To statistically assess the degree of fluctuation of movement-related PAC for each task quantitatively, we first computed the coefficient of variance (CoV) across the time series. A two-way mixed ANOVA test (CoV ∼ group × task) showed a significant interaction [F(3,111) = 3.07, P = 0.031], which indicated that PAC varies differently between groups. In pressing, post hoc testing revealed that the fluctuation was stronger for both patients and controls compared with the resting state (Fig. 4B). In slow tapping, the fluctuation was only larger than in the resting state in controls before FDR correction (Wilcoxon signed-rank, P = 0.033). Since the resolution of the PAC calculation does not permit assessment of modulation across very short movement cycles, fluctuation of PAC in fast tapping task (tapping rate around 4 Hz) across a movement cycle must be interpreted with caution. The larger time-series variance of PAC in pressing and slow tapping tasks compared with the resting state of controls may hint at the possibility that PAC may be modulated at particular movement events. However, direct comparisons of CoV between patients and
