When people perform generous acts for each other, they can balance out relative benefits by alternating who is generous. When and why do they do this? Here we test the explanation that sequences of generosity regulate social relationships. We find that people selectively expect reciprocal generosity in equal (vs. hierarchical) relationships, use reciprocal generosity to infer the presence of an equal relationship, and critically expect that people reciprocate generosity in order to communicate a desire for a (more) equal relationship. In a formal planning model, reciprocal generosity can emerge from the value of communicating desired equality.

In two-person asymmetric coordination dilemmas, both people are better off if they coordinate, but one person benefits more than the other. When these interactions recur, people can form expectations to balance who is better off over time. What does it mean when asymmetric social interactions recur, and what can we learn from how people solve these dilemmas? We hypothesize that people expect social interactions to recur when two people are in a social relationship, and that knowing about the symmetry of the social relationship influences the stable solution to asymmetric coordination dilemmas over time. We report two experiments where participants read stories and answered questions about social interactions between two people. In Experiment 1, participants infer that two people are in a social relationship when there is a sequence of altruistic interactions between them, and specifically infer an asymmetric relationship when one person always performs the altruistic action, and a symmetric relationship when the two people alternate performing the altruistic action. In Experiment 2, participants equally expect alternating and repeating altruistic actions when the relationship is symmetric, but expect repeating actions (following a precedent) when the relationship is asymmetric. Our results suggest that people are able to use knowledge of relationships to generate shared expectations for coordinating on recurrent altruistic social interactions, and vice versa.

Language is simultaneously discrete (symbolic) and continuous (e.g., speech), and meaning-form associations are largely arbitrary. How and why did these properties emerge? To address this question, we study how people develop novel communication systems to refer to a continuous domain (color) using a continuous signal space (whistles). We conducted an experiment in which participants need to generalize from five learned signal-color pairings to a larger range of colors during an online communication game with another participant. We find that: (i) both discreteness and systematicity tend to emerge, such that signaling systems that reflect an underlying symbolic structure as well as systematic association with colors emerge more frequently; and (ii) these emergent systems achieve better communicative performance compared to emergent systems that exhibit only discreteness or only systematicity. These findings suggest a human cognitive bias not only toward symbolic communication, but also toward non-arbitrary meaning-form associations.