Potential participants were recruited from a local university. Twenty young healthy participants (Group 1: age = 27.40 ± 2.07, two women and eight men; Group 2: age = 27.10 ± 2.08, two women and eight men) were recruited. All of them were postgraduate and met all of the following criteria: (1) 18 to 30 years old; (2) right-handed, according to the Edinburgh handedness inventory (Oldfield, 1971); and (3) normal or corrected-to-normal vision. Participants were excluded if they met any of the following criteria: (1) any contraindication to TBS, such as a history of seizures, metal implants, and pregnancy. All participants were screened by a standard safety checklist before enrollment (Rossi et al., 2011); (2) previous history of any neurological or psychiatric diseases; (3) presence of upper limb injuries in the past 3 months; and (4) presence of congenital upper limb deformities. This study was approved by the Human Subjects Ethics Committee, The Hong Kong Polytechnic University (reference number: HSEARS20190326003). All participants voluntarily consented to participate in this study and their written informed consent was obtained before participation commenced.

Participants were randomly allocated to one of the following two groups by drawing lots: Group 1: cTBS followed by iTBS; and Group 2: sham cTBS followed by iTBS. All participants had to attend four consecutive TBS sessions and two EEG assessments before and immediately after 4 daily sessions of stimulation.

The stimulation site for iTBS was the right primary motor cortex (M1). The optimal position was defined as the coil position eliciting the most stable and the largest MEP, with the coil rotated 45° from the sagittal plane. The stimulation position was maintained by a neuro-navigation system (Localite, Bonn, Germany). Resting motor threshold (RMT) is defined as the minimum intensity over the hot spot that elicits an MEP of no <50 μv in three out of six trials over the contralateral first dorsal interosseous (FDI) (Groppa et al., 2012). MEPs were visualized and measured through the MEP monitor (MagVenture, Denmark), with an inter-pulse interval of at least 5 s.

Daily serial sessions of iTBS were delivered by MagPro X100 stimulator (MagVenture, Denmark) with a standard butterfly-shape coil (C-B60), over the right M1 for 4 consecutive days. iTBS can induce the changes in corticomotor excitability and such effects are likely to be solidified over multiple-day stimulation (Wassermann and Zimmermann, 2012). Therefore, we decided to use repeated sessions of stimulations in order to stabilize the modulatory effects and also to imitate the intervention design commonly used in clinical applications (Perellon-Alfonso et al., 2018). We followed previous studies, using four daily sessions of iTBS for healthy adults (Hamzei et al., 2012; Lappchen et al., 2015; Zhang and Fong, 2019). The post EEG measurement was arranged on the same day after the 4th session of iTBS intervention.

The standard 600-pulse TBS protocol was followed (Huang et al., 2005). The stimulation intensity of iTBS was set at a safety limit of 70% of the individual's RMT (Goldsworthy et al., 2012). We did not set the intensity based on the active motor threshold (AMT), because a previous study has shown that pre-stimulation muscle contraction during the measure of AMT could influence the after-effects of TBS (Goldsworthy et al., 2012). Participants in the priming group (Group 1) received 600-pulse cTBS at the intensity of 70% RMT, followed by 600-pulse iTBS at the intensity of 70% RMT. Participants in the non-priming group (Group 2) received 600-pulse cTBS at the intensity of 20% RMT (i.e., sham cTBS), followed by 600-pulse iTBS at the intensity of 70% RMT. Sham stimulation was delivered using the same coil that delivers only 20% of the individual RMT. The reduction of intensity is a simple way for sham stimulation which has been used in previous studies (Dieler et al., 2014; Zhang and Fong, 2019). A previous neurophysiological experiment confirmed that no effect on MEPs can be observed when the stimulation intensity of TBS was decreased to ~62% (55–69%) of AMT. Therefore, we hypothesized that TBS at 20% of RMT could be served as a sham stimulation without causing significant modulatory effect to the stimulated cortex. All participants were told that TBS was a subthreshold stimulation that could not induce significant arm movements or somatosensory perception. The interval between priming and stimulation sessions was 10 min. We choose the 10 min interval based on a previous study about reversal of synaptic plasticity in response to TBS (Zhou et al., 2003), and followed a recent human neurophysiological study about priming iTBS (Opie et al., 2017). Participants were asked to complete a questionnaire regarding the side effects of iTBS they experienced upon the completion of each stimulation session. We assessed the treatment belief of each subject, upon the completion of the post EEG measurement. We asked the participants the question “did you believe that you have been applied brain stimulation in the past 4 sessions?” The treatment belief was assessed by a 10-point Likert scale, from fully disbelieve (rated as 0) to fully believe (rated as 10).

EEG was captured with a 64-channel cap, using a Digital DC EEG Amplifier and Curry 7 (Compumedics Neuroscan, USA). Electrode impedance was kept below 10 kOhm and the signal was sampled at 1,000 Hz. Participants were seated upright in an electromagnetic shielded room and required to minimize any body movements during the recording. Movement-related and MVF-induced sensorimotor ERD were evaluated. Left index finger tapping and incongruent (i.e., mirrored) visual observation of the right index finger tapping were used to elicit the ERD over bilateral sensorimotor areas, with a possibly right dominant lateralization (Pfurtscheller and Neuper, 1994; Zhang and Fong, 2019). For movement-related ERD, participants were instructed to tap on a computer keyboard three times with their left index finger, in response to 60 auditory cues (i.e., 300 ms beep sounds) delivered at random intervals (from 7 to 10 s) and to relax the finger after the completion of the movement (reference to finger tapping tasks). Participants were asked to focus on a centrally located fixation cross in a computer monitor placed in front of them. For MVF-induced ERD, participants were asked to tap on a computer keyboard three times with their right index finger, in response to 60 auditory cues delivered at random intervals (from 7 to 10 s), and to relax the finger after completing the movement (reference to finger tapping tasks). Movements were performed under two conditions. (1) A mirror view of the movement: Participants performed right-index tapping while simultaneously looking at the MVF of their moving finger. The MVF was created using a physical mirror (406 × 432 mm) placed over their midsagittal plane, between both arms. A black curtain was used to block the view of their moving hand. (2) A direct view of the movement: Participants performed right-index tapping while looking at the direct visual feedback (DVF) of their moving finger. Their left hand was hidden by a non-transparent board (see Figure 1 for the EEG set-up). The order of conditions was randomized by drawing lots. A total of 60 movements were collected for each condition, with 180 movements in total. To avoid the potential effects of ordering, we counterbalanced the order of these 3 conditions. The inter-trial interval was similar to previous studies about movement-related and MVF-induced ERD (Rossiter et al., 2015; Espenhahn et al., 2017; Fong et al., 2019), which allowed us to detect the ERD pattern elicited by movement execution or observation. The randomly given inter-trial interval was applied to avoid that the brain activity in association with participants' anticipation to auditory cues was synchronized with the presentation of the given stimulus.

Signals captured were processed offline using EEGLab (Delorme and Makeig, 2004) and custommade Matlab scripts. Raw EEG signals were band-pass filtered between 1 and 80 Hz and then down-sampled at 250 Hz. Additionally, a 50 Hz notch filter was applied. Data were referenced to bilateral mastoid electrodes. Signals with significant movement artifacts and long-term eye closure were rejected during the visual inspection. Then EEG was segmented into 5,000 ms epochs (pre-stimulus −2,000 ms and post-stimulus 3,000 ms, with 0 as the 1st index finger tap to the keyboard). Eye movement artifacts were corrected using an independent component analysis algorithm (Delorme and Makeig, 2004). Typical components reflecting blinking and horizontal eye movement were rejected.

Clean trials were analyzed in a time-frequency domain. The event-related spectral perturbation (ERSP) method was used to compute ERD power (Delorme and Makeig, 2004). For ERSP calculation, the power spectrum was calculated on each epoch and normalized each of them by its respective mean baseline spectra. We selected a baseline period from −1,500 to −1,000 ms for correction, to avoid the contamination of neural oscillations caused by auditory cues delivered prior to the execution of movement. Subsequently, the power was averaged across all included trials and converted to log power (see the following formula).

where n is the number of trials, and F_k(f, t) is the spectral estimation of the kth trial at frequency f and time t. ERD was further computed using the following formula (Makeig, 1993):

where F represents the frequency band of interest. We defined four frequency bands of interest, including mu-1 (8–10 Hz), mu-2 (10–12 Hz), beta-1 (12–16 Hz), and beta-2 (16–30 Hz) based on our previous studies (Fong et al., 2019; Zhang and Fong, 2019). T represents the time interval of interest and a window from 0 to 1,000 ms was selected to reflect the movement stage. N is the number of time-frequency bins in a selected two-dimension rectangular matrix. Following previous literature (Bartur et al., 2015; Lee et al., 2015; Zhang and Fong, 2019), we extracted averaged ERD powers at two electrodes C3 and C4 to represent the left and right sensorimotor activation, respectively. In accordance with previous studies, an asymmetric index (i.e., no hemispheric effect) was calculated with the following formula and used in further statistical analyses (Fong et al., 2019). A more positive value indicates more activation of the right sensorimotor area.

A statistical analysis was performed using SPSS version 23.0. GraphPad Prism version 7 and custom Matlab scripts were used for the figure visualization. Analysis of variance (ANOVA) was performed separately for each frequency band and the asymmetric index was used as the dependent variable. The level of significance was p < 0.05. Violation of sphericity was corrected by Green-Geisser. Potential between-group difference of the dependence variable at baseline was tested by independent t-tests. Two-way repeated measures analysis of variance (ANOVA) with time (pre vs. post) as a within-subject factor and group (Group 1 vs. Group 2) as a between-subject factor was used to analyze the movement-related ERD. Three-way repeated measures ANOVA with time (baseline vs. post-stimulation) and condition (mirror view vs. direct view) as within-subject factors and group (Group 1 vs. Group 2) as a between-subject factor was used to analyze the MVF-induced ERD. In case of any significant effect found, paired t-tests were used for the post hoc comparisons. If any of the dependent variable showed significant between-group difference at baseline, analysis of covariance (ANCOVA) with the baseline value as the covariance would be used instead. Missing data were imputed using a last observation carried forward (LOCF) method; that is, if a subject dropped out, the missing value was replaced by the last assessment results.

The current study used a small sample size and the Bayesian procedures may enhance the statistical information of our results (Quintela-del-Río et al., 2019). Thus, we included a Bayesian analysis to help highlight the relative strength of the evidence in support of either the null or alternative hypothesis (Biel and Friedrich, 2018) using JASP program (Wagenmakers et al., 2018). Bayesian repeated measures ANOVA and paired t-tests were performed. The Bayes Factors (BF) were reported, which evaluated the conditional probability between 2 competing hypotheses (null and alternative hypotheses) and quantify the support levels for each hypothesis. We reported the values as BF₁₀, with a value >1 indicating increased evidence in favor of the alternative hypothesis. The BF₁₀ values for the main and interaction effects in ANOVA were computed using the BF inclusion values output with the “across matched models” option (Koen et al., 2018). BFs were interpreted using the categorical labels, with BF₁₀ values between 1 and 3 correspond to anecdotal evidence, between 3 and 10 correspond to moderate evidence, 10 and 30 correspond to strong evidence, and > 100 correspond to decisive evidence.

The results of an ANOVA examining movement-related ERD are shown in Table 1 and the descriptive data are graphically depicted in Figure 4A. No baseline between-group difference was found in any of the frequency band (all p > 0.05). There were significant time effects noted in both the high mu [F_(1,18) = 6.52, p = 0.020, η² = 0.27, BF₁₀ = 3.01] and low beta bands [F_(1,18) = 6.00, p = 0.025, η² = 0.25, BF₁₀ = 3.54]. The Bayesian statistics showed anecdotal and moderate evidence in favor of the alterative hypothesis for the main effect of time. A significant interaction effect was noted in the high mu band [F_(1,18) = 4.47, p = 0.049, η² = 0.20, BF₁₀ = 1.96] and the mean changes (±standard deviation) of asymmetric index in the high mu band were 0.65 ± 0.47 in Group 1 and 0.06 ± 0.75 in Group 2, indicating that a more obvious shift in sensorimotor high mu ERD toward the right hemisphere was noted in participants who received priming iTBS, compared with those who received iTBS alone. The Bayesian statistics showed anecdotal evidence for the alternative hypothesis for the interaction. Figure 5 shows the topographical distributions of movement-related high mu ERD at baseline and post-stimulation. No other significant effects were found in the two-way ANOVA.

The results of the ANOVA examining MVF-induced ERD are shown in Table 2 and the descriptive data are graphically depicted in Figure 4B. No baseline between-group difference was found in any of the frequency band (all p > 0.05). Significant time effects were observed in both the high mu [F_(1,18) = 4.65, p = 0.045, η² = 0.21, BF₁₀ = 1.55] and low beta [F_(1,18) = 6.10, p = 0.024, η² = 0.25, BF₁₀ = 4.56] bands, and significant condition effects were observed in the low mu [F_(1,18) = 20.84, p < 0.001, η² = 0.54, BF₁₀ = 3556.80], high mu [F _(1,18) = 16.12, p = 0.001, η² = 0.47, BF₁₀ = 2603.50], and low beta bands [F_(1,18) = 11.72, p = 0.003, η² = 0.39, BF₁₀ = 127.33]. No other significant effects were found in the three-way ANOVA. Within-condition comparisons tested by paired t-tests showed that a dominant right-lateralized sensorimotor ERD was found in the low mu (baseline: t = 2.78, p = 0.012, BF₁₀ = 4.41; post-stimulation: t = 4.58, p < 0.001, BF₁₀ = 145.93), high mu (baseline: t = 2.47, p = 0.023 BF₁₀ = 2.58; post-stimulation: t = 4.35, p < 0.001, BF₁₀ = 91.95), and low beta bands (baseline: t = 2.31, p = 0.032, BF₁₀ = 1.978; post-stimulation: t = 3.40, p = 0.003, BF₁₀ = 14.02) under the MVF condition, in contrast to the DVF condition. Significant differences across time were only found in the high mu (t = 2.35, p = 0.030, BF₁₀= 2.09) and low beta bands (t = 2.79, p = 0.012, BF₁₀ = 4.47) under the MVF condition, but not in other frequency bands under the DVF condition.

Voluntary hand movements could attenuate the activities of mu and beta rhythms, as reported by several human EEG experiments carried out by Pfurtscheller and his colleagues (Pfurtscheller and Neuper, 1994; Stancak and Pfurtscheller, 1995; Pfurtscheller et al., 2000). The power suppression of mu and beta bands over the central electrodes induced by voluntary movement has been confirmed to be correlated with the activation of the sensorimotor area (Pfurtscheller and Lopes da Silva, 1999). iTBS is a facilitator of cortical excitability (Huang and Rothwell, 2004). A priming session of cTBS has been shown to intensify the facilitatory effect of subsequent iTBS on the motor cortex, as suggested by an increased MEP amplitude (Murakami et al., 2012; Opie et al., 2017) and a reduction of SICI (Murakami et al., 2012), in contrast to iTBS without priming. In the present study, we found that both priming iTBS and non-priming iTBS enhanced movement-related ERD only in high mu and low beta, but not in low mu and high beta bands, which supported our hypothesis that an inhibitory priming stimulation could intensify the facilitatory effects of subsequent iTBS on sensorimotor areas. Sensorimotor mu ERD has been found to be correlated with both MEP and SICI (Takemi et al., 2013, 2015; Thies et al., 2018); however, the functional dissociation between low mu and high mu rhythms has been reported by a previous study (Frenkel-Toledo et al., 2013). The authors found that high mu ERD, but not low mu ERD, had a clear response to movement execution. Indeed, some early studies have demonstrated that movement-related high mu ERD was found to be topographically restricted to the sensorimotor area, while low mu ERD was relatively topographically widespread (Pfurtscheller et al., 2000). Movement-related high mu ERD is likely to be a more sensitive marker of the activation of sensorimotor activation caused by voluntary movement than low mu ERD (Pfurtscheller and Lopes da Silva, 1999; Frenkel-Toledo et al., 2013). This could explain why we could only observe the facilitatory effect of two motor cortex stimulations on high mu but not low mu ERD. Sensorimotor beta ERD during movement is also thought to be correlated with voluntary movement and motor control. In the present study, low beta ERD was increased by both stimulation protocols, while high beta ERD remained stable at pre- and post-stimulation. The impairment of beta oscillation during movement has been found in previous studies on healthy older adults (Schmiedt-Fehr et al., 2016) and patients with motor impairments due to a stroke (Rossiter et al., 2014). These studies analyzed beta rhythms from 15 to 30 Hz, thus neglecting the low beta band. Our findings showed that high beta ERD was stable during movement in healthy adults with intact motor functions, after repetitive excitatory motor cortex stimulation, while low beta ERD varied along with the stimulation. Low beta ERD is more likely to be correlated with low-level motor control related to corticomotor excitability, while high beta ERD may be correlated with high-order motor functions, such as cognitive-motor control (Adam et al., 2015). In future, the effects of TBS on high beta oscillation and its relationship with the level of motor deficits and the ability of an individual to relearn motor skills warrant investigation in older populations and with patients with neurological conditions, such as stroke.

In this study, priming iTBS seems to be superior to iTBS in enhancing movement related ERD in the high mu band. Although the effect was only modest, we found that a priming session of cTBS did not abolish the excitatory effect of the subsequent iTBS session; it may even potentially boost the effects, found in high mu ERD. These findings support the potential role of metaplasticity in modulating the cortical response to excitatory motor cortex stimulation (Muller-Dahlhaus and Ziemann, 2015). However, the possibility of cortical inhibitory functions induced by the priming and non-priming iTBSs was not known, and further study may also explore the potential effect of priming protocol on cortical inhibitory functions, as measured by concurrent TMS-EEG.

Previous experiments have found that MVF induced a shift in ERD toward the sensorimotor area ipsilateral to the moving hand, compared with DVF, in healthy adults (Bartur et al., 2015; Lee et al., 2015; Zhang and Fong, 2019). In the present study, the effects of MVF were found in low mu, high mu, and low beta bands, which is in line with previous investigations conducted among healthy adults (Bartur et al., 2015; Lee et al., 2015; Zhang and Fong, 2019). Moreover, we found that both protocols enhanced the MVF-induced ERD in the high mu and low beta bands, indicating that both TBS protocols delivered to the motor cortex could make the brain more receptive to MVF, which provide neurophysiological evidence to explain the behavioral benefits from excitatory motor cortex stimulation on the observation-based motor learning (Hoff et al., 2015; von Rein et al., 2015; Zhang and Fong, 2019). However, we did not find any significant difference between the two TBS protocols. MVF-induced ERD is thought to be a summation of the activation of the sensorimotor cortex, presumed MNS, and other neural networks related to attention and cognitive control (Deconinck et al., 2015; Zhang et al., 2018). The magnitude of MVF-induced ERD may also be influence with the level of perception of embodiment during the observation, thus resulting in greater variability in response (Alimardani et al., 2016). The small between-group differences in the motor area may not be clearly reflected on the MVF-induced ERD. The facilitatory effect of priming protocol on MVF-induced oscillation may become observable when it is in combination of observation-based behavioral training (e.g., mirror training) (Zhang and Fong, 2019). Moreover, for populations with reduced responses to MVF and a limited ability to relearn motor skills via observation—patients who have suffered a stroke, for example (Bartur et al., 2018)—the alteration of this marker also needs to be examined and correlated with the potential functional recovery caused by priming iTBS.

It should be recognized that our experiment along with previous studies did not systematically investigate the potential variation in the effects of the priming TBS caused by the parameter difference, such as the delay interval between the priming stimulus and the conditioning stimulus, and the stimulation intensity. An interval of 10 min between cTBS and subsequent iTBS was used in our experiment and Opie et al. (2017); however, an interval of 15-min was applied in Murakami et al. (2012) and a 30 min interval was applied in Mastroeni et al. (2013). Most studies used an identical intensity for both priming and stimulation sessions (Hassanzahraee et al., 2018). However, one experiment showed that a priming cTBS session at a lower intensity (AMT = 70%) followed by a conditioning iTBS session at a higher intensity (AMT = 80%) could also induce the metaplastic effects (Murakami et al., 2012). The optimal selection of the parameters in the priming protocol is still unknown which needs to be further investigated.