Norenzayan et al. provide a valuable review of ethnographic records, experimental research, and cross-cultural studies related to the cultural evolution of prosocial religions. However, clarification is needed about the causal role that different kinds of gods have played in the evolution of big societies. Previous work has focused on the role of powerful, all-knowing, moral, and punishing gods – Big Gods. Big Gods are central to the Abrahamic religions, which include Christianity and Islam – the two most successful religions in the world today. As Norenzayan et al. note, priming, economic games, and self-reported charitable giving studies are all based on the effects of the Abrahamic religions, rather than prosocial religions generally. The Abrahamic religions have also driven the results of previous cross-cultural studies. The cross-cultural studies cited (Johnson Reference Johnson2005; Peoples & Marlowe Reference Peoples and Marlowe2012; Roes Reference Roes1995; Roes & Raymond Reference Roes and Raymond2003) are based on the “high god” variable in the Ethnographic Atlas (EA). In a refined subset of these cultures, known as the Standard Cross Cultural Sample (SCCS), there are 40 moralising high gods, 32 of which (80%) belong to Christian or Islamic cultures (variable 2002) (Murdock & White Reference Murdock and White1969). The remaining eight are either part of another branch of the Abrahamic religions or are plausibly influenced by Abrahamic religions (Atkinson et al. Reference Atkinson, Latham and Watts2014).

There are three major reasons to interpret these studies with caution. First, the moralising high gods considered in these studies have spread with Christianity and Islam, and these religions have a range of unique features that might explain their success. These features include active proselytization, the exclusivity of a single god, belief in an end time, and eternal hell for heresy – none of which are necessarily prosocial. It is therefore not clear that Big Gods have spread because they are prosocial. Second, the Abrahamic religions have spread though processes of cultural diffusion such as trade networks, conquest, and missionisation. These processes led to the transmission of a range of features, including political structures, education systems, subsistence technologies, language, and religion. Cultural diffusion violates the independence assumptions of the statistical methods used in cross-cultural studies, and the diffusion of this suite of traits means that the associations found between Big Gods and big societies do not necessarily reflect a direct causal relationship (Atkinson et al. Reference Atkinson, Latham and Watts2014). Third, because the Abrahamic religions arose within the past 3,000 years, around 9,000 years after humans began the transition to large cooperative societies, they cannot have played a causal role in the emergence of the earliest large human societies.

Though previous research has tended to focus on the role of Big Gods, Norenzayan et al. acknowledge a continuum of supernatural agents and allow for a broad range of moralising supernatural agents. However, as it stands, only the Big Gods corner of this “multidimensional continuum” (sect. 2.5) is explored. If the theory put forward is a general theory about the evolution of prosocial religions, rather than a theory about the recent spread of the Abrahamic religions, then there needs to be more focus on the kinds of religions that were present in early human societies, including premodern societies at various scales of complexity and stratification.

In a recent study, we tested the scope of claims about the role of supernatural agents in the evolution of big societies using a sample of indigenous Austronesian cultures as they were before conversion to Christianity and Islam (Watts et al. Reference Watts, Greenhill, Atkinson, Currie, Bulbulia and Gray2015). We sought to tease apart two different kinds of supernatural punishment: first, broad supernatural punishment (BSP), which includes punishment by a wide range of supernatural agents, and second, the specific belief in punishment by a moralising high god (MHG). Our sample included 96 cultures, with social structures ranging from small, lineage-based communities such as Arosi, in which reciprocity is likely sufficient to sustain cooperation (Scott Reference Scott2007), to large, politically complex societies such as Hawaii, which faced the challenge of maintaining cooperation among multiple large communities (Kirch Reference Kirch2010). Language-based trees of Austronesian cultures are a good proxy for the cultural history of the cultures and enable the use of phylogenetic methods (Gray et al. Reference Gray, Drummond and Greenhill2009). Phylogenetic methods address Galton's problem by controlling for cultural ancestry (Mace & Holden Reference Mace and Holden2005) and are able to get at the direction of causality by inferring the order that traits tend to evolve (Pagel & Meade Reference Pagel and Meade2006). A broad range of supernatural agents were believed responsible for punishment in Austronesia, including the spirits of recently departed ancestors and powerful deities. However, beliefs in the kinds of Big Gods found in Abrahamic religions were scarce. Our results indicated that BSP facilitated the evolution of political complexity in Austronesia, and that MHGs arose after political complexity, though neither sustained political complexity once it had arisen. We do not take our study to be a test of the model put forward in the target article, which includes a complex array of mechanisms that drive the evolution of social complexity. Although further work is needed to establish the mechanism by which BSP facilitates social complexity, the results highlight the potential importance of small gods in the evolution of big societies.

In summary, just as Norenzayan et al. acknowledge a bias in their experimental work towards Abrahamic religions, Abrahamic religions also drive the correlations in most of the comparative studies cited as supporting a role of Big Gods. As a result, these studies cannot tease apart which elements of religion are causally important, nor whether religion is playing a role at all. Future work should (1) systematically catalogue religious variation both within and outside the Abrahamic traditions, (2) test whether and how specific features play a causal role in promoting prosociality and, (3) identify at what points in the evolution of big societies each kind of supernatural agent plays a functional role.