Transcranial direct current stimulation (tDCS) has demonstrated its efficacy in enhancing the acquisition and mastery of surgical skills when administered to specific cortical regions, such as the primary motor cortex (Cox et al., 2020; Gao et al., 2021a) and the prefrontal cortex (Ashcroft et al., 2020). A comprehensive meta-analysis of randomized control studies revealed that tDCS led to significantly superior performance in surgical tasks compared to sham control conditions (Hung et al., 2021; Walia et al., 2021). In a study on bimanual pattern cutting involving tDCS of the primary motor cortex (M1) and portable neuroimaging (Gao et al., 2021a), Gao et al. (2021b) observed a significant effect of M1 tDCS on performance errors (p < 0.001; t-test for normal distribution or Mann–Whitney U-test for non-normal distribution) after day 7 compared to the sham group. Notably, a delayed effect of M1 tDCS emerged after day 7, postulated to be linked to the significant activation of M1 observed only during the later learning stage (day 7–12) in contrast to the initial learning stage (day 2–6). This aligns with established in vivo effects of tDCS, which do not alter the firing rates of cortical neurons (Krause et al., 2017) but instead modulate endogenous task-specific brain activity (Woods et al., 2019). Therefore, neuroimaging can offer insight into the cortical “target” associated with endogenous task-specific brain activity (Walia et al., 2021).

Given that complex motor activities engage diverse brain regions associated with motor control and attention (Alahmadi et al., 2016), a thorough mechanistic understanding of the task-specific brain activity for region-specific tDCS targeting is necessary (Walia et al., 2021). For instance, the dorsolateral prefrontal cortex (PFC) plays a key role in goal-directed behavior and learning, involving coordinated activity within a broader frontoparietal network (James et al., 2013). Changes associated with visuomotor learning may encompass reduced activation in the PFC, coupled with increased activation in the primary and secondary motor regions, the cerebellum, and the posterior parietal cortex (James et al., 2013). In the context of novel environments for skill learning, such as laparoscopic surgery (FLS Trainer System Accessories, 2010), visuomotor learning requires perceptual learning (Kamat et al., 2022) engaging multiple brain regions (Dosenbach et al., 2007). This includes the dorsal attention network for visuospatial awareness and the salience network to focus attention on pertinent stimuli, such as visuomotor task-errors. Consequently, appropriate brain targets for tDCS necessitates consideration of relevant nodes within brain networks (Bressler and Menon, 2010) that are directly linked to the performance and learning of the specific visuomotor skill. In this study, we targeted two relevant brain nodes with tDCS, based on our published model (Walia et al., 2022a), to facilitate error-based learning that constitutes a fundamental mechanism in skill acquisition, encompassing error detection, correction, and subsequent performance adjustments (Seidler et al., 2013). Our application of tDCS in this study was guided by insights gained from analyzing error-related brain states by utilizing a combination of electroencephalography (EEG) and functional near-infrared spectroscopy (fNIRS) during laparoscopic suturing with intracorporeal knot tying task (Walia et al., 2022b). Fundamentals of Laparoscopic Surgery (FLS) suturing with intracorporeal knot tying is a complex motor task requiring high precision hand-eye coordination, depth perception from the 2D view (Dilley et al., 2020), and tool control for optimal performance (Hannah et al., 2022). Top panel of Figure 1 shows the six microstate prototypes discerned from the amalgamated EEG data encompassing both experts and novices performing the FLS complex task, elucidating 77.14% of the overall variance (Walia et al., 2022a). Figures 1A–E show color coded transition probabilities between microstate (MS) classes at the group level with Figure 1A showing that during the 10-s at the start of the FLS task in novices, Figure 1B show that during the 10-s at the start of the FLS task in experts, Figure 1C showing that during the 10-s in the error epoch in novices, and Figure 1D showing that during the 10-s in the error epoch in experts. Here, transition probabilities between MS classes during 10-s in the error epoch showed typical patterns when compared to 10-s at the start of the FLS task. Notably, differences in task-error perception and attention redirection for corrective actions were posited to vary between experts and novices (Walia et al., 2022a) where transition to MS 3 was highlighted in experts (Figure 1D) when compared to novices (Figure 1C) during the 10-s in the error epoch. Here, MS 3 exhibited topographical similarity to Bréchet et al. (2019) microstate D, with primary activity sources located bilaterally in the inferior frontal gyrus (IFG), dorsal anterior cingulate cortex (dACC), and the superior parietal lobule (SPL)/intraparietal sulcus (IPS). In addition, transition to MS 1 was highlighted in both experts (Figure 1D) and novices (Figure 1C) during the 10 s in the error epoch where MS 1 bears the closest topographical resemblance to Brechet and colleagues' microstate C (Bréchet et al., 2019), with sources situated bilaterally in the lateral portion of the parietal cortex, encompassing both the supramarginal gyrus (SMG) and angular gyrus (AG). Therefore, we proposed a model for task-error perception and attention redirection for corrective actions (Walia et al., 2022a) that posited attention driven by unexpected stimuli (errors), subserved by the engagement of the SMG and AG from ventral attention stream in both experts and novices – see Figure 1E. Then, in the experts, an adaptive control of goal-directed attention within the frontoparietal network system was posited, shown in Figure 1E, subserved by the activation of SPL and IPS due to the involvement of dorsal attention systems pertinent to the reorientation of visuospatial attention. Since the frontoparietal network, believed to support adaptive control, exhibited a 2.3-fold overrepresentation in the cerebellum compared to the cerebral cortex (Marek et al., 2018); therefore, cerebellar node was selected as one of the brain nodes for tDCS to facilitate error-based learning. Then, the other brain node for tDCS was selected as left ventrolateral prefrontal cortex (PFC) to facilitate subjective task-error awareness (Wessel, 2012) by modulating the left lateralized brain network for visuospatial attention (De Schotten et al., 2011; Vossel et al., 2014). Our model for task-error perception and attention redirection for corrective actions (Walia et al., 2022a) was supported recently by Choo et al. (2023) who identified contributions from the left lateral prefrontal, occipital, sensorimotor, and parietal channels for post-error adjustments. Specifically, the left lateral prefrontal channels played a predominant role in post-error adjustments, manifesting from around 900 ms into extended post-error periods (Choo et al., 2023). Consequently, modulating task-error perception and attention redirection for corrective actions using tDCS, assessed through portable brain imaging methods such as fNIRS (Gao et al., 2021a) and EEG (Ciechanski et al., 2019), and correlating these changes with tDCS modulated behavioral effects can offer mechanistic insights. In our multi-modal fNIRS-EEG approach to assess tDCS effects on the brain (Guhathakurta and Dutta, 2016), regularized temporally embedded canonical correlation analysis developed in our prior work (Walia et al., 2022a) was applied in the current study to find EEG band power changes that corresponded with the oxyhemoglobin (HbO) concentration changes from fNIRS data.

Post-tDCS change in the FLS performance measures, namely, task time, deviation, incision gap, knot security, accuracy error, from pre-tDCS baseline values are provided in the Supplementary Figure S4 while Supplementary Figure S5 presents the corresponding descriptive statistics table. A significant (q < 0.05) change in the oxyhemoglobin (HbO) hemodynamic response during the FLS task was found for change post-tDCS (POST) from pre-tDCS (PRE) for CER tDCS when compared to SHM tDCS – see Figure 4A. Then, CER tDCS effect on the change (POST-PRE) in the FLS score for CER tDCS when compared to SHM tDCS was significant (p < 0.05) – see Figure 4B. Figure 4A shows that the CER tDCS modulation of the HbO hemodynamic response was the highest (q < 0.05) at the fNIRS channel (source 4 and detector 1) primarily overlying left middle frontal gyrus (AAL: 'frontal_mid_l') and fNIRS channel (source 10 and detector 9) primarily overlying right inferior frontal gyrus, triangular part (AAL: 'frontal_inf_tri_r') based on the sensitivity (weights) of the fNIRS probe (Aasted et al., 2015) as listed in Table 2 (also see Supplementary Table S1). In contrast, the HbO hemodynamic response and the FLS behavioral effects (FLS score) of the PFC tDCS were not significant, as shown in Figure 5. Table 3 presents the descriptive statistics of post-tDCS change in the FLS score from pre-tDCS baseline values (ScoreDiff) for various interventions namely CER tDCS, PFC tDCS, SHM tDCS. Post hoc power analysis for normal distribution of FLS score (ScoreDiff for CER tDCS and PFC tDCS – see normality test results in Supplementary Figure S3) provided a post-hoc power of 0.05 for CER tDCS vs. PFC tDCS intervention.

Figures 6A–E show the Independent Components (IC) clusters for the CER tDCS session that are averaged group dipoles with density at the baseline (PRE) for the FLS task epoch. ERSP at the baseline (PRE) for the FLS task epoch (right bottom panel) indicated activity starting after 1 s where the averaged 1Hz−40Hz scalp topography changes post-tDCS (POS) reached significance (p < 0.05, FDR corrected) within the 5 s, as shown in the Figure 6F. Figure 6G shows the averaged 1HZ−8Hz scalp topography changes post-tDCS (POS) from baseline that also reached significance (p < 0.05, FDR corrected) within the 5 s. Then, the hemodynamic response due to neurovascular coupling from 5 s to 15 s based on the cortical HbO activation (Aasted et al., 2015) at the start of the FLS task post-tDCS are shown in Figure 7 for CER and SHM tDCS. Supplementary Figure S6 shows the corresponding comparison between PFC tDCS (see Supplementary Figure S6a) and for SHM tDCS (see Supplementary Figure S6b).

Our results showed a significant effect of cerebellar (CER) tDCS on POST-PRE HbO hemodynamic response during 0 s−20 s of the FLS suturing with intracorporeal knot tying task when compared to sham (SHM) tDCS. This significant response was primarily (highest optode sensitivity weight) at the left middle frontal gyrus and right inferior frontal gyrus (triangular part) which was expected due to CER tDCS targeting the posterior lobules [see Supplementary Figure S1, also shows the regions of the human cerebellum with color coding their functional connectivity to 7 major networks in the cerebrum (Buckner et al., 2011)] (Rezaee et al., 2021), in particular Crus I and II, that has projections to the prefrontal cortex (however, spared anatomical connections to motor cortex) (Buckner et al., 2011). Table 2 shows AAL regions, 'frontal_inf_tri_l,' 'frontal_inf_tri_r,' 'frontal_mid_l,' 'frontal_mid_r,' 'frontal_sup_l,' 'precentral_r,' that are all underlying the significant fNIRS optode pairs (source 4 and detector 1, source 10 and detector 9) with different sensitivity weights. CER tDCS facilitated FLS task related ERSP difference seems to be driven by the left prefrontal scalp topography – see the associated p-value map at the extreme right for CER tDCS (Figure 6F) when compared to SHM tDCS (Figure 6G). Here, right CER tDCS facilitated the left prefrontal activation ('frontal_inf_tri_l,' 'frontal_mid_l,' 'frontal_sup_l') at the start of the FLS task that was found underlying the fNIRS source 4 and detector 1, as evident in Figure 7A, when comparing post-tDCS cortical activation from pre-tDCS baseline.

CER tDCS modulation can be associated with the frontoparietal facilitation (Lückmann et al., 2014) of higher cognitive functions (du Boisgueheneuc et al., 2006) and proactive control (Hu et al., 2016) respectively – both relevant in error-based learning known to be associated with cerebellum. Here, frontal and parietal regions are attributed roles in attentional mechanisms, suggesting their involvement in allocating attentional resources to perceptual representations necessary at the beginning of the FLS task learning [perceptual learning (Kamat et al., 2022)]. We found that the CER tDCS effects on the behavior was significant (p < 0.05), as shown in the Figure 4B, with an improved POST-PRE change in the FLS score when compared to sham tDCS. Then, left ventrolateral prefrontal (PFC) tDCS modulation was postulated to facilitate task-error perception; however, that did not have a significant effect on the behavior when compared to sham tDCS, as shown in the Figure 5B. Also, left PFC tDCS effect on the HbO hemodynamic response was not significantly different from sham tDCS (see Figure 5A). Supplementary Figure S3 shows PFC post-tDCS effect at the left prefrontal areas ('frontal_mid_l,' 'frontal_sup_l') underlying fNIRS optode pair, source 4 and detector 15, when compared to pre-tDCS baseline. However, left PFC tDCS was postulated to facilitate the perception of the task-error but was not sufficient to facilitate brain activation at the start of the FLS task when compared to sham tDCS. Here, CER tDCS facilitation of the prediction model of the sensory outcome of action is postulated to be more important in skill-naïve novices involved in perceptual learning (Kamat et al., 2022) when compared to task-error perception with sensory error prediction model not available in skill-naïve (Tsay et al., 2022). Therefore, we postulate based on our study results that CER tDCS facilitation of sensory outcome prediction of action was helpful in novices who are FLS skill-naive (Crihfield et al., 2023) and are at the start of the skill learning (Mizuguchi et al., 2018). As the novices gain skill, the left PFC tDCS may facilitate task-error perception to modulate implicit motor recalibration driven together with sensory prediction errors (Tsay et al., 2022). Nevertheless, the effect size of CER tDCS vs. PFC tDCS was relatively small (post-hoc power of 0.05), and both verum tDCS interventions improved FLS score from SHM tDCS's median ScoreDiff of 35 to left PFC tDCS's median ScoreDiff of 65 and right CER tDCS's median ScoreDiff of 92 (see Table 3), i.e., 85% improvement for left PFC tDCS and 162% improvement for right CER tDCS from SHM tDCS's median ScoreDiff. Notably, bilateral prefrontal cortex (anode over F3, cathode over F4) tDCS (Ashcroft et al., 2020; Patel et al., 2021) resulted in a POST-PRE performance score difference in open knot-tying task of 17 for verum tDCS and −17.5 for sham tDCS.

In the current study, right CER tDCS and the related HbO activity at the left inferior and middle frontal gyrus (AAL: 'frontal_mid_l') during the start of the FLS task can be related to the encoding of sensory prediction errors (Schlerf et al., 2012; Ficco et al., 2021), that can be postulated to interact with the task-errors (Tsay et al., 2022) at later stages in skill learning for an improved FLS score. Then, the activation of the right inferior frontal gyrus, triangular part (AAL: 'frontal_inf_tri_r') can be related to predictive error related stopping specifically under situations with high uncertainty (Levy and Wagner, 2011); therefore, a possible coordination with the HbO activity at the left inferior and middle frontal gyrus for error-related perception action coupling can be postulated. Indeed, CER tDCS improved FLS accuracy (see descriptive statistics on ErrorDiff in Supplementary Figure S5) that trended toward improvement better than SHM and PFC tDCS but was not found to be statistically significant (p < 0.05). Here, future studies can investigate transcranial alternating current stimulation of the cortico-cerebello-thalamo-cortical loop that has been shown feasible in other studies (Singh et al., 2021; Walia et al., 2022c).

Our results on tDCS-modulated brain activation and behavioral effects aligned with numerous functional magnetic resonance imaging (fMRI) and fNIRS studies that have been published on skill learning (Roberts et al., 2006; Leff et al., 2007, 2008a,b; Wanzel et al., 2007; Ohuchida et al., 2009; Khoe et al., 2020; Gao et al., 2021a,b). Published fMRI studies have shown that a large-scale brain network encodes motor learning and transfer of learning from related past experiences (Heitger et al., 2012; Gerraty et al., 2014). The prefrontal cortex has been found to integrate the information necessary for action generation and perception (Raos and Savaki, 2017) relevant to error processing. Here, successful skill acquisition leads to an internal forward model (Wolpert et al., 1998) that can simulate the perceptual consequences of planned and executed motor commands. An intact action-perception coupling has been shown to depend on the integrity of the cerebellum (Christensen et al., 2014) that underpins the internal model (Ebner, 2013) and error-based learning (Popa and Ebner, 2019). Error-based sensorimotor learning also involves other areas of the brain, including the parietal cortex, striatum, and anterior cingulate cortex (Seidler et al., 2013). Furthermore, the cingulate and pre-supplementary motor areas are the generator sites of error-related negativity that are time-locked to an erroneous response (Seidler et al., 2013). Here, the role of bilateral inferior frontal junction (IFJ) activation and their coordination in task-error perception action coupling, i.e., an interaction between top-down behavioral goals and bottom-up capture of relevant stimulus, needs further investigation since IFJ responds generally to the onset of behaviorally relevant stimuli (Levy and Wagner, 2011), e.g., task-error in the current study. Indeed, early efferent error prediction can lead to preemptive adjustments, e.g., skilled typists execute errors with lighter keystrokes than novices. Then, fMRI studies can show additional activation of deeper brain structures (Roberts et al., 2006) that cannot be measured using fNIRS. Here, it is known from skill training with fMRI studies that the hierarchy of cognitive control shows a rostrocaudal axis in the frontal lobe, where a shift from posterior to anterior is postulated to mediate progressively abstract higher-order control. In the current study, CER tDCS effect on FLS task related cortical activation included region 6vr+ anterior to the map of premotor area 6 (Thomas Yeo et al., 2011), i.e., the microstructural border between the motor and the cognitive domain (Geyer, 2004; Amunts et al., 2010). The region 6vr+ has greatest connectivity between the extent (as percentage) of the cerebral surface area and the cerebellar gray matter volume (Buckner et al., 2011), which needs future investigation based on fNIRS directed functional connectivity [specifically, subserved by ventral superior longitudinal fascicle–SLF III (Kamat et al., 2022)] vis-à-vis error-related perception action coupling. Also, tDCS effects on the neurovascular unit (Arora and Dutta, 2023) needs investigation where a POS-PRE decrease in the left prefrontal scalp EEG activity (within 1Hz−40Hz and 1 s−5 s at the start of the FLS task) was found to be related to an increase in the left prefrontal HbO cortical activation (from 5 s to 15 s) – see Figures 6F, 7A. Notably, the left prefrontal scalp ERSP changes were partly driven by the EEG frequencies below 8Hz [e.g., theta activity as mechanism of cognitive control (Cavanagh and Frank, 2014)] as seen in the Figures 6E, G.

The limitations of the current study methods include small subject size and portable fNIRS with limited spatial and depth sensitivity (Strangman et al., 2013) that could provide a partial view of brain activation. High-density EEG can be provide deeper localization (Michel and Brunet, 2019) and can be combined with fNIRS for better depth sensitivity. Therefore, the limitations include a low-density fNIRS and EEG sensor montage that limited spatial resolution and depth sensitivity necessary to capture the complete hierarchy of the task related brain areas. Another limitation was unequal gender balance (two males, 10 females) in the small number of participants with only one female subject with left hand dominance; however, removal of those subjects did not affect the trend of the brain imaging and FLS performance measure outcomes to tDCS interventions. In conclusion, our current brain-behavior study provided mechanistic insights into the right cerebellar tDCS facilitated activation of the bilateral prefrontal areas at the start of the FLS suturing with intracorporeal knot-tying task that was related to a significantly improved FLS performance score when compared to sham tDCS. In contrast, left ventrolateral prefrontal cortex tDCS failed to improve FLS performance score significantly when compared to sham tDCS although there was a trend toward improvement. Importantly, in the current study we did not study tDCS effects longitudinally during skill training, in correspondence with the skill retention study by Gao and colleagues (Gao et al., 2021a), which may be helpful in delineating CER tDCS vs. PFC tDCS effects in the future.

The data that support the findings of this study are available on request from the corresponding author.

The studies involving humans were approved by the Institutional Review Board of the University at Buffalo, USA. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

PW: data curation, investigation, experiments, and writing–draft preparation. YF: data curation, investigation, and experiments. JN, SS, and XI: supervision. SD and LC: supervision and writing—reviewing and editing. AD: study conceptualization, methodology, and writing—reviewing and editing. All authors contributed to the article and approved the submitted version.