Abstract
Three-dimensional interpolation occurs when observers perceive surfaces that vary smoothly in depth despite sparse or absent image disparity. The neural mechanism(s) responsible for 3D interpolation are unknown. One possibility is that local disparity or depth information is propogated into blank image regions (Mitchinson & McKee, 1985). An alternate possibility is that surface-based 3D shape detectors mediate interpolation (Domini et al., 2001; Wilcox & Duke, 2003). Can a stereoscopic after-effect be obtained from the interpolated region of a 3D surface? Stereoscopic after-effects are explained by fatigue among neural mechanisms tuned to different disparities. The interpolated region of a 3D surface, however, contains no disparity. A stereoscopic after-effect obtained by adapting to an interpolated region would suggest that surface-based mechanisms mediate 3D interpolation. Observers first adapted to either random dot (density = 25%) or contour stereograms (which contain disparity only at the vertical edges). Adaptation stimuli were 6 sq. deg 3D surfaces curved about a horizontal axis and containing a maximum of either10 min or 20 min of disparity. Observers then judged whether a 3 sq. deg random dot or contour test stereogram in the fixation plane was convex or concave. Thus, observers were making a direction of 3D curvature judgment for a test stimulus that appeared well within the interpolated region of the adaptation stimulus. The means (PSEs for a flat surface) of the resulting psychometric functions were different from baseline (un-adapted), indicating a stereoscopic shape after-effect. The magnitude of the after-effect was not consistently related to the adapt-test surface types nor to the maximum disparity of the adapting stimulus. Both outcomes are consistent with depth interpolation that is mediated by a surface-based mechanism responding to the spatial derivatives of disparity rather than a mechanism which propogates local disparity signals.