Conviction Narrative Theory (CNT) resonates with the anthropological evidence that societies use narratives to frame uncertain decisions (Douglas, Reference Douglas2013). People evaluate narratives with a blend of emotion and intuition (Zinn, Reference Zinn2016), and CNT convincingly outlines how this might be part of an adaptive decision-making system under radical uncertainty. And yet anthropological evidence also shows that high-risk and uncertain decisions frequently involve religious narratives with supernatural causes and other falsehoods (Evans-Pritchard, Reference Evans-Pritchard1937; Malinowski, Reference Malinowski1932). If our narratives are adaptive for dealing with uncertain fitness-relevant problems, then why are so many narratives false?

A popular answer is that our evolved cognition is like a flawed superpower: Humans might be generally sensitive to the statistical and causal structure of their environments (Sperber, Premack, & Premack, Reference Sperber, Premack and Premack1995), but they are also prone to biases and misleading intuitions that disrupt these capabilities (Kahneman, Reference Kahneman2013). An alternative answer, which I advocate here, is that religious falsehoods are part of a broader tendency to generate useful, ecologically rational narratives (Lightner & Hagen, Reference Lightner and Hagen2022). Under radical uncertainty, where information is scarce and the data-generating processes are unknown, how should people think about fitness-relevant challenges like natural hazards or social conflicts? I will discuss three ingredients that supernatural falsehoods might add to decision-making narratives.

The first ingredient is that they simplify complex problems. Contra Johnson et al. in section 6.1, there are strong theoretical reasons to suggest that simple narratives explain more of the data under radical uncertainty, because they are less sensitive to variance around future observations (i.e., they improve a bias-variance tradeoff; Brighton & Gigerenzer, Reference Brighton, Gigerenzer and Schulkin2012). When explaining complex phenomena, a simpler narrative that ignores the idiosyncratic details of individual observations is more likely to generalize to other scenarios with different background circumstances (Quillien, Reference Quillien2020; Woodward, Reference Woodward2006).

A familiar example is the intentional stance (Dennett, Reference Dennett1987). Humans can predict and navigate social behavior because we ascribe simple, reductive mental state concepts to a truly complex and mysterious type of data-generator: other minds (Gerstenberg & Tenenbaum, Reference Gerstenberg and Tenenbaum2017). Religious falsehoods similarly tend to impute oversimplified, anthropomorphic falsehoods onto complex phenomena. For example, many societies depend on forest resources. Forests are complex ecological systems (Filotas et al., Reference Filotas, Parrott, Burton, Chazdon, Coates, Coll and Messier2014), where many species interact in a dynamic “web-like structure” resembling social causation (target article, sect. 6.1). Animistic religions often construe these forest systems as having person-like entities that purposefully engage in helpful, antagonistic, and communicative relationships with each other (Atran et al., Reference Atran, Medin, Ross, Lynch, Vapnarsky, Ek and Baran2002; Ingold, Reference Ingold2006). This perspective might be dismissed as superstitious, but it can serve the genuinely useful and otherwise difficult task of making sense of a complex, dynamic ecological network (ojalehto mays, Seligman, & Medin, Reference ojalehto mays, Seligman and Medin2020).

The second ingredient is that supernatural falsehoods are often sensitive to long-term outcomes that matter. Consider, for example, the Australian Martu landscape burning practices that ultimately lead to continuing food availability (Bird, Tayor, Codding, & Bird, Reference Bird, Tayor, Codding and Bird2013). These practices create a collective action problem because the payoffs are only high if people participate in the required labor (Lightner & Purzycki, Reference Lightner, Purzycki, Purzycki and Bendixen2023). Rather than resorting to decision or game theory, Martu view their burning practices as part of a sacred narrative where failing to participate would bring about the end of the world (Bird, Bird, Codding, & Taylor, Reference Bird, Bird, Codding and Taylor2016). In a real sense, this has useful causal information: In their harsh and arid environment, a collapse of the food supply would be a dire outcome.

Even explanations for “randomness” vary across cultures in ways that seem sensitive to long-term ecological risks (Tucker, Reference Tucker and Stevens2017). Howell (Reference Howell2012), for example, observed that the Malaysian Chewong explain misfortune in terms of spiritual retribution for norm violations rather than chance. In a sense, this seems irrational because it posits a tapestry of non-human agents who sanction inauspicious practices. In another sense, Howell argues, there's an ecologically relevant logic to their view of randomness, because Chewong social rules “prevent risk situations from occurring” while recognizing the fact that “in performing their daily tasks, humans are not separate from the natural world” (p. 9).

Interestingly, Johnson et al. discuss an unpublished experiment that also seems to suggest that people think of randomness as an explanation, rather than a lack thereof (Johnson, Matiashvili, & Tuckett, Reference Johnson, Matiashvili and Tuckett2019; sect. 7.1.1). They also mention that participants tended to think that unexplained price changes were signals when positive. If participants had a long enough time horizon in mind, then this does not seem like a particularly irrational heuristic because major stock market indices do tend to increase over a long enough timeframe (Kim & Ryoo, Reference Kim and Ryoo2011). Like the Martu and Chewong, when given a noisy process with scarce information, participants might have been sensitive to the feedback they have received in the long run.

The third ingredient is that supernatural falsehoods can effectively manipulate behavior by evoking strong emotions when they are communicated. Consider the many sources of danger that forests can contain at night (e.g., predators, enemies, falling hazards). Rather than enumerating and weighing all the possibilities, many societies use narratives to evoke salient fears that effectively keep others from harm's way (e.g., “avoid the forest at night because monsters lurk in there”) (Morin & Sobchuk, Reference Morin and Sobchuk2022; Sugiyama, Sugiyama, Slingerland, & Collard, Reference Sugiyama, Sugiyama, Slingerland and Collard2011). Supernatural punishment is an especially common and effective tool for motivating cooperative behavior because it presents decision-makers with a distressing imagined future for breaking the rules (Bendixen & Purzycki, Reference Bendixen and Purzycki2017; Fitouchi & Singh, Reference Fitouchi and Singh2021). Indeed, supernatural appeals across cultures are not about random subject matter; they are usually about locally important socioecological challenges like resource management and cooperative conflicts (Bendixen et al., Reference Bendixen, Apicella, Atkinson, Cohen, Henrich, McNamara and Purzycki2021; Purzycki, Bendixen, Lightner, & Sosis, Reference Purzycki, Bendixen, Lightner and Sosis2022). They are also sensitive to long-term causation: Cheating might pay in the short-term, but losing a cooperative partner can genuinely present a long-term opportunity cost (Delton, Krasnow, Cosmides, & Tooby, Reference Delton, Krasnow, Cosmides and Tooby2011).

Finding the narrative that “feels right,” as CNT puts it, can help bypass a considerable amount of noise when responding to a complex and mysterious problem. More importantly, religious narratives, though false or seemingly irrational, often turn out to reflect concerns about behaviors that matter for people's livelihoods. With radical uncertainty, they are often all we have.