Abstract

ABSTRACT Human faces are located easily in peripheral vision, but it is unclear which facial information drives this process. We employed a gaze-contingent paradigm to investigate face detection in the periphery and at fixation. Participants searched complex scenes for faces. As they made a saccade towards a peripherally-presented face, it sometimes changed in appearance before reaching foveal view. Changes in facial identity (Experiment 1) or the addition of internal features (e.g., eyes, nose, mouth) to an initially-featureless face (Experiment 2) did not affect face detection, yet performance was hindered by the removal of internal features before fixation occurred. While the presence of internal features in peripheral faces could be perceived, the organization of these features was difficult to extract (i.e., when internal features were rotated 90 degrees) (Experiments 3 and 4). These findings point to a two-stage model of face detection. In the initial stage of encountering faces in the periphery, where visual acuity is low, observers are insensitive to internal facial features and their configuration. Once a face is located and can be viewed in high-acuity regions of central vision, a second stage may use detail from internal features to confirm that a face has been detected.