Some cross-cultural datasets show that countries with the highest levels of legal and political gender equality also show the largest sex differences in diverse measures such as personality traits, mate choice preferences, and subject or career choice (e.g., Falk and Hermle, Reference Falk and Hermle2018). This “paradoxical” finding has been taken as evidence that legal gender equality “frees” women and men to express different, evolved, traits (Schmitt, Reference Schmitt, Shackelford and Hansen2015). We argue, instead, that this pattern is similar to the complex relationship between vitamin D, skin cancer risk, and latitude discussed in the target article: It all makes sense if you consider cultural evolution and its power to shape the phenotypic landscape more broadly.

Humans are a cooperatively breeding species inhabiting diverse ecological niches (Sear, Reference Sear2016). For instance, forager groups differ in the proportion of both direct care (Kramer & Veile, Reference Kramer and Veile2018) and calories provided by fathers (Marlowe, Reference Marlowe2001). They also vary in reliance on forms of subsistence which are less compatible with care for very young infants (e.g., open sea hunting; Marlowe, Reference Marlowe2007). Furthermore, it is likely that humans have adopted diverse mating systems across our evolutionary history. As such, the ability to acquire gendered behaviour flexibly is likely essential.

Tight hormonal control of specific sex-typed behaviours is absent in humans. Instead, hormones such as testosterone (among other behavioural effects) appear to bias the process by which behaviour is acquired, by facilitating a tendency to copy the behaviour of own-gender groups (Hines, Reference Hines2020). It's important to note that this apparent “own-gender” bias can itself be understood as a more simple “copy self-similar others” bias (Meltzoff, Reference Meltzoff2007; Wood, Kendal, & Flynn, Reference Wood, Kendal and Flynn2013), where children select those “others” from among the available categories within their culture. It may also be reinforced by a tendency to copy one's playmates in societies where children are segregated by sex and/or gender (Wood et al., Reference Wood, Kendal and Flynn2013).

Where the cultural learning of locally adaptive gendered behaviour guides individuals into one of a few niches, patterns of sex differences might be relatively easy to predict. Research on spatial cognition, for instance, has demonstrated that patterns of geographical mobility in hunting and trading may be responsible for adult sex differences in spatial cognition in some cultural contexts (Twe and Himba: Vashro, Padilla, and Cashdan, Reference Vashro, Padilla and Cashdan2016) but not others (Tsimane: Trumble, Gaulin, Dunbar, Kaplan, and Gurven, Reference Trumble, Gaulin, Dunbar, Kaplan and Gurven2016). When populations undergo changes, however, predicting the effects of these changes will be complex. While the introduction of schooling and concomitant reduced differences in mobility patterns among Twa children reduced sex differences in spatial cognition (Davis, Stack, & Cashdan, Reference Davis, Stack and Cashdan2021), long-term settlement in the Agta increased sex differences in child domestic labour (Page et al., Reference Page, Hassan, Viguier, Dyble, Smith and Migliano2021). Given that both of these changes could be thought of as making a population more “WEIRD” (western, educated, industrialized, rich, and democratic), we see here that predicting the effects of a construct such as “development” or “gender equality” on sex differences is complex.

Populations with high gender equality indices would mostly be characterised as “WEIRD” societies in the target article. They represent very different ecologies from those we mentioned above. Consider industrialised labour markets; not only are gender roles less constrained by physiology than in foraging societies, but also the number of roles and behavioural and social niches an individual can occupy is vast (Smaldino, Lukaszewski, von Rueden, & Gurven, Reference Smaldino, Lukaszewski, von Rueden and Gurven2019). The search for locally adaptive behaviour is less constrained and therefore more complex. Where decisions are difficult, social learning will play more of a role (Morgan, Rendell, Ehn, Hoppitt, & Laland, Reference Morgan, Rendell, Ehn, Hoppitt and Laland2012) and choices may be arbitrary. We, therefore, expect individuals to be more influenced by factors such as gender in industrialised societies than in foraging and subsistence societies, even when those industrialised societies are relatively gender egalitarian (see Anker [Reference Anker1998] for discussion of gendered labour in Nordic countries). The dramatic shift in computing from being a female profession to a male-dominated one in the United States and United Kingdom (Abbate, Reference Abbate2017) illustrates that the association of gender with behaviour in such societies is malleable and can be arbitrary.

The target paper also argues that typical units of analysis for group comparison effects, such as nation states, or racial groups, are unlikely to capture the actual cultural clustering in a population. This is likewise true for studies of variation in sex differences. However, in addition to the possibility that lower-level clusters are critical, we note that industrialised countries exist within higher-level clusters created by shared media culture (TV, advertising, social media, etc.). For instance, the outputs of film industries in Hollywood and Bollywood (which are highly gendered in both similar and different ways, e.g., Ghaznavi, Grasso, and Taylor, Reference Ghaznavi, Grasso and Taylor2017) are both consumed across large, sometimes overlapping, swathes of the global population.

Our proposition here is in some ways similar to Wood & Eagly's (Reference Wood and Eagly2012) argument that a society's division of labour drives stereotypes and thereby children's learning. While the actual distribution of adult roles might be observable in small groups, however, children in industrialised nations likely have their perceptions of appropriate behaviours shaped more by visual media than by the social roles of the adults around them. We have already demonstrated that, during market integration, globalised media alter preference for sexually relevant traits, reducing idealised female body and waist size (Boothroyd et al., Reference Boothroyd, Jucker, Thornborrow, Barton, Burt, Evans and Tovée2020; Swami et al., Reference Swami, Frederick, Aavik, Alcalay, Allik, Anderson and Zivcic-Becirevic2010) even where ecological pressures such as nutritional stress should direct preferences in the opposite direction (Jucker et al., Reference Jucker, Thornborrow, Beierholm, Burt, Barton, Evans and Boothroyd2017). Given that high gender equality indices are associated with economic development, access to visual media (particularly advertising) might be at its most potent in shaping gendered preferences (Fisher & Jenson, Reference Fisher and Jenson2017) when structural inequality is relatively small.

In other words, we support the claim in the target article that cultural clusters “explain intergroup differences better than genes do.” By considering gender as a “self-organizing trajectory of environmental experience” like those described in the target article, we believe this insight can make sense of the “paradoxical” relationship between gender equality indices and sex differences in behaviour and personality.