Patterns of dark dominance in cortical input.
(A) Two schematics of retinal location illustrate the layout of 8 RGC subpopulations: P pathway and M pathway, foveal and peripheral, ON and OFF. ON (bright center, dark surround) and OFF (dark center, bright surround) cells are illustrated as alternating for clarity, however each subpopulation actually fully tiles the retinal space. Bright values indicate excitatory regions and dark values indicate suppressive regions. P cells are smaller and more numerous than M cells, and foveal cells of both types are smaller than peripheral cells. Four colors are used throughout to indicate each subpopulation: P foveal (yellow), P peripheral (green), M foveal (purple), and M peripheral (orange). (B) Illustrations of the spatial receptive fields of the simulated retinal ganglion cells. Each of the four plots shows the receptive fields for ON and OFF cells of one subpopulation. Each line shows a middle slice through the isotropic 2D DOGs used to simulate RGC receptive fields. Solid lines show the extent of ON receptive fields and dashed lines show the extent of OFF receptive fields, offset laterally for visualization. The black dashed line indicates zero response. OFF receptive fields respond positively when the center is darker than the surround and visa versa. Because the surrounding Gaussian has a large standard deviation, the suppressive surrounds appear very weak in these plots. Icons located within each plot show the ratio of center-to-surround standard deviations. Details of the receptive field parameters can be found in the Methods. (C) Models of the contrast response nonlinearities previously measured for retinal ganglion cells. (D) The overall ratio of OFF to ON cortical input from each pathway for natural images and white noise images. The horizontal dashed line at 1 indicates equal OFF and ON input, values greater than 1 indicate an OFF bias. (E-H) Normalized amplitude (Amp.) distributions for each visual feature are shown for each subpopulation. For Weber contrast, ON and OFF responses were only aggregated for positive contrasts and negative contrasts, respectively. (I-L) The OFF bias was computed as the ratio of the summed OFF responses to the ON responses over all input images.