Fig. 2: Reorganization of layer 2-3 PN activity in M1 during motor learning at the single neuron and the population levels.

a Population data showing average ±SEM activity (across cells, time, and trials) for control (n = 12) and manipulated (n = 10) mice. b Examples of individual neuron activity vs. time, averaged across consecutive trials per training session (1-7). Color encodes the percent change in fluorescence (ΔF/F). c Example traces of the mean activity along the trial time (average across cells and trials) per training session. Same mice as in B. d Population data showing histograms of correlation values between the average activity trace of each cell per session and its own average trace in session 7. Gray dashed vertical line indicates correlation = 0.5. e Fraction of cells having correlation >0.5 with their average activity trace in session 7 (corresponding to data in e). f Fraction of trials being classified as related to the 7^th day in training by a binary classifier trained to separate ensemble activity of the 1^st day (beginner profile) from the 7^th day profile, trained using activity profile averaged across a 1 sec time window centered at t = 1 sec. g Heat map showing probability for trials to be classified as related to the 7^th day (vs. the 1^st) based on ensemble activity for a sliding window analysis (window size 1 sec, window hop 0.5 sec), n = 12 for controls and n = 10 for manipulated mice. Results for sessions 1 and 7 are a product of 10-fold cross-validation. For days 2-6 results are a product of the trained models. Scale bar represents fraction of trials classified as 7^th session. For Fig. 2a, e, and f, dots are population data (per mice), orange and blue lines are mean ( ± SEM) across mice, n = 12 for controls and n = 10 for manipulated mice, correspondingly. Shaded data in the blue box indicate the manipulated session. Source data are provided as a Source Data file.