Abstract
When presented with sets of similar stimuli, the visual system extracts a summary representation of the entire set. This strategy is employed not only for basic features of simple stimuli, but also for high-level object properties such as facial expression. Summary representation allows efficient access to information about the set as a whole, but how does it affect search for an individual within the set? We tested whether observers more readily detected a face that deviated substantially from the mean expression of a set of faces, versus one whose expression was close to the mean. Participants searched for a specific identity within sets of faces. The face sets could be either homogeneous or heterogeneous in expression. Critically, we varied the distance of the expression of the target face from the mean expression of the set. Search times were reduced when the expression of the target face was far from the mean expression of the set. Thus, deviation from the mean, in this case along an irrelevant dimension (expression), influenced search efficiency for the identity. These results demonstrate that the summary representation of a set property significantly impacts target detection.