Abstract
Previous research has suggests that the symmetry of a 3D object facilitates shape constancy across changes in viewpoint (Vetter et al., 1994; Pizlo & Stevenson, 1999; Saunders & Lee, 2011). The aim of this study is to show that symmetry only facilitates shape constancy by cueing the direction and magnitude of the change in viewpoint. Discrimination performance is equivalent when similar cuing is provided for asymmetrical shapes. Symmetric and asymmetric globally convex 3D shapes were generated to have similar complexities by phase scrambling their spherical harmonics while retaining the frequency and amplitude of the shape properties. The shapes were shaded by a diffused point light source. The same symmetrical and asymmetrical shapes were rendered with and without surface contours in addition to the shading. The versions of the shapes with surface contours were circumscribed with a single contour. For symmetrical shapes, the contour followed the plane of symmetry. Shape pairs were presented sequentially in a same/different shape discrimination task with monocular viewing. The second shape presented was rotated about its central vertical axis by a maximum of ±60°. Observers’ ability to discriminate shape was negatively correlated with the magnitude of rotation in all conditions. There was an interaction between the shape symmetry and whether or not surface contours were present. Observers performed better when shapes without contours were symmetrical compared to asymmetrical. The addition of surface contours to asymmetrical shapes increased the performance to match that of symmetrical shapes. These results suggest that both symmetry and surface contours are used as diagnostic features to identify the magnitude of rotation. Once the magnitude of rotation has been determined, symmetrical shapes are no easier to discriminate between than asymmetrical shapes.
Meeting abstract presented at VSS 2012