Abstract
During natural viewing in a 3D scene, there are situations where one eye sees more of a partially occluded object relative to the other eye’s view, producing interocular conflict. Rather than experiencing binocular rivalry (BR), as would typically happen in 2D dichoptic displays, an observer sees a partially occluded object situated further in depth relative to a nearer one. This raises an interesting question: is the key to resolving potential conflict critically dependent on partial occlusion (which implies depth differences from an occluding object) or does it depend on depth separation per se (regardless of occlusion). To evaluate these alternatives, we varied the virtual depth between rivalling stimuli (drifting or stationary gratings) and presented them centrally in a cue-rich, 3D hallway model rendered in virtual reality. Subjects (n=14) reported periods of exclusive motion (Exp 1A) or orientation (Exp 1B) during extended BR viewing. In both experiments, we found that depth separation does not eliminate BR. In contrast, for motion we found an increase in BR coherence, with fewer periods of mixed dominance for drifting gratings at different depths (15.3% versus 21%, Bayes Factor = 3.54, p = .02). Depth also did not have a significant effect on BR coherence for stationary gratings (i.e., 18.3% versus 18.5%, Bayes Factor = 0.27, p = .88). Initial results from Experiment 2 showed that rivalry is still observed when depth separation is conducted in a partial occlusion layout, suggesting that additional information, apart from those associated with partial occlusion (cf. Nakayama & Shimojo, 1990) are required to resolve naturalistic interocular conflict. This additional information was explored in the context of a 3D simulation linking the area of interocular conflict and the amount of defocus blur.