How cultural evolution went blind

Cultural evolution was developed by population geneticists and biologists (Boyd & Richerson, Reference Boyd and Richerson1988; Cavalli-Sforza & Feldman, Reference Cavalli-Sforza and Feldman1981), and it has always been modeled closely after biological evolution. In biological evolution, inheritance alone does not cause systematic evolutionary change, except in rare cases like meiotic drive (Mesoudi, Reference Mesoudi2016). Similarly, animals do not change their genes through conscious will. These principles, translated to cultural evolution, imply that people should not systematically select cultural traits during the transmission process, nor should they re-engineer traits after inheritance. Instead, cultural traits should change through processes that are “blind” to their function, just like genetic mutation is blind to the function of the mutated genes. A classic case of “blind” cultural evolution is prestige-based transmission, in which people copy all traits of prestigious people without knowing which traits have made them prestigious (Henrich & Gil-White, Reference Henrich and Gil-White2001; Richerson & Boyd, Reference Richerson and Boyd2008).

No cultural evolutionist has claimed that cultural evolution unfurls solely through blind processes. In Culture and the Evolutionary Process, Boyd and Richerson (Reference Boyd and Richerson1988) introduced the process of “guided variation,” in which an individual acquires a cultural trait, intentionally modifies it, and then passes it on. In our own work, we have applied guided variation to model the cultural evolution of language (Jackson et al., Reference Jackson, Lindquist, Drabble, Atkinson and Watts2022), religion (Caluori et al., Reference Caluori, Jackson, Gray and Gelfand2020; Jackson et al., Reference Jackson, Caluori, Abrams, Beckman, Gelfand and Gray2021), and emotion (Lindquist et al., Reference Lindquist, Jackson, Leshin, Satpute and Gendron2022).

Nevertheless, blind processes are undoubtedly more popular and more synonymous with cultural evolutionary thinking than “sighted” processes like guided variation. Perhaps blind processes became popular because they were not already well-established in the social sciences. Perhaps it is just more counterintuitive to think of cases where people copy and preserve a cultural trait without understanding its function than to think of cases where people copy a trait they perceive as valuable.

The mind in the cultural machine

Despite the value of blind processes, they clearly do not explain all forms of cultural change (Pinker, Reference Pinker2012), and Singh’s target article is right to observe their limits. We agree with Singh that it is a good time for cultural evolution science to regain its sight. Fortunately, there is also no need to build models of subjective selection from scratch, as psychologists have been studying subjective selection for many decades.

Reviewing three insights from these studies shows how humans go beyond classic blind processes like prestige-based learning, but also reveals curious new forms of blindness. The first insight, for example, is that people do not always know what will benefit them (Kahneman, Knetsch, & Thaler, Reference Kahneman, Knetsch and Thaler1991). Humans are Bayesian, which means that we would rather stick with what we know rather than learn new tricks (Lieder & Griffiths, Reference Lieder and Griffiths2020). This tendency can produce cultural regularities that we would not expect if humans simply selected cultural traits with the highest payoff (Kirby, Dowman, & Griffiths, Reference Kirby, Dowman and Griffiths2007). Cultural systems can amplify this inertia, especially when strong norms raise the coordination costs of cultural change (Gelfand, Harrington, & Jackson, Reference Gelfand, Harrington and Jackson2017). When norms are strong, people may opt for conformist transmission, even when they recognize there are non-conformist alternatives.

Second, even when people try to do what benefits them, they may focus on the wrong time horizon. Temporal discounting – choosing strategies with high payoffs in the short run but lower payoffs in the long run – is the classic example of this error (Samuelson, Reference Samuelson1937). But problems with the time horizon can be more insidious. For example, people can learn to prioritize strategies that work well in few-shot dilemmas (“disengage after people are hostile”) but ultimately lead to inaccurate beliefs (“hostility is common”) because they prevent learning (you never find out that hostile people can also be nice) (Denrell, Reference Denrell2005). Similarly, people show little ability to learn from short-term failure, showing poorer memory for failures than successes (Eskreis-Winkler & Fishbach, Reference Eskreis-Winkler and Fishbach2019). Even “sighted” subjective selection can thus be blind to ultimate function.

Third, people do not always know how something will benefit them. Learning can be “model-free” (using immediate memory of reward) or model-based (involving a causal model of how behaviors yield reward) (Daw et al., Reference Daw, Gershman, Seymour, Dayan and Dolan2011). Psychological studies show that low-volatility environments select for model-free learning, meaning that people learn and inherit cultural traits without a causal model of how they work (Roberts-Gaal & Cushman, Reference Roberts-Gaal and Cushman2023; Roberts-Gaal, Bolic, & Cushman, Reference Roberts-Gaal, Bolic and Cushman2025). Through the same process of reinforcement learning, people may even subjectively select social learning strategies (e.g., conformist versus prestige-based learning) without knowing why (Schultner, Molleman, & Lindström, Reference Schultner, Molleman and Lindström2025).

The benefits of blindness

Many other insights should be listed here, but reviewing them comprehensively would require another target-length article. However, even this brief review reveals a common theme: Even in cases of “subjective selection,” people may be blinder to ultimate cultural payoffs than we might intuitively think (Gigerenzer & Selten, Reference Gigerenzer and Selten2002).

Singh’s article may foreshadow a new wave of cultural evolutionary science that models “sighted” processes, but sight is not everything. Blind processes – either the classic learning biases of cultural evolution or the modes of bounded rationality that have been described by psychologists – show good evidence, and they undoubtedly contribute to both “super” and “minor” cultural attractors. Just because cultural evolutionists have discovered the value of sight, they need not forget the benefits of blindness.