The Encoding of Fine Spatial Information in Salamander Retinal Ganglion Cells

old_uid7325
titleThe Encoding of Fine Spatial Information in Salamander Retinal Ganglion Cells
start_date2009/08/27
schedule12h-13h
onlineno
summaryClassical ganglion cell receptive fields have a smooth, approximately Gaussian spatial profile with high overlap among nearby cells of different morphological type. Models of retinal ganglion cell signaling predict that the receptive fields of neighboring ganglion cells should overlap and the information contained in overlapping sets of retinal ganglion cells should be redundant. In whole-cell voltage clamp recordings from pairs of cells in the salamander retinal ganglion cell layer, however, the receptive fields of nearly all cells tend to be irregular and the receptive fields of neighboring cells vary in their degree of overlap. Using small flashed spots presented near the center of the receptive field (50msec, L-cone stimulating on a rod-suppressing background) as a probe, we directly measure the visual information carried in neighboring ganglion cells. The noise amplitude scales linearly with signal amplitude for almost all cells. The information encoded among neighboring cells are more independent than expected from either a naive model of receptive field overlap or a more detailed Gaussian receptive field model derived from the data and noise model. The performance is close to predicted by the ‘blotchy’, measured receptive fields. These findings suggest that receptive fields are not as smooth or regular as assumed in classical models, and that irregularities in receptive field shape contribute to the encoding of fine spatial information.
responsibles<not specified>