Original ArticleBeliefs about God, the afterlife and morality support the role of supernatural policing in human cooperation☆
Introduction
Group living in modern humans is characterized by a unique level of cooperation and exchange among large numbers of unrelated individuals. We rely on others for information, aid and resources, and we are willing to share information, aid and resources with others whom we may never see again. Despite advantages as a survival strategy, this system of trust and reciprocity is vulnerable to exploitation by free riders or cheats who reap the benefits of the group without contributing their share to the common pool. Nevertheless, humans appear to have overcome, or at least mitigated, the free-rider problem and are able to maintain cooperative social networks for indefinite periods.
Recently, there has been increasing interest in the role played by religion in the origin and evolution of human cooperation and prosociality (Alcorta and Sosis, 2005, Dunbar, 2009, Johnson, 2005, Johnson and Bering, 2006, Johnson and Krüger, 2004, Monsma, 2007, Norenzayan and Shariff, 2008, Pyysiäinen and Hauser, 2010, Richerson and Boyd, 1998, Roes and Raymond, 2003, Rossano, 2007, Ruffle and Sosis, 2007, Snarey, 1996, Sosis and Alcorta, 2003, Wilson, 2002). Whilst some argue that religion is simply a cultural parasite (Blackmore, 1999, Dawkins, 1976, Dawkins, 2006, Dennett, 2006) or evolutionary by-product of other adaptive processes (Atran, 2002, Barrett, 2000, Boyer, 2001, Guthrie, 1993, Pyysiäinen and Hauser, 2010), others see it as providing fitness advantages by guarding against free-riding and facilitating group cohesion, cooperation and trust (Alcorta and Sosis, 2005, Dunbar, 2009, Richerson and Boyd, 1998, Sosis and Alcorta, 2003, Sosis et al., 2007, Wilson, 2002).
The claim that religion increases prosocial behaviour is supported from a number of sources. Members of religious congregations and regular churchgoers are more likely to report giving time and money to charities than non-members or those who attend church less regularly (Monsma, 2007). Experimental work indicates that religious individuals are also both more trusting and more trusted in cooperative economic games (Tan & Vogel, 2008). Perhaps the most convincing evidence, however, comes from studies of religious organizations themselves. Controlling for other relevant predictors, studies found that males in religious Kibbutzim are more likely to cooperate in economic games than males from secular Kibbutzim, with the highest rates of cooperation among those males who most regularly engage in collective rituals (Ruffle and Sosis, 2007, Sosis and Bressler, 2003). A historical survey of 19th century communes showed that religious communes were four times as likely to survive each year than secular communes (Sosis & Bressler, 2003) and that those religious (but not secular) communes with stricter taboos and prohibitions lasted longest.
It remains unclear exactly why religion should have this effect. Dunbar (2009) argues that endorphin release during intensely arousing rituals, such as communal singing or trance dancing, may directly enhance bonding within small groups. Irons, 1996a, Irons, 1996b has used signalling theory from biology to argue that restrictive taboos or costly rituals (that are risky, unpleasant or demanding of time and resources) promote trust and cooperation more indirectly by providing reliable signals of commitment to the group. However, the finding that religious groups are more prosocial and robust than their statistically controlled secular counterparts suggests that there is more to religious cooperation than participation and proscription.
The nature of religious belief itself is also thought to influence levels of cooperation (Bering, 2006, Johnson and Krüger, 2004, Norenzayan and Shariff, 2008, Roes and Raymond, 2003, Snarey, 1996, Stark, 2001). Stark (2001) has shown that strength of belief in God is a better predictor of prosocial attitudes than church attendance. This raises the question of what it is about the beliefs religious individuals hold that could promote prosociality. One mechanism that has been put forward is that belief in the existence of a supernatural agent or agents can increase prosocial behaviour merely by creating the perception of being watched. Reputational concerns are known to be crucial for motivating and maintaining cooperation towards public goods in human groups (Lotem et al., 1999, Milinski et al., 2002, Nowak and Sigmund, 1998a, Nowak and Sigmund, 1998b). ‘Supernatural monitoring’ is hypothesized to activate cognitive architecture associated with reputation management and so promote prosocial behaviour (Johnson and Bering, 2006, Rossano, 2007). Consistent with this proposal, even subtle, subliminal primes of monitoring, such as an image of ‘watching eyes’ (Bateson et al., 2006, Haley and Fessler, 2005) or three dots oriented to reflect a face (Rigdon, Ishii, Watabe, & Kitayama, 2009), can increase some prosocial behaviours (cf. Fehr & Schneider, 2009), particularly towards in-group members (Mifune, Hashimoto, & Yamagishi, 2010). Regarding supernatural primes, Bering, McLeod, and Shackelford (2005) found that subjects told that a ghost had been seen in the lab were significantly less likely to cheat on a competitive task. Similarly, Shariff and Norenzayan (2007) have shown that implicitly priming God concepts is at least as effective at increasing generosity in an anonymous dictator game as priming secular moral institutions. If such subtle monitoring cues can affect prosocial tendencies, it seems plausible that strongly held belief in an ever-present God or spirits that are always watching could have a similar effect.
Belief in supernatural agents may also promote prosociality by providing a threat of punishment for non-cooperation (Johnson & Krüger, 2004). The viability of cooperation within human groups is thought to rely on the potential to punish free riders or reward prosocial behaviour (Boyd, Gintis, Bowles, & Richerson, 2003). However, such enforcement generally incurs a cost that must be borne by those in the group. Enforcement, then, is itself a public good subject to exploitation by free riders. Economic games under anonymous laboratory conditions have revealed that humans are willing to pay a cost to punish free riders (Fehr and Gachter, 2002, Gintis et al., 2003) or reward cooperators (Rand, Dreber, Ellingsen, Fudenberg, & Nowak, 2009), but it is not clear to what extent this generalizes to real-world social interactions. Such a strategy remains vulnerable to exploitation by ‘second-order free riders’ who avoid the cost of punishment, unless those who shirk punishing duties are themselves punished (and those who shirk the punishing of non-punishers are punished, and so on) or there is some external policing mechanism (Henrich & Boyd, 2001). The supernatural punishment hypothesis (Johnson & Krüger, 2004) holds that the threat of supernatural punishment (in this life or the afterlife) arising from belief in morally concerned supernatural agents can help enforce cooperative norms by exporting the cost of enforcement to ostensibly infallible supernatural forces beyond the group. Belief in a punitive supernatural agent can, in principle, exert this effect without requiring that the imagined agent actually punishes free riders — it is enough that group members perceive such a threat.
By deterring free riders and reducing enforcement costs, supernatural policing may have played an important role in human evolution, increasing group stability and cooperation towards public goods (Johnson and Krüger, 2004, Norenzayan and Shariff, 2008). This hypothesis finds some support from cross-cultural data. Johnson (2005) has shown that the presence of moralizing ‘high gods’ — defined as active in human affairs and specifically supportive of human morality (Swanson, 1960) — is associated with some indices of societal cooperation such as taxation, policing and measures of norm compliance, although only two of these relationships remain significant after controlling for regional effects and influence of world religion. To the extent that supernatural policing can promote prosocial behaviour, belief in a morally concerned deity should be selected for or stabilized in societies where free-riding is more likely to be a problem. Again, cross-cultural evidence is consistent with this claim, with moralizing high gods significantly more likely to occur in larger societies, where enforcement costs are likely to be high (Roes & Raymond, 2003), and in regions of water scarcity, where free riding may be especially costly to the group (Snarey, 1996).
One limitation of these cross-cultural studies is that they examine variation at the societal level, but can say little about whether a widely held belief affects those individuals within a society in the manner predicted. Hence, as Johnson (2005) notes, the causal mechanisms linking the presence of moralizing high gods to other societal characteristics remain obscure. Conversely, experimental work supports a link between priming supernatural agent concepts and individual cooperation, but we do not yet know whether such effects apply to long-standing beliefs about supernatural agents and enduring moral dispositions. Furthermore, experimental work has been carried out in only a handful of WEIRD (Western, Educated, Industrialized, Rich and Democratic) cultures (Henrich, Heine, & Norenzayan, 2010). The effect of supernatural monitoring and punishment is put forward as universal, but without a large cross-cultural sample we cannot tell whether any observed association between supernatural belief and morality is the result of underlying psychological universals or some other factor peculiar to the cultural context of the experiments. In the United States for example, a culture has developed in which religious beliefs are particularly closely tied to moral judgements about abortion. To disentangle the effects of universal tendencies from cultural context, we need individual-level data sampled across many cultures and belief systems.
In the present article, we bridge these two levels of analysis, using the World Values Survey (WVS, 1981–2008) to assess individual differences across 87 countries covering a diverse range of social and religious backgrounds. Whilst previous sociological studies have established an association between self-reports of religiosity and prosociality (Monsma, 2007, Stark, 2001), predictions relating specifically to supernatural monitoring and punishment have not been tested using individual survey data. Here, we test these predictions using questionnaire data from the WVS on individuals' beliefs about God and the afterlife, together with ratings of the justifiability of a standard set of moral transgressions. This direct approach — asking individuals to state their beliefs about the supernatural and about conformity to the moral order — is of course subject to the potential biases and limitations of any self-report data. The advantage of such an approach, however, is that a large amount of information can be collected from across many different cultural settings. We can therefore use the questionnaire data to investigate whether supernatural beliefs are related to moral thinking consistently across cultures, controlling for possible cultural differences or the effect of other potential confounds such as religious activity or education.
First, we predict that, due to the effect of supernatural monitoring and/or supernatural punishment, those individuals who profess belief in God will rate moral transgressions as less justifiable than those who do not (Hypothesis 1). Second, following the fear of supernatural punishment hypothesis, we predict that stronger beliefs about the unjustifiability of moral transgressions will be present in individuals who profess belief in heaven and/or hell (Hypothesis 2) — implying belief in reward and punishment in the afterlife. Note that the supernatural punishment argument also applies to rewards, since withholding a reward can be seen as equivalent to punishment. Third, following the supernatural monitoring hypothesis, we predict that, among those who believe in God, those individuals who profess belief in a personal God will rate moral transgressions as less justifiable than those who believe in a Spirit or Life Force, since believing in a personal God is more likely to imply active monitoring and a sense of ‘being watched’ (Hypothesis 3). These predictions should hold after controlling for variation in religious participation (discussed above). Similarly, the predictions should hold after controlling for the effect of cultural or religious background in the form of level of education, reported religious denomination and cultural variation across countries.
Section snippets
Data sources and variables
To test our hypotheses we used cross-cultural survey data from five waves of the World Value Survey (WVS, 1981–2008), covering the period from 1981 to 2008. This aggregate includes standardized surveys conducted in 87 countries all over the world, comprising a total of 355,298 individual face-to-face interviews. Each interview consists of the administration of a questionnaire, covering a total set of 1079 variables. The questionnaires for the five waves of the survey as well as the data and the
General predictions
Table 1 summarizes the results of the ordinal regressions relating supernatural beliefs to beliefs about the justifiability of moral transgressions. Consistent with Hypothesis 1, belief in God predicts stronger belief in the unjustifiability of all 14 moral transgressions examined (all p values <.001). Likewise, consistent with Hypothesis 2, those individuals who believe in heaven or hell rated all 14 moral transgressions as less justifiable than those who do not believe in heaven or hell (all p
Discussion
The results we present here are consistent with and provide support for specific predictions of the supernatural monitoring and fear of supernatural punishment hypotheses. As predicted by both theories and consistent with our Hypothesis 1, individuals who professed belief in God rated moral transgressions as less justifiable than those who did not. Consistent with Hypothesis 2 and the supernatural punishment hypothesis, stronger beliefs about the unjustifiability of moral transgressions were
Acknowledgments
We wish to thank Robin Dunbar, Nicolas Baumard and two anonymous reviewers for helpful advice and comments.
References (58)
Exploring the natural foundations of religion
Trends in Cognitive Sciences
(2000)- et al.
Explaining altruistic behavior in humans
Evolution and Human Behavior
(2003) - et al.
Nobody's watching? Subtle cues affect generosity in an anonymous economic game
Evolution and Human Behavior
(2005) - et al.
Why people punish defectors. Weak conformist transmission can stabilize costly enforcement of norms in cooperative dilemmas
Journal of Theoretical Biology
(2001) - et al.
Altruism toward in-group members as a reputation mechanism
Evolution and Human Behavior
(2010) - et al.
The dynamics of indirect reciprocity
Journal of Theoretical Biology
(1998) - et al.
The origins of religion: Evolved adaptation or by-product?
Trends in Cognitive Science
(2010) - et al.
Minimal social cues in the dictator game
Journal of Economic Psychology
(2009) - et al.
Belief in moralizing gods
Evolution and Human Behavior
(2003) - et al.
Scars for war: Evaluating alternative signaling explanations for cross-cultural variance in ritual costs
Evolution and Human Behavior
(2007)
Religion and trust: An experimental study
Journal of Economic Psychology
Ritual, emotion, and sacred symbols the evolution of religion as an adaptive complex
Human Nature
In gods we trust: The evolutionary landscape of religion
Cues of being watched enhance cooperation in a real-world setting
Biology Letters
The folk psychology of souls
Behavioral and Brain Sciences
Reasoning about dead agents reveals possible adaptive trends
Human Nature
The meme machine
The evolution of altruistic punishment
Proceedings of the National Academy of Sciences of the United States of America
Religion explained: The human instincts that fashion gods, spirits and ancestors
Morality play, or playing morality? Intrinsic religious orientation and socially desirable responding
Self and Identity
The selfish gene
The God delusion
Breaking the spell
Mind the bonding gap: Constraints on the evolution of hominin societies
The relationship between the judged desirability of a trait and the probability that the trait will be endorsed
Journal of Applied Psychology
The nature of human altruism: Proximate patterns and evolutionary origins
Nature
Altruistic punishment in humans
Nature
Eyes are on us, but nobody cares: Are eye cues relevant for strong reciprocity?
Proceedings of the Royal Society of London, Series B
Faces in the clouds: A new theory of religion
Cited by (134)
God or good health? Evidence on belief in God in relation to public health during a pandemic
2023, Journal of Behavioral and Experimental EconomicsReligion and cooperation across the globe
2023, Journal of Economic Behavior and OrganizationReligiousness and adolescent risk behavior
2023, Encyclopedia of Child and Adolescent Health, First EditionCultural differences in neurocognitive mechanisms underlying believing
2022, NeuroImageCitation Excerpt :Beliefs may serve a purpose by being linked to personal intuitive judgments about subjective certainty of sensory perceptions and mental constructs (Harris et al., 2008). Beliefs may also serve as a foundation of concepts about oneself and others (Blaine et al., 1998; Han et al., 2008) and influence reliant commitment to others and large-scale cooperation (Atkinson and Bourrat, 2011; Rotter,1980; Weingast,1995). Brain imaging studies have investigated the neural correlates of religious or secular beliefs and revealed that the mPFC is engaged in religious experiences (Azari et al., 2001) and in maintaining own political beliefs (Kaplan, Gimbel, and Harris, 2016).
Supernatural punishment beliefs as cognitively compelling tools of social control
2022, Current Opinion in PsychologyDeterminants of social norms II - Religion and family as mediators
2024, Journal of Institutional Economics
- ☆
Grants: This work was supported by the EC-FP6 EXREL project grant 43225 and two grants (to Q.A. and P.B.) from the John Templeton Foundation-funded Cognition, Religion and Theology Project at Oxford University.
- 1
Both authors contributed equally to this work.