Abstract
The most prominent social psychological account of dehumanization, infrahumanization theory, argues outgroups are dehumanized to the extent they are denied uniquely human emotions. Recent critiques have identified a confound in previous research whereby uniquely human emotions used as stimuli tend to be more prosocial than the emotions shared with other species. Consequently, apparent evidence for subtle dehumanization may be better explained by intergroup preference. While there is growing appreciation that some studies are confounded this way, the extent of this problem has proved controversial. To gauge prevalence of the confound, we systematically reviewed the infrahumanization literature and extracted all emotion terms used. Participants rated the extent to which these emotions appeared unique to humans and prosocial. From these data, we calculated the percentage of studies that confound humanness with prosociality. In the 10 most cited papers, 95.5% of reported studies were confounded in the predicted direction. Across all 152 studies, 79.6% showed the same issue. These findings point to a pervasive methodological problem, impacting our understanding of discrimination and the reliability of social psychological data. To facilitate progress moving forward, we introduce a freely accessible tool, powered by our emotion rating database, to help researchers generate rigorously controlled stimulus sets.