Tomasello proposes that the concept of obligation emerges in a manner that I will call, “from the inside-out.” A sense of obligation emerges first in children's experiences of shared agency – where they feel a sense of obligation to their collaborative partners – and is then generalized outward to develop a more abstract sense of obligation among group members. On this account, children begin to feel obligated to their group members and to see their group members as obligated to them – simply because they are in the same pack – by around their third birthday.

Here I consider that the concept of obligation might also (or alternately) develop “from the outside-in.” That is, that children have abstract intuitions about social structure that develop separately from their own interpersonal interactions.

Already by age 3, and perhaps earlier in infancy (Ting et al. Reference Ting, Dawkins, Stavans, Baillargeon and Decety2019), children have an abstract understanding that social groups specify who is obligated to one another and who is not (Chalik & Rhodes, Reference Chalik, Rhodes and Bensonforthcoming). We know that this understanding is already abstract – not tied to children's own perspective – because children rely on it to guide their social understanding even when they are not members of the relevant groups.

To illustrate, in Rhodes (Reference Rhodes2012; also described in the target article, sect. 2.2.2.3, para. 2) children were introduced to two made-up social groups (called “flurps” and “zazes”); children had never heard of these groups before and were not assigned to be members of either group. In these studies, by age 3, children predicted that a flurp, for example, would harm a zaz, rather than another flurp. In Rhodes (Reference Rhodes2012), I speculated that these findings had to do with a sense of obligation – that children expected the flurp to avoid harming another flurp because, as group members, they were obligated to protect one another. Indeed, in Rhodes and Chalik (Reference Rhodes and Chalik2013), children judged flurps to hold intrinsic obligations to one another but not to zazes.

If the concept of obligation that drove children's responses in these studies developed from the inside-out, then we might expect children to solve these problems by placing themselves in the shoes of a flurp. That is, they might identify with the agent, and think of the other flurps as their in-group members. We might expect this to be particularly so around age 3, when children are just beginning to generalize from their own experiences outward. But this is not how children thought about these problems. When children are themselves put into made-up groups during experiments (e.g., Dunham et al. Reference Dunham, Baron and Carey2011), the first (and sometimes only) thing that happens is that they feel and respond more positively toward members of their own group (for review, see Dunham Reference Dunham2018). But children did not show any evidence of in-group positivity in these studies – when asked who a flurp would do something nice for, they responded at chance. Children's responses in these third party scenarios – where they reliably expected flurps to harm zazes but to be equally nice to everyone (Chalik & Rhodes Reference Chalik and Rhodes2018; Rhodes Reference Rhodes2012) – is exactly opposite to how children respond when they themselves are placed in made-up groups (where they are nicer to in-group members but not particularly mean to out-group members (Buttelmann & Boehm Reference Buttelmann and Boehm2014; Dunham et al. Reference Dunham, Baron and Carey2011). Thus, when children predicted that flurps would harm zazes, they relied on abstract intuitions about how group memberships specify social obligations, rather than their own first-person perspective.

From where would children get such an early developing abstract understanding of obligation if not by generalizing their own social experiences? One possibility is that children's concept of obligation develops as a specific instantiation of a more general conceptual bias to treat categories as constraining what their members are supposed to do. This is a domain-general feature of early concepts. For instance, children do not just think that cheetahs usually do run fast, they think they are supposed to (Haward et al. Reference Haward, Wagner, Carey and Prasada2018; Foster-Hanson & Rhodes Reference Foster-Hanson and Rhodes2019). Children think there is something wrong with a category member who does not follow the norms of their group (Roberts et al. Reference Roberts, Gelman and Ho2017) – and they hold this intuition just as strongly for categories of animals as for categories of people (Foster-Hanson et al. Reference Foster-Hanson, Roberts, Gelman and Rhodes2018). Children also think that the clearest and most informative example of an animal category is the one that best illustrates these prescribed properties, even if such an instance is rare (e.g., the very fastest cheetah in the world; Foster-Hanson & Rhodes Reference Foster-Hanson and Rhodes2019; Foster-Hanson et al Reference Foster-Hanson, Moty, Cardarelli, Ocampo and Rhodes2019). As another example, children also think that artifacts are supposed to fulfill their intended functions, and again, that there is something defective with one that does not (Diesendruck et al. Reference Diesendruck, Markson and Bloom2003).

Thus, by early childhood, children have abstract, domain-general intuitions that categories prescribe how their members are supposed to be. In the social domain, this is instantiated as an expectation that categories constrain how people are supposed to treat one another, whereas in other domains it is instantiated in other ways (e.g., regarding how animals are supposed to get food, avoid predators, and so on). From this perspective, the general processes that drive conceptual development – in which children actively build hierarchical representations to make sense of various domains of experience (Gopnik & Wellman Reference Gopnik and Wellman2012) – can lead children to develop an abstract understanding that members of a group hold special obligations to one another, separately from their own experiences with shared or collective agency.

Of course, both types of developmental processes could simultaneously be at play – children could both develop a sense of their own obligations via their experiences with shared agency and begin to generalize them out, while at the same time, the mechanisms that underlie conceptual development lead them to construct an abstract understanding of the normative implications of group membership. A full developmental theory of how this critical concept arises would need to examine all of these processes and how they might relate to one another across early childhood development.