|
Abstract:
|
A key challenge in neuroscience is to understand how large networks of neurons collectively encode information. It has recently become possible to record simultaneously from complete mosaics of ~100 ON and OFF parasol cells over a 4x8 degree region of peripheral retina, providing the opportunity to better understand the message the eye sends the brain. We present a new generalized linear multivariate point process model of the visual response properties of networks of these retinal ganglion cells. We discuss model fitting and validation (challenging due to the high dimensionality of the data, but computationally tractable due to the concavity of the loglikelihood), demonstrate the model's accuracy in predicting retinal responses to novel visual stimuli, and describe Bayesian techniques for decoding the network responses. With J. Pillow, E. J. Chichilnisky, E. Simoncelli, and J. Shlens.
|
