Observing Others Give & Take: A Computational Account of Bystanders’ Feelings and Actions

Abstract

Social interactions influence people’s feelings and behaviour. This is true not only when people are directly involved, but also when observing others interact. Little is known, however, about how others’ interactions impact bystanders. Here, we developed a computational model that relates others’ (un)selfish acts to observers’ emotional reactions (which we call a ‘feelings function’) and punishment decisions (‘punishment function’). Our models enabled us to quantify the impact of two social values: ‘selfishness aversion’ and ‘inequity aversion’ on feelings and punishment. The results revealed diminishing sensitivity to actions that violated these social norms. That is, small violations from equality and generosity had a disproportionately large impact on feelings and punishment. We then used our ‘feelings function’ to predict observers’ feelings on out-of-sample trials and found that those estimated feelings were strongly correlated with observers’ punishment decisions. This suggests that observers were often acting in accordance with their affective responses to selfishness and inequity when punishing. We further show that when affective responses indeed align with punishment decisions, participants feel better about their decisions. The study characterizes computational rules by which social interactions between other people are transformed into bystanders’ reactions.