Fig. 3: Sequential decodability of cognitive contributors to information gathering from whole-brain MEG activity.
From: Indecision and recency-weighted evidence integration in non-clinical and clinical settings
a, Cognitive factors are predicted from MEG activity using lasso regression following an iterative method described previously57 (N = 105). The decoded variables of current number of draws, horizon condition, cumulative evidence strength from the start of the game until the previous draw ESd−1 and ΔES were all significantly decodable (P < 0.05, two-sided cluster-based permutation test at threshold t > 3.1; denoted by horizontal lines). Peak decodability timepoints, quantified as the group average of individual maximum values, are indicated by lollipops. Correlation values were normalized by dividing these by the maximum correlation value for visualization purposes. Note that all statistical tests were performed before normalization. Shaded areas represent the standard error. b, Maximum decodability is achieved earlier in the trial for factors independent of the currently presented stimulus (current number of draws, horizon condition, the cumulative evidence from the start of the game until the previous draw) than for those dependent on current evidence processing (ΔES, all P < 0.001, N = 105). Boxplots show the median (centre line) and IQR; whiskers extend to the largest value within 1.5× IQR. Pairwise t-tests are one-sided and not corrected for multiple comparisons.
