Decades of research across a diversity of academic disciplines has attempted to characterize and explain human morality. Resource-rational contractualism -- or the triple-theory of moral cognition -- provides a compelling integration of existing theoretical perspectives. Keeping in mind that the ultimate function of morality is to structure interdependent choice, Levine et al. propose that the proximate mechanisms may involve the creation and application of contracts between people. Specifically, they highlight two axes of resource rationality, emphasizing (i) the process of agreement (i.e., how we find agreement), and (ii) the terms of the agreement (i.e., what the agreement covers).

Yet, we propose that an axis is missing from their model, one that considers the partners in the agreement (i.e., who one should enter into agreements with). We agree that one of the primary challenges of coordinated behavior involves knowing how to act; however, knowing who to coordinate with poses similarly consequential challenges. That is, people vary in the extent to which they act in accordance with shared moral principles (e.g., Baumard, André, & Sperber, Reference Baumard, André and Sperber2013). Notably, free-riders, or those who accept others’ positive social behaviors without ever bearing the costs, pose well-documented challenges to the maintenance of cooperative networks (e.g., Trivers, Reference Trivers1971). Consequently, considerable attention has been paid to identifying how individuals react to others’ moral violations and how individuals choose who to interact with. We believe that the robust and growing body of research examining how children and adults identify and preferentially interact with good social partners provides an important complement to Levine et al.’s proposal.

Social evaluation and partner choice allow for the identification of actors who are more likely to form effective, sustainable contracts in the first place. Models of the cognitive processes that underlie partner choice have been developed from developmental and comparative frameworks (e.g., Barclay, Reference Barclay2004; Bshary & Noe, Reference Bshary, Noë and Hammerstein2003; Kuhlmeier et al., Reference Kuhlmeier, Karasewich, Dunfield and Decety2020). These processes, like those proposed by Levine et al., can be “model-based” or “model-free”, and provide for efficient and flexible identification of prosocial partners.

Starting in infancy and extending across the lifespan, people prefer prosocial over anti-social others and differentially invest in prosocial others (Barclay & Willer, Reference Barclay and Willer2007; Hamlin, Wynn, & Bloom, Reference Hamlin, Wynn and Bloom2007; Margoni & Surian, Reference Margoni and Surian2018; Rand, Arbesman, & Christakis, Reference Rand, Arbesman and Christakis2011). This tendency is not limited to the moral domain but is also evident in children’s reasoning about accurate and inaccurate informants (Dunfield, Kuhlmeier, & Murphy, Reference Dunfield, Kuhlmeier and Murphy2013), those who uphold or violate conventional rules (Riggs & Kalish, Reference Riggs and Kalish2016), and those who support or subvert gender norms (Riggs, Kinard, & Long, Reference Riggs, Kinard and Long2023). Importantly, children and adults use their knowledge of an individual’s behavior in one domain to make decisions about how to interact with them in another domain (Dunfield et al., Reference Dunfield, Isler, Chang, Terrizzi and Beier2023), suggesting that people are highly attuned to cues signaling their partners’ likelihood of upholding a contract. Even without observing others’ behavior, people engage in social evaluation and partner choice. For example, people prefer others who are familiar, socially engaged (Barragan & Dweck, Reference Barragan and Dweck2014; Carpenter, Uebel, & Tomasello, Reference Carpenter, Uebel and Tomasello2013), or signal in-group membership (Hilton, O’Neill, & Kuhlmeier, Reference Hilton, O’Neill and Kuhlmeier2021), possibly suggesting that people are attuned to others who are likely to share their expectations of an appropriate contract.

Though there is considerable debate regarding the mechanisms that underlie social evaluations and selective prosociality, we summarize our model here, as we too have focused on the proximate, cognitive mechanisms. Specifically, Kuhlmeier and colleagues (see Kuhlmeier, Karasewich, & Dunfield, Reference Kuhlmeier, Karasewich, Dunfield and Decety2020, for a review) have identified and described three “cognitive routes” that support partner choice: affective valence matching, affective predictions, and dispositional evaluations.

Under the affective valence matching route, the affective valence associated with another individual’s behavior (or physical characteristics) becomes associated with the individual (e.g., an individual who enables one’s goals is, by association, a positive person), Yet, strict valence matching involves immediate behavioral matches and does not support predictions of future behaviours, likely limiting its applicability in the maintenance of contracts. Affective predictions are similar to affective valence matching, but instead of relying on immediate matches, this cognitive route includes predictions that the those who have acted positively in the past will continue to do so in the future. Both of these routes may support a model-free way to generalize from past interactions and enter into moral contracts with those who are likely to uphold them.

Dispositional evaluations allow individuals to identify good partners for moral agreements by making relatively specific predictions about what type of person an individual is based on their past behaviours. In this route, knowledge about an individual’s past acts allows one to make an attribution about the internal states or tendencies (i.e., dispositions) that supported that behavior. These dispositions can then be used to support stronger and more nuanced expectations about how that individual will behave in the future. Moreover, these dispositional evaluations would support model-based simulations of another person’s likelihood of upholding a moral contract.

Importantly we do not suggest that once a person is cognitively capable of making disposition attributions, that person never uses either of the other two routes; both developmental and situational factors would determine which route is taken. Even adults may rely on the second route when limited information is available, when under cognitive load, or when quick decision-making is required. But, we propose that the social evaluation and partner choice that these decision-making processes result in is an integral step to ensure the efficiently and efficacy of reaching a mutually agreed-upon contract.