Empathy and Prosocial Learning: Evidence for Correlation but Not for Causation

Empathic concern promotes prosocial behavior, but the effect of empathy is not the same for everyone. Scholars have extensively mapped the situational and personality factors that modulate the empathy-helping relationship, but only a few studies have examined the cognitive processes that link empathy with helping. Lockwood et al. (2016) found that individual differences in online simulation (the deliberate intention to imagine what another person is feeling) were associated with individual differences in the rates at which people learned how to benefit an anonymous other person in a two-armed bandit task (which we term prosocial learning). Across three experiments (N = 1,188), we critically tested and extended this finding by asking participants to learn on behalf of real people in need. Causal manipulations of empathy and social evaluation had no effect upon prosocial learning, and we could not replicate the finding that online simulation was associated with learning. Although each individual experiment provided only mixed evidence for the hypothesis that empathic concern is associated with prosocial learning, a mega-analysis that included data from all three experiments revealed that both state and trait empathic concern were weakly (but significantly) positively associated with prosocial learning. We discuss our results in light of problems with the methods that researchers commonly use to test the association between performance on prosocial learning tasks and individual differences in empathy.