Pietraszewski's target article provides a fascinating new approach to describing what the mind represents as a “social group” and how such representations are formed. Essentially, the present model proposes that “group membership is a relational property (…), who ‘belongs’ to what group is borne out of a calculation of the relative relationships among the agents involved” (sect. 8.2, para. 2). While welcoming this new approach, in this commentary, we would like to highlight a few phenomena observed about group psychology (especially, in the field of developmental psychology) for which it is not quite clear how the current model would account. We suggest that either the theory should be extended so that it provides a framework for interpreting these phenomena as well or the limitation of scope should be made more explicit.

Specifically, representations generally appear to be more elaborate than merely involving information about specific roles in certain social interactions: Ample evidence in psychology suggests that these representations of social groups are conceptually rich not only in the minds of adults, but in those of very young children as well (Liberman, Woodward, & Kinzler, Reference Liberman, Woodward and Kinzler2017). In fact, perceived group membership allows even young children to draw inferences not only about how people will relate to each other (Rhodes & Chalik, Reference Rhodes and Chalik2013), but also about, for example, what knowledge (e.g., Liberman, Gerdin, Kinzler, & Shaw, Reference Liberman, Gerdin, Kinzler and Shaw2020; Soley, Reference Soley2019) or preferences (Shutts, Kinzler, McKee, & Spelke, Reference Shutts, Kinzler, McKee and Spelke2009) they possess. Moreover, even young children are selective in what kind of inferences they draw from different group memberships. For example, they expect friends to share knowledge of personal affairs, while they expect members of a cultural group to share knowledge of cultural norms (e.g., Liberman et al., Reference Liberman, Gerdin, Kinzler and Shaw2020). Importantly, these inferences seem to arise as early as 12 months of age (Shutts et al., Reference Shutts, Kinzler, McKee and Spelke2009) and based on cues that are not presented to children in the context of social relations (e.g., language). These examples illustrate that from very early on, humans apply information arising from perceived group membership for predicting the behavior of others not specifically in intergroup conflict situations, but rather in preparation to engage in – mostly collaborative – interactions with fellow ingroups. Although it is acknowledged in the paper that the relational model of groups should be extended to contexts other than conflict, evidence in developmental psychology suggests that a sensitivity to behavioral cues that are not necessarily manifested in interpersonal contexts at all (such as similarity in linguistic behavior) may precede the emergence of a sensitivity to relational cues. Although Pietraszewski's model would allow linguistic cues to be considered “ancillary” attributes, it is not evident how 12-month-old infants would come to encode them as such.

Thus, although the model presented in Pietraszewski's target article is promising in describing how the human mind represents social groups and how these representations may be used to predict behavior in specific interpersonal situations, it is not as clear how it can account for inferences that are not pertinent to how people will interact with each other. It is especially challenging for this model to give an account for such group membership-based inferences where the inferences are drawn from the assumption that the individuals are members of the same group, and not from a perceived relational contrast in interactions (e.g., generalizing food preferences within, but not between groups or expecting a person to understand a specific language based on knowledge of their group membership).

Pietraszewski distinguishes the cognitive processes taking place during these computations from simple “categorization” as he claims that the latter only speaks to the containment issues (sect. 8.1). However, category representations are, in fact, used to store a large body of knowledge of the given class that can later be used to make predictions of different properties of the tokens. Although this is generally true for any ontological field, one robust phenomenon that seems to be specific to social category representations is that the inferences and generalizations that are drawn from them tend to run wild, resulting in robust stereotypes. It is yet unclear to us, how such group representations allow for this phenomenon to occur. Although the paper gives a very elaborate description of how these group representations are formed, what cues (direct or ancillary) may be taken as input for the computational process, it is less clear how they would feed into further inferences. Even if the model's main goal is to explain the formation of such representations, we believe that these questions should be touched upon, because – especially taking an evolutionary approach – function and computational process cannot be perfectly separated.

We do not claim that these questions would invalidate the presented model, rather, we would like to highlight the need to consider how the abovementioned phenomena relate to the computation model described here. It is possible that “group” representations only refer to those collectives where members are likely to interact with each other in some way and thus, the interaction pattern can predict membership and vice versa. In this case, it would be necessary to consider how these group representations are different from other social “category” representations where intergroup conflict (or even other types of social interactions, such as reciprocity) are not at the core of the category (e.g., “women”). In the presented model, for example, features typical of the collective of “women” would be considered ancillary attributes, whereas we would suggest that conflicts arising between the sexes are on most occasions consequences and not essential features of belonging to this specific social group (or “category”). Possibly, an elaborate differentiation between “social group” and “social category” representations would help to disambiguate the issue at hand; however, as of yet, this is unfortunately missing from the literature.