Influence of surface attitude and curvature scaling on discrimination of binocularly presented curved surfaces.

We report on the ability of human observers to discriminate local second-order shape of quadratic stereo-defined surfaces. Local second-order shape can be specified by two parameters: the curvedness (a scale-dependent quantity describing overall curvature of a shape) and the shape index (a scale-independent quantity describing the shape's appearance). We measured shape index discrimination thresholds of shapes subject to attitude and curvedness transformations. Results show that neither slant nor curvedness manipulations affect or bias observers' judgements of surface shape. Furthermore, ideal detector simulations show that observers do not perform optimally: they do not exploit all available information.