Steels & Belpaeme (S&B) take technical and conceptual shortcuts that have significant negative consequences on simulation implementation and agent behavior. Justifiably, their model represents a proof of concept of the role of culturalism in category formation; nevertheless, the absence of detailed information concerning the embodiment of agents and the superficial implementation of learning approaches make their results less relevant than they could be.

Christiansen & Chater (C&C) focus solely on general-purpose cognitive processes in their elegant conceptualization of language evolution. However, numerous developmental facts attested in L1 acquisition confound C&C's subsequent claim that the logical problem of language acquisition now plausibly recapitulates that of language evolution. I argue that language acquisition should be viewed instead as a multi-layered construction involving the interplay of general and domain-specific learning mechanisms.