Skip to main content
PMC home page
eLife logoLink to eLife
. 2021 Nov 23;10:e67355. doi: 10.7554/eLife.67355

Increasing human motor skill acquisition by driving theta–gamma coupling

Haya Akkad 1,2,✉, Joshua Dupont-Hadwen 1, Edward Kane 1, Carys Evans 1, Liam Barrett 3, Amba Frese 3, Irena Tetkovic 3, Sven Bestmann 1,4,†,✉, Charlotte J Stagg 2,5,6,†,✉
Editors: Thorsten Kahnt7, Richard B Ivry8
PMCID: PMC8687660  PMID: 34812140

Abstract

Skill learning is a fundamental adaptive process, but the mechanisms remain poorly understood. Some learning paradigms, particularly in the memory domain, are closely associated with gamma activity that is amplitude modulated by the phase of underlying theta activity, but whether such nested activity patterns also underpin skill learning is unknown. Here, we addressed this question by using transcranial alternating current stimulation (tACS) over sensorimotor cortex to modulate theta–gamma activity during motor skill acquisition, as an exemplar of a non-hippocampal-dependent task. We demonstrated, and then replicated, a significant improvement in skill acquisition with theta–gamma tACS, which outlasted the stimulation by an hour. Our results suggest that theta–gamma activity may be a common mechanism for learning across the brain and provides a putative novel intervention for optimizing functional improvements in response to training or therapy.

Research organism: Human

Introduction

The acquisition of motor skills is a central part of our everyday lives, from learning new behaviours such as riding a bike to the recovery of function after brain injury such as a stroke (Yarrow et al., 2009; Krakauer et al., 2019; Dhawale et al., 2017; Diedrichsen and Kornysheva, 2015). Better understanding of the mechanisms underpinning skill acquisition, to develop mechanistically informed strategies and tools to promote skill learning in healthy and pathological movement, is therefore a high-priority scientific and clinical goal.

Acquisition of motor skills is linked to a number of cortical and subcortical brain regions, but among these, primary motor cortex (M1) is thought to play a central role (Yarrow et al., 2009; Krakauer et al., 2019; Diedrichsen and Kornysheva, 2015; Sanes and Donoghue, 2000), making this a key target for neurorehabilitative interventions (Ward et al., 2019; Kang et al., 2016; Allman et al., 2016). However, the neurophysiological changes through which one might be able to promote skill acquisition in M1 are poorly understood, substantially hampering the development of novel interventions.

Outside the motor domain, the mechanisms underpinning learning have been extensively studied in the hippocampus, where theta-amplitude-coupled mid-gamma frequency activity (θ–γ phase-amplitude coupling [PAC]) has been hypothesized as a key learning-related mechanism. A prominent feature of hippocampal theta (4–8 Hz) activity is its co-incidence with higher-frequency activity in the γ range (30–140 Hz). Gamma coherence in the hippocampus alters during learning (Montgomery and Buzsáki, 2007) and memory retrieval (Yamamoto et al., 2014), and its relative synchrony during task predicts subsequent recall (Fell et al., 2001; Headley and Weinberger, 2011).

Hippocampal activity at different gamma frequencies is coupled to distinct phases of the underlying theta rhythm, suggesting that the precise relationship between gamma activity and theta phase may be important for function (Bragin et al., 1995; Lasztóczi and Klausberger, 2014; Colgin, 2015). For example, 60–80 Hz activity, which increases significantly during memory encoding, is coupled to the peak of the underlying theta oscillation (Lopes-Dos-Santos et al., 2018).

θ–γ PAC appears to be a conserved phenomenon across the cortex, and has been hypothesized as a fundamental operation of cortical computation in neocortical areas (Fries, 2009). For example, In the sensory cortices, it provides a neural correlate for perceptual binding (Lisman and Jensen, 2013). In the pre-frontal cortex, externally driven θ–γ PAC directly influences spatial working memory performance and global neocortical connectivity when gamma oscillations are delivered coinciding with the peak, but not the trough of theta waves (Muellbacher et al., 2001). It is proposed that the theta rhythm forms a temporal structure that organizes gamma-encoded units into preferred phases of the theta cycle, allowing careful processing and transmission of neural computations (Watrous, 2015). In the motor cortex, gamma oscillations at approximately 75 Hz are observed during movement (Crone et al., 1998; Pfurtscheller and Lopes da Silva, 1999; Pfurtscheller et al., 2003; Muthukumaraswamy, 2011; Crone et al., 2006; Muthukumaraswamy, 2010; Nowak et al., 2017), and an increased 75 Hz activity has been observed in dyskinesia, suggesting a direct pro-kinetic role (Swann et al., 2016; Swann et al., 2018). As in the hippocampus, M1 gamma oscillations are modulated by theta activity, with 75 Hz activity in human M1 being phase locked to the peak of the theta waveform (Canolty et al., 2006).

However, whether theta–gamma coupling plays a similar role in non-hippocampal-dependent skill learning in neocortical regions as it does in the hippocampus has not yet been determined. We therefore wished to test the hypothesis that θ–γ PAC is a conserved mechanism for learning across the brain, and therefore may provide a target for influencing the acquisition of new behaviour. To investigate the functional role of θ–γ PAC in learning outside the hippocampus, we modulated local theta–gamma activity via externally applied alternating current stimulation (tACS), a non-invasive form of brain stimulation that can interact with and modulate neural oscillatory activity in the human brain in a frequency-specific manner (Ali et al., 2013; Feurra et al., 2011; Zaehle et al., 2010), over M1 during learning of an M1-dependent ballistic thumb abduction task (Dupont-Hadwen et al., 2019) learning outside the hippocampus, we modulated local theta–gamma activity via externally applied alternating current stimulation (tACS), a non-invasive form of brain stimulation that can interact with and modulate neural oscillatory activity in the human brain in a frequency-specific manner (Ali et al., 2013; Feurra et al., 2011; Zaehle et al., 2010), over M1 during learning of an M1-dependent ballistic thumb abduction task (Dupont-Hadwen et al., 2019). We chose this task because it shows robust behavioural improvement in a relatively short period of time and performance improvement is underpinned by plastic changes in M1 (Classen et al., 1998; Muellbacher et al., 2001; Muellbacher et al., 2002). This encoding of kinematic details of the practiced movement is commonly regarded as a first step in skill acquisition (Classen et al., 1998).

We reasoned that if θ–γ PAC is a key mechanism for motor skill learning then interacting with θ–γ PAC, specifically with 75 Hz gamma activity applied at the theta peak (Lopes-Dos-Santos et al., 2018; Canolty et al., 2006), via tACS should have the capacity to modulate skill acquisition in healthy human participants, putatively via a change in local excitability. Moreover, if the functional role of this theta–gamma PAC is indeed critically dependent on the gamma activity occurring at a specific phase of theta activity then any behavioural effect should be specific to the theta phase at which the gamma was applied. To address this question, we therefore derived a waveform with gamma applied during the trough of the theta activity as an active control.

We first conducted an exploratory single-blinded experiment, in which we tested for the influence of theta–gamma-coupled stimulation on skill acquisition. This experiment revealed that when applied externally over right M1, gamma coupled to the peak of a theta envelope (TGP) substantially enhanced motor skill acquisition, compared to sham and an active stimulation control. Based on these results, we conducted a second, double-blind, pre-registered, sham-controlled experiment, which confirmed the beneficial effect of TGP on motor skill acquisition.

Results

One hundred and four healthy participants performed a M1-dependent ballistic thumb abduction task with their left hand (Dupont-Hadwen et al., 2019; Rogasch et al., 2009; Rosenkranz et al., 2007) while tACS was applied over the right M1. Volunteers trained to increase thumb abduction acceleration in their left, non-dominant, thumb over 5 blocks of 70 trials each (Figure 1).

Figure 1. Theta–gamma transcranial alternating current stimulation (tACS) protocol and task.

Figure 1.

Open in a new tab

(A) Electrode montage: the theta–gamma tACS montage was delivered with one electrode centred over right M1 (red, C4) and the other over the parietal vertex (blue, Pz). Insert: electrical field distribution projected on a rendered reconstruction of the cortical surface in a single individual, demonstrating significant current within M1. (B) tACS waveform: a 75 Hz gamma rhythm was amplitude modulated by the peak (theta–gamma peak [TGP]; upper panel) or trough (theta–gamma trough [TGT]; lower panel) envelope of a 2 mA peak-to-peak 6 Hz theta rhythm. (C) Experimental design: all subjects performed a baseline block, followed by five task blocks. In experiment 2, to assess the duration of behavioural effects, subjects performed an additional two task blocks 75 min after the initial task was complete. Each block consisted of 70 trials with an inter-block interval of 2 min, apart from a 10 min and 1 hr break after blocks 4 and 5, respectively. Stimulation was delivered for 20 min during the first three blocks. (D) Trial design: all trials began with three auditory warning tones acting as a ready-steady-go cue. At the third tone, participants abducted their thumb along the x-axis as quickly as possible and were given online visual feedback of their performance via a screen positioned in front of them. Feedback was presented as a scrolling bar chart with the magnitude of acceleration displayed on a trial-by-trial basis; a green bar indicated acceleration was higher than the previous trial and a red bar indicated the opposite.

Fifty-eight participants (age: 24 ± 5.1 years, 37 females) participated in experiment 1, and were randomly assigned to one of three experimental groups, which received either 20 min of tACS over right primary motor cortex, or sham. Similar to a previous study in the spatial working memory domain (Alekseichuk et al., 2016), in the active tACS condition, participants received (1) theta–gamma peak (TGP) stimulation (Figure 1A), whereby gamma frequency (75 Hz) stimulation was delivered during the peak of a 6 Hz theta envelope as is found naturally in the human motor cortex (Canolty et al., 2006), or (2) an active control, theta–gamma trough (TGT) stimulation, whereby the gamma stimulation was delivered in the negative half of the theta envelope. For sham stimulation, 6 Hz theta was briefly ramped up for 10 s, and then ramped down again. Participants performed the skill learning task during the stimulation, and for approximately 15 min after cessation of stimulation.

TGP stimulation improves motor skill acquisition

We first wished to assess whether participant performance improved with training, regardless of stimulation. As expected, skill increased in all three groups over the course of the experiment (repeated measures analysis of variance [ANOVA] with one factor of block [1–6] and one factor of condition [TGP, TGT, and sham], main effect of block F(2.203,121.187) = 85.122, p < 0.001). However, the stimulation groups differed significantly in their skill acquisition (main effect of condition F(2,55) = 3.396, p = 0.041; condition × block interaction F(4.407,121.187) = 2.692, p = 0.03). Post hoc tests (using Tukey correction for multiple comparisons) revealed a significant difference between TGP and sham (p = 0.04, 95% CI [0.50, 21.78]) and no significant difference between TGT and sham (p = 0.766, 95% CI [−7.52, 13.79]) or TGP and TGT (p = 0.162, 95% CI [−2.40, 18.35]). To further explore the interaction effect, we ran an analysis of simple effects to determine the effect of the condition factor (TGP, TGT, and sham) at each level of the block factor (1–6). This revealed a significant simple effect of condition during stimulation blocks F(2,55) = 4.13, p = 0.021. In line with our primary hypothesis, follow-up analyses demonstrated a 26% larger acceleration gain from baseline during TGP stimulation, compared with sham condition (independent t-test t(36) = 3.052, p = 0.004, Cohen’s d = 0.98, Figure 2A).

Figure 2. Theta–gamma peak (TGP)-transcranial alternating current stimulation (tACS) enhances motor skill acquisition.

Mean ballistic thumb abduction acceleration for each stimulation condition. Each point represents the mean of 10 trials across participants and the error bars depict the standard error between participants. (A) Experiment 1: during stimulation, TGP significantly increased skill acquisition over the course of the experiment (i.e. acceleration gain), compared to sham and theta–gamma trough (TGT). (B) Experiment 2: when replicated in an independent sample, skill acquisition was again significantly greater in the TGP stimulation group compared with sham. This effect was maintained for at least 75 min after stimulation.

Figure 2.

Figure 2—figure supplement 1. Transcranial alternating current stimulation (tACS) does not modulate behavioural variability.

Figure 2—figure supplement 1.

There was no effect of tACS on variability in terms of acceleration within blocks in either (A) experiment 1 or (B) experiment 2.
Open in a new tab

There were no significant differences in baseline performance between TGP, TGT, and sham conditions as demonstrated by a simple effects analysis of the factor of Condition (TGP, TGT, and sham) at the level of baseline block 1 F(2,55) = 0.30, p = 0.743.

This first experiment established the relevant role of theta–gamma-coupled tACS over M1 on motor skill learning in healthy participants, here expressed through an increase in learning. This effect was most effective when gamma frequency stimulation was coupled to the peak of the underlying theta frequency stimulation waveform, as opposed to when it was coupled to the trough of theta. We next sought to confirm this result in an independent cohort, and to further assess the duration of this improvement post-stimulation.

Behavioural effects of TGP stimulation are replicable

In order to try to replicate our results from experiment 1, we conducted a double-blind, pre-registered (https://osf.io/xjpef) replication experiment in an independent sample of 46 participants (age 24 ± 4.1, 32 females, all right handed). Because our first experiment had shown the largest effect on skill learning with TGP stimulation, we now focussed on this condition. Participants were randomised to either TGP stimulation or sham. The experimental protocol was identical to experiment 1, except that we additionally included a probe to test retention at 1 hr after the end of stimulation. There was no significant difference in baseline performance between TGP and sham conditions (t(44) = 0.734, p = 0.467).

As in experiment 1, participants in both conditions showed an improvement in performance throughout the experiment (repeated measures ANOVA, one factor of block [1–8], one factor of condition [TGP and sham]; main effect of block F(3.302,145.239) = 72.912, p < 0.001; Figure 2B).

However, there was a significant difference in skill acquisition between the two conditions (main effect of condition (F(1,44) = 27.241, p < 0.001); block × condition interaction F(3.302,145.239) = 7.258, p < 0.001). The TGP group achieved significantly greater acceleration gain compared to sham during stimulation (t(44) = 4.201, p < 0.001, Cohen’s d = 1.24).

Bang’s blinding index (BI; Bang et al., 2004) indicated successful blinding in both real and sham stimulation groups. Blinding indices were 0.07 and −0.03 in the TGP and sham groups, respectively.

Motor skill gains are retained post-stimulation

We next wished to explore whether the behavioural effects of stimulation outlasted the stimulation period, or whether skill in this group returned to baseline after stimulation had ceased. Comparing the two groups at 75 min post-stimulation demonstrated that the TGP group had a significantly faster acceleration than the sham group (t(44) = 3.430, p = 0.001, Cohen’s d = 1.01).

tACS does not significantly modulate the variability or latency of responses

tACS may increase skill acquisition by changing one or more different aspects of behaviour. Non-invasive brain stimulation approaches have previously been demonstrated to increase behavioural variability in tasks similar to that implemented here (Teo et al., 2010). First, we investigated whether tACS significantly modulated the variability in the maximum acceleration achieved. We ran ANOVAs on the coefficient of variation for each subject for each block, with a within-subject factor of block and between-subject factor of condition for each experiment separately. This revealed a main effect of block in both experiments (E1: F(5,275)=17.1, p < 0.001; E2: F(7,308) = 13.8, p < 0.001), reflecting a general decrease in variability during the task, but no main effect of condition (E1: F(2,55) = 2.36, p = 0.104; E2: F(1,44) = 0.14, p = 0.90) and no block × condition interaction (E1: F(10,275) = 1.15, p = 0.329; E2: F(7,308) = 1.60, p = 0.14; Figure 2—figure supplement 1).

Second, we wished to investigate whether tACS-modulated response time. We therefore ran an ANOVA with a within-subject factor of block and between-subject factor of condition. In experiment 1, there was a significant main effect of block (F(2.18,120) = 11.68, p < 0.001) and condition (F(2,55) = 4.66, p = 0.013), but no significant block × condition interaction (F(4,63,120) = 0.195, p = 0.95). However, when we repeated this analysis for experiment 2 there was no significant effect of block (F(1,44) = 0.014, p = 0.905), and no significant block by condition interaction (F(3.60,158.45) = 0.372, p = 0.30).

Discussion

Theta-amplitude-modulated gamma activity may provide an important mechanism for non-hippocampal-dependent skill acquisition. We used non-invasive brain stimulation to modulate θ–γ PAC in human primary motor cortex in two separate cohorts, one a pre-registered, double-blind study and demonstrated that externally applied θ–γ PAC during a motor task increases skill acquisition in healthy adults. This behavioural improvement was critically dependent on the phase relationship of the theta and gamma components of the stimulation.

Behavioural improvements depend on the phase of theta–gamma coupling

Our results suggest that driving γ activity during the peak, but not the trough, of θ oscillations improves motor skill acquisition. θ–γ PAC has consistently been demonstrated to relate to learning in the rodent CA1 (Bragin et al., 1995; Lasztóczi and Klausberger, 2014; Colgin, 2015; Lopes-Dos-Santos et al., 2018), where oscillations in the θ (5–12 Hz) band become dominant during active exploration (O’Keefe and Recce, 1993), and have been widely hypothesized to allow information coming into CA1 from distant regions to be divided into discrete units for processing (Buzsáki, 2002; Buzsáki and Moser, 2013). A prominent feature of hippocampal theta activity is its co-incidence with higher-frequency activity in the γ range (30–140 Hz). Gamma coherence in the hippocampus alters during learning (Montgomery and Buzsáki, 2007) and memory retrieval (Yamamoto et al., 2014), and its relative synchrony during task predicts subsequent recall (Fell et al., 2001; Headley and Weinberger, 2011). Non-invasively stimulating the human temporal cortex during memory encoding using tACS to increase θ–γ coupling has been variously shown to impair (Lara et al., 2018) or strengthen (Reinhart and Nguyen, 2019) hippocampal memory formation.

Hippocampal activity at different frequencies within the gamma band is coupled to distinct phases of the underlying theta rhythm, suggesting that the precise relationship between gamma activity and theta phase may be important for function (Bragin et al., 1995; Lasztóczi and Klausberger, 2014; Colgin, 2015). For example, 60–80 Hz activity, which increases significantly during memory encoding, is coupled to the peak of the underlying theta oscillation (Lopes-Dos-Santos et al., 2018).

θ–γ PAC appears to be a conserved phenomenon across the cortex and has been hypothesized as a fundamental operation of cortical computation in neocortical areas (Fries, 2009; Lisman and Jensen, 2013; Alekseichuk et al., 2016). Supporting this hypothesis, a recent human study demonstrated an improvement in working memory using tACS (Reinhart and Nguyen, 2019). However, no study to date has shown that θ–γ PAC can modulate non-hippocampal-dependent learning as we do here.

75 Hz activity has a pro-kinetic role in M1 and relates to skill acquisition

Our experiments indicate that 75 Hz activity, coupled to 6 Hz oscillations, can improve motor skill acquisition. We chose 75 Hz stimulation for two reasons: it is implicated in learning in the hippocampus and physiologically, M1 gamma activity centred around 75 Hz occurs at the peak of ongoing theta activity (Canolty et al., 2006) and is ubiquitous in studies of human movement. 75 Hz activity only occurs during actual, rather than imagined, movement (Muthukumaraswamy, 2010), and shows topographical specificity within M1 (Crone et al., 2006). Its hypothesized pro-kinetic role is further supported by the finding of a pathological increase in narrow-band 75 Hz activity within M1 in hyperkinetic patients with Parkinson’s disease (Swann et al., 2016). Our group have previously shown that the degree of response to 75 Hz tACS predicts subsequent learning potential, further highlighting a role for 75 Hz activity not only in movement but in skill acquisition (Nowak et al., 2017). Here, we demonstrate that a more physiological approach to delivering gamma stimulation by coupling it to theta rhythms leads to theta-phase-specific improvements in skill acquisition – something that may allow the development of more targeted therapeutic interventions.

Behavioural benefits of theta–gamma PAC may be mediated by decreases in inhibition

Decreases in M1 GABAergic activity are a central mechanism for motor plasticity (Stagg et al., 2009; Clarkson et al., 2010; Stagg et al., 2011; Blicher et al., 2015; Bachtiar et al., 2015; Traub et al., 1996). However, it is not yet clear how these decreases alter behaviour. θ–γ PAC may be a candidate mechanism for this: M1 gamma activity arises from GABAergic inter-neuronal micro-circuits involving layer V Parvalbumin+ ve neurons (Whittington and Traub, 2003; Bartos et al., 2007; Cabral et al., 2011; Chen et al., 2017; Sohal et al., 2009; Ni et al., 2016; Whittington et al., 2011; Masamizu et al., 2014) thought to be involved in motor learning (Johnson et al., 2017). In slice preparations, theta–gamma coupling within M1 arises spontaneously from layer V when GABA activity is blocked (Johnson et al., 2020). In humans, modulating M1 75 Hz activity in humans using tACS leads to a decrease in local GABAergic activity, the magnitude of which predicts motor learning ability on a subject-by-subject basis (Nowak et al., 2017). The effects of low-frequency tACS may be mediated through cyclically inducing a phase of enhanced excitation (peak) followed by a phase of reduced excitation (trough). If decreases in M1 GABAergic activity is necessary for motor plasticity (Bang et al., 2004; Teo et al., 2010; O’Keefe and Recce, 1993; Buzsáki, 2002; Buzsáki and Moser, 2013), then phases of enhanced excitation (or reduced inhibition) would offer an optimal entrainment window for excitatory rhythms, such as pro-kinetic 75 Hz gamma.

Given the extensive evidence for decreases in GABAergic activity for motor cortical plasticity, it may be that gamma activity, particularly synchronization of gamma activity via theta oscillations, represents an emergent signature of learning that might be targeted to improve behaviour, though the cellular and layer specificity of our findings remain to be determined.

The behavioural effects of tACS are not driven by changes in variability or latency of responses

There are a number of potential mechanisms by which the behavioural improvements we observed might have arisen. Previous studies have demonstrated that skill improvement due to non-invasive brain stimulation might occur via an increase in the variability of behavioural responses (Teo et al., 2010), but this does not seem to be the case here. Additionally, it is possible that our measure of skill learning was confounded by a stimulation-induced change in response time, but the data do not support this hypothesis. However, given the strongly pro-kinetic role of 75 Hz activity in M1, further studies should look at the specific components of motor behaviour this tACS protocol may modulate to identify the precise aspects of motor skill acquisition theta–gamma tACS may modulate.

Anatomical- and frequency specificity of behavioural effects

θ–γ PAC has been suggested as a mechanism by which anatomically distant brain regions become functionally connected (Fries, 2009). We deliberately chose a task that is M1 dependent, thereby providing us with a cortical target for our stimulation, and have not set out to target more than one node in the network. We are confident that we are actively stimulating M1: our tACS protocol induces excitability changes in M1, suggesting a significant physiological effect in this region, and the electrical field simulation demonstrates a significant current within M1 due to tACS. However, this does not rule out that the behavioural effects we observe arise from multiple nodes, and that there is a contribution of the parietal electrode: indeed as with all tACS studies, the current is relatively wide-spread across the cortex. This hypothesis remains to be tested.

Here, we tested an a priori hypothesis about theta–gamma PAC, and its role in non-hippocampal-dependent skill acquisition. We did not test other frequency couplings, and so we cannot claim that similar effects would not be seen with other cross-frequency stimulation paradigms. Similarly, we did not directly test whether θ–γ PAC was superior to either θ or γ stimulation alone. However, by using an active TGT control condition, which delivered the same θ and γ stimulation, and only varied the phase of the θ at which the γ was present, if the behavioural effects seen were solely dependent on either frequency alone then both the peak and trough conditions would have improved learning, which was not the case. Previous studies have shown that gamma stimulation alone can improve learning (Asamoah et al., 2019; Moisa et al., 2016), but not to such as degree as θ–γ PAC (Alekseichuk et al., 2016).

Lack of TGT behavioural effect supports the hypothesis that tACS directly modulates neural activity

There has been some recent controversy about the contribution of direct stimulation of the underlying neural tissue versus other mechanisms (Krause et al., 2019; Vieira et al., 2020; Vöröslakos et al., 2018 ) to the behavioural and physiological effects of tACS, although recent work strongly supports the argument that tACS directly entrains ongoing neural activity (Krause et al., 2019; Vieira et al., 2020). While this paper does not aim to directly address this question, we are confident that our behavioural effects result from direct effects of the current in the brain. Firstly, tACS at the current densities used here have been demonstrated to entrain single-neuron activity in non-human primates (Krause et al., 2019), suggesting at least that direct neuronal entrainment is a possible mechanism. Secondly, although stimulation of peripheral scalp nerves has recently been suggested as a putative explanation for behavioural effects of tACS (Asamoah et al., 2019), in experiment 1, we used an inverted waveform as an active control to rule out effects driven by peripheral stimulation, and in experiment 2, there was successful blinding to stimulation type. Collectively, this suggests that the sensory sensations that may arise from stimulation did not substantially differ between active and sham conditions.

Conclusions

In conclusion, we wished to test whether theta–gamma PAC was an important mechanism in non-hippocampally dependent learning in humans. Using a novel non-invasive brain stimulation approach in humans that emulates known neurophysiological activity patterns during learning (Lopes-Dos-Santos et al., 2018; Lisman and Jensen, 2013; Canolty et al., 2006), we demonstrated, and then replicated, a substantial behavioural improvement due to stimulation. While the neural underpinnings of this functional outcome need to be explored, this result offers a new technique not only to understand physiological mechanisms of human neuroplasticity, but also potentially a putative novel adjunct therapy for promoting post-stroke recovery.

Materials and methods

Experiment 1

Fifty-eight participants (24 ± 5.1 years, 37 females) gave their written informed consent to participate in the experiments in accordance with local ethics committee approval. Participants were right handed and had no contraindications for tACS. Participants were randomly assigned to one of three tACS conditions (N = 20 per condition): (1) TGP stimulation (Figure 1A), whereby gamma frequency (75 Hz) stimulation was delivered during the peak of a 6 Hz theta envelope, (2) an active control: TGT stimulation where the gamma stimulation was delivered in the negative half of the theta envelope, and (3) sham stimulation. Participants were blinded to the type of stimulation delivered and naive to the purpose of the experiment.

Experimental setup

Participants performed a ballistic thumb abduction training task requiring abduction of their left (non-dominant) thumb with maximal acceleration (Dupont-Hadwen et al., 2019; Rogasch et al., 2009; Rosenkranz et al., 2007). Participants were seated with their left arm slightly abducted, with the elbow flexed to 45° (where 0° is full extension) and the forearm semi-pronated with the palm facing inwards. The left hand was chosen to avoid ceiling effects that might be present in the dominant hand. The arm, wrist, and proximal interphalangeal joints were secured in a plastic custom-built arm fixture to prevent the unintentional contribution of whole hand movement to the ballistic acceleration, though the thumb was left free to move (Figure 1C).

The acceleration of the thumb was measured in the x-axis (abduction plane) using an accelerometer (ACL300; Biometrics Ltd, UK) attached to the distal phalanx of the thumb. Recording from the accelerometer was confined to one axis to allow for good skill improvement by providing simplified feedback for the participant (Dupont-Hadwen et al., 2019; Rogasch et al., 2009; Rosenkranz et al., 2007).

Behavioural task

Participants performed ballistic thumb abduction movements of their left hand at a rate of 0.4 Hz indicated by a ready-steady-go procedure, with each of three auditory tones (400 Hz, 300 ms duration) spaced at 500 ms intervals. Participants were instructed to move their thumbs at the onset of the third auditory tone. The behavioural task was separated into six blocks (Figure 1B). Participants performed an initial baseline block of 30 trials. This was followed by four blocks separated by a break of at least 2 min to minimize fatigue, and a final block separated by a 10 min break. Each of these five blocks consisted of 70 trials with a 30 s break between every 35 trials to avoid within block fatigue. Participants were asked to remain at rest during breaks, avoiding any thumb movement.

In all blocks except the baseline block, participants were instructed to move as fast as possible and were encouraged to try to increase their acceleration on every trial. Participants were given visual feedback about the acceleration of their movements on a trial-by-trial basis (Figure 1C). Feedback was presented as a scrolling bar chart with the magnitude of the current acceleration plotted after each trial. If the acceleration on the current trial was greater than on the previous trial, the bar was plotted in green, and if it was less the bar was plotted in red. If a movement was made too early or too late (i.e. movement outside a 300 ms window centred on one second after the first tone), no acceleration feedback was given; instead, the message ‘too early’ or ‘too late’ was presented. Additionally, participants were informed of their progress by displaying a moving average of acceleration values over the preceding 10 trials, indicated by a line plotted on the screen over the locations of the 10 consequential trials.

In the baseline block, participants were told to move as closely as possible to the onset of the third tone, and feedback about the temporal accuracy of the movement was given by the experimenter.

Behavioural data analysis

Data were analysed via Matlab (Mathworks). The maximal acceleration was calculated for each trial, and any trials with a maximum acceleration less than 4.9 ms2 were rejected (Dupont-Hadwen et al., 2019). Additionally, if movements were made too early or too late, that is the onset of acceleration of the movement lay more than 300 ms before or after the expected movement time, they were also rejected (Dupont-Hadwen et al., 2019). Together, this approach led to 1.45 ± 0.94 (mean ± standard deviation [SD]) trials being removed per block of 70 trials in experiment 1, and 0.88 ± 0.99 (mean ± SD) trials removed per block of 70 trials in experiment 2. There was no statistical difference between the number of trials being removed per block in each condition (experiment 1: mixed ANOVA, block × condition [F(5.409,148.742) = 1.649, p = 0.145]; experiment 2: mixed ANOVA, block × condition [F(2.8,137.4) = 1.05, p = 0.396]).

Transcranial alternating current stimulation

tACS was delivered via a DC stimulator in AC mode (NeuroConn DC-Stimulator Plus) through a pair of sponge surface electrodes (5 × 5 cm2). Saline was used as a conducting medium between the scalp and the electrodes. The anode was centred over the right primary motor cortex (C4) and the cathode was positioned over the parietal vertex (Pz), in accordance with the international 10–20 EEG system. Impedance was kept below 10 kΩ. The electrode positions were based on simulation of current flow across the brain, using HD-Explore software (Soterix Medical Inc, New York) which uses a finite-element-method approach to model electrical field intensities throughout the brain (Datta et al., 2013). This confirmed that current was directed to the primary motor cortex (Figure 1A).

The TGP condition consisted of 20 min continuous, sinusoidal 6 Hz (theta) stimulation at an intensity of 2 mA peak-to-peak, coupled with bursts of a sinusoidal 75 Hz (gamma) rhythm amplitude modulated by the positive theta phase (0–180°; Figure 1A). The TGT condition consisted of 20 min continuous, sinusoidal 6 Hz (theta) stimulation at an intensity of 2 mA peak-to-peak, coupled with bursts of a sinusoidal 75 Hz (gamma) rhythm amplitude modulated by the negative theta phase (180–360°). Finally, the sham condition consisted of a 10 s continuous sinusoidal 6 Hz stimulation.

The theta–gamma waveforms were custom coded on the Matlab software and delivered to the NeuroConn stimulator via a data acquisition device (National Instruments USB-6259 BNC). Theta–gamma stimulation was then delivered to the scalp surface electrodes through the NeuroConn stimulator in ‘remote’ mode. Sham stimulation was delivered directly through the NeuroConn stimulator. tACS was administered in a between-subject design. Participants were randomized to receive either 10 s of sham stimulation during the first training block or 20 min of TGP or TGT stimulation during the first three training blocks. Participants were blinded to the stimulation condition used and naive to the purpose of the experiment.

Statistical analyses

Data were tested for normality using the Kolmogorov–Smirnov test. Statistical analyses were performed using SPSS. We used a two-way mixed ANOVA with two independent variables, ‘condition’ (between-subject variable) and ‘block’ (within-subject variable). Acceleration in ms2 was our only dependent variable. Where there was a significant block × condition interaction, we analysed the simple effect of condition within levels of block. Post hoc t-tests were conducted as appropriate and multiple comparisons were corrected for using the Tukey HSD test. When sphericity assumptions were violated, results are reported with a Greenhouse–Geisser correction.

Experiment 2

Forty-four participants (age 24 ± 4.1 years, 32 females) gave their written informed consent to participate in the experiments in accordance with local ethics committee approval. Participants were right handed and had no contraindications for tACS. We performed a pre-registered, double-blinded replication of experiment 1 (TGP and sham only) in an independent sample. The experimental design was pre-registered in full on the Open Science Framework (https://osf.io/xjpef). The experimental design was identical to experiment 1, except in the following aspects.

Power Calculation

Sample size was calculated based on the Cohen’s d effect size of the mean improvement in performance from baseline between the TGP and sham conditions in experiment 1. Given a Cohen’s d = 0.98, 1 − β = 0.95 and α = 0.05 this gave a sample size of 24 per group (G*Power), and allowing for a 10% loss of data, we recruited 27 participants per condition.

Blinding

On the day of testing, a researcher not involved in data analysis and blinded to experimental protocol and rationale (AF, IT, and LB) collected the data and interacted with the participant. Another researcher (HA), not involved in data collection and blinded during data analysis, setup the stimulation condition on the day of testing, but did not interact with the participant. Unblinding was performed following the completion of data collection and analysis. Participants were naive to the purpose of the experiment.

Participants completed a blinding questionnaire at the end of the experiment that required them to identify whether they believed they had received real or sham stimulation. To assess the effectiveness of our blinding, we used Bang’s BI, where a BI of 1 suggests complete unblinding, a BI of 0 random guessing and a BI of −1 opposite guessing.

Behavioural task

The behavioural task parameters were identical to those in experiment 1, but now with an additional two training blocks separated from the previous six blocks by a break of 1 hr (Figure 1B). During the 1 hr break, participants remained seated and at rest while watching a documentary (Planet Earth, season 1 episode 10). The additional seventh and eighth blocks were separated by a break of at least 2 min to minimize fatigue and each consisted of 70 trials, with a 30 s break between every 35 trials to minimize within block fatigue. Participants were asked to remain at rest during breaks, avoiding any thumb movement.

Transcranial alternating current stimulation

Stimulation parameters were identical to those in experiment 1, but only included the TGP and sham conditions. Both the participant and the experimenter were blinded to the stimulation condition used.

Acknowledgements

HA holds a doctoral fellowship funded by Brain Research UK (552175). SB was funded by Brain Research UK (201617-03) and Dunhill Medical Trust (RPGF1810/93). CJS holds a Sir Henry Dale Fellowship, funded by the Wellcome Trust and the Royal Society (102584/Z/13/Z). The work was supported by the NIHR Biomedical Research Centre, Oxford and the NIHR Oxford Health Biomedical Research Centre. The Wellcome Centre for Integrative Neuroimaging is supported by core funding from the Wellcome Trust (203139/Z/16/Z). The Wellcome Centre for Human Neuroimaging is supported by funding from the Wellcome Trust (203147/Z/16/Z).

Funding Statement

The funders had no role in study design, data collection, and interpretation, or the decision to submit the work for publication.

Contributor Information

Haya Akkad, Email: haya.akkad.14@ucl.ac.uk.

Sven Bestmann, Email: s.bestmann@ucl.ac.uk.

Charlotte J Stagg, Email: charlotte.stagg@ndcn.ox.ac.uk.

Thorsten Kahnt, Northwestern University, United States.

Richard B Ivry, University of California, Berkeley, United States.

Funding Information

This paper was supported by the following grants:

  • Royal Society Sir Henry Dale Fellowship 102584/Z/13/Z to Charlotte J Stagg.

  • Brain Research UK 201617-03 to Sven Bestmann.

  • Brain Research UK Graduate Student Fellowship to Haya Akkad.

  • Wellcome Trust Sir Henry Dale Fellowship - 102584/Z/13/Z to Charlotte J Stagg.

Additional information

Competing interests

No competing interests declared.

No competing interests declared.

Author contributions

Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing - original draft, Writing – review and editing.

Methodology.

Data curation.

Data curation.

Data curation.

Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – review and editing.

Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Visualization, Writing – review and editing.

Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – review and editing.

Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Visualization, Writing – review and editing.

Ethics

Human subjects: Ethical permission for this study was granted by the University College London Research Ethics Committee (UCLREC: 6285/001). Written informed consent was obtained from all volunteers prior to data collection.

Additional files

Transparent reporting form

Data availability

All data generated or analysed during this study are included in the manuscript and freely available on the open science framework (https://osf.io/xjpef). Details of data analysis, experimental design and protocol were pre-registered prior to data collection and freely available on the open science framework - Registration form: osf.io/xjpef; Files: osf.io/452f8/files/.

The following previously published datasets were used:

Akkad H, Dupont-Hadwen J, Bestmann S, Stagg CJ. 2018. Improving motor learning via phase-amplitude coupled theta-gamma tACS. Open Science Framework.

References

  1. Alekseichuk I, Turi Z, Amador de Lara G, Antal A, Paulus W. Spatial Working Memory in Humans Depends on Theta and High Gamma Synchronization in the Prefrontal Cortex. Current Biology. 2016;26:1513–1521. doi: 10.1016/j.cub.2016.04.035. [DOI] [PubMed] [Google Scholar]
  2. Ali MM, Sellers KK, Fröhlich F. Transcranial alternating current stimulation modulates large-scale cortical network activity by network resonance. The Journal of Neuroscience. 2013;33:11262–11275. doi: 10.1523/JNEUROSCI.5867-12.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Allman C, Amadi U, Winkler AM, Wilkins L, Filippini N, Kischka U, Stagg CJ, Johansen-Berg H. Ipsilesional anodal tDCS enhances the functional benefits of rehabilitation in patients after stroke. Science Translational Medicine. 2016;8:330re1. doi: 10.1126/scitranslmed.aad5651. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Asamoah B, Khatoun A, Mc Laughlin M. tACS motor system effects can be caused by transcutaneous stimulation of peripheral nerves. Nature Communications. 2019;10:266. doi: 10.1038/s41467-018-08183-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bachtiar V, Near J, Johansen-Berg H, Stagg CJ. Modulation of GABA and resting state functional connectivity by transcranial direct current stimulation. eLife. 2015;4:e08789. doi: 10.7554/eLife.08789. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bang H, Ni L, Davis CE. Assessment of blinding in clinical trials. Controlled Clinical Trials. 2004;25:143–156. doi: 10.1016/j.cct.2003.10.016. [DOI] [PubMed] [Google Scholar]
  7. Bartos M, Vida I, Jonas P. Synaptic mechanisms of synchronized gamma oscillations in inhibitory interneuron networks. Nature Reviews. Neuroscience. 2007;8:45–56. doi: 10.1038/nrn2044. [DOI] [PubMed] [Google Scholar]
  8. Blicher JU, Near J, Næss-Schmidt E, Stagg CJ, Johansen-Berg H, Nielsen JF, Østergaard L, Ho Y-CL. GABA levels are decreased after stroke and GABA changes during rehabilitation correlate with motor improvement. Neurorehabilitation and Neural Repair. 2015;29:278–286. doi: 10.1177/1545968314543652. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bragin A, Jandó G, Nádasdy Z, Hetke J, Wise K, Buzsáki G. Gamma (40-100 Hz) oscillation in the hippocampus of the behaving rat. The Journal of Neuroscience. 1995;15:47–60. doi: 10.1523/JNEUROSCI.15-01-00047.1995. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Buzsáki G. Theta oscillations in the hippocampus. Neuron. 2002;33:325–340. doi: 10.1016/s0896-6273(02)00586-x. [DOI] [PubMed] [Google Scholar]
  11. Buzsáki G, Moser EI. Memory, navigation and theta rhythm in the hippocampal-entorhinal system. Nature Neuroscience. 2013;16:130–138. doi: 10.1038/nn.3304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cabral J, Hugues E, Sporns O, Deco G. Role of local network oscillations in resting-state functional connectivity. NeuroImage. 2011;57:130–139. doi: 10.1016/j.neuroimage.2011.04.010. [DOI] [PubMed] [Google Scholar]
  13. Canolty RT, Edwards E, Dalal SS, Soltani M, Nagarajan SS, Kirsch HE, Berger MS, Barbaro NM, Knight RT. High gamma power is phase-locked to theta oscillations in human neocortex. Science. 2006;313:1626–1628. doi: 10.1126/science.1128115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Chen G, Zhang Y, Li X, Zhao X, Ye Q, Lin Y, Tao HW, Rasch MJ, Zhang X. Distinct Inhibitory Circuits Orchestrate Cortical beta and gamma Band Oscillations. Neuron. 2017;96:1403–1418. doi: 10.1016/j.neuron.2017.11.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Clarkson AN, Huang BS, Macisaac SE, Mody I, Carmichael ST. Reducing excessive GABA-mediated tonic inhibition promotes functional recovery after stroke. Nature. 2010;468:305–309. doi: 10.1038/nature09511. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Classen J, Liepert J, Wise SP, Hallett M, Cohen LG. Rapid plasticity of human cortical movement representation induced by practice. Journal of Neurophysiology. 1998;79:1117–1123. doi: 10.1152/jn.1998.79.2.1117. [DOI] [PubMed] [Google Scholar]
  17. Colgin LL. Theta-gamma coupling in the entorhinal-hippocampal system. Current Opinion in Neurobiology. 2015;31:45–50. doi: 10.1016/j.conb.2014.08.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Crone N E, Miglioretti DL, Gordon B, Sieracki JM, Wilson MT, Uematsu S, Lesser RP. Functional mapping of human sensorimotor cortex with electrocorticographic spectral analysis. I. Alpha and beta event-related desynchronization. Brain. 1998;121 (Pt 12):2271–2299. doi: 10.1093/brain/121.12.2271. [DOI] [PubMed] [Google Scholar]
  19. Crone NE, Sinai A, Korzeniewska A. High-frequency gamma oscillations and human brain mapping with electrocorticography. Progress in Brain Research. 2006;159:275–295. doi: 10.1016/S0079-6123(06)59019-3. [DOI] [PubMed] [Google Scholar]
  20. Datta A, Zhou X, Su Y, Parra LC, Bikson M. Validation of finite element model of transcranial electrical stimulation using scalp potentials: implications for clinical dose. Journal of Neural Engineering. 2013;10:036018. doi: 10.1088/1741-2560/10/3/036018. [DOI] [PubMed] [Google Scholar]
  21. Dhawale AK, Smith MA, Ölveczky BP. The Role of Variability in Motor Learning. Annual Review of Neuroscience. 2017;40:479–498. doi: 10.1146/annurev-neuro-072116-031548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Diedrichsen J, Kornysheva K. Motor skill learning between selection and execution. Trends in Cognitive Sciences. 2015;19:227–233. doi: 10.1016/j.tics.2015.02.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Dupont-Hadwen J, Bestmann S, Stagg CJ. Motor training modulates intracortical inhibitory dynamics in motor cortex during movement preparation. Brain Stimulation. 2019;12:300–308. doi: 10.1016/j.brs.2018.11.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Fell J, Klaver P, Lehnertz K, Grunwald T, Schaller C, Elger CE, Fernández G. Human memory formation is accompanied by rhinal-hippocampal coupling and decoupling. Nature Neuroscience. 2001;4:1259–1264. doi: 10.1038/nn759. [DOI] [PubMed] [Google Scholar]
  25. Feurra M, Bianco G, Santarnecchi E, Del Testa M, Rossi A, Rossi S. Frequency-dependent tuning of the human motor system induced by transcranial oscillatory potentials. The Journal of Neuroscience. 2011;31:12165–12170. doi: 10.1523/JNEUROSCI.0978-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Fries P. Neuronal gamma-band synchronization as a fundamental process in cortical computation. Annual Review of Neuroscience. 2009;32:209–224. doi: 10.1146/annurev.neuro.051508.135603. [DOI] [PubMed] [Google Scholar]
  27. Headley DB, Weinberger NM. Gamma-band activation predicts both associative memory and cortical plasticity. The Journal of Neuroscience. 2011;31:12748–12758. doi: 10.1523/JNEUROSCI.2528-11.2011. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Johnson NW, Özkan M, Burgess AP, Prokic EJ, Wafford KA, O’Neill MJ, Greenhill SD, Stanford IM, Woodhall GL. Phase-amplitude coupled persistent theta and gamma oscillations in rat primary motor cortex in vitro. Neuropharmacology. 2017;119:141–156. doi: 10.1016/j.neuropharm.2017.04.009. [DOI] [PubMed] [Google Scholar]
  29. Johnson L, Alekseichuk I, Krieg J, Doyle A, Yu Y, Vitek J, Johnson M, Opitz A. Dose-dependent effects of transcranial alternating current stimulation on spike timing in awake nonhuman primates. Science Advances. 2020;6:aaz2747. doi: 10.1126/sciadv.aaz2747. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Kang N, Summers JJ, Cauraugh JH. Transcranial direct current stimulation facilitates motor learning post-stroke: a systematic review and meta-analysis. Journal of Neurology, Neurosurgery & Psychiatry. 2016;87:345–355. doi: 10.1136/jnnp-2015-311242. [DOI] [PubMed] [Google Scholar]
  31. Krakauer JW, Hadjiosif AM, Xu J, Wong AL, Haith AM. Motor Learning. Comprehensive Physiology. 2019;9:613–663. doi: 10.1002/cphy.c170043. [DOI] [PubMed] [Google Scholar]
  32. Krause MR, Vieira PG, Csorba BA, Pilly PK, Pack CC. Transcranial alternating current stimulation entrains single-neuron activity in the primate brain. PNAS. 2019;116:5747–5755. doi: 10.1073/pnas.1815958116. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Lara GA, Alekseichuk I, Turi Z, Lehr A, Antal A, Paulus W. Perturbation of theta-gamma coupling at the temporal lobe hinders verbal declarative memory. Brain Stimulation. 2018;11:509–517. doi: 10.1016/j.brs.2017.12.007. [DOI] [PubMed] [Google Scholar]
  34. Lasztóczi B, Klausberger T. Layer-specific GABAergic control of distinct gamma oscillations in the CA1 hippocampus. Neuron. 2014;81:1126–1139. doi: 10.1016/j.neuron.2014.01.021. [DOI] [PubMed] [Google Scholar]
  35. Lisman JE, Jensen O. The Theta-Gamma Neural Code. Neuron. 2013;77:1002–1016. doi: 10.1016/j.neuron.2013.03.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Lopes-Dos-Santos V, van de Ven GM, Morley A, Trouche S, Campo-Urriza N, Dupret D. Parsing Hippocampal Theta Oscillations by Nested Spectral Components during Spatial Exploration and Memory-Guided Behavior. Neuron. 2018;100:940–952. doi: 10.1016/j.neuron.2018.09.031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Masamizu Y, Tanaka YR, Tanaka YH, Hira R, Ohkubo F, Kitamura K, Isomura Y, Okada T, Matsuzaki M. Two distinct layer-specific dynamics of cortical ensembles during learning of a motor task. Nature Neuroscience. 2014;17:987–994. doi: 10.1038/nn.3739. [DOI] [PubMed] [Google Scholar]
  38. Moisa M, Polania R, Grueschow M, Ruff CC. Brain Network Mechanisms Underlying Motor Enhancement by Transcranial Entrainment of Gamma Oscillations. The Journal of Neuroscience. 2016;36:12053–12065. doi: 10.1523/JNEUROSCI.2044-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Montgomery SM, Buzsáki G. Gamma oscillations dynamically couple hippocampal CA3 and CA1 regions during memory task performance. PNAS. 2007;104:14495–14500. doi: 10.1073/pnas.0701826104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Muellbacher W, Ziemann U, Boroojerdi B, Cohen L, Hallett M. Role of the human motor cortex in rapid motor learning. Experimental Brain Research. 2001;136:431–438. doi: 10.1007/s002210000614. [DOI] [PubMed] [Google Scholar]
  41. Muellbacher W, Ziemann U, Wissel J, Dang N, Kofler M, Facchini S, Boroojerdi B, Poewe W, Hallett M. Early consolidation in human primary motor cortex. Nature. 2002;415:640–644. doi: 10.1038/nature712. [DOI] [PubMed] [Google Scholar]
  42. Muthukumaraswamy SD. Functional properties of human primary motor cortex gamma oscillations. Journal of Neurophysiology. 2010;104:2873–2885. doi: 10.1152/jn.00607.2010. [DOI] [PubMed] [Google Scholar]
  43. Muthukumaraswamy SD. Temporal dynamics of primary motor cortex gamma oscillation amplitude and piper corticomuscular coherence changes during motor control. Experimental Brain Research. 2011;212:623–633. doi: 10.1007/s00221-011-2775-z. [DOI] [PubMed] [Google Scholar]
  44. Ni J, Wunderle T, Lewis CM, Desimone R, Diester I, Fries P. Gamma-Rhythmic Gain Modulation. Neuron. 2016;92:240–251. doi: 10.1016/j.neuron.2016.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Nowak M, Hinson E, van Ede F, Pogosyan A, Guerra A, Quinn A, Brown P, Stagg CJ. Driving Human Motor Cortical Oscillations Leads to Behaviorally Relevant Changes in Local GABAA Inhibition: A tACS-TMS Study. The Journal of Neuroscience. 2017;37:4481–4492. doi: 10.1523/JNEUROSCI.0098-17.2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. O’Keefe J, Recce ML. Phase relationship between hippocampal place units and the EEG theta rhythm. Hippocampus. 1993;3:317–330. doi: 10.1002/hipo.450030307. [DOI] [PubMed] [Google Scholar]
  47. Pfurtscheller G, Lopes da Silva FH. Event-related EEG/MEG synchronization and desynchronization: basic principles. Clinical Neurophysiology. 1999;110:1842–1857. doi: 10.1016/s1388-2457(99)00141-8. [DOI] [PubMed] [Google Scholar]
  48. Pfurtscheller G, Graimann B, Huggins JE, Levine SP, Schuh LA. Spatiotemporal patterns of beta desynchronization and gamma synchronization in corticographic data during self-paced movement. Clinical Neurophysiology. 2003;114:1226–1236. doi: 10.1016/s1388-2457(03)00067-1. [DOI] [PubMed] [Google Scholar]
  49. Reinhart RMG, Nguyen JA. Working memory revived in older adults by synchronizing rhythmic brain circuits. Nature Neuroscience. 2019;22:820–827. doi: 10.1038/s41593-019-0371-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Rogasch NC, Dartnall TJ, Cirillo J, Nordstrom MA, Semmler JG. Corticomotor plasticity and learning of a ballistic thumb training task are diminished in older adults. Journal of Applied Physiology. 2009;107:1874–1883. doi: 10.1152/japplphysiol.00443.2009. [DOI] [PubMed] [Google Scholar]
  51. Rosenkranz K, Kacar A, Rothwell JC. Differential modulation of motor cortical plasticity and excitability in early and late phases of human motor learning. The Journal of Neuroscience. 2007;27:12058–12066. doi: 10.1523/JNEUROSCI.2663-07.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Sanes JN, Donoghue JP. Plasticity and primary motor cortex. Annual Review of Neuroscience. 2000;23:393–415. doi: 10.1146/annurev.neuro.23.1.393. [DOI] [PubMed] [Google Scholar]
  53. Sohal VS, Zhang F, Yizhar O, Deisseroth K. Parvalbumin neurons and gamma rhythms enhance cortical circuit performance. Nature. 2009;459:698–702. doi: 10.1038/nature07991. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Stagg CJ, Best JG, Stephenson MC, O’Shea J, Wylezinska M, Kincses ZT, Morris PG, Matthews PM, Johansen-Berg H. Polarity-sensitive modulation of cortical neurotransmitters by transcranial stimulation. The Journal of Neuroscience. 2009;29:5202–5206. doi: 10.1523/JNEUROSCI.4432-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Stagg CJ, Bachtiar V, Johansen-Berg H. The role of GABA in human motor learning. Current Biology. 2011;21:480–484. doi: 10.1016/j.cub.2011.01.069. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Swann NC, de Hemptinne C, Miocinovic S, Qasim S, Wang SS, Ziman N, Ostrem JL, San Luciano M, Galifianakis NB, Starr PA. Gamma Oscillations in the Hyperkinetic State Detected with Chronic Human Brain Recordings in Parkinson’s Disease. The Journal of Neuroscience. 2016;36:6445–6458. doi: 10.1523/JNEUROSCI.1128-16.2016. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Swann NC, de Hemptinne C, Thompson MC, Miocinovic S, Miller AM, Gilron R, Ostrem JL, Chizeck HJ, Starr PA. Adaptive deep brain stimulation for Parkinson’s disease using motor cortex sensing. Journal of Neural Engineering. 2018;15:046006. doi: 10.1088/1741-2552/aabc9b. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Teo JTH, Swayne OBC, Cheeran B, Greenwood RJ, Rothwell JC. Human Theta Burst Stimulation Enhances Subsequent Motor Learning and Increases Performance Variability. Cerebral Cortex. 2010;21:1627–1638. doi: 10.1093/cercor/bhq231. [DOI] [PubMed] [Google Scholar]
  59. Traub RD, Whittington MA, Stanford IM, Jefferys JG. A mechanism for generation of long-range synchronous fast oscillations in the cortex. Nature. 1996;383:621–624. doi: 10.1038/383621a0. [DOI] [PubMed] [Google Scholar]
  60. Vieira PG, Krause MR, Pack CC, Hanslmayr S. tACS entrains neural activity while somatosensory input is blocked. PLOS Biology. 2020;18:e3000834. doi: 10.1371/journal.pbio.3000834. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Vöröslakos M, Takeuchi Y, Brinyiczki K, Zombori T, Oliva A, Fernández-Ruiz A, Kozák G, Kincses ZT, Iványi B, Buzsáki G, Berényi A. Direct effects of transcranial electric stimulation on brain circuits in rats and humans. Nature Communications. 2018;9:483. doi: 10.1038/s41467-018-02928-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Ward NS, Brander F, Kelly K. Intensive upper limb neurorehabilitation in chronic stroke: outcomes from the Queen Square programme. Journal of Neurology, Neurosurgery, and Psychiatry. 2019;90:498–506. doi: 10.1136/jnnp-2018-319954. [DOI] [PubMed] [Google Scholar]
  63. Watrous AJ. Phase-amplitude coupling supports phase coding in human ECoG. eLife. 2015;4:e07886. doi: 10.7554/eLife.07886. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Whittington MA, Traub RD. Interneuron diversity series: inhibitory interneurons and network oscillations in vitro. Trends in Neurosciences. 2003;26:676–682. doi: 10.1016/j.tins.2003.09.016. [DOI] [PubMed] [Google Scholar]
  65. Whittington MA, Cunningham MO, LeBeau FEN, Racca C, Traub RD. Multiple origins of the cortical gamma rhythm. Developmental Neurobiology. 2011;71:92–106. doi: 10.1002/dneu.20814. [DOI] [PubMed] [Google Scholar]
  66. Yamamoto J, Suh J, Takeuchi D, Tonegawa S. Successful execution of working memory linked to synchronized high-frequency gamma oscillations. Cell. 2014;157:845–857. doi: 10.1016/j.cell.2014.04.009. [DOI] [PubMed] [Google Scholar]
  67. Yarrow K, Brown P, Krakauer JW. Inside the brain of an elite athlete: the neural processes that support high achievement in sports. Nature Reviews. Neuroscience. 2009;10:585–596. doi: 10.1038/nrn2672. [DOI] [PubMed] [Google Scholar]
  68. Zaehle T, Rach S, Herrmann CS. Transcranial alternating current stimulation enhances individual alpha activity in human EEG. PLOS ONE. 2010;5:e13766. doi: 10.1371/journal.pone.0013766. [DOI] [PMC free article] [PubMed] [Google Scholar]

Editor's evaluation

Thorsten Kahnt 1

This study provides evidence that increasing theta-gamma phase-amplitude coupling, which is thought to be critical for hippocampal memory, can improve non-hippocampal motor learning. This conclusion is based on two experiments showing that transcranial alternating current stimulation over M1 improves motor learning relative to sham stimulation and an active control condition. The findings will be interesting to neuroscientists and clinicians, as they elucidate mechanisms of motor learning and have implications for improving outcomes for patients recovering from motor impairments.

Decision letter

Editor: Thorsten Kahnt1
Reviewed by: Michael A Nitsche

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Decision letter after peer review:

Thank you for submitting your article "Increasing human motor skill acquisition by driving theta-gamma coupling" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Richard Ivry as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Michael A Nitsche (Reviewer #3).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1. Reviewers agreed that the TMS experiment (Figure 3) does not provide sufficient evidence to draw strong conclusions about the mechanisms by which theta-gamma tACS enhances motor skill acquisition. Reviewers agreed that the sample (N=10) is too small, the effect too weak, and the interaction effect difficult to interpret. Our preference would be to have you repeat this experiment in a larger sample. However, we also felt that there was enough meat in the paper that this should not be required as part of revision. If you are in a position to run a better powered TMS experiment, great-we certainly think it would be a nice addition to the paper. Otherwise, we ask that you discuss this limitation and tone down the conclusions about mechanisms in a revision.

2. Reviewers would also like to see additional analyses of these TMS data. Specifically, please combine blocks 1-3 into "stimulation blocks" and 4-7 as "non-stimulation blocks" and run the ANOVA this way. A significant interaction effect should be followed up by post-hoc tests, comparing TGP vs sham and TGP vs TGT in stimulation and non-stimulation blocks.

Reviewer #1:

This manuscript reports the results from two between-subject experiments testing the role of theta-gamma phase amplitude coupling (PAC) for non-hippocampal motor skill learning. Both experiments use transcranial alternating current stimulation (tACS) over right M1. Experiment 1 (single-blind, N=58) tests three groups with theta-gamma stimulation, where gamma stimulation is delivered either at the peak (TGP) or the through of the theta envelope (TGT), or sham. The results show that TGP relative to sham and TGT increases the acceleration of left thumb abduction. This basic effect is replicated in a second (preregistered, double-blind) experiment (N=46), which further shows that these effects last for at least 1 hour after stimulation offset. Additional analyses show that results are not driven by effects on variability or response times. These experiments are rigorous and the results provide convincing support for the idea that theta-gamma PAC during the peak of the theta rhythm facilitates non-hippocampal skill acquisition.

An additional within-subject experiment (N=10) tests whether this effect is driven by increased cortical excitability. This is assessed using motor evoked potentials (MEP), measured using TMS. The authors report a significant stimulation by block interaction, but no main effect of stimulation. These MEP effects are less convincing for several reasons. First, the experiment is potentially underpowered and the effects are weak. Second, the stimulation/condition by time interaction is difficult to interpret without a main effect of condition at the time of stimulation.

In summary, the study is rigorous and the results are novel and important. They convincingly show that theta-gamma tACS improves non-hippocampal skill learning. These findings are important in two ways. First, they show that theta-gamma PAC plays a role beyond hippocampal-dependent learning. Second, they may have implications for therapy and treatment. However, the results from the TMS experiment are less convincing, leaving it open whether increased cortical excitability is the mechanism by which theta-gamma tACS facilitates motor skill acquisition.

Comments for the authors:

1. My only major concern is related to the TMS experiment. The interaction effect is not very robust and I am not sure the interaction is even the right effect to look at. The interaction appears to be driven just as much by a higher MEP amplitude in the TGP condition during stimulation that decreases afterwards, as it is driven by a low MEP amplitude in the sham condition that increases after stimulation offset. Is there even a difference between conditions during the stimulation period? Also with N=10 this experiment is likely underpowered. I don't think these results are conclusive enough to inform the mechanism by which the stimulation works. My recommendation is to either collect a larger sample or to remove these results and pertaining discussions entirely.

2. Experiment 1: The authors show a difference between TGP and sham, but is there also a difference between TGP and TGT?

3. Is the p value in line 116 is a typo? Should it be p=0.041 instead of p=0.41?

Reviewer #2:

Scientifically, this is a very nice piece of work. The active control is well-chosen to emphasize the specific importance of the theta phase of stimulation. The pre-registered replication improves confidence in the reliability of the effect. Finally, showing this stimulation increases cortical excitability offers a potential mechanism underlying these effects.

I believe the authors could do a better job framing the paper, for which I have some suggestions below. I also point out several small issues with the writing, which should be fixed upon any revision.

In general, the discussion of the hippocampus is too prominent. For instance, it is mentioned in the second sentence of the abstract, and it is a bit jarring coming to the third sentence and subsequently finishing the abstract to realize that the paper has nothing to do with the hippocampus but rather leans on that literature for motivation (which is fine!). For instance, the authors could change the abstract to the following: "Some learning paradigms are closely associated with gamma activity that is amplitude-modulated by the phase of underlying theta activity, but whether such nested activity patterns also underpin skill acquisition is unknown." I also believe the emphasis from lines 51-63 should be on learning and theta-gamma coupling first, with references to research on the hippocampus coming as an ancillary point rather than centering the hippocampus. Such a shift would be subtle, but I believe it would be effective here. Additionally, some aspects of the framing, such as mentioning subregion CA1, are too granular for the motor study here.

Figure 3: What do the authors make of the reduction in the TGP condition after the stimulation? Is this significant on its own? If so, the authors should speculate about whether this might reflect some sort of "fatigue" or "refractory" effect following the enhanced stimulation.

It seems appropriate to describe something about the motor task used in the Introduction. I am familiar with many motor tasks, but not this one (until this paper), so the authors may improve the paper by describing why this one was used here (rather than others).

Reviewer #3:

In this contribution, the authors explored the importance of theta-gamma coupling, an oscillatory brain activity pattern shown to be relevant for hippocampus-dependent learning, for motor skill acquisition, which does critically involve neocortical areas, but not the hippocampus, and thus the general importance of this oscillatory brain activity for learning and memory formation, in healthy humans. The main results of the study show that non-invasive brain stimulation with transcranial alternating currents (tACS) over the sensorimotor cortex indeed improved motor skill learning, and this effect lasted relevantly beyond stimulation. This thus establishes a new mechanistic foundation for learning and memory formation in humans, which might have future applications in rehabilitation. One relevant limitation of the study is that the respective theta gamma protocol also enhanced motor cortex excitability, and the experiment does not allow to conclude if the effects depend critically on the excitability enhancement, which might also be induced by other stimulation protocols not involving theta gamma coupling.

Strengths

1. Innovative concept exploring the relevance of specific brain oscillation patterns for learning and memory formation beyond hippocampal learning.

2. Causal test of the relevance of theta-gamma coupling for motor skill acquisition by non-invasive brain stimulation.

3. The introduction gives a comprehensive overview about the contribution of theta-gamma activity to various cognitive and behavioral functions.

4. The experimental design does involve not only behavioral tests, but also physiological exploration of tACS-induced excitability alterations via TMS-generated MEPs, which helps to clarify physiological mechanisms further.

5. The experimental design explores the specificity of the effects by coupling theta and gamma activity in different ways.

6. The results of the main study were replicated in a second experiment, which increases reliability.

7. Important control measures were conducted for variability, and latency of responses, which helps to clarify specificity of the results.

8. A mechanistic explanation for the effect of theta gamma coupling on motor skill acquisition in given.

9. In the discussion, limitations of the study are clearly mentioned.

Weaknesses

1. In the introduction, the contribution of theta gamma coupling to working memory performance is not mentioned, which is a pity, because here theta gamma tACS over prefrontal areas was shown to alter performance. These results show thus the functional relevance of these oscillations in neocortical areas for cognitive functions.

2. The non-specialized reader is not introduced into mechanisms of action of tACS.

3. It does not become clear if successful blinding was tested in experiment 1.

4. For the TMS measures, the statistics do not allow to identify the specific timing-dependency of differences between the TGP and sham conditions.

5. It would have been good to obtain physiological and behavioral measures in the same group of subjects. This would have allowed to correlate results, and thus make stronger conclusions about the interdependency of excitability, and behavioral alterations.

6. Since only the TGP protocol improved learning, but this protocol also enhanced excitability, it cannot be excluded that not the oscillatory entrainment, but the excitability alteration caused behavioral improvement, which might mean that the oscillations themselves are not critical for the effects, as far as I can see.

7. In the discussion, it does not become quite clear why gamma activity specifically at the theta peak, but not at the trough, should reduce GABA activity.

8. For application of tACS, by the experimental design it cannot be completely excluded that the Pz electrode contributed to the effects, although this is unlikely.

Comments for the authors:

1. It might be good to add information in the introduction about the relevance of theta gamma coupling, also in connection with tACS, for working memory performance in humans.

2. It might be good to add a para to the introduction which explains tACS for non-specialized readers.

3. Please add the blinding index, if available, also for experiment 1.

4. For the TMS measures, it might make sense to add post hoc tests comparing at least TGP, and sham conditions to explore the timing-dependency of the effects.

5. The authors might want to add information, why gamma activity specifically at the theta peak, but not at the trough, should reduce GABA activity, if such information is available.

6. The authors might want to discuss a possible contribution of the parietal electrode to the effects, or exclude this by a control experiment.

7. The authors might want to discuss if it could be the case that not the oscillations, but the excitability enhancement could have been critical for the induced effects. One option to test this would have been a control condition with other stimulation protocols not including oscillations tested here.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "Increasing human motor skill acquisition by driving theta-gamma coupling" for further consideration by eLife. Your revised article has been reviewed by 3 peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Richard Ivry as the Senior Editor.

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

All reviewers agreed that you have adequately addressed several of the essential issues. However, there were lingering concerns regarding the results from the TMS experiment and whether they provide evidence for potential mechanisms of tACS.

Specifically, the additional analyses reveal that there was no significant effect of stimulation condition during the stimulation period and that the condition x block interaction was driven by an effect of TGT vs. sham after the stimulation. Furthermore, there was no evidence for differences between TGP and either sham or TGT, either during or after the stimulation. It is unclear whether this is a true null result or whether there are no effects because of insufficient power. In any case, given this experiment does not provide clear evidence for an effect of the critical tACS protocol (i.e., TGP), reviewers agreed that these data do not help to inform potential mechanisms of tACS.

Unless the situation in your lab has changed such that you are now able to collect additional TMS data to run a well-powered TMS experiment, we'd ask you to remove the TMS experiment and any reference to it from the manuscript.

Reviewer #1:

The authors have been responsive to several of my initial concerns, but my major point about the TMS experiment was not sufficiently addressed.

The authors re-analyzed these data and found no effect of TGP or TGT during the stimulation, but reduced MEPs after TGT stimulation compared to sham and TGP. Importantly, there was no difference between TGP and sham during or after the stimulation. It is therefore unclear why the authors conclude that they found "a pattern of enhanced cortical excitability during TGP stimulation compared to sham and TGT stimulation." As far as I can see, there was no evidence for such a pattern during (not reported) or after the stimulation (p>0.14). So don't think the author's conclusion that "these TMS data provide valuable initial support for phase-entrainment as a potential mechanism of action for tACS" is supported by the data.

Reviewer #2:

I commend the authors for performing a nice revision and again for submitting an interesting paper. They have done an especially good job at this stage by offering more measured interpretations. I recommend publication.

PS. There are still a few instances of beginning sentences with numerals. Please fix these.

Reviewer #3:

All major issues were solved appropriately.

eLife. 2021 Nov 23;10:e67355. doi: 10.7554/eLife.67355.sa2

Author response

Essential revisions:

1. Reviewers agreed that the TMS experiment (Figure 3) does not provide sufficient evidence to draw strong conclusions about the mechanisms by which theta-gamma tACS enhances motor skill acquisition. Reviewers agreed that the sample (N=10) is too small, the effect too weak, and the interaction effect difficult to interpret. Our preference would be to have you repeat this experiment in a larger sample. However, we also felt that there was enough meat in the paper that this should not be required as part of revision. If you are in a position to run a better powered TMS experiment, great-we certainly think it would be a nice addition to the paper. Otherwise, we ask that you discuss this limitation and tone down the conclusions about mechanisms in a revision.

We thank the editor and reviewers for allowing us to review and clarify our TMS findings. We would agree with the reviewers that the results of the TMS study are limited by the small sample size and therefore provide inconclusive evidence regarding mechanism of action of tACS. Unfortunately, in the current circumstances we are unable to collect more TMS data and are grateful for the editor’s understanding of this. However, despite the small sample, we strongly believe that these findings provide valuable preliminary evidence that is worth highlighting for future studies.

In line with reviewer’s suggestions, we have made substantial changes to the manuscript to fully acknowledge this limitation and the preliminary nature of the TMS findings (see reviewer responses below for more detail). We have toned down the conclusions about the mechanisms in the revision. We have additionally run new statistical analyses of the TMS data, which we hope the reviewers and editor agree have made the results clearer to interpret (see the responses to individual reviewers below for more detail).

2. Reviewers would also like to see additional analyses of these TMS data. Specifically, please combine blocks 1-3 into "stimulation blocks" and 4-7 as "non-stimulation blocks" and run the ANOVA this way. A significant interaction effect should be followed up by post-hoc tests, comparing TGP vs sham and TGP vs TGT in stimulation and non-stimulation blocks.

We thank the reviewers for this recommendation, which helps clarify the TMS data. In the revised manuscript, we now combine the data into ‘stimulation blocks’ and ‘post-stimulation blocks’ and run a RM ANOVA followed by post-hoc t-tests examining the interaction effect (see reviewer responses below for more detail).

Reviewer #1:

This manuscript reports the results from two between-subject experiments testing the role of theta-gamma phase amplitude coupling (PAC) for non-hippocampal motor skill learning. Both experiments use transcranial alternating current stimulation (tACS) over right M1. Experiment 1 (single-blind, N=58) tests three groups with theta-gamma stimulation, where gamma stimulation is delivered either at the peak (TGP) or the through of the theta envelope (TGT), or sham. The results show that TGP relative to sham and TGT increases the acceleration of left thumb abduction. This basic effect is replicated in a second (preregistered, double-blind) experiment (N=46), which further shows that these effects last for at least 1 hour after stimulation offset. Additional analyses show that results are not driven by effects on variability or response times. These experiments are rigorous and the results provide convincing support for the idea that theta-gamma PAC during the peak of the theta rhythm facilitates non-hippocampal skill acquisition.

An additional within-subject experiment (N=10) tests whether this effect is driven by increased cortical excitability. This is assessed using motor evoked potentials (MEP), measured using TMS. The authors report a significant stimulation by block interaction, but no main effect of stimulation. These MEP effects are less convincing for several reasons. First, the experiment is potentially underpowered and the effects are weak. Second, the stimulation/condition by time interaction is difficult to interpret without a main effect of condition at the time of stimulation.

In summary, the study is rigorous and the results are novel and important. They convincingly show that theta-gamma tACS improves non-hippocampal skill learning. These findings are important in two ways. First, they show that theta-gamma PAC plays a role beyond hippocampal-dependent learning. Second, they may have implications for therapy and treatment. However, the results from the TMS experiment are less convincing, leaving it open whether increased cortical excitability is the mechanism by which theta-gamma tACS facilitates motor skill acquisition.

Thank you. We are pleased to see that the reviewer appreciates the important implications of theta-gamma coupling in non-hippocampal skill learning. In what follows, we address the reviewer’s concerns, particularly with regards to the TMS experiment where we have made substantial effort to clarify findings and address limitations.

Comments for the authors:

1. My only major concern is related to the TMS experiment. The interaction effect is not very robust and I am not sure the interaction is even the right effect to look at. The interaction appears to be driven just as much by a higher MEP amplitude in the TGP condition during stimulation that decreases afterwards, as it is driven by a low MEP amplitude in the sham condition that increases after stimulation offset. Is there even a difference between conditions during the stimulation period? Also with N=10 this experiment is likely underpowered. I don't think these results are conclusive enough to inform the mechanism by which the stimulation works. My recommendation is to either collect a larger sample or to remove these results and pertaining discussions entirely.

We thank the reviewer for this comment, which was highlighted in the editor’s summary above. Unfortunately given the current circumstances we are not able to acquire more TMS data. We agree that in this limited sample size, the mechanism of action of theta-gamma tACS remains inconclusive. However, we strongly believe that these preliminary findings offer valuable insight into a potential mechanism of action that compliments recent findings from single-unit recordings in non-human primates that is worth exploring in future studies. We have substantially revised our manuscript to clearly address the limitations of our TMS findings. In addition we have run new analyses to clarify the interaction effect presented. We hope that these revisions clarify the findings from, and limitations of, the TMS data, and present it in an appropriate context.

As per reviewers’ recommendations, we combined the TMS data into ‘stimulation blocks’ and ‘post-stimulation blocks’ and ran a RM ANOVA followed by post-hoc t-tests examining the interaction effect i.e. the timing-dependency of the effect. These changes are included in the revised Results section as follows (lines 206-216):

‘MEPs were normalised to baseline and we performed a repeated-measures ANOVA, with one factor of condition (TGP, TGT and Sham) and one factor of block (stimulation blocks and post-stimulation). […] However, MEPs were significantly reduced after TGP stimulation compared with during TGP stimulation (t(9)=2.861, p=0.019), which might reflect a refractory period following enhanced cortical excitability, though this remains speculative.’

We further interpret these TMS results and address the limitation of our small sample size in the revised discussion (Lines 261-271):

‘To explore changes in cortical activity in response to tACS, we quantified changes in TMS-evoked MEPs during theta-gamma peak, trough and sham stimulation. […] However, these TMS data provide valuable initial support for phase-entrainment as a potential mechanism of action for tACS and more generally, for TMS-tACS protocols as an effective tool to explore the phase-specific pattern of neural responses in the human cortex.’

2. Experiment 1: The authors show a difference between TGP and sham, but is there also a difference between TGP and TGT?

We apologise for not including this. There was no significant difference between TGP and TGT. We have now clarified the Results section as follows (lines 140-144):

‘Post-hoc tests (using Tukey correction for multiple comparisons) revealed a significant difference between TGP and sham (p=0.04) and no significant difference between TGT and sham (p=0.766) or TGP and TGT (p=0.162). To further explore the interaction effect, we ran an analysis of simple effects to determine the effect of Condition (TGP, TGT, sham) at each level of Block (1-6).’

3. Is the p value in line 116 is a typo? Should it be p=0.041 instead of p=0.41?

We thank the Reviewer for pointing out this typo. We have now corrected this in the Results section (lines 139):

‘…Effect of Condition F(2, 55)=3.396 , p=0.41 p=0.041…’

Reviewer #2:

Scientifically, this is a very nice piece of work. The active control is well-chosen to emphasize the specific importance of the theta phase of stimulation. The pre-registered replication improves confidence in the reliability of the effect. Finally, showing this stimulation increases cortical excitability offers a potential mechanism underlying these effects.

I believe the authors could do a better job framing the paper, for which I have some suggestions below. I also point out several small issues with the writing, which should be fixed upon any revision.

In general, the discussion of the hippocampus is too prominent. For instance, it is mentioned in the second sentence of the abstract, and it is a bit jarring coming to the third sentence and subsequently finishing the abstract to realize that the paper has nothing to do with the hippocampus but rather leans on that literature for motivation (which is fine!). For instance, the authors could change the abstract to the following: "Some learning paradigms are closely associated with gamma activity that is amplitude-modulated by the phase of underlying theta activity, but whether such nested activity patterns also underpin skill acquisition is unknown." I also believe the emphasis from lines 51-63 should be on learning and theta-gamma coupling first, with references to research on the hippocampus coming as an ancillary point rather than centering the hippocampus. Such a shift would be subtle, but I believe it would be effective here. Additionally, some aspects of the framing, such as mentioning subregion CA1, are too granular for the motor study here.

We thank the reviewer for giving us the opportunity to re-frame the manuscript and clarify the motivation of the work. We have revised our discussion of the hippocampus in the abstract and introduction. We hope this helps guide the reader more clearly through the work.

We have made the following changes to the abstract (lines 31-36):

‘Hippocampal learning is closely associated with gamma activity, which is amplitude-modulated by the phase of underlying theta activity. […] Some learning paradigms, particularly in the memory domain, are closely associated with gamma activity that is amplitude-modulated by the phase of underlying theta activity, but whether such nested activity patterns also underpin skill learning is unknown.’

In the introduction, we added the following to emphasise the importance of theta-gamma coupling outside the hippocampus (lines 69-74):

‘In the pre-frontal cortex, externally-driven θ-γ PAC directly influences spatial working memory performance and global neocortical connectivity when gamma oscillations are delivered coinciding with the peak, but not the trough of theta waves [35]. It is proposed that the theta rhythm forms a temporal structure that organizes gamma-encoded units into preferred phases of the theta cycle, allowing careful processing and transmission of neural computations [Watrous et al., 2015].’

We have removed mention of subregion CA1 (lines 56-59):

‘In rodent hippocampal area CA1, oscillations in the θ (5-12 Hz 4-8 Hz) band become dominant during active exploration [9], and have been hypothesised to allow information coming into CA1 from distant regions to be divided into discrete units for processing [10,11].’

With regards to the order, we believe the introduction flows most logically from the hippocampus first – where the fundamentals of cross-frequency coupling and its role in learning were first established – followed by more recent work identifying cross-frequency coupling in the neocortex (although no work has been performed to date on the role of theta-gamma coupling in skill learning in these regions). We believe that this framing helps establish a clear motivation to explore whether theta-gamma coupling in the neocortex plays a similar role in non-hippocampal learning as it does in the hippocampus. We hope the changes in the amended manuscript have helped clarify the motivation of the work, and have removed the concentration on the hippocampus, which we agree did not aid the relevant framing of the paper.

Figure 3: What do the authors make of the reduction in the TGP condition after the stimulation? Is this significant on its own? If so, the authors should speculate about whether this might reflect some sort of "fatigue" or "refractory" effect following the enhanced stimulation.

We thank the reviewer for making this observation. We have added the following to address this point (lines 214-216):

‘However, MEPs were significantly reduced after TGP stimulation compared with during TGP stimulation (t(9)=2.861, p=0.019), which might reflect a refractory period following enhanced cortical excitability, though this remains speculative.’

It seems appropriate to describe something about the motor task used in the Introduction. I am familiar with many motor tasks, but not this one (until this paper), so the authors may improve the paper by describing why this one was used here (rather than others).

We thank the reviewer for allowing us to clarify this important point. We have added the following to the introduction (lines 88-92):

‘We chose this task because it shows robust behavioural improvement in a relatively short period of time and performance improvement is underpinned by plastic changes in M1 [Classen et al., 1998; Muellbacher et al. 2001; Muellbacher et al. 2002]. This encoding of kinematic details of the practiced movement is commonly regarded as a first step in skill acquisition [Classen et al., 1998].’

Reviewer #3:

[…] 1. It might be good to add information in the introduction about the relevance of theta gamma coupling, also in connection with tACS, for working memory performance in humans.

We have added the following to address this point (lines 69-74):

‘In the pre-frontal cortex, externally-driven θ-γ PAC directly influences spatial working memory performance and global neocortical connectivity when gamma oscillations are delivered coinciding with the peak, but not the trough of theta waves [35]. It is proposed that the theta rhythm forms a temporal structure that organizes gamma-encoded units into preferred phases of the theta cycle, allowing careful processing and transmission of neural computations [Watrous et al., 2015].’

2. It might be good to add a para to the introduction which explains tACS for non-specialized readers.

Thank you for this important point – we have added the following to the manuscript (lines 83-87):

‘…we modulated local theta-gamma activity via externally applied alternating current stimulation (tACS), a non-invasive form of brain stimulation that can interact with and modulate neural oscillatory activity in the human brain in a frequency-specific manner [Ali et al., 2013; Feurra et al., 2011; Zaehle et al., 2010], over M1 during learning of an M1-dependent ballistic thumb abduction task skill [32].’

3. Please add the blinding index, if available, also for experiment 1.

Unfortunately, a blinding index is not available for experiment 1 as blinding data was not collected.

4. For the TMS measures, it might make sense to add post hoc tests comparing at least TGP, and sham conditions to explore the timing-dependency of the effects.

We thank the reviewer for this suggestion, which has helped clarify the TMS results. We have made substantial changes to the analyses of the TMS data. As per reviewers’ recommendations, we combined the data into ‘stimulation blocks’ and ‘post-stimulation blocks’ and ran a RM ANOVA followed by post-hoc t-tests examining the interaction effect. These changes are revised in the Results section as follows (lines 206-216):

‘MEPs were normalised to baseline and we performed a repeated-measures ANOVA, with one factor of condition (TGP, TGT and Sham) and one factor of block (stimulation blocks and post-stimulation). […] However, MEPs were significantly reduced after TGP stimulation compared with during TGP stimulation (t(9)=2.861, p=0.019), which might reflect a refractory period following enhanced cortical excitability, though this remains speculative.’

5. The authors might want to add information, why gamma activity specifically at the theta peak, but not at the trough, should reduce GABA activity, if such information is available.

We have added the following to address this point (lines 279-285):

‘The effects of low frequency tACS may be mediated through cyclically inducing a phase of enhanced excitation (peak) followed by a phase of reduced excitation (trough). […] This hypothesis would be in line with the increase in cortical excitability we observed during TGP, but not TGT stimulation, and the tACS peak phase-preference demonstrated in single-unit recording studies [Johnson et al., 2020]’

6. The authors might want to discuss a possible contribution of the parietal electrode to the effects, or exclude this by a control experiment.

Thank you for this important point. We agree with the reviewer that there might be a possible contribution of the parietal electrode. However, we deliberately chose an M1-dependent task to minimise the potential confounds. We have amended the discussion to include a greater comment on the potential contribution of other nodes (lines 304-309):

‘We are confident that we are actively stimulating M1: our tACS protocol induces excitability changes in M1, suggesting a significant physiological effect in this region, and the electrical field simulation demonstrates a significant current within M1 due to tACS. […] This hypothesis remains to be tested.’

7. The authors might want to discuss if it could be the case that not the oscillations, but the excitability enhancement could have been critical for the induced effects. One option to test this would have been a control condition with other stimulation protocols not including oscillations tested here.

We thank the reviewer for allowing us to expand on this important point. It would seem parsimonious to suggest an explanation where the oscillations led to an increase in excitability, which led to the behavioural effects. Indeed, it seems likely that this is the mechanism by which the oscillations exert their behavioural effects. We have added the following to the Discussion section to clarify this line of argument (lines: 279-285):

‘The effects of low frequency tACS may be mediated through cyclically inducing a phase of enhanced excitation (peak) followed by a phase of reduced excitation (trough). […] This hypothesis would be in line with the increase in cortical excitability we observed during TGP, but not TGT stimulation, and the tACS peak phase-preference demonstrated in single-unit recording studies [Johnson et al., 2020]’

References:

Ali MM, Sellers KK and Fröhlich F (2013). Transcranial alternating current stimulation modulates large-scale cortical network activity by network resonance. Journal of Neuroscience 33(27), 11262-11275.

Classen J, Liepert J, Wise SP, Hallett M and Cohen LG (1998). Rapid plasticity of human cortical movement representation induced by practice. Journal of Neurophysiology 79(2):1117-23.

Feurra M, Paulus W, Walsh V and Kanai R (2011). Frequency specific modulation of human somatosensory cortex. Frontiers in Psychology 2, 13.

Johnson L, Aleksiechuck I, Krieg J, et al. (2020). Dose-dependent effects of transcranial alternating current stimulation on spike timing in awake nonhuman primates. Science Advances 6: eaaz2747.

Muellbacher W, Ziemann U, Boroojerdi B, Cohen L and Hallett M (2001). Role of the human motor cortex in rapid motor learning. Experimental Brain Research 136(4): 431-8.

Muellbacher W, Ziemann U, Wissel J, Dang N, Kofler M, Facchini S, Boroojerdi B, Poewe W and Hallett M (2002). Early consolidation in human primary motor cortex. Nature 415(6872): 640-4.

Watrous AJ, Deuker L, Fell J and Axmacher N (2015). Phase-amplitude coupling supports phase coding in human ECoG. eLife, 4 e07886

Zaehle T, Rach S, and Herrmann CS (2010). Transcranial alternating current stimulation enhances individual α activity in human EEG. PLoS One, 5(11), e13766.

[Editors' note: further revisions were suggested prior to acceptance, as described below.]

The manuscript has been improved but there are some remaining issues that need to be addressed, as outlined below:

All reviewers agreed that you have adequately addressed several of the essential issues. However, there were lingering concerns regarding the results from the TMS experiment and whether they provide evidence for potential mechanisms of tACS.

Specifically, the additional analyses reveal that there was no significant effect of stimulation condition during the stimulation period and that the condition x block interaction was driven by an effect of TGT vs. sham after the stimulation. Furthermore, there was no evidence for differences between TGP and either sham or TGT, either during or after the stimulation. It is unclear whether this is a true null result or whether there are no effects because of insufficient power. In any case, given this experiment does not provide clear evidence for an effect of the critical tACS protocol (i.e., TGP), reviewers agreed that these data do not help to inform potential mechanisms of tACS.

Unless the situation in your lab has changed such that you are now able to collect additional TMS data to run a well-powered TMS experiment, we'd ask you to remove the TMS experiment and any reference to it from the manuscript.

We are very pleased that the Reviewers felt that we had addressed the substantial majority of their comments. In response to the main comment remaining from Reviewer 1, and highlighted by the Editor, we have removed all mention of the TMS experiment from the manuscript. We have responded to all the Reviewers’ comments point-by-point below.

Reviewer #1:

The authors have been responsive to several of my initial concerns, but my major point about the TMS experiment was not sufficiently addressed.

The authors re-analyzed these data and found no effect of TGP or TGT during the stimulation, but reduced MEPs after TGT stimulation compared to sham and TGP. Importantly, there was no difference between TGP and sham during or after the stimulation. It is therefore unclear why the authors conclude that they found "a pattern of enhanced cortical excitability during TGP stimulation compared to sham and TGT stimulation." As far as I can see, there was no evidence for such a pattern during (not reported) or after the stimulation (p>0.14). So don't think the author's conclusion that "these TMS data provide valuable initial support for phase-entrainment as a potential mechanism of action for tACS" is supported by the data.

We understand the Reviewer’s point and have removed the TMS experiment from the manuscript as the Editor requested.

Reviewer #2:

I commend the authors for performing a nice revision and again for submitting an interesting paper. They have done an especially good job at this stage by offering more measured interpretations. I recommend publication.

PS. There are still a few instances of beginning sentences with numerals. Please fix these.

We apologise that we had missed these: the remaining instances were all in the TMS sections and have therefore now been removed.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Akkad H, Dupont-Hadwen J, Bestmann S, Stagg CJ. 2018. Improving motor learning via phase-amplitude coupled theta-gamma tACS. Open Science Framework. [DOI]

    Supplementary Materials

    Transparent reporting form
    elife-67355-transrepform1.docx (111.5KB, docx)

    Data Availability Statement

    All data generated or analysed during this study are included in the manuscript and freely available on the open science framework (https://osf.io/xjpef). Details of data analysis, experimental design and protocol were pre-registered prior to data collection and freely available on the open science framework - Registration form: osf.io/xjpef; Files: osf.io/452f8/files/.

    The following previously published datasets were used:

    Akkad H, Dupont-Hadwen J, Bestmann S, Stagg CJ. 2018. Improving motor learning via phase-amplitude coupled theta-gamma tACS. Open Science Framework.

    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES