A Continuous Version Of The Barber-pole Illusion (1991)
In the well-known "barber pole" illusion, a grating seen behind an elongated rectangular aperture is seen to move along the length of the aperture, although the actual motion of the grating is ambiguous. This effect is sometimes explained in terms of the "terminator" features that occur at the intersections of the grating edges with the aperture. In the present experiments, subjects were presented with sinusoidal luminance gratings presented through an asymmetrical two-dimensional Gaussian contrast window, eliminating the sharply defined terminators. For small angular deviations, the direction of perceived motion depends on both the grating orientation and the window orientation. A typical stimulus consists of a vertically oriented window with an aspect ratio of 2:1, in which about 4 cycles of a drifting horizontal grating can be seen; in this case, the change in perceived direction which occurs when the grating is tilted slightly can be nulled by tilting the window in the opposite direction by approximately the same angle.
These data can be explained by models in which satisfaction of the "intersection-of-constraints" solution of Adelson and Movshon (\fINature\fP, 1982) competes against another factor, such as a preference for a solution in the direction of the luminance gradient (the "normal flow"). Such models also make predictions concerning judgments of perceived 2-D rigidity, such as those studied by Nakayama and Silverman (\fIVision Res.\fP 1988).
Barber-pole, Continuous, Illusion, Version
Inv. Ophth. Vis. Sci. (suppl.), 32, 829 |