Occlusion-based stereopsis with alternating presentation of the stereo half images

摘要

Recently, it was reported that stereoscopic phenomena, including stereopsis and binocular luster, could be differentiated according to dichoptic alternation frequency thresholds (Ludwig, et al., 2007, AP&P 69(1), 92-102; Rychkova & Ninio, 2011, iPerception (2), 50-68). In these studies, left and right eye images were alternated at various rates and the threshold was defined as the minimum alternation frequency for which participants reliably reported depth or luster. In these two studies, alternation thresholds for various stereoscopic phenomena ranged from approximately 3Hz to approximately11Hz. We applied this technique to explore temporal integration for several occlusion-based stereoscopic phenomena (e.g. da Vinci stereopsis (Nakayama & Shimojo, 1990, Vis Res 30(11), 1811-1825), monocular gap stereopsis (Gillam et al., 1999, Vis Res 39(3), 493-502), and phantom stereopsis (Anderson, 1994, Nat 367, 365-368)). Perceived depth in these stimuli is thought to result from the binocular combination of non-corresponding monocular occlusion zones that are geometrically consistent with an occlusion resolution. Participants viewed alternating dichoptic images in a mirror stereoscope. Each stereo-pair depicted two depth orders of the same stimulus, separated spatially. One depth order was geometrically consistent with a nearer target and the other was either consistent with a farther target or inconsistent with occlusion altogether. Using the method of adjustment, participants either increased the alternation rate of the stereo half images until they could identify which of the two stimuli depicted the surface arrangement with the nearer target, or they reduced the alternation rate until they could not make this discrimination. We computed mean alternation rates between randomized ascending and descending trials for each participant. Depth discrimination was reliable across a range alternation rates from approximately 3Hz to approximately 10Hz. Temporal integration of alternating stereo half images of occlusion-based stereograms differs across occlusion-based stereo phenomena but falls within the range observed for disparity-based stereopsis and binocular luster.