Alpha-frequency stimulation strengthens coupling between temporal fluctuations in alpha oscillation power and default mode network connectivity

Abstract

Alpha (8-12 Hz) oscillations and default mode network (DMN) activity dominate the brain's intrinsic activity in the temporal and spatial domains, respectively. They are thought to play crucial roles in the spatiotemporal organization of the complex brain system. Relatedly, both have been implicated, often concurrently, in diverse neuropsychiatric disorders, with accruing electroencephalogram/magnetoencephalogram (EEG/MEG) and functional magnetic resonance imaging (fMRI) data linking these two neural activities both at rest and during key cognitive operations. Prominent theories and extant findings thus converge to suggest a mechanistic relationship between alpha oscillations and the DMN. Here, we leveraged simultaneous EEG-fMRI data acquired before and after alpha-frequency transcranial alternating current stimulation (α-tACS) and observed that α-tACS tightened the dynamic coupling between spontaneous fluctuations in alpha power and DMN connectivity (especially, in the posterior DMN, between the posterior cingulate cortex and the bilateral angular gyrus). In comparison, no significant changes were observed for temporal correlations between power in other oscillatory frequencies and connectivity in other major networks. These results thus suggest an inherent coupling between alpha and DMN activity in humans. Importantly, these findings highlight the efficacy of α-tACS in regulating the DMN, a clinically significant network that is challenging to target directly with non-invasive methods.Significance Statement Alpha (8-12 Hz) oscillations and the default mode network (DMN) represent two major intrinsic activities of the brain. Prominent theories and extant findings converge to suggest a mechanistic relationship between alpha oscillations and the DMN. Combining simultaneous electroencephalogram-functional-magnetic-resonance imaging (EEG-fMRI) with alpha-frequency transcranial alternating current stimulation (α-tACS), we demonstrated tightened coupling between alpha oscillations and DMN connectivity. These results lend credence to an inherent alpha-DMN link. Given DMN dysfunctions in multiple major neuropsychiatric conditions, the findings also highlight potential utility of α-tACS in clinical interventions by regulating the DMN.

Figure 1.

Experimental paradigm (A), ROIs (B), and dynamic analysis pipeline (C). A, Experimental design. Participants underwent simultaneous RS EEG–fMRI recordings before and after tACS/sham stimulation. Each recording session lasts 10 min. B, ROIs. Abbreviations: (l/r) AI, (left/right) anterior insula; dACC, dorsal anterior cingulate cortex; (l/r) dlPFC, (left/right) dorsolateral prefrontal cortex; (d/v) PCC, (dorsal/ventral) posterior cingulate cortex; (l/r) PPC, (left/right) posterior parietal cortex; mPFC, medial prefrontal cortex; (l/r) AG, (left/right) angular gyrus. C, Analysis pipeline illustrating the methods used (see text for more details).

Figure 2.

α-tACS strengthened dynamic coupling between DMN connectivity and alpha power. A, Mean changes (Post–Pre) in the dynamic coupling matrix for the active (top right) and sham control (bottom left) groups across subjects. B, C, Boxplots illustrate increased coupling (from Pre to Post) between fluctuation of alpha power and PCC–rAG (B) and PCC–lAG (C) connectivity in the active (vs sham) group. The red- and blue-shaded areas correspond to the mean ± 1.96 × SEM and the mean ± 1.96 × SD, respectively. For the double contrasts, * = p < 0.05 FDR corrected; p < 0.1 FDR corrected. For the follow-up simple contrasts, p < 0.1; *p < 0.05. The dynamic coupling between alpha power and connectivity of DMN with the CEN and SN, and baseline alpha–DMN coupling are shown in Extended Data Figures 2-1 and 2-2, respectively.

Figure 3.

Data from a representative participant. A, B, For each pair of ROIs (here, vPPC and right AG), average BOLD signals were extracted using a sliding Gaussian window of 60 TRs (∼2 min; A). The two BOLD timeseries for each sliding window were then correlated, resulting in a r coefficient (i.e., FC strength). Congregating the r values for all windows (sliding in increments of 1 TR) across the Pre (left) and Post (right) recordings, we obtained the respective FC timeseries (B). C, D, Similarly, we obtained an alpha power timeseries for the Pre (left) and Post (right) sessions (C). Specifically, we extracted alpha power from each sliding window and congregated alpha power from all windows for Pre and Post sessions, respectively (D). Finally, we correlated the timeseries of FC and alpha power for each session (B, C) and obtained an index of dynamic alpha–FC coupling. In this participant, the coupling increased from −0.62 Pre (left) to 0.82 Post (right).

Figure 4.

α-tACS modulation of dynamic coupling across brain networks and frequency bands. A, No effect of α-tACS on the dynamic coupling between DMN connectivity and beta or theta power. Mean differential (Post–Pre) dynamic coupling matrices were comparable between active (top right) and sham (bottom left) groups for beta (left) and theta (right) band power. B, No effect of α-tACS on the dynamic coupling between alpha power and CEN/SN connectivity. Mean differential (Post–Pre) dynamic coupling matrix of connectivity within and between CEN and SN was comparable between active (top right) and sham (bottom left) groups. No effects survived the significance threshold (p < 0.05 FDR corrected). C, Effects of the lag size. P and corrected p values for the effect of tACS on the dynamic coupling of alpha (blue lines) with vPCC–rAG (left) and vPCC–lAG (right) connectivity were consistently significant (gray shaded area) over lags of zero to five TRs. In contrast, beta (red lines) and theta (black lines) showed stable, nonsignificant patterns over the same lags.

Figure 5.

α-tACS strengthened dynamic coupling between adjusted DMN connectivity and alpha power. A, Changes (Post–Pre) in the dynamic coupling matrix for the active (top right) and sham control (bottom left) groups. B–D, Boxplots illustrate increased coupling (from Pre to Post) between fluctuation of alpha power with mPFC–lAG connectivity (B), PCC–rAG connectivity (C) and PCC–lAG connectivity (D) in the active (vs Sham) group. The red- and blue-shaded areas correspond to the mean ± 1.96 × SEM and the mean ± 1.96 × SD, respectively. For the double contrasts, p < 0.1 FDR corrected; *p < 0.05 FDR corrected. **p < 0.01 FDR corrected. For the follow-up simple contrasts, ·p < 0.1 uncorrected; *p < 0.05 uncorrected; **p < 0.01 uncorrected.