Extended Data Fig. 3: Cell type identification in the mouse retina. | Nature

Extended Data Fig. 3: Cell type identification in the mouse retina.

From: Nonlinear receptive fields evoke redundant retinal coding of natural scenes

Extended Data Fig. 3

The retina was stimulated with a barcode stimulus. Cell responses were then clustered along with information from white-noise stimulation (receptive field size, temporal filter, autocorrelation). a, Responses to the barcode stimulus of four identified clusters. These responses were aligned to a seed cell (first row) to show the match. b, Receptive-field mosaics of the four identified clusters. c, Temporal filters. d, Spike-train autocorrelations (bin size is 0.5 ms). e, Clustering of identified cell types, shown in similar fashion as in Extended Data Fig. 1e. f, Cell-type-specific information rates for different response frequencies under the natural movie (colored) or spatiotemporal white noise (grey). Colored lines represent averages for pairs at similar x-coordinates (with 95% confidence intervals) within the same ganglion cell type. g, Fano factor histograms for cell responses to the natural movie, calculated for each cell by considering spike counts for individual fixations. The reported Fano factors are medians across cells (95% robust confidence intervals in parentheses). h, Histograms of response reliability under natural movies and white noise, measured by the coefficient of determination between firing rates of even and odd stimulus repeats. Data for f-h are from 8 retina pieces.

Back to article page