Modulation of interhemispheric inhibition and motor learning by different intensities of bilateral transcranial direct current stimulation in healthy adults

Abstract

Interhemispheric inhibition (IHI) plays a crucial role in performing unilateral hand movements. Transcranial direct current stimulation (tDCS) has been shown to modulate IHI and enhance motor learning, but the most effective stimulation intensity remains uncertain. This study aimed to investigate how different intensities (1.0 mA, 1.5 mA, 2.0 mA, and sham) of bilateral tDCS affect IHI and motor learning in healthy adults. Twenty right-handed participants completed four randomized sessions, each involving bilateral tDCS application while performing a serial reaction time task. Resting-state electroencephalogram (EEG) was recorded before and after stimulation to assess cortical activity. The brain symmetry index (BSI) was used as an indicator of IHI. The result demonstrated that 2.0 mA produced the most prominent changes in BSI, particularly in frontal and central brain regions. Correspondingly, reaction times improved during and after stimulation at 2.0 mA. Furthermore, the degree of BSI modulation was associated with improvements in reaction time. These findings indicate that 2.0 mA may be a choice for enhancing both IHI and motor learning in healthy individuals. This study also highlights the potential of using BSI as a biomarker to monitor the effects of brain stimulation and guide the selection of intensity in both research and clinical applications.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-025-33298-8.

Introduction

Motor learning refers to practice-related changes in motor performance resulting from the repetition of voluntary motor tasks¹. It involves the acquisition, and retention of motor skill through practice that are supported by neuroplastic changes. Several studies have employed unilateral hand movements to assess motor learning², and it has been suggested that unilateral hand movements are associated with neural modulation in the cortical motor networks of both hemispheres³. Functional magnetic resonance imaging has shown that the primary motor cortex (M1) plays a significant role in motor learning and performance⁴. At rest, the M1 exhibits interhemispheric inhibition (IHI) mediated by transcallosal pathway, where each hemisphere exerts inhibitory influence on the other, maintaining a balance between two hemispheres^5–7. However, during unilateral hand movement, this inhibition temporarily shifts; reduced inhibition from the non-dominant hemisphere results in an increase in the excitability of the dominant hemisphere^8,9. This modulation of IHI is thought to influence the performance of unilateral hand movements¹⁰. Additionally, IHI modulation has been shown to benefit individuals with neurological conditions, such as stroke, where imbalances in IHI negatively affect motor recovery¹¹.

Transcranial direct current stimulation (tDCS) is a non-invasive brain stimulation technique that delivers a weak direct current through the scalp via two electrodes. Within a limited dose, tDCS modulates cortical excitability in a polarity-specific effects: anodal stimulation increases cortical excitability, while cathodal stimulation decreases it^12,13. tDCS is commonly applied using two difference montages: unilateral tDCS, in which one active electrode is applied over the target hemisphere, and bilateral tDCS, where both electrodes are simultaneously applied over each cerebral hemisphere. Bilateral tDCS is often used to modulate IHI between the two hemispheres¹⁴. A meta-analysis reported that the population type was a key factor influencing tDCS effect: bilateral tDCS was more effective for enhancing motor learning and performance in healthy individuals, whereas unilateral tDCS showed greater effects in stroke population¹⁵. For unilateral tDCS, some studies have reported that its effects were intensity-dependent, with higher intensity inducing greater cortical modulations and motor improvement^12,16,17, whereas others have suggested that higher intensity was not lead to better outcomes^18–20. For bilateral tDCS, intensities between 1.0 and 2.0 mA has been used in previous studies to enhance motor learning in healthy individuals²¹, however there remains limited understanding of the dose response relationship between current intensity and both cortical and motor outcomes. The present study addressed this gap by comparing different bilateral tDCS intensities and examining their effects on both cortical activities, assessed by electroencephalography (EEG)-derived brain symmetry index (BSI) and motor learning measured by the serial reaction time task (SRTT). BSI reflects IHI by quantifying asymmetry in EEG power between the two hemispheres²². It can detect global IHI across different brain regions. Unlike paired-pulse transcranial magnetic stimulation (TMS), which primarily measures IHI between specific area (i.e., the M1)²³. Previous study has found that BSI is more asymmetrical during unilateral hand movement and more symmetrical at rest²⁴. The SRTT is a widely used method for assessing motor learning of unilateral hand movement by tracking response times²⁵. These two levels investigation provides novel insights into how stimulation intensity modulates IHI and motor learning efficiency. These findings would have significant implications for future studies and the clinical application of tDCS in conditions involving IHI imbalances.

In the present study, we investigated the effects of different intensities of bilateral tDCS (1.0 mA, 1.5 mA, 2.0 mA, and sham stimulation) in healthy individuals. We hypothesized that bilateral tDCS at an intensity of 2.0 mA would induce a greater change in BSI from the non-dominant to the dominant hemisphere and reduce reaction time during unilateral hand movements compared to other intensities in healthy individuals.

Methods

Trial design

This randomized, double-blind, sham-controlled with cross-over study involved four sessions per participant, receiving 1.0, 1.5, 2.0 mA, and sham stimulation. A minimum 2-week washout period was implemented to prevent carry-over effect, based on previous finding that the aftereffect of a single 20-minute tDCS session combined with motor training resolved within 11 days post-intervention²⁶. Stimulation order was randomized using a computer-generated random sequence, which was securely stored by an independent investigator to ensure researcher independence and prevent reproducibility bias. A partial counterbalancing by simply randomize the order of conditions (sham, 1.0, 1.5, 2.0 mA) for each participant was used. Figure 1 illustrates the study flow diagram.

Fig. 1.

The study flow diagram.

Participants

Potential participants were recruited through advertisement boards at the Faculty of Physical Therapy, Mahidol University. Twenty healthy, right-handed participants, as determined by the Edinburgh handedness inventory²⁷, (9 males; aged 22.60 ± 2.54 years) were recruited. All participants received every stimulation condition. Therefore, participant characteristics (e.g., sex, age) were fully balanced across all stimulation conditions by study design. Exclusion criteria included metal implants, scalp wound, neurological conditions, body mass index (BMI) > 30 kg/m², and recent use of hormones or drugs affecting sleep or movement. All participants provided informed consent. The study was approved by the Mahidol University Central Institutional Review Board (COA No. MUCIRB 2023/051.2803) and registered on ClinicalTrials.gov (25/06/2023, NCT05929014).

Interventions

Transcranial direct current stimulation (tDCS)

Participants sat comfortably during stimulation delivered via tDCS (Ybrain, MINDD STIM, Republic of Korea) using 5 × 5 cm (25 cm²) saline-soaked sponge electrodes. The anodal electrode was placed centered over C3 (left M1) and the cathodal electrode centered over C4 (right M1) according to the international 10–20 EEG system. Electrodes were oriented with their edges parallel to the anterior-posterior axis. Electrode position and current-flow modeling was shown in the Fig. 2. Active stimulation (1.0, 1.5, or 2.0 mA) was applied for 20 min with 30-second ramp-up/down. Stimulation intensities of 1.0, 1.5, and 2.0 mA were selected based on previous studies demonstrating modulation of motor cortex excitability²⁸ and confirming that these intensities are safe and tolerable in healthy participants²⁹. In the sham condition, only the ramp periods were used, as Nitsche and Paulus in 2000¹² indicated that at least 3 min of tDCS is necessary to induce after-effects. During both active and sham stimulations, participants heard a beep sound throughout all sessions. Impedance was auto monitored throughout stimulation period, and the device automatically alarmed and transiently stopped the stimulation if impedance exceeded the safety threshold of 13 (for a reference see Ko et al., 2022³⁰. In such case, possible sources of high impedance such as skin impedance, electrode attachment, and soaking condition of the sponges were checked before restarting the stimulation. After the stimulation, participants reported sensations and guessed the intensity. Both participants and assessors were blinded, an external investigator prescribed the device.

Fig. 2.

Current flow model simulated using ROAST simulation of bilateral tDCS at 2 mA. Slide positions are taken at MNI coordinates (x = − 37.4, y = − 19.1, z =52.4 mm), which functionally localized the hand motor area³¹.

Serial reaction time task (SRTT)

The SRTT, developed using E-Prime software version 3 (Psychology Software Tools, Inc., USA) based on Shilo and colleagues in 2021²⁵, included pseudorandom and sequence blocks. In the sequence blocks, a 12-digit sequence (121423413243) repeated 10 times (120 trials/block). Pseudorandom blocks avoiding sequences of four consecutive numbers (i.e., 1234 and 4321) or repeating patterns (i.e., 1212). Four boxes horizontally displayed on a monitor, and participants responded to asterisk positions using a four-button keypad operated with their dominant hand, with each finger assigned to a specific number: index finger (1), middle finger (2), ring finger (3), and little finger (4) (Fig. 3). A baseline block of 60 trials was completed before stimulation. During tDCS, six blocks were administered (block 1 and 6: pseudorandom; blocks 2–5: sequence), followed by two post-stimulation sequence blocks (blocks 7–8). The SRTT comprised eight experiment blocks (blocks 1–8), with a 30-second break between each block to avoid fatigue. SRTT began 2 min after tDCS onset, allowing for initial stabilization of cortical modulation, as at least 3 min of stimulation are required to induce measurable after-effects¹², and continued for 30 min. The total duration of each session, including stimulation, SRTT, and rest periods, was approximately 35 min.

Fig. 3.

Procedure of the serial reaction time task (left), along with a sample sequence of sequence block (right).

Outcome measurements

Cortical activity

EEG recording

EEG data were recorded before tDCS and after the completion of the SRTT (offline recording). Participants sat comfortably during EEG recording using a 64-channel monopolar amplifier (ANT Neuro, Enschede, Netherlands). The scalp was cleaned, and conductive gel was applied to maintain electrode impedance below 20 kΩ (slightly above typical impedance value, but still acceptable given the EEG amplifier’s input impedance of over 1 GΩ, which may help limit the impact of slightly higher scalp-electrode impedance³². Participants remain relaxed with eyes closed for 5 min, avoiding movement or cognitive activity. Data were collected at a sampling frequency (Fs) Fs = 1024 Hz using EEGO software (ANT Neuro, Enschede, Netherlands) and stored for offline analysis.

EEG data processing

The EEG signal processing was performed using Matlab version 9.14.0 (The MathWorks Inc, Natick, MA, USA) with EEGLAB toolbox. The EEG data was bandpass filtered from 1 to 45 Hz and a notch filter was set at 50 Hz (low/high cutoff frequencies: 49–51 Hz) using a zero-phase finite impulse response filter in both forward and reverse directions. The EEG signal was then re-referenced to the two mastoids electrodes (M1 and M2). For bad segment rejection, we used a fully automated approach based on absolute amplitude thresholding of the pre-processed EEG signal to ensure reproducibility. Multiple thresholds were evaluated: overly permissive thresholds led to noisy spectra with excessive low-frequency power, while overly stringent thresholds resulted in substantial data loss³³. Based on empirical evaluation, signal segments were rejected when their absolute amplitude exceeded 50 V, a threshold that provided a suitable balance between signal quality and data retention³⁴. On average, fewer than 10% of EEG segments were rejected per participants. The Power Spectral Density (PSD) was computed for each channel using Welch’s method with non-overlapping segments of length Fs=1024 data points (1 s) with a Hamming window applied. A modified Matlab pwelch function was derived not to consider segments where data had been rejected to avoid artefacts due to signal discontinuity.

EEG data analysis

To determine a primary outcome for cortical activity, the BSI was calculated. The BSI is an index that evaluates the symmetry across homologous channel pairs. In the present study, we focused on the comparison between spatial regions, not specific sub-bands (delta, theta, alpha, beta), as previously described by Fanciullacci et al., 2017³⁵. The symmetry is evaluated based on the comparison between the EEG power spectral density averaged over frequencies and channel pairs for each hemisphere. The BSI is a numerical value ranging from 0 (perfect symmetry) to 1 (maximal asymmetry). Data from twenty-five channel pairs were used for analysis: Fp1-Fp2, AF3-AF4, AF7-AF8, F1-F2, F3-F4, F5-F6, F7-F8, FC1-FC2, FC3-FC4, FC5-FC6, FT7-FT8, C1-C2, C3-C4, C5-C6, T7-T8, CP1-CP2, CP3-CP4, CP5-CP6, P1-P2, P3-P4, P5-P6, P7-P8, PO3-PO4, PO7-PO8, O1-O2. The frequency range for the analysis was set to 1–45 Hz. We implemented the formula of a revised version of BSI, initially proposed by van Putten³⁶ defined as Eq. (1) for full formula please see Additional file 1.

1

The average square module of the Fourier coefficients of channels on the right hemisphere (same formula is applied for for the left hemisphere); = A time index representing a time segment and = The number of Fourier coefficients.

We used the PSD as described above to estimate the Fourier coefficients over the duration of a complete eyes closed phase. We calculated the BSI in four different brain areas composed of whole brain, frontal, central, and posterior area as suggested in a previous study³⁵.

Reaction time (RT)

RT included both correct and incorrect responses, but incorrect or outlier RTs (±3 SD from the mean) were excluded³⁷. On average, fewer than 4 trials (approximately 3.3% of 120 trials) were removed per block. Pseudorandom block (blocks 1 and 6) were excluded from the RT analyses to minimize potential confounding effects related to initial learning and fatigue, as previous SRTT study has shown that random blocks could reflect practice or non-learning processes rather than sequence learning proper^38,39. Mean RTs were computed for each participant from correct trials within each block (Pre: baseline, During: blocks 2–5, Post: blocks 7 and 8) and entered into group-level analyses.

Statistical analysis

In cross-over design, the sample size calculation was based on the behavioral outcome (RT). Partial eta-squared ( = 0.171) from a previous tDCS study using the SRTT in healthy population was used, corresponding to an effect size of 0.45²⁵. With = 0.05, and power = 0.80, the required sample size was 16. To accommodate a 20% dropout rate, 20 participants were randomized. All analyses were conducted using SPSS software (IBM SPSS Statistics, Version 23, USA). Baseline variables were analyzed with repeated measures ANOVA. Normality was assessed using the Shapiro-Wilk and Kolmogorov-Smirnov tests. A two-way repeated measures ANOVA examined effects of intensity (sham, 1.0, 1.5, 2.0 mA) and time (Pre, During, Post). Mauchly’s test was used to assess the assumption of sphericity, and the Greenhouse-Geisser corrections were applied when this assumption was violated to adjusted degrees of freedom. Post-hoc Bonferroni comparisons were performed for pairwise comparison (within-group and between group comparisons). Additionally, both the between- and within-group effect size were calculated from the mean and SD using Cohen’s formula^40,41. Cohen’s d-values of 0.20, 0.50, and 0.80 were interpreted as small, moderate, and large effect size, respectively⁴¹. A learning index was calculated to quantify sequence-specific learning as the difference between mean RT of random blocks (Blocks 1 and 6) and sequence blocks (Block 2–5, 7–8), with higher values indicating greater sequence-specific learning. This index was used for subsequence repeated measures ANOVA to assess the effect of tDCS intensity on motor sequence learning. A mixed-effects model was used to assess order effect. Statistical significance was set at p < 0.05.

Pearson correlation was used to assess the relationships between changes in RT and BSI. Correlation analyses across multiple brain regions were corrected for multiple comparisons using the Benjamini-Hochberg method⁴² with a false discovery rate (FDR) of q = 0.1, as applied in previous tDCS study⁴³. Correlations with FDR-adjusted p-value below 0.1 were considered significant.

Intention-to-treat analysis was applied, and missing data were handled using mean imputation.

Results

Participant demographics are presented in Table 1. All participants completed the 4 bilateral tDCS conditions. Order effect analysis revealed no significant interaction effect (p > 0.05). Blinding was assessed via end-of-study guesses; 45% of participants correctly identified the sham condition, which was not significantly different from chance (χ² = 16.488, p = 0.057). Adverse events are shown in Fig. 4. The most common events were mild itching and tingling observed in all stimulation conditions, and these effects disappeared after stimulation ended.

Table 1.

Participants characteristics.

Variables	Mean SD or n (n = 20)
Sex (male/ female), n	9/ 11
Age (years)	22.60 2.54
BMI (kg/m2)	21.25 3.28

Fig. 4.

Adverse events after the experiment following sham, 1.0 mA, 1.5 mA, and 2.0 mA.

BSI evaluated by EEG

There was no significant difference in global BSI, frontal BSI, central BSI, and posterior BSI among the 4 intensities at baseline (Additional file 1: Table S1).

Global BSI: As shown in Table 2; Fig. 5A, a two-way repeated measures ANOVA with a Greenhouse-Geisser correction revealed a significant time x intensity interaction (F_{(2.018, 38.338)} = 6.347, p = 0.004, = 0.250). Within-group comparisons indicated a significant increase from Pre to Post in the 1.0 mA (p = 0.028, Cohen’s d = 0.769) and the 2.0 mA (p = 0.006, Cohen’s d = 0.912) groups. Between-group comparisons at Post indicated higher global BSI in the 2.0 mA group vs. sham (p = 0.013, Cohen’s d = 1.241), 2.0 mA vs. 1.0 mA (p = 0.021, Cohen’s d = 1.283), and 2.0 mA vs. 1.5 mA (p = 0.004, Cohen’s d = 1.390).

Table 2.

Raw data of BSI for global, frontal, central, and posterior areas and p-value for overall analysis and post-hoc comparison.

Brain area	Intensity	Mean ± SD		Time effect (p-value)		Intensity effect (p-value)			Interaction effect (Time x Intensity) (p-value)
		Pre	Post	Overall analysis	Post-hoc analyses	Overall analysis	Post-hoc analyses
					Pre vs. Post		Pre	Post
Global	Sham	0.0300.019	0.0290.019	0.016*	0.884	0.002**	All pairwise comparaisons p > 0.050	0.013* (Sham vs. 2.0 mA) 0.021* (1.0 vs. 2.0 mA) 0.004** (1.5 vs. 2.0 mA) Other pairwise comparaisons p > 0.050	0.004**
	1.0 mA	0.0210.012	0.0310.014		0.028*
	1.5 mA	0.0330.023	0.0240.020		0.115
	2.0 mA	0.0340.029	0.0650.039		0.006**
Frontal area	Sham	0.0560.037	0.0440.036	0.768	0.342	0.006**	All pairwise comparaisons p > 0.050	0.021* (1.5 vs. 2.0 mA) Other pairwise comparaisons p > 0.050	0.043*
	1.0 mA	0.0300.024	0.0380.026		0.353
	1.5 mA	0.0630.050	0.0490.037		0.345
	2.0 mA	0.0480.032	0.0740.046		0.015*
Central area	Sham	0.0440.031	0.0500.039	0.001**	0.600	0.289	All pairwise comparaisons p > 0.050	All pairwise comparaisons p > 0.050	0.034*
	1.0 mA	0.0360.043	0.0400.025		0.736
	1.5 mA	0.0410.026	0.0450.024		0.560
	2.0 mA	0.0310.019	0.0710.043		0.001**
Posterior area	Sham	0.0640.049	0.0810.075	0.084	0.399	0.197	All pairwise comparaisons p > 0.050	All pairwise comparaisons p > 0.050	0.539
	1.0 mA	0.0640.050	0.0990.096		0.180
	1.5 mA	0.0560.042	0.0650.046		0.463
	2.0 mA	0.0590.035	0.0590.042		0.952

Testing by a two-way repeated measures ANOVA with post-hoc analysis.

*p-value < 0.050, **p-value < 0.010.

Fig. 6.

Direction of hemisphere asymmetry. Positive values indicate greater right hemisphere activity, whereas negative values indicate greater great to left hemisphere activity (see Additional File 1).

Frontal BSI: A two-way repeated measures ANOVA revealed a significant time x intensity interaction (F_{(3, 57)} = 2.903, p = 0.043, = 0.133). Within-group analysis indicated a significant increase from Pre to Post only in the 2.0 mA group (p = 0.015, Cohen’s d = 0.667). Between-group comparisons at Post indicated higher BSI in the 2.0 mA group compared to 1.0 mA (p = 0.021, Cohen’s d = 1.000).

Central BSI: A two-way repeated measures ANOVA revealed a significant time x intensity interaction (F_{(3, 57)} = 3.093, p = 0.034, = 0.140). Post-hoc analysis revealed a significant increase from Pre to Post only in the 2.0 mA group (p = 0.001, Cohen’s d = 1.290).

Posterior BSI: A two-way repeated measures ANOVA reported no significant interaction between time and intensity (p > 0.05).

Figure 5 illustrates BSI of different brain areas change over time, and Fig. 6 summarizes the direction of hemisphere asymmetry.

Fig. 5.

Changes in BSI over time across global (A), frontal (B), central (C), and posterior (D) regions. * p-value < 0.05, ** p-value < 0.01.

Serial reaction time task (SRTT)

Table 3 presents the comparison of RT across stimulation intensities. No significant differences in RT were observed at baseline among the 4 intensities (Additional file 1: Table S1). A two-way repeated measures ANOVA revealed a time x intensity interaction (F_{(6, 114)} = 2.222, p = 0.046, = 0.105). Within-group comparisons indicated a significant RT reduction in the 2.0 mA group from Pre vs. During (p = 0.008, Cohen’s d = 0.663) and Pre vs. Post (p = 0.001, Cohen’s d = 0.864). Between-group comparisons showed significantly faster RT at during stimulation for 2.0 mA compared with sham (p = 0.037, Cohen’s d = 0.889). Post-stimulation RT was faster for 2.0 mA than sham (p = 0.008, Cohen’s d = 1.196) and than 1.0 mA (p = 0.002, Cohen’s d = 1.068).

Table 3.

Raw data of RT for all groups and p-value for overall analysis and post-hoc comparison.

Intensity	Mean ± SD			Time effect (p-value)				Intensity effect (p-value)				Interaction effect (Time x Intensity) (p-value)
	Pre	During	Post	Overall analysis	Post-hoc analyses			Overall analysis	Post-hoc analyses
					Pre vs. During	Pre vs. Post	During vs. Post		Pre	During	Post
Sham	393.8153.10	390.0748.93	392.0144.37	0.027*	1.000	1.000	1.000	< 0.001***	All pairwise comparaisons p > 0.050	0.037* (Sham vs. 2.0 mA) Other pairwise comparaisons p > 0.050	0.008** (Sham vs. 2.0 mA) 0.002** (1.0 vs. 2.0 mA) Other pairwise comparaisons p > 0.050	0.046*
1.0 mA	384.6153.39	374.9564.44	384.4241.89		0.333	1.000	0.135
1.5 mA	374.2156.78	358.3643.40	369.6136.90		0.228	1.000	0.211
2.0 mA	377.5348.11	341.7459.78	333.4853.86		0.008**	0.001**	1.000

Testing by a two-way repeated measures ANOVA with post-hoc analysis.

*p-value < 0.050, **p-value < 0.010, ***p-value < 0.001.

Repeated measures ANOVA on the learning index demonstrated a significant of stimulation intensity (F_{(3, 57)} = 5.304, p = 0.003, = 0.218). Post-hoc comparisons showed higher learning-index value in the 2.0 mA condition compared with sham (p = 0.019).

Correlation between change of motor performance and brain symmetry index

Pearson correlation analysis across the 4 brain regions were corrected for multiple comparisons using the Benjamini-Hochberg FDR procedure with q = 0.10. The negative correlation between changes in BSI and RT in the central area remained significant after FDR correction (r = -0.518, raw p = 0.019, FDR-adjusted p = 0.076), suggesting that greater asymmetry was associated with better motor performance (Fig. 7). No significant correlations were found in other regions (Additional file 1: Figure S1-3).

Fig. 7.

The correlation between change of motor performance and change of BSI in the central area (p-value is shown without adjustment).

Discussion

In this cross-over randomized controlled trial, we reported BSI values obtained by EEG data and motor performance during the SRTT following bilateral tDCS at 1.0, 1.5, 2.0 mA and sham stimulation in healthy individuals. An order analysis showed no effect of session sequence. Blinding was evaluated through end-of-study guesses. 45% of participants identified the sham condition correctly, a rate consistent with chance-level guessing, indicating successful blinding. Our findings revealed that a single session of 2.0 mA bilateral tDCS for 20 min significantly increased BSI modulation in the global, frontal, and central areas at immediately post-intervention compared to 1.0 mA, 1.5 mA, and sham, with large effect sizes, with the direction of BSI indicating higher cortical activity in the left hemisphere (under anodal stimulation) across all brain areas. Additionally, the 2.0 mA group exhibited a significant reduction in RT compared to other intensities and sham stimulation, also with large effect sizes. Furthermore, the changes of RT induced by 2.0 mA was significantly correlated with changes in BSI.

Intensity-response relationship

Previous studies showed both linear and non-linear relationship between tDCS intensity and neurophysiological or motor responses, resulting in conflicting findings. For unilateral anodal tDCS, cortical excitability increases linearly with intensity up to 1.0 mA^12,44 and up to 3.0 mA¹⁷, similar to the increase in cerebral blood flow (CBF) observed with intensity ranging from 0.5 to 2.0 mA⁴⁵. However, Jamil et al., 2017 showed no clear evidence for a linear relationship of anodal stimulation, showing facilitation of cortical excitability for all intensities (0.5–2.0 mA) relative to sham, with no difference between stimulation intensities. Moreover, a non-linear effect was found on motor performance, as assessed by reaction time following unilateral anodal tDCS. An inverted U-shaped dose-response curve was observed, with 1.0 mA showing the strongest impact on RT compared to lower (0.8 mA) and higher (2.0 mA) intensities⁴⁶. For unilateral cathodal tDCS, it typically induces inhibitory effects at lower intensities but may reverse to excitatory effects at high intensities. For instance, cortical excitability was reduced following 1.0 mA^47,48, whereas increases were reported at 2.0 mA^28,47,49,50 and 3.0 mA⁵⁰. However, mix effects have been reported at 2.0 mA^51,52, while a linear response from 0.5 to 2.0 mA was observed with CBF⁴⁵. In our previous work, unilateral cathodal tDCS at 1.0 and 1.5 mA reduced upper and lower limb muscle strength bilaterally, whereas 2.0 mA reversed this effect¹⁸. To our knowledge, no previous study has examined the dose–response relationship of bilateral tDCS on cortical or behavioural outcomes. In the present study, bilateral tDCS at 2.0 mA induced greater cortical and motor performance responses than lower intensities, while no differences were observed among 1.0, 1.5 mA, and sham conditions. Our results demonstrated a non-linear intensity-response, where only the highest intensity (2.0 mA) was sufficient to induce significant modulation of both cortical activity and motor performance.

Interhemispheric inhibition

Evidence in human studies shows that tDCS could induce depolarization or hyperpolarization of the resting membrane potential depending on the polarity used, thereby modulating cortical excitability and spontaneous firing rate^48,53. A previous paired-pulse TMS study demonstrated that bilateral tDCS with anodal electrode over the M1 increased cortical excitability, while cathodal electrode over the opposite M1 simultaneously reduced it. Additionally, this configuration increased IHI from the hemisphere under the anodal electrode and reduced IHI from the hemisphere under the cathodal electrode¹⁴. A previous fMRI study showing that bilateral tDCS over the M1 modulated interhemispheric functional connectivity via transcallosal pathways during stimulation and increased intracortical connectivity within the M1 region under the anodal electrode after stimulation. In the present study, we used BSI as an indicator of IHI. The baseline global BSI observed in this study (0.033 ± 0.027) was consistent with values reported by van Putten and Tavy⁵⁴, who documented a mean global BSI of 0.042 ± 0.005 in healthy controls. The direction of BSI found in the present study indicated higher cortical activity in the left hemisphere (under anodal stimulation) across all brain areas following bilateral tDCS at 2.0 mA. This effect may be induced by bilateral tDCS over both M1 regions. This finding aligns with the paired-pulse TMS and fMRI studies mentioned above. Moreover, statistical analysis indicated significant changes in the global, frontal, and central areas. This widespread BSI aligned with the current-flow modeling showing that bilateral tDCS at 2.0 mA affected a widespread area, not only the M1 (see Fig. 2). This aligns with a previous fMRI study reporting anodal tDCS over the M1 (unspecified intensity) modulated brain networks not only the stimulated motor cortex but also bilateral frontal, parietal, and premotor cortical areas compared to sham stimulation⁵⁵. Moreover, wide current spread was observed only with larger electrodes (4 × 4.5 cm), not with smaller electrodes (3 × 3 cm) when using 1.0 mA anodal-tDCS⁵⁶. In our protocol, the used of 5 × 5 cm electrodes may have contributed to wide current distribution. However, it should be noted that anatomical factors in individual’s subjects (i.e., thickness of the skull and the cerebrospinal fluid, and the gyral depth) could also affect strength of electric field⁵⁷.

Considering the mechanism underlying unilateral hand movement, it requires interhemispheric interaction involving network-level communication between cortical and subcortical areas, including the supplementary motor area (SMA), the dorsal premotor cortex, the ipsilateral M1, and the basal ganglia. These interactions are necessary to restrict motor output to the contralateral M1 that controls the hand movement⁸. These network communications may account for the widespread BSI changed observed in the present study.

Motor performance

Significant reductions in RT were observed between Pre vs. During and Pre vs. Post only in the 2.0 mA group. Moreover, significant between-group differences were found both During and Post between 2.0 mA and sham stimulation and as well as between 2.0 mA and 1.0 mA at Post. These results suggest that 2.0 mA bilateral tDCS induced greater RT improvements compared to other intensities. This finding aligns with previous studies showing no enhancement in RT during the SRTT following 1.0 mA anodal-tDCS applied over the M1⁵⁸ or over the cerebellum⁵⁹. However, another study suggested that applying 1.5 mA of both anodal-tDCS and bilateral tDCS over the M1 for three consecutive days enhanced motor skill learning⁶⁰. A recent systematic review and meta-analysis indicated that multiple sessions of anodal tDCS, compared to a single session, significantly improve the SRTT performance⁶¹. The lack of significant effects at lower intensities in the current study may be attributed to the use of only a single session of tDCS.

Correlation between BSI and motor outcome

A significant improvement in RT following 2.0 mA stimulation was correlated with IHI modulation assessed by BSI specifically in the central area. Previous studies have highlighted the role of corpus callosum-mediated IHI in motor-related cortical areas⁶² supporting our findings that enhanced IHI facilitation may contribute to improved motor performance⁶³, Studies in healthy individuals indicate that during unilateral hand movements, IHI favors facilitation in the contralateral hemisphere, enhancing motor performance as measured by RT for tasks such as unilateral voluntary finger abduction^64,65. This suggests that modulation of IHI, particularly in the dominant hemisphere during right-hand tasks, may release motor cortical representations more effectively compared to nondominant tasks⁶⁵.

In healthy right-handers, IHI is balanced at rest. However, during right-hand movements that require increased brain asymmetry, the right hemisphere reduce its inhibition onto the left hemisphere. This reduction allows the left hemisphere to become more active, facilitating the initiation of right-hand movement^8,66. In this study, an increase in BSI was associated with improved motor performance, suggesting that modulating IHI through bilateral tDCS could potentially benefit unilateral movements.

Strengths, limitations, and suggestion for further study

The present study employed a cross-over design, which could minimize inter-subject variability when comparing conditions. This design allows for the same level of accuracy as a parallel-group design⁶⁷. However, the four-condition setup may have introduced a learning effect, as participants’ performance in later conditions could improve due to familiarity with the task⁶⁸. To avoid this, a two-week washout period, longer than the 7-days washout period used in previous study, was implemented to eliminate any residual stimulation effects and minimize learning effects of the motor learning task²⁵. An order analysis was statically performed indicating that the session sequence did not influence the outcomes. Thus, this reduces the possibility of potential carryover effects of multiple tDCS sessions in our crossover design. Moreover, a previous study using a similar protocol with 2.0 mA tDCS for 20 min during a motor task found no retention effects 14 days after the intervention⁶⁹. Additionally, baselines performance across all four conditions showed no significant differences, suggesting that any observed effects were not influenced by baseline variability. Next, it should be noted that BSI reflects interhemispheric balance based on differences in hemispheric activity, rather than providing a direct measure of inhibitory interactions between hemispheres as obtained from paired-pulse TMS. Another limitation is the absence of long-term effect investigation, thus the reported effect was acute and may be transient. Further study should address this limitation. Lastly, this study was conducted in healthy young adults, and the results may differ in other populations. Further studies involving older adults or individuals with motor impairments, such as stroke, are crucial for better understanding the therapeutic potential of tDCS in enhancing motor recovery.

Conclusion

A single session of bilateral tDCS (anodal over the left hemisphere and cathodal over the right hemisphere) for 20 min at an intensity of 2.0 mA produced greater modulation of BSI, which was used to assess the IHI between the two hemispheres, showing greater cortical activity in the left hemisphere particularly in the global, frontal, and central areas at the end of stimulation, compared to 1.0 mA, 1.5 mA, and sham stimulation in healthy individuals. There were no differences between 1.0, 1.5 mA and sham stimulation. At the performance level, bilateral tDCS at 2.0 mA greater reduced reaction time performed by unilateral dominant-hand movement, compared to other intensities and sham stimulation. Moreover, changes in reaction time were significantly correlated with BSI modulation. These findings may help in selecting tDCS intensity, particularly when targeting IHI modulation, and unilateral motor performance.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

WK supervised the study. TP and WK contributed to the study planning. TP, IA and SH performed data collection. TP, JD, IA, AT and WK contributed to the data analysis and drafted the manuscript. BA and VS provided critical feedback and helped shape the study. All the authors have read and agreed to the published version of the manuscript.

Funding

This research project has been supported by Mahidol University (Fundamental Fund: fiscal year 2025 by National Science Research and Innovation Fund (NSRF)) (awarded to WK). TP was supported by Scholarships for Ph.D. Student from Mahidol University. SH was supported by research assistance grant by Faculty of Physical Therapy, Mahidol University.

Data availability

Data will be made available on request. The request for analyses of data from this study should be directed to the corresponding author (wanalee.klo@mahidol.edu).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval and consent to participant

A study description was provided for all participants prior to the experiment. All participants received a participant information sheet and provided written informed consent. The study was conducted in accordance with the Declaration of Helsinki, approved by the Mahidol University Central Institutional Review Board (COA No. MU-CIRB 2023/051.2803), and the trial was registered on ClinicalTrials.gov (25/06/2023, NCT05929014).