Abstract
The goal of this study was to investigate at which stage in visual processing that stereo-slant adaptation occurs. We tested the predictions of two different hypotheses utilizing a property of depth from binocular disparity, namely that slant scales with distance. If adaptation occurs at a low (disparity) level, then we predict that the after-effect expressed in units of disparity gradient will be independent of distance. If adaptation occurs at a high (perceived slant) level, then we predict that the after-effect, expressed in units of disparity gradient, will increase with decreasing test distance, because a particular slant requires a larger disparity gradient at a short distance than at longer distance.
Subjects adapted to a stereo-defined slanted surface at a viewing distance of 57 cm for five minutes. The slanted surface was random-dot pattern (30 × 24 deg) of which the two half-images, viewed by the left and right eye, were horizontally magnified relative to each other. The after-effect was measured by a nulling method. The subjects judged the direction of slant of a test stimulus (24 × 24 deg). In a sequence, presentation of the same adaptation stimulus (for 2 sec) was followed by presentation of a test stimulus (for 300 msec) of which the horizontal magnification was varied according to an adaptive method (MUEST). In different sessions, we presented the test stimulus at three different viewing distances (28, 57 and 85 cm). In order to have good binocular alignment, a fixation cross was presented for 1.5 sec between presentation of the adaptation and the test stimulus.
We found that the amount of horizontal magnification that was required to null the after-effect and perceive an unslanted test stimulus increased with decreasing distance. However, the change in distance (scaling effect) was smaller than the prediction that adaptation only occurs at a perceived slant level. Thus, adaptation occurs on both low (disparity) and high (perceived slant) levels.