The effect of transcranial alternating current stimulation (tACS) at alpha and beta frequency on motor learning

Behavioural Brain Research 293 (2015) 234–240

Contents lists available at ScienceDirect

Behavioural Brain Research journal homepage: www.elsevier.com/locate/bbr

Research report

The effect of transcranial alternating current stimulation (tACS) at alpha and beta frequency on motor learning Bettina Pollok ∗ , Ann-Christin Boysen, Vanessa Krause Heinrich-Heine-University Duesseldorf, Medical Faculty, Institute of Clinical Neuroscience and Medical Psychology, Universitaetsstr. 1, 40225 Duesseldorf, Germany

h i g h l i g h t s • • • • •

Motor leaning is associated with changes of primary motor cortex (M1) oscillations. TACS (10 Hz, 20 Hz, sham, 35 Hz) was applied to left M1 during SRTT learning. 10 Hz and 20 Hz tACS likewise facilitated motor sequence learning compared to sham and 35 Hz tACS. 20 Hz tACS yielded less susceptibility to interference by a random pattern. Motor-cortical beta oscillations may stabilize motor control.

a r t i c l e

i n f o

Article history: Received 7 April 2015 Received in revised form 21 July 2015 Accepted 24 July 2015 Available online 28 July 2015 Keywords: Neuroplasticity Neuromodulation Oscillatory activity Primary motor cortex (M1) Serial reaction time task (SRTT) Interference

a b s t r a c t At present it remains elusive to what extent motor-cortical alpha (8–12 Hz) and beta (13–30 Hz) oscillations are associated with motor sequence learning. In order to interact with motor-cortical oscillations, the present study applied transcranial alternating current stimulation (tACS) at 10 Hz, 20 Hz and sham stimulation over the left primary motor cortex (M1) during a serial reaction time task (SRTT) in 13 healthy volunteers. In a control experiment, tACS at 35 Hz was applied in another sample of 13 volunteers. The participants performed the task with the right hand. A sequential pattern was interleaved by a randomly varying pattern serving as interference from sequence learning. Reaction times were determined as dependent variable. Both 10 and 20 Hz tACS facilitated SRTT acquisition in contrast to sham and 35 Hz tACS. After acquisition, the interfering condition led to increased reaction times comparable to baseline level during 10 Hz, sham and 35 Hz tACS. In contrast, during 20 Hz tACS the initial learning success was retained despite interference. While motor-cortical tACS at 10 and 20 Hz likewise facilitates the acquisition, tACS at 20 Hz frequency additionally stabilizes the newly learned motor sequence indicated by less susceptibility to interference. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Motor learning is essential for the interaction with our environment and adjustments to changes. Initial learning is characterized by quick skill acquisition. After acquisition the skill is consolidated and stabilized being transferred to motor memory. Finally, the skill has reached an automatic level during subsequent retrieval [1,2]. Neuroimaging studies point out that training-related neuroplastic changes already occur very early during motor skill acquisition [1]. These changes appear to take place within the motor net-

∗ Corresponding author. Fax: +49 211 81 13015. E-mail address: [email protected] (B. Pollok). http://dx.doi.org/10.1016/j.bbr.2015.07.049 0166-4328/© 2015 Elsevier B.V. All rights reserved.

work consisting of the primary motor cortex (M1) and functionally connected cortical and subcortical regions [3–7]. Synchronized oscillatory activity at alpha (8–12 Hz) and beta (13–30 Hz) frequencies within and between these brain regions is assumed to facilitate neuronal plasticity thereby promoting motor learning [8,9]. In addition, alpha and beta frequencies may mediate different motor control modes [10]. While automatic motor control is likely to be reflected by synchronized alpha oscillations, complex motor control rather is associated with oscillatory synchronization at beta frequency [11,12]. In addition, altered beta oscillations have recently been associated with motor sequence learning [9]. At present, it has fairly been investigated whether and to what extent implicit motor sequence learning can be affected by non-invasive modulation of alpha and beta oscillatory brain activity.

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

To study motor sequence learning, the serial reaction time task (SRTT) is a well-established paradigm eliciting learning over the course of time [13]. SRTT learning has been shown to be modulated by non-invasive electrical brain stimulation such as transcranial direct current stimulation (tDCS) [14–16]. By application of a weak electrical constant current, tDCS is suitable to modulate cortical excitability in a polarity-dependent way via modulation of the neuronal resting membrane potential and/or synaptic activity [17]. In comparison to tDCS, transcranial alternating current stimulation (tACS) applies a sinusoidal waveform at specific frequencies alternating between the anode and cathode. Although the exact underlying mechanisms are not well understood, tACS is hypothesized to alter the power of oscillatory brain rhythms by (de) synchronizing neuronal networks in a frequency-dependent manner [18,19]. TACS may entrain endogenous brain oscillations during stimulation [20–23] and may also be associated with neuroplastic changes outlasting stimulation cessation [20,24–26]. Both tACS and tDCS are suitable to investigate the neurophysiological underpinnings of cognitive and motor processes as well as their potential implementation in neuro(psycho) logical and psychiatric therapy [27,28]. Antal et al. applied tACS at 10 Hz during SRTT acquisition and observed an improvement of reaction times [29]. Since tACS at other frequencies was not shown to improve reaction times in that study, one may hypothesize a frequency-specific effect on SRTT acquisition. Assuming that tACS interacts with oscillatory brain activity, the present study elucidates the effect of motor-cortical alpha band tACS at 10 Hz and beta band tACS at 20 Hz on the acquisition of a new motor sequence. To this end, 10 Hz and 20 Hz tACS were applied over the left M1 during SRTT acquisition with the right hand. In order to control for placebo and frequency-unspecific effects, sham tACS and 35 Hz tACS were applied. 2. Material and methods 2.1. Participants The study was accomplished by a sham-controlled, double-blind design. General exclusion criteria were history or family history of epileptic seizures, history of migraine, unexplained loss of consciousness, or brain related injury, history of other neurological or psychiatric disorders, pregnancy, intake of central nervous systemeffective medication, cardiac or brain pacemaker, or other metal implants that could not be removed for the experiment. All participants were naïve regarding the exact purpose of the study and most participants had never received transcranial electrical stimulation before. The main investigator was blinded regarding the type of stimulation in each session. A second investigator who was not involved in the interaction with participants was responsible for stimulation only. The main investigator was informed about each session’s type of stimulation after raw data analysis. Participants provided written informed consent prior to participation. The study was accomplished with the approval of the local ethics committee and is in accordance with the Declaration of Helsinki. 2.1.1. Main experiment – 10 Hz vs. 20 Hz vs. sham tACS TACS was applied at 10 Hz vs. 20 Hz vs. sham in three sessions counterbalanced across participants. Sessions were separated by at least one week. A priori sample size calculation revealed that n = 12 participants were needed to achieve effect sizes of .80 with a repeated-measures design and a significance level of ˛ < .05. 14 healthy participants were included in this design to compensate for potential dropouts. Since one participant was unavailable after the first session and excluded from data analysis, data from 13 par-

235

ticipants (6 male, 7 female) with an average age of 22.08 (±.71; mean ± standard error of the mean, SEM) years were considered for the final analysis. All participants were classified as right-handed (87.69 ± 4.82) by means of the Edinburgh Handedness Inventory (EHI) [30] and had normal or corrected to normal sight. For righthandedness a minimum EHI score of 40 was required. 2.1.2. Control experiment – 35 Hz Another group of 13 healthy, right-handed participants (6 male, 7 female; age: 23.23 ± 1.04; EHI: 87.31 ± 3.70) received tACS at 35 Hz in order to control for frequency-unspecific stimulation effects. Age and handedness were comparable between participants of the main and control experiment (independent samples t-test: age: t(23) = −.89, p = .39; EHI: t(24) = .06, p = .95). 2.2. Apparatus and materials 2.2.1. Transcranial alternating current stimulation (tACS) In a first step, the hand region of the left M1 was localized by single transcranial magnetic stimulation (TMS) pulses using a standard figure of eight coil (MC-B70) connected to a MagPro stimulator (Medtronic, Minneapolis, USA). The coil was placed tangentially to the scalp with the handle pointing backwards and laterally at about 45◦ away from the midline inducing an initial posterior-anterior current flow in the brain. We localized the optimal cortical representation of the right hand first dorsal interosseus (FDI) muscle by eliciting motor-evoked potentials (MEP). By moving the coil in 0.5 cm steps anterior, posterior, medial, and lateral to this area, the exact localization of the spot which elicited the maximal FDI motor response was determined as motor hot spot. TACS was applied via two saline-soaked sponge electrodes (7 cm × 5 cm) on the skin surface (DC-Stimulator Plus, Eldith, NeuroConn, Ilmenau, Germany) with an intensity of 1 mA (peak-to-peak-amplitude; sinusoidal waveform) corresponding to 0.0286 mA/cm2 current density under the electrode. The electrodes were placed above the left M1 hot spot and right orbita. Impedance was kept below 5 kOhm. Stimulation parameters were applied with respect to current safety guidelines of tDCS [31]. During active stimulation sessions, tACS duration was individually adjusted to SRTT duration lasting 12 min 12 s (±4.7 s) on average. During sham stimulation, active tACS was applied only at the beginning of SRTT performance within the first 30 s in order to elicit a short tingling and flickering sensation usually perceived at stimulation onset. The stimulator switched off automatically. Sham tACS was applied for 30 s with either 10 Hz or 20 Hz tACS in half of participants, respectively, in a counterbalanced order. Participants and investigator were blinded with respect to stimulation (10 Hz vs. 20 Hz vs. sham). Since tACS at frequencies below 40 Hz is likely accompanied by cutaneous and visual flicker perception due to retinal stimulation [32,33], at the end of each session participants filled in a stimulation questionnaire. In the first step, participants were to rate the type of stimulation by indicating whether they received active or sham stimulation. If active stimulation was estimated, they needed to indicate whether stimulation was carried out with 10 Hz, 20 Hz, 30 Hz (main experiment) or 10 Hz, 20 Hz, 35 Hz (control experiment). In addition, participants rated their subjective confidence in each of the two questions. A numerical rating scale from 1 (totally uncertain) to 10 (totally certain) was used. 10 Hz and 20 Hz stimulation was correctly identified in 31% of active stimulation sessions with a mean confidence rating of 4.25 (±0.62). 35 Hz tACS was not correctly identified in any session. Sham stimulation was correctly depicted in 31% of sham sessions with a mean subjective confidence of 7.25 (±0.95). Participants were also asked to indicate any unusual perception during or immediately after stimulation (e.g. itching, tingling, flickering). In 4/13 sessions with 10 Hz, 6/13 with 20 Hz,

236

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

2.3. Experimental design

Fig. 1. (A) Schematic illustration of the sequential pattern within the serial reaction time task (SRTT) and corresponding button presses on the response button box: The eight-digit SRTT was presented in four target locations on a projection screen. Participants were required to react as fast and as accurately as possible when indicated by color change at one target location at a time. Reaction times were measured by button presses with the first four fingers of the right hand using a right-hand response button box (e.g. sequential pattern 4-1-3-2-1-4-3-2). B) The SRTT was divided into blocks of sequential (S) interleaved by random (R) stimulus presentations. In each block (R1 –R4 and S1 –S8 ) the respective sequence was repeated five times. Statistical analysis was focused on the learning index at four time points t1 = R1 − S1 , t2 = R2 − S6 , t3 = R3 − S7 and t4 = R4 − S8 . Active tACS was individually adjusted to SRTT duration.

and 1/13 with 35 Hz tACS participants perceived visual flickering at any time during stimulation. 2.2.2. Motor learning – serial reaction time task (SRTT) The SRTT is a well-established experimental paradigm eliciting motor sequence learning over time. It involves choice-reactiontime trials of stimuli presented in several target locations either at a sequential or a random series. The SRTT was presented as a 4digit keyboard (Fig. 1A). Stimuli were four dark blue keys arranged horizontally along the middle of a screen in front of the participants against a black background at a distance of 2.66 m and with a visual angle of 12.87◦ . The task was introduced to the participants as a simple reaction time task. Participants were not given hints about the sequential order of the stimuli. They were instructed to react as fast and as accurately as possible by pressing the corresponding button of a response box when one key on the screen changed its color from dark blue to light blue. The target key remained light blue until the correct button was pressed. The next target key changed its color after a 1 s delay. We applied three parallel versions of an eightdigit SRTT requiring participants to react with the first four fingers (thumb (1), index finger (2), middle finger (3), ring finger (4)) of the right hand using a custom-made response button box placed on the participants’ right side and connected to a standard Windows PC (Fig. 1A). Parallelism of SRTT versions was confirmed in a pilot study prior to the present experiment. SRTT presentation and recording of reaction times was carried out using E-Prime® (Psychology Software Tools Inc.). In order to elucidate whether participants learned the sequence explicitly, participants were verbally asked at each session’s end whether they recognized anything during the task. Two participants reported to have detected a sequential pattern eventually during one session. We did not ask these participants to verbally recall the detected pattern in order to keep them as naïve as possible to the hidden SRTT sequence until all sessions were completed. The data imply that motor sequence learning may have potentially become explicit in those participants who reported to have detected a sequential pattern. No further information was collected when participants did not report anything at the session’s end. In those participants the type of learning (implicit vs. explicit) remained open.

Participants were stimulated while performing the task (Fig. 1B). In order to control for carry-over effects, sessions were separated by at least one week, and participants performed three parallel SRTT versions. Order of versions was counterbalanced across participants and sessions. The SRTT started with a random pattern (2 blocks (R1 + R2 ) à 5 randomly varying repetitions à 8 button presses) followed by the sequential pattern eliciting motor sequence learning (6 blocks (S1 –S6 ) à 5 sequential repetitions à 8 button presses). After sequence acquisition, a randomly varying pattern was again presented (R3 + R4 ). This design implements interference of the random pattern with subsequent retrieval of the sequence (1 block S7 à 5 sequential repetitions à 8 button presses) serving as index of stabilization or maintenance of training-related improvements after interference [2,34]. Sequence reacquisition was determined as reaction time improvement (1 block S8 à 5 sequential repetitions à 8 button presses). Within each block reaction times were averaged. 2.4. Data analysis Reaction time was measured by the onset of each button press and logged in EPrime® data files. The first sequence of the trial was discarded from data analysis to allow participants for adaptation with the button box. Reaction times of the sequential pattern were subtracted from the random pattern (Ri − Si ) in order to normalize the individual performance over time representing an index for motor learning. The analysis was focused on four time points of motor sequence acquisition: t1 = R1 − S1 (baseline), t2 = R2 − S6 (end of initial acquisition), t3 = R3 − S7 (immediately after interference), and t4 = R4 − S8 (subsequent sequence repetition after interference). Data were checked for normal distribution by Kolmogorov–Smirnov test. All data were shown to be equal to Gaussian distribution apart from baseline during sham stimulation (p = .02, not corrected). Analysis of variance (ANOVA) with within-subject-factors stimulation (10 Hz vs. 20 Hz vs. sham) and time (t1 vs. t2 vs. t3 vs. t4) was calculated. In a first step, ANOVA for the random condition was performed to rule out an impact of random reaction times on the learning index. In the second step, we performed ANOVAs for the learning index. Post-hoc analyses were performed by separate univariate ANOVAs and paired t-tests. P-values were corrected for multiple testing with the sequential Bonferroni procedure [35]. All statistical comparisons were calculated with IBM SPSS Statistics 22. 3. Results 3.1. Main experiment – 10 Hz vs. 20 Hz vs. sham tACS Reaction times during the random condition were controlled for differences between stimulation conditions. Neither the factor stimulation (F(2,20) = .69, p = .39) nor the stimulation × time interaction (F(6,60) = 1.30, p = .27) were found to be significant ruling out an effect of changes in the random condition on the learning index. ANOVA of the learning index revealed a significant main effect of factor time (F(3,36) = 17.55, p < .001) and a significant stimulation × time interaction (F(6,72) = 2.24, p = .049) but no significant main effect of factor stimulation (F(2,24) = 2.70, p = .088). In order to further analyze the stimulation × time interaction, we performed separate ANOVAs for each of the four time points, respectively with factor stimulation (10 Hz vs. 20 Hz vs. sham). At t1 (F(2,24) = .87, p = .432) and t4 (F(2,24) = .20, p = .820), no significant differences between stimulation conditions were found. But, stimulation-specific differences were observed at t2 (F(2,24) = 5.56, p = .010) and t3 (F(2,24) = 4.03, p = .031). At t2, participants learned

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

237

Comparison between t2 and t3 revealed that after interference reaction times remained on the same level during 20 Hz tACS (t(12) = .96, p = .716), while they significantly slowed down during 10 Hz tACS (t(12) = 4.94, p < .001) – indicating less susceptibility to interference following 20 Hz tACS. No significant changes of the learning index during sham stimulation were found (t(12) = .04, p = .967). Table 1 summarizes paired t-tests regarding the effects over time within each stimulation condition. Mean and individual stimulation effects on the learning index are illustrated in Fig. 2A and B.

3.3. Control experiment – 35 Hz Separate ANOVAs with within-subject-factor time (t1 vs. t2 vs. t3 vs. t4) and between-subject-factor stimulation (10 Hz vs. 35 Hz; 20 Hz vs. 35 Hz; sham vs. 35 Hz) were performed in order to determine whether significant differences between 35 Hz and stimulation conditions from the main experiment occur. The comparison between 20 Hz and 35 Hz tACS revealed a significant main effect of time (F(3,72) = 6.37, p = .001) and a significant stimulation × time interaction (F(3,72) = 4.68, p = .005), but no significant main effect of stimulation (F(1,24) = 2.93, p = .100). The significant interaction is due to differential stimulation effects at t2 and t3: In contrast to 35 Hz tACS (t1 vs. t2 t(12) = −1.55, p = .292), 20 Hz tACS facilitated initial SRTT acquisition (t(12) = − 3.42, p = .02). The acquisition level was retained at t3 during 20 Hz tACS (t2 vs. t3 t(12) = .96, p = .716) still superior to baseline level (t1 vs. t3 t(12) = −2.91, p = .039) whereas 35 Hz tACS resulted in a significant decline (t2 vs. t3 t(12) = 3.55, p = .02) comparable to baseline level (t1 vs. t3 t(12) = 2.60, p = .092). Comparing 10 Hz and 35 Hz tACS a significant main effect of time (F(3,72) = 13.07, p < .001) was observed but no significant main effect of stimulation (F(1,24) = 1.63, p = .215). Although the stimulation × time interaction marginally exceeded the .05 significance level (F(3,72) = 2.65, p = .055) we performed post-hoc t-tests showing that in contrast to 35 Hz tACS, SRTT acquisition was facilitated during 10 Hz tACS (t1 vs. t2 t(12) = −4.54, p = .004). Interference was evident at t3 with performance subsiding back to baseline level during both 10 Hz (t2 vs. t3 t(12) = 4.94, p < .001; t1 vs. t3 t(12) = −.39, p = .707) and 35 Hz tACS (t2 vs. t3 t(12) = 3.55, p = .02; t1 vs. t3 t(12) = 2.60, p = .092). ANOVA for sham and 35 Hz tACS – again – revealed a significant main effect of time (F(3,72) = 4.42, p = .007) and a significant stimulation × time interaction (F(3,72) = 5.27, p = .002), but no significant main effect of stimulation (F(1,24) = .14, p = .710). Comparable to 35 Hz tACS, SRTT acquisition did not significantly benefit from sham stimulation (t(12) = −2.10, p = .171). While 35 Hz tACS was associated with a significant decline of reaction times from t2 to t3 (t(12) = 3.55, p = .02) back to baseline level (t(12) = 2.60, p = .092), reaction times during sham stimulation remained on baseline level (t1 vs. t3 t(12) = −1.38, p = .386). The comparison between the stimulation conditions revealed a significant differential effect of 20 Hz and 35 Hz tACS at t3 (independent samples t-test t(24) = 5.05, p < .001) – indicating less

Fig. 2. (A) Changes of the learning index over time (t1–t4). Positive values indicate faster reaction times during presentation of the sequential as compared to the random pattern. Participants received 10 Hz vs. 20 Hz vs. sham tACS or 35 Hz tACS during SRTT. 10 Hz and 20 Hz tACS facilitated the initial acquisition of the motor sequence from t1 to t2, while sham and 35 Hz tACS yielded no significant effect. At t3, 20 Hz tACS was associated with no significant interference by a preceding random pattern whereas performance significantly declined during 10 Hz and 35 Hz tACS subsiding back to baseline level. Shown are mean values. Error bars indicate standard error of the mean (SEM). Asterisk indicates p < .05. B) Individual data: Learning index for t1 to t4 in individual participants during 10 Hz, 20 Hz, 35 Hz and sham tACS.

to a significantly greater extent when receiving active as compared to sham stimulation (10 Hz vs. sham: t(12) = 2.97, p = .036; 20 Hz vs. sham: t(12) = 2.85, p = .03). 10 Hz and 20 Hz effects did not differ significantly from each other (t(12) = .59, p = .567) – suggesting a facilitating effect of both 10 Hz and 20 Hz tACS on sequence acquisition. At t3, reaction times during SRTT were significantly faster during 20 Hz tACS as compared to sham (t(12) = 3.12, p = .027; Fig. 2A) whereas they did not differ significantly between 10 Hz and sham (t(12) = .28, p = .787).

Table 1 Paired samples t-tests for the learning index for t1–t4. P-values were corrected for multiple testing by the sequential Bonferroni procedure (Holm, 1979). Significant p-values are highlighted. 10 Hz

t1–t2 t2–t3 t3–t4 t1–t3 t1–t4

20 Hz

Sham

35 Hz

t

p

t

p

t

p

t

p

t(12) = −4.54 t(12) = 4.94 t(12) = −3.15 t(12) = −.39 t(12) = −3.25

.004 .001 .016 .707 .021

t(12) = −3.42 t(12) = .96 t(12) = −.56 t(12) = −2.91 t(12) = −3.71

.020 .716 .584 .039 .015

t(12) = −2.10 t(12) = .04 t(12) = −2.51 t(12) = −1.38 t(12) = −2.87

.171 .967 .112 .386 .070

t(12) = −1.55 t(12) = 3.55 t(12) = −2.25 t(12) = 2.60 t(12) = .47

.292 .020 .132 .092 .650

238

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

susceptibility to interference during 20 Hz compared to 35 Hz tACS. Performance was comparable at each time point when comparing 35 Hz with sham (p > .126) and 10 Hz tACS (p > .18). 4. Discussion The present study investigates the effect of tACS at 10 Hz and 20 Hz on the acquisition of a motor sequence. While sham tACS served as control condition for unspecific stimulation effects, 35 Hz tACS was applied in order to determine the specificity of stimulation frequency. TACS was applied while participants reacted as fast as possible to a sequential pattern interleaved by a random pattern of button presses serving as interference from sequence learning. The data suggest a facilitating effect of 10 Hz and 20 Hz tACS on sequence acquisition while following sham and 35 Hz stimulation no significant learning occurred. Stabilization of the newly learned motor sequence at t3 as determined by less interference by the random pattern was favoured by 20 Hz tACS only. 4.1. Facilitated initial acquisition by 10 Hz and 20 Hz tACS The data suggest a beneficial effect of tACS at 10 Hz and 20 Hz. Since 35 Hz tACS did not result in this facilitation from t1 to t2, the present data imply a specific modulation of motor sequence acquisition by alpha and beta frequencies. But, we would like to stress that descriptively learning occurred even during 35 Hz tACS and that the non-significant effect might at least partly be explained by slightly faster baseline performance during 35 Hz tACS. But, no significant difference was found at baseline between stimulation conditions. Thus, we here would not argue that 35 Hz stimulation does not at all favour motor sequence acquisition, but that this effect is weaker than that following 10 Hz and 20 Hz tACS. Behavioral tACS effects have been related to entrainment of ongoing oscillatory brain activity by stimulation frequency [19]. Such interaction between externally applied stimulation and brain oscillations has been shown to manipulate brain functions by promoting neural activity to resonate at stimulation frequency. The effects may vary with the amount of oscillatory activity prior to stimulation For instance, alpha band tACS over the occipital cortex was most effective in darkness when alpha oscillations are prevailing, while beta band tACS was most effective during light when alpha are switching to beta oscillations [36]. All in all, it would be reasonable to assume that behavioral effects found here are due to such entrainment. The present results partly replicate the findings of Antal et al. who showed a facilitation of implicit motor sequence learning during alpha band tACS at 10 Hz [29]. Interestingly enough, in that study stimulation at 15 Hz and 30 Hz yielded no significant effects on motor performance. Since the efficacy and direction of tACS effects strongly depend on the endogenous oscillatory state [37], the concordance between applied stimulation frequency and individual endogenous oscillatory activity may contribute to discrepant effects between studies. We here would argue that 15 Hz probably does not match sensorimotor beta oscillations and therefore failed to show an effect on motor learning. This interpretation agrees with the notion that behavioral effects measured online i.e. during tACS are likely to rely on the individual brain state e.g. frequency- and site-specific patterns of oscillatory activity [38] and go along with an entrainment of prevailing oscillations [38]; reviewed in in vitro and in vivo animal studies [18]. Moreover, methodological differences might have contributed to this possible discrepancy: Antal et al. delivered tACS for five minutes with a 16 cm2 stimulation electrode and 0.4 mA intensity (current density 0.025 mA/cm2 ) and used a 12-digit SRTT possibly demanding higher and/or distinct cognitive control processes

than the present eight-digit SRTT [29]. All in all, the present study supports the hypothesis that tACS at alpha and beta frequencies facilitates sequence learning. Since this effect was not anymore evident at t4 it might be limited to initial acquisition and/or is rather short-lived. Applying tACS over several minutes results in changes of excitability of brain areas being stimulated as well as functional interaction of these areas [19]. While alpha band tACS applied over the occipital cortex is associated with inhibition [23], tACS applied at beta frequency over M1 is likely to increase excitability [39,40]. Since we did not monitor alterations of oscillatory activity or cortical excitability during stimulation, the present results do not allow a conclusion about the neurophysiological mechanisms yielding the behavioral effects found here. One may speculate that the beneficial effect of tACS at 10 and 20 Hz on initial acquisition is not likely to be explained by frequency-specific entrainment. Rather, these effects might be due to increased excitability as suggested by tDCS studies [2,15]. Increased M1 excitability by anodal tDCS improved initial acquisition when applied during acquisition of the SRTT [14,15,41] and also early consolidation when applied immediately after learning [42], whereas reducing cortical excitability by cathodal tDCS during learning impaired acquisition [15]. In the present study, tACS at alpha and beta frequency improved reaction times in line with anodal tDCS leading to the speculation that the behavioral effects may occur due to increased motor-cortical excitability promoting synaptic plasticity [20,26]. Interestingly enough, both frequencies yield comparable effects on SRTT acquisition. 4.2. Less interference by 20 Hz tACS Stabilization of a newly learned motor skill between successive training sessions and less interference are characteristic indices for successful consolidation [2,34]. One of our recent magnetoencephalography (MEG) studies linked early consolidation within ten minutes after initial motor sequence acquisition to changes of motor-cortical beta oscillatory activity [9]. In the present study, 20 Hz tACS favoured stabilization already immediately after acquisition as indicated by less susceptibility to an interfering random pattern suggesting that boosting beta oscillations promotes functional reorganization associated with motor sequence learning. In support of this conclusion, Engel and Fries postulated that motorcortical beta oscillations may represent the maintenance of the current motor and cognitive states [43]. While motor-cortical alpha oscillations most likely reflect reduced attentional demands after initial sequence learning, beta oscillations might indicate functional reorganization associated with motor sequence learning and early consolidation [9]. Taken together, tACS at beta frequency may interact with motor-cortical beta oscillations reflecting functional reorganization over the time course of motor learning [8]. Accordingly, it has been shown that rhythmic brain stimulation applied over several minutes likely induces long-term neuroplastic changes [23] possibly representing the neurophysiological basis of motor stabilization of a newly learned sequence as shown here. Alternatively, 20 Hz tACS may have also affected directly the processing of the visuomotor association between the visual target and respective button press. In accordance, electroencephalography recordings from a force-tracking task have shown that functional interaction within long-range visuomotor neuronal networks is reflected by synchronized oscillatory activity at beta frequency [44]. Since t3 implied a switch from the randomly varying back to the previously learned sequential pattern, one might speculate that less interference observed at t3 might be due to reduced task-switching costs associated with 20 Hz tACS. Leite et al. previously showed reduced task-switching costs in a cognitive task following anodal tDCS of M1 and dorsolateral prefrontal cortex, while cathodal tDCS

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

increased task-switching costs in a motor task [45]. Although we cannot exclude that the present effect at t3 may partly be due to changes of cortical excitability, one might wonder why this effect was induced solely by 20 Hz tACS. Assuming that the effects found at t2 may be explained by non-specific changes of cortical excitability, we would expect that 10 Hz tACS and 35 Hz tACS would yield comparable effects. Thus, we here would argue that the effect of 20 Hz tACS at t3 might reflect current motor stabilization [43]. 4.3. Limitations Since this study focused on motor sequence learning, it remains elusive to what extent the present finding can be transferred to motor learning in general and its possible application for neurorehabilitation. We point out that the core network of SRTT learning has been found to involve primary motor and sensory cortices (M1/S1), dorsolateral premotor cortex (PMC), supplementary motor area (SMA), superior parietal cortex, thalamus, putamen, and cerebellum [6]. Stimulation electrodes were placed above the left M1 and right orbita with equal size of 7 × 5 cm2 . We need to acknowledge a possible contribution of overlapping left PMC stimulation to the present effects. But, PMC rather seems to be involved in offline improvements occurring several hours after learning [46,47]. Those data support the relevance of PMC for late consolidation of a motor sequence rather than initial acquisition or early consolidation. Moreover, we cannot rule out the possibility that the right prefrontal cortex may have been stimulated due to the electrode montage applied here. Modeling of the electrical current flow within the brain has revealed that its maximum might not be directly underneath the electrodes but spread toward frontal cortices [48]. Although the prefrontal cortex is proposed to be involved in SRTT learning [49], this presumption is weakened since it is rather dedicated to explicit learning of visuomotor associations than implicit learning [1]. Importantly, Nitsche et al. previously showed a beneficial impact of anodal M1 tDCS on SRTT acquisition while tDCS of premotor and prefrontal cortices had no significant effect [14]. As we did not monitor oscillatory activity or excitability changes during stimulation, the present data do not allow a conclusion about the underlying neurophysiological processes but provide evidence that frequency-specific tACS is associated with differential effects on motor behavior. 5. Conclusion While in contrast to 35 Hz tACS motor-cortical 10 and 20 Hz tACS boost the acquisition of a motor sequence, 20 Hz tACS additionally results in reduced susceptibility to interference. These data support the hypothesis that M1 beta oscillations may stabilize motor control. Conflict of interest The authors declare no competing financial, personal or otherwise related interests which could have influenced the present work. Funding This work was supported by the German Research Foundation (Deutsche Forschungsgemeinschaft (DFG); PO806/3-1 to BP). The funding source was not involved in performance of the research and preparation of the manuscript.

239

Acknowledgement We would like to thank Dr. Markus Butz for his valuable expertise with tACS application and Mitjan Morr for assistance with raw data analysis.

References [1] U. Halsband, R.K. Lange, Motor learning in man: a review of functional and clinical studies, J. Physiol. Paris 99 (2006) 414–424. [2] J. Reis, E. Robertson, J.W. Krakauer, J. Rothwell, L. Marshall, C. Gerloff, E. Wassermann, A. Pascual-Leone, F. Hummel, P.A. Celnik, et al., Consensus: can tDCS and TMS enhance motor learning and memory formation? Brain Stimul. 1 (2008) 363–369. [3] W. Muellbacher, U. Ziemann, J. Wissel, N. Dang, M. Kofler, S. Facchini, B. Boroojerdi, W. Poewe, M. Hallett, Early consolidation in human primary motor cortex, Nature 415 (2002) 640–644. [4] W. Muellbacher, U. Ziemann, B. Boroojerdi, L. Cohen, M. Hallett, Role of the human motor cortex in rapid motor learning, Exp. Brain Res. 136 (2001) 431–438. [5] J. Doyon, H. Benali, Reorganization and plasticity in the adult brain during learning of motor skills, Curr. Opin. Neurobiol. 15 (2005) 161–167. [6] R.M. Hardwick, C. Rottschy, R.C. Miall, S.B. Eickhoff, A quantitative meta-analysis and review of motor learning in the human brain, Neuroimage 67 (2013) 283–297. [7] J.W. Krakauer, R. Shadmehr, Consolidation of motor memory, Trends Neurosci. 29 (2006) 58–64. [8] T.W. Boonstra, A. Daffertshofer, M. Breakspear, P.J. Beek, Multivariate time-frequency analysis of electromagnetic brain activity during bimanual motor learning, Neuroimage 36 (2007) 370–377. [9] B. Pollok, D. Latz, V. Krause, M. Butz, A. Schnitzler, Changes of motor-cortical oscillations associated with motor learning, Neuroscience 275C (2014) 47–53. [10] P. Fries, A mechanism for cognitive dynamics: neuronal communication through neuronal coherence, Trends Cogn. Sci. 9 (2005) 474–480. [11] B. Pollok, V. Krause, M. Butz, A. Schnitzler, Modality specific functional interaction in sensorimotor synchronization, Hum. Brain Mapp. 30 (2009) 1783–1790. [12] N.J. Davis, S.P. Tomlinson, H.M. Morgan, The role of beta-frequency neural oscillations in motor control, J. Neurosci. 32 (2012) 403–404. [13] M. Nissen, P. Bullemer, Attentional requirements of learning: evidence from performance measures, Cogn. Psychol. 19 (1987) 1–32. [14] M.A. Nitsche, A. Schauenburg, N. Lang, D. Liebetanz, C. Exner, W. Paulus, F. Tergau, Facilitation of implicit motor learning by weak transcranial direct current stimulation of the primary motor cortex in the human, J. Cogn. Neurosci. 15 (2003) 619–626. [15] C.J. Stagg, G. Jayaram, D. Pastor, Z.T. Kincses, P.M. Matthews, H. Johansen-Berg, Polarity and timing-dependent effects of transcranial direct current stimulation in explicit motor learning, Neuropsychologia 49 (2011) 800–804. [16] S.S. Kantak, C.K. Mummidisetty, J.W. Stinear, Primary motor and premotor cortex in implicit sequence learning–evidence for competition between implicit and explicit human motor memory systems, Eur. J. Neurosci. 36 (2012) 2710–2715. [17] C.J. Stagg, M.A. Nitsche, Physiological basis of transcranial direct current stimulation, Neuroscientist 17 (2011) 37–53. [18] D. Reato, A. Rahman, M. Bikson, L.C. Parra, Effects of weak transcranial alternating current stimulation on brain activity-a review of known mechanisms from animal studies, Front. Hum. Neurosci. 7 (2013) 687. [19] G. Thut, C. Miniussi, J. Gross, The functional importance of rhythmic activity in the brain, Curr. Biol. 22 (2012) R658–R663. [20] C.S. Herrmann, S. Rach, T. Neuling, D. Struber, Transcranial alternating current stimulation: a review of the underlying mechanisms and modulation of cognitive processes, Front. Hum. Neurosci. 7 (2013) 279. [21] A. Antal, W. Paulus, Transcranial alternating current stimulation (tACS), Front. Hum. Neurosci. 7 (2013) 317. [22] A. Pogosyan, L.D. Gaynor, A. Eusebio, P. Brown, Boosting cortical activity at Beta-band frequencies slows movement in humans, Curr. Biol. 19 (2009) 1637–1641. [23] T. Zaehle, S. Rach, C.S. Herrmann, Transcranial alternating current stimulation enhances individual alpha activity in human EEG, PLoS One 5 (2010) e13766. [24] V. Krause, C. Wach, M. Sudmeyer, S. Ferrea, A. Schnitzler, B. Pollok, Cortico-muscular coupling and motor performance are modulated by 20 Hz transcranial alternating current stimulation (tACS) in Parkinson’s disease, Front. Hum. Neurosci. 7 (2014) 928. [25] C. Wach, V. Krause, V. Moliadze, W. Paulus, A. Schnitzler, B. Pollok, The effect of 10 Hz transcranial alternating current stimulation (tACS) on corticomuscular coherence, Front. Hum. Neurosci. 7 (2013) 511. [26] A. Antal, W. Paulus, Investigating neuroplastic changes in the human brain induced by transcranial direct (tDCS) and alternating current (tACS) stimulation methods, Clin. EEG Neurosci. 43 (2012) 175. [27] F. Frohlich, Endogenous and exogenous electric fields as modifiers of brain activity: rational design of noninvasive brain stimulation with transcranial alternating current stimulation, Dialogues Clin. Neurosci. 16 (2014) 93–102.

240

B. Pollok et al. / Behavioural Brain Research 293 (2015) 234–240

[28] Y.I. Shin, A. Foerster, M.A. Nitsche, Transcranial direct current stimulation (tDCS) – application in neuropsychology, Neuropsychologia 69 (2015) 154–175. [29] A. Antal, K. Boros, C. Poreisz, L. Chaieb, D. Terney, W. Paulus, Comparatively weak after-effects of transcranial alternating current stimulation (tACS) on cortical excitability in humans, Brain Stimul. 1 (2008) 97–105. [30] R.C. Oldfield, The assessment and analysis of handedness: the Edinburgh inventory, Neuropsychologia 9 (1971) 97–113. [31] M.A. Nitsche, D. Liebetanz, N. Lang, A. Antal, F. Tergau, W. Paulus, Safety criteria for transcranial direct current stimulation (tDCS) in humans, Clin. Neurophysiol. 114 (2003) 2220–2222, author reply 2222–2223. [32] W. Paulus, On the difficulties of separating retinal from cortical origins of phosphenes when using transcranial alternating current stimulation (tACS), Clin. Neurophysiol. 121 (2010) 987–991. [33] Z. Turi, G.G. Ambrus, K. Janacsek, K. Emmert, L. Hahn, W. Paulus, A. Antal, Both the cutaneous sensation and phosphene perception are modulated in a frequency-specific manner during transcranial alternating current stimulation, Restor. Neurol. Neurosci. 31 (2013) 275–285. [34] E.M. Robertson, A. Pascual-Leone, R.C. Miall, Current concepts in procedural consolidation, Nat. Rev. Neurosci. 5 (2004) 576–582. [35] S. Holm, A simple sequentially rejective multiple test procedure, Scand. J. Stat. 6 (1979) 65–70. [36] R. Kanai, L. Chaieb, A. Antal, V. Walsh, W. Paulus, Frequency-dependent electrical stimulation of the visual cortex, Curr. Biol. 18 (2008) 1839–1843. [37] T. Neuling, S. Rach, C.S. Herrmann, Orrating neuronal networks: sustained after-effects of transcranial alternating current stimulation depend upon brain states, Front. Hum. Neurosci. Chest 7 (2013) 161. [38] G. Thut, C. Miniussi, New insights into rhythmic brain activity from TMS–EEG studies, Trends Cogn. Sci. 13 (2009) 182–189. [39] M. Feurra, G. Bianco, E. Santarnecchi, M. Del Testa, A. Rossi, S. Rossi, Frequency-dependent tuning of the human motor system induced by transcranial oscillatory potentials, J. Neurosci. 31 (2011) 12165–12170.

[40] D.J. Schutter, R. Hortensius, Brain oscillations and frequency-dependent modulation of cortical excitability, Brain Stimul. 4 (2011) 97–103. [41] Y.H. Kim, J.W. Park, M.H. Ko, S.H. Jang, P.K. Lee, Facilitative effect of high frequency subthreshold repetitive transcranial magnetic stimulation on complex sequential motor learning in humans, Neurosci. Lett. 367 (2004) 181–185. [42] F. Tecchio, F. Zappasodi, G. Assenza, M. Tombini, S. Vollaro, G. Barbati, P.M. Rossini, Anodal transcranial direct current stimulation enhances procedural consolidation, J. Neurophysiol. 104 (2010) 1134–1140. [43] A.K. Engel, P. Fries, Beta-band oscillations–signalling the status quo? Curr. Opin. Neurobiol. 20 (2010) 156–165. [44] J. Classen, C. Gerloff, M. Honda, M. Hallett, Integrative visuomotor behavior is associated with interregionally coherent oscillations in the human brain, J. Neurophysiol. 79 (1998) 1567–1573. [45] J. Leite, S. Carvalho, F. Fregni, O.F. Goncalves, Task-specific effects of tDCS-induced cortical excitability changes on cognitive and motor sequence set shifting performance, PLoS One 6 (2011) e24140. [46] L.A. Boyd, M.A. Linsdell, Excitatory repetitive transcranial magnetic stimulation to left dorsal premotor cortex enhances motor consolidation of new skills, BMC Neurosci. 10 (2009) 72. [47] M.A. Nitsche, M. Jakoubkova, N. Thirugnanasambandam, L. Schmalfuss, S. Hullemann, K. Sonka, W. Paulus, C. Trenkwalder, S. Happe, Contribution of the premotor cortex to consolidation of motor sequence learning in humans during sleep, J. Neurophysiol. 104 (2010) 2603–2614. [48] M. Bikson, A. Datta, A. Rahman, J. Scaturro, Electrode montages for tDCS and weak transcranial electrical stimulation: role of return electrode’s position and size, Clin. Neurophysiol. 121 (2010) 1976–1978. [49] E.M. Robertson, The serial reaction time task: implicit motor skill learning, J. Neurosci. 27 (2007) 10073–10075.