Can the imaging process explain ganglion cells anisotropies? D Pamplona1, J Triesch1, C A Rothkopf2 |
|---|
1Johann Wolfgang Goethe University, Frankfurt Institute for Advanced Studies, Germany |
| The statistics of the natural environment have been characterized to gain insight in the processing of natural stimuli under the efficient coding hypothesis. However, much less work has considered the influence of the imaging system itself. Here we use a model of the human imaging process that shapes the local input signal statistics to the visual system across the visual field. Under this model, we have shown that the second order statistics of naturalistic images vary systematically with retinal position [Pamplona et al, 2013, Vision Research, 83,66-75]. In the present study, we investigate the consequences of the imaging process on the properties of retinal ganglion cells according to a generative model encompassing two previous approaches [Dong and Attick, 1995, Network: Computation in Neural Systems, 6, 159-178; Doi, 2006, Advances in neural information processing]. Our results agree with previous empirical data reporting anisotropies in retinal ganglion cells' receptive fields and thereby provide a functional explanation of these properties in terms of optimal coding of sensory stimuli [Croner and Kaplan, 1995, Vision Research, 35,7-34; Passaglia et al, 2002, Vision Research, 42, 683-694]. We conclude by providing a detailed quantitative analysis of model retinal ganglion cells across the visual field. |
| Up Home |
|---|