Retinal Image or Perceived Features as Determinants of Error in Geometric Illusions?

Abstract

An examination of the implications of the theory of geometric illusions proposed by Gregory (1963) leads to the prediction that the illusion error should be determined by the retinal image projected by the figure rather than by its perceived characteristics where these differ. This prediction is tested in two experiments, using a version of the Zöllner illusion. In the first experiment, distance is used to manipulate the retinal image characteristics and, in the second, slant. In both cases, error is found to follow changes in the retinal image rather than being held relatively constant by the perceived size and shape. It is suggested that the effect of slant and distance on illusions, as reported by others, can be attributed to the change in retinal image characteristics involved in such operations.