We agree with CC on the necessity of copying. In this regard, their model differs from some other recent ones that background or deny the importance of copying (e.g., Baumard & André, Reference Baumard and André2025; Osiurak et al., Reference Osiurak, Lasserre, Arbanti, Brogniart, Bluet, Navarro and Reynaud2021). Yet, CC claim that there is evidence for copying in apes’ “tool-use and food-processing skills.” They also imply that the manufacture of early stone tools by hominins required copying. Together, these claims place copying as one of the early baseline conditions from which human-style culture evolved. This assumption matters far more than CC acknowledge, because once it is granted, much of the hard explanatory work is already done.

Are these claims of CC correct? The question is ill-posed as formulated. Social learning comes in different varieties. On one taxonomy: there is social learning of “know-w” (i.e., -what, -where, and -when) and social learning of know-how (Tennie et al., Reference Tennie, Bandini, van Schaik and Hopper2020a). Know-how relates to the production of behavioural and artefact forms – often in sequential ways. There is also a distinction between know-how copying and know-how triggering (sensu Sperber, Reference Sperber and Aunger2000). Both involve know-how, but only the former depends on fine-grained causal sensitivity (sensu Woodward, Reference Woodward2010) with respect to the structure of the behaviour/artefact that is used as a model (Buskell & Tennie, Reference Buskell and Tennie2025).

These distinctions matter because the social-learning processes they label enable different things. As regards behaviours/artefacts that are too complex or arbitrary to arise in any way other than being causally based on model structures, only know-how copying fits the bill. In general, this is the type of social learning demanded by CC’s subsequent stages (i.e., > 2). Thus, as we see it, the key question is whether apes and early hominins possess(ed) know-how copying.

Current evidence suggests that know-how copying in apes is extremely limited – if it exists at all – unless humans train or enculturate them. The overwhelming majority of chimpanzee behaviours are individually reinventable (Gunasekaram et al., Reference Gunasekaram, Battiston, Sadekar, Padilla-Iglesias, van Noordwijk, Furrer, Manica, Bertranpetit, Whiten, van Schaik, Vinicius and Migliano2024; see also Whiten et al., Reference Whiten, Goodall, McGrew, Nishida, Reynolds, Sugiyama, Tutin, Wrangham and Boesch2001; Motes-Rodrigo & Tennie, Reference Motes-Rodrigo and Tennie2021). Thus, they are latent solutions, by definition. The few remaining candidates for know-how copying are much debated (e.g., Koops et al., Reference Koops, Soumah, van Leeuwen, Camara and Matsuzawa2022 vs. Tennie & Call, Reference Tennie and Call2023). Moreover, apes untrained and/or unacculturated by humans do not spontaneously copy know-how under well-controlled conditions (reviewed in Tennie, Hopper, & van Schaik, Reference Tennie, Hopper, van Schaik, Hopper and Ross2020b). Know-how copying plausibly exists elsewhere in the animal kingdom (though it is rare, and usually restricted to the song domain). But treating it as present in apes in any robust sense is unjustified.

Turn now to stone tools. At least as regards early stone tools (i.e., > 800 kya), the claim that they evidence know-how copying is dubious. Knapping-naive humans reinvent early stone tool making techniques (Snyder, Reeves, & Tennie, Reference Snyder, Reeves and Tennie2022), and both wild and captive apes can reinvent much of the relevant “knapping” know-how (Masi et al., Reference Masi, Pouydebat, San-Galli, Meulman, Breuer, Reeves and Tennie2022; Motes-Rodrigo et al., Reference Motes-Rodrigo, McPherron, Archer, Hernández-Aguilar and Tennie2022). The social learning signals of the early stone tool record are more comparable to the ape pattern than the later human stone-tool pattern. This suggests that both early stone tool and ape cultures are minimal cultures comprised of latent solutions (Tennie et al., Reference Tennie, Premo, Braun and McPherron2017).

Finally, consider language. Whatever one’s view about the role of know-how copying in technological evolution, its importance to language is obvious. CC clearly acknowledge that language is not plausibly transmitted and retained without robust copying capacities (see CC’s emphasis on “miming” in their Stage 4; see also Tennie, Call, & Tomasello, Reference Tennie, Call and Tomasello2009; Planer, Bandini, & Tennie, Reference Planer, Bandini, Tennie, Raviv and Boeckx2025). Language catapults know-how copying onto center stage. It becomes a necessary cultural evolutionary requirement; not just for language learning itself, but also for the many downstream forms of culture that depend on language for their transmission and retention (Moore & Tennie, Reference Moore and Tennie2025).

So here is a hard question that CC’s account elides: how and why did robust, cross-domain know-how copying evolve in the human lineage? There is reason to think this is a deep puzzle, not least because it appears to have evolved only once on this planet. For example, there are costs to be paid for the cognitive machinery (to build up or even to exapt other parts; Enquist, Lind, & Ghirlanda, Reference Enquist, Lind and Ghirlanda2023), with little immediate direct pay-off – there is a start-up problem (Andersson & Tennie, Reference Andersson and Tennie2023). For miming (CC’s stage 4; see also Planer et al., Reference Planer, Bandini, Tennie, Raviv and Boeckx2025), these costs may be especially high, as miming requires solving the correspondence problem (i.e., the ability to match visual representations to corresponding motor representations; Nehaniv & Dautenhahn, Reference Nehaniv, Dautenhahn, Dautenhahn and Nehaniv2002). Some radical ways of thinking may be required to solve this problem and others like it. In this vein, Dennett (Reference Dennett2017) highlights the role of “digitization schemes” in making cultural copying of behavioural and artefact forms possible, where these schemes emerge bottom-up out of selfish competition among rival cultural variants (see Planer (Reference Planer2025) for discussion). Others have zoomed out and applied Maynard Smith and Szathmáry’s (Reference Maynard Smith and Szathmáry1995) transitional framework to this problem (e.g., Andersson & Tennie, Reference Andersson and Tennie2023). Still others point to special ecological circumstances in interaction with hominin evolution (Richerson & Boyd, Reference Richerson and Boyd2005). However one ultimately resolves this issue, the evolution of robust know-how copying looks like one of the crucial bottlenecks in CC’s story and in any story of human cultural evolution.

Where does all this leave CC’s model? If CC intended their model merely as a sequential how-possibly model, then that could have come through more clearly. Read this way, CC offer an interesting but mostly speculative (and as they themselves state, not fully novel) scenario about human socio-cultural evolution. If, instead, CC intended their model as a how-actually model, then it does not meet the relevant evidential standards. For it is silent on the evolutionary emergence of one of its most crucial enabling mechanisms – namely know-how copying. Their account then passes over what may be the most important step: the transition from ape-like cultures without robust know-how copying to human lineages in which such copying became sufficiently strong and widespread to underwrite language – but also complex ritual practices, technologies with deep-reaching histories, etc. These copying-dependent forms account for the bulk of the cumulative and collaborative culture CC aimed to explain.