Poster at CogSci 2026

Abstract

People are sensitive to unequal outcomes. Recent studies show that people’s ability to learn action-reward mappings depends on thepercentage of the reward amount they receive (compared to another person), even when such information does not change which actions are the most rewarding for the learner. However, questions remain about the degree to which this sensitivity to inequity depends on the broader inequity context. We report results from two additional experiments in which participants had the opportunity to learn action-reward mappings, with only part of the reward going to the participant. Crucially, we manipulated whether participants encountered both advantageous and disadvantageous inequity, or only one type of inequity, within the same learning block. We found that the effect of inequity differed between these two designs, suggesting that inequity affects learning in at least partly a context-dependent way. We formalized context-dependency into a new computational model which outperformed its context-independent counterpart.