Psychophysical receptive fields of edge detection mechanisms

Theories of edge detection generally assume a front-end linear stage involving some population of neural filters. Here we study these early mechanisms using psychophysical techniques, and evaluate a number of models for edge detection. We measured psychophysical efficiency for detection of noisy luminance edge stimuli over a range of stimulus sizes and shapes. The data suggest a diversity in receptive field shape and orientation bandwidth, consistent with physiological evidence, but inconsistent with standard multi-channel models of visual processing.