Abstract
Several psychophysical studies with humans have reported that attention affects the contrast response function via contrast gain, response gain or both. Correspondingly, several neurophysiological studies have shown that attention modifies the contrast response functions of neurons in early visual areas, such as V4 and MT, by either contrast- or response-gain mechanisms. Whereas contrast gain corresponds to a change in sensitivity of the contrast response function, response gain corresponds to a multiplicative increase in response.
Although the pattern of results are consistent between psychophysical and neurophysiological studies, comparisons need to be made with caution, and an explicit link between psychophysical data and single-unit recordings should be established. To narrow this gap, we propose a simple biologically plausible model based on an extension of Jazayeri & Movshon's model (2006), which predicts optimal representation of sensory information given neural population response. To assess the effect of attention on contrast response, the model we propose makes specific quantitative predictions regarding gain changes in neural response as a function of performance in 2AFC orientation discrimination tasks. By changing the parameters of the assumed population - orientation tuning, threshold and saturation of the contrast response function- the model can predict attentional modulation at several visual areas (e.g., V1, V4).