Journal of Physiology - Paris 99 (2006) 813 www.elsevier.

com/locate/jphysparis

How the brain controls repetitive nger movements

Bettina Pollok, Joachim Gross, Alfons Schnitzler
*
Department of Neurology, MEG-Laboratory, Heinrich-Heine, University, Moorenstr. 5, 40225 Duesseldorf, Germany

Abstract Adequate interaction with our physical and social environment requires accurate timing abilities. Since planning and control of movements is closely related to sensorimotor synchronization, the investigation of synchronization abilities may allow insights into fundamental principles of motor behaviour. The nger-tapping task has frequently been used to study the synchronization of ones own movements in relation to external events. Data from behavioural studies gave rise to the assumption that it is not the peripheral event (i.e., nger-tap or pacing signal) that is synchronized but its central representation. The neural foundations of sensorimotor synchronization have only recently been investigated and are still poorly understood. The present article reviews data from neurophysiological studies investigating sensorimotor synchronization to shed light on the neurophysiological processes associated with sensorimotor synchronization. This review focuses on studies investigating neuroelectric and neuromagnetic activity associated with simple repetitive synchronization tasks. 2005 Elsevier Ltd. All rights reserved.
Keywords: Coherence; MEG; Review; Sensorimotor; Synchronization; Tapping

1. Synchronization with external eventsbehavioural data Timing of ones own movements with respect to external events has frequently been used to investigate sensorimotor timing mechanisms. In several studies subjects synchronized nger-taps with a regular auditory pacing signal [18]. Usually, subjects report the feeling of exact synchrony between nger-taps and pacing signal. However, the tap leads over the pacer by about 2060 ms. This so-called negative asynchrony is a stable behavioural phenomenon, which has been demonstrated in a large number of studies (for an overview see [9]). This observation gave rise to the hypothesis that it is the central representation of the pacing signal and the nger-tap, that is synchronized rather than the peripheCorresponding author. Tel.: +49 211 8117893; fax: +49 211 8119032. E-mail addresses: bettina.pollok@uni-duesseldorf.de (B. Pollok), jgross@uni-duesseldorf.de (J. Gross), schnitza@uni-duesseldorf.de (A. Schnitzler). 0928-4257/$ - see front matter 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.jphysparis.2005.06.002
*

ral events. Because of presumably dierent processing times of tactilekinaesthetic and auditory feedback, the tap has to lead over the pacing signal to achieve synchrony on a cortical level. An alternative approach to account for this eect is based on the assumption that dierences of the processing time needed to establish a central representation, may contribute to the observed synchronization error [3]. Studies varying the amount of somatosensory feedback substantiate the signicance of somatosensory information for tapping accuracy. It has been shown that increased tactile and kinaesthetic feedback due to tapping with higher forces results in a decrease of the negative asynchrony [3]. Accordingly, reduced somatosensory feedback due to anaesthesia of the nger tip [10] and somatosensory deaerentation [11] is associated with an increase of the negative asynchrony. In the latter case, central or peripheral lesions forestall the processing of somatosensory feedback information. Taken together, these data suggest a specic signicance of somatosensory feedback information for accurate

B. Pollok et al. / Journal of Physiology - Paris 99 (2006) 813

synchronization. However, it still remains an open question how the brain uses somatosensory information to synchronize ones own movements with respect to external events.

2. Neurophysiology The investigation of central processes associated with sensorimotor timing mechanisms requires methods with a temporal resolution in the range of milliseconds. Electroencephalography (EEG) as well magnetoencephalography (MEG) comply with this condition. However, dierent analysis strategies reveal insights into dierent aspects of brain function. 2.1. Evoked responses One possibility for the investigation of EEG- and MEG-data is the analysis of evoked responses. This procedure investigates neurophysiological responses timelocked to a certain event like the onset of a movement or the onset of a sensory stimulus. The analysis of evoked responses mainly reveals information about cortical processes. Neuroelectric [1214] as well as neuromagnetic [1518] studies demonstrated that the execution of simple repetitive nger movements of one hand is associated with activity within the Rolandic ssure of the contralateral hemisphere corresponding to the primary sensorimotor cortex. This activity was time-locked to movement onset whereas evoked responses time-locked to the onset of the pacing signal were localized within the superior temporal sulcus of both hemispheres corresponding to the auditory cortex [16]. Since other studies focused on tap-related activity, no activation of the auditory cortex was reported [12,15,17,18]. However, it is likely that comparable auditory evoked responses were evident in these studies as well. It has been shown that activity within the primary sensorimotor cortex can be decomposed into dierent components. First, at about 100 ms before movement onset a movement eld localized with the primary motor cortex (M1) occurred. This activity most likely represents the neurophysiological correlate of the motor command [1418]. Further activity was detected around tap-onset [1618]. The source was localized within the primary somatosensory cortex (S1) and most likely represents the neuromagnetic correlate of kinaesthetic feedback due to nger movements. Interestingly, this source was not detected by other studies [1215]. However, this might be simply due to slightly dierent analysis procedures. In the studies by Gerlo and co-workers [14,15] and Kopp et al. [13] neurophysiological activity was measured with respect to movement onset. On the contrary, in the studies by Pollok et al. [17,18] and Mu ller

et al. [16], neuromagnetic activity was analysed with respect to tap-onset. Therefore, in these studies the nger has to be moved downwards before tap-onset. Since the movement is associated with proprioceptive feedback, activity around tap-onset most likely represents somatosensory feedback. Finally, activity was detected around 100 ms after movement initiation, respectively, 100 ms after tap-onset [1218]. MEG studies revealed that the corresponding dipole was localized within the primary somatosensory cortex [1418] but inferior to the rst S1 source [16 18]. Therefore, Mu ller et al. [16] labelled this source S1 inferior. This activity most likely represents the neurophysiological correlate of somatosensory feedback due to nger-taps and nger-movements. This interpretation is supported by the observation that this activity occurs even when the nger is moved passively [1922]. A study by Kristeva-Feige and co-workers [23] investigating selfpaced movements at much slower frequencies (i.e., every 815 s) revealed that cutaneous input also contributes to this activity. However, the study of Mu ller et al. [16] revealed evidence for the assumption that the functional meaning of S1 inferior might go beyond the sole processing of somatosensory feedback. Specically, the authors hypothesized that this source might also represent the neuromagnetic correlate of an evaluation process keeping the subject in time with the pacing signal. This hypothesis is based on two observations. First, only S1 inferior was equally well time-locked to the onset of nger-taps and the auditory pacing signal, whereas both other sources (i.e., S1 and M1) were more related to the tap-onset. Second, the authors demonstrated that it is the last event that triggers S1 inferior. Usually, the tap leads over the pacing signal (i.e., negative asynchrony), but in some cases the tap follows the pacer (positive asynchrony). In these cases the tap is the last event, whereas mostly the pacer occurs last. Mu ller and co-workers [16] compared amplitude and latency of the three detected sources between positive and negative asynchronies. They showed that in the case of taprelated activity S1 inferior arose at about 80 ms earlier for positive asynchronies as compared to negative asynchronies, whereas latencies of both other sources did not vary depending on the direction of the asynchrony. This result implies that only S1 inferior becomes active after both relevant events (tap and pacing signal) have occurred. However, the latter observation was based on data from one subject only, since the number of positive asynchronies is usually small. Taken together, both results revealed evidence for the assumption that S1 inferior might be crucial for the evaluation of the temporal distance between tap and pacer. Since this result might lead to a fundamental re-denition of the functional meaning of S1 activity, further investigation of this phenomenon was necessary.

10

B. Pollok et al. / Journal of Physiology - Paris 99 (2006) 813

In a following study the same experiment was performed in order to replicate the ndings of Mu ller et al. [17]. Interestingly, in this study all three sources were more related to the tap than to the pacer weakening the hypothesis that the functional meaning of S1 inferior goes beyond the sole processing of somatosensory information. Furthermore, the comparison between positive and negative asynchronies was performed to validate the results from Mu ller et al. This comparison did not show any dierences of the three tap-related sources depending on positive and negative asynchronies. However, because of the small numbers of positive asynchronies this result was based on data from two subjects only. To elucidate the debated functional role of S1 inferior, the aim of the following study was to enhance the number of positive asynchronies in order to validate the demonstrated dierences of the time course of S1 inferior depending of the direction of the asynchrony. For this reason, subjects performed the same nger-tapping task as in the studies of Mu ller et al. [16] and Pollok and co-workers [17]. However, at the time of tap-onset an additional auditory feedback signal was delivered [18]. Previous studies have shown that delivery of auditory feedback at the time of tap-onset leads to a reduced negative asynchrony [2,7]. Therefore, a larger number of positive asynchronies was expected. Indeed, data analysis demonstrated that in seven subjects a sucient number of positive asynchrony was evident. Comparison between positive and negative asynchronies revealed no latency dierences of S1 inferior. Therefore, this study reveals further evidence against the hypothesis that S1 inferior might represent a kind of evaluation process. Rather, comparison of amplitudes showed a signicantly reduced S1 inferior amplitude associated with positive asynchronies. Although movement kinematics were not controlled in this study, analysis of tap duration showed signicantly shorter duration times associated with positive asynchronies. This implies that the demonstrated amplitude reduction might be due to a reduced amount of tactile feedback, because shorter tap durations might be associated with lesser tactile feedback, resulting in reduced S1 inferior amplitude. This result demonstrates that S1 inferior is modulated by tactilekinaesthetic feedback. Taken together the studies evidenced that S1 inferior most likely exclusively represents somatosensory feedback processing rather than an evaluation process. For methodological reasons insights from analysis of evoked elds are largely restricted to the cortex. However, functional imaging studies demonstrate that the execution of even simple nger movements is associated with neural activity in spatially distributed brain areas [2426]. On the other hand, these studies did not reveal information about the temporally precise interaction between participating brain areas, which reects an impor-

tant coding scheme within the brain [27]. Using frequency analysis of EEG and MEG recordings recent studies have revealed a more detailed picture of the dynamic interplay between cortical and subcortical brain areas during repetitive movement tasks. 2.2. Frequency analysis Analysis in the frequency domain allows to investigate both local activity and neural coupling between different brain sites. Local activity is represented by the local power, whereas synchronized oscillatory activity reects temporal interaction between brain areas. Synchronized oscillatory activity reects functional connectivity and is assumed to act as a binding mechanism that integrates spatially distributed neural activity [27,28]. EEG allows the investigation of coherence between dierent electrode sites. So far EEG studies investigating simple unimanual repetitive movements studied coherence within and between both cerebral hemispheres [2932]. Gerlo and co-workers [29] compared the coupling pattern associated with internally and externally paced movements. In this study, coupling was evident between electrodes covering the primary sensorimotor cortex (S1/M1) of both hemispheres and between S1/ M1 and mesial premotor areas. In both conditions, coupling occurred at alpha (911 Hz) and beta frequency (2022 Hz). During the internally paced movement, coupling was signicantly stronger between bilateral S1/M1 and between S1/M1 contralateral to the moving hand and mesial premotor cortex. According to the interpretation of the authors, enhanced coherence during selfpaced movements might reect higher demands of the motor system during this condition. Interestingly, coherence dierences between both conditions occurred at beta frequency, whereas maximal power dierences were primarily evident at alpha frequency. This implies that local activity and interarea coupling, as represented by coherence between dierent electrode sites, reects dierent aspects of information processing within the human motor system. It remains an open question whether these dierences between alpha and beta frequency reect independent or related processes. Manganotti and co-workers investigated changes of coherence between electrode sites associated with dierent degrees of task complexity [30]. Signicant coherence was observed between bilateral S1/M1, parietal, lateral and mesial premotor and prefrontal areas. In accordance with Gerlo et al. [29], coherence increased when the tasks became more complex. Interestingly, in this study coherence changes occurred at alpha as well as at beta frequency but was stronger in the alpha band. Power changes depending on task complexity were evident at alpha frequency only. Although this result does not agree with the results of Gerlo and co-workers [29], we should stress that the

B. Pollok et al. / Journal of Physiology - Paris 99 (2006) 813

11

tasks subjects had to perform dier between both studies. Whereas in the work of Gerlo et al. [29] subjects performed alternating exions and extensions of one nger only, in the study of Manganotti et al. [30] subjects performed sequential movements of at least three ngers of one hand. However, results from both studies point to three important aspects: First, the execution of simple unimanual nger movements is associated with modulation of oscillatory activity within bilateral S1/M1. Second, coherence between participating brain areas increases with increased task complexity. Third, the functional meaning of dierent frequencies, specically whether they represent related or independent aspects of motor control, remains to be determined. Toma and co-workers [31] investigated the eect of movement rate on local power and coherence during a repetitive thumb-movement of the right hand. They found that coherence between electrodes covering bilateral S1/M1 and a frontal mesial brain area, most likely corresponding to supplementary motor area (SMA) was followed by transient decoupling during slow movements (i.e., up to 1 Hz) but was sustained for fast repetitive movements (i.e., between 2 and 4 Hz). Behaviourally, subjects switched from synchronization to syncopation at movement frequencies between 1 and 2 Hz. These data suggest that at slow movement rates each single movement might be separately controlled, whereas at higher rates the rhythm or the whole stream might be controlled instead of a single event. Interestingly, coupling strength increased with faster movements. However, task diculty did not increase with higher movement velocities. Therefore, coherence strength is modulated by dierent aspects of the experimental procedure like task complexity and movement velocity. This is an important information, which should be taken into account when comparing results from different studies. All in all, the demonstrated studies evidence that the analysis of power and coherence reveals important insights into the central organisation of motor behaviour. However, localization of EEG activity is poor [33] and, more specically, the contribution of subcortical and cerebellar activity cannot be studied by EEG-data. However, the specic signicance of the cerebellum as well as the thalamus for timing mechanisms is well known. MEG together with the analysis tool Dynamic Imaging of Coherent Sources (DICS; [34]) allows a tomographic mapping of power and coherence, which is mainly achieved by using spatial lter algorithms. Using this approach, a recent study investigated the oscillatory network associated with simple auditorily paced nger-taps [35]. Results suggest that task execution is associated with a cerebello-thalamic-cortical network comprising ipsilateral cerebellum, thalamus, S1/M1, posterior parietal (PPC) and lateral (PMC) as well as

mesial premotor cortex (SMA) contralateral to the moving hand. Furthermore, oscillatory activity was detected within the superior temporal sulcus corresponding to the auditory cortex. Fig. 1 illustrates the oscillatory network resulting from cerebro-muscular and cerebro-cerebral coherence analysis. Analysis of the directionality index (DI) [36] revealed that coupling predominantly led from cerebellum to thalamus, from thalamus to PPC, from PPC to premotor areas (i.e., lateral as well as mesial) and from S1/M1 to cerebellum. Furthermore, bi-directional coupling occurred between thalamus and S1/M1, between thalamus and premotor areas, between premotor areas and contralateral S1/M1, and nally between bilateral S1/M1. This result oers further support for the hypothesis that the execution of even simple unimanual nger movements is associated with bilateral S1/M1 activity. Interestingly, coupling occurred between auditory cortex and posterior parietal cortex. Coherence mainly led from the auditory cortex to PPC. This result suggests that auditory pacing information might enter the central motor system via PPC. This hypothesis is supported by the observation that in the posterior parietal cortex of the rat polysensory neurons are localized, responding to tactile and auditory input [37]. Since the above mentioned behavioural data gave rise to the hypothesis that somatosensory feedback information due to the ngertap might be synchronized with the pacing signal, coupling between PPC and auditory cortex oers a highly speculative but interesting possibility to explain how movements might be synchronized with external events. With respect to left hand nger-taps, a comparable network was detected. However, one important

Fig. 1. Group result of cerebro-cerebral coherence analysis during right-hand tapping. Lines indicate consistent coherences between areas across all subjects. Arrows demonstrate predominant coupling direction, whereas conjunctions without arrowheads illustrate bi-directional coupling.

12

B. Pollok et al. / Journal of Physiology - Paris 99 (2006) 813 [7] J. Mates, G. Aschersleben, Sensorimotor synchronization: the impact of temporally displaced auditory feedback, Acta Psychol. (Amst.) 104 (2000) 2944. [8] J. Mates, T. Radil, E. Po ppel, Cooperative tapping: time control under dierent feedback conditions, Percept. Psychophys. 52 (1992) 691704. [9] G. Aschersleben, P. Stenneken, J. Cole, W. Prinz, Timing mechanisms in sensorimotor synchronization, in: W. Prinz, B. Hommel (Eds.), Common Mechanisms in Perception and Action: Attention and Performance, vol. XIX, University Press, Oxford, 2002, pp. 227244. [10] G. Aschersleben, J. Gehrke, W. Prinz, Tapping with peripheral nerve block, Exp. Brain Res. 136 (2001) 331339. [11] P. Stenneken, G. Aschersleben, J. Cole, W. Prinz, Self-induced versus reactive triggering of synchronous movements in a deafferented patient and control subjects, Psychol. Res. 66 (2002) 40 49. [12] C. Gerlo, C. Toro, N. Uenishi, L. Cohen, L. Leocani, M. Hallett, Steady-state movement-related cortical potentials: a new approach to assessing cortical activity associated with fast repetitive nger movements, Electroencephalogr. Clin. Neurophysiol. 102 (1997) 106113. [13] B. Kopp, A. Kunkel, G. Mu ller, W. Mu hlnickel, H. Flor, Steadystate movement-related potentials evoked by fast repetitive movements, Brain Topogr. 13 (2000) 2128. [14] C. Gerlo, U. Uenishi, M. Hallett, Cortical activation during fast repetitive nger movements in humans: dipole sources of steadystate movement-related cortical potentials, J. Clin. Neurophysiol. 15 (1998) 502513. [15] C. Gerlo, N. Uenishi, T. Nagamine, T. Kunieda, M. Hallett, H. Shibasaki, Cortial activation during fast repetitive nger movements in humans: steady-state movement-related magnetic elds and their cortical generators, Electroencephalogr. Clin. Neurophysiol. 109 (1998) 444453. [16] K. Mu ller, F. Schmitz, A. Schnitzler, H.J. Freund, G. Aschersleben, W. Prinz, Neuromagnetic correlates of sensorimotor synchronization, J. Cogn. Neurosci. 12 (2000) 546555. [17] B. Pollok, K. Mu ller, G. Aschersleben, F. Schmitz, A. Schnitzler, W. Prinz, Cortical activations associated with auditorily paced nger tapping, Neuroreport 14 (2003) 247250. [18] B. Pollok, K. Muller, G. Aschersleben, A. Schnitzler, W. Prinz, The role of the primary somatosensory cortex in an auditorily paced nger tapping task, Exp. Brain Res. 156 (2004) 111117. [19] R. Lange, H. Nowak, C. Weiller, Passive nger movement evoked elds in magnetoencephalography, Exp. Brain Res. 136 (2001) 194199. [20] B. Lee, H. Lu ders, R. Lesser, D. Dinner, H. Moris, Cortical potentials related to voluntary and passive movements recorded from subdural electrodes in humans, Ann. Neurol. 20 (1986) 32 37. [21] H. Shibasaki, G. Barrett, E. Halliday, A. Halliday, Cortical potentials following voluntary and passive nger movements, Electroencephalogr. Clin. Neurophysiol. 50 (1980) 201213. [22] J. Xiang, M. Hoshiyama, S. Koyama, Y.H. Suzuki, S. Watanabe, D. Naka, R. Kakigi, Somatosensory evoked magnetic elds following passive nger movement, Cogn. Brain Res. 6 (1997) 73 82. [23] R. Kristeva-Feige, S. Rossi, V. Pizzella, A. Sabato, F. Tecchio, B. Feige, G. Romani, J. Edrich, P. Rossini, Changes in movementrelated brain activity during transient deaerentation: a neuromagnetic study, Brain Res. 714 (1996) 201208. [24] K. Lutz, K. Specht, N.J. Shah, L. Ja ncke, Tapping movements according to regular and irregular visual timing signals investigated with fMRI, Neuroreport 11 (2000) 13011306. [25] C.H. Moritz, V.M. Haughton, D. Cordes, M. Quigley, M.E. Meyerand, Whole-brain functional MR imaging activation from a

dierence between left- and right-hand tapping was evident. Coupling between bilateral S1/M1 occurred in the right hand condition only. This result might shed some light on the functional meaning of coupling between both primary sensorimotor cortices. Coupling between left- and right-S1/M1 is most likely due to callosal bres. Data from transcranial magnetic stimulation (TMS) studies suggest that these connections have inhibitory effects. The study of Netz and co-workers [38] showed that in right handed subjects the dominant left hemisphere has stronger inhibitory impact on the right hemisphere than vice versa. In line with these results, coupling between bilateral S1/M1 demonstrated by Pollok et al. [35] can be best interpreted as inhibition of the right hemisphere via transcallosal connections. The lack of signicant coupling between both S1/M1 during the left hand condition agrees well with the asymmetry of inhibition observed by Netz et al. [38] and suggests a stronger inhibitory inuence of the dominant on the non-dominant motor cortex than vice versa.

3. Conclusion The studies reviewed in the present article indicate that dierent analysis strategies reveal information about dierent aspects of brain functions. While evoked responses allow detailed insights into the time course of brain activation, frequency analysis allows the detection of brain networks and therefore sheds light on how brain areas interact with each other.

Acknowledgements Bettina Pollok was funded by the Deutsche Forschungsgemeinschaft (PO 806/2-1). Alfons Schnitzler was supported by the VolkswagenStiftung (I/73240).

References
[1] G. Aschersleben, W. Prinz, Synchronization actions with events: the role of sensory information, Percept. Psychophys. 57 (1995) 305317. [2] G. Aschersleben, W. Prinz, Delayed auditory feedback in synchronization, J. Mot. Behav. 29 (1997) 3546. [3] G. Aschersleben, Temporal control of movements in sensorimotor synchronization, Brain Cogn. 48 (2002) 6679. [4] M. Billon, C. Bard, M. Fleury, J. Blouin, N. Teasdale, Simultaneity of two eectors in synchronization with a periodic external signal, Hum. Mov. Sci. 15 (1996) 2538. [5] M. Billon, A. Semjen, J. Cole, G. Gauthier, The role of sensory information in the production of periodic nger-tapping sequences, Exp. Brain Res. 110 (1996) 117130. [6] P.A. Kolers, J.M. Brewster, Rhythms and responses, J. Exp. Psychol. Hum Percept. Perform. 11 (1985) 150167.

B. Pollok et al. / Journal of Physiology - Paris 99 (2006) 813 nger-tapping task examined with independent component analysis, Am. J. Neuroradiol. 21 (2000) 16291635. N. Sadato, V. Ibanez, M.P. Deiber, G. Campbell, M. Leonardo, M. Hallett, Frequency-dependent changes of regional cerebral blood ow during nger movements, J. Cereb. Blood Flow Metab. 16 (1996) 2333. F. Varela, J.P. Lachaux, E. Rodriguez, J. Martinerie, The brainweb: phase synchronization and large-scale integration, Nat. Rev. Neurosci. 2 (2001) 229239. W. Singer, Neuronal synchrony: a versatile code for the denition of relations, Neuron 24 (1999) 4965. C. Gerlo, J. Richard, J. Hadley, A.E. Schulman, M. Honda, M. Hallett, Functional coupling and regional activation of human cortical motor areas during simple, internally paced and externally paced nger movements, Brain 121 (1998) 15131531. P. Manganotti, C. Gerlo, C. Toro, H. Katsuta, N. Sadato, P. Zhuang, L. Leocani, M. Hallett, Task-related coherence and taskrelated spectral power changes during sequential nger movements, Electroencephalogr. Clin. Neurophysiol. 109 (1998) 5062. K. Toma, T. Mima, T. Matsuoka, C. Gerlo, T. Ohnishi, B. Koshy, F. Andres, M. Hallett, Movement rate eect on activation and functional coupling of motor cortical areas, J. Neurophysiol. 88 (2002) 33773385.

13

[26]

[27]

[28] [29]

[30]

[31]

[32] D.J. Serrien, M.J. Cassidy, P. Brown, The importance of the dominant hemisphere in the organization of bimanual movements, Hum. Brain Map. 18 (2003) 296305. [33] R. Hari, Magnetoencephalography as a tool of clinical neurophysiology, in: E. Niedermeyer, F. Lopes da Silva (Eds.), Electroencephalography, Williams & Wilkins, Baltimore, 1987, pp. 10351061. [34] J. Gross, J. Kujala, M. Ha ma la inen, L. Timmermann, A. Schnitzler, R. Salmelin, Dynamic imaging of coherent sources: studying neural interactions in the human brain, Proc. Natl. Acad. Sci. USA 98 (2001) 694699. [35] B. Pollok, J. Gross, K. Mu ller, G. Aschersleben, A. Schnitzler, The cerebral oscillatory network associated with auditorily paced nger movements, Neuroimage 24 (2005) 646655. [36] M. Rosenblum, A. Pikovsky, Detecting direction of coupling in interacting oscillators, Phys. Rev. E Stat. Nonlinear Soft Matter Phys. 64 (2001) 045202. [37] S. Di, B. Brett, D.S. Barth, Polysensory evoked potentials in rat parietotemporal cortex: combined auditory and somatosensory responses, Brain Res. 642 (1994) 267280. [38] J. Netz, U. Ziemann, V. Homberg, Hemispheric asymmetry of transcallosal inhibition in man, Exp. Brain Res. 104 (1995) 527 533.