Fig. 2: Functional cell typing reveals the dependence of axonal propagation speed on eccentricity in midget and parasol cells and in human and macaque retinae.

a, From top to bottom: schematic of HD-MEA recordings with light stimulation. ‘ON–OFF’ light stimulus (contrast over time); raster plot of the spiking response of a representative ON midget RGC. Each row represents a single stimulus presentation, and each small vertical dash represents a spike; average firing rate over trials is depicted below. F.R., firing rate. b–d, Clustering of light-induced RGC responses to identify functional cell types of the human fovea (b), human periphery (c) and macaque periphery (d). Left: normalized firing rates (averaged over trials) of all RGCs depicted as rows in response to the stimulus depicted above. Cells were grouped by clusters (number in circle). Labels on the left indicate the putative cell type for groups of clusters. Right: functional clusters (UMAP projection). UMAP coordinates in b–d were rotated to reflect similarity in the cluster structure. Each dot represents an RGC. The colors correspond to the cell type (left). The numbers in the circle indicate the cluster number. Dashed lines are visual aides that separate cell types of different response polarities. e, Distributions of action potential speeds in midget (purple) and parasol (green) cells in human fovea and periphery. Foveal (midget, 0.56 ± 0.17 m s⁻¹; parasol, 0.68 ± 0.17 m s⁻¹; median ± s.d.); peripheral (midget, 0.91 ± 0.22 m s⁻¹; parasol, 1.10 ± 0.20 m s⁻¹; median ± s.d.). Two-sided Wilcoxon rank-sum test—foveal, **P < 0.01; peripheral, ***P < 0.001. Intratype comparison shows lower speeds in the fovea than in the periphery. Group medians are indicated. f, Same as e but for macaque periphery (median speed midget, 0.92 ± 0.24 m s⁻¹; median speed parasol, 1.08 ± 0.21 m s⁻¹; two-sided Wilcoxon rank-sum test, ***P < 0.001).

Source data