Two types of EEG recordings were taken: (1) resting-state recordings; (2) event-related recordings. For the resting-state recordings, participants were instructed to keep their eyes open and look straight ahead at a fixation cross for 3 min. For the event-related recordings, participants were instructed to perform self-paced isometric flexions of the right index finger at approximately 10 s intervals. An index finger flexion was chosen, as it has been shown to elicit the three event-related changes in beta activity over the sensorimotor cortex (e.g., Rimbert et al., 2018). Movements were self-paced to engage the neural processes involved in generating internally motivated voluntary movements, which slightly differ from externally motivated movements (e.g., reduced motor preparation; Brass and Haggard, 2008; Haggard et al., 2006). The tip of the index finger was placed on an 8 mm 10 N/2.2 lb. SingleTact force sensor (SingleTact, Glasgow, UK) that was permanently secured on a computer mouse. Participants performed a total of 60 index finger flexions, which were split into two blocks of 30 flexions. Before the experimental session began, participants completed one 60 s finger flexion training block with a go signal (“press” presented on a screen every 10 s) and one block that was self-paced, with verbal feedback from the experimenter. For the duration of event-related recordings, participants were instructed to look straight ahead at a fixation cross to avoid random eye movements, and minimize any other movement (e.g., blinking, swallowing, etc.).

As stated above, the event-related recordings required participants to press into a force sensor. This was done so that the onset of movement could be registered as a NetStation (Electrical Geodesics, Inc.) digital input event for data analysis. The force sensor output was digitized at a sampling value of 5,000 Hz (CED Power1401), data were acquired using Signal (version 6.02, Cambridge Electronic Design), and event-triggers were generated in NetStation at the onset of movement, using a custom-written Signal sequencer. The onset of movement was defined as the moment at which the force-trace exceeded a pre-determined threshold (which was determined for each participant during practice trials).

EEG data were pre-processed using the EEGLAB toolbox (Delorme and Makeig, 2004) through the MATLAB environment (MathWorks, R2020b). All EEG data were down-sampled to 500 Hz, bandpass filtered from 0.5 to 95 Hz, and notch filtered at 50 Hz. The data were then epoched: resting-state data were divided into 2,000 ms segments; event-related data were segmented from −2,500 to 4,500 ms relative to stimulus onset. Bad channels and noisy epochs were then visually identified and manually removed, and all removed channels were interpolated. The data were then re-referenced to the average, using the fullRankAveRef EEGLAB plugin. Next, independent component analysis (ICA) was performed, using the Infomax algorithm. Following ICA, components containing artifacts clearly distinguished from brain-driven EEG signals (e.g., ocular, vascular, and myogenic artifacts) were visually identified and subtracted from the data. During EEG pre-processing, two resting-state data sets and three event-related data sets (across 4 participants) were identified as having a large number of artifact contaminated epochs. These data sets contained <11 useable trials, below the recommended minimum of 20 trials (Cohen, 2014). Participants with an insufficient number of trials were excluded from further analysis, resulting in sample sizes of N = 21 for the resting-state analyses and N = 20 for the event-related analyses.

Using custom MATLAB scripts, power values were computed from the pre-processed resting-state data and the pre-processed event-related data. Power values were calculated for the left M1 ROI (tACS target) and the right M1 ROI. The right M1 ROI was analyzed to explore whether tACS effects extended to this functionally connected region. While electrodes C3 and C4 in international 10–20 EEG system correspond to the approximate position of the left M1 and right M1, respectively (Jurcak et al., 2007), we analyzed the activity of electrode clusters—a cluster of seven electrodes centered at C3 (C3 cluster: C3, FC3, C1, FC5, CP1, C5, CP3), and a cluster of seven electrodes centered at C4 (C4 cluster: CP2, CP4, C6, C2, FC6, FC4).

In this study, the data were analyzed with generalized linear mixed models (GLMMs) instead of ANOVAs, due to mixed models accounting for inter-individual variability, as inter-individual response variability is a known issue in NIBS (for a review, see Guerra et al., 2020). Statistical analyses and visualization of the results were performed via customized scripts in MATLAB, and the software package R for Statistical Computing version 2023.09.0 + 463 (R Core Team, 2023), using packages “tidyverse” (Wickham, 2023), “DescTools” (Signorell, 2023), “janitor” (Firke, 2023), “car” (Fox et al., 2023), and ggplot2 (Wickham et al., 2023). The “lme4” package (Bates et al., 2023) was used to construct GLMMs, fitted by means of the glmer() function. For post-hoc analyses, we used the “emmeans” package (Lenth, 2023) to perform pairwise comparisons between estimated marginal means derived from our GLMM models. These comparisons were executed using the contrast() function with Bonferroni correction to account for multiple comparisons. To quantify the magnitude of differences, effect sizes were calculated using the eff_size() function, providing standardized measures of the differences between means.

Overall, the theta, alpha, and beta frequency bands showed a similar post-tACS pattern of change across stimulation intensities, while responses in gamma power differed. Here, we focus on describing the post-hoc analyses of resting-state power in the beta band, as this was the target frequency. Detailed results for each of the other frequency bands can be found in the Supplementary Results. As shown in the left panel of Figure 3, all four stimulation intensities showed increased beta power at the C3 cluster from pre to post₁ (|zs| ≥ 3.578, ps ≤ 0.001, |ds| ≥ 0.008) and from pre to post₂ (|zs| ≥ 4.267, ps < 0.001, |ds| ≥ 0.010). Most stimulation intensities did not show a significant change from post₁ to post₂ (|zs| ≤ 0.951, ps ≥ 0.608, |ds| ≤ 0.002), with only 0.5 mA stimulation showing a significant change: a further increase in beta power (z = −2.689, p = 0.022, d = −0.006). These increases in power at C3 were unlikely due to tACS, as the extent of changes following real stimulation was not greater than the changes following sham stimulation (|zs| ≤ 1.493, ps = 1.000, |ds| ≤ 0.029).

Region-specific changes in beta power were observed following 1.0 mA and 1.5 mA stimulation (|zs| ≥ 2.652, ps ≤ 0.024, |ds| ≥ 0.050), but not 0.5 mA or sham stimulation (|zs| ≤ 2.064, ps ≥ 0.117, |ds| ≤ 0.041). Following 1.0 mA stimulation, from pre to post₁, beta power increased at C3 (z = −4.713, p < 0.001, d = −0.011) and decreased at C4 (z = 3.014, p = 0.008, d = 0.007). At post₂, C3 beta power remained elevated relative to baseline (z = −4.267, p < 0.001, d = −0.010), while C4 beta power shifted back toward baseline levels (z = −3.579, p = 0.001, d = −0.008). In contrast, 1.5 mA stimulation increased C3 beta power from pre to post₁ (z = −6.384, p < 0.001, d = −0.015) without affecting C4 (z = −1.670, p = 0.285, d = −0.004). By post₂, 1.5 mA stimulation showed delayed increases in beta power at both C3 (z = −5.411, p < 0.001, d = −0.013) and C4 (z = −2.783, p < 0.016, d = −0.007).

The GLMM analysis of pre-movement power found significant main effects for INTENSITY and FREQUENCY (χ²s ≥ 13.626, ps ≤ 0.004), but not TIME or REGION (χ²s ≤ 1.142, ps ≥ 0.565). The analysis also revealed a significant two-way interaction between INTENSITY and TIME (χ² = 19.434, p = 0.004). All other interactions were non-significant (χ²s ≤ 17.640, ps ≥ 0.479). These results indicate that tACS induced stimulation intensity-specific changes in pre-movement power. However, these changes were not frequency-specific, indicating that beta tACS did not selectively modulate the pre-movement beta ERD. Additionally, these changes were not region-specific, indicating that tACS exerted similar effects bilaterally, across the stimulated C3 cluster (left M1 ROI) and the unstimulated C4 cluster (right M1 ROI). The highest level of interaction—the two-way INTENSITY × TIME interaction—was further investigated with post-hoc comparisons.

Figure 4 shows the pre-movement power values for each time-point and stimulation intensity. It is important to note that there were significant differences in power between some of the stimulation intensities at baseline. Tables of these results can be found in the Supplementary Results. For this reason, comparisons between stimulation intensities were only performed on the relative change between two time points, though these comparisons should be interpreted with caution. From pre to post₁, none of the stimulation intensities showed a significant change in pre-movement power (|zs| ≤ 1.360, ps ≥ 0.362, |ds| ≤ 0.001). From pre to post₂, there was a significant increase in pre-movement power only following 1.0 mA stimulation (|z| = −3.045, p = 0.007, |d| = −0.003). This increase was likely due to tACS, as there were no changes following sham stimulation from pre to post₂ (z = 1.917, p = 0.134, d = 0.002). From post₁ to post₂, the sham condition showed a significant decrease in power (z = 2.961, p = 0.009, d = 0.003), while the real stimulation intensities showed no significant changes in power (|zs| ≤ 1.697, ps ≥ 0.206, |ds| ≤ 0.002). It is possible that the real stimulation intensities promoted stability in pre-movement power, or that the decrease following sham stimulation was a chance occurrence.

The GLMM analysis of movement power found significant main effects for all four factors: INTENSITY, TIME, FREQUENCY, and REGION (all χ²s ≥ 10.46, all ps ≤ 0.005). The analysis also revealed a significant three-way interaction of INTENSITY, TIME, and FREQUENCY (χ² (18, N = 20) = 30.83, p = 0.030). All other interaction effects were non-significant (χ²s ≤ 16.34, ps ≥ 0.060). These results indicate that tACS induced frequency-specific and intensity-specific, but not region-specific, changes in movement power. The highest level of interaction—the three-way INTENSITY × TIME × FREQUENCY interaction—was further investigated with post-hoc comparisons.

Overall, none of the real stimulation intensities induced a change in beta or gamma power, relative to sham stimulation (|zs| ≤ 2.668, ps ≥ 0.138, |ds| ≤ 0.238). The lack of change in beta power indicates that the movement beta ERD was not modulated by 20 Hz tACS (see Figure 5). In contrast, intensity-dependent changes in theta and alpha power were observed, indicating that cross-frequency neuromodulation occurred. Detailed results for the theta, alpha, and gamma frequencies can be found in the Supplementary Results.

The GLMM analysis of post-movement power revealed significant main effects for all four factors: INTENSITY, TIME, FREQUENCY, and REGION (all χ²s ≥ 11.43, all ps ≤ 0.010) and a significant two-way INTENSITY × TIME interaction (χ² (6, N = 20) = 34.51, p < 0.001), with all higher-order interactions non-significant (χ²s ≤ 20.05, ps ≥ 0.330). Similar to the pre-movement period, tACS induced stimulation intensity-specific changes in post-movement power, however, these changes were not frequency-specific or region-specific. As these changes were not frequency-specific, these results indicate that beta tACS did not selectively modulate the post-movement beta ERS. The highest level of interaction—the two-way INTENSITY × TIME interaction—was further investigated with post-hoc comparisons.

Figure 6 shows the post-movement power values for each time-point, and stimulation intensity. Unlike the other two movement periods, there were no significant differences between the stimulation intensities at post-movement baseline. Tables of these results can be found in the Supplementary Results. From pre to post₁, none of the four stimulation intensities showed a significant change in post-movement power (|zs| ≤ 1.296, ps ≥ 0.397, |ds| ≤ 0.001). From pre to post₂ and from post₁ to post₂, there were significant increases in post-movement power following 1.0 mA and 1.5 mA stimulation (|zs| ≥ 3.803, ps < 0.001, |ds| ≥ 0.004), but there was no difference between these intensities in the magnitude of power increase (|zs| ≤ 0.468, ps = 1.000, |ds| ≤ 0.068). These increases were likely induced by tACS, as changes were not observed following sham stimulation (z = −0.591, p = 0.825, d = −0.001). In contrast, 0.5 mA stimulation showed a significant decrease in post-movement power from pre to post₂ (z = 2.460, p = 0.037, d = 0.002), but no change from post₁ to post₂ was observed (z = 2.034, p = 0.104, d = 0.002).

All stimulation intensities, including sham, showed increases across all frequency bands, except gamma, from pre to post₁ (~5 min post-tACS) and from pre to post₂ (~25 min post-tACS). These power increases were unlikely due to tACS, as the extent of changes following real stimulation was not greater than the changes following sham stimulation. There are four possible explanations for this broad (4–30 Hz) power increase. First, the observed changes in power may simply reflect spontaneous fluctuations in EEG activity. Second, these changes in power may be attributed to improvements in the signal-to-noise ratio from the pre- to post-tACS measurements. This phenomenon is frequently encountered in EEG research (Kappenman and Luck, 2010). As the signal-to-noise ratio increases, the true neural signal becomes more prominent relative to background noise, potentially leading to apparent increases in power across frequency bands. Third, the increase in resting-state power might have been due to repeated activation of the sensorimotor cortex during the movement recordings, which involved ~60 self-paced index finger flexions over 10 min. However, the absence of change in resting-state power in sham from post₁ to post₂ indicates that this was unlikely. The post₁ resting-state measurement followed a long (>20 min) period of inactivity, while the post₂ resting-state measurement immediately followed a movement block. If the movement caused the broad increase in resting-state power, it is likely that we should have seen an increase between post₁ and post₂ within the sham condition. Indeed, though changes in resting-state power have been observed following substantial motor training (e.g., Moisello et al., 2015), studies that have implemented simple movement protocols, similar to the current study (without tACS), have not observed changes in resting-state power (e.g., Dornowski et al., 2018; Sasaki et al., 2018). Fourth, the increase in resting-state power might have been due to mental fatigue. Each experimental session lasted 2–2.5 h, and some participants anecdotally reported feeling fatigued during both post-tACS EEG recordings. A recent meta-analysis showed that mental fatigue was positively associated with a broad (1–30 Hz) increase in EEG power, similar to what was observed in the current study, which reflect homeostatic processes (Tran et al., 2020). We considered additional analyses to further investigate whether these broad power increases might be related to fatigue accumulation across sessions (i.e., if beta power increases were greater in later sessions), but did not pursue this due to statistical power limitations. Our counterbalanced design resulted in only 5–6 participants receiving sham in any given sequence position, providing insufficient statistical power for reliable analysis of order-specific effects. We also lacked standardized pre/post fatigue measurements that would allow us to quantify fatigue development throughout each session or correlate it with the observed power increases. Future studies comprising larger sample sizes should consider taking measures of fatigue (e.g., heart rate variability or self-report questionnaires) to account for this variable.

When applying tACS at 1.0 mA, numerous studies have observed significant post-stimulation increases in resting-state M/EEG power of the stimulated frequency band (e.g., Battaglini et al., 2020; D’Atri et al., 2019; Zaehle et al., 2010). Accordingly, we expected that 1.0 mA tACS (but not sham) would increase resting-state beta power. However, we found that the increase in beta power following 1.0 mA stimulation was comparable to the increase in power following sham stimulation. There are a few possible explanations for the non-significant difference between 1.0 mA and sham stimulation.

Although we did not record EEG during tACS, it is possible that the combined frequency and current intensity of tACS used in this study were not adequate to effectively modulate beta oscillations during the stimulation period. Results from rodent studies suggest that stimulation at frequencies >10 Hz might require higher current intensities to induce a response (e.g., Anastassiou et al., 2011; Deans et al., 2007). This might explain the discrepancy between the null finding in the current study and significant findings in previous human M/EEG studies, where 1.0 mA tACS was applied at ≤10 Hz. However, this notion is contradicted by evidence of 1.0 mA beta tACS inducing significant changes (e.g., greater motor learning: Pollok et al., 2015). With conflicting findings, it remains unclear as to whether the tACS protocol used in the current study was able to effectively modulate the neural oscillations of the left M1 ROI during the stimulation period. If real tACS modulated neural activity as it was delivered, the effect might not have outlasted the stimulation period. This study was limited to recording EEG before and after stimulation because, during stimulation, the tACS electrodes obstruct the EEG electrodes that record activity from the target site and the tACS stimulation artifact contaminates activity recorded from nearby electrodes (Asamoah et al., 2019; Kasten and Herrmann, 2019; Neuling et al., 2017; Noury et al., 2016; Noury and Siegel, 2018). Future studies might overcome this limitation with devices that support concurrent stimulation and recording, as well as developments in artifact removal techniques (Fehér and Morishima, 2016).

Another possible explanation is that real tACS induced a lasting effect on left M1 neural activity, but the measures used in the current study were unable to capture the effect. A recent in vivo study of non-human primates demonstrated that tACS could entrain the firing of individual neurons (Johnson et al., 2020). However, across all stimulation intensities (0.5 mA, 1.0 mA, and 1.5 mA), a maximum of 9 neurons out of 34 were simultaneously entrained, demonstrating that only a relatively small proportion of neurons responded to tACS. Measuring changes in left M1 neural activity non-invasively with EEG power requires a large population of neurons to be modulated in order to be recognized. Therefore, in the current study, it is possible that real tACS modulated neural activity beyond the duration of stimulation, but the extent of neuromodulation was not in a sufficiently large population of neurons to be captured by EEG power spectral analysis. Our results did, however, suggest that 1.0 mA beta tACS might have influenced areas that include the M1-M1 network. While sham stimulation resulted in similar increases in beta power at both the left and right M1 ROIs, 1.0 mA stimulation produced distinct regional differences. Specifically, beta power increased in the left M1 ROI but decreased in the right M1 ROI, suggesting stimulation-dependent changes in the M1-M1 network. This interpretation was supported by follow-up EEG connectivity analyses, reported in Wansbrough et al. (2024a), which demonstrated that 1.0 mA stimulation significantly reduced M1-M1 connectivity compared to sham. These findings suggest that 1.0 mA beta tACS might reduce coupling between the stimulated left M1 ROI and the unstimulated right M1 ROI. Given the potential functional implications of altered interhemispheric connectivity, future research should investigate the functional consequences of unifocal tACS on both unilateral and bilateral movement.

Finally, the null-finding here might be explained by both intra- and inter-individual variability. Between-session intra-individual variability contributed to baseline differences in resting-state and event-related power. The GLMMs accounted for this variability by incorporating participant-specific random intercepts, allowing for appropriate modeling of individual differences. Given the centered random intercepts, intra-individual variability is not believed to have significantly biased the interpretation of tACS-induced changes. Additionally, the results revealed substantial inter-individual differences in tACS response. For example, at post₁, ~50% of participants showed a decrease in resting-state beta power at left M1 ROI following 1.5 mA stimulation, while ~25% showed an increase and ~25% showed no change (defined as <10% change). This inter-individual response variability is a common issue in NIBS literature (for a review, see Guerra et al., 2020). Various factors may contribute to inter-individual response variability, including differences in anatomical, neurochemical, and demographic characteristics (for a review of these factors, see Vergallito et al., 2022). Consequently, despite applying identical intensities across participants, variations in skull thickness, cortical folding patterns, and tissue conductivity properties likely resulted in inter-individual differences in electric field distributions reaching neural targets (Hunold et al., 2023). Another potential source of inter-individual variability relates to dynamic systems theory (Pikovsky et al., 2001), which suggests that the alignment between endogenous oscillations and the exogenous stimulation frequency influences entrainment outcomes. We made an effort to reduce the impact of inter-individual variability by implementing a within-subjects design, recruiting a sample of healthy right-handed young adults, and by taking individual differences into consideration in the statistical models as a random intercept. However, sources of variability would have still impacted the individual response to tACS, which might have contributed to the null finding. Future studies may overcome these issues through individualized stimulation approaches that account for both anatomical differences (through electric field modeling) and oscillatory characteristics (through frequency-calibrated stimulation) (Wansbrough et al., 2024b).

Importantly, our results align with a growing body of literature demonstrating inconsistent effects of beta tACS on sensorimotor oscillations. While several studies have reported significant increases in beta power following 20 Hz stimulation (Battaglini et al., 2020; Lyzhko et al., 2023; Nakazono et al., 2020; Suzuki et al., 2022), others have failed to observe significant modulation (Akaiwa et al., 2024; Harada et al., 2020; Lafleur et al., 2021; Rumpf et al., 2019; Sugata et al., 2018). These discrepancies may be related to methodological variations including electrode montages, stimulation durations, and measurement approaches. These reproducibility challenges underscore the need for more systematic parameter investigations and standardized methodological approaches in beta tACS research. Multi-site studies with standardized protocols could provide important insights into the reproducibility of beta tACS effects, both physiologically (on neural oscillations) and functionally (on motor performance). Additionally, reporting individual-level data alongside group-level analyses may help identify factors contributing to response variability and inform more reliable stimulation protocols. Limited evidence of intensity-dependent changes in resting-state power of the target frequency (beta) band.

Previous in vivo research of both rodents and non-human primates has shown that the entrainment of neural firing depends on the stimulation intensity. When applying tACS at a set frequency, higher stimulation intensities entrained more neurons to fire at the tACS frequency compared to lower intensities (Asan et al., 2020; Huang et al., 2021; Johnson et al., 2020; Krause et al., 2022). Several human studies have also demonstrated intensity-dependent changes in EEG power of the stimulated frequency band (De Koninck et al., 2021; Wang Y. et al., 2022), as well as changes in corticospinal excitability (Moliadze et al., 2012; Shorafa et al., 2021). The hypothesis of linear intensity-dependent changes in resting-state beta power at the left M1 ROI was not supported by the changes in beta power, which did not significantly differ between intensities. This non-significant finding might be explained by the reasons discussed in the previous section. Interestingly, the previously noted region-specificity of the tACS response appeared to be intensity-specific: increases in beta power were significantly different between the stimulation site (left M1 ROI) and the contralateral site (right M1 ROI), following only 1.0 mA and 1.5 mA stimulation.

Only three studies have previously examined the effect of beta tACS (when applied at rest over the left M1) on event-related M/EEG power (Akaiwa et al., 2024; Harada et al., 2020; Sugata et al., 2018). Following 1.0 mA beta tACS, no changes event-related beta power were observed during movement (Harada et al., 2020; Sugata et al., 2018), or post-movement (Akaiwa et al., 2024). In the current study, 1.0 mA stimulation did not selectively modulate the beta ERD/ERS of any movement period. Instead, there was a broadband increase (4–90 Hz) in pre-movement and post-movement power, but not movement power. The pre- and post-movement increases in power were likely due to tACS, as sham stimulation did not show the same effect. It is unclear why responses within each movement period differed. One possible explanation might be that the brain state underlying each movement period affected the response to tACS. Indeed, these movement periods are underpinned by slightly different neural processes (e.g., differences in excitation/inhibition; Kilavik et al., 2013), and research suggests that differences in brain states can affect the tACS response (e.g., Shorafa et al., 2021; Wang Y. et al., 2022). Interestingly, the increases in pre- and post-movement power were delayed, being only observed at post2 (~25 min post-tACS). The delayed responses may reflect late plasticity-like mechanisms, which have been observed in corticospinal excitability following 20 and 250 Hz tACS (Moliadze et al., 2010). Further, current results did not show a significant effect of region, indicating that 1.0 mA tACS might have had a similar effect on the stimulated left M1 ROI and the non-stimulated right M1 ROI. The pre-movement ERD and post-movement ERS are observed bilaterally (Kilavik et al., 2013), thus, the bilateral response observed following tACS might reflect a functional spread of neuromodulation. Further research is needed to clarify the exact mechanisms underpinning the observed effect of tACS on movement-related neural activity, as well as the functional effects of these changes. Beta tACS did not modulate event-related peak beta frequencies.

While we did not see any tACS-induced changes specific to the beta band, we observed intensity-dependent changes in broadband (4–90 Hz) power of the pre- and post-movement periods, but not the movement period. In the pre-movement period, 1.0 mA stimulation induced a delayed increase in power (from pre to post₂), while sham, 0.5 mA and 1.5 mA stimulation showed no change. For post-movement power, we observed significant increases in power following both 1.0 mA and 1.5 mA stimulation (from pre to post₂ and from post₁ to post₂), though the extent of changes did not significantly differ between 1.0 mA and 1.5 mA stimulation, indicating that there might have been a plateau. Together, these results indicate that tACS induced a non-linear dose–response effect on pre- and post-movement power. The non-linear response is somewhat in line with the findings of De Koninck et al. (2021), who found that alpha tACS at 1.0 mA induced a greater increase in alpha power compared to stimulation at 4.0–6.0 mA. The authors suggested that the non-linear intensity-dose response might be explained by compensatory or homeostatic mechanisms – a finding that is also observed in transcranial direct current stimulation (Batsikadze et al., 2013). It is unclear why we observed broadband effects rather than beta-specific ones; this may be related to cross-frequency coupling mechanisms, which we discuss in detail in the Supplementary Material along with the additional results within other frequency bands.

Interestingly, although there was no significant change in post-movement power following sham stimulation, there was a significant decrease in power following 0.5 mA stimulation (from pre to post₂). This finding might have been a result of spontaneous fluctuations or might suggest that 0.5 mA beta tACS induced a suppression of power. Suppressed EEG power following 0.5 mA tACS aligns with the results from a recent study in non-human primates. Krause et al. (2022) found that applying tACS at a lower intensity decreased entrainment of neural firing. The authors reasoned that the weak current was unable to override the rhythm of neural firing, resulting in a competition for control between the natural endogenous rhythm and the exogenous current. Similar mechanisms might explain the suppression observed in the current study, though it is not clear why suppression occurred across a broad range of frequencies. Indeed, Krause et al. (2022) only observed effects within a narrow window around the stimulation frequencies. Comparisons between these two studies should be taken with caution as there are significant methodological differences between them. For example, Krause et al. (2022) analyzed local field potentials recorded during tACS. Without more comprehensive and extensive analyses, the exact mechanisms underpinning the suppression of post-movement in this study remain uncertain.

We also observed instances where power values decreased following sham stimulation, but not real stimulation. For example, in the pre-movement period, there was a decrease in power following sham stimulation from post₁ to post₂, but no changes following 0.5 mA, 1.0 mA, or 1.5 mA stimulation. A speculative interpretation of this might be that the real stimulation promoted stability of sensorimotor oscillations, though the mechanisms underlying this are unknown. Overall, while tACS did not appear to selectively modulate pre-movement, movement, and post-movement beta oscillations, the results suggest that tACS can modulate broad sensorimotor activity, and that the stimulation intensity is an important consideration for the modulation of this activity. While the current study did not directly assess behavioral outcomes, the observed intensity-dependent changes in broadband power may have functional significance. The increases in pre-movement power following 1.0 mA stimulation suggest that this protocol could potentially influence motor readiness and response selection (Kilavik et al., 2013). Similarly, the bidirectional modulation of post-movement power (increases after 1.0 mA and 1.5 mA; decreases after 0.5 mA) may affect motor inhibition processes, which are critical for movement termination (Heinrichs-Graham and Wilson, 2016). Notably, the fact that different movement phases showed distinct responses to stimulation suggests that adaptive stimulation protocols, which take specific movement phases into account, might be necessary to achieve targeted functional effects. Future studies incorporating behavioral measures are needed to directly test these hypotheses and determine the optimal stimulation parameters for specific functional outcomes. Additionally, future research should test a broader range of intensities, particularly above 1.5 mA, to provide a more comprehensive understanding of the dose–response relationship. Several studies have already explored higher tACS intensities: De Koninck et al. (2021) found low intensity (1.0 mA) stimulation induced greater and longer-lasting increases in alpha power than higher intensities (4.0–6.0 mA); Wang Y. et al. (2022) demonstrated that 2.0 mA alpha tACS increased resting-state alpha power more effectively than 1.0 mA, but only when participants had their eyes open; and clinical work by Wang H. et al. (2022) successfully applied 15 mA currents in depression treatment. Similarly, investigating different stimulation durations (e.g., >20 min) could help determine whether longer stimulation periods produce stronger or more persistent effects. Indeed, extended protocols have shown promising outcomes, such as Del Felice et al.’s (2019) 30 min tACS sessions for Parkinson’s disease, which modulated sensorimotor oscillations and improved motor performance. However, there remains a need for systematic investigation directly comparing different stimulation durations. Such extended parameter exploration would be especially valuable given the non-linear intensity-dependent effects observed in our study and others, and would contribute to developing more effective tACS protocols for both research and potential clinical applications.