Abstract
INTRODUCTION. Perception of a test pattern composed of two contrasts depends dramatically on the contrast of the adapt pattern immediately preceding it. If the immediately preceding contrast is straddled by the two test contrasts, the test pattern is very difficult to perceive correctly. If, however, the immediately preceding contrast is a bit greater than both the two test contrasts (or a bit less), performance improves dramatically. Here we show that this Straddle Effect (originally nicknamed Buffy adaptation) is spatially local and specific for orientation. METHODS. In the experiments here, the adapt and test patterns were regularly-spaced 2x2 grids of Gabor patches. The adapt-pattern Gabors could differ from the test-pattern Gabors in one of three spatial characteristics: orientation, phase, or exact position. The spatial frequency and size of all Gabor patches were constant throughout. All the Gabors in the adapt pattern had contrast 50%. Two of the four Gabors in each test pattern were at one contrast, and two were at another. The observer had to say whether the arrangement of the two different test contrasts formed horizontal or vertical stripes. The adapt pattern was on for about 1 sec before and after the 100 ms test pattern. RESULTS. When the spatial characteristics of the adapt and test patterns were identical, there was always a Straddle Effect. When they differed in orientation, phase, or position, the Straddle Effect was much diminished. The way the Straddle Effect diminished was NOT that the straddling test pattern became easier to perceive correctly. Rather test patterns having contrasts both a bit above the adapt contrast (or both a bit below) became much harder to perceive correctly. DISCUSSION. Perhaps the contrast comparison process proposed to account for the Straddle Effect occurs at the output of something as localized as a V1 receptive field.
Meeting abstract presented at VSS 2012