Stereopsis from motion-defined contours

Abstract

Random-dot stereograms demonstrate that monocularly visible contours are not necessary for stereopsis, although in the absence of point-for-point correspondence, they are sufficient for stereoscopic combination. The quality of stereopsis from interocularly uncorrelated motion-defined forms was examined here. Results indicate that perceived magnitude of depth is not veridical, and that more depth is seen for crossed than uncrossed disparities. The difficulty in perceiving “behind” depth is due to a monocular depth cue which conflicts with binocular disparity in specifying depth only in the absence of interocular correlation. The overall reduction in depth is not the result of binocular rivalry from the lack of interocular correlation, and so appears to be a function of the type of feature being matched.