A computational account of multiple motives guiding context-dependent prosocial behavior

Prosocial behaviors play a pivotal role for human societies, shaping critical domains such as healthcare, education, taxation, and welfare. Despite the ubiquity of norms that prescribe prosocial actions, individuals do not adhere to them consistently but often behave selfishly, thereby harming the collective good. Interventions to promote prosociality would therefore be beneficial but are often ineffective because we lack a thorough understanding of the various motives that govern prosocial behavior across different contexts. Here we present a computational and experimental framework to identify motives behind individual prosocial choices and to characterize how these vary across contexts with changing social norms. Using a series of experiments in which 575 participants either judge the normative appropriateness of different prosocial actions or choose between prosocial and selfish actions themselves, we first show that while most individuals are consistent in their judgements about behavior appropriateness, the actual prosocial behavior varies strongly across people. We used computational decision models to quantify the conflicting motives underlying the prosocial judgements and decisions, combined with a clustering approach to characterize different types of individuals whose judgements and choices reflect different motivational profiles. We identified four such types: Unconditionally selfish participants never behave prosocially, Cost-sensitive participants behave selfishly when prosocial actions are costly, Efficiency-sensitive participants choose actions that maximize total wealth, and Harm-sensitive participants prioritize avoiding harming others. When these four types of participants were exposed to different social environments where norms were judged or followed more or less by the group, they responded in fundamentally different ways to this change in others’ behavior. Our approach helps us better understand the origins of heterogeneity in prosocial judgments and behaviors and may have implications for policy making and the design of behavioral interventions.