Abstract

Gamma-aminobutyric acid (GABA) inhibition shapes motor cortex output, gates synaptic plasticity in the form of long-term potentiation, and plays an important role in motor learning. Remarkably, recent studies have shown that acute cardiovascular exercise can improve motor memory, but the cortical mechanisms are not completely understood. We investigated whether an acute bout of lower-limb high-intensity interval (HIT) exercise could promote motor memory formation in humans through changes in cortical inhibition within the hand region of the primary motor cortex. We used TMS to assess the input–output relationship, along with inhibition involving GABAA and GABAB receptors. Measures were obtained before and after a 20-min session of HIT cycling (exercise group) or rest (control group). We then had the same participants learn a new visuomotor skill and perform a retention test 5 hr later in the absence of sleep. No differences were found in corticomotor excitability or GABAB inhibition; however, synaptic GABAA inhibition was significantly reduced for the exercise group but not the control group. HIT exercise was found to enhance motor skill consolidation. These findings link modification of GABA to improved motor memory consolidation after HIT exercise and suggest that the beneficial effects of exercise on consolidation might not be dependent on sleep.

INTRODUCTION

Motor skill learning is the process by which movement becomes effortless, faster, and more accurate through repeated practice. This time-dependent process involves online skill acquisition during task performance and offline motor memory consolidation, that is, skill stabilization or improvement that occurs while not actively practicing (Dayan & Cohen, 2011; Robertson, Press, & Pascual-Leone, 2005). A brain region critical for both aspects of motor learning is the primary motor cortex (M1) (Muellbacher et al., 2002). Plasticity within M1, including the modulation of gamma-aminobutyric acid (GABA) inhibition, is essential for motor learning (Chen, Kim, Peters, & Komiyama, 2015; Coxon, Peat, & Byblow, 2014; Dayan & Cohen, 2011; Stagg, Bachtiar, & Johansen-Berg, 2011; Robertson et al., 2005; Classen & Cohen, 2003; Muellbacher et al., 2002; Butefisch et al., 2000; Rioult-Pedotti, Friedman, Hess, & Donoghue, 1998). At the level of the M1 synapse, a reduction of GABA inhibition is necessary for long-term potentiation (LTP) to occur (Rioult-Pedotti et al., 1998; Hess, Aizenman, & Donoghue, 1996) and for learning-induced reorganization of dendritic spines (Chen et al., 2015), which suggests a role for GABA in regulating the early consolidation of motor memory.

One way to enhance learning is through noninvasive brain stimulation. For example, anodal transcranial direct current stimulation (a-tDCS) applied to M1 during motor practice has been shown to enhance learning of a sequential visual isometric pinch task (SVIPT) through a time- but not sleep-dependent, offline consolidation effect (Reis et al., 2009, 2015). a-tDCS promotes synaptic plasticity that requires synthesis of brain-derived neurotrophic factor (BDNF; Fritsch et al., 2010) and, in conjunction with a reduction of GABAergic inhibition after stimulation (Stagg, Bachtiar, et al., 2011; Stagg, Bestmann, et al., 2011; Stagg et al., 2009), is likely to contribute to the enhancement of motor memory (Reis et al., 2009). Interestingly, similar effects have been observed after cardiovascular exercise; however, the cortical mechanisms remain unclear. It is possible that exercise similarly promotes brain plasticity via a reduction of GABAergic inhibition.

There is some evidence that acute cardiovascular exercise facilitates motor learning (Mang, Snow, Campbell, Ross, & Boyd, 2014; Roig, Skriver, Lundbye-Jensen, Kiens, & Nielsen, 2012). For example, Roig et al. (2012) found that exercise improved the retention of a fine motor skill for up to 7 days yet had no effect on the rate of motor skill acquisition. The neural mechanisms were not investigated in this study. Other purely physiological studies have used TMS of the nonexercised M1 hand region to show that exercise causes a transient reduction of inhibition (Singh, Duncan, Neva, & Staines, 2014; Smith, Goldsworthy, Garside, Wood, & Ridding, 2014). The effect of exercise on inhibition may be one mechanism underlying the enhanced ability to experimentally induce plasticity subsequent to exercise (Mang et al., 2014; Singh, Neva, & Staines, 2014; McDonnell, Orekhov, & Ziemann, 2006), but this remains to be demonstrated. Overall, these results suggest that the cardiovascular challenge of exercise, and/or the associated movement of the lower limbs, has a generalized effect on M1 that extends beyond the representations of the exercised limbs and that exercise may boost the capacity for plasticity associated with motor learning.

The studies reporting that acute exercise facilitates the acquisition and retention of novel motor skills involved high-intensity interval (HIT) exercise (Mang et al., 2014; Roig et al., 2012). In contrast, TMS investigations of M1 plasticity have generally involved lower-intensity exercise. This is important because moderate-intensity and HIT exercises have very different physiological effects, with the latter associated with higher levels of lactate and BDNF in the circulation (Saucedo Marquez, Vanaudenaerde, Troosters, & Wenderoth, 2015; Skriver et al., 2014). We investigated the effect of lower-limb HIT exercise on inhibition in the hand representation of M1 and, in the same participants, their subsequent ability to learn and consolidate a fine motor skill. We anticipated that HIT exercise would improve motor learning compared with a no-exercise control group, through an effect on offline consolidation (Mang et al., 2014; Roig et al., 2012). Our novel hypothesis was that HIT exercise would result in a period of disinhibition within M1 that would be associated with the magnitude of offline skill gains. This is what we found, suggesting a mechanistic link between exercise-induced disinhibition mediated by GABAA synapses and time- but not sleep-dependent early consolidation processes critical for the formation of a new motor memory.

METHODS

Participants

Twenty-four healthy participants took part in this study (10 women, mean age = 23.63 ± 4.38 years, range = 19–41 years) without any known history of neurological impairment or current physical illness or injuries. The study was approved by the Monash University Human Research Ethics Committee, and participants provided informed consent before commencement of the study. Participants were screened for contraindications to TMS and exercise (Physical Activity Readiness Questionnaire). Handedness was assessed using the Edinburgh Handedness Inventory (Oldfield, 1971). Current self-reported physical activity was assessed using the International Physical Activity Questionnaire (IPAQ). Participants were asked to wear a Polar H7 heart rate monitor (Polar Electro, Kempele, Finland) throughout the assessment. They were also asked to refrain from strenuous physical activity for the 24 hr leading up to the assessment. Participants were naive to both TMS and the motor task and to the specific hypotheses of the study relating to disinhibition and offline consolidation.

Design

Participants were pseudorandomized to either a control group (n = 12, two left-handed) or an exercise group (n = 12, two left-handed), while minimizing variance in sex, age, height, weight, and self-reported physical activity (IPAQ) using an online algorithm (www.rando.la). This between-participant design consisted of a morning session involving TMS measures before and after 20 min of exercise (exercise group) or an equivalent period of rest (control group), followed by the acquisition of a novel motor skill (Figure 1A). Retention was assessed approximately 5 hr later, as offline effects require a minimum of 4 hr to develop (Kantak & Winstein, 2012) and may be differentially affected by sleep (Diekelmann & Born, 2010).

Figure 1. 

Illustrative summary of study design and its components. (A) Timeline of the single-day assessment. Participation involved baseline and postexercise or rest TMS measures, eight acquisition blocks, and four follow-up retention blocks approximately 5 hr later. (B) Mean heart rate during exercise session with 1 SD, expressed as a percentage of each participant's heart rate reserve (HRR) (target heart rate calculated as HRR * 0.5 + resting heart rate [RHR] for low-intensity epochs and HRR * 0.9 + RHR for high-intensity epochs). (C) SVIPT motor learning task depicted with example pseudorandomized color sequences for individual trials and feedback graph. The required sequence order for force pulses was always red–blue–green–yellow–white, returning to the home position between each pulse.

Figure 1. 

Illustrative summary of study design and its components. (A) Timeline of the single-day assessment. Participation involved baseline and postexercise or rest TMS measures, eight acquisition blocks, and four follow-up retention blocks approximately 5 hr later. (B) Mean heart rate during exercise session with 1 SD, expressed as a percentage of each participant's heart rate reserve (HRR) (target heart rate calculated as HRR * 0.5 + resting heart rate [RHR] for low-intensity epochs and HRR * 0.9 + RHR for high-intensity epochs). (C) SVIPT motor learning task depicted with example pseudorandomized color sequences for individual trials and feedback graph. The required sequence order for force pulses was always red–blue–green–yellow–white, returning to the home position between each pulse.

EMG and TMS

Surface EMG was recorded from the first dorsal interosseous (FDI) of the dominant hand, after skin preparation, using Ag–AgCl electrodes in a belly-tendon montage (Covidien, Dublin, Ireland). A ground electrode was placed on the back of the hand. The EMG signal was amplified, bandpass filtered 10–1000 Hz (PowerLab; ADInstruments, Dunedin, New Zealand), and then digitized at 4000 Hz using LabChart 7.0 (ADInstruments). Each EMG trace included 1 sec before the TMS trigger and 1 sec after.

Single and paired TMS pulses were delivered using a MagVenture MagPro X100 with MagOption unit (MagVenture Ltd., Farum, Denmark) and a monophasic waveform, with a C-B60 figure-of-eight coil (70-mm wing diameter). The TMS coil was held tangentially to the skull and at a 45° angle to the midline over the dominant M1 hand area, with a posterior to anterior current flow induced in the cortex. After finding the optimal coil position for motor evoked potentials (MEPs) in FDI, resting motor threshold (RMT) was determined as the lowest stimulator output needed for the TMS pulse to elicit an MEP of >50 μV in four of eight trials. Active motor threshold (AMT) was defined as the lowest intensity required to elicit MEPs of >200 μV in four of eight trials, whereas the dominant FDI was voluntarily activated. The following TMS measures were obtained before and commencing 10 min after exercise/rest, with the participants sitting and both hands resting on a pillow positioned on their lap:

  • 1. 

    The input–output (IO) relationship was used to assess the effect of exercise on corticomotor excitability. Four different TMS intensities were delivered over dominant M1, with 12 pulses per intensity: 100%, 110%, 120%, and 130% RMT. The order of intensities was randomized across participants but held constant over time for a given participant.

  • 2. 

    Measures of GABAergic inhibition within M1 were obtained. These included short-interval intracortical inhibition (SICI) at both 2- and 1-msec ISIs and long-interval intracortical inhibition (LICI) with a 120-msec ISI. SICI2msec is known to reflect synaptic inhibition via GABAA receptors (Ziemann et al., 2015), whereas SICI1msec is accepted as being mechanistically distinct, attributed to extrasynaptic GABA tone (Stagg, Bestmann, et al., 2011) or axonal refractoriness (Fisher, Nakamura, Bestmann, Rothwell, & Bostock, 2002). LICI was taken as a measure of slow-acting inhibition dependent on the GABAB receptor (Ziemann et al., 2015).

For SICI measures, the test stimulus (TS) intensity was set to produce nonconditioned (NC) MEPs of 1-mV peak-to-peak amplitude at rest (typically between 120% and 130% RMT), and the conditioning stimulus (CS) was set to 80% AMT to produce approximately 50% inhibition of the NC response for SICI2msec at baseline. Thereafter, the CS was held constant. At each time point, 12 NC MEPs and 12 conditioned MEPs for each of SICI2msec and SICI1msec were acquired. For the postmeasures of SICI, the TS intensity was adjusted where necessary to within 0.3 mV of the baseline NC MEP amplitude to ensure valid comparison of SICI across premeasures and postmeasures. Six of 24 participants (four in the exercise group) had altered corticomotor excitability at post. Adjustments were in the range of 1–5% of the maximum stimulator output (intensity was increased for four of the six participants, three of which were in the exercise group).

For LICI, both pulses were set at the TS1mV stimulus intensity and delivered 120 msec apart. A silent period duration of at least 150 msec when stimulating at TS1mV during voluntary contraction was confirmed at the beginning of the experiment. LICI was determined from 12 trials. LICI measures were not obtained for one participant in the control group because of technical difficulties during data acquisition.

SVIPT

All participants were new to the SVIPT task and were given standardized instructions. Participants were seated in front of a computer screen and held the force transducer in their dominant hand between their index finger and thumb. Squeezing the force transducer generated horizontal cursor movement with more force moving the cursor further to the right of the screen. The goal was to produce five individual pulses of force moving the cursor to five targets on the screen in a specific color sequence (red–blue–green–yellow–white), returning to the home position after each pulse (rest). The positions of the five colors were pseudorandomly alternated from trial to trial (Figure 1C); however, the target sequence order was always the same (red–blue–green–yellow–white), and each block consisted of the same pseudorandomized order of color positions. Participants were instructed to start as soon as the colors appeared, land the peak of their force pulses in the center of the targets, and try to be as fast and accurate as possible when completing the five-target sequence.

The furthest target was set at 45% of the participant's maximum voluntary pinch contraction (MVC), and the width of each target was 2.5% of their MVC. MVC was determined from the participant's maximal brief pinch on the force transducer, with the force pulse lasting approximately 0.5 sec. For the learning task, a logarithmic transformation was applied to the relationship between force and cursor movement, thereby increasing task difficultly because of high cursor gain at low levels of force.

A block consisted of eight trials of the five-color sequence, and visual feedback was provided graphically at the end so that participants could track their progress and use the information to improve on subsequent blocks. Before obtaining baseline TMS measures, participants were introduced to the task and completed one block to ensure they were familiar with the task requirements. The acquisition phase consisted of a further seven blocks; and the retention phase, four blocks. Before the retention test, participants performed one block to counteract the known warm-up decrement for this task (Reis et al., 2015). Participants were reminded of the sequence order at the beginning of both the acquisition and retention phases. There was approximately 5 sec between each trial and a 1-min break between blocks while feedback was displayed.

Motor learning is sensitive to positive and negative feedback (Galea, Mallia, Rothwell, & Diedrichsen, 2015). Thus, in addition to speed, error, and trial–trial accuracy for each target (Figure 1C), one of four possible messages was displayed at the end of each block, designed to provide positive and negative feedback on performance: (1) “Well done! You were both faster and more accurate compared to the previous block”; (2) “You were faster compared to the last block, but at the expense of force error. Try to be both fast and accurate”; (3) “Your force error decreased compared to the previous block, but at the expense of speed. Try to be both fast and accurate”; and (4) “Try harder. Both speed and force error deteriorated compared to the previous block.”

HIT Exercise Protocol

HIT exercise can be differentiated from low- and moderate-intensity aerobic exercises by the energy-generating process within the muscles being exercised. HITs require an increased rate of energy supply, which exceeds the rate at which glycogen can be broken down by aerobic glycolysis. This results in increased reliance on anaerobic glycolysis and accumulation of lactate in the circulation.

The HIT exercise session was performed on a stationary exercise bike (Wattbike, Geelong, Australia). Resting heart rate (RHR) was obtained while seated using a Polar H7 heart rate monitor (Polar Electro). The current study included participants with a range of fitness levels; hence, exercise intensity was appropriately tailored to each individual based on heart rate reserve (HRR):
formula
where
formula
(Tanaka, Monahan, & Seals, 2001).

Participants in the exercise group alternated between periods of low-intensity cycling for 3 min at approximately 50% HRR (HRR * 0.5 + RHR) and high-intensity cycling for 2 min with a target HR of up to 90% HRR (HRR * 0.9 + RHR), for a total duration of 20 min, followed by a 3-min low-intensity cool-down period (a similar protocol to that of Roig et al., 2012). During exercise, participants were not permitted to grip the handlebars but were allowed to rest their hands and forearms on a towel placed over the Wattbike's aero handlebars (time trial position).

Data Analysis

Neurophysiological Measures

For all TMS pulses, the MEP peak-to-peak amplitude was calculated and used as a measure of corticomotor excitability. MEP data were discarded online if they were contaminated with prestimulus muscle activation on the EMG recording or if postprocessing revealed a root mean squared value of >10 μV in the 100-msec window before the TMS pulse. MEP data for each block of 12 single or paired-pulse TMS trials were then subjected to Grubb's test, and significant outliers were removed before averaging.

The IO relationship was analyzed with a three-way mixed model ANOVA. The between-participant factor was Group (exercise or control), and the within-participant factors were Time (baseline and post-10 min), and Stimulus intensity (100%, 110%, 120%, and 130% RMT).

SICI measures (SICI2msec and SICI1msec) were expressed as the ratio of the NC response (C/NC) and subjected to two-way mixed model ANOVAs, with Group (exercise and control) as the between-participant factor and Time (baseline and postexercise) as the within-participant factor.

Similarly, LICI was measured by comparing the ratio of peak-to-peak amplitudes of the two suprathreshold stimuli (S2/S1) and analyzed using a two-way mixed model ANOVA, with Group (exercise and control) as the between-participant factor and Time (baseline and postexercise) as the within-participant factor.

SVIPT Motor Learning Measures

Performance on the SVIPT was assessed from the speed of each trial and accuracy of the force peaks. Trial time (speed) was calculated from target onset (appearance of the colored targets) to the last force peak. Trial force error (the inverse of accuracy) was calculated as the sum of differences between the center of each target and the participant's five respective force peaks. Block means were used to determine skill, as described below.

To account for the trade-off between speed and accuracy during motor performance (Fitts, 1954), a quantifiable measure of overall skill level is needed. Following the procedure outlined by Reis et al. (2009), we used a separate data set (n = 10 participants) to determine the speed–accuracy trade-off function for our task. Participants completed 10 blocks, paced at different rates using a metronome (randomized across blocks) on two separate days, and performance at each rate was then averaged across days. Systematically varying speed and recording accuracy derived a speed–accuracy trade-off function (r2 = .99):
formula
where a and b are dimensionless parameters and “duration” refers to the block mean trial time. Rearranging the function to solve for the skill parameter, a, and fixing the value of b at 1.627 allowed calculation of the “skill parameter”:
formula
where larger values of the skill parameter reflect greater skill. Finally, as per Reis et al. (2009), our skill measure was the logarithm of the skill parameter, which serves to homogenize variance. This skill measure was used in all statistical analyses.

Baseline skill (Block 1) was compared across groups using an independent-samples t test. Next, the entire skill acquisition phase was examined with a two-way mixed model ANOVA, with Group (exercise and control) as the between-participant factor and Acquisition block (2–8) as the within-participant factor.

Skill was then decomposed into “online effects,” which were the performance gains within participants between Block 1 and the average of Blocks 7–8, and “offline effects,” which were the difference in skill level between the average of Blocks 7–8 and the average of Blocks 9–10. To ensure that our analysis is comparable with other research using the SVIPT task, which included more trials in each block (Reis et al., 2009, 2015; Statton, Encarnacion, Celnik, & Bastian, 2015), we averaged across two blocks at the end of acquisition and at the beginning of retention. A clear outlier was detected that violated normality for the exercise group. Inspection of the data revealed that this participant was the first to complete the study. For this participant only, an error in the task feedback script meant that he or she received a disproportionate amount of negative reinforcement. Considering the effects of reinforcement on motor learning (Galea et al., 2015; Sternad & Kording, 2015), this participant was removed from all analyses. Two independent-measures t tests compared the exercise and control groups on their online and offline effects for the skill measure, and to better understand whether the observed effects could be ascribed to speed or accuracy, the tests were also conducted on the raw data used to calculate skills. For all statistical analyses, α was set to .05, and unless otherwise stated, results are reported as mean ± standard error.

RESULTS

Participant and Exercise Characteristics

Groups were well balanced on age, sex, handedness, body mass index, self-reported physical activity levels (IPAQ), and RHR, with all ps > .05. Participant means and standard deviations can be found in Table 1. Heart rate was monitored throughout the exercise (Figure 1B), and on average, the maximum heart rate recorded during exercise was 190.54 ± 4.10 beats/min. The power-to-weight ratio during the final two HITs provides an indication of participant fitness (Hawley & Noakes, 1992) and ranged from 0.84 to 4.37. In six participants, we confirmed that our exercise protocol resulted in high levels of lactate by obtaining capillary blood samples from the fingertip (Lactate Pro2, Arkray, Japan). At the end of the HIT protocol, lactate reached 12.17 ± 1.89 mmol/L.

Table 1. 

Participant Means and Standard Deviations for the Control and Exercise Groups

ControlExercise
n (female) 12 (6) 12 (4) 
Right-handedness (left) 10 (2) 10 (2) 
Age (years) 22.42 ± 1.73 24.83 ± 5.83 
Body mass index 22.08 ± 2.68 24.42 ± 2.68 
Self-report physical activity (IPAQ) 2139 ± 1214 2770 ± 1602 
RMT (%MSO) 67.08 ± 4.98 61.67 ± 8.87 
AMT (%MSO) 57.67 ± 4.91 52.08 ± 9.04 
TS (%RMT) 126 ± 7.51 125.67 ± 9.97 
CS (%AMT) 79 ± 9.07 77 ± 8.18 
RHR (bpm) 69.44 ± 7.38 65.58 ± 4.52 
HRR (bpm) – 124.92 ± 7.19 
HR (high intensity) – 190.54 ± 4.10 
RPE (high intensity) – 17.08 ± 0.63 
ControlExercise
n (female) 12 (6) 12 (4) 
Right-handedness (left) 10 (2) 10 (2) 
Age (years) 22.42 ± 1.73 24.83 ± 5.83 
Body mass index 22.08 ± 2.68 24.42 ± 2.68 
Self-report physical activity (IPAQ) 2139 ± 1214 2770 ± 1602 
RMT (%MSO) 67.08 ± 4.98 61.67 ± 8.87 
AMT (%MSO) 57.67 ± 4.91 52.08 ± 9.04 
TS (%RMT) 126 ± 7.51 125.67 ± 9.97 
CS (%AMT) 79 ± 9.07 77 ± 8.18 
RHR (bpm) 69.44 ± 7.38 65.58 ± 4.52 
HRR (bpm) – 124.92 ± 7.19 
HR (high intensity) – 190.54 ± 4.10 
RPE (high intensity) – 17.08 ± 0.63 

There were no significant group differences (all ps > .05). RMT (%MSO) = RMT as a percentage of maximal stimulator output (MSO); AMT (%MSO) = AMT as a percentage of MSO; RPE = Borg's Rating of Perceived Exertion; TS (%RMT) = TS intensity as a percentage of RMT; CS (%AMT) = CS intensity as a percentage of AMT.

Neurophysiological Measures

The groups were also similar for the TMS stimulation parameters RMT, AMT, TS intensity as a percentage of RMT, and CS intensity as a percentage of AMT (all ts < 1.84, ps > .05). Averages can be found in Table 1. Pretrigger root mean squared EMG was recorded and inspected to ensure that there was no muscle activity during TMS measures acquired at the baseline and post time points (mean ± standard deviation, control: mean = 3.87 μV and SD = 1.22 μV [at baseline], mean = 3.73 μV and SD = 1.34 μV [post]; exercise: mean = 3.36 μV and SD = 0.74 μV [at baseline], mean = 2.49 μV and SD = 0.58 μV [postexercise]).

Corticomotor Excitability

The group mean MEP amplitudes recorded at baseline and 10 min after exercise, or rest, are shown in Figure 2A. As expected, MEP amplitude increased with increasing stimulation intensity, F(3, 63) = 27.14, p < .001, ηp2 = .56. There were no other main effects (Group: F(1, 21) = 0.42, p = .53, ηp2 = .02; Time: F(1, 21) = 0.53, p = .48, ηp2 = .03) and no interactions (all Fs < 1.57, ps > .23).

Figure 2. 

Changes in intracortical excitability and inhibition. (A) IO relationship for the control and exercise groups, at stimulation intensities of 100%, 110%, 120%, and 130% RMT. (B–D) Paired-pulse measures of GABA inhibition before and after exercise, SICI2msec, SICI1msec, and LICI, respectively. For both SICI and LICI, a y axis value of 0 reflects total suppression of the TS MEP with addition of the CS, whereas a value of 1 reflects no suppression of the TS MEP. Disinhibition is present after exercise for SICI2msec, but not SICI1msec or LICI. All data are presented as group means and SEM. *p < .05, significant reduction of inhibition.

Figure 2. 

Changes in intracortical excitability and inhibition. (A) IO relationship for the control and exercise groups, at stimulation intensities of 100%, 110%, 120%, and 130% RMT. (B–D) Paired-pulse measures of GABA inhibition before and after exercise, SICI2msec, SICI1msec, and LICI, respectively. For both SICI and LICI, a y axis value of 0 reflects total suppression of the TS MEP with addition of the CS, whereas a value of 1 reflects no suppression of the TS MEP. Disinhibition is present after exercise for SICI2msec, but not SICI1msec or LICI. All data are presented as group means and SEM. *p < .05, significant reduction of inhibition.

SICI

To assess the effect of exercise on SICI, two-way mixed model ANOVAs were conducted with SICI2msec and SICI1msec as the dependent variables. For the SICI2msec analysis, one participant in the exercise group showed the desired 50% inhibition at baseline, but inhibition was dramatically reduced postexercise to the extent that facilitation was observed. Despite this observation being consistent with our disinhibition hypothesis, this participant was removed from all analyses involving SICI2msec.

The SICI2msec analysis revealed a main effect of Time (F(1, 20) = 7.11, p = .02, ηp2 = .26), no main effect of Group (F(1, 20) = 0.05, p = .83, ηp2 = .002), and an interaction between Time and Group (F(1, 20) = 7.76, p = .01, ηp2 = .28). As seen in Figure 2B and confirmed with post hoc tests, the exercise group showed a reduction in SICI after exercise (p = .002, d = 1.53), whereas SICI2msec was unchanged in the control group (p = .94).

For SICI1msec, there were no main effects and no interaction (Figure 2C; Time: F(1, 21) = 0.62, p = .44, ηp2=.03; Group: F(1, 21)=0.17, p=.68, ηp2= .01; Time × Group: F(1, 21) = 1.87, p = .19, ηp2 = .08).

LICI

Again, there were no main effects and no interaction (Figure 2D; Time: F(1, 21) = 0.02, p = .88, ηp2 = .001; Group: F(1, 21) = 1.99, p = .17, ηp2 = .009; Time × Group: F(1, 21) = 0.03, p = .87, ηp2 = .001).

In summary, there was no effect of exercise on corticomotor excitability, SICI1msec, or LICI. However, HIT exercise significantly reduced SICI2msec, indicating a specific effect on synaptic inhibitory neurotransmission involving the GABAA receptor.

SVIPT Task Performance

Group means for speed (Figure 3A) and force error (Figure 3B) are included for descriptive purposes only. These variables were used in the computation of the skill measure (Figure 3C), which takes into consideration the known interdependency between speed and accuracy.

Figure 3. 

Learning data for the motor learning task. (A) Decrease in SVIPT trial time for both groups over 12 blocks. (B) Decrease in trial force error rate on SVIPT in both groups over 12 blocks. (C) Speed and force error were used for the calculation of a motor skill measure, which increased in both groups over the 12 blocks. (D–F) Analysis of online and offline learning effects. There was no difference between groups in online skill gain. In contrast, the exercise group demonstrated a positive offline effect for the skill measure. Total indicates the sum of online and offline effects. All data are presented as means ± SEM. Level of significance: *p < .05.

Figure 3. 

Learning data for the motor learning task. (A) Decrease in SVIPT trial time for both groups over 12 blocks. (B) Decrease in trial force error rate on SVIPT in both groups over 12 blocks. (C) Speed and force error were used for the calculation of a motor skill measure, which increased in both groups over the 12 blocks. (D–F) Analysis of online and offline learning effects. There was no difference between groups in online skill gain. In contrast, the exercise group demonstrated a positive offline effect for the skill measure. Total indicates the sum of online and offline effects. All data are presented as means ± SEM. Level of significance: *p < .05.

Baseline Performance and Skill Acquisition

Skill was not significantly different between groups at baseline (t(1, 21) = −1.31, p = .21, 95% CI [−0.43, 0.10], d = 0.55). Both groups showed improvement of skill over the acquisition phase (Figure 3C). A two-way mixed model ANOVA, with Group and Block (2–8) as factors, were used to compare skills across the groups during acquisition. As sphericity was violated, the Greenhouse–Geisser correction was applied (ε = 0.55). There was a main effect of Block (F(6, 126) = 21.70, p < .001, ηp2 = .51) but no main effect of Group (F(1, 21) = 1.95, p = .18, ηp2 = .09) and no interaction between Block and Group (F(6, 126) = 1.14, p = .34, ηp2 = .05).

For each individual, the SVIPT skill measure was decomposed into online and offline effects, as per Reis et al. (2009, 2015).

Online Effects

There was no difference between the exercise and control groups for the online effect for the skill measure (Figure 3F; t(1, 21) = −0.45, p = .66, 95% CI [−0.35, 0.23], d = 0.19) or for either speed (Figure 3D; t(1, 21) = −0.53, p = .60, 95% CI [−0.83, 0.49], d = 0.22) or force error (Figure 3E; t(1, 21) = 0.16, p = .88, 95% CI [−0.10, 0.12], d = 0.07) when considered separately.

Offline Effects

The exercise group showed an improvement in skill at the retention test (Figure 3F; M = 0.084, SE = 0.038), whereas the control group did not (M = −0.034, SE = 0.035), and this difference was significant (t(1, 21) = −2.29, p = .03, 95% CI [−0.22, −0.01], d = 0.96). Eight of 11 participants in the exercise group showed a positive offline effect for skill, compared with 5 of 12 in the control group. Looking at the subcomponents of the skill measure, the exercise group showed an offline improvement for force error (Figure 3D; t(1, 21) = −2.60, p = .017, 95% CI [−0.80, −0.01], d = 1.08) but not for speed (Figure 3E; t(1, 21) = 0.16, p = .87, 95% CI [−0.24, 0.28], d = 0.07).

Association between SICI and Offline Effects

To then assess whether the modulation of inhibition (SICI2msec) was associated with the offline learning effect, a Pearson correlation and linear regression analysis was conducted. First, the change in SICI (ΔSICI2msec) was calculated, with positive values indicating reduced inhibition. The analysis found that, when considering the groups separately, ΔSICI2msec was not significantly associated with the offline effect (Figure 4A; control: F(1, 10) = 1.57, R2 = .14, r = .37, p = .24; exercise: F(1, 8) = 0.23, R2 = .03, r = .17, p = .64). As the slopes were not significantly different from each other (F(1, 18) = 0.45, p = .51), the data were pooled to increase statistical power. This exploratory analysis suggests that ΔSICI2msec may well be positively associated with the offline effect (Figure 4B; F(1, 20) = 6.70, R2 = .25, r = .50, p = .018), accounting for 25% of the variance. Although this establishes an association between disinhibition and the offline learning effect, the result is not specific to exercise and will require replication in a larger sample.

Figure 4. 

Scatterplot representing the relationship between changes in SICI and offline effects. Linear regressions are shown for each group separately (A) and for all participants combined (B). People in whom there is a greater release of GABAA-receptor-mediated SICI2msec showed a greater offline skill improvement (R2 = .25, r = .50, p = .018). Positive values for ΔSICI and Δoffline indicate a reduction of inhibition and an offline skill gain. Dashed lines represent 95% confidence intervals for the line of best fit.

Figure 4. 

Scatterplot representing the relationship between changes in SICI and offline effects. Linear regressions are shown for each group separately (A) and for all participants combined (B). People in whom there is a greater release of GABAA-receptor-mediated SICI2msec showed a greater offline skill improvement (R2 = .25, r = .50, p = .018). Positive values for ΔSICI and Δoffline indicate a reduction of inhibition and an offline skill gain. Dashed lines represent 95% confidence intervals for the line of best fit.

DISCUSSION

Our study had three main findings. First, an acute bout of HIT exercise involving the lower limbs resulted in improved consolidation of a novel motor skill (a positive offline effect) involving the hand. Second, lower-limb HIT exercise resulted in a period of GABAA-receptor-mediated synaptic disinhibition in the hand representation of M1 (SICI2msec), whereas measures of GABAB-receptor-mediated inhibition (LICI) and nonsynaptic inhibition (SICI1msec) were unchanged. Finally, when considering all participants together, we found a novel association between the magnitude of disinhibition and positive offline effects.

HIT Exercise Improves Motor Memory through an Effect on Early Consolidation

We found evidence to support our hypothesis that exercise would improve offline, but not online, effects. This prediction derived from a previous study reporting improved motor memory at 24-hr and 7-day retention tests (Roig et al., 2012). In our study, retention was tested on the same day after a period of 5-hr wakefulness. Although we did not assess long-term retention, a noteworthy implication is that the observed offline gain in our study cannot be explained by sleep-dependent consolidation (Barakat et al., 2013; Walker, Brakefield, Morgan, Hobson, & Stickgold, 2002). There are known benefits of acute exercise on sleep (Kredlow, Capozzoli, Hearon, Calkins, & Otto, 2015), which could have contributed to previous results (Mang et al., 2014; Roig et al., 2012). The observed offline skill loss in our study is also often seen in control groups (Reis et al., 2009, 2015; Cantarero, Lloyd, & Celnik, 2013; Roig et al., 2012). a-tDCS applied to M1 during motor practice has been reported to prevent offline skill loss and promote gains (Reis et al., 2009) that do not depend on sleep (Reis et al., 2015). Our results suggest that, similar to a-tDCS, HIT exercise may be beneficial for early consolidation in the absence of sleep.

High-intensity exercise did not benefit motor task acquisition (online effects). On average, the control group appeared to have slightly better skill than the exercise group when first exposed to the task (a nonsignificant difference at baseline, before exercise), but both groups improved to the same degree, with no Group main effects or interactions. Although, at first glance, it appears that participants favored speed over accuracy after exercise, this was not the case with online effects in fact similar across groups for both speed and force error (accuracy). We used a 20-min HIT protocol similar to Roig and colleagues (2012), who also showed no effect of exercise on acquisition, whereas two subsequent studies have reported that exercise enhances motor skill acquisition. One used a cycling HIT protocol (Mang et al., 2014), whereas the other involved steady-state treadmill exercise of moderate intensity (Statton et al., 2015). Thus, to date, the literature investigating whether exercise enhances motor skill acquisition is mixed and may be dependent on factors such as exercise intensity or the timing and duration of task performance (Chang, Labban, Gapin, & Etnier, 2012). Postexercise fatigue, or the delay incurred to obtain TMS measures, could have counteracted the beneficial effects of exercise on motor skill acquisition, but such explanations are speculative and require further investigation. It is also possible that the physiological effects of high-intensity exercise interact more with the mechanisms underlying consolidation than acquisition.

HIT Exercise Reduces GABAA Receptor-Dependent Synaptic Inhibition

HIT exercise resulted in disinhibition in the hand representation of M1, with an effect found for SICI2msec, but not SICI1msec or LICI. Our finding for SICI2msec is consistent with two previous studies of moderate-intensity lower-limb exercise (Singh, Duncan, et al., 2014; Smith et al., 2014) and extends these observations to include HIT exercise. Our results also provide some insight into the specific inhibitory circuits affected by exercise.

Somewhat in contrast with our results, others have reported reduced LICI after moderate-intensity exercise (Mooney et al., 2016; Singh, Duncan, et al., 2014). The high-intensity exercise used in our study elevates cortisol (Rojas Vega et al., 2006), which is known to negatively affect GABAB inhibition (Sale, Ridding, & Nordstrom, 2008). Intensity could mediate the effects of exercise on GABAB circuitry, but this remains to be determined.

We also established that the reduction in SICI depends on the ISI. When the TMS pulses are separated by 2–3 msec, SICI is known to reflect inhibition that depends on GABAA receptors (Ziemann et al., 2015), whereas SICI1msec is mechanistically distinct and considered a marker of extrasynaptic GABA tone (Stagg, Bestmann, et al., 2011) or axonal refractoriness (Fisher et al., 2002). Interestingly, SICI1msec correlates with sensorimotor cortex GABA concentration derived from magnetic resonance spectroscopy (Stagg, Bestmann, et al., 2011). It has been reported that high-intensity exercise increases GABA concentration in the primary visual cortex (Maddock, Casazza, Fernandez, & Maddock, 2016)—an effect ascribed to de novo neurotransmitter synthesis. The apparent contrasting results of exercise on measures of GABA derived from TMS and magnetic resonance spectroscopy are intriguing and warrant further investigation, especially considering that these measures tap into different aspects of GABA function (Stagg, 2014). Our finding that HIT exercise resulted in the modulation of SICI2msec, but not SICI1msec, supports the conclusion that exercise has a specific effect on synaptic GABAA neurotransmission.

Corticomotor excitability, indexed by MEPs to single-pulse TMS delivered to the M1 hand region, was unchanged in our study. Previous studies have similarly reported that moderate-intensity (Singh, Duncan, et al., 2014; Smith et al., 2014) and HIT lower-limb cycling exercises (Mang et al., 2014) have no effect on M1 hand-region excitability. Collectively, these observations suggest that lower-limb exercise, on its own, does not induce LTP-like effects in the M1 hand region. This is important because LTP-like effects could occlude subsequent motor learning (Cantarero, Tang, O'Malley, Salas, & Celnik, 2013). In fact, there is evidence from repetitive TMS protocols that the capacity for plasticity induction in the M1 hand region is enhanced after both moderate (Singh, Neva, et al., 2014; McDonnell, Buckley, Opie, Ridding, & Semmler, 2013) and HIT exercises (Mang et al., 2014). Disinhibition may explain the enhanced capacity for plasticity induction after exercise.

HIT Exercise-induced Disinhibition Is Associated with Offline Gains in Motor Learning

A novel result of our study was that reduced SICI2msec was significantly associated with positive offline gains. This result suggests that GABAA-mediated inhibition may contribute to the effect of HIT exercise on motor learning; however, the association was only significant when combining participants from both groups and will require investigation in a larger sample. The association across all participants suggests that disinhibition is relevant for understanding the mechanisms that regulate memory consolidation more generally. Indeed, it has recently been proposed that inhibitory mechanisms function as a switch to determine which memories are consolidated (Breton & Robertson, 2014). The evidence to support the theory of an inhibitory mechanism (Tunovic, Press, & Robertson, 2014) was inferred from single-pulse TMS, which is sensitive to changes in both excitatory and inhibitory circuits. The convergent evidence from our paired-pulse TMS results strengthen the argument that disinhibition enables memory consolidation and specifically implicate the GABAA receptor. LTP-like plasticity is strongly implicated in the consolidation of motor memory (Cantarero, Lloyd, et al., 2013; Reis et al., 2009; Muellbacher et al., 2002; Rioult-Pedotti, Friedman, & Donoghue, 2000), and the removal of GABA inhibition is a known requirement for the induction of LTP in M1 (Rioult-Pedotti et al., 1998; Hess et al., 1996) and other brain circuits (Letzkus, Wolff, & Luthi, 2015). Disinhibition, in the absence of altered corticomotor excitability, may result in a state that is favorable for laying down new memories within M1.

Limitations

Changes in SICI after exercise within M1 explained only part of the variance in offline effects, which suggests that other brain regions may have also contributed to task learning and consolidation or that factors other than SICI may have contributed to the interaction between exercise and motor learning. High-intensity exercise is associated with increased levels of neurochemicals, such as BDNF and lactate (Skriver et al., 2014; Cotman & Berchtold, 2007). BDNF in blood plasma has been shown to play a role in LTP-like plasticity (Fritsch et al., 2010) and is elevated more so after HIT exercise than steady-state continuous exercise (Saucedo Marquez et al., 2015). Elevated levels of BDNF and lactate in blood after HIT exercise have also been associated with improved motor skill retention (Skriver et al., 2014). Although there is debate regarding whether blood BDNF reflects brain BDNF, circulating lactate is a significant source of energy for neurons (van Hall et al., 2009; Kemppainen et al., 2005). However, the mechanism through which lactate might support the consolidation of motor memory is not well understood (Skriver et al., 2014). Our study establishes a role for reduced GABAA-mediated inhibition after HIT exercise, but how this relates to elevated BDNF and lactate is not known as these measures were not obtained.

It is difficult to generalize our TMS findings to the entire M1, considering we measured inhibition changes from one of several muscles critical to motor task execution (i.e., FDI). We reason that similar results are likely in other hand muscles as previous studies have recorded from abductor pollicis brevis (Singh, Neva, et al., 2014) or extensor carpi radialis (Singh, Duncan, et al., 2014). Seeing as offline learning effects were found for a motor task involving activity of several muscles, it is reasonable to assume that the effects of exercise are somewhat generalizable to other nonexercised cortical representations within M1 and perhaps even to other cortical regions. Combining TMS with EEG (TMS-EEG) to investigate transcranial evoked potentials could provide a way to test whether the effects we observe are restricted to the motor cortex or reflect a global disinhibition.

Although the FDI hand muscle was not involved in producing the force necessary to sustain cycling exercise, it is possible that movement of the lower limbs contributed to disinhibition of the hand region in our study and in previous studies (Singh, Duncan, et al., 2014; Smith et al., 2014). For example, voluntary lower-limb movements can modulate inhibition in the upper-limb motor cortex, likely because of common input from the premotor cortex (Byblow et al., 2007). In the study by Byblow and colleagues, dynamic spatiotemporal modulation of forearm muscles was observed during rhythmical movements about the ankle (a negligible cardiovascular challenge), and the effect was only demonstrated for muscle–limb pairs with a known preference for isodirectional coupling. Others have looked at the aftereffects of upper-limb passive movement. No effects were found for the measures of GABAA inhibition as a result of rhythmic unimanual passive movement for 1 hr (Mace, Levin, Alaerts, Rothwell, & Swinnen, 2008) or rhythmic bimanual active–passive movement for 20 min (Byblow et al., 2012), although measures were taken from the limbs being moved. Thus, we think movement per se is a possible, yet unlikely, explanation for our results, and more research is clearly needed. Systematic manipulation of movement volition and intensity is required to determine whether it is movement, or effects that result from cardiovascular challenge, that results in generalized disinhibition within M1.

HIT exercise resulted in a positive offline effect, but in our study, the control group attained a higher absolute level of skill. This is probably due to interindividual variability in baseline motor skill that was not accounted for despite randomization of participants to the exercise and control groups. In addition, multiple sessions may be necessary for the full benefits of HIT exercise on motor learning to be revealed. Reis et al. (2009) showed that five consecutive days of a-tDCS during motor practice lead to overall better total learning via the cumulative effect of offline gains between each practice session. It remains to be determined whether multiple sessions of HIT exercise before motor learning produce additive offline gains and increased total learning. As high-intensity exercise releases different neurochemicals into the blood stream compared with low- or moderate-intensity exercise, future research should investigate the dose–response relationship to further elucidate how exercise-induced disinhibition and changes in BDNF and lactate interact with LTP-like plasticity to promote learning.

Conclusion

We show that lower-limb HIT exercise results in synaptic GABAA disinhibition within the hand representation of M1. We also show that these effects on cortical physiology may have functional relevance, with exercise associated with an offline skill gain supporting early consolidation of a novel fine motor skill in the absence of sleep.

Acknowledgments

We are grateful to Nigel Rogasch for comments on the manuscript; Karrin Kuyumcian, Chris Alarcón, and Kim Le for their assistance with data collection and analysis; and the staff at the Monash Instrumentation Facility and Monash Biomedical Imaging.

Reprint requests should be sent to James P. Coxon, Faculty of Medicine, Nursing, and Health Sciences, School of Psychological Sciences, Monash University, 18 Innovation Walk, Clayton 3800 VIC, Australia, or via e-mail: james.coxon@monash.edu.

REFERENCES

REFERENCES
Barakat
,
M.
,
Carrier
,
J.
,
Debas
,
K.
,
Lungu
,
O.
,
Fogel
,
S.
,
Vandewalle
,
G.
, et al
(
2013
).
Sleep spindles predict neural and behavioral changes in motor sequence consolidation
.
Human Brain Mapping
,
34
,
2918
2928
.
Breton
,
J.
, &
Robertson
,
E. M.
(
2014
).
Flipping the switch: Mechanisms that regulate memory consolidation
.
Trends in Cognitive Sciences
,
18
,
629
634
.
Butefisch
,
C. M.
,
Davis
,
B. C.
,
Wise
,
S. P.
,
Sawaki
,
L.
,
Kopylev
,
L.
,
Classen
,
J.
, et al
(
2000
).
Mechanisms of use-dependent plasticity in the human motor cortex
.
Proceedings of the National Academy of Sciences, U.S.A.
,
97
,
3661
3665
.
Byblow
,
W. D.
,
Coxon
,
J. P.
,
Stinear
,
C. M.
,
Fleming
,
M. K.
,
Williams
,
G.
,
Muller
,
J. F.
, et al
(
2007
).
Functional connectivity between secondary and primary motor areas underlying hand-foot coordination
.
Journal of Neurophysiology
,
98
,
414
422
.
Byblow
,
W. D.
,
Stinear
,
C. M.
,
Smith
,
M. C.
,
Bjerre
,
L.
,
Flaskager
,
B. K.
, &
McCambridge
,
A. B.
(
2012
).
Mirror symmetric bimanual movement priming can increase corticomotor excitability and enhance motor learning
.
PLoS One
,
7
,
e33882
.
Cantarero
,
G.
,
Lloyd
,
A.
, &
Celnik
,
P.
(
2013
).
Reversal of long-term potentiation-like plasticity processes after motor learning disrupts skill retention
.
Journal of Neuroscience
,
33
,
12862
12869
.
Cantarero
,
G.
,
Tang
,
B.
,
O'Malley
,
R.
,
Salas
,
R.
, &
Celnik
,
P.
(
2013
).
Motor learning interference is proportional to occlusion of LTP-like plasticity
.
Journal of Neuroscience
,
33
,
4634
4641
.
Chang
,
Y. K.
,
Labban
,
J. D.
,
Gapin
,
J. I.
, &
Etnier
,
J. L.
(
2012
).
The effects of acute exercise on cognitive performance: A meta-analysis
.
Brain Research
,
1453
,
87
101
.
Chen
,
S. X.
,
Kim
,
A. N.
,
Peters
,
A. J.
, &
Komiyama
,
T.
(
2015
).
Subtype-specific plasticity of inhibitory circuits in motor cortex during motor learning
.
Nature Neuroscience
,
18
,
1109
1115
.
Classen
,
J.
, &
Cohen
,
L. G.
(
2003
).
Practice-induced plasticity in the human motor cortex
. In
S.
Boniface
&
U.
Ziemann
(Eds.),
Plasticity in the human nervous system
(pp.
90
107
).
Cambridge
:
Cambridge University Press
.
Cotman
,
C. W.
, &
Berchtold
,
N. C.
(
2007
).
Physical activity and the maintenance of cognition: Learning from animal models
.
Alzheimers Dement
,
3(2 Suppl.)
,
S30
S37
.
Coxon
,
J. P.
,
Peat
,
N. M.
, &
Byblow
,
W. D.
(
2014
).
Primary motor cortex disinhibition during motor skill learning
.
Journal of Neurophysiology
,
112
,
156
164
.
Dayan
,
E.
, &
Cohen
,
L. G.
(
2011
).
Neuroplasticity subserving motor skill learning
.
Neuron
,
72
,
443
454
.
Diekelmann
,
S.
, &
Born
,
J.
(
2010
).
The memory function of sleep
.
Nature Reviews Neuroscience
,
11
,
114
126
.
Fisher
,
R. J.
,
Nakamura
,
Y.
,
Bestmann
,
S.
,
Rothwell
,
J. C.
, &
Bostock
,
H.
(
2002
).
Two phases of intracortical inhibition revealed by transcranial magnetic threshold tracking
.
Experimental Brain Research
,
143
,
240
248
.
Fitts
,
P. M.
(
1954
).
The information capacity of the human motor system in controlling the amplitude of movement
.
Journal of Experimental Psychology
,
47
,
381
391
.
Fritsch
,
B.
,
Reis
,
J.
,
Martinowich
,
K.
,
Schambra
,
H. M.
,
Ji
,
Y.
,
Cohen
,
L. G.
, et al
(
2010
).
Direct current stimulation promotes BDNF-dependent synaptic plasticity: Potential implications for motor learning
.
Neuron
,
66
,
198
204
.
Galea
,
J. M.
,
Mallia
,
E.
,
Rothwell
,
J.
, &
Diedrichsen
,
J.
(
2015
).
The dissociable effects of punishment and reward on motor learning
.
Nature Neuroscience
,
18
,
597
602
.
Hawley
,
J. A.
, &
Noakes
,
T. D.
(
1992
).
Peak power output predicts maximal oxygen uptake and performance time in trained cyclists
.
European Journal of Applied Physiology and Occupational Physiology
,
65
,
79
83
.
Hess
,
G.
,
Aizenman
,
C. D.
, &
Donoghue
,
J. P.
(
1996
).
Conditions for the induction of long-term potentiation in layer II/III horizontal connections of the rat motor cortex
.
Journal of Neurophysiology
,
75
,
1765
1778
.
Kantak
,
S. S.
, &
Winstein
,
C. J.
(
2012
).
Learning-performance distinction and memory processes for motor skills: A focused review and perspective
.
Behavioural Brain Research
,
228
,
219
231
.
Kemppainen
,
J.
,
Aalto
,
S.
,
Fujimoto
,
T.
,
Kalliokoski
,
K. K.
,
Langsjo
,
J.
,
Oikonen
,
V.
, et al
(
2005
).
High intensity exercise decreases global brain glucose uptake in humans
.
Journal of Physiology
,
568
,
323
332
.
Kredlow
,
M. A.
,
Capozzoli
,
M. C.
,
Hearon
,
B. A.
,
Calkins
,
A. W.
, &
Otto
,
M. W.
(
2015
).
The effects of physical activity on sleep: A meta-analytic review
.
Journal of Behavioral Medicine
,
38
,
427
449
.
Letzkus
,
J. J.
,
Wolff
,
S. B.
, &
Luthi
,
A.
(
2015
).
Disinhibition, a circuit mechanism for associative learning and memory
.
Neuron
,
88
,
264
276
.
Mace
,
M. J.
,
Levin
,
O.
,
Alaerts
,
K.
,
Rothwell
,
J. C.
, &
Swinnen
,
S. P.
(
2008
).
Corticospinal facilitation following prolonged proprioceptive stimulation by means of passive wrist movement
.
Journal of Clinical Neurophysiology
,
25
,
202
209
.
Maddock
,
R. J.
,
Casazza
,
G. A.
,
Fernandez
,
D. H.
, &
Maddock
,
M. I.
(
2016
).
Acute modulation of cortical glutamate and GABA content by physical activity
.
Journal of Neuroscience
,
36
,
2449
2457
.
Mang
,
C. S.
,
Snow
,
N. J.
,
Campbell
,
K. L.
,
Ross
,
C. J.
, &
Boyd
,
L. A.
(
2014
).
A single bout of high-intensity aerobic exercise facilitates response to paired associative stimulation and promotes sequence-specific implicit motor learning
.
Journal of Applied Physiology
,
117
,
1325
1336
.
McDonnell
,
M. N.
,
Buckley
,
J. D.
,
Opie
,
G. M.
,
Ridding
,
M. C.
, &
Semmler
,
J. G.
(
2013
).
A single bout of aerobic exercise promotes motor cortical neuroplasticity
.
Journal of Applied Physiology
,
114
,
1174
1182
.
McDonnell
,
M. N.
,
Orekhov
,
Y.
, &
Ziemann
,
U.
(
2006
).
The role of GABA(B) receptors in intracortical inhibition in the human motor cortex
.
Experimental Brain Research
,
173
,
86
93
.
Mooney
,
R. A.
,
Coxon
,
J. P.
,
Cirillo
,
J.
,
Glenny
,
H.
,
Gant
,
N.
, &
Byblow
,
W. D.
(
2016
).
Acute aerobic exercise modulates primary motor cortex inhibition
.
Experimental Brain Research
,
234
,
3669
3676
.
Muellbacher
,
W.
,
Ziemann
,
U.
,
Wissel
,
J.
,
Dang
,
N.
,
Kofler
,
M.
,
Facchini
,
S.
, et al
(
2002
).
Early consolidation in human primary motor cortex
.
Nature
,
415
,
640
644
.
Oldfield
,
R. C.
(
1971
).
The assessment and analysis of handedness: The Edinburgh inventory
.
Neuropsychologia
,
9
,
97
113
.
Reis
,
J.
,
Fischer
,
J. T.
,
Prichard
,
G.
,
Weiller
,
C.
,
Cohen
,
L. G.
, &
Fritsch
,
B.
(
2015
).
Time- but not sleep-dependent consolidation of tDCS-enhanced visuomotor skills
.
Cerebral Cortex
,
25
,
109
117
.
Reis
,
J.
,
Schambra
,
H. M.
,
Cohen
,
L. G.
,
Buch
,
E. R.
,
Fritsch
,
B.
,
Zarahn
,
E.
, et al
(
2009
).
Noninvasive cortical stimulation enhances motor skill acquisition over multiple days through an effect on consolidation
.
Proceedings of the National Academy of Sciences, U.S.A.
,
106
,
1590
1595
.
Rioult-Pedotti
,
M. S.
,
Friedman
,
D.
, &
Donoghue
,
J. P.
(
2000
).
Learning-induced LTP in neocortex
.
Science
,
290
,
533
536
.
Rioult-Pedotti
,
M. S.
,
Friedman
,
D.
,
Hess
,
G.
, &
Donoghue
,
J. P.
(
1998
).
Strengthening of horizontal cortical connections following skill learning
.
Nature Neuroscience
,
1
,
230
234
.
Robertson
,
E. M.
,
Press
,
D. Z.
, &
Pascual-Leone
,
A.
(
2005
).
Off-line learning and the primary motor cortex
.
Journal of Neuroscience
,
25
,
6372
6378
.
Roig
,
M.
,
Skriver
,
K.
,
Lundbye-Jensen
,
J.
,
Kiens
,
B.
, &
Nielsen
,
J. B.
(
2012
).
A single bout of exercise improves motor memory
.
PLoS One
,
7
,
e44594
.
Rojas Vega
,
S.
,
Struder
,
H. K.
,
Vera Wahrmann
,
B.
,
Schmidt
,
A.
,
Bloch
,
W.
, &
Hollmann
,
W.
(
2006
).
Acute BDNF and cortisol response to low intensity exercise and following ramp incremental exercise to exhaustion in humans
.
Brain Research
,
1121
,
59
65
.
Sale
,
M. V.
,
Ridding
,
M. C.
, &
Nordstrom
,
M. A.
(
2008
).
Cortisol inhibits neuroplasticity induction in human motor cortex
.
Journal of Neuroscience
,
28
,
8285
8293
.
Saucedo Marquez
,
C. M.
,
Vanaudenaerde
,
B.
,
Troosters
,
T.
, &
Wenderoth
,
N.
(
2015
).
High-intensity interval training evokes larger serum BDNF levels compared with intense continuous exercise
.
Journal of Applied Physiology
,
119
,
1363
1373
.
Singh
,
A. M.
,
Duncan
,
R. E.
,
Neva
,
J. L.
, &
Staines
,
W. R.
(
2014
).
Aerobic exercise modulates intracortical inhibition and facilitation in a nonexercised upper limb muscle
.
BMC Sports Science, Medicine and Rehabilitation
,
6
,
23
.
Singh
,
A. M.
,
Neva
,
J. L.
, &
Staines
,
W. R.
(
2014
).
Acute exercise enhances the response to paired associative stimulation-induced plasticity in the primary motor cortex
.
Experimental Brain Research
,
232
,
3675
3685
.
Skriver
,
K.
,
Roig
,
M.
,
Lundbye-Jensen
,
J.
,
Pingel
,
J.
,
Helge
,
J. W.
,
Kiens
,
B.
, et al
(
2014
).
Acute exercise improves motor memory: Exploring potential biomarkers
.
Neurobiology of Learning and Memory
,
116
,
46
58
.
Smith
,
A. E.
,
Goldsworthy
,
M. R.
,
Garside
,
T.
,
Wood
,
F. M.
, &
Ridding
,
M. C.
(
2014
).
The influence of a single bout of aerobic exercise on short-interval intracortical excitability
.
Experimental Brain Research
,
232
,
1875
1882
.
Stagg
,
C. J.
(
2014
).
Magnetic resonance spectroscopy as a tool to study the role of GABA in motor-cortical plasticity
.
Neuroimage
,
86
,
19
27
.
Stagg
,
C. J.
,
Bachtiar
,
V.
, &
Johansen-Berg
,
H.
(
2011
).
The role of GABA in human motor learning
.
Current Biology
,
21
,
480
484
.
Stagg
,
C. J.
,
Best
,
J. G.
,
Stephenson
,
M. C.
,
O'Shea
,
J.
,
Wylezinska
,
M.
,
Kincses
,
Z. T.
, et al
(
2009
).
Polarity-sensitive modulation of cortical neurotransmitters by transcranial stimulation
.
Journal of Neuroscience
,
29
,
5202
5206
.
Stagg
,
C. J.
,
Bestmann
,
S.
,
Constantinescu
,
A. O.
,
Moreno
,
L. M.
,
Allman
,
C.
,
Mekle
,
R.
, et al
(
2011
).
Relationship between physiological measures of excitability and levels of glutamate and GABA in the human motor cortex
.
Journal of Physiology
,
589
,
5845
5855
.
Statton
,
M. A.
,
Encarnacion
,
M.
,
Celnik
,
P.
, &
Bastian
,
A. J.
(
2015
).
A single bout of moderate aerobic exercise improves motor skill acquisition
.
PLoS One
,
10
,
e0141393
.
Sternad
,
D.
, &
Kording
,
K. P.
(
2015
).
Carrot or stick in motor learning
.
Nature Neuroscience
,
18
,
480
481
.
Tanaka
,
H.
,
Monahan
,
K. D.
, &
Seals
,
D. R.
(
2001
).
Age-predicted maximal heart rate revisited
.
Journal of the American College of Cardiology
,
37
,
153
156
.
Tunovic
,
S.
,
Press
,
D. Z.
, &
Robertson
,
E. M.
(
2014
).
A physiological signal that prevents motor skill improvements during consolidation
.
Journal of Neuroscience
,
34
,
5302
5310
.
van Hall
,
G.
,
Stromstad
,
M.
,
Rasmussen
,
P.
,
Jans
,
O.
,
Zaar
,
M.
,
Gam
,
C.
, et al
(
2009
).
Blood lactate is an important energy source for the human brain
.
Journal of Cerebral Blood Flow and Metabolism
,
29
,
1121
1129
.
Walker
,
M. P.
,
Brakefield
,
T.
,
Morgan
,
A.
,
Hobson
,
J. A.
, &
Stickgold
,
R.
(
2002
).
Practice with sleep makes perfect: Sleep-dependent motor skill learning
.
Neuron
,
35
,
205
211
.
Ziemann
,
U.
,
Reis
,
J.
,
Schwenkreis
,
P.
,
Rosanova
,
M.
,
Strafella
,
A.
,
Badawy
,
R.
, et al
(
2015
).
TMS and drugs revisited 2014
.
Clinical Neurophysiology
,
126
,
1847
1868
.