Extended Data Fig. 2: Granular movement tuning of SNr firing changes.

a, Scatter plot displaying the fraction of positively versus that of negatively modulated neurons for each task time window (Fig. 2a) across the dataset (left) and in each single mouse (right). The median fraction of positively modulated units (cyan vertical line) is higher than that of negatively modulated units (magenta horizontal line) for each behaviour and across mice. b, Heatmaps displaying the computed task time window modulation (left) and z-scored average firing rate aligned to the onset of retraction of SNr neurons with positive task modulation (right). Neurons were sorted based on their window of maximum positive modulation in the modulation heatmap. Note that increases in spike rate in heatmap (yellow) tile movement execution. c, Heatmaps displaying the z-scored average firing rate of SNr neurons gathered from 2 single mice (mouse 1 and 2), with negative task modulation aligned to the onset of retraction and sorted by trough time. Note decreases in spike rate (blue) tiling movement execution similar to the full dataset across all mice. d, Single neuron modulation to task time windows (top; color code and sequence of boxes as in Fig. 2b), single trial raster plots (middle) and average firing rate with standard error of the mean (bottom) of SNr neurons (additional examples to the three units shown in Fig. 2, except for rightmost plot aligned to manipulation start) aligned to reach start, retract start or manipulation start. Each of the neurons displays a very precise pause in firing aligned with specific task time windows. Note for neuron 3, its negative modulation during manipulation, paralleling its pause aligned to retraction start. e, (left) Single trial raster plots (top) and average firing rate with standard error of the mean (bottom) of the same SNr neurons (7) and (3) as displayed in Fig. 2a, Extended Data Fig. 2d, aligned to the second retraction start in trials where two reaches were executed in a 0.3-0.6 s interval. The pauses in firing repeat twice as the movements are executed twice. (right) Heatmap displaying the z-scored average firing rate aligned to the onset of retraction of SNr neurons with positive task modulation. Neurons were sorted as in Extended Data Fig. 2b. Note that increases related to reaching and retraction repeat twice as movements are executed twice while handling-related spiking rate changes are absent since mice do not collect the food pellet in these trials. f, (left) Heatmaps displaying pairwise correlation of average neuronal activity of tonically firing SNr neurons from one mouse in 4 s windows around retraction start in isolated trials (left), first retraction start in dual reach trials (centre-left), second retraction start in dual reach trials (centre-right) and 100 random timestamps (right). (right) Bar plot (mean and standard error) displaying the correlation slope and R value resulting from regressing the correlation matrix for isolated retraction start and second retraction start in dual reach trials compared to those resulting from correlating the correlation matrix for isolated retraction start and random timestamps (n = 17 mice). Note preservation of correlations in neuronal population activity when aligning to retraction in isolated trials or upon repetitions of the movement and their disappearance when considering random timestamps indicating the movement-related nature of these firing patterns.

Source Data