Is belief in supernatural agency an example of evolved “misbelief”? McKay & Dennett (M&D) consider recent psychological experiments that have investigated whether religious beliefs cause prosocial behavior such as generosity and honesty (for reviews, see Norenzayan & Shariff Reference Norenzayan and Shariff2008; Shariff et al. Reference Shariff, Norenzayan, Henrich, Schaller, Norenzayan, Heine, Yamagishi and Kameda2010). In M&D's philosophical analysis, whether or not religion supplies a case of evolved misbelief turns out to depend on the psychological mechanism that best accounts for these effects. We therefore revisit the experimental evidence and discuss in some depth the ideomotor and supernatural watcher accounts for these effects.
M&D cite Randolph-Seng and Nielsen (Reference Randolph-Seng and Nielsen2008), who critiqued Shariff and Norenzayan (Reference Shariff and Norenzayan2007), questioning the plausibility of the supernatural watcher hypothesis because the data could not conclusively distinguish between the ideomotor and supernatural watcher explanations. These two mechanisms gain plausibility given two distinct but well-supported empirical literatures. There is considerable evidence showing that prosocial behavior can be facilitated both by activating nonconscious altruistic thoughts (e.g., Bargh et al. Reference Bargh, Gollwitzer, Lee-Chai, Barndollar and Troetschel2001), and by heightened reputational concerns (e.g., Fehr & Fischbacher Reference Fehr and Fischbacher2003). These two mechanisms are not mutually exclusive, however, and may even reinforce each other in everyday life.
The interesting question therefore is: What kind of laboratory evidence can provide support for the supernatural watcher account above and beyond behavioral-priming processes? First, if the priming effects of God concepts are weaker or nonexistent for non-believers, then the effect could not be solely due to ideomotor processes, which are typically impervious to prior explicit beliefs or attitudes. Second, if God primes make religious participants attribute actions to an external source of agency, these effects could not be explained by ideomotor processes, as such manipulations disambiguate the felt presence of supernatural watchers from their alleged prosocial consequences. Finally, if the supernatural watcher explanation is at play, religious primes should arouse social evaluation of the self. Moreover, such reputational awareness should moderate the magnitude of the prime's effect on prosocial behavior.
As M&D note, evidence on the first point is currently mixed. However, close examination of the findings betrays a revealing pattern. All but one of these priming studies recruited student samples, which can be problematic since beliefs, attitudes, and social identity among students can be unstable, raising questions about the reliability of chronic individual difference measures of religious belief and identity measures for students who are still in transition to adulthood (Sears Reference Sears1986; Henrich et al., in press). Thus, student atheists might be at best “soft atheists.” In the only religious priming experiment we are aware of that recruited a non-student adult sample (Shariff & Norenzayan Reference Shariff and Norenzayan2007, Study 2), the effect of the prime emerged again for theists, but disappeared for these “hard” atheists (see Fig. 1). In addition, Henrich et al. (Reference Henrich, Ensminger, McElreath, Barr, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Henrich, Lesorogol, Marlowe, Tracer and Ziker2009) found that across 14 small-scale societies of varying group size, where there is variability in whether supernatural agents are morally concerned, belief in the moralizing Abrahamic God (along with degree of market integration) predicted larger offers in the dictator and ultimatum games. These initial findings speak against an exclusively ideomotor account of the results, and suggest that belief – not just alief – is involved in religious prosociality.
Results from the dictator game in Shariff and Norenzayan (2007, Study 2) indicate that priming God concepts increased generosity for religious believers but not for atheists. Error bars represent standard error of the mean.
Regarding the second question, one experiment clearly separates the felt presence of a supernatural agent from prosocial outcomes. Dijksterhuis et al. (Reference Dijksterhuis, Preston, Wegner and Aarts2008) found that after being subliminally primed with the word “God,” believers (but not atheists) were more likely to ascribe an outcome to an external source of agency, rather than their own actions. In addition, religious belief positively correlates with greater concern with social evaluation of the self (Trimble Reference Trimble1997), and recent experimental evidence points to this being a causal relationship. Gervais and Norenzayan (Reference Gervais and Norenzayan2009) found that priming God concepts (using the same sentence unscrambling task of Shariff and Norenzayan [2007]) increased public self-awareness (Govern & Marsch Reference Govern and Marsch2001) – a measure that taps into feelings of being the target of social evaluation. In contrast, and as predicted, the prime had no effect on private self-awareness. Ongoing research is examining whether prosocial effects of religious primes are moderated by measures of evaluative concern, a key prediction of the supernatural watcher hypothesis, which would be incompatible with a purely ideomotor account. Thus, although M&D are right that more research is needed to reach firm conclusions, the evidence regarding the supernatural watcher hypothesis is more compelling than M&D's cautious approach suggests. But does that mean that belief in supernatural agents is an example of adaptive misbelief?
M&D briefly mention both by-product theories of religion and cultural evolutionary explanations for cooperation. We have argued elsewhere (Norenzayan & Shariff Reference Norenzayan and Shariff2008; Norenzayan, in press; Shariff et al. Reference Shariff, Norenzayan, Henrich, Schaller, Norenzayan, Heine, Yamagishi and Kameda2010) that integrating these two frameworks yields a more cogent explanation for the rise and persistence of religious beliefs than theories which invoke a more direct genetic evolutionary argument (e.g., Bering et al. Reference Bering, McLeod and Shackelford2005; Johnson & Bering Reference Johnson and Bering2006). Once belief in supernatural agency emerged as a by-product of mundane cognitive processes, cultural evolution favored the spread of a special type of supernatural agent – moralizing high Gods. Growing evidence is converging on the conclusion that sincere belief in these omniscient supernatural watchers facilitated cooperation and trust among strangers (Norenzayan & Shariff Reference Norenzayan and Shariff2008). Not surprisingly, this cultural spread coincided with the expansion of human cooperation into ever larger groups over the last 15 millenia (Cauvin Reference Cauvin2000). This evolutionary scenario has the virtue of explaining an otherwise puzzling feature of religious prosociality – namely, the systematic cultural variability in the prevalence of moralizing Gods across societies that correlates with group size (e.g., Roes & Raymond Reference Roes and Raymond2003). Contrary to a genetic adaptation account, the deities of most small-scale societies, which more closely approximate ancestral conditions, are neither fully omniscient nor morally concerned. It is the evolutionarily recent anonymous social groups, facing the breakdown of reputational and kin selection mechanisms for cooperation, which most strongly espouse belief in such Gods. Thus, beliefs in moralizing supernatural agents may not qualify as genetically evolved misbeliefs. But they could instead be seen as examples of culturally evolved ones that played a key historical (although not irreplaceable) role in the rise and stability of large cooperative communities.
