Impersonal prosociality is considered a cornerstone of a thriving civic society, well-functioning institutions, and a growing economy. Previous research has documented substantial cross-societal variation in prosociality using tasks such as dictator games, where individuals allocate money between themselves and others. In such tasks, individuals typically receive full information about how decisions impact others and make decisions privately. Here, we propose that different societies rely on distinct mechanisms—guilt and internalized norms versus shame and external pressures—to support prosociality. In 20 culturally diverse countries, we will administer dictator games and experimentally induce guilt, by varying information about the consequences of participants’ decisions, and shame, by varying observability. Additionally, we will measure guilt- and shame-proneness at the individual and societal level. We will test the hypotheses that activating guilt (by varying information) more strongly increases prosociality among guilt-prone individuals and societies, whereas activating shame (by varying observability) more strongly increases prosociality among shame-prone individuals and societies.