Abstract
Humans are often confronted with many faces at once, for example when interacting with a group of people. Compared to seeing a face by itself, processing groups of faces can evoke a different style of information processing, in which people efficiently encode the mean ("ensemble") information of the group. Gaining information about characteristics for an entire group at once might be useful, for example when extracting the average emotion of the crowd at a given moment. Previous research examining ensemble representations of facial expression has been limited because the "groups" contained different expressions of a single person, thus being highly artificial. In the present study, we demonstrate that ensemble representations for expressions can be extracted when faces in the group are of different people. Participants saw happy or angry groups of four face identities with different expression intensities. One individual's face was presented again, either with the exact same expression as seen in the group, or with its intensity shifted towards or away from the average expression of the group. Participants judged whether the expression intensity was the same as it had been in the group. They made errors that were skewed towards the mean expression of the group. These results suggest that people extract an ensemble expression for naturalistic groups of individuals, which biases their memory for individual face expressions. These results extend previous findings by demonstrating that observers encode high-level ensemble information, such as the current mood of a crowd, despite natural variation in a second characteristic, such as the identity of the individual faces in the group. Since crowds experienced outside the laboratory always consist of different people's faces, our findings provide important evidence for the idea that ensemble perception is involved in coding high-level information for groups of faces in everyday life.
Meeting abstract presented at VSS 2016