The Ponzo Illusion in Pictures with and without Suggested Depth

Abstract

figure was incorporated into two pictures depicting concrete scenes. The standard, abstract illusion remained unaltered when the Ponzo figure in the picture repre- sented the vertical face of an object viewed in the frontoparallel plane, but the illusion was increased when the figure represented an object extending in depth. The results are interpreted as failing to support Gregory's theory of 'misapplied' constancy scaling. The theory of geometrical illusions proposed by Gregory (1963, 1968) has not been confirmed by the results of several experiments ostensibly testing predictions derived from the theory. Gregory (1967) and Horrell (1971) have, however, defended the theory and clarified certain fre- quent misconceptions about the mechanisms involved in constancy scaling as proposed by the theory. The present experiment is a new test of Gregory's theory and is concerned with the role of primary depth cues in the Ponzo illusion in pictures with and without suggested depth. Gregory's theory should most clearly apply to the Ponzo illusion. In the present experiment, the extent of this illusion was measured under three conditions in which the converging lines of the Ponzo figure were incorporated into line drawings depicting different concrete scenes. In one condition, the converging lines of the Ponzo figure formed the sides of a gymnastic jumping box and thus represented the edges of a vertical (or nearly vertical) face of an object viewed in the frontoparallel plane. In the second condition, the converging lines of the Ponzo figure repre- sented the parallel sides of a road drawn in perspective. Finally, the Ponzo figure was presented alone, as an abstract drawing. The implications for Gregory's theory of the outcome of this procedure are considered in the discussion.