The inappropriate constancy scaling notion of geometric illusions was explored by employing a textural analogue of the Ponzo figure. Ten Ss estimated the length of a horizontal line by equating it with varying companion lines in the context of the Ponzo figure, a textural analogue, and a baseline control in which the lines appeared with no surrounding contours. The textural analogue had the added feature of imposing no contours at the ends of the horizontal lines. It was found that length estimates were significantly different between the horizontals of the Ponzo figure and control stimuli, but not between the texture figure and a context-free control. The results suggest that inappropriate constancy scaling plays a minor role at best in the perception of geometric illusions.One category of theory that attempts to explain the operation of geometric illusions assumes that the failure to equate physicaUy identical elements in two-dimensional displays may be attributed to inappropriate constancy scaling (Day, 1972; Gergory, 1963;Tausch, 1954). Such a theory explains the Ponzo illusion by proposing that the angular portion of the figure suggests depth in the manner of linear perspective, and so the more "distant" of the two lines is perceived as larger when both lines subtend equal visual angles.Other theories, the assimilation theory in particular, discount the importance of constancy scaling and emphasize the presence of contours at the end points of the lines to be equated (Fisher, 1968(Fisher, , 1969Pressey, 1971).Demonstrations and experiments that support the misplaced scaling notion have not always been careful to isolate depth features of a display while controlling for the occurrence of end contours. For example, Leibowitz, Brislin, Perlmutter, and Hennessy (1969) presented evidence that the size of two lines of equal retinal extent were misjudged when superimposed on a photograph of an open field, a scene which supposedly portrayed strong textural cues. It is contended that their context may not have eliminated the occurrence of end contours, particularly because texture was more densely concentrated around the upper horizontal line than around the lower line. In fact, a close examination of the figure suggests that linear perspective cues may have been present as well. The general problem is that linear perspective cues generally produce end contours at the same time.In the present experiment, an attempt was made to evaluate size estimations of physically identical horizontal lines when embedded in a depth context created by texture alone. The stimulus was arranged in such a way that end contours were not present and so *Requests for reprints may be sent to: Mark B. Fineman, Southern Connecticut State College, Department of Psychology, 501 Crescent Street, New Haven, Connecticut 06515.
31afforded an opportunity to assess the role of constancy scaling alone. It was felt that this procedure would help to isolate individual factors contributing to the Ponzo illusion and would more clearly delineate the re...