Stimulus-evoked retinal ensembles.

(A) Top. Population activity (spikes per bin) of 319 RGC’s in time during ON light stimulation (white background) and OFF (gray background). The black arrow shows the ON stimulus started. As expected for the retina, the population activity transiently increases when the stimulus switches and then decays and stabilizes. Middle. Each population vector is coloured according to the ensemble to which they belong. Bottom. The activation of each ensemble in time is represented by a coloured dot (matching the population vector color code). The method detects a consistent stimulus-locked ensemble activity with ON (ensembles from 1 to 5) and OFF (ensembles from 6 to 10) ensembles. Ensemble 0 represents all the population vectors that were discarded according to rejection criteria (see Methods). (B) The spatial distribution of the estimated receptive fields for the corresponding core-cells of each ensemble. (C) The stacked trial-by-trial responses of each ensemble in time show their reliable and selective responses to specific stimulus features. (D) The PSTH for each ensemble shows the stimulus preference of the ON and OFF ensembles. The different responses can be classified as transient and sustained, showing detailed stimulus tuning. (E) The stacked trial-by-trial responses of the most correlated core-cell for each ensemble shows that the temporal activity of the ensemble is poorly determined by the activity of their core-cells. (F) For each ensemble, the average PSTHs of their corresponding core-cells grouped by cell type (colors) shows no clear tuning preference. (G) The coloured matrix shows the RGCs in rows and the ensembles in the columns. Each column is colored in the entries corresponding to their core-cells, according to which ensemble they belong (A).