Abstract
In the face processing literature, the parts/wholes, inversion and composite paradigms are regarded as the gold standards of holistic face processing. Although converging results from these paradigms support the holistic perception of faces, surprisingly, little is known about the precise onset of this process. To address this issue, in Experiment 1, participants viewed a study face for brief exposures of 17, 50, 250 and 500 ms, followed by a diffeomorphic mask that eliminates the low-level properties (e.g., contrast, luminance) of the face stimulus (Stojanoski & Cusack, 2014). The parts/wholes paradigm was employed to test recognition of the eyes (or mouth) presented in isolation or in the whole face. The main finding was that 17 ms and 50 ms of encoding time was sufficient to produce holistic recognition of the eye and mouth parts, respectively. In Experiment 2, study faces were randomly presented in either their upright or inverted orientation for 17, 50, 250, 500 ms. Replicating the findings of Experiment 1, the eyes and mouth were better recognized in the whole face than in isolation in presentation times as brief as 17 ms. Critically, the whole face advantage was abolished when the same faces were presented and tested in their inverted orientations. In Experiment 3, we adopted a composite paradigm where participants viewed a study face for 17, 50, 150 ms and asked to make a same/different judgement to the cued top (or bottom) half of a test face. In exposure durations of 17 ms, judgement was better in the congruent condition than in the incongruent condition (i.e., the congruency effect). The congruency effect was stronger when the cued and uncued halves were aligned than misaligned. Thus, results from the three gold standard tests of holistic processing show that 17 ms is sufficient to elicit the holistic perception of faces.
Acknowledgement: Natural Sciences and Engineering Research Council of Canada