Response characterization of a single representative RGC. a) Characterization with standard methods. The left panel shows the raster plot for all the repetitions of a flash at full contrast (digitized intensity from 0 to 255 for \(400\text{ms}\)); the background color indicates the luminance profile (arbitrary scaling) along time. The panels on the right show the STA computed from the response to checkerboard stimulus. The left panel shows the temporal profile of the receptive field, plotting the contrast at the point of largest variation along time (black crosses), determined by SVD, and its fit to a set of two low-pass filters (red line). The plot on the right shows the spatial profile of the STA as the frame corresponding to peak contrast and its fit to a 2-D Gaussian function (red ellipse at 1 s.d.); the black bar is \(0.2\text{mm}\) long. While the STA shows the typical response of an OFF type cell, the flash response shows a brief response to stimulus onset, meaning that the cell response has an ON component. Together, these results show that the recorded unit corresponds to a standard RGC. b) The response to speed at each spatial frequency (sf) was fitted to a skewed gaussian. All cells that had a good fit (\(\chi^{2}<0.05\)) at every sf were classified as speed responsive cells. The fit allows to interpolate the preferred speed and to measure the tuning bandwidth. c) Spatiotemporal tuning of the cell for the different types of stimuli. The intensity at each square denotes the average firing rate for the corresponding stimulus parameters, relative to the average firing rate in response to the stimulus with maximum response. As expected, speed preference decreases with spatial frequency, as seen by the slanted response profile. For the MC, the response profile is smaller, evidencing a narrower tuning. d) Raster and PSTH for each type of stimulus and to the different speed (in \(\mu m/s\)) at the preferred sf of the cell (the sf that elicited the largest response in C, denoted by the red cross). This cell, like many others in the sample, increases its firing rate for higher speeds. The firing rates are stable across repetitions, so the narrower tuning is not due to drifts or decay in cell activity. Supplementary figures show other representative cells using the same layout.