Abstract
The fusion of interocularly paired stimuli with offsets in position can yield depth information with high precision. Similarly, the interocularly unpaired points of a partially occluded object – seen in one eye but not the other – also provides important binocular depth cues (Nakayama and Shimojo, 1990). A stereo mechanism estimates the relative depth relations between objects using such unpaired points. It has been proposed that when interocularly unpaired points do not satisfy natural occlusion geometry, stable depth assignment fails, resulting in binocular rivalry (Hayashi et al., 2004). If binocular rivalry is indeed the erroneous output of occlusion relation estimation, then one might expect that the relative depths of rivaling stimuli could affect their visibility. Here we hypothesize that perceptual dominance of a stimulus during binocular rivalry is associated with that stimulus having a shallower depth. We tested this hypothesis by measuring whether a relatively shallower stimulus would be more dominant than a deeper ("occluded") stimulus in two experiments. A convex or concave contextual depth was first established by manipulating the positional disparity of two annuli. Rivaling stimuli (Experiment 1: red vs green patches; Experiment 2: left- vs right-tilted Gabor patches) were then presented for 500ms, 1000ms, or 1500ms between the two annuli, at the same distance from fixation. 6 subjects reported that the shallow stimulus was more frequently dominant for both color and Gabor patches in the convex context at 1000ms and 1500ms. No differences were found in the concave depth context. Our results provide partial support for the hypothesis that contextual depth can affect stimulus dominance during rivalry.
Meeting abstract presented at VSS 2014