The role of the angle components in the wings-in and wings-out forms of the Müller-Lyer illusion.

The effect of selective amputations of the angle components in the wings-in (underestimated) and wings-out (overestimated) forms of the Müller-Lyer illusion was examined in two experiments. The stimulus figures consisted of one, two, or four angles. In experiment 1 the method of paired comparisons was used to scale the figures on the psychological continuum of length, and in experiment 2 the method of reproduction was used to obtain quantitative measures of illusion magnitude. There was good agreement between the scaling and the length-reproduction measures of the illusion. The illusory effects in all figures were significant, and the extent of the underestimation and overestimation of the wings-in and wings-out figures, respectively, increased as the number of angles increased. In general, selective amputation of the angle components produced similar patterns of illusory effects in the wings-in and wings-out figures. These findings are discussed with reference to the issue of whether the two forms of the conventional (ie four-angle) Müller-Lyer illusion are similar or distinct illusion types.