Extended Data Fig. 3: Movement-contingent SNr firing changes.

a, Alternative hypotheses for multiphasic modulation of SNr neurons in the form of schematic neuronal activity traces. As movements are executed (A and B), consider a bidirectionally modulated neuron increasing for movement A and pausing its firing subsequently as movement B is executed (left). In Model 1 (middle), as the duration of movement A is altered (e.g. duration), the entire dynamical activity pattern could change, representing a different bound action. Similarly, when movement A is executed without movement B (ablation), the entire dynamical activity pattern could be altered to reflect the difference in the bound action or the different phases of the activity could occur similarly as if the activity pattern were temporally bound and the sequence of activity changes represented a bound movement sequence. Alternatively, in Model 2 (right), each of the phases of the activity pattern would be linked to the execution of particular movements, and hence, specifically increasing the duration of movement A would widen the activity phase linked to the execution of the movement A and leave the modulation during an unaffected movement B unchanged. Similarly, when movement B is specifically ablated from the sequence of movements being executed, and A is unchanged, the pattern of neural activity linked to the execution of movement B would be absent, while leaving dynamics related to movement A unchanged. b, Median distance from the slit at the start of reach (top) and start of retraction (bottom) for long and short duration reaches (n = 17 mice), and their probability density reflecting no structured covariation between the distance from slit and reach durations taken across mice (left). Comparison of median reaching velocity for short and long duration reaches revealed no consistent trend in reaching hand speed across mice (n = 17) (right). c, Relative median timing of task events and relative probability densities for aborted vs complete reaches revealing similar probability densities of reach starts (left) and similar velocities of the reaching hand (right) for both trial types across mice (n = 17). d, Relative median timing of task events and relative probability densities for reaches followed by handling and those without handling revealing similar probability densities of reach starts for both trial types across mice (n = 17). e, Perievent time histograms of three example SNr neurons (left, middle and right) aligned to retraction start on respective movement variation trials in cyan and magenta (short and long reach, reach and aborted reach, handle and no handle trials). f-h, Z-scored average firing rate of the three SNr neuronal populations aligned to retract start in the movement variation trials in cyan and magenta, overlaid with 99.9% confidence interval of the distribution of average firing rates computed over shuffled trial pairs in grey: short and long reach (left), reach and aborted reach (middle), handle and no handle trials (right). Shaded rectangles represent intervals of significant difference from the null distribution. Note, in addition to the changes in movement related pauses in the respective variations, the same neuronal populations increase in firing in a movement dependent manner as select other movements are executed. i, Z-scored average firing rate of the SNr neuronal population negatively modulated during retraction/grasping/handling aligned to retract start on respective movement variation trials in cyan and magenta, overlaid with 99.9% confidence interval of the distribution of average firing rates computed over shuffled trial pairs in grey: short and long reach (left), reach and aborted reach (middle), handle and no handle trials (right). Shaded rectangles represent intervals of significant difference from the null distribution. Note the increase in firing rate of this population of neurons during reaching showed an increase in its duration as the duration of the reach was increased, while the other phases of the modulation were unaffected. Similarly, the modulation at the onset of retraction was significantly different between the complete and abbreviated reaches as described in Fig. 3. Further, the negative modulation during handling was also contingent on whether the mice handled the food pellet subsequent to reaching.

Source Data