Fig. 3: Temporal structure of the reconstructed representations.
From: Representation of probabilistic outcomes during risky decision-making
a Autocorrelation of the reconstructed probabilities (dark blue line) and its s.e.m. (light blue shade) obtained from n = 21 participants, against the mean autocorrelation obtained by repeating the entire analysis (n = 21) 100 times after randomly permuting the class labels (grey transparent lines). Stars indicate the time lags at which the autocorrelation differed from chance (two-sided cluster-level permutation test, p < 0.010). b Distribution of the duration of steady outcome representation epochs (blue) against the null distribution derived from permuted classifiers (red; two-sample Kolmogorov–Smirnov test, p < 0.001). c Consistently with the epoch durations in b, number of transitions between most likely outcome over the 1500 ms deliberation interval is lower than chance (permutation test, p < 0.010). The error bar represents the s.e.m. of the number of transitions for the correct classifier.
