Abstract
Purpose: Chan, Stevenson & Pizlo's experiments (submitted) showed that binocular perception of a 3D shape critically depends on simplicity constrains (e.g. planarity of contours, symmetry of an object). They formulated a model in which perceptual processing begins with 3D shape reconstruction from one retinal image. This reconstruction involves simplicity constrains. The role of binocular disparity in this model is limited to correcting the monocularly reconstructed shape. To test this model, we performed two experiments on 3D shape recognition, in which 3D objects characterized by various degrees of simplicity were viewed binocularly or monocularly. If binocular disparity is of secondary importance, binocular and monocular performance should be correlated. In particular, binocular performance is expected to be high only in the case of objects for which monocular performance is high, as well. Method: We used six types of objects with different degrees of “simplicity”: (a) polyhedron represented by edges of the faces, (b) vertices of a polyhedron, (c) a polygonal line connecting the vertices of polyhedron in a random order, (d) partially non-planar, symmetric polyhedron (some visible contours were not planar), (e) planar, asymmetric polyhedron, and (f) non-planar, asymmetric polyhedron. Hidden edges of the polyhedra were removed. Three subjects (one naïve) were tested. Results: The correlation between binocular and monocular performance was fairly high (r > 0.8). In particular, high binocular performance was observed only in cases where monocular performance was high, as well. For all six stimuli, binocular performance was better than the monocular one. Conclusion: These results provide strong support for Chan et al.'s model, in which binocular and monocular perception of 3D shapes involves the same set of a priori constraints. Binocular disparity seems to be a secondary factor.