The role of symmetry in shape perception

Abstract

Five experiments are reported that test a symmetry theory of contextual effects on the perceived shape of a class of ambiguous figures. The theory assumes that at least some aspects of shape perception necessarily take place within a perceptual reference frame and that the choice of frame is strongly biased toward axes of global symmetry. The amount of bias produced by configural symmetries was determined for a shape discrimination task by measuring reaction times to distriminate the shape of a target figure (as a square or a diamond) when it is embedded in contexts with horizontal and vertical symmetries versus left and right diagonal ones. Experiments 1 and 2 used configural lines of squares/diamonds to demonstrate that performance is much worse when symmetries exist only along diagonals and to establish that the present configural bias in shape perception is closely related to those previously demonstrated in perceived pointing of ambiguous triangles (Palmer and Bucher 1981). Experiment 3 generalizes the phenomenon along lines predicted by the symmetry theory by showing that the same results are obtained for discriminating a plus-sign (+) from the letter ‘X’ in configural arrays. Experiments 4 and 5 generalize it further by showing that the same kinds of bias effects can be produced by other symmetrical contexts: rectangular frames, bisecting lines, and stripes inside a single square/diamond. These results are interpreted as supporting the assumptions of the symmetry theory with the provision that local as well as global axes of symmetry are important.