Fig. 3: Decreased neuronal dimensionality and trial variability during learning.

a Diagram of neuron variance space. PCA was performed with neurons as variables. b Percentage of principle components (PCs) that explains 70% of variance in neuron subspace, over the trial structure. Vertical dashed lines indicate task windows. Colors represent learning phases. c Quantification of the percentage of PCs in neuron space during each task window. Colors represent task window. d Diagram of trial variance space. PCA was performed with trials as variables. e Percentage of PCs that explains 70% of variance in trial space, during the trial time. Vertical dashed lines indicate task windows. Colors represent learning phases. f Quantification of the percentage of PCs in trial space, during each task window. Colors represent task window. g Discrimination index of the top 5 PCs during each task window, regarding the corresponding task variables. Line color saturation represents the PC number. Naïve, N; learning; L, expert, E. (S1: 13 mice, 130 [N], 183 [L], 138 [E] sessions; PPC-RL: 13 mice, 70, 74, 98 sessions; PPC-A: 9 mice, 60, 109, 40 sessions; sessions with less than 30 neurons on average across the resampling procedure were excluded; *p < 0.05, **p < 0.01, ***p < 0.001; 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.