Figure 7. Accurate decoding of the dorsostriatal brain region identity in early training but not after extended training.

(A) Mean decoding accuracy (± standard deviation) of the dorsostriatal subregion identity, DLS or DMS, of individual neurons based on their sequence-related activity (True - red) is compared with the decoding accuracy expected from chance (Shuffled - dark blue) across early training sessions (left) or during extended training, as a function of ensemble size (right). (B) Decoding based on the first three principal components following principal component analysis. Average accuracy (± standard deviation) is presented as a function of early training sessions (left) or ensemble size during extended training (right).

Figure 7—figure supplement 1. Principal Component Analysis (PCA) followed by Linear Discriminant Analysis (LDA) allows prediction of brain region identity in early training but not extended training.

(A) Decoding accuracy of brain region identity as a function of the number of principal components (PCs) included in the analysis, during four selected early training sessions. Mean accuracy (± s.d.) in each condition was compared to chance (blue – shuffled condition for session 10) (B) Decoding accuracy of brain region identity as a function of the number of principal components (PCs) included in the analysis, and across different ensemble sizes, during extended training. Mean accuracy (± standard deviation) in each condition was compared to chance (blue – shuffled condition for 600 units).