Over the past decade, several accounts have been proposed for the core system that underlies infants’ ability to reason about social interactions (Baillargeon et al., Reference Baillargeon, Scott, He, Sloane, Setoh, Jin, Bian, Mikulincer and Shaver2015; Spelke, Reference Spelke2022; Ting et al., Reference Ting, Buyukozwer Dawkins, Stavans, Baillargeon and Decety2020; Woo, Tan, & Hamlin, Reference Woo, Tan and Hamlin2022). In Thomas’s account, this system is limited to the representation of interpersonal relationships. It allows infants to detect when such relationships exist, to categorize them into types (communal, asymmetrical, and reciprocal), and to compute their strength (a strong relationship is one to which individuals remain committed over time and across contexts). Here, we suggest that this relationship-based account does not do justice to the available evidence on early sociomoral reasoning. We first introduce an alternative, principle-based account and then discuss some of its implications for Thomas’s account.

According to the principle-based account (Ting et al., Reference Ting, Buyukozwer Dawkins, Stavans, Baillargeon and Decety2020), infants’ social world is not a uniform landscape of interpersonal relationships but a varied landscape structured by several types of social distinctions, each laden with rich implications for how interactions may unfold. Do individuals belong to the same or to different moral circles? Do they belong to the same or to different social groups within their moral circle? Do they possess the same status within their group, or are some leaders and others followers? Importantly, at least five normative principles guide early sociomoral reasoning. They include harm avoidance, which regulates interactions within moral circles (Jin & Baillargeon, Reference Jin and Baillargeon2025; Ting & Baillargeon, Reference Ting and Baillargeon2021); ingroup support, which regulates interactions within social groups (Bian & Baillargeon, Reference Bian and Baillargeon2022; Jin & Baillargeon, Reference Jin and Baillargeon2017; Ting, He, & Baillargeon, Reference Ting, He and Baillargeon2019); authority, which regulates interactions between leaders and followers (Margoni, Baillargeon, & Surian, Reference Margoni, Baillargeon and Surian2018; Margoni & Thomsen, Reference Margoni and Thomsen2024; Stavans & Baillargeon, Reference Stavans and Baillargeon2019); fairness, which regulates social exchanges (Bian, Sloane, & Baillargeon, Reference Bian, Sloane and Baillargeon2018; Buyukozer Dawkins, Sloane, & Baillargeon, Reference Buyukozer Dawkins, Sloane and Baillargeon2019; Sloane, Baillargeon, & Premack, Reference Sloane, Baillargeon and Premack2012); and reciprocity, which regulates partnerships or friendships (Jin et al., Reference Jin, Ting, He and Baillargeon2024). Together, these five principles and their attendant concepts form the “first draft” of human moral cognition (Graham et al., Reference Graham, Haidt, Koleva, Motyl, Iyer, Wojcik, Ditto, Devine and Plant2013). The first four principles are depersonalized and extend to all individuals who fall within the purview of their application: Individuals are expected to refrain from doing severe harm to any member of their moral circle, to help any ingroup member in need, to protect any of their followers from harm, and to distribute resources fairly between any recipients, whether or not they have formed interpersonal relationships with these other individuals. By contrast, the final principle, reciprocity, is personalized in that it regulates the development of interpersonal relationships between specific individuals.

The principle-based account overlaps with that of Thomas in that both encompass infants’ expectations about reciprocal relationships. Beyond this similarity, however, differences abound, giving rise to several issues.

A first issue is that it is unclear how Thomas can account for the wide range of sociomoral expectations infants demonstrate. For example, she writes that while studies using resource-distribution tasks (Buyukozer Dawkins et al., Reference Buyukozer Dawkins, Sloane and Baillargeon2019; Sloane et al., Reference Sloane, Baillargeon and Premack2012) show that “infants seem to expect that two individuals will be given the same amount, … they do not tell us whether seeing an equal resource distribution prompts infants to predict future interactions…. By contrast, a [reciprocal] relationship should be marked by an even balance over time and across contexts” (p. 22). We agree that there is no evidence that infants in these resource-distribution studies are reasoning about interpersonal relationships and that reciprocal relationships should be balanced – but none of this helps explain why infants would expect equal distributions in the first place. In the principle-based account, such expectations are supplied by infants’ sense of fairness, but there is little in Thomas’s account to explain them.

A similar point arises with respect to leader–follower tasks. By the second year of life, infants expect a leader to intervene when a follower steals from another follower (Stavans & Baillargeon, Reference Stavans and Baillargeon2019), and they expect followers to obey a leader’s instruction even in the leader’s absence (Margoni et al, Reference Margoni, Baillargeon and Surian2018; Margoni & Thomsen, Reference Margoni and Thomsen2024). Thomas suggests that infants are reasoning about asymmetrical relationships between high-ranking and low-ranking individuals, but why would infants expect protection from high-ranking individuals and obedience from low-ranking individuals when they first see them interact? In the principle-based account, such expectations are supplied by infants’ sense of authority.

A second issue is that in Thomas’s account, infants are incapable of representing social structures such as groups; these representations develop later in childhood “as children are exposed to more people in and outside of their social networks” (p. 30). This claim is surprising given the extensive evidence that infants can represent a wide variety of social groups, including animal puppets that differ in kind (Bian et al., Reference Bian, Sloane and Baillargeon2018; Rhodes et al., Reference Rhodes, Hetherington, Brink and Wellman2015; Ting & Baillargeon, Reference Ting and Baillargeon2021), geometric figures that differ in shape or color (Powell & Spelke, Reference Powell and Spelke2013; Pun, Birch, & Baron, Reference Pun, Birch and Baron2021), and individuals who speak different languages or perform different conventional actions (Liberman, Kinzler, & Woodward, Reference Liberman, Woodward and Kinzler2017, Reference Liberman, Kinzler and Woodward2018). Categorization of individuals into such groups, or even into minimal groups based on novel labels or outfits (Bian & Baillargeon, Reference Bian and Baillargeon2022; Jin & Baillargeon, Reference Jin and Baillargeon2017; Rhodes et al., Reference Rhodes, Hetherington, Brink and Wellman2015; Ting et al., Reference Ting, He and Baillargeon2019), elicits rich expectations of ingroup support.

Thomas argues (without marshalling evidence for her interpretation) that infants in these studies are representing communal relationships rather than groups; the latter characterization would fundamentally undermine her central claim that infants’ core system is limited to representing interpersonal relationships. Although we reject this interpretation, we believe that more should be done to understand how infants identify and reason about interpersonal and group interactions. For example, would infants expect any member of a group to help an ingroup victim – but also expect a member who is a friend of the victim to do so more promptly? At the same time, would infants expect a leader in a group to ignore interpersonal ties when regulating access to a group resource among followers? Such investigations would help deepen our understanding of how infants represent, integrate, and rank-order the varied elements of their social world.