Abstract
Are there neural mechanisms that are selective for 3-D curvatures, and are they invariant to the patterns that form the curvature? We test for such mechanisms by adapting to curvatures defined by plaids differing in spatial frequency by a factor of three. Gratings differing by frequency octaves do not raise contrast thresholds for each other, indicating that they are detected by independent mechanisms. Observers adapted monocularly to perspective images of carved, sinusoidal surfaces textured with horizontal-vertical plaids at either 2 or 6 cpd. Each adapting image spanned 1.5 cycles and was concave or convex at the central fixation. Shape aftereffects were measured using 1 flat, 4 concave, and 4 convex test images varying in amplitude between the concave and convex adapting stimuli, and textured with either the 2 or 6 cpd plaid. Observers were presented with a 2 minute initial adaptation, followed by 200 msec test images each preceded by a 5 second top-off adaptation. They were asked to judge the shape of the test as concave or convex. The perceived flat point was estimated as the curvature that was perceived as convex 50% of the time, and was extracted from psychometric fits. To examine frequency selectivity, we measured shape aftereffects with test stimuli of the same or different frequencies as the adaptation stimulus. Results from six observers show significant shape aftereffects: adapting to a convex surface causes a flat test to appear concave, and thus the perceived flat point is shifted to convex values, and vice versa for concave adaptation. Aftereffects were obtained in both within and across frequency conditions, but were larger for the within-frequency conditions. Pattern adaptation cannot account for aftereffects in the across frequency conditions. These results thus show that there are neural mechanisms selective for 3-D curvatures whose responses are invariant to the texture pattern.
NIH R01 EY13312 to Q. Zaidi and PSC-CUNY 60066-35 36 to A. Li.