Abstract
Previous studies have shown that the disparity threshold Dmin depends on the interocular contrast ratio. Dmin is lowest (best), when the interocular contrast ratio is 1. However, little is known about the dependence of Dmax on the interocular contrast ratio. In this study, we measured both Dmin and Dmax using random-Gabor-patch stereograms, in which, Gabor patches had random positions and phases, but with a fixed spatial frequency (3 cpd). The two eyes had two identical arrays of patches except that one eye's array could be shifted horizontally, and they could differ in contrast. We found that, for both Dmin and Dmax, performance was best when the two eyes had equal contrast, and declined when the contrast was reduced in either or both eyes. Ironically, reducing the contrast in one eye resulted in worse performance than reducing it in both eyes, consistent with previous studies on Dmin (Legge & Gu 1989; Hou et al 2011). We tested two models: (1) Energy model: the disparity energy is proportional to the product of the two eyes' contrast, and is then normalized (divided) by monocular contrast energy; (2) Double gain-control model (Ding & Sperling 2006): the contrasts in the two eyes are first mutually suppressed by each other (double layered), and then multiplied to compute disparity energy. Both models assume that the disparity energy is a function of disparity, the product of disparity power and exponential decay functions, which increases with stimulus disparity at small disparities but decreases at large disparities. Both models predict Dmin equally well (reduced-chi-square less than 1.5); however, the gain-control model performs much better (reduced-chi-square less than 2.5) than the energy model (reduced-chi-square greater than 5.5) for Dmax. Further work is needed to determine whether Dmin and Dmax share a similar network.
Meeting abstract presented at VSS 2016