Phase synchrony among neuronal oscillations in the human cortex

Abstract

Synchronization of neuronal activity, often associated with network oscillations, is thought to provide a means for integrating anatomically distributed processing in the brain. Neuronal processing, however, involves simultaneous oscillations in various frequency bands. The mechanisms involved in the integration of such spectrally distributed processing have remained enigmatic. We demonstrate, using magnetoencephalography, that robust cross-frequency phase synchrony is present in the human cortex among oscillations with frequencies from 3 to 80 Hz. Continuous mental arithmetic tasks demanding the retention and summation of items in the working memory enhanced the cross-frequency phase synchrony among alpha (approximately 10 Hz), beta (approximately 20 Hz), and gamma (approximately 30-40 Hz) oscillations. These tasks also enhanced the "classical" within-frequency synchrony in these frequency bands, but the spatial patterns of alpha, beta, and gamma synchronies were distinct and, furthermore, separate from the patterns of cross-frequency phase synchrony. Interestingly, an increase in task load resulted in an enhancement of phase synchrony that was most prominent between gamma- and alpha-band oscillations. These data indicate that cross-frequency phase synchrony is a salient characteristic of ongoing activity in the human cortex and that it is modulated by cognitive task demands. The enhancement of cross-frequency phase synchrony among functionally and spatially distinct networks during mental arithmetic tasks posits it as a candidate mechanism for the integration of spectrally distributed processing.

Figure 2.

Mental arithmetic tasks strengthen local CF phase synchrony. A, Schematic illustration of the tasks R, T1, and T2 used in the study (see Materials and Methods). B, Relative changes in the PLF_1:m (color scale) are averaged over gradiometers and subjects (right panels). Asterisks indicate the statistical significance of changes in the PLF_1:m (left of/in each element) and in the percentages of statistically significant channels (right of/in, as in Fig. 1 E). The topographies of the most prominent changes in PLF_1:m (color scale) and their statistical significances (transparency levels) are shown in the left panels. Note that p_bin was <0.02 for finding statistically significant (p = 0.01) task effects by chance in more than three gradiometer locations. NS, Not significant.

Figure 1.

Local CF phase synchrony in the human cortex recorded with magnetoencephalography. Aa, The gradiometer signal (black) over the left occipitoparietal region shows a burst of ∼10 Hz oscillations (blue) and intermittent 30-40 Hz oscillations (red). Ab, Amplitude spectrum of the gradiometer signal in Aa. B, Oscillation phases (red and blue; normalized with 1/π) and their 1:3 phase difference (purple). C, 1:3 Phase synchrony, quantified in sliding windows of 500 ms. The green line indicates 95% confidence limits. The gray arrows mark a period of prominent synchrony. D, 1:3 Phase synchrony occurred intermittently with 1:3 phase synchrony. E, 1:m Phase synchrony mapped with Morlet wavelets over the 20 min recording sessions. The color scale indicates the percentage of gradiometers detecting statistically significant (p < 0.01) synchrony between oscillations at frequencies f_x and f_y (f_y = f_x/m) averaged over subjects (N = 17). The lower end of the color scale is set by the binomial probability (p_bin) that is <0.02 for finding by chance >3% of the gradiometers to show statistically significant synchrony. F, Topographies of β-α (left) and γ-α (right) phase synchronies. The color scale indicates the normalized (Norm.) phase-locking factor, averaged over subjects. The levels of transparency indicate p_bin for finding by chance the proportion of subjects with a statistically significant synchrony [note that p_bin < 0.0001 corresponds to p_bin,B < 0.01 with Bonferroni's correction for the number of gradiometers (102)]. NS, Not significant. The MEG helmet covering the whole scalp is flattened, and anterior direction is upward.

Figure 3.

CF phase synchrony and CF amplitude correlations. A, Phase synchrony (1:m) mapped with broadband filtering and Hilbert transform (see Materials and Methods) over the 20 min recording sessions (compare with Fig. 1 E). The color scale indicates the percentage of gradiometers detecting significant (p < 0.01) synchrony. B, Task effects on the PLF_1:m obtained with broadband filtering and the Hilbert transform (color scale), averaged over gradiometers and subjects (compare with Fig. 2 B). Asterisks indicate the statistical significance of changes in the PLF_1:m (left to/) and in the percentages of statistically significant channels (right to/). C, CF amplitude correlations within gradiometers quantified with cross-amplitude histograms and averaged over gradiometers and subjects. D, The task effects on CF amplitude correlations and the significance levels (compare with Fig. 2 B).

Figure 4.

A, B, Interareal phase synchrony within (A) and between (B) α, β, and γ oscillations in R. The color scale of the matrices indicates the normalized (Norm.) PLF_1:m (averaged over subjects) for each pair of gradiometers that are on the x- and y-axes. The gradiometers are ordered so that the axes run from the most anterior (A) to the most posterior (P) left-hemispheric gradiometer and then conversely for the right hemispheric gradiometers. The levels of transparency indicate the binomial probability (p_bin) for finding by chance the observed number of subjects with a statistically significant synchrony (as in Fig. 1 F). The lines over the MEG helmet topographies indicate the 500 gradiometer pairs with strongest synchrony for each distance scale, averaged over subjects (see Materials and Methods). The underlying maps show the topography for the mean phase synchrony of each gradiometer with all other gradiometers. The strengths of all synchronies during R decayed exponentially with increasing distance (data not shown), and all synchronies were also stronger within than between the hemispheres. NS, Not significant.

Figure 5.

Task effects on interareal phase synchronies. A-D, Lines show the 500 gradiometer pairs with the most statistically significant task effects (for complete data, see Fig. 6). The underlying topographical maps show the task effects on the mean phase synchrony for each gradiometer (as in Fig. 4). The levels of transparency indicate the statistical significance of the task effects on mean phase synchrony. Note that there were also highly significant differences between T1 and T2 (C, D), although they were not significant in the topographies of mean phase synchrony (Figs. 6, 8). NS, Not significant.

Figure 6.

Task effects on interareal phase synchronies. A, C, E, G, Matrices showing the task effects on PLF_1:m for all gradiometer pairs are ordered as in Figure 4. The color scale indicates the magnitude, and the levels of transparency indicate the statistical significance of task effects. B, D, F, H, The contribution of spurious significances to data in A, C, E, and G. Binomial statistics were used to evaluate the probability of finding by chance the number of statistically significant gradiometer pairs in the corresponding matrices. The numeric values indicate the exponent a for Bonferroni-corrected (16 matrix elements) p_B,bin < 10^a.(a values more than -2 were not considered significant; a values less than -53 are marked with s.) The underlying color map shows the magnitude of the task effect, averaged over the respective gradiometer pairs. A, Anterior; P, posterior; LA, left anterior; LP, left posterior; RA, right anterior; RP, right posterior; NS, Not significant.

Figure 7.

Task effects on interareal phase synchronies as a function of gradiometer pair separation; T1/T2 versus R. To corroborate and extend the findings in Figures 5 and 6, we quantified the task effects on phase synchrony within and between the gradiometer groups in the four quadrants of the MEG helmet. The lines show the sliding average PLF_1:m (y-axis) of 125 gradiometer pairs as a function of their mean pair separation (see x-axis). A, B, The lines are colored so that the right posterior (RP) quadrant is marked with red, the left posterior (LP) quadrant is marked with yellow, the right anterior (RA) quadrant is marked with blue, and the left anterior (LA) quadrant is marked with green. Lines with a single color show the task effects on synchrony within the quadrant corresponding to that color. Lines with two colors show the task effects on synchrony between the two quadrants indicated by the colors. The horizontal bars indicate the significance level of the task effect; light color, p < 0.05; medium color, p < 0.01; dark color, p < 0.001. Note that p < 0.01 corresponds to p_B < 0.05 after Bonferroni's correction with the number of independent averaging windows (5).

Figure 8.

Task effects on interareal phase synchronies as a function of gradiometer pair separation; T1 versus T2. Line colors, scales, and statistics areas in Figure 7. A, Phase synchronies (1:1). B, Phase synchronies (1:2 and 1:3). LA, Left anterior; LP, left posterior; RA, right anterior; RP, right posterior.

Figure 9.

Mental arithmetic tasks enhance frontal 5-8 Hz and posterior 8-40 Hz oscillations and attenuate the posterior 3-8 Hz oscillations. A, The traces show the wavelet amplitudes (for frequencies in Figs. 1 and 2) averaged within recording sessions and across the gradiometers and subjects (R, blue; T1, green; T2, red). B, C, Topographies of the oscillation amplitudes in R (B) and the topographies and significance levels of task effects (C) averaged over the frequency bands that are indicated by the black horizontal bars. NS, Not significant.