Does inversion disrupt averaging of emotional expressions?
Schimmel, Sarah Elyse
MetadataShow full item record
Previous research suggests subjects compute statistical summary representations (SSRs) to represent the average emotion of a set of faces without representing the individual faces comprising the set. Evidence for this claim relies on the finding that subjects can identify changes to average expression even when they cannot localize any face that changed. However, previous work in our lab suggests subjects must perceive changes to individual faces to infer changes in average expression. Thus, it is unclear whether subjects can compute average expression without encoding individual faces. Inverting faces impairs recognition of some individual facial expressions (sadness, disgust) more than others (fear, neutral). Subjects judged which of two consecutive arrays of upright or inverted faces displayed more sadness or disgust, or were more fearful or neutral. Finding that 1) inversion interferes with recognition of fear/neutral in ensembles of faces and 2) inversion has no effect on recognition of average sadness/disgust in ensembles would demonstrate a double dissociation between the processing of individual items and ensembles, and provide strong support for the claim that computing the average expression of a set of faces does not rely on individual exemplars. Results showed a double dissociation between the processing of individual items and ensembles. The inversion effect was statistically significant for items of disgust, but not for ensembles, and was not statistically significant for neutral items, but was for ensembles. The inversion effect was statistically significant for items and ensembles of both fear and sadness.