Linear encoding model reveals similar single neuron activity profiles between visually guided and visually independent decision tasks.

(a) Firing rate distribution of single units recorded in visually independent decision task. (b) Distribution of maximum relative contribution of a single regressor to single neuron activity in the visually independent decision task, by epoch. The same cutoff threshold separating “specialized” from “mixed” neurons as in the visual decision task is shown in shaded regions. (c) Proportions of cells with “specialized” vs. “mixed” selectivity profiles in the visually independent task, as classified using the threshold in (a). (d) Proportion of variance explained by linear encoding model in the visually independent task, across behavioral epochs. (e) Comparison of variance explained by linear model between visual decision task vs. visually independent decision task, across behavioral epochs. Points indicate mean, error bars indicate standard deviation. Median variance explained is slightly, but not significantly, higher in the visual decision task than in the visually independent decision task within each epoch (hierarchical permutation test, see Methods, all p > 0.05). (f) Measure of cluster stability (adjusted Rand Index) when clustering single neuron feature encoding profiles between pairs of epochs, compared to stability over independent partitions in the same epoch (diagonal). (g, h) Pairwise correlation structure between regressors in (g) visual decision task and (h) visually independent task. Regressors are as follows: (1) stimulus, (2) choice, (3) reaction time, (4) movement latency, (5) correctness, (6) reward delivery, (7) previous trial last port visited, (8) previous trial choice, (9) previous trial outcome, and (10) previous trial stimulus. The underlying data for this figure are available for download from 10.17632/5ms7gcb67j.1.

(TIF)