The work of Norenzayan et al. on the evolution and psychology of “prosocial religions” impressively integrates theory and data from multiple disciplines, including economics, anthropology, history, evolutionary biology, and social psychology. Although this is in many respects a rich and fertile approach, the fact that notions of “religious prosociality” differ across disciplines can give rise to serious conceptual confusions.

In an influential review, Norenzayan and Shariff (Reference Norenzayan and Shariff2008, p. 58) defined “religious prosociality” as “the hypothesis that religions facilitate costly behaviors that benefit other people.” Although they noted that such behaviours can produce victims as well as beneficiaries, their focus on nice, “neighbourly” aspects such as generosity and trust was consistent with a standard social psychological conception of “prosociality” (Batson & Powell, Reference Batson, Powell, Millon, Lerner and Weiner2003). Subsequently, other authors (e.g., Galen, Reference Galen2012; Preston et al. Reference Preston, Ritter and Hernandez2010) have reinforced this usage, contrasting religion's “prosocial” effects with its “antisocial” or “nonprosocial” effects, the latter including aggressive and prejudicial behaviours. Thus, evidence that participants passing a Christian landmark express more negative attitudes toward Christian out-groups than those passing a civic landmark (LaBouff et al. Reference LaBouff, Rowatt, Johnson and Finkle2012) has been taken as evidence against the religious prosociality hypothesis (Galen Reference Galen2012).

This might have less serious consequences theoretically were it not for Norenzayan et al.’s cultural evolutionary argument. According to Norenzayan et al., the advent of cultural notions that intertwine the “supernatural” with the “prosocial” has been a key factor in the human transition from small-scale, kin-based groups to complex large-scale societies. On this view, the “religious prosociality” hypothesis is not the hypothesis that religion promotes indiscriminate sharing and caring, but rather the hypothesis that religion fosters social cohesion within religious groups – favouring their “stability, survival, and expansion, at the expense of less successful rivals” (Norenzayan Reference Norenzayan2013, p. 30). As the current target article makes clear, Norenzayan et al. view “prosocial religions” as religious groups that encourage cooperation among their adherents, and – when intergroup threat is perceived – hostility and aggression towards out-groups.

From this perspective, it is no paradox that the holy books of the two most dominant “prosocial” religions, Christianity and Islam – whose adherents include the majority of the world's people (Central Intelligence Agency 2015) – contain numerous exhortations to violence against out-group members. As unpalatable as it may seem, even the barbaric treatment of out-group members by groups such as ISIS/DAESH is not necessarily “antisocial” on this conception. Indeed, aggression, murder, and even genocide can be viewed as prosocial acts insofar as they facilitate success in intergroup competition and conflict (McKay & Whitehouse Reference McKay and Whitehouse2015). By contrast, a paradigmatically antisocial act might be a cyberattack on social institutions carried out for mere personal satisfaction, rather than in the service of some group cause.

So, does the evidence indicate that religiously motivated altruism is always parochial – that is, preferentially directed toward in-group members? Interestingly, several lines of recent evidence suggest otherwise. Reddish et al. (Reference Reddish, Bulbulia and Fischer2013) found that social synchrony, a key feature of many religious rituals, evoked cooperation with both in-group members (with whom the synchronous action was performed) and members of a non-performance group. Meanwhile, Everett et al. (Reference Everett, Haque and Rand2015) found that religious participants gave significantly more money to other players in an economic game than did atheist participants – irrespective of whether the recipients were coreligionists or atheists. In fact, only the atheists in this study discriminated between religious and atheistic recipients, transferring marginally more money to atheist recipients.

Do such findings count against Norenzayan et al.’s cultural evolutionary story? Not necessarily. In the case of the synchrony study, cooperation with out-group members may represent a spillover effect. That is, it may be that synchronous behaviours promote generalized cooperative sentiment, ordinarily applied toward co-religionists in the immediate performance vicinity but here – in the artificial context of the experiment – extended also to out-group members. As for Everett et al. (Reference Everett, Haque and Rand2015), one possibility is that as prosocial religions grow and prosper, the decidedly parochial mores of their initial manifestations transmute into more benign, universal forms, forms that contemporary adherents adopt (see also Clobert et al. Reference Clobert, Saroglou and Hwang2015; Johnson et al. Reference Johnson, Memon, Alladin, Cohen and Okun2015b). According to Hartung (Reference Hartung1995), whereas the biblical context of the injunction to “love thy neighbour” clearly indicates that one's neighbour is a fellow in-group member, most contemporary Jews and Christians view the law as applying to everybody – that is, everybody is “thy neighbour.”

For Hartung (Reference Hartung1995), attempts to present religious in-group morality as universal morality are disingenuous, defying the clear intent of the texts upon which such moralities are based. As he documents, certain religious texts (e.g., Maimonides’ Codes) have been “strategically mistranslat[ed]” to obscure the parochial intent of the original variants (e.g., replacing the words “single Israelite” with “human being”). Although we understand the impulse to expose the parochial underbelly of prosocial religions, we should also be exploring ways of making the ideals of universalistic prosociality achievable (Whitehouse Reference Whitehouse, Clarke, Powell and Savulescu2013a; Reference Whitehouse2013b). The interesting research question is whether certain elements of the universal religious repertoire (e.g., notions of hell; kinship cues) are especially geared toward motivating parochial as opposed to universal conceptions of morality. If so, do these mechanisms exhibit plasticity such that, for example, religiously motivated “prosociality” is more parochial in the presence of out-group threat and more universalist in conditions of “existential security”? We hope that future research will elucidate the prospects for harnessing the various cognitive and cultural mechanisms that Norenzayan et al. discuss in the service of implementing a more universally applicable conceptualisation of “prosociality,” in which the “neighbourhood” of “love thy neighbour” expands without limit.