Abstract
Objects in a scene are often partially occluded without causing the viewer any problem: the occluded parts are apparently represented via amodel completion. To evaluate human ability to perceive and remember partially occluded pictures, we showed sequences of pictures using rapid serial visual presentation (RSVP) for durations of 53ms, 107ms, 213ms or 426ms/picture. The pictures were 1296 color photographs (size = 300×200 pixels) with widely varied content. Partially occluded pictures were covered by 90 red disks (size = 16×16 pixels each) at random positions, covering 30% of the picture area. Whether a given sequence was presented normally or partially occluded was counterbalanced between participants. In one condition participants attempted to detect a named target (e.g., “businessmen at table”) which was present on 2/3 of the trials; in a second condition, participants were given a yes-no memory test of one item; the correct answer was “yes” on 2/3 of the trials. In the detection condition the occluded pictures were almost as successfully detected as the normal pictures, even at the shortest duration. In contrast, in the memory condition performance was markedly worse with partially occluded pictures than normal pictures, particularly at the two shorter durations. The results suggest that amodal completion can occur rapidly when the observer has gist information about a picture, but occurs more slowly when the viewer has no prior conceptual information. The effects of inverting normal and partially occluded pictures will also be discussed.
Grant MH47432 from the National Institute of Mental Health