A computational model for the impact of altruism on happiness

The pursuit of well-being is fundamental to human experience. Helping others is widely believed to be an important route to happiness, but the relevance of individual social preferences remains unclear. Happiness is also thought to depend on past prediction errors, whether events are better than expected. We conducted two experiments (total N = 235) in which participants took altruistic actions (either giving to others or charity) and selfish actions (keeping for themselves) and periodically reported their happiness. The outcomes of only some actions were realized, leading to prediction errors. In both experiments, greater generosity was not associated with greater happiness. However, the outcomes of altruistic actions had a greater hedonic impact in more generous individuals. Using computational modelling, we found that momentary happiness was predicted by the history of prediction errors for both altruistic and selfish actions with greater weight for actions aligned with individual social preferences, whether individuals were overall more altruistic or selfish. Notably, outcome size did not affect happiness, suggesting that multiple small donations should increase happiness more than a single equivalent donation. We confirmed this prediction in a third experiment (N = 198): giving to three charities with separate actions increased happiness more than giving the same amount with a single action. Our results explain why generosity is not always associated with greater happiness and offer new strategies to maximize happiness through giving.