The evolution of religion
Abstract and Keywords
This article reviews recent theoretical and experimental work in the cognitive and evolutionary psychology of religion, and motivates an adaptationist stance. By ‘religious cognition’ and ‘religiosity’, the article means belief and practice relative to supernatural beings and powers — call them ‘gods’. It suggests that religiosity — god commitment and the behaviour it prompts and that motivate it — bears the hallmark of adaptive design. Adaptive religiosity is more like an achievement than a harmless mistake, as spandrelists urge, or catching a cold, as meme theorists urge. It begins with a serious obstacle to any evolutionary treatment of religion. Science assumes methodological naturalism. The article does not stipulate the gods out of existence. However, it suggests the idea that nature is secular and provides a good explanation from minimal assumptions about the complexity of the world.
In this chapter, I review recent theoretical and experimental work in the cognitive and evolutionary psychology of religion, and motivate an adaptationist stance. By ‘religious cognition’ and ‘religiosity’ I mean belief and practice relative to the supernatural beings and powers—call them ‘gods’.
Much of the work in the naturalistic study of religion views religiosity as an after-effect or spandrel of a mind not designed for religion. Some theorists explain religion as the product of viral-like ideas working for their own success; on this view, religion is an adaptation but not ours (Dawkins, 1976; Blackmore, 2000; Dennett, 2006). Other researchers urge that religion's adaptive features emerge at the level of religious groups; these features (where they exist) are best studied in virtue of cultural evolution. We inherit religious ideas and practices (some maladaptive) from our parents through a flow of ideas, not genes (Boyd and Richerson, 1985; Richerson and Boyd, 1989; Wilson, 2002; Richerson and Boyd, 2005). Here religion may be adaptive without being an adaptation, a species trait. Others maintain that genetic resources predispose some individuals to spirituality but there is individual variation in these dispositions. The relevant traits do not describe god commitment but rather describe ‘spirituality’ (Hamer, 2004). Presumably, selection maintains variation in these dispositions through frequency-dependent selection on the relevant alleles. It is not yet clear how to integrate the inchoate studies on spirituality genes with the separate (though perhaps related) phenomena of belief and practices relative to the supernatural powers and agents; for spirituality genes are not god genes.
Below I suggest that religiosity—god commitment and the behaviour it prompts and that motivate it—bears the hallmark of adaptive design. Religious cognition solves functionally useful and enduring problems that would have confronted ancestral hominids, and there is evidence that we possess a very specific and otherwise improbable mental architecture to do its business. Adaptive (indeed non-lethal) religiosity is more like an achievement than a harmless mistake, as spandrelists urge, or catching a cold, as meme theorists urge. And the cognitive demands on the intergenerational flow of religious ideas are substantial, allowing selection to target and amplify dedicated cognitive architecture to facilitate the transmission and improvement of religious ideas and practices. So I will contend.
Let's begin with a serious obstacle to any evolutionary treatment of religion. Science assumes methodological naturalism. We do not stipulate the gods out of existence. But we begin with the idea that nature is secular and see how far this assumption takes us. For all we know naturalists of the distant future will appeal to gods in their explanations for our belief in them. But for now, our aim is to produce a good explanation from minimal assumptions about the complexity of the world.
(p. 622) From the naturalistic stance religiosity is striking. If the gods do not exist then religious persons systematically err in their judgements about the world. This is puzzling because cognition evolved to enable agents to get the world right. For misunderstanding the world typically comes at a cost to an organism's inclusive fitness. Given the fitness costs of error, it seems that selection should have weeded out cognitive features that allow us to produce religious mistakes (Godfrey-Smith, 1996; Sterelny, 2003).
Yet religious commitment is rampant. It appears in all cultures, is nearly universal among foragers (whose life-ways resembles those of our distant ancestors), and even remains popular in scientifically advanced societies. Where ancient religious traditions are losing their hold, new-age spiritualities, astrologies, and spiritualist fusions are rushing in (e.g. Reader and Tanabe, 1998). Whatever one's position on the spandrel/adaptation question, we know natural selection did not substantially cull the psychological structures that allow for supernatural commitment. Religion remains persuasive.
More puzzling, religious commitment stimulates reproductively costly behaviour. Some of these costs are substantial. Ritual participants subject themselves to fatigue, exposure, injury, disease, bodily discomfort, and impairment. At the extreme, practices of suicide bombing, genital mutilation and celibacy are straightforwardly devastating to gene lineages prone to them. Yet even mundane religious practices come with opportunity costs: sitting in silence, bowing before statues, flattering priests, defers the commerce of life. Religion saddles the committed with practical costs. And if you are a religious eunuch, self-immolator, suicide soldier, or celibataire, these saddles can be heavy indeed.
It is easy to imagine that selection permitted religiosity because its benefits outweigh these harms. Naturalists have discerned important functionality: religion enhances solidarity, provides existential purpose, explains our origins and fates, and activates placebo healing (Preus, 1987; Sosis and Alcorta, 2003). Some question the adequacy of these functional interpretations (Boyer, 2001; Dennett, 2006). Yet even conceding this functionality, naturalists have underrated the explanatory problem. For there appear to be less costly, evolvable designs. It remains mysterious why functions related to health would hinge on priestcraft: our interests do not converge perfectly. Similarly baffling is how trials by fire or witch hunts or celibate lifestyles feed the need for scientific explanation. Selection should prefer no science to bad science because the costs of bad science are manifold (e.g. arsenic as a treatment for hysteria, etc.: Shapiro and Shapiro, 1997). Likewise it is hard to understand how fellow feeling and solidarity might benefit from silent meditation, walking over coals, or livestock sacrifice. The strong bonds of friendship, for example, require no extraneous theology, trials by fire, or slaughter. Whether religion provides meaning and solace is debatable. For the gods can be capricious, arbitrary, and terrifying (Berger, 1990). More problematically, existential worries themselves need to be explained, for philosophical anxiety distracts us from the tasks of life.
The epistemic and practical costs of religion leave us with the cost problem. Organisms shouldn't think and act as we do (Atran, 2002, 2004; Bulbulia, 2004a; see also Dennett, 2006, for thorough discussion). Why then, did natural selection not reduce or eliminate religious tendencies?
43.2. Spandrel theories
The standard picture of cognitive evolution presents selection as incrementally assembling ever more intricate and discerning cognitive designs. Gradually, lineages get better at appreciating and responding to stable features of their environments relevant to survival. Perceptual systems become more resolved, reflexes more finely attuned, and behavioural responses more elaborate, flexible, and precise. Selection gradually makes animals more god-like.
Much of our cognitive complexity comports to this picture, but the picture needs to be revised. Selection must factor the evolutionary costs and accessibility of accurate registration against alternative designs. Getting the world right is no easy task. Our evolutionary interests frequently diverge from those around us. Indeed, large chunks of the world are hostile to our interests. Organisms pollute the epistemic environments of their rivals; the signals an agent receives are often noisy. Filtering the noise to (p. 623) accurately represent the world is computationally intensive, and there may not be enough time, information, or cognitive resource to get it right. Frequently error will satisifice a fitness function. Assigning cost and benefits to response strategies is difficult, for not all parameters are known; yet abstractly we know that where:
Average reproductive costs of perfection of accuracy < Average costs of inaccuracy,
All things being equal, selection will favour inaccuracy.
We have assumed that there are no gods in nature. This is true of the gods we worship, and it is also true of ourselves. The first wave of cognitive research on religion began with an appreciation of human imperfection. The field's vanguards understood natural selection's power to build intricate psychological architecture, but they also understood that this power is constrained. In an early work in the field, Stewart Guthrie argued that religious experience emerges from dispositions to overgeneralize agency (Guthrie, 1993). We believe the world is filled with human agents because hair-trigger perceptual-response strategies animate the world around us. Our tendency to anthropomorphism optimizes a tough problem; the world is ambiguous. When sighting persons, false positives bring little cost. Yet false negatives can be disastrous. Other persons are critical to our existence and progress in life; they also provide the most significant threats. Hence, we need to find other agents when they are there. Given the benefits of agent detection, the low costs of false positives when compared with false negatives, and the stability of this equation over phylogenic time, a ‘Hyperactive Agent Detection Device’ (HADD) looks evolvable (‘HADD’ is a term coined by Justin Barrett, 2000).
This conjecture is amenable to empirical testing, and Guthrie (1993, 2001) marshals much evidence that we anthropomorphize. We see faces in the clouds, dramas in inkblots, night shadows as villains; and we see the supernatural everywhere.
Yet clearly this story cannot explain all religious thought and behaviour. In response to ambiguous person sightings, we do not typically tattoo our faces, or spend decades building Gothic cathedrals, or give up sex for life. At best, HADD explains an aspect of a more elaborate causal process. Furthermore, HADD mischaracterizes the perceptual aspects of religious cognition. If religion were simply based on fast and dirty perceptual heuristics, we could easily correct religious errors. You leap, note it was only the wind, and cease worrying. Yet we generally don't correct for supernatural mistakes in this way. A Hindu won't give up Shiva worship when you tell her it was only the wind. Guthrie's book did not ignite a worldwide movement to atheism—‘Alas God was only a face in the cloud.’ Instead, religious judgement becomes easily entrenched. It resists fast and substantial revision (Berger and Luckmann, 1972). Yet such entrenchment usually does not hold off inferences to the best explanation in other domains, certainly not in most perceptual domains.
Perhaps a more significant problem here is that the gods are never experienced as human agents. We do not, strictly speaking, project humanity into the world. We project super-humanity. And why we do so remains mysterious on Guthrie's account. For if religious experience happens because it is important to find people, why does religious understanding never turn on people-sightings?
About the time Guthrie was developing his HADD account, the cognitive anthropologist Pascal Boyer was developing an alternative that addresses this explanatory lacuna (Boyer, 1992, 1994; for similar accounts see also Pyysiainen, 2001; Atran, 2004). The central theme running throughout Boyer's work is that religious ideas are conceptually configured to activate multiple features of the intuitive inference systems that govern our ordinary understandings of the world. For Boyer, religious ideas are memorable precisely because they violate our intuitive expectations—that is, because they are supernatural.
Notice: culture matters to how we understand and engage with the world and each other. Yet psychologists have discovered that many features of cognitive processing and development closely resemble the function and growth of biological organs. Children know more than they learn. The perception and inference engines that guide us through hostile reality need to operate quickly. And for invariant aspects of our world, richly structured and rapid cognitive processing can emerge in a lineage without strong external scaffolding. Large parts of our (p. 624) tracking and response systems function implicitly. Children appear to acquire folk physics, much of language and agent-psychology, and basic biological reasoning without explicit or heavily structured learning environments (Gelman and Markman, 1987; Gelman and Kremer, 1991; Spelke et al., 1995) though see (Sterelny, 2003, 2006).
Boyer has shown experimentally that ‘supernatural’ concepts are intrinsically memorable (Boyer and Ramble, 2001). He conjectures that because supernatural concepts violate our intuitive expectations for the relevant natural kinds, they are both striking and present fewer memory demands. The gods are unusual and uncanny agents. We heed them precisely because they are so counter-intuitive and strange. Boyer further observes there are constraints on the degree to which intuition-violating concepts are memorable. For counter-intuitive concepts are subject to threshold effects. Minimally counterintuitive concepts come with low memory demands: intuitive psychology dictates how we think about them, save the striking, attention-grabbing difference. God thinks like a man, and is in most respects like a man, but unlike a man created everything and can destroy everything. The notion that there is such a being is interesting and thus memorable. However, the architectonic multiple-intuition-violating gods of the philosophers—those superconfigurations of properties described in lengthy theological tomes—are not memorable. For we cannot rely on folk psychology to guide our intuitions about all the relevant properties of these matrix gods. Once supernatural concepts become too unusual and extraordinary, we are neither interested nor influenced (Boyer and Ramble, 2001).
It is crucial to Boyer's (2001) story that no single cognitive mechanism is responsible for religious thought. ‘Religiosity’ is not a natural kind. Nor has religiosity evolved to benefit religious agents or groups (p. 330): “There is no religious instinct, no specific inclination in the mind, no particular disposition for these concepts, no special religion center in the brain, and religious persons are not different from nonreligious ones in essential cognitive functions.” Instead, many different systems grinding through the same conceptual information produces behaviour we in ordinary language loosely describe under a common adjective ‘religious’. But this language is inadequate to science (p. 330): “Instead of a religious mind, what we have found is a whole frustration of invisible hands.”
There is much experimental evidence to support Boyer's account (reviewed by Boyer, 2003). The theory predicts that religious persons will not process religious commitments as elaborate theological doctrines of the kind religious elites promote, and in which, during reflective moments, religious persons claim to believe. Instead the theory predicts that religious agents will think of their gods as persons, save for a few extraordinary properties. In an important experiment, Justin Barrett and Frank Keil (1996, 1998) confirmed that religious devotees in America and India represent their gods anthropomorphically. Indeed, Barrett and Keil found that such representations conflict with believers' expressed theological commitments. For example, participants in these experiments assent to the theological proposition that their god is all knowing, but nevertheless believe that prayer is sometimes necessary to communicate thoughts to the divine. So it appears Guthrie was almost right: minimally adjusted anthropomorphic thought is religiosity's universal mode (see Barrett, 2004).
Does Boyer's spandrel theory adequately explain why we believe in the gods, and commit passionately to them? Spandrel theory manifests the theoretical virtue of parsimony. It is better if religion can be explained without reference to a special-purpose mind design, for we should not look for complexity beyond necessity. Yet I am among a growing number of naturalists who are not satisfied that religion is an evolutionary accident (Irons, 1996, 2001; Sosis, 2003, 2004; Bering, 2004; Bulbulia, 2004a,b; Dunbar, 2005; Johnson, 2005). In our view, spandrel explanations fail to explain why religious agents strongly commit to supernatural reality in a natural world lacking supernatural inputs. Memory of a concept is one thing. Giving over your life is another. Adaptationists do not believe that spandrelizing religion solves the cost problem. In our view, it is nearly miraculous (if you will) that agents come to identify so strongly with gods. The poverty of stimulus here is almost complete. There are no dragon-tongued gods to generate dragon-tongue god beliefs. Moreover selection is (p. 625) highly conservative. Were religious cost maladaptive, selection should have reduced and eliminated religious inclinations as it compared alternative designs over hundreds of generations. Indeed there are at least three evolutionary pathways bypassing costly religious cognition. (i) Any gradual improvement in epistemic accuracy would have displaced religious errors, with flow-on effects to practical costs. And so were religion maladaptive, selection would have targeted dispositions to meet counter-intuitive information with scoffing incredulity rather than pious commitment. Furthermore, (ii) by holding accuracy fixed but minimizing the error rate, selection would have reduced epistemic costs, thereby undercutting the basis for religious practice. If we only thought to believe in Zeus twice in our lives etc., we'd have saved much sacrificial cattle. Finally, (iii) severing the motivational connections linking belief to practice would have rendered religious beliefs, however common, inert and nearly harmless. If religious belief were dislodged from motivational processing it would cause less maladaptive harm. Without religious passion there would be no heavy reproductive cross to drag. Yet selection did not nudge religious cognition in any of these three directions. If anything, the motivational knobs of religious conviction and emotion are set too high.
To shed light on religion's adaptive functionality, I urge viewing the costs surrounding religion as adaptive signals. Religiosity may be adaptive in other ways (for example, see an account of healing functions in McClenon, 2002). Here I will focus only on the signalling dimension. To motivate this adaptationist stance I need to introduce two conceptual tools: (i) elementary game theory, from economics, and (ii) costly signalling theory, from behavioural ecology.
43.3. Game theory and the evolutionary dynamics of cooperation
An important factor in the biological success of our lineage has been our capacity to secure reliable cooperation among non-kin (Sterelny, 2003). Working together, agents can pool their resources to achieve more than going it alone. For many organisms it makes sense to team up with others, and many lineages exhibit powerful social tendencies. Fish school to reduce the risks of predation and to increase vigilance. A large cohort brings lower odds that the local predator will eat you; lots of eyes are better than two. Gray wolves coordinate their attacks on large prey; lots of jaws bring down the beast, and a shared kill is better than none at all.
We are an impressively cooperative primate. Indeed, our encephalization appears to have been driven by the demands of social life (Dunbar, 1998). We think well to cooperate well. By co-coordinating our plans, and anchoring individual fates to collective outcomes, agents are able to promote their interests more effectively than by going it alone. Yet our alliances are notoriously unstable. Defectors benefit by stealing the advantages of collective life without sacrifice. Exchange usually pays better than it costs. But defecting from exchange, taking its spoils without paying a price, often pays better still. And where many defect, the benefits of cooperation are lost (Skyrms, 1996).
To understand the issues, consider an exchange problem among a pair of unrelated individuals. This problem is called a ‘generalized prisoner's dilemma’ (PD; there are many types of cooperation problem but PDs are probably the easiest to understand). Assume that P is the fitness pay-off (the relevant utility) of non-cooperative individual action. Call R, the reward for cooperative action (give and take). Let T = the temptation (i.e. the pay-off) to cheat (to take but not give), and let S = the pay-off for cooperating when an exchange partner cheats (to give but not take).
Though the prisoner's dilemma is not an a priori given in nature, a large class of interactions will conform to the following pay-off matrix:
T (cheating) > R (reciprocity) > P (solitary action) > S (getting cheated).
Where an exchange is structured in this way, an agent built to optimize self-interest will choose to defect. No matter what other agents do, prisoner dilemma exchange is structured so that a defector maximizes utility by cheating. ‘Defection’ strongly dominates cooperation. It is what economists call the ‘Nash equilibrium’ for this exchange (Nash, 1951; Schelling, 1960). The tragedy of such interactions is that that cooperation is ‘strictly efficient’. If agents were simply to opt for (p. 626) reciprocal exchange, they would be better off than by going it alone (Skyrms, 1996).
Nevertheless cooperation is not lost. The problem is solved when defection is made too costly, too risky, or too difficult to arrange. There are numerous means by which agents can alter pay-off schedules so that R > T. In iterated dilemmas agents with memories can punish defectors by avoiding them in future encounters (Axelrod, 1990). Moreover agents can develop various commitment devices and technologies for securing exchange. By empowering Bob to cut my arm off should I not keep to the arrangement bargain, I've ensured that I will. Yet how do I grant and maintain these powers? The policing of exchange can be costly and unreliable in ever so many ways (Axelrod, 1997; Bulbulia, 2004a; Johnson, 2005).
Moralizing religious understandings can solve cooperative problems cheaply. They do so by altering the expected returns of cooperative action. Consider self-interested agents who believe that cooperation always pays better than defection. Assuming their belief-desire psychology remains in tact, such agents will have motivations to exchange cooperatively. This remains true whatever the actual (i.e. natural) pay-off matrix. Suppose the reward for a cooperative response is perceived to be eternal bliss and the punishment for defection is perceived to be reincarnation as a dung beetle. Such belief is at odds with nature, for it inaccurately depicts probable outcomes. We assume you won't come back as anything. Here the relevant causal processes are supernatural. Yet because cognition errs in precisely this way, it cheaply motivates reciprocity. The belief that what goes around comes around polices behaviour without police.
Perceived supernatural pay-off matrix: R > P > S > T
If you believe in karma, or that infidels are punished, or that religion is its own best reward (for religious motivations need not be extrinsic), then your religiosity when shared can solve prisoner's dilemmas. The key word here is ‘shared’. Selection will equip agents for moralizing religion only where it is possible to reliably recognize the presence of common moralizing religious understanding. Thus, solving the recognition constraint is critical to the evolution of religious cognition. For discerning and signalling religious commitments would have been an essential precondition to establishing moralizing religious dispositions in our lineage. The supernatural reward hypothesis therefore predicts that reliable signalling mechanisms will emerge to solve the recognition constraint. For in the absence of such mechanisms tendencies to moralizing religion would be quickly eliminated.
43.4. Costly signalling theory
Cooperators seek to exchange only with other cooperators. In order to do this, cooperators must identify each other. But how? An arbitrary signal of cooperative intention will not work. For it is always in a defector's interest to mimic such a signal, only to defect when favourable (Zahavi, 1993). In an evolutionary arms race, defectors will always match the arbitrary signals of religious cooperators. Selection therefore needs to equip religious agents with the capacity to produce signals that only religious persons can reliably produce. Note that the recognition constraint is far from trivial. For what could count as such a signal? In answering this question, we can begin to understand how the many of the practical costs of religion have adaptive value (Irons, 2001; Sosis, 2003; Sosis and Alcorta, 2003; Sosis and Ruffle, 2003). For these costs identify cognitive and motivational resources that religious persons possess, but which non-religious persons do not possess.
The study of animal signalling systems has shed much light on how selection outfits organisms with biological devices capable of indicating and measuring cooperative commitment (Zahavi and Zahavi, 1997). Organisms use ‘handicaps’ to authentically signal their properties and intentions (Zahavi, 1975). A reliable signal will be structured so that only a signaller that actually possesses the relevant property or intention will be able to afford producing the signal. Suppose I want to know if you're rich. You can say you're rich—anyone can—but the Ferrari provides real evidence.
Costly signals of intention and worth are rampant in nature. The songs of male bulbuls (Pycnonotidae) are carefully arranged to signal strength and fitness. The musical ornamentation correlates with fitness, so is able to attract mates (p. 627) and frighten rivals. The whistling of skylarks when chased by predatory merlins is also a costly signal of aerobic fitness, one that generally forestalls the chase. Skylarks able to sing as predators stalk are unlikely to be caught, so merlins generally do not bother them. The stotting of springbok gazelle in the presence of lions is another hard-to-fake signal of fitness and speed. The gazelle would like the lions to starve; the lions would prefer gazelle to lie down without a chase. Stotting allows for a cooperative compromise, one that averts constant chasing (the option in this PD of ‘going it alone’; for discussion see Bulbulia, 2004b).
With respect to religious signals, consider how religious emotions function as hard-to-fake signals of religious commitment. Emotions are not private affairs of the heart locked away from view, not even in England. They can be detected in facial expressions, intonations, trembling, perspiration, blushing, violent outbursts and other perceptual traces. Furthermore emotions are difficult to manage. They are processed in the limbic system, outside of the executive control centres of the cerebral cortex (Frank, 1988; Ramachandran and Blakeslee, 1998). It is precisely because emotions are difficult to manipulate that they function as reliable signals. We can lie with words but faking tears is harder. I conjecture that religious emotions function as signals that identify the presence of moralizing motivations and commitments (Bulbulia, 2004a). Weeping before a statue of Shiva suggests commitment to a cosmos in which Shiva reigns. Arm waving and ecstatic grimacing at a Christian rock concert are hard-to-fake markers of Christian commitment. It may well be that such emotional signals modulate to assess the degree to which a believer is committed. For informally, it seems we can distinguish between powerful and weak religious displays. Lots of tears and arm waving suggests strong commitments; moderate halfhearted flinching, perhaps, less. If there is a link between motivations and expressions, it would appear that the magnitude of a devotion signal is inversely proportional to the odds for defection (Bulbulia, 2004a). This conjecture can be tested by assessing how religious persons factor emotional signals into their exchange forecasts and plans. (Below I will come back to empirical results that bear on these conjectures.)
Emotional signals predict future behaviour. But for a display to be effective, there must be an audience. A significant facilitator of religious display is public religious ritual, occasions that both elicit and enable the scrutinizing of hard-to-fake religious emotions. Rituals appear to have many functions—to conventionally communicate intentions, inculcate doctrines, promulgate laws, forge alliances, build hope and solace, mark transitions, calm, excite, or entertain participants—and these functions vary with contexts and individuals. Evolutionary researchers also suggest that ritual functions to provide forums in which religious emotions, and other costly signals of commitment, can be produced and evaluated (Sosis, 2003). Though no theorist, I am aware of suggestions that ritual behaviour emerges from genetic endowment (as does the exquisite dance of male peacocks or the display propensities of bowerbirds). Our love for ritual activity and the disposition to scan and process the information we acquire from them may well be guided by internally structured design. While the jury is out on how far biological endowment shapes and constrains ritual design, even if rituals are construed as fully technological artefacts (invented, and transmitted in the flow of cultural ideas), their presence and improvement would have greatly enhanced the evolution of tendencies to form supernatural convictions and commitments. For I repeat, such tendencies will not be selected in the absence of a solution to the recognition constraint. (For non-adaptationist accounts that posit strong psychological constraints on the design of religious rituals see: Whitehouse, 1992, 2000, 2004; McCauley and Lawson, 2002.)
A second means by which the religious can reliably signal supernatural commitment is by undertaking practical costs. Turning up at a ritual or hazarding a religious practice minimally involves an opportunity cost. And beyond such costs, rituals demand sacrifices of attention and resource, and some religious practices are exhausting, dangerous, painful, and supremely boring to the undevout. Again, were religious practice not adaptive, it is hard to understand why selection would tolerate religious costs. Adaptationists urge that such practices are generally adaptive. Participation, under the scrutinizing eye of an audience, serves to signal (p. 628) religious, and so cooperative, commitment. Adaptationists predict that ritual trials will be fashioned so that only those committed to moralizing super-nature will be willing to endure them frequently. The practical costs of ritual participation will render expected utilities explicit in ways directly related to supernatural belief. This condition is especially important where groups compete against each other for resources, and so where defection incentives are especially high.
Consider Bob a religious believer in a moralizing supernatural order. Bob faces a decision whether to partake of a strenuous or tedious ritual. Bob must weigh the costs of ritual participation times their frequency against the conditional probability the gods will bring about an outcome whose benefit outweighs the expected total practical cost of participation. If Bob thinks that participation brings (intrinsic or extrinsic) supernatural reward then:
The cost of religious practice × frequency < conditional probability of value through supernatural reward.
Now consider Roz the unbeliever. Roz wants to fake religious commitment and defect. Roz doesn't think the gods will give her anything for her time, energy, and resource outlay. She anticipates only further rituals everlastingly with no supernatural commission. Beyond ongoing practical costs, Roz must factor the conditional risk of getting caught, and, in the relevant setting, punished. Hence Roz's expected utility from costly religious practice may well exceed her perceived utility in cheating the devout. For she has no supernatural surplus weighting the religious option. Hence for Roz:
The conditional probability of value from cheating the devout < costs of ritual participation × frequency
So the practical costs of religion are neither arbitrary nor maladaptive. While verbal expressions of religious conviction are cheap; religious practice assesses whether the religious are willing to back locutions with deeds (for further discussion see Sosis and Alcorta, 2003; Bulbulia, 2004a).
Notice that where the pay-offs from defection are high, religious imposters can nevertheless invade religious communities. The expectation, therefore, is for the costs surrounding religious practice to modulate to assess the strength of religious devotion. Adaptationists hypothesize that ritual costs and frequency will correlate with defection incentives. This leads to a surprising prediction. When facing crisis, religious individuals will tend to devote more time and material resources to costly ritual participation than they normally do. At first glance such outlays are puzzling, for with adversity we would ordinarily predict a tendency to conserve. Daniel Chen observed precisely this correlation in his analysis of the Indonesian financial crises of the late 1990s (Johnson, 2003; D. Chen, unpublished data). As money and resources dried up, Muslim Indonesian families devoted a greater share of their dwindling reserves to religious observance. Though apparently mal-adaptive, Chen observes that religious institutions provide social insurance. They minimize the risks by collectively supporting the most needy. Costly observance in insurance societies makes the join—defect—leave strategy more expensive.
There is further experimental support for the ritual adaptationist hypothesis. Bill Irons and his colleagues Lee Cronk and Shannon Steadman found that among the people of Utila (a Bay Island of Honduras), men prefer to marry women who frequently attend church. However, this preference is not reversed: women are less interested in the religious piety of husbands. Why the difference? Irons notes that the men of Utila spend much time away from home in maritime work. Given frequent separation, the risk of infidelity is especially high. The risks of false parenting fall disproportionately on men. (Women know when they are biological mothers. The signs are unmistakable). Irons explains the data by conjecturing that religious signalling is favoured among Utilan women as a signal of sexual virtue (Irons, 2001).
In a comparative study of 200 religious and secular communes in the nineteenth century, Rich Sosis determined that the religious communities were far more likely to outlast their non-religious counterparts—four times as likely in any given year (Sosis, 2000). In a subsequent study, Sosis and Bressler showed that religious communes imposed over twice as many costly requirements on their members than did secular communes, and further that the number of costly requirements was positively correlated with (p. 629) group lifespan. Interestingly a similar effect did not hold for secular communes, where costly requirements did not correlate with secular commune lifespan (Sosis and Bressler, 2003). Sosis and Ruffle have further shown that religious ritual influences cooperation in contemporary religious kibbutzim. Using common-pool resource games, the authors found that religious males were significantly more altruistic in their play than were religious and secular females, and secular males. The authors discovered no sex differences in cooperation among the secular kibbutz members, eliminating the possibility that there were broader differences in the ways males and females play the game. Noting that only orthodox men are expected to participate in communal prayer three times a day, the authors conclude that costly ritual participation (rather than any inherent differences between the sexes) accounts for the discrepancy (Sosis and Ruffle, 2003).
43.5. Religiosity as a cognitive adaptation
Religious behaviour may be adaptive; even so, it does not follow that dedicated mental architecture produces it. It may well be that religious thought is a commonly discovered invention, shaped for specific tasks, and passed down as future generations use their culturally inherited religion as a platform for further improvements. We do not require dedicated architecture for reading, fire cognition, and water extraction. Yet these skills are probably more widespread in the hominid lineage than religion, and arguably they are more adaptive. Moreover, when we consider the substance of religious life, we see the trail of the cultural serpent everywhere; children are not born knowing what to do with chalices and menorahs, how to meditate, how Ganesh got his elephant head. Indeed the conventionality of religion—its cultural dependence—facilitates costly theological signalling by insulating religious communities from religious imposters who don't know all the dogmas and rules. The more elaborate and esoteric are theological and ritual conventions, the harder are they to counterfeit. Local theologies and folklores serve as encryption technologies. Thus religiosity's functionality benefits from cultural variation. Given that dependency, why should we think that religious cognition emerges from a cognitive design selected for it?
Undeniably substantial aspects of an individual's religiosity come by way of contingent social involvements. We cannot fully understand an individual's religiosity without understanding how these cultural settings bear on psychological development. The nature/nurture dualism has no place in the cognitive study of religion (indeed the dualism misleads nearly everywhere in the cognitive study of mind). Furthermore, whether religiosity is a psychological adaptation is itself of little theoretical interest (Atran, 2004). For the goal is to produce a better theory of religious minds, not to speculate about ancestral history. The adaptationist stance is interesting because, if correct, then reverse-engineering techniques likely will prove fruitful.
These provisos aside, four considerations suggest the presence of dedicated religiosity architecture that serves the adaptive functions just described.
First, religious cognition must be encapsulated, and it is hard and risky to rely on teaching for the monitoring of information flow. Notice that though religiosity relies on error, we cannot become globally impoverished at assessing the contents of distal reality. Cognitive adaptationism predicts that god belief will not inform inferences to explanation in all domains. Nor will supernatural error impinge on practical task solving and priority assessment. For unconstrained religious error is an evolutionary death sentence. An organism that allows error to corrupt all empirical inferences about the world will be quickly absorbed by its hostile elements. So over time, selection will eliminate unconstrained error-prone designs. Functional religiosity must be insulated from unrelated problem domains. Thus religiosity demands specific meta-representational structuring to curtail inferential haemorrhaging. It requires a scope-syntactic architecture of the kind that enables us to consider fictional worlds, engage in counter-factual reasoning and planning, and remember the past without confusing these imagined worlds (p. 630) with reality (for an excellent general discussion of the architecture required for meta-representational thought see Cosmides and Tooby, 2000).
The encapsulation constraint leads to specific, testable predictions. For example, functional religious representations need to be tagged with high degrees of subjective confidence—for we have seen how such confidence when shared solves generalized prisoner's dilemmas. Yet such representations must not factor into decisions that mediate a religious agent's practical involvements with the world, except to meet signalling demands. The theory therefore predicts that we will believe with our hearts that the gods will provide, while striving with our hearts to provide for ourselves. You may believe in karma—that moral causality guarantees just outcomes in this life or the next. Yet unless punishment is also viewed as a karmic duty, an unbounded architecture might well permit the inference: do not punish enemies (… for they shall get theirs anyway, as the wheel turns). Or it may warrant complacency about death among the well-behaved. However, to give up caring about punishment or life is a bad idea for survival.
While I am unaware of studies that assess the scope, magnitude, and variation to which religiosity is encapsulated, there is evidence that religiosity is not too reproductively damaging. Religious persons are generally not less successful, healthy and fertile than their secular counterparts; indeed religious persons in North America frequently score higher on measures of physical and mental health (Koenig et al., 2001; Pargament, 2002; Bulbulia, 2006). We know that erroneous religious understandings do not impair individuals as they would if these understandings were inferentially robust. However, we need better meta-representational analyses to show how (Bulbulia, 2006; but see Pyysiainen, 2003).
Second, religious representations need to be precisely integrated to other cognitive systems. Though inferentially encapsulated, cognitive adaptationism predicts that religiosity will not be inferentially inert. For religious representations are functional precisely because they strategically mediate an agent's engagement with her world in a few domains. Religiosity cannot be insulated to the degree that it is thought free-spinning in a void. Its rubber must meet the road at strategic moments, for example, by the production of costly signs to mark and verify religious motivations. Again, at present we do not have a clear picture of the scope-grammars that structure religious cognition. Informally, we know that religious concepts are linked to emotional and motivational processing. Unlike esoteric false beliefs (in the reality of possible worlds or that colours are intrinsic properties) agents form motivational commitments to their gods, and these motivations powerfully affect behaviour. We know further that religious beliefs strongly motivate social norms. Understanding this moral functionality enables us to hazard more precise predictions about the religiosity's design.
For example, the theory predicts that religious representations will not merely consist of minimally violating anthropomorphic concepts. Rather, supernatural concepts will be arrayed to police behaviour, either through their perceived intrinsic or extrinsic properties. Indeed, there is much anecdotal evidence that religious concepts are moralizing (Malinowski, 1935). And there is some recent quantitative evidence as well. In a study of 186 cultures, Dominic Johnson (2005) found that the salience of ‘high gods’ correlates positively with enhanced reciprocity. While Johnson notes that the category of high gods and punishing gods do not perfectly overlap (the relevant ethnographic data contain gaps) nevertheless the study yields an intriguing result. In societies where gods are imagined as powerful there appears to be less selfish behaviour. However, if religious cognition is a cognitive adaptation then Johnson needs to explain how the cultural variant ‘weak gods’ becomes popular in some populations. (“non-moralizing” gods may be artefacts of informal description; one virtue of cognitive adaptationism is that it enables more precise descriptive practices: see Bulbulia, 2004b).
Cognitive adaptationism further predicts that evidence of supernatural commitment will factor into exchange decisions. We have already reviewed Sosis's studies showing that ritual participation correlates positively with altruism and group longevity. In a recent study, Bulbulia and (p. 631) Mahoney found that enhanced altruism strategies were more common in cooperation games among Christian New Zealanders (n = 61) when compared to New Zealand ‘citizen’ controls (n = 55). Out of a possible $5 ‘gift’ (enough to buy lunch) we found mean gifting by Christians was $2.84 (SD = $1.56) whereas in the control group it was only $0.73 (SD = $1.34). There is, however, a dark side to the religious altruism of Christian New Zealanders. The games were structured so that the altruistic gifting was always to an anonymous group member observed to incur a financial cost by punishing an out-group member, merely for having an out-group status. We know of no explicit Christian teaching that reinforces rewarding fellow religionists for such punishing behaviours, so the behaviour is probably not learned from Christian sources (at least not directly). Such behaviours appear less anomalous if core structural features of religious morality are features of biological endowment, for such punishing is another instance (and signal) of reciprocity: a giving to one who has given to the group. We believe more empirical work is needed before we can be confident that religious morality is special; the causally important differences between the dominant exchange strategies differentiating individuals in these groups may lie elsewhere. (For discussion of non-religious factors that influence strategic norms see Henrich et al., 2004.) Nevertheless the data so far strongly indicate that religiosity is especially morally motivating.
Thirdly, cognitive adaptationism conjectures that religious commitment is supported through a cognitive architecture that biases and distorts information flow to bolster subjective confidence in the relevant group-organizing religious concepts. Remember, nothing in our experience should make the prospect of supernatural existence a live question, any more than our experience should lead us to suppose unicorns exist or that Mickey Mouse is real. To repeat: the poverty of supernatural stimulus is nearly complete. Getting agents to believe in super-nature presents a significant design problem. Even a very careful structuring of a cultural environment by religious elites (for example, see Roes and Raymond, 2003) could not easily do the trick, for the requisite stimuli—projecting an eight-armed goddess, or Hermes—are substantial.
A prediction of the conjecture that biasing and distortion machines underlie religious commitment is that people genuinely believe in the supernatural. [This prediction may sound odd, yet Dennett (2006) and Palmer and Steadman (2004) challenge it; for these theorists, religionists only ‘believe in belief’.] It seems to me, however, a good bet that incredulity is fairly rare, perhaps itself only a cognitive prospect in very recent cultures—and even in these, still fairly rare. How to test this prediction? We hardly need to. Costly emotional signals and dispositions to participate in religious trials assess these beliefs for us, naturally; for we have seen how such signals are hard to fake absent the relevant supernatural commitments. Of course, to convince the sceptics we require quantitative measures. These remain on the horizon of inquiry.
Another prediction of the theory is that religionists will be disinclined to accept functional explanations for their commitments. For it would appear that the functional utility of religious representations is optimized through self-deception about the nature of that functionality. There are good reasons for thinking that the architecture supporting religious commitment yields this obscurantism. It is not merely that religious agents have no need to be aware of the functionalist nature of their commitments (as language speakers have no need for explicit access to universal grammar), the functionality of the system may be threatened by self-conscious knowledge of its design; such knowledge can lead to a failure of confidence in religious morality. If we can explain moralizing supernatural commitment without appeal to moralizing forces, then altruists require non-religious reasons to motivate cooperative behaviour. If the gods are perceived to motivate exchange then disenchantment with the gods, all things being equal, will threaten those motivations.
Moreover, because moralistic condemnation of ulterior motives is a hard-to-fake signal of genuine supernatural (and so moral) commitment, cognitive adaptationists predict that such condemnation will form part of religiosity's functional design. In particular, we predict that (p. 632) whatever its merits, cognitive-adaptationist theory will have few adherents among fundamentalistically committed religionists. We furthermore observe that it is within their power to refute cognitive adaptationism by embracing and praising it (for discussion see Bulbulia, 2006).
Of course, exchange partners may possess a raft of motivations apart from religion to moderate selfishness, including a powerful distaste for cheating, natural affection, sensitivity to reputation, and worry over natural policing or retribution. We claim only that religiosity supplements these ordinary moral motivations, but to do so, we must believe in non-natural rewards.
Fourth, religious commitments appear to come on-line early in childhood development, before robust and explicit religious education directs these commitments in cultural-specific directions. This emergence is consistent with religiosity's internal structuring. Barrett and colleagues have shown that children reason about God as different from other persons (Barrett and Richert, 2003). After the age of 5 years children begin to understand that human agents do not know everything, but children do not give up that assumption about God. God belief is resilient to theory-of-mind correction. Conversely before the age of 5 years, children credit all agents with omniscience. The authors suggest that children are better prepared to conceptualize god properties than they are human properties. Children appear born to believe. [Knight et al. (2004) have replicated this result in a cross-cultural study.]
The developmental psychologist Deb Keleman has taken this line further by showing that American children before the age of 5 years are biased to reason about the natural world in terms of intention and purpose; they are ‘intuitive theists’ who prefer explanations of designing intention everywhere (for a review see Kelemen, 2004). Children see clouds as “for raining” and mountains as “pointy so that animals won't sit on them and smash them” or “so that animals could scratch on them when they got itchy” (Kelemen, 1999b). These biases endure into late childhood, a pattern also observed in British children (Kelemen, 2003). Children also prefer to explain items in the world as being caused by supernatural agents, a bias that endures into early adolescence, even in children from non-religious households (so they're not straightforwardly getting this bias from their parents; Evans, 2000, 2001). Keleman and DiYanni moreover have recently shown that rampant teleology and preference for supernatural explanation is connected. Not only do children prefer to find purpose in nature, they prefer explanations in which agents are responsible for providing these purposes (DiYanni and Kelemen, 2005; Kelemen and DiYanni, 2005).
It may be that this intuitive theism has moral dimensions. Bering (2004) has shown that after being primed to believe in a supernatural agent (an invisible ‘Princess Alice’), children display inhibitive responses when tempted to cheat in reward games. Children strongly police their behaviour when natural anomalies (for example, a picture falling off the wall) suggest the presence of a supernatural being. “Even the youngest children behave as if they've been caught red-handed.”
In sum, while there remain significant gaps in our understanding of how children and adolescents come to commit to gods, the data so far are consistent with the view that all children possess a religiosity faculty with entrenched features. The data suggest that this faculty emerges naturally in the course of childhood development, without the need for a robustly scaffolded religious education (for discussion see Bulbulia, 2005).
Time to briefly summarize the state of play. First, I used game theory and the prisoner's dilemma to illustrate how the epistemic costs of religious belief may facilitate cooperation where there are incentives for individuals to defect. I observed that because religious cooperators perceive the world as strongly discouraging cooperation, religious commitment when shared and recognized helps to build stronger coalitions. In short, because the gods police social contracts, the epistemic costs of religious consciousness have adaptive value. Second, I used costly signalling theory to show how religious cooperators satisfy the recognition constraint. In order for religious (p. 633) cooperation to work, religious agents need to find each other. Because defectors will always have incentives to mimic religious signals, such signals must incur costs that only genuine religionists will bear. Emotional displays provide hard-to-fake signals of authentic commitment to the gods, because emotions are difficult to control consciously. Moreover, rituals provide venues for the committed to advertise conviction. The various practical costs of religious ritual are, for the committed, understood as investments: for by their lights, ritual brings supernatural rewards. Cost is here perceived to be an investment. Defectors, on the other hand, will perceive no such investment incentive. Therefore they generally will not be willing to expend as much as those genuinely committed to the gods. Ritual costs thus act as a social filter. Generalizing, because the practical costs of religion signal religious commitment, they have adaptive value. I then noted that religiosity may be adaptive without relying on a dedicated cognitive architecture, but examined four separate data streams converging to the view that religiosity is at least partially controlled by internal cognitive resources we do not explicitly learn.
My focus in this chapter has been on internal psychological architecture, one I believe underlies our capacity for religion. I think that one of the most important horizons of future inquiry lies in developing a more precise and detailed account of that architecture. Yet I also believe that progress will be impaired unless we integrate this evolutionary psychological approach with the growing psychological literature on religious development in children, with archaeological studies, experimental economics, and with the vast and expanding ethnographies that record variation in religious cultural groups. For fruitful integration to occur, we need to begin thinking about our religious traditions not as snake trails of mistakes and costly maladaptations—ʼbarking madʼ as one researcher puts it—but as practices for human flourishing, sculpted by psychological adaptations, and every bit as intricate and functional as a forager's spear or an Eskimo's canoe.
Atran, S. (2002) In Gods We Trust: The Evolutionary Landscape of Religion. Oxford University Press, New York.Find this resource:
Atran, S. (2004) Religion's evolutionary landscape: counterintuition, commitment, compassion, communion. Behavioral and Brain Sciences, 27: 713–730.Find this resource:
Axelrod, R. (1990) The Evolution of Co-operation. Penguin, London.Find this resource:
Axelrod, R. (1997) The Complexity of Cooperation. Princeton University Press, Princeton, NJ.Find this resource:
Barrett, J. L. (2000) Exploring the natural foundations of religion. Trends in Cognitive Sciences, 4(1): 29–34.Find this resource:
Barrett, J. L. (2004) Why Would Anyone Believe in God? AltaMira Press, Lanham, MD.Find this resource:
Barrett, J. L. and Keil, F. (1996) Conceptualizing a nonnatural entity. Cognitive Psychology, 31: 219–247.Find this resource:
Barrett, J. L. and Keil, F. C. (1998) Cognitive constraints on Hindu concepts of the divine. Journal for the Scientific Study of Religion, 37: 608–619.Find this resource:
Barrett, J. L. and Richert, R. A. (2003) Anthropomorphism or preparedness? Exploring children's God concepts. Review of Religious Research, 44: 300–312.Find this resource:
Berger, P. (1990) The Sacred Canopy: Elements of a Sociological Theory of Religion. New York, Doubleday.Find this resource:
Berger, P. and Luckmann, T. (1972) The Social Construction of Reality: A Treatise in the Sociology of Knowledge. Doubleday, New York.Find this resource:
Bering, J. M. (2004) The evolutionary history of an illusion: religious causal beliefs in children and adults. In B. Ellis and D. F. Bjorklund (eds) Origins of the Social Mind: Evolutionary Psychology and Child Development. Guilford Press, New York.Find this resource:
Blackmore, S. (2000) The Meme Machine. Oxford University Press, New York.Find this resource:
Boyd, R. and Richerson, P.J. (1985) Culture and the Evolutionary Process. University of Chicago Press, Chicago.Find this resource:
Boyer, P. (1992) Explaining religious ideas: elements of a cognitive approach. Numen, XXXIX: 27–57.Find this resource:
Boyer, P. (1994) The Naturalness of Religious Ideas: A Cognitive Theory of Religion. University of California Press, Berkeley.Find this resource:
Boyer, P. (2001) Religion Explained: The Evolutionary Origins of Religious Thought. Basic Books, New York.Find this resource:
Boyer, P. (2003) Religious thought and behaviour as by-products of brain function. Trends in Cognitive Sciences, 7: 119–124.Find this resource:
Boyer, P. and Ramble, C. (2001) Cognitive templates for religious concepts: cross-cultural evidence for recall of counter-intuitive representations. Cognitive Science, 25: 535–564.Find this resource:
Bulbulia, J. (2004a) Religious costs as adaptations that signal altruistic intention. Evolution and Cognition, 10: 19–38.Find this resource:
Bulbulia, J. (2004b) The cognitive and evolutionary psychology of religion. Biology and Philosophy, 18: 655–686.Find this resource:
Bulbulia, J. (2005) Are there any religions? Method and Theory in the Study of Religion, 17: 71–100.Find this resource:
Bulbulia, J. (2006) A scope syntax for religious representations. In Preparation.Find this resource:
Bulbulia, J. A. (2006) Nature's Medicine: empirical constraint and the evolution of religious healing. (p. 634) In P. MacNamara (ed.) Where Man and God Meet: The New Sciences of Religion and Brain. Greenwood Publishers, Westwood, CT.Find this resource:
Chen, D. Economic Distress and Religious Intensity: Evidence from Islamic Resurgence During the Indonesian Financial Crisis. Submitted 2005. American Economic Review.Find this resource:
Cosmides, L. and Tooby, J. (2000) Consider the source: the evolution of adaptations for decoupling and metarepresentation. In D. Sperber (ed.) Metarepresentation. Oxford University Press, New York, pp. 53–116.Find this resource:
Dawkins, R. (1976) The Selfish Gene. Oxford University Press, New York.Find this resource:
Dennett, D. C. (2006) Breaking the Spell: Religion as a Natural Phenomenon. Viking, New York.Find this resource:
DiYanni, C. and Kelemen, D. (2005) Time to get a new mountain? The role of function in children's conception of natural kinds. Cognition, 97: 327–335.Find this resource:
Dunbar, R. I. (1998) The social brain hypothesis. Evolutionary Anthropology, 6: 178–190.Find this resource:
Dunbar, R. I. (2005) The Human Story: A New History of Mankind's Evolution. Faber & Faber, London.Find this resource:
Evans, E. M. (2000) The emergence of beliefs about the origin of species in school-age children. Merrill Palmer Quarterly, 46: 221–254.Find this resource:
Evans, E. M. (2001) Cognitive and contextual factors in the emergence of diverse belief systems: Creation versus evolution. Cognitive Psychology, 42: 217–266.Find this resource:
Frank, R. (1988) Passions Within Reason: The Strategic Role of the Emotions. Norton, New York.Find this resource:
Gelman, S. and Kremer, K. (1991) Understanding natural cause: children's explanations of how objects and their properties originate. Child Development, 62: 396–414.Find this resource:
Gelman, S. A. and Markman, E. (1987) Young children's inductions from natural kinds: the role of categories and appearances. Child Development, 58: 1532–1540.Find this resource:
Godfrey-Smith, P. (1996) Complexity and the Function of Mind in Nature. Cambridge, Cambridge University Press.Find this resource:
Guthrie, S. (1993) Faces in the Clouds: A New Theory of Religion. New York, Oxford University Press.Find this resource:
Hamer, D. (2004) The God Gene: How Faith is Hardwired into our Genes. New York, Doubleday.Find this resource:
Henrich, J., Boyd, R. et al. (eds) (2004) Foundations of Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen Small-Scall Societies. New York, Oxford University Press.Find this resource:
Irons, W. (1996) Morality, religion, and evolution. In W. M. Richardson (ed.) Religion and Science: History, Method, and Dialogue, pp. 375–399. Routledge, New York.Find this resource:
Irons, W. (2001) Religion as hard-to-fake sign of commitment. In R. Nesse (ed.) Evolution and the Capacity for Commitment. Russell Sage Foundation, New York.Find this resource:
Johnson, C. (2003) During economic turmoil, religion is ‘insurance’. Science and Theology News, 4.Find this resource:
Johnson, D. D. P. (2005) God's punishment and public goods: a test of the supernatural punishment hypothesis in 186 world cultures. Human Nature, 16: 410–446.Find this resource:
Kelemen, D. (1999b) Why are rocks pointy? Children's preference for teleological explanations of the natural world. Developmental Psychology, 35: 1440–1453.Find this resource:
Kelemen, D. (2003) British and American children's preference for teleo-functional explanations of the natural world. Cognition, 88: 201–222.Find this resource:
Kelemen, D. (2004) Are children intuitive theists?: Reasoning about purpose and design in nature. Psychological Science, 15: 295–230.Find this resource:
Kelemen, D. and DiYanni, C. (2005) Intuitions about origins: purpose and intelligent design in children's reason about nature. Journal of Cognition and Development, 6: 3–31.Find this resource:
Knight, N., Sousa, P. et al. (2004) Children's attributions of beliefs to humans and God: cross-cultural evidence. Cognitive Science, 28: 117.Find this resource:
Koenig, H. G., McCullough, M. E. et al. (2001) Handbook of Religion and Health. Oxford University Press, New York.Find this resource:
Malinowski, B. (1935) The Foundations of Faith and Morals: An Anthropological Analysis of Primitive Beliefs and Conduct with Special Reference to the Fundamental Problem of Religion and Ethics. Oxford University Press, London.Find this resource:
McCauley, R. N. and Lawson, E. T. (2002) Bringing Ritual to Mind. Cambridge University Press, New York.Find this resource:
McClenon, J. (2002) Wondrous Healing: Shamanism, Human Evolution, and the Origin of Religion. Northern Illinois University Press, DeKalb, IL.Find this resource:
Nash, J. (1951) Noncooperative Games. Annals of Mathematics, 54: 289–295.Find this resource:
Palmer, C. T. and Steadman, L. B. (2004) With or without belief: a new approach to the definition and explanation of religion. Evolution and Cognition, 10: 138–147.Find this resource:
Pargament, K. I. (2002) The bitter and the sweet: an evaluation of the costs and benefits of religiousness. Psychological Inquiry, 13: 168–189.Find this resource:
Preus, J. S. (1987) Explaining Religion: Criticism and Theory from Bodin to Freud. Yale University Press, New Haven.Find this resource:
Pyysiainen, I. (2001) How Religion Works: Towards a New Cognitive Science of Religion. Brill, Leiden.Find this resource:
Pyysiainen, I. (2003) True fiction: philosophy and psychology of religious belief. Philosophical Psychology, 16: 109–125.Find this resource:
Ramachandran, V. S. and Blakeslee, S. (1998) Phantoms in the Brain: Probing the Mysteries of the Human Mind. Quill William Morrow, New York.Find this resource:
Reader, I. and Tanabe, G. J. J. (1998) Practically Religious: Worldly Benefits and the Common Religion of Japan. University of Hawaii Press, Honolulu.Find this resource:
Richerson, P. and Boyd, R. (1989) The role of evolved predispositions in cultural evolution. Or, human sociobiology meet Pascal's Wager. Ethology and Sociobiology, 10: 195–219.Find this resource:
Richerson, R. and Boyd, R. (2005) Not by genes alone: how culture transformed human evolution University of Chicago Press, Chicago.Find this resource:
Roes, F. L. and Raymond, M. (2003) Belief in moralizing gods. Evolution and Human Behavior, 24: 126–135.Find this resource:
Schelling, T. (1960) The Strategy of Conflict. New York, Oxford University Press.Find this resource:
(p. 635) Shapiro, A. and Shapiro, E. (1997) The placebo: is it much ado about nothing? In A. Harrington (ed.) The Placebo Effect, pp. 12–36. Harvard University Press, Cambridge, MA.Find this resource:
Skyrms, B. (1996) Evolution of the Social Contract. Cambridge University Press, New York.Find this resource:
Sosis, R. (2000) Religion and intragroup cooperation: preliminary results of a comparative analysis of Utopian communities. Cross-Cultural Research, 34: 77–88.Find this resource:
Sosis, R. (2003) Why aren't we all Hutterites? Human Nature, 14: 91–127.Find this resource:
Sosis, R. (2004) The adaptive value of religious ritual. American Scientist, 92: 166–172.Find this resource:
Sosis, R. and Alcorta, C. (2003) Signalling, solidarity, and the sacred: the evolution of religious behavior. Evolutionary Anthropology, 12: 264–274.Find this resource:
Sosis, R. and Bressler, E. (2003) Co-operation and commune longevity: a test of the costly signaling theory of religion. Cross-Cultural Research, 37: 11–39.Find this resource:
Sosis, R. and Ruffle, B. (2003) Religious ritual and cooperation: testing for a relationship on Israeli religious and secular kibbutzim. Current Anthropology, 44: 713–722.Find this resource:
Spelke, E., Phillips, A. et al. (1995) Infants knowledge of object motion and human action. In D. Sperber, D. Premack and A. Premack (eds) Causal Cognition. Clarendon Press, Oxford.Find this resource:
Sterelny, K. (2003) Thought in a Hostile World: The Evolution of Human Cognition. Blackwell, Oxford.Find this resource:
Sterelny, K. (2006) The evolution and evolvability of culture. Mind and Language, 21: 137–165.Find this resource:
Whitehouse, H. (1992) Memorable religions. Man, 27: 777–797.Find this resource:
Whitehouse, H. (2000) Arguments and Icons. Oxford, Oxford Universit y Press.Find this resource:
Whitehouse, H. (2004) Modes of Religiosity: A Cognitive Theory of Religious Transmission. AltaMira Press, Lanham, MD.Find this resource:
Wilson, D. S. (2002) Darwin's Cathedral: Evolution, Religion, and the Nature of Society. University of Chicago Press, Chicago.Find this resource:
Zahavi, A. (1975) Mate selection: a selection for a handicap. Journal of theoretical Biology, 67: 603–605.Find this resource:
Zahavi, A. (1993) The fallacy of conventional signalling. Philosophical Transactions of the Royal Society of London, B, 338: 227–230.Find this resource:
Zahavi, A. and Zahavi, A. (1997) The Handicap Principle: A Missing Piece of Darwin's Puzzle. Oxford University Press, New York. (p. 636) Find this resource: