Fig. 4: Improved task representation in S1 and PPC encoding subspaces during learning.

a Diagram of generating encoding axis, within-area interaction axis, and inter-area interaction axes in the same reduced 30-dimensional neural space for each subpopulation. b Example encoding axis scheme for the texture axis. Colored dots represent neural activity in texture 1 and texture 2 trials, during the texture window. Crosses represent the means of responses for the two texture types. c Discrimination index of population activity projection on the encoding axes, for each task variable, in S1, PPC-RL, and PPC-A. Gray lines represent confidence intervals (CI) of shuffled data. d Color-coded pairwise similarity of encoding axes for all task variable combinations. Symbols on the x- and y-axis indicate task windows. Red and black asterisks indicate that the data from the expert condition is significantly higher or lower than the naive condition. e Quantification of pairwise encoding axes similarity. (S1: 13 mice, 122, 171, 119 sessions [naïve, N; learning, L; expert, E]; PPC-RL: 13 mice, 66, 72, 86 sessions; PPC-A: 9 mice, 56, 99, 33 sessions; (c and e): *p < 0.05, **p < 0.01, ***p < 0.001; (d): *p < 0.05; two-sided Wilcoxon rank-sum test against naive condition; p-values were corrected for multiple comparisons using FDR method; mean ± SEM). Source data are provided as a Source Data file.