Shape constancy as a function of selected physical stimulus dimensions.

Abstract

By a random method the right half of a figure was constructed; then a mirror image was drawn on the left side to produce a perfectly symmetrical figure. This basic figure was manipulated in such a way as to produce five figures (instances) each on the four physical dimensions of symmetry, number of points, ratio of curved to angled turns, and extension. The hypothesis that the instances of the physical dimensions would be related to degree of shape constancy in an ordinal manner was not supported in any simple way. However there was enough evidence to warrent further investigation. The dimension of extension was selected. Three new stimulus forms were constructed in a manner commensurate with Exp. I, and varied over the dimension of extension. The functions from instance to instance of the three random figures were not significantly similar. However, the homegeneity of mean Brunswik ratios, and individual responses decreased with increased extension; also, values of the ratio closer to 1.00 were generated with increased extension. This suggests that the interaction effects of particular figure and location on the dimension of extension may be of fundamental importance.