Abstract
A remarkable ability of our visual system is to perceive occluded stimuli as whole and complete immediately and effortlessly, which is known as amodal completion. In this study, we combined psychophysics and functional magnetic resonance imaging (fMRI) to investigate how amodal completion is implemented in the visual hierarchy. Identical face fragments from 5 deg face side views (either left tilted or right tilted) were presented stereoscopically either behind or in front of an occluder. When the fragments were behind the occluder, they were amodally completed and perceived to be a coherent face. When the same fragments were in front of the occluder, they were perceived as disjoint face patches. These two kinds of stimuli were presented with various durations (50-350ms) and followed by a mask. Subjects were asked to judge the orientation of face views. The time course of face amodal completion was assessed using a performance-based measure (Murray et al., 2001). We found that the completion took place between 100 and 300 ms. Meanwhile, subjects were scanned to measure their cortical responses to the stimuli throughout the visual hierarchy. Early visual areas (V1 and V2) showed a higher response to the perceived fragmented faces than to the perceived completed faces when the stimuli were presented for 50 and 150 ms. However, higher visual areas (OFA and FFA) showed the opposite pattern when the stimuli were presented for 250 and 350 ms. In a second experiment, subjects were asked to do an RSVP task at fixation point and ignore the face stimuli. We found that this attentional manipulation largely abolished all the differential responses. These findings suggest that face amodal completion results from early suppression in early visual areas and late enhancement in higher visual areas and attention plays a critical role in these neural events.
This work is supported by the National Natural Science Foundation of China (Project 30870762, 90920012 and 30925014).