ANRV355-AN37-07 ARI 7 May 2008 17:26 V I E E W R S I E N C N A D V A Evolutionary Perspectives on Religion Pascal Boyer1,2 and Brian Bergstrom1 Departments of Psychology1 and Anthropology2, Washington University, St. Louis, Missouri 63130; email: [email protected], [email protected] Annu. Rev. Anthropol. 2008. 37:111–130 Key Words The Annual Review of Anthropology is online at cognition, cooperation, beliefs, morality, institutions anthro.annualreviews.org This article’s doi: Abstract 10.1146/annurev.anthro.37.081407.085201 Recent work in biology, cognitive psychology, and archaeology has re- Copyright c 2008 by Annual Reviews. newed evolutionary perspectives on the role of natural selection in the All rights reserved emergence and recurrent forms of religious thought and behavior, i.e., 0084-6570/08/1021-0111$20.00 mental representations of supernatural agents, as well as artifacts, rit- ual practices, moral systems, ethnic markers, and specific experiences associated with these representations. One perspective, inspired from behavioral ecology, attempts to measure the fitness effects of religious practices. Another set of models, representative of evolutionary psy- chology, explain religious thought and behavior as the output of cog- nitive systems (e.g., animacy detection, social cognition, precautionary reasoning) that are not exclusive to the religious domain. In both per- pectives, the question remains open, whether religious thought and behavior constitute an adaptation or a by-product of adaptive cognitive function. 111 ANRV355-AN37-07 ARI 7 May 2008 17:26 INTRODUCTION nonphysical agents or communicate with them; Understanding religious thought and behav- ethnic affiliation and coalitional processes ior as consequences of human nature is an old linked to nonphysical agents; and project, perhaps as old as the systematic, reflec- evolutionary models, such as other ex- tive examination of belief in gods and spirits. A planatory models in anthropology, as- notion of “natural religion” as the result of fun- sume cross-cultural commonalities in damental human propensities is familiar, for in- each of these domains of thought and be- stance, in both Western and Muslim traditions havior. (Pailin 1984, Reinhart 1986). Understanding Does this collection of features constitute a religion as a result of evolution by natural se- domain of “religion”? As we demonstrate be- lection is obviously a more recent research pro- low, the models described here do not assume gram, even though the first rudiments of such a that the features listed above are always found project can be found in Darwin himself (Darwin together. The only assumption is that evolution 1871). In the past 20 years, a number of psy- provides the context for understanding some or chologists, anthropologists, religious scholars, many of the phenomena listed. In this sense the and evolutionary biologists have put forward term religion is to an evolutionary anthropolo- a new evolutionary perspective, understanding gist what “tree” is to an evolutionary botanist, religion as one among the many domains of cul- a common prescientific category that may need tural activity that were shaped by human evo- to be replaced with other, causally grounded, lutionary history (Hinde 1999). scientific categories. Are religious phenomena sui generis? A re- THE RELIGIOUS DOMAIN lated but distinct question is whether “reli- gious” is a specific property of the phenomena What the term religion denotes is widely dis- listed above, such that they would all differ from puted in contemporary anthropology and reli- their “nonreligious” counterparts. For instance, gious studies (Saler 1993), so it may be of help is a religious ritual different from a nonreligious to start with a rough demarcation of the field one, and religious morality intrinsically differ- of inquiry. Evolutionary models are supposed ent from the nonreligious kind? The models re- to explain a whole collection of behaviors and viewed here make no strong assumption about mental representations that are found in many that. Indeed, most of them lead to deflationary different human groups, including the follow- accounts of religion, in which the phenomena in ing: question are explained in terms of processes that mental representations of nonphysical would operate in the same way in other contexts agents, including ghosts, ancestors, spir- (Lawson & McCauley 1990). In this sense, what its, gods, ghouls, witches, etc., and beliefs explains religious thought or behavior may also about the existence and features of these explain many other domains of cultural thought agents; and behavior (Saler 1993). This pits evolution- artifacts associated with those mental rep- ary models, like most other anthropological resentations, such as statues, amulets, or accounts, against classical assumptions in the other visual representations or symbols; study of religion, following which there is a spe- ritual practices associated with stipulated cific quality of religious phenomena (thought, nonphysical agents; experience, emotion, etc.) that requires expla- moral intuitions as well as explicit moral nation [see Eliade (1959), Otto (1959) and a understandings that people in a particular discussion in Wiebe (1998)]. group connect to nonphysical agency; Are there religious universals? Some fea- specific forms of experience intended to tures of religion may well constitute substantive either bring about some proximity to cultural universals (Brown 1991). This however 112 Boyer · Bergstrom ANRV355-AN37-07 ARI 7 May 2008 17:26 is not crucial to using evolutionary models, ism is the most archaic form of religious be- whose aim is to explain the variance observed in havior (Eliade & Trask 1964, Shirokogoroff terms of some common factors. Thus explana- 1935). However, such inferences are fraught tions should be sought for all cultural phenom- with problems, as we discuss below (see Experi- ena whose recurrence is clearly above chance. ence and Commitment). Cues to special experi- In many domains, evolution resulted in disposi- ence do not necessarily indicate that such expe- tions that render humans sensitive to particular rience was interpreted in terms of nonphysical contextual input, but evolved human disposi- agency or that religious specialists like modern tions do not always result in uniform behaviors shamans existed. or cultural outputs. Most evolutionary approaches to religion Was religion present in ancestral times? If actually eschew speculation on archaic forms religion, or at least the phenomena described of religious behavior and experience. Rather, a above, were influenced by natural selection, we common strategy is to identify the capacities should expect at least some of these behaviors to and behaviors universally involved in religious be ancient as well as widespread. In the archae- thought or behavior and to relate them to plau- ological record, we find evidence for a variety sible selective pressure. In this sense, evolution- of nonpragmatic behaviors, such as elaborate ary anthropologists of religion adopt the same burial procedures, from the earliest stages of the strategy as other evolutionary behavioral scien- Paleolithic and perhaps also in Neanderthals tists do—that of measuring the contributions (Trinkaus & Shipman 1993). Also, we find ev- of various behaviors and behavioral strategies idence for supernatural concepts, chimeras for to fitness. instance, from early stages of modern human cultures (Mithen 1999). Did these behaviors and concepts already constitute “religion”? The EVOLUTIONARY BACKGROUND question only makes sense if we assume that Genetic and Cultural Evolution “religion” stands for a natural kind, an inte- grated package, which is probably not the case. The theoretical background to models of re- More important, the archaeological evidence ligion is a specific approach to cultural phe- shows that many of the phenomena discussed nomena and was developed in the past 20 here appear at the same point (about 50 kya) years by evolutionarily inspired cultural anthro- along with other phenomena typical of modern pologists (Boyd & Richerson 1985, Durham humans, such as regional “cultural” differences 1991, Lumsden & Wilson 1981, Sperber 1985). (Richerson & Boyd 2006), sophisticated tool- A common assumption among these differ- equipment (Mithen 1996), body ornamentation ent, partly overlapping frameworks is that and make-up (Hovers et al. 2003), and probably what we observe as cultural representations the first musical instruments (Falk 2000). and practices are variants (of cultural traits), What was early “religion” like? There is found in roughly similar forms in a particu- no straightforward way of connecting inferred lar place or group because they have resisted past practices to their modern equivalents, es- change and distortion through innumerable pecially to the beliefs and experiences associated processes of acquisition, storage, inference, and with religious thoughts. Some of the evidence, communication. such as rock art similar to the visual phenom- In these models, the spread of specific vari- ena induced by trance and altered states of con- ants of cultural representations (such as a par- sciousness (Pearson 2001), would point to a ticular religious belief or concept represented Paleolithic form of shamanism (Hayden 2003) by a human mind) is seen as partly analogous and ecstatic experience (Dornan 2004). This to the spread of alleles in a gene pool. In par- would be consistent with a traditional assump- ticular,