We agree with Ashley Thomas regarding the importance of social relationships in human development, but we disagree with how such knowledge should be conceptualized and explained. Despite her title referring to “developmental origins,” Thomas provides no real developmental account. She asserts that infants possess abstract representations of social relationships, built on core knowledge of relationships that is either innate or early developing, in the sense that it is already present before the infant engages with the domain of relationships and thus does not depend on experience.

Elsewhere, we and others have critiqued claims that knowledge is representational and instead argue that knowledge develops through action and interaction (e.g., Carpendale & Lewis, Reference Carpendale and Lewis2021). A further problem is how such knowledge could be present before experience. There is a missing biological story here without references to indicate that this could be biologically plausible. Are we to assume that genes fill in the missing story in getting from the zygote to infants’ innate knowledge of relationships? Although decades ago this claim was accepted, geneticists explain that genes do not determine outcomes and are part of a complex developmental system. Whether genes are expressed or not depends on many internal and external factors (e.g., Meaney, Reference Meaney2010; Moore & Lewkowicz, Reference Moore and Lewkowicz2024). Furthermore, to get to the neural interconnectivity that is presumably required to support the claimed core knowledge of relationships requires experience to form and strengthen or prune away synapses (e.g., Stiles, Reference Stiles2017). Thomas also assumes a problematic conception of evolutionary theory that overlooks development (Moore & Lewkowicz, Reference Moore and Lewkowicz2024; Witherington, Lickliter, & Moore, Reference Witherington, Lickliter and Moore2023). In addition to these problems, in this commentary, we focus on difficulties concerning the evidence Thomas draws on.

Thomas argues that infants possess abstract representations of social relationships (e.g., communal sharing, authority ranking, equality matching) and can also determine the strength of these relationships. However, the research cited is based on looking-time methodology, which can be over-interpreted and potentially misleading (Houston-Price & Nakai, Reference Houston-Price and Nakai2004; Paulus, Reference Paulus2022). Looking-time studies often rely on the idea that if an infant looks longer at an event, this is due to a violation of expectations, presumably based on pre-existing social representations. For example, Powell and Spelke (Reference Powell and Spelke2013, Reference Powell and Spelke2016) reported that infants looked longer when members of synchronized groups imitated out-group members. Thomas interprets this as evidence of an understanding of group-based affiliation. Looking-time paradigms, at best, provide correlational data and do not offer explanatory power (Aslin, Reference Aslin2007; Haith, Reference Haith1998). Moreover, the interpretation of infant looking time can shift depending on the study. On the one hand, longer looking time is often taken as a sign of surprise at a violation of expected social behavior (e.g., Powell & Spelke, Reference Powell and Spelke2013). On the other hand, it can be interpreted as evidence of preference for an expected fair social outcome (e.g., Geraci & Surian, Reference Geraci and Surian2011). This inconsistency shows that looking time is susceptible to interpretive bias and lacks methodological clarity. In a meta-analysis of looking time studies, Hollich and colleagues (Reference Hollich, Newman and Jusczyk2000) concluded that infant looking behavior is influenced by multiple factors, such as perceptual salience and familiarity, yet these studies are used to claim infants’ conceptual understanding despite minimal and ambiguous indicators.

A further difficulty with claims of innate knowledge based on young infants’ looking time (e.g., Powell & Spelke, Reference Powell and Spelke2016) is that infants’ cortical control of their movements develops slowly over their first year of life. Yet these actions are assumed to demonstrate knowledge that is continuous with that of adults, which is located in the cortex (Blumberg & Adolph, Reference Blumberg and Adolph2023).

If infants possess core knowledge of relationships, this should be evident in how they act. Yet studies based on behavioral evidence suggest a different developmental story. Play is an early form of self-initiated social activity, and its structure can reflect a child’s capacity for relationship recognition. If infants understand communal or authority relationships early on, socially directed play should be evident early in life. But until about 12 months, play is predominantly solitary. Later, children begin playing side by side, but this parallel play is still not a shared or social activity. Around 24–30 months, children begin engaging in associative and cooperative play, with turn-taking, sharing roles, and explicitly referencing one another’s behavior (Howes & Matheson, Reference Howes and Matheson1992). Thus, real-world behaviors like play suggest that social relationships are not represented abstractly from birth, as core knowledge theory suggests, but gradually emerge through repeated interaction and shared activity.

Similarly, Thomas discusses infants’ expectations of sharing behaviors (e.g., food sharing as evidence of communal relationships), yet when considering actual sharing, the developmental trajectory suggests that these behaviors emerge gradually in conjunction with growing socio-emotional understanding (Brownell, Svetlova, & Nichols, Reference Brownell, Svetlova and Nichols2009), and sharing becomes relational as the child participates in social interactions and learns to recognize roles and intentions (Brownell, Reference Brownell2013; Carpendale et al., Reference Carpendale, Müller, Wallbridge, Broesch, Cameron-Faulkner and Ten Eycke2021). However, the implication of the author’s core knowledge theory is that infants already possess models like “communal sharing” or “authority ranking,” which attributes adult reasoning to infants who are still developing a sense of self and other. It also downplays the role of culture, where what counts as “sharing” or “deference” is shaped by social practices, not abstract categories (Gaskins & Miller, Reference Gaskins and Miller2009). Behavioral studies of play and sharing, both grounded in action, reveal a developmental progression rooted in social interaction, cultural norms, and emergent understanding. The evidence supports a view of infants as active participants, not passive holders of pre-existing knowledge.

In contrast to Thomas’s approach, we propose that infants are active participants, not passive holders of pre-existing relational templates. We argue that social understanding develops from participation in routines with others in the developmental systems elicited by infants’ biological embodiment, and through these interactions, infants gradually develop an understanding of social relations, not the other way around (Carpendale & Lewis, Reference Carpendale and Lewis2021). That is, rather than Thomas’s assumption that infants begin with cognition that makes relations possible, it is experience in relating to others in which communication and language emerge, which makes the development of cognition possible (Jopling, Reference Jopling and Neisser1993).