A communication-theoretic formulation of a continuous linear-nonlinear model of retinal ganglion cells

The analysis of mutual information has been widely applied to the nervous system, but it is complicated by the extensive nonlinearities of neurons. The linear-nonlinear system with a continuous input and output is a common framework representing sensory systems, including the retina and early visual system, auditory system, and psychophysical behavior. Here we derive an analytical formulation for mutual information in continuous linear-nonlinear (LN) systems with signal-dependent noise. We expect this formulation to facilitate the understanding of how the parameters of an LN system influence information transmission.