Spatial frequency discrimination in cyclopean vision.

It is well known that inspecting a cyclopean grating causes threshold for detecting a subsequently presented test cyclopean grating to be elevated, and that the threshold elevation is greatest at the adapting frequency. We report here that spatial frequency discrimination threshold is also elevated, but the elevation is greatest at frequencies offset from the adapting frequency, and the elevation at the adapting frequency is near-zero. We conclude that discrimination threshold is determined by the relative activity of cyclopean frequency-tuned channels, and suggest that relative activity is computed at an opponent-frequency stage. Discrimination threshold for cyclopean gratings was 2.5-4% for two observers, and remained approximately constant over the range 0.16-2.0 cycles/ degrees. Discrimination threshold for luminance-defined gratings was only slightly lower. Discrimination threshold was approximately independent of the grating's peak-to-peak disparity over a range of 45:1 for one observer and 17:1 for another. This finding as well as the low value of discrimination threshold are consistent with an opponent-process model. The dot density of every cyclopean grating used was chosen bearing in mind our finding that three or more spatial samples per grating cycle are required before sampling effects can be ignored.