Reinvigorating the Comparative, Cooperative Ethnographic Sciences of Religion

Benjamin Grant Purzycki, Joseph Watts

From weekly Catholic masses to the sacrifice of goats by the Nuer of Sudan, people around the world devote considerable time and resources to gods and spirits. Over time, natural selection is supposed to increase the frequency of functional traits and decrease the frequency of maladaptive traits. Does this mean that nature “decided” that the expenses of religious belief and practice are worthwhile investments? Do returns from religious investments come from other people? Why are gods involved at all? Couldn’t such a successful species as ours come up with less bizarre ways of dealing with each other and our environments? If religion does yield important returns, features of religion should acclimate to new conditions. Do they?

If they serve functions, then religious systems should indicate a system that pays off more than it pays out. As with many other social science fields seeking evolutionary explanations, the most interesting debates in the field of evolutionary religious studies right now are about whether or not religion contributed to human survival and reproduction and how we can even go about determining an answer. We wager that systematic, well-organized ethnographic research can provide us with the means required to resolve many of these debates.

The past few decades have seen numerous attempts at explaining supernatural beliefs and practices naturalistically in a variety of fields. Despite their many successes, important fields of research remain lacking in crucial ways. For example, social psychologists and experimental economists with exquisite methodological and analytical training primarily derive generalizations from convenient samples of university students. Cultural anthropologists are still hungover from the sweet, self-indulgent stupor of postmodernism and get nauseous from the slightest whiff of functionalism. Evolutionary biological anthropologists largely ignore religion and—let’s face it—often trivialize the importance of soft stuff such as beliefs, symbolic culture, and ritual. The most influential progenitors of the cognitive science of religion movement spent the past twenty years largely dismissing religion as merely a cognitive glitch while other pundits are all too quick to characterize religion as maladaptive (that is, that religious systems pay out more than they pay off). While virtually all such contemporary endeavors speak to it, they simply do not empirically address the question of whether or not variation in threats to human social life helps explain variation in our religious beliefs and behaviors. So many people seem to have strong opinions despite the answers being far from clear.

To make genuine progress toward a naturalistic understanding of supernatural beliefs and practices, we need an approach that provides a scientific, cross-cultural perspective. We also need some consensus on how to go about assessing competing views. At least in the social sciences, it’s remarkable how quickly some researchers see evidence confirming their existing views. To make sense of religion more reliably, we need well-organized teams of researchers who are open to a variety of views.

In one sense, this sounds like an impossible task. Organizing social scientists is like herding cats and is rarely thought of as a scientific deliverable, even though it can be a full-time job. All too often the importance of good organization gets outshined by data sets, fancy publications, and figureheads. In another sense, however, there’s never been an easier time to pull off something like this. Wealthy foundations dole out millions toward research; global travel has never been easier and more far-reaching; social scientific methods and statistical analyses have unprecedented transparency and accessibility; and communication streams are now so sophisticated that colleagues can e-mail images of themselves on Fijian beaches to friends enduring Siberian winters. The seeds of progress are already planted—we just have to deal with ourselves.

These days, there are three main modes that ongoing cross-cultural empirical research investigating the payoffs religion might have. One is particularistic; some researchers focus on their specific question, conduct studies in their field site, and hope to add to the greater conversation. This has been the main approach for cultural anthropologists for quite some time: live with and study a small population and try to make sense of it. While ethnography is all too often associated with clumsy, qualitative methods and an over-reliance on researchers’ intuitions, some younger ethnographers such as Eleanor Power and John Shaver have executed exquisite, precise studies that inform what we know about religious behavior and its payoffs. These and other exemplar cases, however, are conducted in isolation and often emphasize local particulars. Another mode is similar but collaborative and team-based. This approach has the benefit of being an organized effort to test specific predictions, but this often comes at the expense of focusing on wider variation and whether or not religious investments pay off. The Cultural Evolution of Religion Research Consortium (CERC) has focused on “moralistic gods” (see below) and their role in the payouts required to build broader social bridges but did not examine the effects of a broader range of beliefs or the form and function of ritual.

The third mode takes advantage of centuries of recorded observations of other people and compiles them into databases. Much of what we think we know about religious variation comes from well-established cross-cultural databases. Notably, these have remarkably scant data about religious beliefs and practices, and the data they do have isn’t great. The two most established cross-cultural databases are the Ethnographic Atlas and the Standard Cross-Cultural Sample (SCCS). These databases are built by filtering through and aggregating the ethnographic descriptions of cultures (typically from books), and then numerically coding whether a set of pre-defined features are present. These features range from the “coldest month of the year” to “post-partum sex taboos.” It is perhaps surprising then that there are only a handful of variables on religion, and only one, called “High Gods,” has received much attention.

We focus on this High Gods variable in some detail here because it has driven so much of what we think we know about the evolution of religion. It also provides a good illustration of what the field needs. High-profile articles and books have relied heavily on this variable and have suggested a functional relationship between moralistic high gods and various measures of social complexity such as the domestication of animals, monetary systems, and political complexity. Despite how much hinges on this solitary variable, a casual glance suggests that we should be very cautious about using it to make grand claims.

In these databases and publications, a high god is defined as: “a spiritual being who is believed to have created all reality and/or to be its ultimate governor, even though his/her sole act was to create other spirits who, in turn, created or control the natural world.” It has a four-option categorical value: 1) absent or not reported; 2) present but not active in human affairs; 3) present and active in human affairs but not supportive of human morality; 4) present, active, and specifically supportive of human morality.

Figure 1. Levels of jurisdictional hierarchy across “high god” coding types for 186 societies in the Standard Cross-Cultural Sample (SCCS).

Let’s take a look at the data (Figure 1). There are 186 cultures in the SCCS. In sixty-eight of them (37 percent), high gods are coded as “absent or not reported.” That doesn’t strictly mean that these populations don’t have high gods. It could also mean that the high god wasn’t important enough to have been noted by ethnographers, ethnographers didn’t bother to note it, or the people being studied didn’t tell the ethnographer about it. In the event that the ethnographies mention moralistic gods (i.e., ancestor spirits) that aren’t creators, this may have also been coded as “absent or not reported.”That’s a bit like asking two hundred people a question and lumping together eighty who didn’t respond, said “no,” changed the subject, or said “yes, but with one major caveat.” Ideally, we really want to know the difference between those who said “no,” those who fell asleep during the interview, those who changed the subject, and those whose responses were more thoughtful than your question.

If we break down who believes what, more striking non-patterns emerge. Remember that the commonly held idea is that social complexity, often measured by levels of “jurisdictional hierarchy beyond local community,” is somehow associated with moralistic high gods; the more complex a society is, the more likely it’ll have a moralistic creator deity. But if you look at the distribution of the high god question across social complexity types, it shows that the highest value of the religion codes (23 percent of all coded societies) are simple societies where moralistic (creator) high gods are either “absent” or “not reported” (or possibly have non-creator deities that do care about moral stuff). The rest are all under 10 percent. The most common society type with moralistic high gods only has one level of social complexity!

Many of these criteria for a “high god” sound suspiciously like the Abrahamic god found in Christianity, Judaism, and Islam. Indeed, when you dig into the small number of moralizing high gods in the SCCS, it turns out that most of these gods either belong to an Abrahamic religion or have likely been influenced by long histories of contact with Abrahamic religions. Another issue stems from this emphasis on the evolution or emergence of high gods that are “specifically supportive of human morality.” Many current evolutionary approaches emphasize the uniqueness of these gods, as though such moralistic deities are especially good at inducing moral behavior. But again, they rely on the SCCS with its focuses on the creator and “ultimate governor” gods. Why does it matter in regard to moral behavior whether a god was also a creator god or not? As it turns out, many deities in the ethnographic world are morally concerned but don’t care much for engineering. These probably got coded as “absent or not reported.” Some theories hinge on the idea of morally concerned gods being unique when they might not be. Other theorists suspect that moral concern is far more prevalent than that reported in the SCCS. Only reliable methods, analyses, and adequate ethnography can really answer this question sufficiently.

This leads to another basic question: What do the non-moralistic gods care about? If we knew, we might be able to address whether or not different gods correspond to different kinds of social problems with different kinds of returns. For example, we do know that some traditional societies claim that gods punish people who hunt too much. Does belief in spirits’ punishment of environmental destruction actually contribute to conservation? More specifically, do such beliefs actually entail the short-term reduction of hunting for long-term, more reliable yields? Here, too, opinions abound while evidence is scant.

We and others have been working on these issues both ethnographically and with the development of cross-cultural databases. In our view, the best ethnography makes sense of why people do what they do by documenting the conceptual assumptions they have and detailing how the data was collected. Well-organized ethnographers can collect data together and then aggregate it to make systematic comparisons. While not without limitations, such a dataset allows researchers to address variation at the level of individuals, groups, and regions and can point to universals and divergences.

The aforementioned CERC project called “The Evolution of Religion and Morality Project” has been one effort to process and publish such data. This project links demographics and beliefs to economic behavior and measures of group cohesion and was composed of an interdisciplinary team of researchers interested in religion’s contributions to cooperation. Some recent efforts include an expansion of the number of field sites and experiments and projects that address some of the variation regarding what other gods care about. Some outputs from this project are growing, publicly available data sets that include data from far-reaching places. This data can be used for a wide variety of tests about religion or other topics. A recently published Nature article reports that beliefs in punitive and omniscient gods are associated with greater fairness toward other people, and a forthcoming special issue in Religion, Brain and Behavior contextualizes some of the results from eight different field sites. In summary, these works suggest that certain strains of religion may facilitate more reliable cooperation on scales much larger than traditional, face-to-face societies.

Recent years have also seen the construction of new cross-cultural databases of religion from populations both past and present. This includes things such as the Database of Religious History and the Pulotu Database of Pacific Religion. Like previous cross-cultural databases, these historical databases are built from existing descriptions of older and/or contemporary religious systems. While in some cases the questions that we can ask are limited by what ethnographers were interested in at the time, this new generation of databases offers many improvements over older databases such as the SCCS. For instance, they include variables on a wider range of religious features, are generally built on a critical analysis of a broader range of source materials, and provide direct citations for each data point.

Research using the Pulotu database shows that belief in moralizing high gods was scarce before the spread of Christianity and Islam, and that many of these cultures were able to build substantial and sophisticated social systems without that belief. However, there was no shortage of morally concerned, morally ambiguous, and morally depraved gods and spirits. Examples of morally concerned gods and spirits included spirits that would punish people for stealing from one another’s gardens, ancestor spirits that made sure their families behaved well toward one another, and gods who got furious at incest. Analyses provided some evidence that cultures with these kinds of morally concerned deities were more likely to gain political complexity than those without, which suggest that these kinds of beliefs might have served some kind of social function.

Together these different strands of cross-cultural research indicate that at least some supernatural beliefs and practices can cause people to behave in ways that serve evolutionary functions. We acknowledge that the results of our research are tentative and that more work is needed to identify the specific beliefs and practices that serve functions, the mechanism by which they provide fitness benefits, and whether they increase the fitness of individuals, sub-communities, or cultural groups. We caution that identifying evolutionary functions of a religious system does not lend any credence to its veracity, does not mean that it could serve these functions in all cultural systems, and does not mean that there are no better solutions out there. In fact, if it turns out that religion has had important functions, it should be all the more reason to strengthen participatory secular institutions. However, we wager that we cannot understand the evolution of religious systems and why they remain so common today without understanding the functions they serve.

Benjamin Grant Purzycki

Benjamin Grant Purzycki writes from the Max Planck Institute for Evolutionary Anthropology.

Joseph Watts

Joseph Watts writes from the Max Planck Institute for the Science of Human History.


The greatest challenge in advancing the social science of religion may lie in better organizing the social scientists.

This article is available to subscribers only.
Subscribe now or log in to read this article.