Paired-pulse transcranial magnetic stimulation (TMS) is used to measure the excitability of interhemispheric inhibition (IHI) between the hand areas of the two motor cortices. It varies from person to person, and is highly predictive of individual differences in callosal anatomy (fractional anisotropy) and even motor behaviour, e.g. the amount of involuntary electromyographic (EMG) ‘mirroring’ in one hand during rapid contraction of the other. The present experiments tested whether it also predicts how well individuals can improve motor performance in a task involving the two hands. Healthy participants were given 100 trials to maximize the initial acceleration of a ballistic finger movement made with one hand while trying to maintain a tonic low level of EMG activity in the other hand. Initially, each movement was accompanied by additional unwanted EMG mirroring in the other hand. However, after practice, participants had on average increased acceleration by approximately one-third without changing the amount of EMG mirroring in the contralateral hand; indeed, in some individuals EMG mirroring activity declined. TMS measures showed that there was an increase in corticospinal excitability in the trained hemisphere, but there was no change in the excitability of short- or long-latency IHI from the trained to non-trained hemisphere. Nevertheless, in each individual, the baseline (pre-practice) excitability of short-latency IHI was highly predictive (r = 0.65; P = 0.0019) of the change in EMG mirroring. The implication is that a physiological measure of brain excitability at rest can predict behaviour in response to training.
It is well known that there is considerable variation between individuals in the response to many non-invasive brain stimulation protocols involving transcranial magnetic stimulation (TMS) or transcranial direct current stimulation (TDCS). Recently, several authors have reported that these can correlate well with individual differences in brain anatomy and even behavioural task performance. For example, the excitability of interhemispheric inhibition (IHI) between the motor cortex hand areas correlates with measures of fractional anisotropy in the region of the corpus callosum carrying connections between the two hemispheres (Wahl et al., 2007; Fling & Seidler, 2011). Similarly, differences in the paired-pulse TMS interactions between ventral premotor and primary motor cortex (M1) during an action selection task correlate with fractional anisotropy of white matter fibres linking the two areas (Boorman et al., 2007). At a behavioural level, IHI correlates with the amount of involuntary electromyographic (EMG) activity in one hand, i.e. EMG mirroring, when people make a rapid or constant forceful contraction of the other hand (Hübers et al., 2008; Fling & Seidler, 2012). Finally, the reduction in levels of γ-aminobutyric acid (GABA) as measured by magnetic resonance spectroscopy produced by anodal TDCS of the motor cortex correlates with an individual's capacity to learn a novel motor task (Stagg et al., 2011a,b).
In the present experiments we tested whether measures of IHI would be predictive of an individual's capacity to adapt behaviour in a simple ballistic motor learning task. Volitional unimanual movements are frequently accompanied by subtle concomitant involuntary activation of the homologous contralateral muscles, which is detectable in healthy human subjects using surface EMG, i.e. EMG mirroring (Giovannelli et al., 2006, 2009; Hübers et al., 2008). In healthy humans, this effect is thought to be due to unwanted activation of the ‘relaxed’ M1, which then drives the mirror EMG (Addamo et al., 2007; Cincotta & Ziemann, 2008). This is compatible with the finding that individuals with the most excitable IHI have the least mirror EMG: more profound inhibition from the active hemisphere suppresses involuntary activation of the ‘relaxed’ hemisphere. The question we ask here is whether the degree of EMG mirroring can be reduced by practice, and whether this relates to baseline measures of IHI or practice-related changes of IHI. Participants made rapid, forceful abduction movements of the index finger of one hand while maintaining a constant low-level contraction of the opposite hand. They were then given 100 practice movements in which they were instructed to maximize the acceleration of the index finger while minimizing any unwanted (mirror) EMG in the other hand. In half of the participants we gave online feedback of the focal task by displaying the acceleration traces of the finger movements on a PC screen (i.e. feedback-provided motor task) in order to encourage participants to increase acceleration as much as possible, while in the other half no feedback was given (i.e. feedback-deprived motor task group), although the instruction to increase acceleration was the same. This ensured that although the first dorsal interosseous (FDI)MIRROR background contraction in the two sessions was the same, there was a range of performance change across individuals on the contralateral side.
Our hypothesis was that practice would focus the motor output to the corresponding M1 and therefore reduce the involuntary spread of contralateral muscle activation, i.e. physiological EMG mirroring. Given the functional relevance of inhibitory interhemispheric pathways in preventing involuntary EMG mirroring and overt mirror movements during focal contraction of one hand (Mayston et al., 1999; Wahl et al., 2007; Hübers et al., 2008; Giovannelli et al., 2009), we tested whether any motor practice-related changes in EMG mirroring would be reflected in baseline measures of IHI or practice-related changes of IHI. Our prediction was that task acceleration would increase while EMG mirroring decreased, and that the extent of the latter would correlate with the magnitude of baseline IHI from the training to the mirror M1. Hence, individuals with greater baseline IHI would be better able to prevent the spread of contralateral motor overflow during the task. An alternative explanation is that reduced EMG mirroring does not depend on baseline IHI but on the ability to increase IHI during the task. In this case we would expect that the greater the increase in IHI, the better the reduction in EMG mirroring.
Materials and methods
Twenty-six subjects (10 females; mean age 28.90 ± 4.65 years, age range: 21–37 years) participated in the study. All subjects were right-handed, scoring above 70 on the Edinburgh Handedness Inventory (Oldfield, 1971), had no history of neurological or psychiatric disorders, and were not taking any CNS active drugs at the time of experiments. None of the subjects had ever engaged in professional training involving the hands. All subjects gave their informed consent to the experimental procedures, which were approved by the local Ethics Committee and conducted in accordance with published international safety recommendations (Rossi et al., 2009) and regulations laid down in the Declaration of Helsinki.
Surface EMG activity and motor-evoked potentials (MEP) elicited by TMS were recorded from both FDIs [i.e. the dominant (FDITASK) and the non-dominant (FDIMIRROR); Figs 1 and 2] using silver-chloride surface cup electrodes (9 mm in diameter), taped in a belly tendon arrangement with the active electrode centred over the muscle belly of the FDIs and the reference electrode over the metacarpo-phalangeal joint of the respective index finger (inter-electrode distance ~2 cm). The EMG raw signals were amplified (1000 ×) and band-pass filtered (20 Hz–2 kHz) by a Digitimer D360 amplifier (Digitimer, Welwyn Garden City, Hertfordshire, UK), digitized at a sampling rate of 4 kHz by an analogue-to-digital interface (Micro 1401; Cambridge Electronic Design, UK) and stored on a laboratory computer for off-line analyses. The EMG traces were analysed using customized Signal® version 4.00 (Cambridge Electronic Design, UK) and matlab® version 7.1 (The MathWorks, Natick, USA) software.
Participants were comfortably seated in a chair with the arms slightly abducted from the trunk (~45–50 °), the elbow flexed (~90 °) and both forearms in prone position. The right forearm and wrist were tightly attached on the armrest with straps. The right wrist was kept in a neutral position. The right thumb was slightly abducted, and fingers 2–5 adducted extended at the inter-phalangeal and flexed at the metacarpo-phalangeal joints (~70–80 °).
The motor training task was adopted from previous studies (Agostino et al., 2007, 2008). Participants were first asked to keep their dominant index finger extended and in line with the forearm. Participants were then instructed to produce ballistic finger abductions of their dominant index finger, so as to achieve the highest initial acceleration possible, in response (but not to react immediately) to a ‘go’ signal, given randomly at ~0.2 Hz, and to return to the neutral position. While performing fast abductions with their dominant index finger, participants were instructed to pinch with the 1st and 2nd finger a cylindrical body in order to isometrically recruit at ~5–10% of the maximal voluntary contraction in the contralateral FDIMIRROR (Fig. 2A; Giovannelli et al., 2006; Hübers et al., 2008). The maintenance of a constant level of isometric contraction in the FDIMIRROR was monitored online by displaying the continuous EMG activity on a PC screen in front of the participants. In each training session 100 movements were collected; 10 consecutive movements were considered as a trial and averaged (Fig. 2A). A rest interval of 10 s was left between trials to avoid fatigue (Fig. 2A). Before starting the motor training, one practice trial was permitted for the participants to become familiar with the experimental setup. In the present study we adopted a simple ballistic motor task with no real requirements for accuracy, just acceleration, as it fitted in well with the possibility to explore the effects of motor practice on the EMG mirroring activity related to fast finger movements. Moreover, although the after-effect of a simple ballistic motor task has been clearly described in terms of changes of corticospinal excitability, i.e. cortical plasticity (Classen et al., 1998; Muellbacher et al., 2001, 2002; Agostino et al., 2007, 2008), it is less clear whether similar tasks also modify IHI (Hortobágyi et al., 2011). Even though a number of studies exploring the asymmetry of the EMG mirroring in healthy humans reported stronger EMG mirroring during voluntary movements of the non-dominant hand (Armatas et al., 1996; Uttner et al., 2007), no difference between the two hands (Hübers et al., 2008) has also been described. Because motor training-related after-effects have been studied more often at the level of the dominant M1 (Classen et al., 1998; Muellbacher et al., 2001, 2002; Agostino et al., 2007, 2008), we selected the dominant M1 as the M1TASK and the non-dominant M1 as M1MIRROR, respectively.
TMS was delivered to both M1s [i.e. the dominant (M1TASK) and non-dominant (M1MIRROR), respectively; Fig. 1] using two Magstim 2002 magnetic stimulators with a monophasic current waveform (Magstim, Carmarthenshire, Wales, UK). Each magnetic stimulator was connected to a focal figure-of-eight-shaped coil (outer diameter of each wing, 70 mm). The intersections of the coils were placed tangentially to the scalp with the handles pointing backward and laterally at ~45 ° angle away from the midline, in this way the monophasic current induced in both M1s was approximately perpendicular to the line of the central sulcus resulting in a predominantly trans-synaptic activation of the corticospinal system (Kaneko et al., 1996; Di Lazzaro et al., 2004). During the experiments, participants wore a swimming cap and the hot spot positions of both FDIs, i.e. the optimal scalp positions for eliciting MEPs of maximal amplitudes in the contralateral FDI, defined as the M1TASK and the M1MIRROR respectively, were marked on it. TMS was delivered with the FDIs at complete rest as confirmed by visual inspection of the EMG record in the 200-ms preceding stimulation. Traces with background EMG activity exceeding 50 μV in this 200-ms window were excluded from analysis (~1% of trials).
Corticospinal excitability was tested delivering single-pulse TMS, on both the M1TASK and the M1MIRROR hot spots (Fig. 1). As a measure of corticospinal excitability on the M1TASK we used the resting motor threshold (RMT), determined to the nearest 1% of the maximum stimulator output (MSO), defined as the minimal stimulus intensity required to produce MEPs larger than 50 μV peak-to-peak amplitude, in the contralateral FDITASK, in at least five out of 10 consecutive trials. As a measure of corticospinal excitability on the M1MIRROR, we adjusted the stimulator intensity to produce, at rest, MEPs of ~1 mV in peak-to-peak amplitude (1 mV-MEP) in the contralateral FDI MIRROR. The measurements of RMT and 1 mV-MEP, over the M1TASK and M1MIRROR, respectively, were followed by measurement of IHI targeting M1MIRROR.
IHI was measured by means of a standard paired-pulse TMS protocol (Ferbert et al., 1992; Hübers et al., 2008; Ni et al., 2009). The conditioning stimulus (CS) and the test stimulus (TS) were delivered over the M1TASK and the M1MIRROR hot spots, respectively (Fig. 1). The CS intensity was adjusted in 10% steps from 110 to 150% of RMT. The TS intensity was set at 1 mV-MEP. Five blocks of IHI measures (one block for each CS intensity: 110–150% of the RMT using a constant TS intensity of 1 mV-MEP) were collected. To investigate short- and long-latency IHI (s-IHI and l-IHI), 12 and 30 ms interstimulus intervals (ISI) were selected (Ferbert et al., 1992; Chen et al., 2003; Ni et al., 2009). It has been suggested that by studying s-IHI and l-IHI, at 12 and 30 ms ISIs, it is possible to test interhemispheric circuits that are supposed to be mediated by different populations of GABAergic interneurons (Irlbacher et al., 2007). Only s-IHI, however, is thought to play a predominant role in the suppression of the EMG mirroring during fast finger movements (Duque et al., 2007; Hübers et al., 2008).Twenty paired-pulse (CS + TS) trials (10 trials each for s-IHI and l-IHI) were randomly intermixed every 4–6 s with 10 trials using TS alone (30 trials in total for each block performed before and after the motor task, 300 trials in total). During this test, high-intensity CSs induced MEPs in the FDITASK, and this was used to plot the input–output properties of the M1TASK.
TMS measures of corticospinal excitability and IHI were collected before and immediately after the motor training task. If the motor task changed RMT or 1 mV-MEP, then the stimulus intensities were readjusted to compensate (Hübers et al., 2008).
The acceleration of index finger abduction was recorded with an accelerometer (model ACL300, voltage sensitivity 100 mV/g; Biometrics, UK) firmly taped to the distal phalanx of the right index finger. The signal was amplified (model ACL300, Biometrics, UK), digitized (A/D rate 4 kHz, CED Micro 1401) and stored in a laboratory computer for online visual display. Later off-line analyses on the acceleration traces were performed using customized Signal® version 4.00 (Cambridge Electronic Design, UK). The first acceleration peak of each index finger abduction was measured in amplitude and expressed in g.
EMG mirroring was measured as detailed elsewhere (Mayston et al., 1999; Giovannelli et al., 2006, 2009; Hübers et al., 2008). The EMG traces from both the FDITASK and the FDIMIRROR were single trial DC corrected and rectified offline. A reference cursor was set to identify the onset of the voluntary EMG bursts onset in the FDITASK (Fig. 2B). The EMG mirroring was quantified according to the following formula:
where α is the mean EMG amplitude (mV) in the FDIMIRROR during the 50-ms window following the FDITASK burst onset, and β is the mean background EMG activity amplitude (mV) in the FDIMIRROR in the time window of 1000 ms preceding the FDITASK burst onset (Giovannelli et al., 2006; Hübers et al., 2008). Thus, a value of 0% indicates absence of EMG mirroring, and a value of 100% indicates that the EMG mirroring is twice as high as background EMG (Fig. 2B). The background EMG activity measurements allowed us to evaluate any contribution of non-specific changes in background muscular contraction to the change in EMG mirroring throughout the motor task. In a series of control measurements, the reference cursor was randomly set between the voluntary EMG burst onsets of the FDITASK. These control measurements showed that the mean ± SD control measure of the EMG mirroring was 0.68 ± 5.2%. EMG mirroring was defined as activity that was more than 2SD (i.e. 11.27%, cut-off value) above the level of background EMG activity.
The MEP amplitude was measured within a time window of 20–40 ms after the TMS artefact. IHI was calculated for each ISI (12 and 30 ms) in each block (CS intensity: 110–150% RMT) by expressing the mean peak-to-peak amplitude of the conditioned test MEP in the paired-pulse trials as a percentage of the test MEP (Ferbert et al., 1992; Hübers et al., 2008).
Group differences in baseline EMG mirroring, background EMG activity and acceleration peak (as calculated in the 1st trial measurements) and TMS parameters (as calculated before the motor training) were evaluated using Student's t-tests.
To evaluate the course of EMG mirroring and background EMG activity and acceleration peak during the motor training task, a normalization procedure was used. Absolute values of each parameter (from the 2nd to the 10th trial of the training) were normalized to their corresponding baseline (1st trial). These data were then entered into separate repeated-measures analysis of variance (anova) using EMG mirroring and background EMG activity and acceleration peak as dependent variables, TRIAL NUMBER (nine levels: 2nd to 10th trial) as within-subject factor and FEEDBACK (two levels: feedback vs. no feedback) as between-group factor.
Motor task-related changes in TMS measures of corticospinal excitability (MEPs amplitude) were evaluated using a repeated-measures anova with within-subject factors CS INTENSITY (five levels: 110–150% RMT) and MOTOR TRAINING (two levels: pre-training vs. post-training). To evaluate the s- and l-IHI measures, the within-subject factor ISI (two levels: 12 vs. 30 ms) was included. To evaluate group differences the between-group factor FEEDBACK (two levels: feedback vs. no feedback) was also included.
Pearson's product-moment correlation coefficient was calculated to evaluate our a priori hypotheses, i.e. possible associations between practice-related changes of EMG mirroring, baseline maximal s-IHI and l-IHI, and overall changes in either s-IHI or l-IHI. Finally, we tested whether changes of EMG mirroring correlated with practice-related changes (%) of other parameters, i.e. acceleration peak of the ballistic movement and average corticospinal excitability of the trained hemisphere (see Results). Tukey honest significant difference test was used for the post hoc analysis in the anovas. Unless otherwise stated, all results are indicated as mean values ± standard error of the mean (SEM). In all tests the level of significance was set at P < 0.05. Bonferroni's correction was applied to multiple correlations (e.g. Bonferroni-adjusted alpha levels: 0.05/6 = 0.0083).
None of the participants reported fatigue or adverse effects during or after the experiments. In none of the experiments did visible mirror movements accompany the EMG mirroring. There were no ipsilateral MEP responses to TMS.
EMG mirroring, background activity and acceleration peak
The average baseline EMG mirroring was 19.4 ± 3.4% (ranging from 38.6 to −4.7%) and 40.3 ± 3.6% (ranging from 144.6 to 3.5%) for the feedback-deprived and feedback-provided motor task sessions, respectively. No significant difference in baseline EMG mirroring was found between the two sessions (P = 0.08). Six subjects (4/13–30.76% of subjects participating in the feedback-deprived motor task session; and 2/13–15.38% of subjects participating in the feedback-provided motor task session) had mean baseline EMG mirroring below the cut-off value (see Materials and methods). Because the aim of this study was to evaluate the practice-related effects on EMG mirroring, we excluded these six subjects. The remainder of the analysis was therefore conducted on nine and 11 subjects participating in the feedback-deprived and feedback-provided motor task sessions, respectively. The average baseline background EMG mirroring was the same in both sessions, being 235 ± 78 μV (ranging from 121 to 419 μV) and 270 ± 33 μV (ranging from 113 to 387 μV) for the feedback-deprived and feedback-provided motor task sessions, respectively (P = 0.51). The average baseline acceleration peak was slightly different between sessions (P = 0.002); it was 0.73 ± 0.06 g (ranging from 0.46 to 1.06 g) and 1.13 ± 0.08 g (ranging from 0.67 to 1.80 g) for the feedback-deprived and feedback-provided motor task sessions, respectively.
Figure 3 (upper panel) depicts the course of the baseline normalized acceleration peak throughout the motor task in the feedback-deprived and feedback-provided sessions. Repeated-measures anova showed a significant effect of MOTOR TRAINING (F8,144 = 3.11, P = 0.002), indicating that participants increased their acceleration during training. There was no effect of FEEDBACK (F1,17 = 0.00, P = 0.97), suggesting that the two groups learned at similar rates. There was a trend towards a significant interaction MOTOR TRAINING × FEEDBACK (F8,144 = 1.98, P = 0.053), which was probably caused by the tendency of performance to plateau in the feedback-deprived sessions.
The middle panel of Fig. 3 shows that there was a trend toward a reduction in EMG mirroring from blocks 1 to 10 in both the feedback-deprived and feedback-provided sessions (−34.1 and −30.9%), although anova disclosed no significant effect of MOTOR TRAINING (F8,136 = 1.26, P = 0.27), FEEDBACK (F1,17 = 0.06, P = 0.80), or MOTOR TRAINING × FEEDBACK interaction (F8,136 = 0.64, P = 0.74).
Finally, there was no significant change in background EMG activity of FDIMIRROR throughout the motor task [Fig. 3, lower panel; MOTOR TRAINING (F8,136 = 0.29, P = 0.96); FEEDBACK (F1,17 = 0.00, P = 0.97); MOTOR TRAINING × FEEDBACK (F8,136 = 0.92, P = 0.50)]. Given there were no significant interaction terms in the lower two panels of Fig. 3, we can conclude that training had the same effect on EMG mirroring and background EMG in both the feedback-provided and feedback-deprived sessions.
Pre-task measurements of RMT50 μV (in the M1TASK) and of 1 mV-MEP (in the M1MIRROR), respectively, were 37.1 ± 4.4 and 44.4 ± 4.8% of MSO for the feedback-deprived motor task session, and 39.1 ± 1.9 and 48.4 ± 6.6% of MSO for the session with feedback. They did not differ between sessions and were unchanged after motor practice (all P > 0.05). As shown in Fig. 4, however, the input–output properties of M1TASK increased after practice, indicating an increase in excitability of the trained hemisphere. This was confirmed by a repeated-measures anova, which showed a significant effect of MOTOR TRAINING (F1,18 = 9.91, P = 0.005) and CS INTENSITY (F4,72 = 20.05, P < 0.0001), but no significant effect of FEEDBACK (F1,18 = 0.06, P = 0.80) or any significant interaction terms between the main factors [CS INTENSITY × MOTOR TRAINING (F4,72 = 0.67, P = 0.61); CS INTENSITY × FEEDBACK (F4,72 = 0.22, P = 0.92); MOTOR TRAINING × FEEDBACK (F1,18 = 0.57, P = 0.46); CS INTENSITY × MOTOR TRAINING × FEEDBACK (F4,72 = 0.38, P = 0.82)]. We conclude that motor training increased excitability of M1TASK, independent of the type of feedback (Fig. 4).
Values of s-IHI and l-IHI obtained at different CS intensities are shown in Fig. 5. Repeated-measures anova revealed a significant main effect of CS INTENSITY (F4,72 = 19.44, P < 0.0001), confirming that the mean magnitude of s-IHI and l-IHI increased with increasing CS intensity. Conversely, the main factors FEEDBACK, MOTOR TRAINING and ISI were not significant (F1,18 = 2.72, P = 0.11; F1,18 = 1.46, P = 0.24; and F1,18 = 0.75, P = 0.39, respectively), and there were no significant interactions between the main factors [FEEDBACK × MOTOR TRAINING (F1,18 = 0.08, P = 0.78); FEEDBACK × ISI (F1,18 = 0.32, P = 0.58); MOTOR TRAINING × ISI (F1,18 = 0.52, P = 0.48); FEEDBACK × CS INTENSITY (F4,72 = 1.20, P = 0.31); ISI × CS INTENSITY (F4,72 = 1.39, P = 0.24); MOTOR TRAINING × CS INTENSITY (F4,72 = 1.13, P = 0.35); FEEDBACK × ISI × MOTOR TRAINING (F1,18 = 0.03, P = 0.85); FEEDBACK × ISI × CS INTENSITY (F4,72 = 1.07, P = 0.37); FEEDBACK × MOTOR TRAINING × CS INTENSITY (F4,72 = 0.07, P = 0.99); ISI × MOTOR TRAINING × CS INTENSITY (F4,72 = 0.70, P = 0.59); FEEDBACK × ISI × MOTOR TRAINING × CS INTENSITY (F4,72 = 0.08, P = 0.98)]. Thus, neither feedback-deprived nor feedback-provided motor training had any effect on s-IHI and l-IHI (Fig. 5).
We combined data from both feedback-deprived and feedback-provided motor training sessions as they had behaved the same way in all preceding anovas. As outlined in the Introduction, we had two hypotheses to test. One was that the reduced EMG mirroring would be related to the baseline IHI; the other was that the reduction in EMG mirroring would correlate with an increase in IHI after practice.
There was a significant correlation between changes in EMG mirroring and the individual maximal s-IHI at baseline (r = 0.65, P = 0.0019; Fig. 6), indicating that the greatest reduction in EMG mirroring was associated with the most effective individual maximal s-IHI. The correlation between changes in EMG mirroring and the average baseline l-IHI was not significant (r = 0.25, P = 0.27; Fig. 6). There was no correlation between overall changes in either s-IHI or l-IHI and practice-related changes in EMG mirroring (r = 0.36, P = 0.11; r = 0.11, P = 0.63).
As outlined in the Materials and methods, we also tested whether the practice-related changes in EMG mirroring were related to the changes in acceleration of the ballistic movement or to the changes of the average corticospinal excitability of the trained hemisphere. There was no correlation between changes in EMG mirroring and acceleration peak (r = 0.32, P = 0.16). Similarly, there was no correlation between changes in EMG mirroring and average corticospinal excitability of the trained hemisphere (r = −0.0081, P = 0.97).
In the present study we found that, as reported by others (Classen et al., 1998; Muellbacher et al., 2001; Agostino et al., 2007, 2008), subjects improved performance in the trained task. Furthermore, this happened even though there was no overall change in EMG mirroring, and even a tendency for it to decline. Physiologically there was an increase in the excitability of corticospinal output from the trained hemisphere, but there was no change in IHI from the trained to the contralateral hemisphere. However, individual changes in EMG mirroring did relate to the basal amount of s-IHI, i.e. the greater the basal levels of s-IHI the greater the reduction in EMG mirroring. The conclusions from this are: (i) that corticospinal excitability and cortico-cortical (interhemispheric) excitability can be modulated independently by motor training, even though they may share some of the same circuitry (Avanzino et al., 2007); and (ii) basal physiology measures of s-IHI give an indication of the overall extent to which EMG mirroring modification is possible, i.e. that the baseline s-IHI is a key factor that determines how successfully participants can learn to focus their motor commands on the task being trained and prevent overflow to the opposite hemisphere.
The reduction in EMG mirroring we observed during motor training in individuals with greater baseline s-IHI was not explained by a change in the level of background EMG activity in the tonically contracting FDIMIRROR as this was constant. Nor is it likely to reflect any fatigue that might have been caused by training as fatigue is known to increase rather than decrease EMG mirroring (Cincotta & Ziemann, 2008). There was also no correlation between the reduced amount of EMG mirroring and the improvement in motor performance, i.e. acceleration peak, in the trained hand. Thus, reduced mirror activity was not a direct consequence of a larger EMG burst from the trained hemisphere causing increased IHI onto the non-trained hemisphere (Hinder et al., 2010). We can also exclude the possibility that the effects are related to attention as they were not influenced by the presence of feedback, which potentially influences the attentional resources. Finally, there was no correlation between the practice-related changes of EMG mirroring and corticospinal excitability of the trained hemisphere. We conclude that reduction of EMG mirroring is a process that is separate from improving performance of the trained hand and practice-related corticospinal plasticity of the trained hemisphere.
As stated above, although there was no overall change of s- and l-IHI after training, the individual maximal level of s-IHI, but not the individual maximal level of l-IHI, prior to training correlated with the reduction in mirror activity that occurred during training. Thus, the present results suggest that the motor training-related effects on the EMG mirroring are specific to one interhemispheric motor pathway, mediated by a population of GABAergic interneurons (Irlbacher et al., 2007), which are thought to play a predominant role in the suppression of EMG mirroring during fast finger movements (Duque et al., 2007; Hübers et al., 2008). We speculate that the excitability of s-IHI measured at rest is a measure of ‘resource’, that is, it gives an indication of what level of IHI is available to the system to employ during voluntary movement. In fact, because IHI is directly related to the structural measures (magnetic resonance imaging fractional anisotropy) of the anatomy of the mid-portion of the corpus callosum (Wahl et al., 2007; Koerte et al., 2009), it may give an indication of the physical limits of IHI. Thus, individuals with greater s-IHI at rest will have a greater potential for controlling EMG mirror activity during training of intentional movement. In this scheme, motor practice does not reduce EMG mirroring by increasing the sensitivity of IHI. Instead, EMG mirroring may decline because the motor command is better targeted at the task being performed. The more ‘resource’ that there is available in s-IHI, the more efficiently this focussing can reduce EMG mirroring activity. Although recent studies have shown that there are similar structure–function relationships when examining GABA-A-mediated IHI, i.e. s-IHI (Wahl et al., 2007), as those found with l-IHI (Fling & Seidler, 2012), the present results confirm that only s-IHI has a functional role in the suppression of the EMG mirroring during fast finger movements (Duque et al., 2007; Cincotta & Ziemann, 2008; Hübers et al., 2008).
In previous studies it has been shown that IHI from the trained to the untrained motor cortex can show plastic changes, mainly seen as a reduction of IHI (Shim et al., 2005; Perez et al., 2007; Camus et al., 2009; Hortobágyi et al., 2011). In the present study, however, we did not detect any practice-related changes in IHI. Methodological differences between our experiments and those of previous study could account for our different findings. In the present study we investigated changes in the IHI targeting the untrained motor cortex after a simple ballistic motor learning task, while previous studies examined different tasks, involving force production (Shim et al., 2005) or motor sequence learning, i.e. the serial reaction time task (Perez et al., 2007; Camus et al., 2009). It is therefore possible that the variable cognitive load or attentional demand involved in different forms of motor learning may influence the results. Additionally, the lack of change in IHI could be due to other specific features of the present experiment, such as the relatively short duration of the motor task. In this regard it is worth noting that Hortobágyi et al. (2011) observed a less profound IHI after 1000 submaximal voluntary contractions of the FDI. Finally, an alternative hypothesis is that our results were influenced by the constant isometric force produced by the left hand during training, as volitional activity in one hand can modulate IHI in the homologous muscle of the contralateral limb (Giovannelli et al., 2009; Hinder et al., 2010).
In theories of optimal motor control (Todorov, 2004), the motor system attempts to achieve a desired level of performance at minimal cost. In the present experiments we might then speculate why motor training leads to reduced EMG mirroring, as it has no direct effect on the task itself, which is to increase acceleration of the opposite hand. One possible explanation is that it is a result of a very generalized ‘cost function’, which is to minimize all activity associated with the task, whether it is relevant or irrelevant to task performance. Effectively this would reduce all overflow of activity that was not relevant to the task. Another explanation is that reduced EMG mirroring is secondary to the motor system's attempts to maximize some other, task-relevant, function, such as focussing the motor command onto only those motor outputs that are strictly required to produce the required movement.
The present study specifically examined the effects of brief motor practice on EMG mirroring, and therefore we do not know the extent to which the effects would carry over to other stages of motor learning, such as consolidation (Brashers-Krug et al., 1996; Muellbacher et al., 2002) or long-term retention (Reis et al., 2009), or whether practice-related changes of EMG mirroring in one hand are associated with similar changes in the untrained as well as in the trained hand, a phenomenon referred to as intermanual transfer (Perez et al., 2007; Camus et al., 2009). It is also important to note that in the present study we adopted a simple, ballistic movement of the finger with no real requirements for accuracy, just acceleration. Hence, we do not know whether similar results would occur with more specific training, i.e. producing specific sub-maximal force patterns, or timing-specified movements. In addition, in the present study we evaluated only one of a range of possible inhibitory interactions between the hemispheres. It is likely that interactions between M1 and other areas, such as the premotor areas (including the supplementary motor area and the anterior cingulum) and cerebellum might also contribute to reduce EMG mirroring (Brinkman, 1984; Giovannelli et al., 2006). Basal ganglia are also thought to be involved in supporting the cortical networks responsible for non-mirror transformation of voluntary movements (Giovannelli et al., 2006). Whether such structures might also play a role in reduced EMG mirroring remains an open question. Finally, we did not record H-reflex or F-waves to monitor changes of spinal motor neuron excitability after the motor task, and therefore we cannot exclude the possibility that changes of spinal cord excitability influenced the training-related reduction in EMG. A comprehensive evaluation, however, of all these neurophysiological measures was beyond the aim of the present study, and a more detailed exploration of these possibilities requires further investigations.
In conclusion, our findings show that motor training of one hand reduces the level of mirror activity in the opposite hand depending on the pre-training level of excitability in interhemispheric pathways connecting the two M1 cortices. However, this does not exclude possible contributions from other cortical motor areas or the basal ganglia, which also may be important. The main implication of the relationship between baseline IHI and behaviour suggests that a physiological measure of brain excitability at rest can predict behaviour in response to training. Second, the present study provides novel information on the complex relationships between motor performance and IHI, and indicates that increased IHI may be either detrimental (Fling & Seidler, 2012) or beneficial to motor performances, according to different contexts. Third, the present study provides additional data to help understand the factors influencing the practice-related plastic changes of the interhemispheric pathways. These may well depend on the precise nature of the task being studied, and are not present in all types of training. Finally, increased understanding of the physiological mechanisms involved in suppression of the EMG mirroring and mirror movements could theoretically help us to develop interventions to avoid the spread of unwanted motor overflow in pathological conditions.
Matteo Bologna was supported by the European Neurological Society (ENS).