Abstract
The dipper function is the typical result of psychophysical contrast discrimination experiments where just-noticeable-difference (jnd) of contrast is plotted against contrast of the grating. The dipper has been successfully modelled by assuming a non-linear contrast response function of a spatial frequency channel sensitive to the grating used in the experiment (Legge and Foley, 1980). However, the single channel model is not successful in modelling experiments in which the observer must perform contrast discrimination of gratings which differ in orientation or spatial frequency. Models which include inhibitory input from other channels have performed better on this count (Foley, 1994) but the results of our experiments, in which spatial frequency is altered, have features that are not accounted for by such models. We introduce a model which takes account of some recent neurophysiological evidence about the detailed structure of inhibitory connections in monkey V1. We show that this model does provide a better fit to the psychophysical data. Finally, we show how the model can be modified to fit data on the contrast coding of natural scenes.