Fig. 6: Performance-based network connectivity supports learning strategy.

a Average normalized power for channels in the right intraparietal sulcus (IPS R) and right middle temporal gyrus (MTG R) of subject 6 across trials. The value for each trial comes from averaging the normalized power in the time-frequency windows identified by the non-parametric cluster statistic (Table 2 and Table 3). This figure uses the results of the error state. b Matrix of channel connectivity strengths found by taking the magnitude of two-tailed Spearman’s correlation value between the average normalized power of channels in IPS R and MTG R. Values close to 1 (red) indicate high correlations and values close to 0 (yellow) indicate low correlations. c Comparison between subject connectivity strength (found by taking the average of the channel connectivity strength) and session performance reveals high-performing subjects engaged in stronger connectivity between IPS R and MTG R (two-tailed Pearson correlation: r = 0.96, p = 0.17). The correlation value represents the performance connectivity strength. Directed graph depicting performance connectivity strength across regions for (d) error state and (e) perturbed state. The regions are colored by the large-scale brain network they belong to: dorsal attention (DAN) in dark blue, default (DN) in red, ventral attention (VAN) in light blue, and visual in yellow. Only pairs with positive relationships between connectivity strength and session performance are shown (i.e., higher performance with higher subject connectivity). The magnitude of the correlation between connectivity strength and session performance is depicted by the thickness of the edge; the higher the correlation, the thicker the edge. Values below 0.5 were truncated. The regions are ordered by when they encode each state and its phase is labeled along the circumference, demonstrating synchrony across time. See Supplementary Fig. 9 for full performance connectivity. Source data are provided as a Source Data file.