Extended Data Fig. 10: Relationship between tongue angle and neural activity in the movement-null and movement-potent subspaces.
From: Separating cognitive and motor processes in the behaving mouse
a. Projections along movement-potent (top) and movement-null (bottom) components of CDaction. Correct trials shown in solid lines and error trials shown in dashed lines. Shaded region depicts 95% CI of bootstrap distribution. b. Tongue angle for an example session for correct and error trials Black values indicate tongue not visible. c. Tongue angle on correct and error right and left trials. Tongue angle was linearly time warped to allow for averaging over trials and sessions. Mean and s.e.m. across sessions shown. d. Tongue angle (left) and predictions from the full population neural activity (middle left), null subspace activity (middle right), and potent subspace activity (right) for an example session. e. Variance explained (R2) of tongue angle by prediction from movement-null (green) and movement-potent (pink) subspaces. Asterisks denote significant differences between predictions from null and potent subspaces and error bars indicate standard deviation across sessions (p = 2 × 10−8, paired two-sided t-test, n = 25 sessions).
