Overgeneralizing Emotions: Facial Width-To-Height Revisited

Abstract

The study investigated facial attribution bias. Instead of asking participants to attribute character to faces, as usually done, we did the opposite: Participants were asked to generate the faces of specified characters, namely an aggressive/dominant male or the opposite (peaceful-submissive male). Participants used three methods: They generated free drawings, selected features from an assembly-kit, or edited facial photographs using PC software. We investigated facial width-to-height ratio in these generated portraits. We found that participants did not model static facial width to express character; instead they modelled expressed emotions, anger in particular. This reduced facial height, thereby increasing fWHR.