Information maximization as a principle for contrast gain control

Contrast gain control has been found to be an important and common mechanism underlying the visual system's adaptation to the statistics of the visual scenes. Yet, the biophysical factors and computational rules governing its operation remain elusive. In this paper, we first studied the basic factors underlying contrast gain tuning in a neuronal model. We found that the nonlinearities (threshold and saturation), which are common to all spiking neurons, determines the preferred contrast sensitivity as well as the maximum information coding capacity of the neuronal model. We then investigated the design principles underlying adaptive gain control in various stimulus conditions, and found that an adaptive rescaling mechanism predicted by information transmission maximization can explain a variety of observed contrast gain control phenomena in neurophysiological experiments, including the divisive adaptation of the input-output function to mean contrast, and the inverse power law relation between response gain and input contrast. Our results indicated that the contrast gain control mechanisms in the visual systems may have a purpose of maximizing information encoding of input signals in varying environmental conditions. Page 3 of 43 ScholarOne, 375 Greenbrier Drive, Charlottesville, VA, 22901 The Journal of Neuroscience For Peer Review Only