MEASURING THE SPATIAL-FREQUENCY SELECTIVITY OF 2ND-ORDER TEXTURE MECHANISMS

Recent investigations of texture and motion perception suggest two ear ly filtering stages: an initial stage of selective linear filtering fo llowed by rectification and a second stage of linear filtering. Here w e demonstrate that there are differently scaled second-stage filters, and we measure their contrast modulation sensitivity as a function of spatial frequency. Our stimuli are Gabor modulations of a suprathresho ld, bandlimited, isotropic carrier noise. The subjects' task is to dis criminate between two possible orientations of the Gabor. Carrier nois es are filtered into four octave-wide bands, centered at m = 2, 4, 8, and 16 c/deg. The Gabor test signals are w = 0.5, 1, 2, 4 and 8 c/deg. The threshold modulation of the test signal is measured for all 20 co mbinations of m and w. For each carrier frequency m, the Gabor test fr equency w to which subjects are maximally sensitive appears to be appr oximately 3-4 octaves below m. The consistent m X w interaction sugges ts that each second-stage spatial filter may be differentially tuned t o a particular first-stage spatial frequency. The most sensitive combi nation is a second-stage filter of 1 c/deg with first-stage inputs of 8-16 c/deg. We conclude that second-order texture perception appears t o utilize multiple channels tuned to spatial frequency and orientation , with channels tuned to low modulation frequencies appearing to be be st served by carrier frequencies 8 to 16 times higher than the modulat ions they are tuned to detect.