Identification of multiple noise sources improves estimation of neural responses across stimulus conditions

Abstract

Most models of neural responses are constructed to capture the average response to inputs but poorly reproduce the observed response variability. The origins and structure of this variability have significant implications for how information is encoded and processed in the nervous system. Here, we present a new modeling framework that better captures observed features of neural response variability across stimulus conditions by incorporating multiple sources of noise. We use this model to fit responses of retinal ganglion cells at two different ambient light levels and demonstrate that it captures the full distribution of responses. The model reveals light level-dependent changes that could not be seen with previous models. It shows both large changes in rectification of nonlinear circuit elements and systematic differences in the contributions of different noise sources under different conditions. This modeling framework is general and can be applied to a variety of systems outside the retina.