I applaud the fresh perspective this paper brings. The argument that child peer groups are largely independent active producers and maintainers of autonomous cultures goes against the grain of much cultural evolution research, which has focused almost exclusively on adults and assumed children to be passive recipients of vertically transmitted cultural knowledge. The authors’ evidence that child peer groups impact how human culture evolves is convincing. Also welcome is the target article’s focus on the role of creative processes in maintaining cultural diversity, which, with a few exceptions (Enquist, Ghirlanda, & Eriksson, Reference Enquist, Ghirlanda and Eriksson2011; Fogarty, Creanza, & Feldman, Reference Fogarty, Creanza and Feldman2015; Gabora, Reference Gabora2008; Gabora & Smith, Reference Gabora, Smith, Henley, Kardas and Rossano2019), receives little attention.

The authors note that, “Rather than just more time for learning adult skills, long childhoods also allow for the generation and maintenance of peer cultural products.” Presumably, the making of these products develops their minds in new ways. Since peer cultures incorporate complex, contextual relationships amongst products, peer cultures – and the mental structures behind them – are most aptly modeled, not as “collections” of cognitive traits, but as networks. Thus, the approach taken here is consistent with network science approaches to cultural evolution (Buskell, Enquist, & Jansson, Reference Buskell, Enquist and Jansson2019; Gabora, Reference Gabora1998) and with the hypothesis that it is integrated cognitive networks – i.e., ways of structuring knowledge – that evolve through culture, as opposed to discrete units of cultural knowledge (Gabora, Reference Gabora2004). The authors write, “Through collaborative exploration, peer interaction can be generative, leading to the discovery of information not initially possessed by either child,” and this too is amenable to network science methods. They have been applied not just to cultural evolution but to the development of children’s mental models (Beckage, Aguilar, & Colunga, Reference Beckage, Aguilar and E. Colunga2015; Beckage, Smith, & Hills, Reference Beckage, Smith and Hills2011; Stella, Beckage, & Brede, Reference Stella, Beckage and Brede2017), including the interactive development of a shared mental model extending across multiple children (Gabora, Beckage, & Steel, Reference Gabora, Beckage and Steel2022).

It was interesting to learn that, prior to age three, children are not adapted for collaborating with peers, only with adults. This is consistent with the proposal that the density of causal and associative links connecting mental representations in the toddler’s mind crosses a percolation threshold, such that it undergoes a phase transition from discrete islands of connectivity to an integrated network (Gabora, Beckage, & Steel, Reference Gabora, Beckage and Steel2022). It seems reasonable that, before this threshold is crossed, obtaining enough knowledge to form an integrated (initially small) world model is best achieved through interactions with adults, while after it is crossed, playful exploration and growth of this primitive cognitive structure is best achieved through interactions with peers. After all, with age, thinking can become more rigid, and views can become entrenched, potentially stifling new ideas and perspectives (Candeias & Galindo, Reference Candeias and Galindo2021; Ross et al., Reference Ross, Lachmann, Jaarsveld, Riedel-Heller and Rodriguez2023). This kind of phase transition in cognitive development can be modeled using autocatalytic networks. Autocatalytic networks have been used to model not just how concepts coalesce into an integrated network in a young child’s mind but also cognitive changes underlying the origins of the capacity for cumulative, adaptive, open-ended cultural evolution (Gabora & Steel, Reference Gabora and Steel2017, Reference Gabora and Steel2020a,Reference Gabora and Steelb). Autocatalytic network models distinguish (1) innate knowledge (nature) and knowledge obtained from the environment, e.g., through social learning (nurture), from (2) derived knowledge generated autonomously by the individual(s) in question. While nature and nurture provide the raw conceptual materials for an integrated understanding of the world, derived knowledge establishes relationships amongst these raw materials, and fills in gaps through the generation of new ideas (Gabora & Robertson, Reference Gabora, Robertson, Pérez-Jara and Ongay2025). This distinction enables us to track the products and changes that individuals (or interacting groups) contribute to cultural lineages. Thus, autocatalytic networks might be particularly suitable for modelling cognitive change through children’s cultural interactions with peers.

This research may help uncover new mechanisms by which cultures evolve. It suggests that child peer cultures facilitate a cycle of dismantling and reconstituting of mental models that enables each new generation to escape the mental ruts of the previous generation, and adapt to environmental changes. If this is the case, then culture may evolve, not just because knowledge accumulates, but because, during childhood, each generation collectively establishes new cognitive scaffolds that constrain and enable what thought trajectories and cultural contributions are possible, both in childhood and later in life. Further investigation of this hypothesis using network-based computer models of cultural evolution, and comparison of results with and without child peer group subcultures, could potentially constitute a landmark contribution to research at the intersection of cognitive development and cultural evolution.

Although the target paper is not the first to do so, the use of the term “inheritance” in reference to cultural traits is ill-advised. An inherited trait is transmitted vertically (e.g., from parent to offspring) by way of a self-assembly code (e.g., DNA) and therefore does not need to be re-learned or re-acquired by the next generation. This is not the case with respect to cultural traits (Gabora, Reference Gabora2011; Smith, Gabora, & Gardner-O’Kearny, Reference Smith, Gabora and Gardner-O’Kearny2018); for example, if you know how to ride a bike, your child doesn’t automatically know how to ride a bike but must learn this skill from scratch. Biologists use the term acquired traits to refer to traits that are obtained not through inheritance but through interaction with the environment (e.g., social learning), and it seems prudent to adhere to this distinction in the cultural evolution literature as well (Gabora, Reference Gabora2011).

“Selection” is also misleading when applied in a cultural context. In its scientific usage, the term selection is applicable when an evolutionary process occurs due to differential replication of randomly generated heritable variations in a population over generations, such that some traits become more prevalent than others. In culture, rarely are multiple variants randomly generated, and distinctions between variants are not heritable (see above). Cultural change is primarily due to whatever is biasing the generation of variants away from random factors, such as intelligence, planning, and intuition, which are not random but based on pattern completion (Bowers, Farvolden, & Mermigis, Reference Bowers, Farvolden, Mermigis, Smith, Ward and Finke1995; Gigerenzer, Reference Gigerenzer2023).