Synchronization of body movements to an external beat is a universal human ability, which has also been recently documented in nonhuman species. The neural substrates of this rhythmic motor entrainment are still under investigation. Correlational neuroimaging data suggest an involvement of the dorsal premotor cortex (dPMC) and the supplementary motor area (SMA). In 14 healthy volunteers, we more specifically investigated the neural network underlying this phenomenon using a causal approach by an established 1-Hz repetitive transcranial magnetic stimulation (rTMS) protocol, which produces a focal suppression of cortical excitability outlasting the stimulation period. Synchronization accuracy between rhythmic cues and right index finger tapping, as measured by the mean time lag (asynchrony) between motor and auditory events, was significantly affected when the right dPMC function was transiently perturbed by “off-line” focal rTMS, whereas the reproduction of the rhythmic sequence per se (inter-tap-interval) was spared. This approach affected metrical rhythms of different complexity, but not non-metrical or isochronous sequences. Conversely, no change in auditory–motor synchronization was observed with rTMS of the SMA, of the left dPMC or over a control site (midline occipital area). Our data strongly support the view that the right dPMC is crucial for rhythmic auditory–motor synchronization in humans.
Rhythmic motor entrainment refers to the ability to synchronize body movements to an external rhythm (Zatorre et al. 2007; Fitch 2009). This ability is particularly evident in music performance and dance and is closely related to auditory cues, as synchronization to rhythmic visual stimuli is less accurate (Repp and Penel 2004). It represents a universal cross-cultural behavior (Levitin and Tirovolas 2009), which spontaneously emerges even in nonmusicians (Drake et al. 2000). A natural predisposition for rhythmic engagement is suggested from recent kinematic data showing rhythmic movement in response to music and other metrically regular sounds in preverbal infants (Zentner and Eerola 2010). Moreover, experimental evidences indicate that rhythmic motor entrainment is not a human prerogative, because spontaneous synchronization of movements to the tempo of a musical beat has been observed in vocal mimicking nonhuman animals, such as parrots (Patel et al. 2009; Schachner et al. 2009), but not in nonvocal learning species that are by far closer to humans, such as nonhuman primates (Schachner et al. 2009). Hence, it has been hypothesized that, in these species, rhythmic auditory–motor entrainment evolved as a by-product of selection for vocal mimicry (Patel et al. 2009; Schachner et al. 2009). This ability requires a precise temporal coordination between the perception of predictable stimuli and motor responses (Repp 2005).
Functional neuroimaging data suggest that the neural processes underlying rhythmic auditory–motor entrainment rely on a widely distributed neural network involving basal ganglia, cerebellum, and neo-cortical areas including auditory regions of the temporal lobe, premotor cortex (PMC) and supplementary motor area (SMA) (Thaut 2003; Chen et al. 2006, 2008a, 2008b, 2009; Grahn and Brett 2007; Zatorre et al. 2007; Thaut et al. 2008; Schwartze et al. 2011). Specifically, recent fMRI studies show that dorsal PMC (dPMC) activation is modulated by the metrical structure of the rhythm and increases with the temporal complexity of the auditory cue (Chen et al. 2006, 2008a, 2008b, 2009). Although these correlational findings suggest a specific role of this cortical area in auditory–motor interaction, the assumption that functional activation necessarily indicates causality still needs to be verified. In addition, the specific contribution of dPMC and SMA to different aspects of rhythmic motor entrainment has been not clarified yet.
The aim of the present study was to address these issues using a well-established low-frequency repetitive transcranial magnetic stimulation (rTMS) protocol, which produces a transient interference with the function of the stimulated cortical area that outlasts the application of the rTMS trains by tens of minutes (Chen et al. 1997). On 4 separate days, rTMS was delivered to either dPMC, to the SMA, and over a midline occipital control site of 14 healthy right-handed participants. In each experimental session, subjects were asked to synchronize their right index tapping to rhythmic auditory cues of different complexity and to continue reproducing the rhythm after cessation of the external cues. All tasks were performed at the baseline and immediately after the low-frequency rTMS train. Our prediction was that a worsening of auditory–motor interaction should be observed after rTMS, if the stimulated brain area was involved in the neural network underlying this behavior.
Materials and Methods
Fourteen healthy volunteers (9 women; mean age 23.9 years, range 21–30 years) with normal hearing and no history of implanted metal devices or neurological disease were included in the study. All subjects were naïve for TMS techniques and blinded to the purpose of the study. No participant had current formal musical training. Two subjects had previous noninstitutional, incomplete formal training. This basic training lasted <1 year and was discontinued at least 8 years before the study. Currently, they rarely play pop-music songs on the piano at amateur level. All participants were right handed according to the Edinburgh handedness inventory (Oldfield 1971) (mean dexterity index 86.4%, range 58–100%). The study was performed according to the Declaration of Helsinki and the local ethics committee approved the use of rTMS. All subjects gave their written informed consent and were asked to report adverse effects experienced during or after rTMS.
Auditory stimuli consisted of rhythmic sequences (Fig. 1A) and were created by the Magix Music Maker 15 software (Magix AG, Berlin, Germany). Simple percussion sounds (woodblock) of 120 ms duration were used to compose the rhythmic sequences. For all sequences, sound parameters (i.e., pitch, intensity, volume) were constant and only the time intervals were manipulated. Therefore, the beat perception emerged uniquely from the temporal structure of the sequences (Povel and Essens 1985).
Rhythmic cues consisted of metrical and nonmetrical sequences (Fig. 1A). An isochronous (ISO) sequence of 40 s duration, in which the auditory cues were repeated at a regular interstimulus interval (ISI) of 250 ms (Fig. 1A), was also included to evaluate whether rTMS effects were specifically related to the patterned temporal structure of the auditory cues. Metrical and nonmetrical sequences were created manipulating the temporal structure of each rhythm. Forty-eight rhythms were selected from the patterns reported by Grahn and Brett (2007). Metrical rhythms were formed by 6 or 7 intervals related with integer ratios of 1:2:3:4, and were simple, with accents occurring at regular intervals (METRICsimple) or complex, with accents occurring at irregular intervals (METRICcomplex). Non-metrical rhythms (NON-METRIC) consisted of 6 or 7 intervals related with noninteger ratios (1:1.4:3.5:4.5). For each rhythm, the minimum time interval was 250 ms, and the duration of the basic sequence was 4 s. The stimuli used in the experimental sessions consisted of 10 consecutive repetitions of the basic 4-s rhythmic sequence. The resulting 40-s-long rhythmic sequences (Fig. 1B) were more similar to the structure of the musical pieces and, therefore, more “ecological,” than the presentation of single 4-s rhythmic sequences. Moreover, preliminary trials showed that auditory–motor synchronization with a single presentation of the target sequence resulted too difficult in some subjects, especially with the more complex auditory stimuli. Four blocks with metrical manipulation including 12 rhythmic stimuli (4 METRICsimple, 4 METRICcomplex, and 4 NON-METRIC) were created and presented to each subject.
Synchronization Continuation Task
All participants took part in 2 separate conditions: 1 ISO condition and 1 metrically manipulated block (patterned rhythmic cues). During the experimental sessions, subjects were seated on a comfortable chair. The auditory stimuli were binaurally presented at comfortable sound level via headphones connected to a personal computer. Subjects were requested to listen to the stimuli and synchronize their movements to the auditory cues by tapping in time with them with the right dominant hand until the end of the sequence (synchronization phase, Fig. 1B). Subjects were instructed to start tapping as soon as they subjectively perceived to have identified the rhythm. When the auditory sequence ended, subjects were asked to continue reproducing the rhythm with no auditory cue (continuation phase, Fig. 1B). During the continuation phase, 10 s in the ISO condition and 3 rhythm repetitions (∼12 s) in the metrical manipulation were, respectively, recorded. The tapping was performed with the index finger on a rigid plastic surface. A microphone positioned underneath the surface was used to record the acoustic correlate of the finger beat. This allowed a real-time recording of each tap. Acoustic tracks of both rhythmic cues and finger beats were analyzed using the Cool Edit Pro 2.0 software (Syntrillium Software Corporation, Phoenix, USA).
Transcranial Magnetic Stimulation Procedure
RTMS was delivered using a Magstim Rapid stimulator with a biphasic current waveform (Magstim Co., UK), connected to a figure-of-eight coil (external diameter of each loop, 9 cm) placed tangentially to the scalp. Prior to rTMS, single magnetic pulses were delivered to the hand area of the primary motor cortex (M1) to establish the optimal position (hot spot) to elicit motor-evoked potentials (MEP) in the contralateral first dorsal interosseous (FDI) muscle. The intensity of rTMS was related to the individual resting motor threshold (RMT) of the FDI, determined over the hot spot. RMT was defined as the minimum stimulus intensity that produced MEP > 50 μV in at least 5 of 10 consecutive trials during muscle relaxation (Rossini et al. 1994).
To stimulate the dPMC, the center of the junction of the coil was moved anterior by 3 cm from the individually determined M1 hot spot, with the handle pointing backwards and 45° away from the midline (Siebner et al. 2003; Cincotta et al. 2004; Rizzo et al. 2004; Giovannelli et al. 2006; Ward et al. 2010). The procedure of using the hot spot as an “anchor point” to identify the site of dPMC stimulation relies on a meta-analysis of various motor activation studies, showing that peak activation in the dPMC, was located on average 2.3 cm anterior to the center of hand movement-related activation in the M1 (Picard and Strick 2001). We positioned the coil at the anterior border of the estimated dPMC area to minimize possible current spread to the adjacent M1 (Siebner et al. 2003, Cincotta et al. 2004; Giovannelli et al. 2006). However, in the last 5 subjects examined, the dPMC location was also estimated by a neuronavigational system (SofTaxic, E.M.S., Bologna, Italy) using digitized skull landmarks (nasion, inion, and 2 preauricular points) and about 50 scalp points provided by a Polaris Vicra optical tracker (Northern Digital, Canada). Coordinates in Talairach space of cortical sites underlying the right (14, −4, 62) and left dPMC (−14, −4, 62) were approximately estimated by the neuronavigator on the basis of an MRI-constructed stereotaxic template (Chen et al. 2008a). In these subjects, dPMC localization estimated by the neuronavigator was 2.0–3.5 cm anterior to the hot spot for the contralateral FDI muscle in the M1 (mean 2.6 and 2.5 cm for the right and left hemispheres, respectively), confirming that placing the center of the coil junction 3 cm anterior to the individual “hot spot” for the contralateral FDI represents an acceptable compromise between the optimal dPMC targeting and the need of minimizing the spread of induced current to the M1. To stimulate the SMA, the center of the junction of the coil was positioned 2.5 cm anterior to the vertex position (Cz of the 10–20 EEG International System) with the handle pointing backwards along the sagittal midline. This location was in line with previous rTMS studies, which report the site of SMA stimulation on average 2–3 cm anterior to Cz (Matsunaga et al. 2005; Del Olmo et al. 2007; Hamada et al. 2009; Mantovani et al. 2010; Tanaka et al. 2010). Finally, as a control condition, the center of the junction of the coil was positioned over Oz of the 10–20 EEG International System, with the handle oriented horizontally.
RTMS parameters used in the present study were in accordance with published international safety recommendations (Rossi et al. 2009).
All subjects participated in 4 experimental sessions on separate days. The interval between 2 consecutive experimental sessions ranged from 1 to 29 days and the mean individual interval was similar across subjects (mean = 7.7 days, range 2.7–14 days). In each session, the synchronization-continuation task was performed before (Baseline) and immediately after the end of the rTMS application (post-rTMS). The stimulation protocol consisted of a train of 15 min of real rTMS at a frequency of 1 Hz at an intensity of 115% RMT. This rTMS protocol has been shown to produce a depression of the excitability of the stimulated cortical area for tens of minutes after the end of stimulation (Chen et al. 1997) and to suppress the movement-related cortical activity when applied to the M1 (Rossi et al. 2000). The synchronization-continuation task was the same in all experimental sessions, whereas the site of rTMS application (right dPMC, left dPMC, SMA, and Oz) varied across the experimental sessions.
In 4 right-handed healthy subjects (2 women; mean age 24.7 years, range 23–27 years) different from those participating in the rTMS experiment, a control experiment was conducted to test whether the slight asynchrony reduction observed after rTMS in some experimental conditions (see Results) could be due to a learning effect. Subjects underwent a single experimental session in which they were asked to perform the synchronization-continuation task with metrically manipulated sequences at baseline and after 15 min without rTMS application.
The experimental phase was preceded by a training phase to familiarize subjects with rTMS and the task (different rhythmic sequences were used in this phase). The order of ISO and metrically manipulated conditions, the order of rhythms in each block, the order of experimental conditions and coupling between blocks of rhythms and experimental conditions were randomized and counterbalanced across subjects.
In each experimental condition, all subjects spontaneously started to synchronize the tapping to the rhythmic cues between the second and the fourth repetition of the basic 4-s rhythmic sequence. For each condition, most of the auditory–motor synchronizations started at the second or third repetition of the basic sequence, whereas a low percentage of them started at the fourth repetition (see Supplementary Table 1). Moreover, 3-way repeated-measures analysis of variance (rmANOVA) with stimulation site (4 levels: rTMS of the right dPMC, left dPMC, SMA, and Oz), metric manipulation (3 levels: METRICsimple, METRICcomplex, and NON-METRIC) and time (2 levels: baseline and post rTMS) as within-subjects factors showed that the number of basic 4-s rhythmic sequence listened before starting to synchronize tapping did not differ across the different stimulation sites (details of the analysis are given in the Supplementary Material). These data show that the whole intra- and interindividual variability in the beginning of tapping occurred between the second and fourth repetition of the basic 4-s rhythmic sequence. Hence, the analysis of auditory–motor synchronization has been performed using the last 5 repetitions and the same number of taps for each subject.
Data of the metrical manipulation condition (METRICsimple, METRICcomplex, and NON-METRIC) and of the ISO condition were separately analyzed. Three dependent variables were evaluated: asynchrony and inter-tap-intervals (ITIs) during the synchronization phase, and ITIs during the continuation phase. The asynchrony (ms) was measured as the absolute value of the time difference between the onset of the auditory cue and the onset of the respective motor response (tap). The ITI (i.e., the time lag between the onsets of 2 consecutive taps) evaluated with respect to the time lag between the target auditory cues is a measure of the accuracy in reproducing the time intervals of the rhythmic sequence. Hence, the deviation (in absolute value) between the ITI and the interval between onsets of the 2 target acoustic cues (inter-cue interval) was calculated and expressed as a percentage of the inter-cue interval (% ITI deviation).
For each metrically manipulated condition (METRICsimple, METRICcomplex, and NON-METRIC), the mean values of asynchrony, ITI deviation in the synchronization phase, and ITI deviation in the continuation phase were calculated. For each measure, the difference between the measures obtained at baseline and after rTMS was expressed as a percentage of the mean value of the 2 measurements (percentage difference). The values obtained were entered in separate two-way rmANOVA with stimulation site (4 levels: rTMS of the right dPMC, left dPMC, SMA, and Oz) and metric manipulation (3 levels: METRICsimple, METRICcomplex, and NON-METRIC) as within-subjects factors.
For the ISO condition, the percentage differences between the mean asynchrony, ITI deviation in the synchronization phase, and ITI deviation in the continuation phase obtained at baseline and after rTMS were entered in separate one-way rmANOVA with stimulation site (4 levels: rTMS of the right dPMC, left dPMC, SMA, and Oz) as within-subjects factor.
To verify whether the 3 dependent variables (asynchrony, ITI deviation in the synchronization phase, and ITI deviation in the continuation phase) at baseline were matched across rTMS sessions, a one-way rmANOVA with stimulation site (4 levels) as within-subjects factor was separately performed for METRICsimple, METRICcomplex, NON-METRIC, and ISO conditions.
Post hoc tests were performed using the Bonferroni test. Significance was set at P < 0.05.
None of the participants reported adverse effects during or after the experimental procedures.
Patterned Rhythmic Cues
For each task, the mean time lag between the onset of the auditory event and the motor response (asynchrony) was used to measure the subject's accuracy in auditory–motor synchronization, whereas the mean percentage deviation between ITI and relative inter-cue intervals expressed the accuracy in the reproduction of the target rhythm. Raw data are summarized in Figure 2 and further details are given in the Supplementary Table S2. At baseline, no statistically significant difference was observed across rTMS sessions with respect to asynchrony (METRICsimple: F3,39 = 0.861, P = 0.429; METRICcomplex: F3,39 = 0.931, P = 0.435; and NON-METRIC: F3,39 = 1.241, P = 0.308), ITI deviation in the synchronization phase (METRICsimple: F3,39 = 1.692, P = 0.184; METRICcomplex: F3,39 = 0.694, P = 0.515; and NON-METRIC: F3,39 = 1.385, P = 0.262), or ITI deviation in the continuation phase (METRICsimple: F3,39 = 0.496, P = 0.687; METRICcomplex: F3,39 = 0.279, P = 0.733; and NON-METRIC: F3,39 = 2.553, P = 0.069).
Two-way rmANOVA showed a significant main effect of “stimulation site” on the percentage asynchrony difference between pre- and post-rTMS performances (F3,13 = 3.846, P = 0.017), whereas the main effect of different rhythmic cues was not significant (F2,13 = 0. 827, P = 0.449). A significant interaction between these two factors (F3,39 = 2.904, P = 0.013) emerged. Post hoc comparisons revealed that, specifically for the METRICcomplex task, rTMS of the right dPMC significantly worsened asynchrony compared with all other stimulation sites (P < 0.001, P = 0.007, and P = 0.013 versus rTMS of left dPMC, of the SMA, and of Oz, respectively) (Fig. 3A). Conversely, no significant asynchrony modification was observed in the METRICsimple and in the NON-METRIC task (Fig. 3A).
For ITI deviation, the two-way rmANOVA did not reveal any significant pre–post rTMS difference either in the synchronization phase (stimulation site: F3,13 = 1.673, P = 0.189; rhythmic cue: F2,13 = 2.185, P = 0.133; interaction: F3,39 = 1.776, P = 0.115) or in the continuation phase (stimulation site: F3,13 = 0.554, P = 0.535; rhythmic cue: F2,13 = 2.358, P = 0.114; interaction: F3,39 = 0.222, P = 0.968).
In the control experiment, when the metrically manipulated tasks were repeated a second time without rTMS, the asynchrony and the ITI deviation were lower than at the baseline. This difference reached significance for asynchrony and ITI deviation during the synchronization phase but not for ITI deviation during the continuation phase (details of data analysis are given in the Supplementary Material).
Raw data are reported in Figure 2 and further details are given in the Supplementary Table S3. At baseline, no significant difference emerged across rTMS sessions for asynchrony, ITI deviation in the synchronization phase, or ITI deviation in the continuation phase (F3,39 = 0.206, P = 0.892; F3,39 = 0.419, P = 0.741; and F3,39 = 0.224, P = 0.879, respectively).
One-way rmANOVA did not reveal significant pre–post rTMS differences in asynchrony (F3,39 = 0.448, P = 0.720) (Fig. 3B), in ITI deviation for the synchronization phase (F3,39 = 0.697, P = 0.560), or in ITI deviation for the continuation phase (F3,39 = 0.633, P = 0.598) across the different stimulation sites.
Results obtained in the group of 14 healthy volunteers did not change even if the 2 subjects who had previous noninstitutional musical training were excluded from the analysis (for details, see the Supplementary Material).
The main result of the current study is that, in healthy right-handed humans, transient rTMS-induced interference with the function of the right dPMC affects the accuracy of synchronization between right-hand movements and auditory cues. This effect was topographically specific because no worsening in synchronization accuracy was seen with rTMS of the left dPMC, of the SMA, or of an occipital control site. Conversely, after rTMS of these 3 sites (left dPMC, SMA, and Oz) a slight, nonsignificant enhancement of the synchrony was observed (Fig. 3A). This trend toward enhanced synchrony in the second performance was likely due to a training effect, as confirmed by the control experiment where the synchronization–continuation task was repeated 15 min after the baseline without rTMS application. In addition, the effect was related to the metrical structure of the auditory cues as it was not observed with nonmetrical and isochronous sequences, and was greater with complex than with simple metrical rhythms. The finding that the transient disruption of either dPMC did not affect synchronization to isochronous auditory cues is in line with previous fMRI data (Rao et al. 1997; Jäncke et al. 2000; Thaut 2003; Chen et al. 2006). In these correlational studies, dPMC activation was not observed when motor activity was synchronized to isochronous cues, whereas it emerged when the metric saliency increased following the manipulation of the accent structure of the rhythms (Chen et al. 2006). Interestingly, ISI of 300 ms used by Rao et al. (1997) and Chen et al. (2006) was similar to that used for isochronous sequences in the present study (250 ms). Finally, the effect was specific for the auditory–motor synchronization process because rTMS of the right dPMC selectively interfered with the synchrony of the taps to the external cues, whereas the ITI was not affected. The ITI deviation measures the ability to reproduce the rhythmic sequence by finger tapping both in presence of an external cue (as in the synchronization phase of our experimental paradigm) or in absence of the auditory cue (as in the continuation phase), whereas the asynchrony specifically assesses the ability to tap in time with the external cue. This dissociation confirms that asynchrony and ITI represent 2 complementary measures of rhythmic-motor entrainment and is in keeping with previous rTMS data focusing on the reproduction of isochronous sequences (Malcolm et al. 2008, see the next paragraph). Therefore, the present causal findings strongly support the view that the right dPMC plays a pivotal role in auditory–motor synchronization.
Previous rTMS studies focused on synchronization to isochronous auditory cues and showed conflicting results. Increased ITI variability (Del Olmo et al. 2007) and asynchrony (Pollok et al. 2008) have been reported with off-line conditioning of the left dPMC at ISI of 500 and 800 ms, respectively. In addition, an effect on asynchony, but not on ITI, has been reported by off-line rTMS-induced disruption of the left superior temporal–parietal cortex during synchronization to isochronous rhythms at an ISI of 500 ms (Malcolm et al. 2008). However, Doumas et al. (2005) did not find any effect of rTMS of the left dPMC using isochronous cues at an ISI of 500 ms. As to the ventral potion of PMC (vPMC), Malcolm et al. (2008) did not observe significant effects on synchronization error after rTMS of the left hemisphere. However, more recent rTMS (Kornysheva and Schubotz 2011; Ruspantini et al. 2011) and theta-burst stimulation (Bijsterbosch et al. 2011) data suggest that the left vPMC may be involved in sensorimotor synchronization to isochronous sequences. Our data support the view that the role of the dPMC in synchronization to isochronous rhythms, if any, is less relevant, regardless of the rate, than dPMC involvement in synchronization to a metrically manipulated rhythm.
The finding that rTMS of the dPMC did not interfere with auditory–motor synchronization with nonmetrical cues is also in keeping with most of the previous neuroimaging studies showing different patterns of brain activation when structured (metrical) and unstructured (nonmetrical) rhythmic sequences were reproduced (Sakai et al. 1999) or listened to (Bengtsson et al. 2008). Specifically, activity in the premotor area and cerebellar anterior lobe was observed during reproduction of structured rhythms, whereas activity in the right prefrontal cortex and cerebellar posterior lobe emerged for unstructured rhythms (Sakai et al. 1999). Sakai et al. (1999, 2004) proposed that the prevalent prefrontal cortex activation for nonmetrical sequences was related to a higher cognitively demanding performance with respect to the more automatic timing processes underlying reproduction of metrical rhythms. However, in another study (Chen et al. 2008a), a correlation between dPMC activation and temporal complexity of the auditory cue was seen in nonmusicians, with maximal dPMC activity during synchronization to nonmetrical rhythms. One limitation of the present study is that besides the preferential transient disruption of the stimulated cortical area, suprathreshold focal rTMS could also partly interfere with other regions, such as the prefrontal cortex, potentially involved in the neural network underlying rhythmic motor entrainment. Nevertheless, the current data support the view that right dPMC involvement in auditory–motor synchronization is related to the metrical structure of the cues, unless one admits that both floor and ceiling effects occur for nonmetrical and isochronous sequences, respectively. Finally, the prefrontal cortex, along with the pre-SMA, is thought to be scarcely engaged in externally cued motor tasks but is involved in internally driven motor performances (Lau et al. 2004a, 2004b). Hence, it will be worth to tailor future studies to investigate the role of prefrontal cortex and pre-SMA on ITI, especially in the continuation phase which, at difference from asynchrony, is not affected by disruption of the dPMC and is less related to the external cue.
PMC is thought to be engaged in planning, selection, and control based on external events of motor programs (Picard and Strick 2001; Hoshi and Tanji 2007). In this framework, dPMC seems to have an important role in indirect sensorimotor processing (Hoshi and Tanji 2007). The present findings provide further evidence of this and can be accounted for in the context of a recent model of the brain mechanisms underlying auditory–motor interactions, suggesting that the dPMC is involved in higher order aspects of movement organization (Zatorre et al. 2007; Chen et al. 2009). As the functions of the ventral and lateral portions of the PMC are thought to be partly distinct (Picard and Strick 2001), further rTMS studies are needed to investigate the role of the vPMC in auditory motor synchronization with metrically manipulated stimuli by an interference approach.
The role of SMA in motor and perceptual timing processes is acknowledged (Macar et al. 2004, 2006). Neuroimaging studies showed SMA activation when rhythm reproduction was performed in the absence of auditory cues during a synchronization-continuation task (Rao et al. 1997; Lewis et al. 2004). In addition, neuropsychological investigations indicated that patients with SMA lesions are impaired in reproducing rhythms from memory (Halsband et al. 1993). Hence, the fact that SMA disruption did not affect the synchronization phase of auditory–motor entrainment is not surprising. In contrast, the fact that rTMS-induced SMA disruption did not interfere with the performance in the continuation phase of our experimental paradigm may be an unexpected finding, although a similar negative result was reported by Del Olmo et al. (2007), using only isochronous auditory cues. Further studies are needed to characterize more specifically the role of SMA in rhythm reproduction from memory.
The current literature on the lateralization of the neural mechanisms underlying rhythm processing shows conflicting results (for review, see Limb 2006). The present study strongly suggests that the involvement of the dPMC in auditory–motor synchronization is highly lateralized to the right hemisphere. This result is consistent with a previous study showing a higher activation of the right dPMC in nonmusicians (Chen et al. 2008a). At present, we can only speculate on this lateralization. However, as neuroimaging data support the view that activation of the left and right hemisphere reflects differential engagement of implicit versus explicit perceptual timing processes (for review, see Coull and Nobre 2008), it can be hypothesized that the synchronization-continuation task with metrical manipulated sequences used in our study mainly relies on explicit rhythm processing in nonmusicians volunteers. An analogous interpretation has been suggested by Grahn and McAuley (2009) in a recent neuroimaging study on individual differences in beat perception.
In conclusion, the current data originally provide novel information about neural processes involved in motor entrainment and may contribute to shed light on the neural basis of music performance and perception but also on brain mechanisms involved in motor timing functions. In addition, characterizing the neural network underlying rhythmic motor entrainment appears to be a promising tool for planning rehabilitative strategies based on auditory cues such as those used in Parkinson's disease (McIntosh et al. 1997; Marchese et al. 2000) and stroke (Thaut et al. 1997; Altenmüller et al. 2009). Potentially, current results may even be useful to establish the cortical areas to be targeted in procedures aiming to use noninvasive neuromodulation techniques such as rTMS and transcranial direct current stimulation to enhance motor learning compared with conventional rehabilitation procedures. Further work is necessary to assess these intriguing hypotheses.
A grant from “Ente Cassa di Risparmio di Firenze,” Florence, Italy supported this project.
Conflict of Interest: None declared.