Methods followed those of our previous work on oscillatory bursts (Lundqvist et al., 2016). Power was computed from a wavelet transform, as described above, and pooled within several frequency ‘‘sub-bands’’: low-beta (10–20 Hz), high-beta (20–32 Hz), low-gamma (40–65 Hz), gamma (55–90 Hz), and high-gamma (70–100 Hz). Within each sub-band, the mean and SD of power across all trials and within-trial time points was computed and used to threshold the data at 2 SDs above the mean. Bursts were defined as periods where power remained above threshold for at least 3 oscillatory cycles (defined at the center frequency of each sub-band). This resulted in a binary representation of time points within versus outside bursts. These were pooled separately within all beta and gamma sub-bands by labeling as bursts time points where a burst occurred within any sub-band (logical OR across sub-bands). Burst rate for each task condition, band, and time point was computed as the proportion of trials containing within-burst time points. Finally, the difference in burst rate at each time point from the fixation baseline was computed and plotted in Figure S2.
