Abstract
Past investigations (e.g., Norman, Todd, & Phillips, 1995) have shown that optical patterns containing only deformations of smooth shading and/or specular highlights are as informative about local aspects of 3-D shape (e.g., surface orientation) as patterns containing surface texture. In the current study, we investigated the informativeness of deforming shading and specular highlight fields and moving texture patterns for the discrimination of global 3-D shape. The stimuli were randomly generated smoothly-curved objects defined by the positions and orientations of 5120 connected triangular polygons, and were similar to those used by Norman & Todd (1996, 1998). For each condition (texture, shading, highlights), thresholds for the discrimination of global 3-D shape were determined. The objects were created by a process analogous to Fourier synthesis: they were formed by modulating a sphere in depth with a variety of unidimensional sinusoidal surfaces with varying orientations and magnitudes. By varying the relative magnitudes of the final modulations on every trial, we created pairs of surfaces that were similar in overall shape, yet were different. This overall shape difference was initially large at the beginning of a block of trials and was reduced over time according to the PEST (parameter estimation by sequential testing) adaptive procedure. Threshold shape differences were determined for all conditions and observers. We used a 2 temporal interval AFC paradigm, and the observers' task was to indicate for each pair of objects, whether their 3-D shape was the “same” or “different”. The results showed that the observers' discriminations were remarkably good for surfaces defined only by shading and/or specular highlights. Indeed, the performance for specular highlights was best, followed by shading and texture. The results demonstrate that shading and specular highlights are valuable sources of information for the human perception of global shape.