Abstract
The combination of ordinal and metric depth cues poses problems for standard models of cue integration, because the two cue types do not have comparable units. When occlusion (ordinal) and binocular disparity (metric) cues conflict, one cue is sometimes ignored; other times, surfaces are perceived as transparent, broken, or bending to reconcile occlusion and disparity signals (Howard 2012). What visual features determine how occlusion and disparity combine remains unclear. It might be expected that sensory occlusion cues, such as T-junctions, are necessary for occlusion to influence stereoscopic depth perception. Here we show, instead, that completely illusory occlusion can trump binocular disparity. When a monocularly presented bar is filled in through the blind spot, it can be perceived as occluding a binocularly presented bar in the same visual region. Observers viewed a cross of textured bars of different colors through a stereoscope. One bar was presented monocularly, reaching through the blind spot; the other bar was presented binocularly. Observers judged the perceived depth of the binocular bar by comparing it to a single binocular bar with variable disparity in a 2IFC task. When the monocular bar was seen to occlude the binocular bar, depth estimates for the binocular bar were farther than in a baseline condition, where a binocular bar was presented in isolation. In contrast, when the binocular bar was perceived to be in front, it appeared as closer. In summary, a task-irrelevant illusory occluder overrides binocular disparity cues. Although the occluding surface is illusory, it is not subject to bending or other surface deformations. Instead, the perceived global depth ordering, which is not based on retinal cues, sets boundaries for the interpretation of metric stereoscopic depth estimates. Interestingly, filling-in seems to produce strong, opaque surface representations that provide functional contributions to other perceptual processes, such as depth estimation.
Meeting abstract presented at VSS 2016