High-definition tDCS of the motor cortex affects prefrontal and primary motor activity differently in young and older adults

Abstract

The ability to suppress irrelevant or distracting inputs that interfere with goal-driven behavior is known to decline with increasing age. Although noninvasive stimulation of the motor cortices has been shown to modulate age-related changes in motor activity, the findings remain preliminary, and it is unknown whether this effect extends beyond the motor cortex. In this study, 125 healthy adults, categorized into young (20–35 years) and older groups (55–72 years) underwent three visits (i.e., anodal, cathodal, and sham). During each visit, they received 20 min of high‐definition transcranial direct current stimulation (HD‐tDCS) applied to their left primary motor cortex (M1) and completed a flanker task during high-density magnetoencephalography (MEG). Statistically significant oscillatory responses were imaged and analyzed using voxel-wise, whole-brain, and point of stimulation approaches. Our results showed increased gamma flanker interference effects within the contralateral M1 in older relative to younger adults following anodal stimulation, and after anodal compared to cathodal HD-tDCS in older adults. We also found polarity-based differences in beta and gamma M1-prefrontal connectivity as a function of age group. Critically, these data indicate distinct spectrally- and polarity-dependent effects of M1 HD-tDCS on the local and network-level neurophysiological responses serving motor performance in young versus older healthy adults.

Introduction

Effective movement execution in the context of response competition is orchestrated through the coordination of motor oscillatory responses in primary cortices, guided by higher cognitive inputs from frontal regions such as the dorsolateral prefrontal cortices (DLPFC) and inferior frontal gyri (IFG) [1, 2]. Both cognitive and motor functioning tend to decline with age, leading to functional dependencies in older adults. The age-related neurophysiological changes that accompany these declines have been observed in the primary motor cortices, as well as higher cognitive hubs [1, 3]. To mitigate such changes, noninvasive brain stimulation interventions like transcranial direct current stimulation (tDCS) are being widely studied to determine their potential in rehabilitation. While many mechanistic details of tDCS are not fully understood, the stimulation is thought to modulate neural responses by altering resting membrane potential [4], and neurotransmitter levels, particularly gamma-aminobutyric acid (GABA) and glutamate [5]. Given its noninvasive nature, low-cost, and ease of use, interest in tDCS has markedly grown with many teams examining the key parameters of its application. For example, recently there has been a marked shift from conventional sponge electrodes to high-definition tDCS (HD-tDCS) to achieve higher spatial precision [6]. Besides electrode configuration, other methodological specifics such as current intensity, duration, and state dependency are among the key factors under study that are known to shape tDCS effects and these may be the cause of the substantial variability in the current brain stimulation literature.

A handful of studies have paired tDCS with cognitive and motor tasks to assess its effects in older adults, but the findings remain heterogeneous [7]. For instance, a recent study combined motor tDCS with transcranial magnetic stimulation to assess age-related neuroplasticity and found evidence for the characteristic anodal-excitation/cathodal-inhibition dichotomy [8]. Another study reported that anodal tDCS of the pre-supplementary motor area resulted in enhanced inhibitory control in older adults [9]. However, a subsequent motor tDCS aging study used titrated intensities (1, 2, and 3 mA) and found no significant effects of stimulation on motor learning [10]. There is also clinical trial work showing that tDCS therapy in older adults leads to improved gait and cognitive function, as well as a reduction in dual-task costs when standing or walking [11, 12]. Of note, none of these studies examined the underlying neural responses, thus the origins of any stimulation effects remain unaddressed. Another concern, commonly seen in studies using conventional tDCS, is the use of a contralateral supraorbital electrode as a reference. This can lead to bidirectional modulation, thereby complicating the assessment of regional stimulation effects on neural activity. This becomes particularly challenging in aging studies, where age-related hemispheric asymmetry reduction and bilateral/contralateral hemispheric compensatory activations are commonly seen [13].

Regarding the neural origins, the planning and execution of voluntary actions are supported by neural population-level oscillations in the beta (~ 15–30 Hz) and gamma bands (~ 65–90 Hz) [14–19]. Specifically, several hundred milliseconds before movement onset, a robust decrease in beta power relative to baseline, termed an event-related desynchronization (ERD), emerges in the primary motor cortices [15, 17–22]. This so-called peri-movement beta ERD has long been associated with both motor planning and execution, and is known to be modulated by the certainty of the motor plan and other factors [15, 23–27]. There is also a transient increase in gamma activity that coincides with movement onset [28]. This response has generally been referred to as the movement-related gamma synchronization (MRGS), and it is strongest in the motor cortex contralateral to movement [14, 17, 29, 30]. Since the MRGS is tightly locked to movement onset, it was initially thought to specifically reflect the motor execution signal [31, 32], but later research has extended its role to higher-order cognitive control and attentional processes [33–36]. Finally, following movement termination, there is a sharp increase in beta power referred to as the post-movement beta rebound (PMBR) [37, 38], which is thought to reflect sensory feedback following movement completion.

As mentioned above, studies using motor tasks that involve cognitive interference have shown that both the beta ERD and the MRGS responses are sensitive to such parameters, even when the output is a simple movement (e.g., finger tap) [23, 34, 35, 39]. This suggests that these neural responses are likely much more than a simple movement execution signal. In addition, recent work has shown that these effects of cognitive interference on motor-related oscillations are accentuated with aging, particularly in higher-order brain regions that are known to interact with M1 during motor performance [40]. While groundbreaking, these aging findings did not make direct connections with the extensive literature showing that aging affects multiple parameters of motor-related oscillations, including the strength of the beta ERD response during motor performance [41, 42]. Thus, whether these well-known effects of aging on cortical motor oscillations underlie the sensitivity to cognitive interference during motor tasks, or whether these effects merely reflect aging-related changes in higher-order cortices, remains poorly understood. Regardless of the precise mechanism, the literature broadly supports that aging leads to a slowing of behavioral responses, including increased sensitivity to interference, and alterations in cortical oscillatory activity in M1 and higher-order areas known to be involved in movement execution. While such declines with aging are well appreciated, effective therapeutic approaches to offset such changes remain limited and are a major goal of clinical neuroscience research in the context of aging.

In the current study, we aimed to quantify the offline effects of HD-tDCS of the left primary motor cortex (M1) on the neural responses serving motor planning and execution during response competition in young versus older adults using magnetoencephalography (MEG). Our approach enables the polarity-specific effects of motor HD-tDCS on systems-level motor dynamics to be probed in young versus older adults with high spatiotemporal precision. Based on the existing aging literature, we hypothesized that cortical motor stimulation would alter local oscillations in the beta and gamma range, as well as connectivity among M1 and higher-order brain regions [43]. In addition, we expected to observe elevated spontaneous beta activity in the M1 cortices of older compared to younger adults, consistent with multiple previous reports [41, 42], but did not have strong directional hypotheses regarding the impact of HD-tDCS on spontaneous activity.

Methods

Participants

A total of 125 participants (65 males) were enrolled, including 63 young (age: 20–35) and 62 older (age: 55–72) healthy adults. Besides age, the two groups were closely matched. Full demographic information is provided in the results. Exclusion criteria included any medical illness, disorder, and/or trauma that may affect CNS function, current substance use, and the standard exclusion criteria for MEG/MRI (i.e., no ferromagnetic implants). The study conformed to the Declaration of Helsinki standards, with participants providing written informed consent, per IRB guidelines.

High-definition transcranial electrical stimulation

A 4 $\times$ 1 configuration (Soterix Medical, NY) was used to administer HD-tDCS to the left M1 [44]. The central electrode was positioned at C3, aligning with the left M1 encircled by four electrodes of opposite polarity (i.e., C1, C5, FC3, and CP3) [45]. Current flow modeling was conducted to determine the focality and intensity of stimulation [46]. Each participant completed three visits, separated by at least 1 week (M = 10.4 days, SD = 5.2 days; Fig. 1). Stimulation conditions were pseudorandomized to include two active (anodal or cathodal) and one sham session in a double-blind design. During the active visits, participants underwent 20 min of 2.0 mA HD-tDCS, plus a 30-s ramp-up period. The sham visit followed the same protocol but without stimulation beyond the ramp-up period. Across all visits, participants performed a multi-source interference task (MSIT) during the stimulation period to keep them cognitively engaged [35, 40, 47, 48]. In this task, each trial started with a central fixation cross presented for 2000–2400 ms. A row of three equally spaced integers between 0 and 3 then replaced the fixation, and these stimuli were presented for 1500 ms. Two of the number stimuli were always identical (task-irrelevant), and the third was unique to that trial (task-relevant). Participants were given a button pad and instructed that the index, middle, and ring finger locations represented the integers 1, 2, and 3, respectively. Participants were instructed that on each trial they would be presented with a row of three integers, and that the objective was to indicate the “odd-number-out” by pressing the button corresponding to its numerical identity (and not its spatial location). Using these stimuli, four interference conditions were possible: (1) control (no interference), (2) spatial interference (e.g., 0,0,2), (3) identify interference (e.g., 1,2,1), and (4) multi-source interference (e.g., 2,3,2). See Table 1 for behavioral performance on the MSIT task. There were 75 trials of each type, and all trials were pseudorandomized so that no trial type occurred more than twice in a row. There was a delay of about 66 min from the end of the stimulation to the initiation of the MEG experiment, which is well within the 2 + hour time window that has been reported for offline HD-tDCS effects [49].

Fig. 1.

(Top) Current flow model; (bottom) experimental paradigm. Participants received 20 min of anodal/cathodal or sham HD-tDCS over the left motor cortex. Current flow modeling revealed focused field intensity values near the precentral gyrus (M1). Following HD-tDCS, participants completed the flanker task during MEG recording. The total visit time from the beginning of stimulation to the end of the MEG task was approximately 100 min

Table 1.

Demographics and behavioral performance

Demographics	Young (N = 58)	Older (N = 56)
Age in years, mean (SD)	25.01 (3.74)	63.53 (4.71)
Sex (Female %)	58	55
Education in years	16	16
Race %
White	70.69	91.07
Black	6.90	7.14
Asian	20.69	0
More than one race	0	1.79
Others	1.72	0
Ethnicity %
Not Hispanic or Latino	93	100
Flanker Task (during MEG)
Reaction time in ms, mean (SD)	561.16 (85.16)	679.31 (81.46)
Accuracy %, mean	98.31	96.74
MSIT (during HD-tDCS)
Reaction time in ms, mean (SD)
Control	837.39 (123.85)	959.70 (116.75)
Identity interference	945.29 (119.56)	1062.88 (110.15)
Spatial interference	893.75 (127.68)	1040.05 (120.51)
Multisource interference	1029.89 (121.14)	1156.71 (116.99)
Accuracy %, mean
Control	99.09	99.21
Identity interference	98.30	98.70
Spatial interference	98.16	98.07
Multisource interference	95.97	96.31

SD standard deviation

MEG experimental paradigm

Following stimulation, participants completed a modified Eriksen flanker task [50] while seated in a nonmagnetic chair with their head positioned within the MEG sensor array (Fig. 1). A fixation cross was presented for 1500 ms (± 50 ms), followed by the target stimulus for 2500 ms, which consisted of five centrally presented arrows. Participants were instructed to respond based on the direction of the middle arrow (left = right index; right = right middle finger), which could be pointing in the same (congruent) or opposite direction (incongruent) as the surrounding arrows. Two hundred trials (100/condition) were presented in a pseudorandomized order, making the overall MEG recording approximately 14 min.

MEG data acquisition

All recordings were conducted in a two-layer magnetically shielded VACOSHIELD room (Vacuumschmelze, Hanau, Germany) using a MEGIN Triux Neo-MEG system (Helsinki, Finland) with 306 sensors (204 planar gradiometers and 102 magnetometers). The acquisition bandwidth was 0.1–330 Hz and the sampling rate was 1 kHz. During data acquisition, participants were monitored via real-time audio-visual feeds from inside the shielded room. Each MEG dataset was individually corrected for head motion and subjected to noise reduction using the signal space separation method with a temporal extension [51].

MEG pre-processing and sensor-level statistics

Ocular and cardio-artifacts were removed from the data using signal space projection (SSP), and the projection operator was accounted for during source reconstruction [52]. The continuous magnetic time series was divided into 4000 ms epochs, with the baseline being -1800 ms to -1000 ms, and 0 ms marking motor response onset. Epochs containing artifacts were rejected based on a fixed threshold method, supplemented with visual inspection. On average, 89 congruent and 88 incongruent trials per participant were used for analysis. A 2 × 2x3 ANOVA (i.e., two task conditions, age as a between-subject categorical variable, and three within-subject stimulation configurations) was performed to ensure the number of accepted trials didn’t differ by stimulation, task condition, age, or their interactions, all ps > 0.5.

The artifact-free epochs were transformed into the time–frequency domain using complex demodulation [53] and the resulting spectral power estimations per sensor were averaged across trials to generate time–frequency plots of mean spectral density. Using 204 gradiometers, each data point was evaluated using a mass univariate approach based on the GLM. To minimize the risk of false positives, paired-sample t-tests were conducted against baseline, with significant time–frequency bins identified at p < 0.05. To control for multiple comparisons, surviving bins were clustered with temporally and/or spectrally neighboring bins that were also above the threshold and a cluster value was derived by summing the t-values across all significant bins per cluster, which were then evaluated using nonparametric permutation testing (1000 permutations). Clusters exceeding the p < 0.001 threshold in the permuted distribution were considered significant.

MEG beamformer imaging and statistics

Each participant’s MEG data were coregistered with their high-resolution T1-weighted structural brain data. Cortical neural responses were imaged using a beamformer [54], which employs spatial filters in the time–frequency domain to calculate source power for the entire brain volume, with units (i.e., pseudo-t) that reflect noise-normalized power differences per voxel. Normalized source power was computed for the selected time–frequency windows over the entire brain volume per participant at 4.0 × 4.0 × 4.0 mm resolution. The resulting beamformer images were grand-averaged across all participants, task conditions, and HD-tDCS configurations to assess data quality and the brain areas generating the strongest oscillatory responses. Voxel time-series (i.e., “virtual sensors”) were then extracted from each participant’s data using the peak voxel in these grand-averaged beamformer images. To compute the virtual sensors, we applied the sensor weighting matrix derived through the forward computation to the preprocessed signal vector, which yielded a time series for the specific coordinate in source space. Note that virtual sensor extraction was done per participant once the coordinates of interest were known. Once the virtual sensor time series were extracted, we computed the envelope of the spectral power in the frequency bin that was used in the beamforming analysis. From this time series, we computed both absolute and relative (i.e., baseline-corrected) time series for each participant. To examine spontaneous activity, values from the absolute power time series from each peak were averaged across the baseline period (i.e., from − 1800 to -1000 ms) followed by 2 × 3 ANOVAs. To examine task-related responses, values from the relative power time series per peak were averaged across the active windows, followed by 2 × 2x3 ANOVAs.

Note that we used a two-way ANOVA (and not a three-way) on the spontaneous data because the baseline period preceded stimulus onset, and thus, the task condition factor was not relevant. Significant interactions were probed using FDR-corrected t-tests. Finally, exploratory Pearson correlations were conducted on significant stimulation-related findings to further clarify links among variables of interest. Since these analyses were exploratory, we did not correct for multiple comparisons. For further methodological details on our MEG analyses, see [55].

Whole-brain connectivity analysis

To assess the polarity-specific effects of motor HD-tDCS on functional connectivity with M1 in young versus older adults, the peak voxels for each oscillatory response were identified from the grand-averaged maps, and these were used for the calculation of beta and gamma coherence beamformers (with source amplitude regressed out, to avoid biasing connectivity estimates) [54], followed by voxel-wise whole-brain 2 × 2x3 ANOVAs. In both cases, the peak voxels were within the left M1 cortex. A threshold of p < 0.005 was used to define clusters in the coherence maps, and interaction effects were probed using FDR-corrected t-tests. Any value ± 3 SD from the mean was considered an outlier and removed before statistical analyses.

Results

Of the 125 enrollees, nine participants were excluded due to MEG artifacts and four due to poor task performance, including two with accuracy below 3 SD and two with reaction time above 3 SD. Notably, the reaction time outliers were removed from behavioral data but kept for neural data, although including/excluding these participants did not affect the findings. The remaining 114 participants consisted of 58 young and 56 older adults (Table 1).

Behavioral analysis

A 2 × 2 × 3 ANOVA on reaction time showed a main effect of the task condition, F(1,110) = 611.291, p < 0.001, indicating that participants responded significantly slower in incongruent (mean reaction time: 648.16 (102.61) ms) compared to congruent trials (mean reaction time: 592.32 (99.52) ms), and a main effect of age, F(1,110) = 59.108, p < 0.001, showing that the older group was slower than the younger group overall (older: mean: 679.31 (81.46) ms; younger mean: 561.16 (85.16) ms). An ANOVA on accuracy revealed a significant age-by-condition interaction, F(1,110) = 5.584, p = 0.020, such that only younger adults exhibited an interference effect on accuracy, t(56) = 2.923, p_FDR = 0.01, reflecting poorer performance on incongruent trials (congruent: 98.67%; incongruent: 97.95%). There was also a main effect of age such that older adults were less accurate overall, F(1,110) = 8.189, p = 0.005 (older: 96.74%; younger: 98.31%). There were no main effects of stimulation nor interaction effects involving stimulation on reaction time or accuracy. See Table 1 for mean reaction times and accuracy per age group.

Sensor- and source-level motor oscillatory dynamics

Significant oscillatory responses were observed in the beta range (16–26 Hz) from -300 to 0 ms and the gamma band (64–82 Hz) from -50 to 100 ms across all participants (ps < 0.001, corrected), with both responses mapping to the left M1 (Fig. 2). Virtual sensors were then extracted from the peak voxels of each response (i.e., beta and gamma) to examine the temporal evolution of these responses for each stimulation condition. These time series were used to compute the mean absolute power spanning the baseline period, as well as the relative power during stimulus processing. Two 2 × 3 ANOVAs indicated significant main effects of age for spontaneous beta, F(1,79) = 28.427, p < 0.001, and gamma power, F(1,75) = 15.447, p < 0.001, such that both spontaneous beta and gamma were significantly stronger in the older group compared to their younger counterparts. Neither the interaction nor main effect of stimulation was significant for spontaneous activity, ps > 0.5. Next, 2 × 2x3 ANOVAs were conducted on the relative power data and these showed a significant task condition x age x stimulation interaction for gamma responses, F(2,216) = 5.049, p = 0.007 (Fig. 3). Follow-up independent- and paired-samples post-hoc t-tests indicated that the gamma flanker interference effect (incongruent-congruent) was stronger in older compared to younger adults following anodal HD-tDCS, t(108) = -2.725, p_FDR = 0.036, as well as following anodal versus cathodal HD-tDCS within the older group, t(52) = 3.033, p_FDR = 0.036. Significant main effects of stimulation, F(2,216) = 3.950, p = 0.021, and age, F(1,108) = 18.955, p < 0.001 were also found but not examined, as the significant interaction effect superseded them. Finally, we further probed the significant stimulation related effects using exploratory Pearson correlations. These indicated that the gamma flanker interference effect (incongruent-congruent) in the left M1 cortex following anodal stimulation was negatively correlated with the reaction time flanker interference effect (r =—0.338, p = 0.013) and positively correlated with the accuracy flanker interference effect (r = 0.273, p = 0.047) in older participants. In addition to behavior, we also probed whether age was correlated with the strength of stimulation effects in each age group, but neither these nor other relationships were significant. The lack of age-related correlations in each group could reflect restricted range effects.

Fig. 2.

Neural responses during the flanker motor task. (Left) Grand-averaged time–frequency spectrograms of MEG sensors exhibiting one or more significant responses, with gamma activity at the top, beta at the bottom. All signal power data are expressed as percent difference from baseline, with color legends on the right shown for each respective spectrogram. In each spectrogram, dashed lines indicate the time–frequency windows that were subjected to beamforming. Note that we did not examine the beta rebound (increase in power at ~ 800 ms) due to the reaction time differences. (Right) Grand-averaged beamformer images (pseudo-t) across all participants and HD-tDCS montages for each time–frequency component. The color bars on the right show response amplitude in pseudo-t values

Fig. 3.

Flanker effect in the left primary motor cortex (M1). The grand-averaged map of the motor-related gamma response during the − 50 to 100 ms window from 64 to 82 Hz is displayed in the top right corner, with the violin and box plots showing that the interference effect was stronger in the older relative to younger adults following anodal stimulation, and stronger in older adults after anodal compared to cathodal HD-tDCS. Stimulation conditions are denoted on the x-axis, while flanker interference relative power in the gamma range is shown on the y-axis in % change from baseline units. The violin/box plot includes the individual data points, mean, and the first and third quartiles (box). *p_FDR < 0.05

Motor HD-tDCS effects on beta and gamma coherence

Whole-brain, voxel-wise 2 × 2x3 ANOVAs were performed on the coherence maps, with amplitude regressed out. These showed significant age-by-stimulation interactions in interhemispheric M1-IFG beta, F(2,158) = 5.929, p = 0.003 and M1-DLPFC gamma connectivity, F(2,148) = 8.837, p < 0.001 (Fig. 4). Follow-up t-tests showed weaker M1-IFG beta connectivity following anodal HD-tDCS in younger compared to older adults, t(80) = -3.063, p_FDR = 0.027. Additionally, within the younger group, connectivity was weaker following anodal compared to cathodal stimulation, t(43) = -271, p_FDR = 0.04. For gamma, cathodal stimulation led to weaker M1-DLPFC connectivity compared to sham in older adults, t(34) = -3.707, p_FDR = 0.001, and stronger connectivity compared to sham in younger adults, t(44) = 2.995, p_FDR = 0.012. Additionally, M1-DLPFC connectivity was stronger in younger compared to older adults following cathodal HD-tDCS, t(78) = -3.067, p_FDR = 0.012, while the opposite pattern was observed following the sham condition, t(78) = -2.852, p_FDR = 0.014. No other main effects or interactions were significant, all ps > 0.05. Pearson correlations on the significant stimulation-related connectivity effects were not significantly related to reaction time or accuracy effects. However, the strength of interhemispheric M1-IFG beta connectivity was negatively correlated with age in the older group after collapsing across stimulation conditions (r = -0.381, p = 0.019). No other relationships were significant.

Fig. 4.

Age by stimulation interaction in coherence analysis using left primary motor (M1) seed. (Left) Whole-brain 2 × 2 × 3 ANOVAs showed decreases in interhemispheric M1-IFG beta connectivity following anodal relative to cathodal HD-tDCS in younger adults, as well as reduced M1-IFG beta connectivity in younger relative to older adults following anodal stimulation. (Right) The ANOVA on gamma coherence maps showed M1-DLPFC connectivity to be decreased in older adults following cathodal HD-tDCS compared to sham, with the opposite pattern being observed in younger adults. Younger adults also had stronger M1-DLPFC connectivity following cathodal stimulation relative to older adults, while the opposite was true for the sham condition. Stimulation conditions are denoted on the x-axis, while connectivity residuals (source amplitude regressed out) are shown on the y-axes. Each violin/box plot includes the individual data points, mean first and third quartiles (box). IFG, inferior frontal gyrus; DLPFC, dorsolateral prefrontal cortex. **p_FDR < 0.01, *p_FDR < 0.05

Discussion

This study combined offline M1 HD-tDCS with MEG and advanced source imaging to examine spectrally- and polarity-specific effects on movement-related neural oscillations in healthy aging. Behaviorally, all participants were slower to respond during incongruent compared to congruent trials [3], with older adults also being generally slower to respond overall and less accurate. Older adults also showed polarity-specific effects of motor stimulation on gamma flanker interference in left M1. Network-level modulatory effects of HD-tDCS diverged among the young and older groups, with effects differing based on the stimulation polarity and frequency of neural responses. Finally, we found that spontaneous beta and gamma activity was elevated in the M1 cortices of older relative to younger adults, although stimulation did not significantly affect levels of spontaneous cortical activity in either group. We expand on these findings below.

A key finding was the stronger gamma interference effect in left M1 cortex following anodal compared to cathodal stimulation in older adults. This gamma response is known to be critical for movement execution, and such an increase may reflect the greater task condition demands (i.e., incongruent). Notably, such increased gamma power with increased interference was limited to the anodal condition in the older group, which may indicate task condition-dependent anodal-excitation and thus neural gain in older adults. However, this amplification did not translate into significantly improved behavioral performance, which could partially stem from the lack of differences compared to the sham condition in older adults. Such null findings of stimulation on behavioral performance were not surprising, especially given the fact that all our participants were healthy adults and could likely compensate for any disruptions to neural processes. Some previous studies using interference-type tasks have found that tDCS affects behavior [56], whereas several other studies have found no significant effects [57, 58]. Interestingly, all of these studies reported robust neural differences, with the primary methodological differences being the location of the stimulation electrode(s). In the current study, we did not expect differences in behavior, and not having such differences was beneficial in many respects, including simplifying the interpretation of neural differences. In short, when a behavioral difference exists, one can never be certain whether the neural differences reflect true processing differences or merely a secondary effect of worse accuracy. Interestingly, while behavior was not significantly improved in older adults during the anodal condition, the neural gamma interference response was correlated with the behavioral reaction time and accuracy interference effects, indicating that these changes have important implications for behavior, including smaller interference effects on reaction time and accuracy effects similar to the younger group. Beyond the behavioral data, such polarity-specific effects on M1 gamma oscillations support previous reports of differential cortical modulation by anodal and cathodal tDCS with aging [59] and require further investigation.

Aside from modulating M1 gamma oscillations, HD-tDCS also differentially affected M1 connectivity with prefrontal cortices within different spectral ranges among young and older adults. Interhemispheric M1-IFG beta connectivity decreased after anodal compared to cathodal stimulation in younger adults, whereas HD-tDCS had no effect on beta connectivity in older adults which exploratory correlations suggested may be declining with increasing age across all stimulation conditions. In contrast, cathodal stimulation led to an increase in M1-DLPFC gamma connectivity in younger and a decrease in older adults compared to sham. This finding is particularly intriguing, as it suggests that the effects of tDCS on motor control dynamics are spectrally-specific in younger versus older adults. Both of these brain regions (IFG and DLPFC) are widely known for their roles in the top-down regulation of goal-directed actions [60]. Notably, two recent studies of response competition in the motor system have focused on such top-down control in healthy young [35] and older adults [1]. Altogether, our data provide empirical evidence for differential neuromodulatory effects on prefrontal-motor networks with healthy aging. Such changes in the interhemispheric M1-IFG/DLPFC coupling also resonate with the existing idea of tDCS-induced modulation of transcallosal inhibition, theorized by a multitude of stimulation studies, [61] which in turn is thought to rely on GABAergic interneurons [62]. The differential effects observed here could be due to declining GABA levels with age [63], leading to alterations in interhemispheric transmission in older adults following stimulation. Cumulatively, our connectivity findings suggest divergent spectrally- and polarity-specific effects of motor HD-tDCS on M1-prefrontal loops in young versus older adults. tDCS-induced, age-dependent network-level changes have also been investigated in studies targeting non-motor cortical regions. For example, a fMRI study showed that a single session of anodal tDCS applied to the IFG could restore connectivity patterns in older adults so that they closely resemble those of younger participants, while also leading to improvements in cognitive performance [64]. Another common tDCS target in aging studies is the DLPFC, with the goal of improving working memory processing; however, the findings to date remain overwhelmingly heterogeneous, with some studies reporting improvements [65] and others no effect of stimulation [66].

Finally, we also found that spontaneous beta and gamma activity were strongly elevated in older relative to younger adults. Studies of spontaneous beta in the motor cortices have reliably shown that power increases with age in adulthood [41, 42, 67, 68], increases throughout the day in healthy adults and is tightly linked to the amplitude of beta ERD responses during movement [18], and that changes in spontaneous M1 beta with development are tied to pubertal testosterone increases as children enter adolescence [29]. While the literature linking spontaneous gamma levels in the sensorimotor cortices with increasing age is more limited [68, 69], studies have linked elevated gamma in M1 [70–74], early visual cortices [75], and prefrontal cortices [76–79] with cognitive impairment in multiple conditions. Thus, we strongly expected spontaneous beta and somewhat expected spontaneous gamma to be elevated in older relative to younger adults, although we were less confident that such spontaneous activity would be amendable to electrical brain stimulation and ultimately our findings in this regard were null. While several studies have shown that spontaneous theta and alpha are altered by tDCS [56, 80–82], all of these studies focused on occipital cortices and used conventional tDCS, with only one showing that spontaneous gamma in the occipital cortices was also modulated [82]. To our knowledge, no studies to date have shown such effects in the motor cortices or demonstrated effects on spontaneous beta or gamma activity using HD-tDCS, with the exception of one study that showed polarity-specific effects on spontaneous gamma in primary visual regions [83]. Thus, future work is warranted in this area to understand the mechanisms that may underlie such spectral and regional specificity, as inducing changes in spontaneous cortical activity may be critical to achieving robust therapeutic effects [72, 84–86].

Before closing, some limitations of this work should be acknowledged. First, our task involved a very simple movement (i.e., finger tap) and studies involving more complex movements, such as sequencing [87–90] could be more sensitive to tDCS-induced changes in the motor cortices and higher-order areas. Future work should consider implementing such multipart movements and/or using tasks that include more conditions to better isolate the critical components driving interference effects and their sensitivity to tDCS neuromodulation. Second, the precise mechanisms of tDCS are unclear, making our interpretations implicating GABAergic neurotransmitters speculative. This warrants future multimodal studies combining magnetic resonance spectroscopy with tDCS. Third, we did not find significant effects of stimulation on behavior (i.e., reaction time or accuracy), and this should be kept in mind when interpreting our neural differences. Of note, some studies using similar paradigms have found stimulation-induced behavioral differences [56], while others have not ^57,58. Thus, future work is needed to identify the key factors driving these inconsistencies. Finally, we focused on motor responses in this brief report, and future work should consider cognitive control in the context of response competition.

Author contribution

Yasra Arif: conceptualization, data curation, formal analysis, software, investigation, methodology, visualization, validation, writing—original draft, writing—review and editing; Peihan J. Huang: data curation, formal analysis, software, visualization, writing—review and editing; Seth D. Springer: software, writing—review and editing, funding acquisition; Hannah J. Okelberry, Jason A. John, Nathan M. Petro, Grant M. Garrison, Kennedy A. Kress, Kellen M. McDonald, and Giorgia Picci: data curation, investigation, project administration, writing—review and editing; Tony W. Wilson: conceptualization, funding acquisition, project administration, validation, visualization, resources, supervision, writing—review and editing.

Funding

This work was supported by the National Institutes of Health (grant numbers RF1-MH117032, S10-OD028751, R01-MH116782, P20-GM144641(TWW), and F30-AG076259 (SDS)). The funders had no role in study design, data collection, and analysis, decision to publish, or preparation of the manuscript.

Data availability

The data used in this article is publicly available through the COINS framework (https://coins.trendscenter.org/).

Declarations

Conflict of interest

The authors declare no competing interests.