Precise Ensemble Face Representation Given Incomplete Visual Input

摘要

Humans can recognize faces in the presence of environmental noise. Here, we explore whether ensemble perception of faces is similarly robust. Is summary statistical information available from crowds of faces that are visually incomplete? Observers viewed sets of faces varying in identity or expression and adjusted a test face to match the perceived average. In one condition, faces amodally completed behind horizontal bars. In another condition, identical facial information was presented, but in the foreground (i.e., face parts appeared on fragmented strips in front of a background). Baseline performance was determined by performance on sets of fully visible faces. The results revealed that the ensemble representation of amodally completing sets was significantly better than the fragmented sets and marginally worse than in the fully visible condition. These results suggest that some ensemble information is available given limited visual input and supports a growing body of work suggesting that ensembles may be represented in the absence of complete visual information.