Figure 3.

Slow cortical dynamics track tone sequences. A, Analysis schematic. We defined “identical” trials as trials in which an identical tone sequence was presented repeatedly and “similar” trials as trials in which tone sequences were statistically similar (same β and σ) but nonidentical. We identified neural activity encoding tone sequences by comparing MEG activity across-trial correlation/coherence for identical versus similar trials. B, Correlation analysis. Across-subject mean F-values corresponding to the effect of “identical” versus “similar” trial pairs from within-subject ANOVAs using across-trial correlation of the MEG activity time course as the dependent variable. A repeated-measures ANOVA across subjects shows a single significant cluster including all sensors after correction by cluster-based permutation test (p < 0.001, n = 11). Left, Result obtained using full-band (0.05–150 Hz) MEG data. Right, Result obtained using 0.05–3 Hz data. Black dots indicate sensors where tone sequence tracking effect in the full band data was strongest (F > 50). C, Phase dissimilarity analysis. Ci, Group-average phase dissimilarity (difference in phase coherence between identical and similar set of trials) in the <3 Hz band is plotted across sensors. Block dots are the same as in B and were used to define left and right sensor clusters for subsequent analysis. Cii, Phase dissimilarity by frequency for the sensor clusters shown in Ci, as well as for all sensors. Dashed vertical lines indicate tone presentation frequency (3.3 Hz) and its harmonics. Shaded areas denote SEM across subjects. Ciii, Phase coherence across identical (green) and similar (magenta) set of trials, respectively, for the sensor clusters shown in Ci, as well as for all sensors. Peaks in the plots correspond to tone presentation frequency and its harmonics.