Receptive field properties of human motion detector units inferred from spatial frequency masking

Abstract

This study was designed to investigate the spatial frequency selectivity and spatial structure of receptive fields of motion sensitive mechanisms in human vision. Spatial frequency selectivity was inferred from masking measurements, using dynamic test and mask stimuli. For test frequencies between 0.025 and 15.0 c/deg, maximal masking occurred when the mask frequency matched that of the test, suggesting that the test was detected by mechanisms tuned to (or near to) that frequency. For tests below 0.025 c/deg or above 15.0 c/deg, maximal masking occurred at 0.025 and 15.0 c/deg, respectively, suggesting that there exist no mechanisms selective to frequencies outside these limits. A masking model, suitable for interpreting results obtained with drifting test stimuli, was developed and used to calculate spatial frequency selectivity functions from masking data. Assuming small signal linearity, and a constant phase spectrum, the selectivity functions were inverse-Fourier transformed to yield estimates of the extent and structure of receptive fields. Field width was found to vary with test spatial frequency from 5.8 deg at 0.03 c/deg to 0.05 deg at 10.0 c/deg. These estimates were compared with width estimates previously obtained by a summation technique (Anderson & Burr, 1987), and found to be similar over a wide range of spatial frequencies (2.5 log units). Gabor functions provided a reasonable fit to the calculated field profiles at high spatial frequencies (above 1.0 c/deg), but not at low frequencies.