Abstract
Introduction: We are interested in how the visual system combines different visual cues to recover depth when one cue is ambiguous. Convex and concave surfaces produce similar texture projections, especially at large viewing distances. Our study considered disparity (an unambiguous cue) and its combination with ambiguous texture information. We investigated the amount of interaction between the two cues. Specifically, we asked whether disparity and texture were processed (a) separately, before linear combination of shape estimates, or (b) jointly, such that disparity effectively disambiguated the texture information.
Methods: Vertical ridges were presented stereoscopically to observers. There were four different texture patterns. Each was consistent (in terms of maximum likelihood) with both a convex and a concave ridge of roughly the same amplitude. The lengths, orientations and distribution of the texture lines were consistent with 0, ±2.5, ±5 or ±7.5 cm ridges. Disparity was consistent with a −7.5, −5, −2.5, 0, 2.5, 5, or 7.5 cm ridge. Observers set a 2D cross-section to match the perceived amplitude. In a second experiment observers judged the profile of the stimuli defined solely by texture under monocular viewing.
Results & Conclusions: When texture was the only available shape cue (monocular presentation), observers consistently reported the convex interpretation. However, in stereoscopic stimuli, texture information modulated the shape from disparity in a way inconsistent with simple linear combination. For example, when disparity indicated a concave surface, a texture pattern which was perceived as highly convex when viewed monocularly caused the stimulus to appear more concave than a ‘flat’ texture pattern. Our data support the notion that different cues can disambiguate each other (e.g. modified weak fusion). The monocular data and the cue interactions are well modelled using a Bayesian approach that incorporates a prior for convexity.