Thomas argues that an intuitive theory of relationships is built on core knowledge that enables infants to categorize relationships into three types based on Fiske’s models of social relations: communal sharing, authority ranking, and equality matching. We agree that infants are able to distinguish between relationships based on these models. However, we argue that equality matching should include both positive and negative forms of direct reciprocity, that is, responding in kind to actors who benefit and harm oneself. We review evolutionary and developmental evidence that suggests that positive and negative direct reciprocity may rely on distinct cognitive representations.

Evolutionary theorists have argued that both positive and negative reciprocity are important in managing social relationships, but that they serve different ultimate functions. Positive direct reciprocity supports mutually beneficial, long-term relationships with non-kin and may thus be a mechanism for the evolution of cooperation (Axelrod & Hamilton, Reference Axelrod and Hamilton1981; Henrich & Muthukrishna, Reference Henrich and Muthukrishna2021; Nowak, Reference Nowak2006; Trivers, Reference Trivers1971). At a social level, positive reciprocity may be a human universal that helps human social groups to function (Fiske, Reference Fiske1992; Gouldner, Reference Gouldner1960). Early theories of negative direct reciprocity proposed that it too could support the evolution or maintenance of cooperation (Boyd & Richerson, Reference Boyd and Richerson1992; Clutton-Brock & Parker, Reference Clutton-Brock and Parker1995). However, more recent theories and empirical work suggest that it serves a more self-protective function that supports “revenge” (McCullough, Kurzban & Tabak, Reference McCullough, Kurzban and Tabak2013) or competition to maintain or increase one’s status (Raihani & Bshary, Reference Raihani and Bshary2019). However, long bouts of retaliation are not beneficial to the individual or the group, and, therefore, humans and other species often opt to change partners after a tit-for-tat exchange or find a means to repair the relationship (McAuliffe, Marshall & McLaughlin, Reference McAuliffe, Marshall and McLaughlin2025). Although theoretical and empirical research in this area is ongoing, positive and negative direct reciprocity likely have different ultimate functions and possibly different phylogenetic histories. If this is true, then it is also likely that different representations subserve these two forms of reciprocity.

The developmental research reviewed by Thomas suggests that infants can track individuals and quantities, two important components of representing equality-matching relationships. Toddlers also distinguish between adults who intend to be helpers and those who do not intend to help, preferring to interact with the would-be helpers (Dunfield & Kuhlmeier, Reference Dunfield and Kuhlmeier2010). By the preschool years, children can engage more robustly in both positive and negative reciprocal interactions, and differences between the two forms of reciprocity emerge. Notably, when children’s options are limited to either a choice between two partners or a response to a single partner, even 3-year-olds show evidence of both positive and negative reciprocity (House et al., Reference House, Henrich, Sarnecka and Silk2013; Vaish, Carpenter, & Tomasello, Reference Vaish, Carpenter and Tomasello2010; Warneken & Tomasello, Reference Warneken and Tomasello2013; Wu & Gao, Reference Wu and Gao2018). However, when children can respond in kind to multiple agents, a clear asymmetry appears: children between 4 and 7 years of age will selectively punish someone who took from them but will not give back to a giver (Chernyak et al., Reference Chernyak, Leimgruber, Dunham, Hu and Blake2019). Instead, they give randomly to all the agents present, showing generalized positive reciprocity. Importantly, this is not due to selective memory – children recall perfectly well who took and who gave to them. Similar results have been found more recently but only among boys (Benozio, House, & Tomasello, Reference Benozio, House and Tomasello2024). In a follow-up experiment, children could quickly learn (or be reminded) to give back to the giver, but this suggests a stronger role for learning of positive than negative reciprocity (Chernyak et al., Reference Chernyak, Leimgruber, Dunham, Hu and Blake2019). Combined, this evidence suggests that there are different developmental pathways for positive and negative direct reciprocity. While basic components such as tracking individuals and benefits and rewards may be similar or the same, the cues that trigger a response, the emotions that motivate action, and the enculturation of contingent responses differ. More research needs to be done to determine the early ontogeny of positive and negative direct reciprocity, but we argue that distinct representations underlie these forms of reciprocity by the time children actively engage in ongoing relationships with peers.

In addition to our brief review of the evidence for distinct forms of equality matching, we suggest two empirical tests that can help to determine whether one or two representations are required. First, infant research can examine whether expectations for positive and negative reciprocal interactions are correlated, suggesting a single core representation of equality matching, or unrelated, suggesting separate core representations that underlie different forms of reciprocity. For example, studies could test whether infants who expect others to reciprocate help also expect them to retaliate against harm or whether these expectations emerge separately. Such findings could offer insights into whether infants’ representations of equality matching are organized under a single representational system. Second, longitudinal research could map individual differences in infants’ expectations of positive and negative direct reciprocity onto later first-person behaviors. Should infant expectations of positive direct reciprocity predict positive reciprocal behaviors at 4 years of age, but not negative reciprocal behaviors, or the opposite for negative direct reciprocity, this would signal distinct representations at work.

In summary, we argue that positive and negative direct reciprocity need to be considered as potentially distinct representations for equality matching. We are excited to engage in this discussion and by the empirical research this updated social relations model can inspire.