The Birth of High Gods How the Cultural Evolution of Supernatural Policing Influenced the Emergence of Complex, Cooperative Human Societies, Paving the Way for Civilization
Azim F. Shariff , Ara Norenzayan , and Joseph Henrich
Voltaire’s well-tread quote, “If God did not exist, it would be necessary to invent him,” was written with direct reference to the effectiveness of God as a supernatural policing agent (Voltaire & Redman, 1977). We argue that indeed supernatural policing was a driving force for the invention of God, but this invention—like so many other cultural products—was not the product of a brilliant religious mind or a committee of Machiavellian priests. Instead, omniscient, moralizing supernatural agents derived from a suite of religious beliefs that were culturally selected for their ability to galvanize cooperation in larger groups, promote in-group cohesion, and foster competition with other social groups. The emergence of religions, and modern world religions in particular, has been a cumulative process involving myriad interacting individuals that stretched over hundreds of generations of interacting individuals within the context of intergroup competition.
Humans are not just social, group-living animals but also highly cultural animals (Henrich, in press; Norenzayan, Schaller, & Heine, 2006). The cognitive and behavioral capacities that make human culture possible—complex communication skills, social learning mechanisms, and biased information processing that favors common traits and prestigious individuals—evolved because they allow individuals to readily adapt their behavior to the novel and changing environments at rates much faster than genetic evolution (Boyd & Richerson, 1998; Henrich & Boyd, 1998; Henrich & Gil-White, 2001; Richerson & Boyd, 2005; Tomasello, 1999; Tomasello, Kruger, & Ratner, 1993).
Natural selection has equipped many species with both individual and social learning capacities. As individuals of these species confront the challenges of survival and reproduction, they use their naturally evolved learning capacities to locally adapt. When encountering an evolutionarily novel food, crows and chimpanzees (just to name two) can individually figure out how to use tools for extracting the food (Hunt, 1996; McGrew, 1974). Chimps and dolphins can learn about these tools from conspecifics, who have already figured out the problems individually (Boesch & Tomasello, 1998; Rendell & Whiten, 2001). This means that evolutionary problems are often tackled first, in many species, by learning. Cultural evolution in humans has solved a vast range of evolutionary challenges, as the insights and accidents of generations accumulate and populations become increasingly better adapted (Boyd & Richerson, 1995). Clothing is a cultural adaptation to cold weather. Fire is an energy-saving and nutrient-releasing cultural adaptation to acquiring high-quality food that was shaped the subsequent evolution of our digestive system (Wrangham, Conkin-Brittain, 2003). The use of different spices across human societies shows that spicing, including tastes and recipes, is a cultural adaptation to meat-borne pathogens that are particularly dangerous in hot climates (Sherman & Billing, 1999). Inuit kayaks are culturally evolved engineering marvels that adapt this tropical primate to arctic hunting. These are true adaptations in the evolutionary psychological sense, because they are complex, functionally integrated solutions to recurrent ecological problems. But they are not directly the product of natural selection acting on genes (Richerson & Boyd, 2005) or evoked from domain-specific modules.
On the one hand, genetically evolved aspects of our minds and bodies can constrain cultural developments. And certainly genetic evolution laid the groundwork for the emergence of cultural learning and cultural evolution. On the other hand, however, cultural traits can arise and spread to address environment social problems, which in other species could be dealt with only by genetic evolution. For example, the omnivore’s dilemma (Rozin, 1987) suggests that the human capacity to eat a wide range of plant and animal products dramatically increased calorie intake and hence survival but also gave rise to selective pressures to avoid harmful substances (such as rotten meat, poisonous plants) that could have been lethal. Along with evolved psychological adaptations (e.g., the emotion of disgust), an interlocking set of culturally evolved beliefs, practices, and institutions (food taboos, hygiene rules, eating rituals) has shaped human diets in adaptive ways. Careful mathematical modeling of the interaction between cultural and genetic evolutionary processes shows that culture need not be on a tight “genetic leash.” Sometimes the cultural tail wags the genetic dog (Rogers, 1988), meaning that cultural evolution can drive genetic evolution by altering the selective environment faced by genes.
In this chapter we explore the idea that some of the central features of religion, and in particular those features that have spread so successfully since the origins of agriculture, have emerged via competition among different cultural groups, bearing different religious beliefs and practices. An integrated suite of religious beliefs, rituals, practices, and institutional forms thus evolved to address the evolutionary challenge of sustaining large-scale cooperation and exchange among nonrelatives. We further consider the possibility that these cultural evolutionary processes, if they have occurred over a sufficiently long time span, may have influenced the course of human genetic evolution in a process known as culture–gene coevolution. Culture and genes may have interacted to make certain aspects of religion— such as big gods—more “thinkable.”
To begin, we lay a foundation for this effort by summarizing an account of the cognitive capacities that underlie supernatural agent beliefs (gods, ghosts, ancestor spirits) as evolutionary by-products—natural selection did not favor these capacities because they gave rise to supernatural beliefs. Then we argue that the human capacity for deep commitment to such beliefs was exploited through the mechanisms of cultural evolution to serve as supernatural policing agents to solve the evolutionary problems associated with cooperative behavior in large, genetically unrelated groups. Although we are not the first to advance the idea that religion galvanizes cooperation within groups (for early discussions of religion and social cohesion, see Durkheim, 1912/1995; for recent treatments, see Irons, 1991; Johnson & Krueger, 2004; Sosis & Alcorta, 2003; Wilson, 2002), our aim is to argue for the central role of belief in supernatural agents (in addition to religious ritual) and a culturally evolved (rather than genetically evolved) explanation for these innovations. To do so, we must first visit the current discussion of religion’s place within the story of human evolution.
The Cognitive Archi tecture of God Concepts
Several theorists of religion (e.g., Johnson & Bering, 2006; Landau, Greenberg, & Solomon, 2005) have argued that religion is a naturally selected genetic adaptation— a trait complex, in the same way that the vertebrate eye, or echolocation in bats, is an adaptation that has conferred a reproductive advantage to ancestral organisms. Such arguments need to fulfill the strict criteria of adaptive design that are the standard in evolutionary biology: compelling adaptive function in ancestral environments, unitary and complex design, efficiency, precision, specificity, economy, and reliability (cf. Williams, 1966). Such a model also needs to rule out both the possibility that religion is a cultural by-product of adaptive design (Atran & Norenzayan, 2004) and the possibility that it is not a product of adaptive cultural learning processes (Henrich & McElreath, 2006; Richerson & Boyd, 2005), of the kind that produced adaptations such as kayaks and spicing in food preparation recipes. As we argue in this chapter, religion fulfills none of these criteria (for similar views, see Atran, 2002; Bloom, 2005; Boyer, 2001; Kirkpatrick, 1999).
Instead, we argue that religion is not an evolutionary adaptation per se. In fact religion is not a unitary thing; it simply points to a family resemblance category of converging sets of cultural by-products, rooted in innate psychological tendencies that constrain and channel the transmission and survival of religious beliefs and practices. These four converging paths are counterintuition (supernatural agents), commitment (motivation belief in counterintuitive agents, displays in costly sacrifices), compassion (relieving existential anxieties), and communion (ritual) (Atran & Norenzayan, 2004). These psychological criteria—the four Cs of religion—are themselves cultural manipulations of psychological adaptations (agency detection, costly commitment) or panhuman existential concerns (fear of death, of social deception), and many belief systems in many places do not even have all four (Johnson, 2003). Religions evolve along culturally distinct though partially convergent paths that are constrained by a complex evolutionary landscape reflecting cognitive, emotional, and material conditions for ordinary social life. Given the mental and social realities of this landscape, certain religious elements are more likely to proliferate. For example, in terms of what supernatural agents come to be believed, there is an optimal balance of how much these beings conform to and how much they violate our intuitive assumptions about physical, biological, and psychological phenomena. The proliferation sweet spot is a minimally counterintuitive supernatural being—super enough to capture attention, and natural enough to still make sense.
The combination of an intuitive conceptual grounding and an interesting nonintuitiveness makes beliefs more likely to be transmitted and retained in a population than random departures from common sense. On the one hand, category violations that shake basic notions of ontology are attention arresting and hence resistant to memory degradation. Only if the resultant impossible worlds remain bridged to the everyday world, however, can information be stored, evoked, and transmitted (Atran & Norenzayan, 2004; Atran & Sperber, 1991; Boyer, 1996). Several lines of experiments support these assertions, indicating that minimally counterintuitive concepts (Barrett & Nyhof, 2001; Boyer & Ramble, 2001) as well as minimally counterintuitive narrative structures such as folktales (Norenzayan, Atran, Faulkner, & Schaller, 2006) have a cognitive advantage over other cognitive templates, be they entirely intuitive or maximally counterintuitive. Once these beliefs are cognitive selected, they are available to undergo cultural selection and stabilization. In what follows, we explore how cultural evolutionary processes may have selected among the potential pool of readily transmittable beliefs to expand and galvanize cooperative behavior in large social groups.
Cooperation in Large Groups
The social environment of religion’s infancy was one likely characterized by relatively small groups. These groups were held together by a few behavioral mechanisms that have genetically evolved in nonhuman species to permit limited amounts of cooperation. Social organisms confront a tension between the stability and cooperativeness of the social group, on the one hand, and the selfishness of the individual, on the other. Although group living conveys many advantages to individual members (e.g., avoidance of and protection from predators), there are many potentially cooperative circumstances in which it is more advantageous for individuals to evade contributing to the collective and free riding on the contributions of others. This strategy will, unchecked, prove so successful that it will overrun an entire population, making group living an impossibility.
As a result, the evolutionary mechanisms of kin selection and reciprocal altruism have favored the emergence of altruism toward relatives and in reciprocal dyads or very small groups. Among humans, indirect reciprocity, wherein reputations can be ascertained by third parties rather than only through personal interactions, has increased the number of potential dyadic partners. Indirect reciprocity, however, does not increase the size of the cooperative group and operates effectively only so far as these reputations can be very reliably transmitted and recalled for most potential partners (Henrich & Henrich, 2007). None of these mechanisms permits large-scale cooperation.
Thus, though humans have evolved to use each of these strategies, the extent of human social interaction was still, for much of human history, limited to cooperation in very small groups. There are two ways in which human sociality was limited. First, kin selection and reciprocity are limited to small cooperative units of two or three individuals and cannot explain interactions in which large of numbers cooperate in the same unit, such as in warfare, group hunting and food sharing, recycling, blood donation, voting, or community house construction. Second, because groups were likely regulated by reputational information and personal relationship, this caps the size at which individuals can maintain a generalized sense of trust toward fellow group members. Extrapolating from neocortex size, Dunbar (2003) estimated that human brains were designed to manage ancestral groups of about 150 members. Beyond this number, unfamiliarity abounds, trust disintegrates, reciprocity is compromised, and groups divide or collapse. Although this specific number can be disputed (e.g., Smith, 1996), it is apparent today from the size of modern human settlements that solutions have been found to the limitations that used to make such settlements unstable. This effect is demonstrated in ethnographic work in part of New Guinea, where villages routinely split once they exceed about 300 people (i.e., 150 adults). Tuzin (1976, 2001) detailed the historical emergence of an anomalous village of 1,500 people and showed how culturally evolved beliefs about social organization, marriage, norms, rituals, and supernatural agents converged to maintain harmony and galvanize cooperation in a locale where this scale was previously unknown.
Archaeological evidence makes clear that human societies had begun to “scale up” group size and the scale of cooperation between 14,000 and 12,000 years ago, as the Pleistocene gave way to the Holocene and the preagricultural villages of the Natufians gave way to towns such as Jericho (Cauvin, 1999). A number of innovations— all necessary, none sufficient—emerged around this time that allowed larger populations to live relatively harmoniously in cohesive groups. Revolutions in agriculture, hierarchical political organization, and, we argue, religious beliefs and associated costly rituals made such settlements sustainable.1
The Role of Gods in Promoting Cooperation
Emerging religious belief systems, we suggest, increased trust among unrelated individuals, allowing cooperation to expand beyond the small groups to which it had been previously limited. There is empirical evidence that religion, today, facilitates trust and cooperation among genetically unrelated individuals. Recently, Tan and Vogel (2005) examined religiosity in the context of a trust game. The results were clear: Religious trustees were trusted more, particularly by religious trusters, and religious trustees were indeed more cooperative in turn toward the trusters. Importantly, these findings were not reducible to ingroup–outgroup processes. Consistent with these results, findings by Gervais, Shariff, and Norenzayan (2007) reveal that prejudice toward atheists is mainly driven by moral distrust rather than by visceral antipathy, as is the case for ethnic prejudice (Allport, 1954). Sosis and Ruffle (2003) examined the link between religion and cooperative behavior in Israeli kibbutzim. They found that religious kibbutz members were more cooperative than secular members, and religious attendance predicted cooperative decision making, controlling for a number of variables. In a different analysis, Sosis and colleagues compared the longevity of religious and secular communes in 19thcentury America (Sosis & Bressler, 2003). For any given year, religious communes were found to outlast those driven by secular ideologies, such as socialism, by a factor of four. The remarkable survival value of religion could be explained by the cooperative advantages that it confers to groups. But what accounts for these seemingly religiously derived cooperation and trust benefits?2
We hypothesize that cultural evolution favored the emergence of an interrelated suite of beliefs about the traits of supernatural agents. As background, the religions of small-scale societies including foragers often do not have one or two powerful gods who are markedly associated with moral behavior (Roes & Raymond, 2003). Many gods are ambivalent or whimsical, even creator gods. Gods, in most smallscale societies, are not omniscient or omnipotent. Notions of a pleasant afterlife appear to be a relatively recent innovation (McNeill, 1991). We suggest that moralizing high gods gradually moved to the forefront of religious systems as cultural evolution—driven by processes favoring larger, more cooperative, more harmonious groups—favored rituals and practices that instill greater degrees of committed belief in people about gods who (a) cared about cooperative- and harmony-enhancing behavior (the group’s moral norms), (b) could and would reward and punish appropriately, and (c) had the power to monitor all behavior all the time. These religious beliefs helped expand the sphere of human cooperation. In particular, we suggest that the fear of imagined supernatural policing agents helped overcome the constraints imposed on the scale of human social interaction and cooperation by our kin and reciprocity-based psychologies.
The omniscience of these agents extends one’s vulnerability of “being caught” to all times and all places. Some gods can even read people’s thoughts. Moreover, there are no restrictions on how many transgressions these supernatural agents can keep track of. The consequence is that “hidden defection,” which was still a viable individual strategy in groups with indirect reciprocity, is markedly reduced.
Partially outsourcing not only the monitoring but also the punishing aspects of cheater detection to supernatural agents also contributes to addressing the problem of costly punishment (Johnson & Bering, 2006). The costliness of punishing cheaters (through both the act of punishing and the potential retribution for this act) itself creates a second order of cheaters—those who free ride on their punishing duties. This is a problem that can extend, at least theoretically, ad infinitum (Henrich & Boyd, 2001). Because supernatural agents are not generally thought to be privy to the same concerns as men, they can be seen to punish without cost or fear of retribution. Finally, the belief that the punishments of moralizing high gods are accurate and complete is favored by cultural evolution. The idea that no one escapes the omniscient judge may help satisfy human intuitions about fairness and justice (Haidt & Joseph, in press). The belief in a supernatural watcher can extend the otherwise limited scope of human cooperation, effectively infinitely, provided that the fear of these supernatural beings reaches a near-ubiquitous distribution in the group.
A growing body of empirical support bolsters these claims (see Norenzayan & Shariff, 2008). Snarey (1996) examined the features of god concepts across cultures as a function of life-threatening water scarcity. Societies with high water scarcity were more likely to have morally concerned deities who encouraged the prosocial use of natural resources. This finding held even when controlling for cultural diffusion of high gods via missionary activities. Thus, high gods were culturally selected when freeloading was particularly detrimental to the cohesiveness of the social group.
In a similar cross-cultural analysis, Roes and Raymond (2003) predicted, and found, that across cultures, large societies are associated with moralizing high gods—group size was correlated with the existence of supernatural watchers who are concerned about the morality of human interactions. This finding held controlling for the cultural diffusion of high gods via missionary activity, as well as for societal inequality.
In societies with moralizing gods, a fear of supernatural agents among individuals can be evoked simply to enforce moral norms. In one study, children were explicitly told not to look inside a box and then left alone in the room with it (Bering, 2003). Those who were previously told that a fictional supernatural agent, Princess Alice, is watching were significantly less likely to peek inside the forbidden box. A later study (Bering, 2006) found a similar effect in university students. Those who were casually told that the ghost of a dead student had been spotted in the experimental room were less willing to cheat on a rigged computer task.
If reminders of a supernatural agent can reduce cheating, reminders of a moralizing high god may reduce selfish behavior and increase generosity, even toward strangers. Shariff and Norenzayan (2007) tested this possibility. Participants who were implicitly primed with god concepts behaved more altruistically in an economic game measuring fair behavior than those receiving either a neutral prime or no prime at all. In an anonymous, non-iterated version of the “dictator game,” participants were randomly assigned to be either the giver or the receiver. Those assigned to the role of the giver were allotted $10, which they were given the opportunity to share—in any amount they saw fit—with the receiver, who would otherwise receive nothing. Assured anonymity from the other player and confidentiality in their decision, 38% of givers in the control conditions kept all the money for themselves. This figure fell to 14% for participants implicitly primed with god concepts. At the same time, the proportion offering $5 to the receiver—an even half of the money—rose from 20% in the control conditions to 48% in the religiously primed condition. Among non-student atheists, however, the god primes had no effect. Subsequent studies showed that this effect is not explainable in terms of changes in positive or negative mood or in terms of increases in feelings of empathy.
Although other interpretations are possible, these results suggest that the imagined presence of supernatural watchers can reduce selfishness and increase the adherence to fairness norms, even among anonymous strangers. Throughout history, this combination of cheating reduction and generosity fostering would have proved even more effective at stabilizing large societies than cheating reduction on its own. But is this suite of beliefs surrounding moralizing high gods a product of long-term cultural evolution or a reliably developing product of genetic evolution and thus a piece of human nature? Like most of human thought and behavior, there will undoubtedly be influences from both genetic evolution and cultural evolution on these beliefs. Certainly, as discussed earlier, the mental capacities that make such beliefs plausible, even thinkable, are the product of the genetic evolution. Equally certainly, the specific content of religious beliefs, such as the belief in Old Man Coyote, Vishnu, or the Abrahamic God, is transmitted culturally. A better question, then, is to what extent and which specific details of religious beliefs in supernatural watchers are culturally rather than biologically evolved. This is where the debate begins.www.psychspace.com心理学空间网
