Abstract
Binocular disparity extraction is well understood for simple stimuli defining single objects or surfaces in depth. When scenes contain a complex pattern of elements, scattered at different depths through a volume, the disparity extraction process, and the combination of disparities for subsequent depth perception, is much less well understood. For example, it is not clear whether the disparity of each element is independently represented. Here we explored how disparity signals are combined to judge volume (or thickness) defined by binocular disparity. We tested two combination rules: (1) the difference between the means of crossed and uncrossed disparities; (2) the variance of the whole depth distribution. Stimuli consisted of lines elements with random orientation, x, and y positions. Elements were located on one of 4 depth planes. In the narrow condition these were located at +/- 5.68 and 8.52 min arc disparity, in the wide, at +/- 9.94 and 4.26 min arc disparity. Mean crossed and uncrossed disparities were the same for these two conditions. Observers were asked to judge the thickness of the volume, compared to a standard stimulus where elements were distributed across a pair of planes in depth (14.2 min arc disparity apart). If rule (1) were used to combine depths for volume perception, we expected no differences between conditions. If rule (2) were used, we expected the wide condition to be perceived as having a thicker depth volume. Or (3), observers might separately represent each plane and make judgements based on the outer planes only. Results were consistent with observers using rule 2. This suggests that the variance of the whole depth distribution is used to obtain an estimate of depth volume but that we do not have independent access to representations of the 4 separate planes.
Meeting abstract presented at VSS 2014