Abstract
Last year we reported a new kind of contrast adaptation, which we called “Buffy adaptation” (Graham & Wolfson VSS 2006). We explain it as adaptation of a contrast comparison level in a “Buffy channel.” Our original stimuli were large: each was a 15x15 grid of Gabor patches. This year we show that the same phenomena exist with much smaller stimuli (presented foveally). After adapting for a short time (1 sec) to a 2x2 grid of Gabors all at the same contrast (e.g. 50%), a test stimulus is presented briefly (82 msec), and then the adapting stimulus is shown again (1 sec). In this 2x2 test stimulus, the Gabor patches in the first row (or column) have one contrast (e.g. 40%), and those in the second row (or column) have another contrast (e.g. 60%). Thus the test stimulus contains horizontal (or vertical) contrast-defined (a.k.a. second-order) stripes. The subject's task is to identify the orientation of these stripes. When the test stimulus' contrasts (e.g. 40%,60%) STRADDLE the adapting contrast (50%), subjects perform very poorly; but when the contrasts are ABOVE (50%,60%) or BELOW (40%,50%) the adapting contrast, subjects perform very well. On the other hand, subjects perform poorly on all three of these stimuli after adapting to a blank gray field (0% contrast). That these effects are seen with small stimuli (2x2) as well as large (15x15) means they can occur within a small region and do not require extensive spatial pooling. Furthermore, the effect is found when the small stimuli are presented peripherally (inner edge at 1, 3, or 5 deg eccentricity) in spite of the differences between foveal and peripheral processing. In addition to showing that our “Buffy channel” explanation works, we will show that a number of other explanations do not.
Supported by NIH grant EY08459