Extended Data Fig. 4: Relationships of dopamine signaling with task-independent locomotion cannot account for trajectory error encoding.

a, Average acceleration coefficients during running in directions contralateral (left) and ipsilateral (right) to the implant in a model describing ΔF/F as a function of continuous acceleration at different time lags during the intertrial interval (see Methods). Acceleration was calculated from the absolute value of the angular velocity for contralateral and ipsilateral running periods independently. Thin lines represent averages for each mouse, thick black line, average across mice (n = 193 sites, 6 mice) b, Left, average signed angular acceleration (+contralateral, − ipsilateral) aligned to cue onset on congruent (green) and incongruent (purple) contralateral cue trials, split into tertiles (dark to light colours indicates fast to slow) for one mouse. Accelerations on incongruent trials are oppositely signed for ipsilateral and contralateral trials, but TE coefficients are positive for both trial types (Extended Data Fig. 3g). c, Average absolute value of angular treadmill velocity aligned to cue onset across all trials (left) and trials where running was contralateral (middle) and ipsilateral (right) to the implant at cue onset during a non-directional, Pavlovian delay conditioning task (top left inset, see Methods). Trial averages are split into tertiles within 0.3 s before cue onset (dark to light colours indicate fast to slow trials). d, Average cue onset aligned ΔF/F for a single recording site on the trial types indicated in c. e, Violin plots showing maximum and minimum angular velocity coefficients across all sites for the trial types in c. For the boxplot, the centre point is the median, the lower and upper bounds are the first and third quartiles, and the whiskers extend from the box bounds to the most extreme value no further than 1.5 x interquartile range from the bounds. Each data point represents a single site. Inset text, proportion of fibers significant (two-sided t-test on model coefficients, p < 0.05 for 3 timepoints in a row within 1 s cue window, Bonferroni corrected). Shaded regions in all plots, 95% confidence intervals.