The relationship between visual illusions and cues to distance

Abstract

The theory of R. L. Gregory that certain visual illusions are caused by the inappropriate action of a constancy-scaling mechanism was critically examined. Several unsuccessful attempts were made to replicate his experimental findings that certain ambiguous figures, such as the M-L illusion, appear 3-dimensional in a particular way when presented in reduced cue conditions. It was noted that the depth effect reported by Gregory was not large enough to explain all the illusory distortion in his figures. It was suggested that this might be because his apparatus allowed certain cues which could be used to determine the true form of the figures and thus destroy or reduce any 3-dimensional effects. The experimental results suggested that this was not so. In later experiments it proved possible to repeat Gregory's results only by inducing Ss to adopt a specific perceptual set. If this was not done Ss tended to see the figures in different ways which often changed over time. Combined analysis of the results of all Ss on many different figures showed a slight tendency for the central part of any Gestalt or figure to appear nearer than other parts. Two possible hypotheses were advanced to explain this result but further experimentation suggested that both were inadequate. Experimental evidence is provided that the Ponzo illusion is the result, of a shrinkage of the lower line rather than an expansion of the upper line, as is generally thought. This and other evidence is interpreted as suggesting that even this illusion may not have a perspective component. Taken as a whole the results suggest that any perspective theory of the illusions will prove inadequate. It is finally suggested that further research be directed towards inhibition type theories.