Abstract
WHY: Increasing the contrast of both left and right eye half-images of a stereoscopic stimulus improves performance in a stereoacuity task. Increasing the contrast of just one half-image degrades performance. This “Contrast Paradox” phenomenon implies some form of gain adjustment in the mechanisms responsible for detecting offsets. This phenomenon also occurs for some vernier and motion tasks when the stimulus components being compared differ in contrast. In this study, we tested whether the contrast paradox occurs for stereoscopic judgments in which each eye's image was equivalent to vernier and motion stimuli that produce a paradox. If the gain adjustment operates on each half-image prior to binocular combination, the stereo task should show a Contrast Paradox. Otherwise, performance should follow the average contrast. HOW: We measured thresholds for detecting relative disparity between the top and bottom halves of a single Gabor patch, in which contrast was varied independently in the two halves. We also measured thresholds for detecting disparity motion of a single Gabor patch, in which contrast was varied independently in two sequential frames. The circular 1 cpd vertical Gabor stimuli had contrasts of 10% and/or 40%. WHAT: Thresholds for detecting relative disparity did not show a contrast paradox in either task. Performance roughly followed average contrast (i.e., when contrasts were unmatched, thresholds were in between those for matched low and matched high contrast). Conversely, offset detection based on one half-image (vernier offset or lateral motion) did show a contrast paradox, as expected. SO WHAT? Are relative disparity thresholds for depth offset and depth motion based on monocular offset and motion signals, respectively? If so, thresholds should have shown a contrast paradox. Instead, our results imply that relative disparity judgments rely on a second-stage comparison of disparity signals extracted independently from each half of the depth offset or each frame of the depth motion stimuli.