One of the central aims of Veissière et al. is to explain how agents learn to respond to the norms and resources of their local cultural niche. Crucially, such an explanation has to be sensitive to the fact that among the most important components of this cultural niche are other people. Insofar, as our social success is about cultural niche construction, it has to involve the construction or shaping of people. Other people do not only scaffold our capacity to acquire norms and conventions, but they actively transform our behavioral dispositions and mental states by shaping them to respect the prevailing norms and stereotypes (De Bruin & Strijbos Reference De Bruin and Strijbos2020; Zawidzki Reference Zawidzki2013). Although this idea seems to be implicit in Veissière et al., the thinking through other minds (TTOM) hypothesis could shed more light on the dynamics by which perception, action, and niche construction led to the acquisition and production of cultural habits, and to the inference and learning about other minds (Fig. 4 in the target article). The following commentary is an invitation to Veissière et al. to expand on this. We offer three possible starting points.

First, Veissière et al. do not go into detail about the forms of enculturation that require us to make inferences about the mental states of other agents and those that do not. On the one hand, they note that some forms of enculturation do not involve mindreading but merely “following the ‘tracks’ laid down in local environments by others, or following the norms and rules presented through institutions, without engaging with others’ interiority” (sect. 2.3, para. 4). On the other hand, they also note that other forms require “inferences to how others think” and that “the process of TTOM consists in inferring the priors or expectations that guide the beliefs of another agent or group of agents” (sect. 3.6, para 5). In other words, some forms of enculturation require only superficial action copying, whereas others require inferences about mental states ranging from intentions to beliefs and desires as well as emotions. Given the variety of forms of enculturation and of mental states, it would be important to specify what forms of enculturation require what forms of theory of mind. A deeper explanation of these dynamics would make it easier to grasp the ontogenetic and phylogenetic implications of TTOM.

Second, and relatedly, it would be valuable if Veissière et al. could explain what sorts of theory of mind capacities they believe enculturation presupposes, and what sorts of theory of mind capacities enculturation makes possible. In some instances (e.g., sect. 3.6, para. 7) they appear to indicate quite broadly that theory of mind capacities are necessary for enculturation, but this is unlikely to be the case for more sophisticated forms of theory of mind, such as explicit reasoning about reasons, as in Mercier and Sperber's (Reference Mercier and Sperber2017) framework (which they cite repeatedly). For Mercier and Sperber, reasoning about reasons depends on theory of mind capacities that involve the attribution of propositional mental states with rich semantic contents. This would seem to presuppose enculturation rather than the other way around. Perhaps more basic forms of mindreading, such as level-1 perspective taking, shared attention, and action tracking are necessary for enculturation. It would be helpful to spell out reasons for thinking what forms of mindreading enable enculturation, and what forms are enabled by enculturation.

The two starting points identified above both concern the question what kind of inferences agents need to make, and which theory of mind capacities they need, in order for enculturation to take off (i.e., the top half of Fig. 4 in the target article). A third starting point would be to elucidate how enculturation contributes to the shaping of agents (i.e., the bottom half of the figure). One possibility, recently put forward by Hohwy and Michael (Reference Hohwy, Michael, de Vignemont and Alsmith2017; cf. also Michael Reference Michael and Munoz-Suárez2015), is that infants and young children progressively refine their agent-models through interaction with others, and also increasingly conform to those models themselves. More concretely, infants and young children apply agent-models to others during the course of development, and use these agent-models to guide imitation and other forms of cultural learning. A result of this is that it becomes increasingly feasible for others in their culture to model them as well. In other words, agents’ models of themselves (their self-models) and their actual selves are fitted together as a natural consequence of modeling and interacting with each other. From the perspective of the prediction error minimization framework, this appears as a form of active inference: infants and young children shape their selves progressively to match the agent models that they have been using to interpret others.

We hope that careful consideration of these three points will help in clarifying the relationship among culture, theory of mind, and predictive coding, and that it will stimulate progress in explaining the dynamics by which minds shape, and are shaped by, other minds through the complex process of enculturation.