(A) Setup for eyeblink conditioning in head-fixed mice on a running wheel, showing an LED as the CS and an air-puff US. (B) Sagittal sections showing the complete elimination of CB1R (red) in CB1KOs and DAPI staining the granule-cell layer (blue). Images of the cerebellum of a representative control (upper panel) and global CB1 knockout (bottom panel). (C) Average %CR learning curves of global CB1 knockout mice (CB1KO, red, N = 11) and their littermate controls (WT, black, N = 12). Error bars indicate SEM. (D) Average running distance of WT (black, N = 12) and CB1KO (red, N = 11) on a self-paced treadmill across all learning sessions. Global CB1 knockout mice were significantly hypoactive (*p=0.015). Box indicates median and 25th to 75th percentiles, whiskers extend to the most extreme data points. (E) Onset session of learning for CB1KO (red) and littermate controls (black), divided into top and bottom runners (open and filled circles, respectively), plotted against the animals’ mean walking distance, averaged across all 20 training sessions. Vertical dashed line indicates the threshold used for dividing the control animals (150 meters/session, on average). Onset session was defined as the session in which the average CR amplitude exceeded 0.1. Each dot represents an animal. Lines are linear robust fits for CB1KO (red, slope = −0.05, *p=0.02) and controls (gray, slope = −0.04, *p=0.03). (F) Average %CR learning curves of CB1KO (red line, N = 11) compared with wildtype littermates with comparable (solid black line, N = 6) or increased locomotor activity levels (dashed black line, N = 6). Error bars indicate SEM. (G) Average eyelid traces of CS-only trials from the last two training sessions for CB1KO (red line, N = 11), hypoactive wildtype animals (solid black line, N = 6) and littermate controls with increased locomotor activity (dashed black line, N = 6). Vertical dashed line represents the time that the US would have been expected on CS+US trials. Shadows indicate SEM. (H) Trial-to-trial correlation between eyelid response amplitude and walking speed. Amplitudes for all trials from six training sessions following learning onset (defined as in (E)) are plotted with lines representing averages across control (black) and global knockout (red) animals; shadows indicate SEM. There was a linear positive relationship for both controls (N = 12, F(1,65.9) = 34.7, ***p=1.5e-07) and global knockouts (N = 11, F(1,43.5) = 43.4, ***p=4.9e-08). Histograms indicate the relative % of trials (averaged across animals) from each genotype that fell in each speed bin. (I) Average amplitude learning curves including only stationary trials of control (WT, black, N = 12) and CB1KO mice (red, N = 11), aligned for each animal’s onset session (defined as in (E)). Error bars indicate SEM. (J) Average eyelid traces of stationary (<0.05 m/s) CS-only trials for control (black) and CB1KO (red) animals, from same sessions in (I). Shadows indicate SEM. Vertical dashed line represents the time that the US would have been expected on CS+US trials. (K) Mean eyelid amplitudes from stationary trials from same sessions as in (I). There was no significant difference in the average amplitude of control (black, N = 12) vs. CB1KO (red, N = 11) animals (p=0.58). Dots represent individual animals. Box indicates median and 25th to 75th percentiles, whiskers extend to the most extreme data points.