-tergroup conflict as the origin of large-scale human coopera-tion. In these models, genes spread as groups prevail in inter-group conflict because inter-groups are genetically differentiated. Specifically, groups in which some members have a genetic predisposition for cooperativeness are more likely to prevail in intergroup conflict than groups of defectors, allowing for such genes to spread. However, this relies on genetic di ffer-ences between groups (Maynard Smith,1976;West, Griffin, & Gardner,2007b; Wrangham & Glowacki,2012), yet the extent of genetic differentiation between groups is debated (Bowles,2009; Hill et al.,2011), as is the frequency of in-tergroup conflict (Fry & Söderberg,2013). Moreover, con-flict between relatives may counteract indirect fitness bene-fits (West et al.,2011). Overall, it appears unlikely that the 1Though limited dispersal can engender local competition
whose costs outweigh the benefits of kin selection (Murray & Ger-rard,1984;West et al.,2006).
assumptions of genetic differentiation due to low migration and simultaneous high mortality due to intergroup conflict accurately reflect the conditions during recent human pre-history (Hill et al.,2011;Fry & Söderberg,2013).
In contrast to genetic group selection models, models of cultural group selection show that an inclination to behave cooperatively can arise from a more general predisposition for conformist social learning (Boyd & Richerson, 1982). Thus, if a general tendency to mimic others evolved (e.g. as a solution to the problem of costly information acquisi-tion), this could give rise to social norms. Some of these norms may prescribe cooperative behaviour. If variation in such norms across populations is associated with victory and defeat in intergroup conflict, cultural group selection would favour norms that promote a group’s fate. These norms would be most likely to be cooperative norms. Then, it has been argued, cooperative norms would create an environment which selects genetically for predispositions that are adaptive in cooperative group life, including cooperation and punish-ment of defectors.
Cultural group selection models have been under particu-lar scrutiny (Hagen & Hammerstein,2006;André & Morin, 2011;Raihani & Bshary,2015b), and more recent work in-dicates that several key assumptions of cultural group se-lection do not hold up. In particular, it is assumed that so-cial learning is conformist, but non-conformist transmission dominates conformist transmission in evolutionary models (Eriksson & Coultas,2009). Worse, conformist transmission may in fact counteract the evolution of cooperative tenden-cies (Lehmann, Feldman, & Foster,2008).Indeed, we know that humans are keenly aware of the situational limits to the benefits of conformity (Morgan, Rendell, Ehn, Hoppitt, & Laland,2012;Morin,2014), and may even behave in strate-gically non-conformist ways to their own benefit (van Kleef, Homan, Finkenauer, Gundemir, & Stamkou,2011;Bellezza, Gino, & Keinan, 2014; van Kleef, Wanders, Stamkou, & Homan,2015;Wanders,2015).Raihani and Bshary(2015b) conclude that “it is apparently easier for costly helping strate-gies to evolve under genetic systems of inheritance rather than via culture”, yet genetic group selection driven by in-tergroup conflict is itself an unlikely explanation for human cooperative behaviour.
Groups as a Container for Indirect Reciprocity
Indeed, it seems unnecessary to invoke group selection to explain parochial cooperation when it can be explained more parsimoniously through individual-level selection. By-product benefits derived from the selfish behaviour of peers may already motivate group-serving actions (Kokko et al., 2001;Clutton-Brock,2002;Bergmüller et al.,2007). In ad-dition, an ability to keep track of the reputations of individ-uals can give rise to directed indirect reciprocity, in which others cooperate preferentially with an actor who has
previ-ously behaved cooperatively towards third parties. The ben-efits of being part of a strong group may motivate individu-als to form, spread, and remember the reputations of others (Alexander, 1974); and the benefits of a positive coopera-tion may motivate cooperative behaviour even towards third parties with whom future interactions are unlikely (Nowak & Sigmund, 1998; Yamagishi et al., 1999; Nowak & Sig-mund,2005). Groups provide a container for such indirect reciprocity mechanisms that enable and sustain cooperation among their members (Yamagishi et al.,1999).
If individuals can distinguish between others who may be more or less likely to cooperate with them, they should pref-erentially and more cooperatively interact with those who are likely to behave cooperatively towards them (Alexander, 1974; Roberts,1998; Nowak & Sigmund,1998). The oth-ers’ history of behaviour towards third parties (their ‘im-age score’, Nowak & Sigmund, 1998) may be an indica-tion of this propensity to cooperate in the future (Nowak & Sigmund, 1998; Lotem, Fishman, & Stone, 1999). In-deed, in laboratory experiments, humans behave more co-operatively towards those who have a history of cooperative behaviour towards third parties (Wedekind & Milinski,2000; Milinski, Semmann, Bakker, & Krambeck,2001; Seinen & Schram,2006;Engelmann & Fischbacher,2009) and prefer-entially choose them as partners for interactions (Sylwester & Roberts, 2010, 2013). It has been argued that a image scoring-based strategy is evolutionarily unstable because it does not allow for defection against uncooperative individu-als (Leimar & Hammerstein,2001). A ‘standing’-based rule takes into account the history of the previous interaction part-ners when forming the reputation of an individual to avoid this problem (Leimar & Hammerstein,2001). However, ex-perimental evidence indicates that more humans rely on im-age scoring rather than on standing when deciding whether to cooperate with others (Milinski et al., 2001; Seinen & Schram,2006;Ule, Schram, Riedl, & Cason,2009).
Where potential interaction partners discriminate between individuals based on their reputation, these individuals have an incentive to behave in such a way as to gain a reputa-tion that ensures successful interacreputa-tions in the future. Indeed, there is strong evidence that individuals behave more coop-eratively when they expect to build a reputation (Yamagishi et al.,1999;Mifune, Hashimoto, & Yamagishi,2010;Nettle et al.,2013;Balliet et al.,2014) or actually can build a rep-utation (Milinski, Semmann, & Krambeck,2002; Seinen & Schram,2006). However, unconditional cooperation would easily be exploited—a more viable strategy might be to be known as ‘nice, but not gullible’ (Trivers, 1971; Brandt, Hauert, & Sigmund,2003). Indeed, reputation is unlikely to be one-dimensional (Horita,2010;Ozono & Watabe,2012; Columbus, 2013; Gordon, Madden, & Lea, 2014; Raihani & Bshary, 2015a); and individuals may strategically build a reputation as being not just cooperative, but also willing
to punish those who behave uncooperatively (Brandt et al., 2003;Raihani & Bshary,2015a). Thus, the ability to build a reputation with others and the expectation of indirect reci-procity motivates individuals to behave cooperatively, but may also lead them to punish others who do not.
For an investment in a positive reputation to pay off, pop-ulations must be stable enough that the ‘investment’ of co-operation is repaid through advantages in coco-operation and partner choice (Nowak & Sigmund, 2005; Roberts, 2015). Group-structured populations with limited migration pro-vide such conditions, allowing for bounded generalised reci-procity (Yamagishi et al., 1999; Masuda, 2012). Models show that withgroup cooperation can arise based on in-direct reciprocity when individuals track the reputation of in-group members, but assign a group-based reputation to out-group members (Masuda,2012), a phenomenon known as out-group homogeneity (Judd & Park,1988; T. M. Os-trom & Sedikides,1992). Indeed, humans are more attentive to the behaviour of in-group members (Reese, Steffens, & Jonas,2013), judge their moral transgressions more harshly (Fessler et al.,2015, but see ;Valdesolo & DeSteno,2007), and may punish their defection more harshly (Shinada, Ya-magishi, & Ohmura,2004, but see ;Bernhard, Fehr, & Fis-chbacher,2006;Bernhard, Fischbacher, & Fehr,2006). Con-versely, they also cooperate more with in-group members than with out-group members, but do not distinguish between out-group members and unattached strangers (Balliet et al., 2014). This in-group favouritism is particularly pronounced when indirect reciprocity is possible (Balliet et al., 2014). Thus, there is strong evidence that groups provide a container for indirect reciprocity, which drives parochial cooperation in humans.
Parochial Cooperation in Humans
Cooperation lies at the heart of human life: it shapes everyday human experience, and enables social life as we know it. However, humans also distinguish between group members and others, and cooperate more with in-group members than with out-in-group members or unattached strangers (95% CI for meta-analytic d: [.27, .38] and [.24, .36], resp. Balliet et al.,2014;Allport,1954;Tajfel, Billig, Bundy, & Flament,1971;Brewer,1999; Hewstone, Rubin, & Willis,2002). Two major explanations have been put for-ward to explain this in-group favouritism: the social identity approach and the theory of bounded generalised reciprocity.
The social identity approach integrates the contributions of social identity theory (Tajfel & Billig, 1974; Tajfel & Turner,1979) and self-categorisation theory (Turner, Hogg, Oakes, Reicher, & Wetherell,1987). This approach centers on the view that categorising oneself and others into in- and out-group(s) gives rise to intergroup discrimination and, in particular, in-group favouritism (Balliet et al.,2014). One of the core elements of the social identity approach is the
meta-contrast principle (Turner et al.,1987), according to which in-group favouritism requires a salient out-group. Indeed, in the absence of an out-group, people would not categorise themselves as members of an in-group, and thus in-group favouritism should not occur (Balliet et al., 2014). How-ever, once an out-group is salient, people self-categorise as members of an in-group and derive their social identity—part of their self-concept—from membership in this social group and the emotional significance they attach to their member-ship (Tajfel & Billig, 1974; Balliet et al., 2014). Because people strive for a positive social identity, self-categorisation motivates them to positively distinguish their in-group from salient out-groups; i.e., to engage in in-group favouritism.
The theory of bounded generalised reciprocity theory pro-vides an alternative explanation of parochial cooperation (Yamagishi et al.,1999). According to this theory, the ben-efits of group membership in human evolutionary ancestry have shaped human social behaviour (Alexander,1974; Bal-liet et al., 2014). Specifically, groups act as containers for reputation-based indirect reciprocity (Yamagishi et al.,1999; Masuda,2012). Consequently, humans identify with groups to which they belong. They are motivated to maintain a rep-utation as good cooperators and to be attentive to the repu-tations of in-group members, and expect their peers to pref-erentially cooperate with those in-group members who have such a positive reputation (Yamagishi et al.,1999;Balliet et al.,2014). Recent work indicates that the theory of bounded generalised reciprocity explains in-group favouritism in co-operation better than the social identity approach; in partic-ular because it can account for in-group favouritism in the absence of an out-group and the lack of distinction between out-group members and unattached strangers (Balliet et al., 2014).
Group Identity, Interdependence, and Reputation in Bounded Generalised Reciprocity
Bounded generalised reciprocity argues that humans are adapted to a) have a generalised trust that in-group members will behave cooperatively; b) be motivated to attain a reputa-tion as good cooperators among in-group members; c) expect that a good reputation will elicit increased cooperation from in-group members, even if they have not directly interacted with them before (Balliet et al., 2014). It follows that in-dividuals will show in-group favouritism even in the absence of outcome interdependence (Yamagishi et al.,1999;Rabbie, Schot, & Visser, 1989), but only when reputation-building is possible. Here, I shortly summarise how bounded gener-alised reciprocity is dependent on, and affected by variation in interdependence and reputation mechanisms.
The theory of bounded generalised reciprocity predicts some in-group favouritism even in the absence of outcome interdependence (Yamagishi et al.,1999;Balliet et al.,2014), as long as the actor has an expectation that he or she is
build-ing a reputation with group members, and that these in-group members will discriminate based on the actor’s rep-utation. Yamagishi et al. (1999) call this multilateral fate control; i.e., one’s outcomes are determined by third parties, who are part of the group. However, it is expected that in-group bias is increased with interdependence. Both of these predictions are confirmed in the meta-analysis by 95% CI for d: [.07, .37] and [.30, .47], resp. Balliet et al. (2014). However, it is also implied that interdependence with the out-group may reverse biases; and indeed, individuals with direct outcome interdependence with out-group members appear to show out-group favouritism (Rabbie et al.,1989).
Thus, parochial cooperation is dependent on the ability to build a reputation with in-group members. Indeed,Balliet et al. (2014) found no in-group favouritism in cooperative games where the actor’s group membership was known to them, but not to the recipient (meta-analytic 95% CI for d: [-.12, .21]). However, the absence of reputation mechanisms in the explicit structure of an experimental set-up should not be taken to imply the absence of reputational concerns. There is ample evidence that subjects fail to fully grasp (or believe) anonymity in laboratory settings. Yamagishi et al.(1999), for example, report that a subset of subjects expected repu-tation effects when these were not possible, and that these subjects, but not others, showed in-group favouritism under anonymity.2 What matters is thus not perceived anonymity, but reputational concerns. However, reputational concerns should matter less if individuals can rely on direct (rather than merely indirect) reciprocity. Indeed, in economic games that allow for direct reciprocity (compared to ones that do not), participants show less in-group favouritism; indicat-ing lesser concern for their reputation as good cooperators with in-group members (meta-analytic 95% CI: [.14, .38] and [.33, .51], resp.Balliet et al.,2014).
Conflict and (or?) Threat as Motivator of Parochial Co-operation
The mechanisms of bounded generalised reciprocity can sustain cooperation within groups and in absence of an out-group. However, it is well-documented that intergroup conflict increases identification with the in-group (Cashdan, 2001;Paolini, Harwood, & Rubin,2010; Whitehouse, Mc-Quinn, Buhrmester, & Swann Jr.,2014) and leads members to subscribe to group norms and practices (Chudek & Hen-rich,2011;Benard,2012), to work together (Erev, Bornstein, & Galili,1993; Bornstein & Ben-Yossef,1994; Bornstein, Gneezy, & Nagel,2002; Bornstein & Gilula,2003), and to sacrifice immediate self-interest to promote their in-group (Rusch,2014). Why would this be? In the following, I will discuss two relevant hypotheses.
A set of group-selection models center on the claim that intergroup conflict meant that members of less cooperative groups had worse chances of survival, thus creating selection
pressures for parochial cooperation (Choi & Bowles,2007; Bowles,2009). If intergroup conflict was the driving force giving rise to parochial cooperation, we would expect that cooperation in conflict would have been particularly selected for; and that parochially cooperative behaviours would be evoked by cues of intergroup conflict. At the proximal level, this is in line with realistic conflict theories, according to which intergroup conflict is founded in deprivation of one group by another of some resource (de Dreu,2010a). Fol-lowing this argument, we would expect intergroup conflict to increase cooperation within groups, relative to situations in which intergroup conflict is absent.
Hypothesis 1a: Intergroup conflict increases cooperation, relative to non-social situations.
We should note that social identity theory makes a similar prediction: since mere presence of an out-group gives rise to identification with the in-group, and this identification mo-tivates attitudes, thoughts, and behaviours that promote the group over the out-group, intergroup conflict should in-crease parochial cooperation relative to a non-social setting, in which no out-group is present (Tajfel & Turner, 1979). However, in contrast to realistic conflict theories, the social identity approach does not require actual intergroup conflict for this effect to occur. Thus, these theories differ in the causes, but not the predictions about parochial cooperation in intergroup conflict.
An alternative argument is offered by (de Dreu et al.,n.d.): parochial cooperation in intergroup conflict is driven by fear and greed (Bornstein & Gilula, 2003; Bornstein, 2003; de Dreu et al., 2010; de Dreu,2010a,2010b; de Dreu, Giffin, Krikeb, Prochazkova, & Columbus,2015) According to this view, threats to the in-group motivate greater parochial coop-eration than do collective opportunities (de Dreu, Giffin, et al.,2015;Rusch & Löscher,2013;Rusch,2014). Threats to the in-group cause fear, which some evidence indicates elic-its fast, automatic helping behaviours towards the in-group (de Dreu et al., 2010; de Dreu, Scholte, van Winden, & Ridderinkhof,2014; de Dreu, Dussel, & ten Velden,2015). In contrast, according to this model, opportunities motivate greed, which appears to be more calculated and slow (de Dreu, Giffin, et al., 2015). This matches with the well-established concept of loss aversion—the general tendency of humans to weight losses more than gains, and falling be-hind more than coming out ahead (Kahneman, Knetsch, & Thaler,1990;Ariely, Huber, & Wertenbroch,2005;Camerer, 2005;Novemsky & Kahneman,2005;Gill & Prowse,2012). Loss aversion would have been adaptive in an environment in 2This may be maladaptive, in the sense of a mismatch between
environment and adapted mind or misapprehension of the situation (Hagen & Hammerstein, 2006), or a cautious strategy given that perceptions of anonymity may be wrong (Tennie,2012;Raihani & Bshary,2015b).
which survival is uncertain (McDermott, Fowler, & Smirnov, 2008; Li, Kenrick, Griskevicius, & Neuberg, 2012; Levy, 2012). As McDermott et al. (2008, p. 337) write, “when survival is uncertain, marginal losses prove more critical for reproductive success than marginal gains”. It is thus that the potential cost of group extinction is greater than the potential gain from collective predation, and consequently we expect fear to motivate greater parochial cooperation than greed.
Hypothesis 1b: Individuals show more parochial cooperation in the face of threats to the in-group rather than collective opportunities, irrespective of the source of the threat or oppor-tunity.
Note that this argument is solely concerned with the costs and benefits to the in-group (resp. its members). In con-trast to group selection models, it does not relate the fit-ness of the members the in-group relative to a particular out-group; but rather to the entire population. Thus, we would expect humans to be sensitive to threats to their in-group, but not to the (social) nature of the threat. This is sup-ported by findings that both intergroup conflict and famine are associated with increased identification with the in-group (Cashdan,2001), and that adverse climatic conditions are as-sociated with higher levels of in-group favouritism across so-cieties (van de Vliert,2010;van de Vliert & Postmes,2012). Indeed, even proponents of the hypothesis that intergroup conflict drove group selection for parochial cooperation ad-mit that other factors than intergroup conflict may have been equally or more frequent causes of group extinction (Bowles, 2008,2009;Trinkaus,2011).3
In-group Love and Out-group Hate in Intergroup Conflict
In-group favouritism can take the form of in-group love and out-group hate (Brewer,1999;Halevy, Bornstein, & Sa-giv, 2008; de Dreu et al., n.d.). However, in-group love dwarfs out-group hate as a motivation for parochial coop-eration in intergroup conflict (Brewer,1999; Halevy et al., 2008; Abbink, Brandts, Herrmann, & Orzen, 2012; Balliet et al., 2014; Weisel & Böhm, 2015; de Dreu et al., n.d.). This is a consequence of bounded generalised reciprocity: helping in-group members, but not harming out-group mem-bers, builds a reputation as a good cooperator (Halevy et al., 2008; Mifune et al., 2010; Halevy, Chou, Cohen, & Liv-ingston,2012).4 The distinction between in-group love and
out-group hate is linked to that between fear-driven defense and greed-driven predation. Although both in-group love and out-group hate may motivate defense and predation, it seems intuitive that in-group love would motivate mostly defense, and out-group hate mostly predation. Indeed, predation ap-pears to be sensitive to individual pay-offs and susceptible to free-riding (de Dreu, Giffin, et al.,2015; Mathew & Boyd,
2011, 2014), motivated by individual greed rather than in-group love. Thus, collective predation is likely opportunis-tic, with a marginal role for out-group hate (Abbink et al., 2012). Consequently, we expect to see more free-riding and a greater variance between subjects among members of preda-tory groups than of defensive ones:
Hypothesis 2: Free-riding is more frequent among members of predatory groups than of de-fending groups.
Hypothesis 3: The between-subjects variance in parochial cooperation is greater among mem-bers of predatory groups than of defending groups.
In contrast, group defense is likely motivated by in-group love. Acts of heroic self-sacrifice, for example, are observed more in defense than in offense (Rusch & Löscher, 2013). However, in-group love is a consequence of bounded gener-alised reciprocity; i.e., motivated by reputational concerns. A consequence of this is that we would expect members of groups under threat to be more sensitive to their reputa-tion than members of predatory groups. Indeed, experimen-tal evidence indicates that expressing in-group love, but not expressing out-group hate, garners a prestigious reputation (Halevy et al.,2012), which provides a path to power (Halevy et al.,2012;Cheng, Tracy, Foulsham, Kingstone, & Henrich, 2013). This provides the second motivation for Hypothesis 1b, as well as Hypothesis 4b:
Hypothesis 4a: Reputational concerns moti-vate cooperation.
Hypothesis 4b: Reputational concerns motivate cooperation more strongly in members of defen-sive than of predatory groups.
It is worth noting a conflicting hypothesis here. The Male Warrior Hypothesis posits that the willingness of males to en-gage in warfare—in other words, warriorship—has been un-der sexual selection (Johnson et al.,2006;van Vugt, de Cre-mer, & Janssen,2007; Sell et al.,2009;McDonald, Navar-rete, & Van Vugt, 2012; Rusch, Leunissen, & van Vugt, 2015;Glowacki & Wrangham,2015). Its proponents argue that women would prefer successful warriors as mates, pro-viding an incentive to engage in attacks. Consequently, male 3However, note that Hypothesis 1b does not rely on group
se-lection; rather, parochial cooperation would have been selected for because of direct and indirect benefits derived from by-product mu-tualisms and indirect reciprocity.
4Note that group-selection theories and social identity treat
in-group love and out-in-group hate as functionally equivalent: both serve to promote the in-group relative to the out-group. In contrast, bounded generalised reciprocity theory provides a motivation for in-group love, but not out-group hate.
members of warrior parties should be attentive to their rep-utation, though not necessarily as cooperators (as bounded generalised reciprocity argues). There is some evidence cor-roborating this, indicating that men are more competitive in intergroup conflict (Johnson et al., 2006; van Vugt et al., 2007), and that women judge war heroes as more attractive (Rusch et al.,2015).
Out-group Empathy and Moral Hypocrisy
Empathy with the members of an out-group may be a way to temper out-group hate. Empathy is the ability to take the other’s perspective—to understand and feel what they expe-rience (Davis,1983). Empathy with an individual may moti-vate cooperative behaviour towards the individual (Batson & Moran,1999;Batson,2008). Though this effect appears less strong when the other is a member of an out-group, it per-sists in intergroup interactions (Batson, Kobrynowicz, Din-nerstein, Kampf, & Wilson,1997; Stürmer, Snyder, Kropp, & Siem,2006;Vorauer & Sasaki,2009). Indeed, empathy to-wards members of the out-group tempers negative attitudes (Stephan & Finlay, 1999; Cikara, Bruneau, & Saxe,2011) and aggression in intergroup conflict (Struch & Schwartz, 1989), and may take precedence over parochial cooperation (Batson et al., 1999). Thus, we expected that more empa-thetic individuals would show less parochial cooperation in predatory attacks on an out-group (but not when the target was non-social, or in defense).
Hypothesis 5: Higher empathy predicts lower parochial cooperation in predatory attacks on an out-group.
While empathy can be genuinely felt, it may not always be. Batson et al.(1997) coined the term moral hypocrisy for behaviour that appears moral, but in fact benefits the ac-tor. Not participating in a predatory attack under the guise of empathetic concern for the out-group may be such a be-haviour, if the “conscientious objector” still benefits from spoils of war. This is the case wherever predatory attacks produce public goods, such as deterrence, territorial gains, or loot that is shared with the entire community (Kohler & Kramer Turner, 2006; Mathew & Boyd,2011; Wrangham & Glowacki,2012; Glowacki & Wrangham, 2013; Rusch, 2014). Then, a non-participating member of a predatory group is free-riding on the efforts of his or her peers. Empa-thy with the out-group provides a plausible moral guise for free-riding (cf.Valdesolo & DeSteno,2007), though only for members of predatory groups. Thus, we expected members of predatory groups to express greater moral hypocrisy when facing an out-group, compared to a non-social target; and that this would increase free-riding.
Hypothesis 6: Members of predator group express more moral hypocrisy in intergroup con-flict than when exploiting a natural resource,
which is associated with more frequent free-riding.
It is a corollary of hypotheses 5 and 6 that members of predatory groups are sensitive to the nature of their tar-get. When exploiting a natural resource, neither genuine nor feigned empathy should affect their willingness to cooperate with their in-group. In intergroup conflict, however, mem-bers of predatory groups face a nested social dilemma (Wit & Kerr,2002;Polzer, Stewart, & Simmons,1999): they can benefit their in-group, but only at the expense of the out-group. They can also refrain from harming the out-group, but will inevitably free-ride on the efforts of their fellow predators, unless all other predators also abstain from inter-group conflict. Thus, social rather than non-social predation adds other-regarding concerns, which we expect to reduce parochial cooperation among members of predatory groups.
Hypothesis 7: Members of predatory groups exhibit less parochial cooperation in predation of social rather than non-social targets.
Asymmetric Intergroup Conflict and the Predator-Prey Game
Explicitly or implicitly, intergroup conflict has often been thought to be symmetric, with both sides vying for the same good (Abbink, 2012; de Dreu, Giffin, et al.,2015). How-ever, intergroup conflict is often asymmetric: a bidder pur-sues a target in a hostile take-over (Andrade, Mitchell, & Stafford,2001; Martynova & Renneboog,2008), a country aims to conquer another in war (Jones, Bremer, & Singer, 1996; Miller, 2009; Wright, 2014; de Dreu, Giffin, et al., 2015), minorities clash with the establishment over civil rights (Lowery, Unzueta, Knowles, & Goff,2006). For ex-ample, we have recently shown that two thirds of militarised interstate conflicts in recent history were fought between a revisionist side seeking appropriation of resources or territo-rial expansion and a non-revisionist side trying to preserve the status quo (de Dreu, Giffin, et al., 2015; Jones et al., 1996). In each of these predator-prey-like conflicts, one side is seeking appropriation and expansion, while the other side is fighting for survival and preservation of the status quo (Bornstein, Kugler, & Zamir,2005; de Dreu, Giffin, et al., 2015). In other words, one side is motivated motivated by greed, and the other by fear. As discussed before, this asym-metry in aims can have profound implications for the will-ingness of individuals to cooperate with their peers.
A wide variety of empirical evidence shows that in such asymmetric intergroup conflict, group defense motivates greater cooperation than predation. Revisionist victory is rare in militarised interstate conflicts (de Dreu, Giffin, et al.,2015; Jones et al.,1996) and hostile take-overs in busi-ness (Andrade et al.,2001;Martynova & Renneboog,2008), indicating greater proclivity towards parochial cooperation
in defensive than in predatory groups. This pattern is not unique to humans: chimpanzees raid neighbouring commu-nities for control over territory and access to females, but are unsuccessful more often than not (Wilson et al.,2014), as are collective hunters such as wolves and killer whales (J ˛edrzejewski et al., 2002; Hoelzel et al., 2007; Packer & Ruttan, 1988). More evidence comes from an analysis of U.S. Medals of Honor: American soldiers decorated for their altruistic actions were most often fighting in the defense rather than an offense. Thus, parochial cooperation appears to be emerge more readily and to be stronger among mem-bers of groups fighting a threat than those pursuing an oppor-tunity.
To study asymmetric intergroup conflict in the labora-tory, we have recently developed a game theoretic paradigm, which we call the Intergroup Predator-Prey Game (de Dreu, Giffin, et al., 2015). It is an n-player extension of the Predator-Prey game pioneered by de Dreu et al. (2010). Members of one side (‘predators’) can make investments to attempt and appropriate the resources of the other side (i.e., are motivated by greed to pursue an opportunity). Mem-bers of the other group (‘prey’) may only make investments to safeguard their resources against this predation (i.e., are motivated by fear to protect against losses). We found that the patterns in this game mirror observational evidence: predatory victory is rare and survival of the defenders the norm; and this frequent survival of groups fighting to pre-serve the status quo is due to greater cooperation, and con-comitantly vastly less frequent free-riding, among the mem-bers of these groups relative to memmem-bers of groups aim-ing for appropriation of resources (de Dreu, Giffin, et al., 2015). Parochial cooperation in defense is voluntary, needs no norm-enforcement, and does not require coordination (de Dreu, Giffin, et al.,2015); whereas predatory groups become more cooperative when peer punishment can reduce free-riding, and more effective with coordination (de Dreu, Giffin, et al.,2015). Here, we are using a modified version of the In-tergroup Predator-Prey Game to study parochial cooperation in defense against and predation of either a social or a non-social target.
Methods Sample
Sampling. Participants were recruited via Amazon Me-chanical Turk (MTurk). Participants could earn US$0.50 as a participation fee (average time on task 15 minutes), plus up to US$2.00 depending on their performance during the experiment. Subjects had to be US residents, and have an 85% approval rate for prior work on the MTurk platform.
248 participants were recruited. Of these, 119 were male. The mean age was 36 (sd= 12.2). Participants were mostly non-Hispanic white (189). Three participants were excluded
for not answering the questionnaire truthfully (all or almost all answer having the same value). Seven subjects were ex-cluded because their location was outside the United States. 13 subjects who completed the study but did not provide their details to receive payment were not excluded. There was no dropout during the study.
Manipulation Check. Only half of the participants passed all five items of the manipulation check (116/237); in particular because they doubted that their own group mem-bers were real (82/237) and that their opponent’s investments were made by real participants (36/110) resp. based on nat-ural phenomena (54/127). Overall, participants in the two “nature” conditions were more likely to fail the manipulation check than participants in the intergroup conditions (χ2-test, p= .012). To avoid discarding more than half of the available data, we performed all analyses on the full data set.
Operationalisation
Intergroup Predator-Prey Game. At the beginning of each round t, each player i receives an endowment of E= 20 tokens. They can decide to make an investment 0 ≤ gi≤ 20 into the group pool C. The group pool C is then compared against the investment of the out-group, respectively nature, N. For prey groups, if their investment CPrey equals or ex-ceeds N, each member of each group retains their remaining endowment. However, if N exceeds CPrey, the prey retain nothing. In contrast, for predator groups, if N exceeds their investment CPred, each member of each group retains their remaining endowment. However, if CPred exceeds N, the predators each receive their individual remaining endowment plus one third of the remaining prey endowment, respectively the remaining natural resource,P3i=1gi− N. This results in the following pay-off functions O:
Ot,prey,i= E − gi if N ≤ Cprey 0 if N > Cprey (1) Ot,pred,i= E − gi if Cpred ≤ N E − gi+60−N3 if Cpred > N (2) This game was repeated ten times. Participants did not re-ceive feedback on their investments during the game or after. After conclusion of the sampling period, participants were matched with a different set of interaction partners for each round (complete stranger matching) to compute pay-offs.
Variables. Role and resource are varied experimentally. Cooperation was operationalised as the average per-round investment of each individual player. Free-riding was op-erationalised as the proportion of rounds in which a player invested 0 MU, per condition.
Questionnaire Measures. Empathy was measured us-ing the 7-item empathetic concern subscale of the interindi-vidual reactivity index (IRI, Davis, 1983). The scale
in-cludes items such as “I often have tender, concerned feelings for people less fortunate than me.” Items were rated on a five-point Likert scale (1= does not describe me well, 5 = describes me well). The scale was reliable (Ωt= .89),5thus
a composite score for empathetic concern was computed. Moral hypocrisy was measured using four scenarios, which described either the participant or another member of the participant’s group as having invested either 0 or 5 monetary units during one round of the predator-prey game. Subjects were asked to rate the fairness and morality of the behaviour on a five-point Likert scale ("I [the other player] acted fairly [morally], 1 = completely disagree, 5 = com-pletely agree). These questions followedBatson et al.(1997) andValdesolo and DeSteno(2007). We also included two further questions to mask the target questions. Only answers from the two zero-investment scenarios were used, as the effect size for moral hypocrisy on the other scenarios was very small. The fairness and morality items of the self- and other-judgments, respectively, were averaged. The moral hypocrisy measure was obtained by subtracting the other-from the self-judgment score (such that a larger score reflects greater pro-self bias).
Reputational concern was measured using a six-item scale adapted from Beersma and Van Kleef (2011). Items were judged on a five-point Likert scale (1= totally disagree, 5 = totally agree). The scale includes items such as “During the decision making task, I thought about how others would think about me.”. The scale was reliable (Ωt = .82) with the exception of one item; thus this item was omitted and a composite score for reputational concern was computed.
For further exploration, we tested situational interdepen-dence using the 30-item situational interdepeninterdepen-dence scale (adapted from unpublished work by Daniel Balliet). The scale has five dimensions: temporal, interdependence, con-flict, information certainty, and power. Items were rated on a five-point Likert scale (1= completely agree, 5 = completely disagree; except for the power dimension, where 1= defi-nitely the other, 5= definitely myself). Exploratory factor analysis indicated a four- rather than five-factor solution and low general factor saturation (Ωh = .54). However, four out of the five scale had good reliability on their own (temporal: Ωt = .83, interdependence: Ωt = .83, conflict: Ωt = .72, information certainty:Ωt= .64, and power: Ωt= .90). Procedures
The study was run entirely online using Qualtrics. Partic-ipants were recruited via Amazon Mechanical Turk. Upon signing up for the study, they were redirected to Qualtrics. They were then given a general introduction and asked to provide consent. Then, participants were assigned to one of the four conditions and given a detailed conceptual introduc-tion. Then, they were given an introduction to the exact pay-off structure of the game.
Comprehension checks consisted of four yes/no items re-garding reputational concerns and outcome interdependence, as well as two scenarios. Each scenario described one poten-tial round of the game, and participants were asked to pick their own pay-off from four plausible options. Participants had as many attempts as necessary to pass the comprehension check, and questions were not changed for new attempts.
Then, participants played ten rounds of the game. Partici-pants did not receive feedback during the game.
The manipulation check consisted of five yes/no items. These items measured whether participants believed that other participants (in- and out-group members) were real people, and that their choices were anonymous.
Then, participants were (in random order) presented with the IRI, SIS, reputational concern, and moral hypocrisy tionnaires. Finally, we asked a number of demographic ques-tions. At the end of the study, participants were debriefed. Payments were made after conclusion of the experiment via the Mechanical Turk platform.
Results Table 1
Parochial cooperation in defense against and predation of social and non-social targets.
Full Data Restricted Data
Mean SD Mean SD Predators 7.48 5.73 8.28 6.07 - Social 7.01 5.70 7.49 6.00 - Non-social 7.90 5.74 9.28 6.02 Prey 9.88 4.26 9.64 3.67 - Social 10.68 4.14 10.09 3.44 - Non-social 9.20 4.24 9.03 3.89
Members of defensive groups made larger investments, on average, than members of predatory groups (Table 1; type II ANOVA on aligned rank transformed data, F(1,233) = 13.005, p < .001, η2p = .053, Bayes factor against inter-cept model= 1036.310). In contrast, there was no main ef-fect of the social/non-social target manipulation (F(1,233) = .095, p > .05, η2p < .001, BF = .167). This supports hy-pothesis 1b over hyhy-pothesis 1a (BF= 6208.389). However, there was a significant interaction between role and target (F(1,233)= 4.525, p = .034, η2
p= .020, BF against main ef-fect model= 1.624). Members of defensive groups invested vastly more than members of predatory groups in intergroup conflict (Wilcoxon Mann-Whitney rank sum test, U = 899.5, p < .001, r = .349, JZS BF(r = .707) = 2331.080), but not when facing a non-social target (U = 1847.5, p > .1, 5McDonald’sΩ is superior to Cronbach’s α, cf. Zinbarg,
r = .070, JZS BF(r = .707) = 0.732).6 This was be-cause parochial cooperation under threat increased when the threat was social rather than non-social (U = 1523, p > .1, r = .174, JZS BF(r = .707) = 4.175); whereas members of predatory groups were less cooperative in this scenario (U = 1748.5, p > .1, r = .086, JZS BF(r = .707) = 0.295). However, neither effect was statistically significant. Thus, in contrast to hypothesis 7, neither defenders nor predators seemed sensitive to the social nature of the target; though stronger statistical tests may elucidate whether a decrease among predators or increase among defenders has outsize in-fluence on the investment gap in intergroup conflict vis-à-vis the non-social context.
Figure 1. Parochial cooperation in defensive and predatory groups, across social and non-social contexts.
Defense Predation 1 2 3 4 5
Non−social Social Non−social Social
Perceived Conflict ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment ** n.s. *** Defense Predation 0 5 10 15 20
Non−social Social Non−social Social
Investment
Members of predatory groups exhibited vastly more fre-quent free-riding than members of defensive groups (test for equality of proportions, X2(1) = 229.49, p < .001, ˆppred = .253, ˆpprey = .036, r = .52, BF = 8.994 × 1052). Consequently, between-subjects variation was much larger in predatory than in defensive groups (Levene’s test, spred = 5.046, sprey = 3.202, W(1,235) = 24.345, p < .001, ra-tio of variances= 2.482). Predators and prey did not dif-fer in within-subjects variation in investments (Welch’s t-test, t(226.36)= -1.058, p > .05, η2p = .005). Thus, hypotheses 2 and 3 were supported.
Participants who expressed greater reputational concern made greater investments on behalf of their group (moder-ated linear regression, t(1, 233)= 2.181, p = .030, η2
p= .019, BF = .901), but the effect of reputational concerns was not conditioned on membership in a defending or attacking group (t(1, 233)= -.723, p > .05, η2
p= .002, BF against main effect model = .262). Thus, hypothesis 4 is only partially supported. Free-riding in predation on a human out-group was not predicted by empathetic concern (logistic regression, p > .05, Tjur’s D7= .013; no support for hypothesis 5).
Participants expressed significant moral hypocrisy relative to their in-group members (type II ANOVA, F(1,233)= 18.972, p< .001, η2g = .016, BF(r = .707) = 3.397), but its extent did not differ between members of predator and defensive groups (F(1,233)= 1.994, p > .05, η2
g = .002, BF again main effect model= 7.336 × 10−7). Differences in the extent of freerid-ing between the social and non-social conditions were not mediated by moral hypocrisy (causal mediation analysis for total effect, p > .05, 95% CI [-.085, .072]), and consequently there was no support for mediation moderated by role (Test of differences in ACME, p > .05, 95% CI [-.041, .022]; test of differences in ADE, p > .05, 95% CI [-.092, .216]; no support for hypothesis 6).
Table 2
Overview of hypotheses and effect size estimates. Where pos-sible, effect sizes are converted to η2for better comparison.
Hypothesis Support Effect BF
Hypothesis 1a Against η2
p< .001 .167
Hypothesis 1b Yes η2
p= .053 1036.310 Hypothesis 2 Yes η2= .270 8.994 ×1052
Hypothesis 3 Yes Ratio= 2.482 NA
Hypothesis 4a Weak η2p= .019 .901 Hypothesis 4b Against η2 p= .002 .262 Hypothesis 5 No D= .013 NA Hypothesis 6 No NA NA Hypothesis 7 No η2= .007 NA
Table 2 summarises the hypothesis tests and effect sizes for hypotheses 1-7. We find support for four hypotheses: that threat increases parochial cooperation (hypothesis 1b), that predatory groups show a higher frequency of free-riding (hypothesis 2) and greater between-subject variance in in-vestments (hypothesis 3), and that reputational concern mo-tivates parochial cooperation (hypothesis 4). However, role explained only about 5.3% of variance in investments, and reputational concerns a mere 1.9%. In the intergroup con-flict condition, role explained 12.2% of variance in parochial cooperation; vastly less than the 65.5% reported byde Dreu, Giffin, et al.(2015).
Discussion
In this study, we found support for our hypothesis that it is threat to the in-group, rather than intergroup conflict per se, that promotes parochial cooperation. Members of groups un-der threat were consistently more cooperative towards their peers than members whose group was in pursuit of an oppor-tunity. However, we also found indications that intergroup 6Applying Bonferroni correction for multiple comparisons, α=
.1.
7FollowingTjur(2009), the coefficient of discrimination D for
conflict plays a specific role, beyond being the situation in which generic threats and opportunities occur. The di ffer-ence in parochial cooperation between defensive and preda-tory groups was more pronounced during intergroup conflict, where these groups potentially directly face each other, than when threats and opportunities had a non-social character. This was due to both higher investments by members of de-fensive groups and lower investments by members of preda-tory groups; though neither effect reached significance and further investigation is in order.
Individual Differences: Reputation and Morality
We hypothesised three mechanisms which might drive differences in parochial cooperation between members of predatory and defensive groups, and between social and non-social contexts: reputational concerns, empathetic con-cern, and moral hypocrisy. Participants who expressed greater concern about their reputation indeed behaved more parochially cooperatively. However, this relationship did not differ between conditions, in contrast to our prediction that group defense would be more motivated by the prospects of a good reputation than collective predation. It may be that the design of the experiment reduced reputation effects too much to be informative about these questions; indeed, the modal reputational concern score was 1 (indicating no con-cern across five questions). A future study with a design ex-plicitly allowing for reputation effects may help to elucidate reputation effects and in particular differences between con-ditions.
However, it is telling of the uncertainty about the role of reputation in the literature that the hypothesis we put for-ward here is only one among several that find support. The male warrior hypothesis, in particular, has argued that men who are successful in predation on out-groups are perceived as more attractive and enjoy better reproductive prospects than their peers (Johnson et al.,2006;van Vugt et al.,2007; Sell et al.,2009;McDonald et al.,2012;Rusch et al.,2015; Glowacki & Wrangham,2015). Glowacki and Wrangham (2015) recently presented findings that extend this, showing that successful members of raiding parties enjoyed not just better reproductive fitness, but also higher social status. This seems at odds with findings that harming an out-group may even reduce prestige and power (Halevy et al.,2012). It could be argued that raids produce benefits to the in-group, and suc-cessful raiders night share their loot with their group. Then, aggression towards an out-group may be perceived as serving the in-group, and rewarded with the veneration otherwise re-served for heroic defenders. However, for our sample we find some indication that this is not so: free-riding is seen as more morally permissible (for both self and others) among mem-bers of predatory groups than among defenders (F(1,233)= 17.843, p < .001, η2
g = .057). Indeed, predation, in the lab as in the field, seems largely opportunistic, and produces few
if any public goods (Mathew & Boyd,2011; Taylor,2014; Rusch,2014;de Dreu, Giffin, et al.,2015).
We had also predicated that moral hypocrisy would oc-cur more in members of predatory groups than of defensive groups, and in particular when facing an out-group. In con-trast, moral hypocrisy was stable across conditions of both role and target. We did, however, observe that free-riding was perceived as more morally permissible among predators than among prey, and more in the social than in the non-social setting (type II ANOVA, F(1,233)= 5.617, p = .019, η2
g = .019))). One explanation may be that these situation are governed by different social norms. However, if that is so, these norms seem equivocal, for in previous research we have found that free-riding members of predatory groups in intergroup conflict are punished by their peers with the ex-plicit motivation to coerce them to cooperate (de Dreu, Gif-fin, et al.,2015, cf. ;Mathew & Boyd,2011). It could be that free-riding is punished even if it is morally permissible, in line with findings that defection, rather than norm violation, is punished (Bone, Silva, & Raihani,2014). Furthermore, interestingly the actual rate of free-riding was predicted by participant’s perception of others’ free-riding as morally per-missible, but not their own (multivariate regression, F(1,231) = 2.151, p = .033, η2
p = .020 and F(1,231) = -.191, p > .05, η2
p < .001, respectively). This, again, points towards so-cial norms rather than self-serving morality as allowing free-riding.
Thus, we observe differences in parochial cooperation in line with the hypothesis that threats to the in-group, rather than intergroup conflict more generally, motivate parochial cooperation. However, individual differences in reputational concern and moral hypocrisy evoked by different conditions do not seem to explain differences in the level of parochial cooperation.
Situational Differences
A second approach is to consider situational differences. Balliet (unpublished work) distinguishes five dimensions on which situations differ: temporal, interdependence, conflict, information certainty, and power. (Balliet,2015) has high-lighted differences in perceived interdependence depending on the situational context, which may give rise to differences in parochial cooperation. Interdependence has a strong, pos-itive effect on in-group favouritism, although it is not neces-sary for parochial cooperation to occur (Balliet et al.,2014). If members of predatory and defensive groups perceive their interdependence with their peers differently, this may give rise to differences in cooperation. Another situational can-didate to explain differences in the predator-prey context is perceived conflict within the in-group. Perceived conflicts of interest between the members of a group may undermine their willingness to engage in parochial cooperation. If the potential for deprivation by in-group members is perceived
as more acute in some contexts than in others, this may give raise to situational differences in parochial cooperation (de Dreu,2010a).
Figure 2. Perceived conflict in defensive and predatory groups, across social and non-social contexts.
Defense Predation 1 2 3 4 5
Non−social Social Non−social Social
Perceived Conflict
Indeed, the social situation was perceived as engendering more intragroup conflict than the non-social situation in both predatory and defensive groups (Figure 2; MANOVA, Role: Wilks’Λ = .938, p = .035; Target: Λ = .938, p = .012; Role × Target: Λ = .981, p > .05. Type II ANOVA for Conflict, Target: F(1,233)= 12.999, p < .001, η2
p = .052; all other post-hoc comparisons p > .01)8. Perceptions of
internal conflict did not differ strongly between defensive and predatory groups (Type II ANOVA for conflict, Role: F(1,233)= .440, p > .05, η2p = .002). As suggested, further exploratory analysis indicated that when controlling for role and target, greater perceived intragroup conflict was associ-ated with reduced parochial cooperation (multivariate linear regression, t(1,232) = -2.812, p = .005, η2
p = .033). Fig-ure 3 highlights that this effect may be stronger in predators than in defenders, in particular in the social situation. This would provide a psychological mechanism for the observed differences in parochial cooperation between predatory and defensive groups; however, including an interaction between perceived intragroup conflict and role and target did not yield any statistically significant results (all p > .05).
In this study, we have focused on the manipulation of threats and opportunities faced by members of groups in both social and non-social settings. It is likely that not only per-ceived intragroup conflict, but also perper-ceived threat to the in-group is affected by differences between social and non-social contexts. This is suggested by our experimental re-sults showing greater parochial cooperation among defend-ers in a social than in a non-social settings. Future research should build upon this finding by explicitly investigating per-ceptions of threat and opportunity in social and non-social
Figure 3. Association between perceived conflict and parochial cooperation in defensive and predatory groups, across social and non-social contexts. Lines indicate linear regression slopes. 0 5 10 15 20 1 2 3 4 5 Perceived Conflict Investment Prey/Non−social Prey/Social Predators/Non−social Predators/Social
settings. In particular, it seems prudent to study the di ffer-ence between social and non-social contexts using iterated games with feedback. If initial differences in expectations about the threat stemming from social and non-social forces have a deeper root, they should persist over time; otherwise, it is likely that investment levels will converge.
Conclusion
In this study, we have shown that people are more will-ing to cooperate with their in-group under threat than in pur-suit of an opportunity, and that intergroup conflict widens this gap. In contrast to theories that propose that intergroup conflict generally increases cooperation, we show that it is threat, rather than conflict, that motivates people to work to-gether for the benefit for their group. We have argued that threats to the in-group rouse fast, automatic, and affective fear, whereas collective opportunities elicit slow and and de-liberate greed (de Dreu et al.,2010,2014;de Dreu, Giffin, et al.,2015;de Dreu, Dussel, & ten Velden,2015). As a conse-quence, parochial cooperation in defense is quick and strong, but in predation, it is easily undermined by selfish incentives to free-ride. This may be an adaptation to an ancestral en-vironment in which group membership was crucial (West et al.,2007a; Choi & Bowles,2007; Bowles, 2009; de Dreu et al.,n.d.) and the potential cost of a weakened group was greater than the potential gains from a strengthened group (McDermott et al.,2008).
In line with the theory of Bounded Generalised Reci-procity, we have found that reputational concerns are asso-8Applying Bonferroni correction for multiple comparisons, α=
ciated with greater parochial cooperation, mirroring the car-rot of praise and the stick of bad-mouthing. However, neither empathetic concern nor moral hypocrisy helped elucidate the mechanism by which the difference in cooperativeness be-tween predators and defenders comes about. This leaves room for further research into these underlying mechanisms. Some promising avenues are differences in the perception of the respective situations as being threatening or providing opportunities, expectations about the cooperative behaviour of in- and out-group members, and the moral norms govern-ing cooperation and free-ridgovern-ing.
Attacks and catastrophes can devastate, but they can also give rise to cooperation and altruism. Predation can be har-rowing; but conflict can also give engender productivity and creativity (Jehn & Mannix, 2001; de Dreu, 2006). At its best, this study, in providing some insights into the nature of parochial cooperation and intergroup conflict, will helps us understand how to bolster resilience and resistance in the face of threats, and how to channel predatory tendencies into productive and creative forces.
References
Abbink, K. (2012). Laboratory Experiments on Con-flict. The Oxford Handbook of the Economics of Peace and Conflict, 532–553. doi: 10.1093/oxfordhb/ 9780195392777.013.0022
Abbink, K., Brandts, J., Herrmann, B., Orzen, H. (2012). Parochial altruism in inter-group conflicts. Economics Letters, 117(1), 45–48. doi: 10.1016/j.econlet.2012.04 .083
Abbot, P., Abe, J., Alcock, J., Alizon, S., Alpedrinha, J. A. C., Andersson, M., . . . Zink, A. (2011). Inclusive fitness theory and eusociality. Nature, 1062(2010), 1057–1062. doi: 10.1038/nature09831
Alexander, R. D. (1974). The evolution of social behavior. Annual Review of Ecology and Systematics, 5, 325– 383.
Allport, G. W. (1954). The Nature of Prejudice. Cambridge, MA: Addison-Wesley.
Andrade, G., Mitchell, M., Stafford, E. (2001). New ev-idence and perspectives on mergers. Journal of Eco-nomic Perspectives, 15(2), 103–120.
André, J.-B., Morin, O. (2011). Questioning the cultural evolution of altruism. Journal of Evolutionary Biol-ogy, 24(12), 2531–2542. doi: 10.1111/j.1420-9101 .2011.02398.x
Ariely, D., Huber, J., Wertenbroch, K. (2005). When Do Losses Loom Larger than Gains? Journal of Market-ing Research, 42(2), 134–138.
Associated Press. (2012, November). What we know about Superstorm Sandy a month later. Retrieved from http://bigstory.ap.org/article/what-we -know-about-superstorm-sandy-month-later
Axelrod, R., Hamilton, W. D. (1981). The Evolution of Cooperation. Science, 211(4489), 1390–1396. Azarow, J., Manley, M. J., Koopman, C., Platt-Ross, A.,
But-ler, L. D., Spiegel, D. (2003). American expres-sions of altruism and generativity in the aftermath of the September 11, 2001 terrorist attacks. Psicología Política, 27, 37–58.
Balliet, D. (2015). Functional Interdependence Theory: Ad-vancing Evolutionary Social Psychology.
Balliet, D., Wu, J., de Dreu, C. K. W. (2014). In-group favoritism in cooperative decision making: A meta-analysis. Psychological Bulletin. doi: http://dx.doi .org/10.1037/a0037737
Batson, C. D. (2008). Empathy-Induced Altruistic Mo-tivation. In Herzliya symposium on prosocial mo-tives, emotions, and behavior(pp. 1–32). doi: http:// dx.doi.org/10.1037/12061-001
Batson, C. D., Ahmad, N., Yin, J., Bedell, S. J., Johnson, J. W., Templin, C. M., Whiteside, A. (1999). Two Threats to the Common Good: Self-Interested Egoism and Empathy-Induced Altruism. Personality& Social Psychology Bulletin, 25(1), 3–16.
Batson, C. D., Kobrynowicz, D., Dinnerstein, J. L., Kampf, H. C., Wilson, A. D. (1997). In a Very Different Voice: Unmasking Moral Hypocrisy. Journal of Personal-ity and Social Psychology, 72(6), 1335–1348. doi: 10.1037/0022-3514.72.6.1335
Batson, C. D., Moran, T. (1999). Empathy-induced al-truism in a prisoner’s dilemma. European Journal of Social Psychology, 29(7), 909–924. doi: 10.1002/ (SICI)1099-0992
Baumeister, R. F., Leary, M. R. (1995). The Need to Belong: Desire for Interpersonal Attachments as a Fundamen-tal Human Motivation. Psychological Bulletin, 117(3), 497–529. doi: 10.1037/0033-2909.117.3.497
Beersma, B., Van Kleef, G. A. (2011). How the Grapevine Keeps You in Line: Gossip Increases Con-tributions to the Group. Social Psychological and Personality Science, 2(6), 642–649. doi: 10.1177/ 1948550611405073
Bellezza, S., Gino, F., Keinan, A. (2014). The Red Sneakers Effect: Inferring Status and Competence from Signals of Nonconformity. Journal of Consumer Research, 41(1), 35–54. doi: 10.1086/674870
Benard, S. (2012). Cohesion from Conflict: Does In-tergroup Conflict Motivate Intragroup Norm Enforce-ment and Support for Centralized Leadership? Social Psychology Quarterly, 75(2), 107–130. doi: 10.1177/ 0190272512442397
Bergmüller, R., Johnstone, R. A., Russell, A. F., Bshary, R. (2007). Integrating cooperative breeding into theoret-ical concepts of cooperation. Behavioural Processes, 76(2007), 61–72. doi: 10.1016/j.beproc.2007.07.001
Bernhard, H., Fehr, E., Fischbacher, U. (2006). Group A ffili-ation and Altruistic Norm Enforcement. The American Economic Review, 96(2), 217–221.
Bernhard, H., Fischbacher, U., Fehr, E. (2006). Parochial altruism in humans. Nature, 442(7105), 912–5. doi: 10.1038/nature04981
Birch, J., Okasha, S. (2014). Kin Selection and Its Crit-ics. BioScience, 65(1), 22–32. doi: 10.1093/biosci/ biu196
Bone, J., Silva, A. S., Raihani, N. J. (2014). Defectors, not norm violators, are punished by third-parties. Biology Letters, 10, 20140388.
Bornstein, G. (2003). Intergroup Conflict: Individual, Group, and Collective Interests. Personality and So-cial Psychology Review, 7(2), 129–145.
Bornstein, G., Ben-Yossef, M. (1994). Cooperation in in-tergroup and single-group social dilemmas.(Vol. 30). doi: 10.1006/jesp.1994.1003
Bornstein, G., Gilula, Z. (2003). Between-Group Com-munication and Conflict Resolution in Assurance and Chicken Games. Journal of Conflict Resolution, 47(3), 326–339. doi: 10.1177/0022002703252367
Bornstein, G., Gneezy, U., Nagel, R. (2002). The effect of intergroup competition on group coordination: An experimental study. Games and Economic Behavior, 41(1), 1–25. doi: 10.1016/S0899-8256(02)00012-X Bornstein, G., Kugler, T., Zamir, S. (2005). One team must
win, the other need only not lose: an experimental study of an asymmetric participation game. Journal of Behavioral Decision Making, 18(2), 111–123. doi: 10.1002/bdm.489
Bowles, S. (2008). Conflict: Altruism’s midwife. Nature, 456, 326–7.
Bowles, S. (2009). Did warfare among ancestral hunter-gatherers affect the evolution of human social behav-iors? Science, 324(5932), 1293–8. doi: 10.1126/ science.1168112
Boyd, R., Richerson, P. J. (1982). Cultural Transmission and the Evolution of Cooperative Behavior. Human Ecology, 10(3), 325–351.
Brandt, H., Hauert, C., Sigmund, K. (2003). Punishment and reputation in spatial public goods games. Proceedings of the Royal Society. Series B, Biological Sciences, 270(1519), 1099–1104. doi: 10.1098/rspb.2003.2336 Brewer, M. B. (1999). The Psychology of Prejudice: Ingroup
Love or Outgroup Hate? Journal of Social Issues, 55(3), 429–444.
Camerer, C. F. (2005). Three Cheers–Psychological, Theo-retical, Empirical–For Loss Aversion. Journal of Mar-keting Research, 42(2), 129–133.
Cashdan, E. (2001). Ethnocentrism and Xenophobia: A Cross-Cultural Study. Current Anthropology, 42(5), 760–765.
Cheng, J. T., Tracy, J. L., Foulsham, T., Kingstone, A., Hen-rich, J. (2013). Two Ways to the Top: Evidence That Dominance and Prestige Are Distinct Yet Viable Av-enues to Social Rank and Influence. Journal of Per-sonality and Social Psychology, 104(1), 103–125. doi: 10.1037/a0030398
Choi, J.-K., Bowles, S. (2007). The Coevolution of Parochial Altruism and War. Science, 318(5850), 636– 640.
Chudek, M., Henrich, J. (2011). Culture-gene coevolution, norm-psychology and the emergence of human proso-ciality. Trends in Cognitive Sciences, 15(5), 218–226. doi: 10.1016/j.tics.2011.03.003
Cikara, M., Bruneau, E. G., Saxe, R. R. (2011). Us and Them: Intergroup Failures of Empathy. Current Direc-tions in Psychological Science, 20(3), 149–153. doi: 10.1177/0963721411408713
Clutton-Brock, T. (2002). Breeding together: kin selec-tion and mutualism in cooperative vertebrates. Science (New York, N.Y.), 296(April), 69–72. doi: 10.1126/ science.296.5565.69
Columbus, S. (2013). Leadership in Altruistic Punishment and Reward. In Proceedings of the student research conference 2013(pp. 157–160). Amsterdam, NL. Conradt, S. (2012, November). 11 Acts of Kindness in
the Aftermath of Hurricane Sandy. Retrieved from http://mentalfloss.com/article/13069/11 -acts-kindness-aftermath-hurricane-sandy Davis, M. H. (1983). Measuring individual differences in
empathy: Evidence for a multidimensional approach. Journal of Personality and Social Psychology, 44(1), 113–126. doi: 10.1037/0022-3514.44.1.113
Dawes, R. M. (1980). Social Dilemmas. Annual Review of Psychology, 31, 169–93.
de Dreu, C. K. W. (2006). When Too Little or Too Much Hurts: Evidence for a Curvilinear Relationship Be-tween Task Conflict and Innovation in Teams. Jour-nal of Management, 32(1), 83–107. doi: 10.1177/ 0149206305277795
de Dreu, C. K. W. (2010a). Social Conflict: The Emer-gence and Consequences of Struggle and Negotiation. In S. T. Fiske, D. T. Gilbert, G. Lindzey (Eds.), Hand-book of social psychology (5th ed.). New York, NY: Wiley.
de Dreu, C. K. W. (2010b). Social value orientation mod-erates ingroup love but not outgroup hate in com-petitive intergroup conflict. Group Processes & In-tergroup Relations, 13(6), 701–713. doi: 10.1177/ 1368430210377332
de Dreu, C. K. W., Balliet, D., Halevy, N. (n.d.). Parochial Cooperation in Humans: Forms and Functions of Self-Sacrifice in Intergroup Conflict. In A. J. Elliot (Ed.), Advances in motivational science (pp. 1–63). New
