- Email: [email protected]
Synchronization of single motor units during voluntary contractions in the upper and lower extremities
Clinical Neurophysiology 112 (2001) 1243±1249
www.elsevier.com/locate/clinph
Synchronization of single motor units during voluntary contractions in the upper and lower extremities Myung-Shin Kim a,b, Yoshihisa Masakado a,*, Yutaka Tomita a, Naoichi Chino a, Young Sook Pae b, KyungEun Lee b a
Department of Rehabilitation Medicine, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan b Department of Pharmacology, Ewha Womans University School of Medicine, Seoul, South Korea Accepted 23 March 2001
Abstract Objective: To investigate motor unit synchronization in the time and frequency domains and compare the amount and nature of this synchronization between upper and lower extremity muscles in human subjects. Methods: A total of 120 motor unit pairs from biceps brachii (BB), ®rst dorsal interosseous (1DI), vastus medialis (VM), and tibialis anterior (TA) on the dominant side were analyzed and compared. Pairs of motor unit spike trains were recorded from two concentric needle electrodes inserted within these muscles in healthy volunteers. Subjects were instructed to maintain a weak isometric contraction of these muscles so that an individual motor unit recorded from each concentric needle discharged at a steady rate of approximately 10 impulses/s. Pairs of motor unit spike trains were cross-correlated in the time domain, and coherence analysis in the frequency domain was performed on the same spike train data. Results: Synchronization was seen in all the muscles studied. Strength of motor unit synchronization, expressed as synchronization index (SI), was greater in 1DI muscles compared to other muscles (P , 0:01). Coherence analysis revealed signi®cant association between motor unit ®rings in the 1-5 and 25±30 Hz frequency ranges in all the muscles studied. The incidence of 25±30 Hz coherence peaks were found to be greater for 1DI muscles compared to other muscles. Conclusion: The above results suggest a possible role for corticospinal projections in producing pre-synaptic inputs responsible for synchronization of motor unit ®rings and 25±30 Hz coherence peaks. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Motor unit synchronization; Cross-correlation analysis; Coherence analysis; Upper extremity muscles; Lower extremity muscles; Human subjects
1. Introduction Motor unit synchronization has been widely studied as a means of investigating synaptic connectivity. Motor units ®re asynchronously under normal conditions such as smooth muscle contractions, but they can also ®re in synchrony (Buchtal and Madsen, 1950; Milner-Brown et al., 1975). Motor unit synchronization is de®ned as the tendency for two motor units to ®re with dependent latencies relative to each other, more often than would be expected if the motor units were to ®re randomly, but independently. Motor unit synchronization can be revealed by cross-correlating the motoneuron discharges of the two spike trains when a narrow central peak is seen in the cross-correlation histogram (Sears and Stagg, 1976). The occurrences of the peaks may be separated into two categories; those that occur * Corresponding author. Tel.: 181-3-3353-1211; fax: 181-3-3225-6014. E-mail address: [email protected] (Y. Masakado).
within a few milliseconds of zero latency are referred to as short-term synchronization, and those that occur at greater latencies as long-term synchronization (De Luca et al., 1993; Kirkwood and Sears, 1978; Sears and Stagg, 1976). The origin of synchronization is unclear: primary afferent ®bers, segmental neurons, propriospinal neurons, and supraspinal structures could be postulated as sources of motor unit synchronization. However, several lines of evidence indicate that the most signi®cant contributor to motor unit synchronization is the last order common stem pre-synaptic input from descending neurons to motoneurons (Bremner et al., 1991; Datta et al., 1991; Farmer et al., 1993). The corticospinal system is known to exert a greater excitatory in¯uence over distal than over proximal upper extremity muscles (Palmer and Ashby, 1992). In man, as in monkeys, the density of corticospinal projections is generally greater to motor nuclei innervating distal muscles than
1388-2457/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1388-245 7(01)00549-1
CLINPH 2000161
1244
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
proximal muscles (Brouwer and Ashby, 1992). Transcranial magnetic stimulation studies revealed that ®rst dorsal interosseous (1DI) muscles had larger representation showing higher amplitude motor evoked potentials and lower thresholds. However, synchronization behavior in upper and lower extremity muscles has not yet been extensively studied. Cross-correlation methods in the time domain can yield useful information about the presence and strength of synchronization between motor unit ®rings, which in turn can be used to make inferences about the pre-synaptic organization, strength, and synaptic time course of common inputs to motoneurons. A useful extension of cross-correlation analysis of motor unit ®ring in man would be to perform the frequency domain equivalent, coherence analysis (Brillinger, 1983). Signi®cant coherence between a pair of motor unit spike trains would imply some common periodicity of the pre-synaptic input and that the periodic component of the pre-synaptic input common to the motoneurons can be identi®ed. No consensus currently exists on which synchronization index derived from the cross-correlation histogram is most appropriate for expressing the strength of the correlated activity in the two motor neurons. Although it is recognized that the synchronous counts must be normalized to baseline discharge to allow for comparison with other cross-correlation histograms, a variety of normalization procedures have been proposed (Ellaway, 1978; Sears and Stagg, 1976; Nordstrom et al., 1992). In the present study, we used a new method of analyzing the synchronization behavior in the time domain to assess the presence and strength of synchronization. We also used coherence analysis to monitor the rhythmic behavior of these motor unit ®rings, and compared these responses in the upper and lower extremity muscles in humans. 2. Methods 2.1. Subjects
mity muscles, 30 tibialis anterior (TA) motor unit pairs and 30 vastus medialis (VM) motor unit pairs were examined. Subjects were instructed to maintain a weak isometric contraction of each muscle so that both motor units recording from each needle discharged at a steady rate of approximately 10 impulses/s, and discharges of these motor units were recorded for 3 min. The subjects were aided by auditory and visual feedback of unit activity and a ratemeter display. The needle positions were adjusted so that the activity of an individual low threshold motor unit could be recorded in each channel. 2.3. Data analysis 2.3.1. Cross-correlation histogram construction Cross-correlation histograms were used to analyze the interdependence between discharge intervals of simultaneously observed motor unit ®ring trains. These histograms were constructed from pairs of motor unit spike trains by measuring only the ®rst order forward and backward recurrence intervals (the nearest forward and backward ®ring times) of the alternate motor unit with respect to each ®ring of a reference motor unit. The preparation for statistical analysis of the spike trains involved converting the raw electromyography (EMG) signals into sequences of standard pulses by passing the EMG signal through a simple level-detection circuit Discovery Version 5.0 (DATAWAVE Technologies). Fig. 1 shows a schematic representation of how these measurements were made. The recurrence times were then used to increment the bins representing the corresponding latencies in the cross-correlation histogram. A bin width of 1 ms and a pre- and post-trigger latency of 200 ms were used. The middle panel of Fig. 1 shows an example of a cross-correlation histogram using this technique. Peaks in the cross-correlation histogram indicate interdependence between two discharging motor units, one ®ring forward or backward relative to the other motor unit.
Six healthy volunteers (5 men and one woman) aged between 27 and 42 years participated in the study. All participants were free of any known neuromuscular, musculoskeletal, or cardiopulmonary disorders and all showed a preference using their right hand for writing and activities of daily living, and their right leg for kicking a ball. All the subjects were informed of the purpose and procedures of the experiments, and their consent to be tested was obtained prior to invitation of the study.
2.3.2. Synchronization in time domain Strength of synchronization, expressed as synchronization index (SI) was calculated from the normalization procedure as depicted in the bottom panel of Fig. 1, was chosen because of its greater reliability than the single-bin measure used in earlier studies. The raw cross-correlation histogram on the middle panel of Fig. 1 is normalized, so that the total number of data points equals 1000. The moving average of the normalized histogram was calculated using the following equation:
2.2. Motor unit recordings
Fi 1=27 Ci-3 1 3Ci-2 1 6Ci-1 1 7Ci 1 6Ci11 1 3Ci12 1 Ci13
Two concentric needle electrodes were inserted into the muscles under study. A total of 120 motor unit pairs on the dominant side were examined. For the upper extremity muscles, we examined 30 1DI motor unit pairs and 30 biceps brachii (BB) motor unit pairs. For the lower extre-
where Fi is the number in the i-th bin after the moving average is calculated, and Ci is the number in the i-th bin before the moving average. SI Fi 2 7.64; where 7.64 is the average (n 5:00) and 3 times standard deviation (SD) of the sum of randomized
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
1245
A Mann±Whitney test was used to compare the magnitude of synchronization among these muscles. 2.3.3. Synchronization in frequency domain Coherence is a measurement showing the correlation strength between the two motor unit spike trains in the frequency domain, which varies between 0 and 1 (Rosenberg et al., 1989). Calculation of coherence provides a bounded measure of the association between the two processes. Coherence may be interpreted as a representation of the frequency component of common synaptic inputs. The coherence spectra were calculated of¯ine from the same data ®les used to calculate the cross intensity estimates using MATLAB Version 5.2 (Mathworks, Natick, MA). 3. Results 3.1. Synchronization in time domain
Fig. 1. Cross-correlation histogram construction from pairs of motor unit action potential trains and synchronization index. (A) Pairs of motor unit spike trains. Schematic description of the forward and backward ®rst order recurrence times. Firing times of each individual motor unit are depicted by vertical bars. Train 1 represents the ®rings of the reference (triggering) motor unit. Train 2 represents the ®rings of the alternate unit. Tb, forward ®rst order recurrence time; Tf, backward ®rst order recurrence time. (B) Example of a cross-correlation histogram that displays short term synchronization in ®rst dorsal interosseous in one subject. Lower red horizontal line shows the average value of occurrences in a bin, whereas the upper red horizontal line represents the 3 SD con®dence level. Synchronization index (SI) is the area belonging to above 3 SD con®dence level in positive Y-axis. X-axis represents the latency differences between two different motor unit action potential trains. (C). (a) Synchronization index (SI). (b) Duration of synchronization.
motor unit ®rings. If the number of occurrences in a bin was found to be greater than the mean value at a 3 SD con®dence level, then the peak was designated as a signi®cant peak, and peak width, synchronization Duration, was determined by ®nding all adjacent bins above the mean at a 3 SD con®dence level. This can be easily detected by using the newly constructed histogram on the bottom panel of Fig. 1. The Mean of synchronization is measured as the median location in which consecutive bins exceeded the 3 SD con®dence level on the bottom panel of Fig. 1. The Mode of synchronization is measured as the location of the highest peak on the bottom panel of Fig. 1.
Motor unit ®ring of the upper and lower extremity muscles were studied during weak voluntary isometric contraction using pairs of needle electrodes in normal human subjects. Cross-correlation histogram of the ®ring time of one event unit before and after the time of ®ring of another reference unit, showed a clear central peak, indicative of synchronization. Synchronization was seen in all the muscles studied. Percentage of motor unit synchronization was more prevalent in muscles that were located distally (1DI and TA, 100%) than those located proximally (BB: 93%, VM: 90%) (Fig. 2). The strength of synchronization, expressed as the synchronization index (SI), was 33.8 ^ 24.7 for 1DI, 12.7 ^ 16.2 for TA, 8.4 ^ 14.5 for BB, and 2.7 ^ 2.2 for VM. Motor unit short-term synchronization was more prominent in the distally located muscles compared to the proximally located muscles in both upper and lower extremities (P , 0:01) (Table 1). SI compared by subjects showed that synchronization was more pronounced in 1DI muscles than any other muscles. The rank order of motor unit synchronization in
Fig. 2. Detection percentage of signi®cant central peaks in cross-correlation histogram. 1DI, ®rst dorsal interosseous; BB, biceps brachii; TA, tibialis anterior; VM, vastus medialis.
1246
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
Table 1 Synchronization in time domain a Muscles
Synchronization index
Checked First dorsal interosseous Biceps brachii Tibialis anterior Vastus medialis
33.8 ^ 24.7* 8.4 ^ 14.5 12.7 ^ 16.2* 2.7 ^ 2.5
Variables Duration (ms) b
Mean (ms) c
Mode (ms) d
10.9 ^ 2.8 5.4 ^ 3.9 7.6 ^ 3.7 4.0 ^ 1.9
2 1.6 ^ 2.3 2 0.5 ^ 3.8 1.0 ^ 5.3 0.7 ^ 4.9
2 0.6 ^ 1.8 2 0.5 ^ 4.0 2.3 ^ 5.0 1.6 ^ 1.8
a *P , 0:01. (A Mann±Whitney test was employed for comparisons of motor unit synchronization index between proximal and the distal muscles. In upper extremity; biceps brachii vs. ®rst dorsal interossei. In lower extremity; vastus medialis vs. tibialis anterior.) b Width of each synchronous event that exceeded the mean at a 3SD con®dence level. c The median location of each synchronous event that exceeded the mean at a 3 SD con®dence level. d The location of the highest signi®cant peak that exceeded the mean at a 3 SD con®dence level. Values are mean ^ SD.
different subjects was found to be the same in all muscle pairings tested, with 1DI being the largest and VM being the smallest. Certain subjects consistently showed more motor unit synchronization than others in equivalent recordings (Fig. 3). The Duration of synchronization was longer in the distally located muscles compared to the proximally located muscles in the following order: 1DI, TA, BB, and VM. The mean and mode of synchronization was centered around 0 ms, indicative of short-term synchronization (Table 1). 3.2. Synchronization in frequency domain Coherence analysis revealed signi®cant association between motor unit ®rings in the 1±5 and 25±30 Hz frequency ranges in all muscles studied. The signi®cant coherence peaks in the 1±5 Hz range were detected in 100% of 1DI and VM, and 97% of BB and TA. The percentage of 25-30 Hz coherence peaks were 50% for 1DI, 20% for BB, 13% for TA, and 3% for VM (Fig. 4). The ranges of the lower frequency coherence peaks were 1±7 Hz for 1DI, 1±6 Hz for BB, 1±5 Hz for TA, and 1±4 Hz for VM. The ranges of the higher frequency coherence peaks were 18±23 Hz for 1DI, 18±21 Hz for BB, 27±34
Fig. 3. Synchronization index for each subject. ; ®rst dorsal interosseous; B, biceps brachii; ; tibialis anterior; ; vastus medialis.
Hz for TA, and 37±39 Hz for VM. The most frequently occurring coherence value in the lower frequency band was 2 Hz for 1DI, 1 Hz for BB,1 Hz for TA, and 1 Hz for VM. The most frequently occurring coherence value in the higher frequency band was 20 Hz for 1DI, 20 Hz for BB, 30 Hz for TA, and 38 Hz for VM (Table 2). 4. Discussion Synchronization analysis using EMG recording and cross-correlation histogram has an advantage over studying pre-synaptic organization without disturbing the body system in human subjects. During isometric contraction, motor units of the same muscle tend to ®re in synchrony more often than expected by chance (Sears and Stagg, 1976). Early applications of cross-correlation analysis involved the recording of extracellular spike activity from invertebrate neurons in which synaptic interaction is mediated by large post-synaptic potentials (Moore et al., 1970). The times of spike occurrence from two neurons were treated as stochastic point processes and used to construct a histogram in which the times of the reference spikes, de®ned as time zero, were correlated with those of
Fig. 4. Detection percentage of signi®cant coherence peaks in each muscle. 1DI, ®rst dorsal interosseous; BB, biceps brachii; TA, tibialis anterior; VM, vastus medialis.
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249 Table 2 Synchronization in frequency domain a Muscles
Coherence peaks
Checked
Lower frequency (Hz) Higher frequency (Hz)
First dorsal interosseous Biceps brachii Tibialis anterior Vastus medialis
1±7 (2) 1±6 (1) 1±5 (1) 1±4 (1)
18±23 (20) 18±21 (20) 27±34 (30) 37±39 (38)
a All the values are mean frequency ranges of coherence peaks and the most frequently occurring frequency values of coherence peaks are in the parentheses.
another spike train, the response. In the absence of synaptic interaction, the neuronal discharges are temporally unrelated and the histogram is ¯at. The appearance of peaks or troughs in the histogram indicates a raising or lowering of response neuron ®ring probability brought about either by direct synaptic communication between the neurons or through the in¯uence of a pre-synaptic input that is common to the neurons. In the situation where one of the neurons acts as an excitatory input to the other, thus modulating its ®ring, a peak of increased ®ring probability appears at a delay corresponding to the time taken for the discharges of the ®rst neuron to in¯uence the second and for this to be recorded. In the case where recordings are made from a pair of neurons that share a common pre-synaptic input, the primary effect is recognized as a peak of increased ®ring probability centered around time zero (Sears and Stagg, 1976). In the present study, we examined a total of 120 motor unit pairs in 4 different upper and lower extremity muscles during gentle isometric contraction in an effort to see the differences in synchronization behavior in time and frequency domain. The discharge characteristics of the motor units differ in every muscle. In 1DI, most motor units are recruited at relatively low force thresholds compared to larger muscles that are composed of slow twitch ®bers such as vastus medialis (Dumitru, 1995). We studied at minimal force of contraction so that each motor unit studied would discharge roughly 10 impulses/s. We compared motor unit synchronization and the periodic component of pre-synaptic input common to the motor unit pairs between upper and lower extremity muscles. Our data demonstrated that motor unit synchronization was a more prominent feature in the 1DI muscles when time domain analysis in a cross-correlation histogram was applied (Fig. 5). 1DI muscles showed the highest synchronization index (SI) and 100% synchronization compared to other muscles. Our data showed a greater amount of synchronization compared to other studies (De Luca et al., 1993; Dietz et al., 1976; Nordstrom et al., 1992). This discrepancy may be due to the analytic procedure previously employed to collect the data, different muscle contraction levels, and different synchronization indexes used. Debate continues about the origin of this motor unit
1247
synchronization and whether it is generated by the peripheral or spinal and supraspinal pathways. Proximal and distal muscles differ in a number of neuromuscular system parameters. Compared to distal muscles, segmental pathways are more in¯uential in producing motor unit ®ring of proximal muscles in which the synchronization index in the cross-correlation histogram and the incidence of 25±30 Hz coherence peaks were found to be smaller. These results suggest that a peripheral afferent origin of synchronization is very unlikely. In contrast to a peripheral afferent origin of this shortterm synchronization, the corticospinal tract is known to monosynaptically activate intrinsic hand muscles powerfully (Clough et al., 1968). Individual corticomotoneuronal cells make more frequent and more potent terminations in the motoneuron pools of the distal muscles than proximal muscles (McKiernan et al., 1998). Supporting evidence for a corticospinal origin of short-term synchronization in man includes loss of short-term synchronization following lesions of the corticospinal pathway, whereas normal motor unit synchronization in a deafferented patient supports a central origin (Datta et al., 1991; Farmer et al., 1993). In our experiment, the duration of synchronization was longer in 1DI muscles. This longer duration of synchronization might suggest increased synchronized activity of interneurons projecting to the distal rather than proximal motoneuron pools. However, this should be reviewed more carefully. From studies of primates and man, one can assume that oscillatory activity exists in the motor system. In humans, individual motor unit discharges within and between muscles and EMGs are commonly modulated by 1±12 and 16±32 Hz drives (Farmer et al., 1993). The magnetoencephalogram recorded from a localized region of the sensorimotor cortex indicated coherence with the EMG from the contralateral 1DI within the 13±35 Hz frequency range, suggesting a possible role for the central pathway in producing motor output (Conway et al., 1995). In our study, signi®cant coherence has been observed between motor unit pairs in the 1±5 and 25±30 Hz ranges, suggesting a common rhythmic input to the motoneuron pools. Coherence peaks in the 25±30 Hz frequency range were found to be more prevalent in the distally located muscles compared to the proximally located muscles. This may indicate a central origin, and supports the hypothesis that stronger corticospinal projections to the distal muscles generate such local rhythmic activity. Toro et al. (1994) suggested a particular role for 20±30 Hz oscillations in the control of distal movements. In the present study, slightly different ranges of higher frequency peaks were observed for each group of muscles. Different frequency ranges for different muscles (18±23 Hz for 1DI, 18±21 Hz for BB, 27±34 Hz for TA, and 37±39 Hz for VM) might suggest different individual oscillating activities in the brain.
1248
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
Fig. 5. Cross-correlation histogram and matched coherence spectra. (A) Cross-correlation histogram constructed between the discharges of two motor units recorded from within ®rst dorsal interosseous (1), biceps brachii (2), tibialis anterior (3), and vastus medialis (4). (B) Coherence spectra for the data used to construct the cross-correlation histogram in (A).
The observations made in the present study indicate that synchronization is more prominent in the distally located muscles, and higher frequency (25±30 Hz) coherence peaks are also more prevalent in the distally located muscles in both upper and lower extremities. These results may be related to a greater number of strong corticospinal projections in distal motoneurons.
References Bremner FD, Baker JR, Cole JD, Stephens JA. Correlation between the discharges of motor units recorded from the same and from different ®nger muscles in man. J Physiol 1991;432:355±380. Brillinger DR. The ®nite Fourier transformation of a stationary process. In: Brillinger DR, editor. Handbook of statistics, Amsterdam: Elsevier, 1983.
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249 Brouwer B, Ashby P. Corticospinal projections to lower limb motoneurons in man. Exp Brain Res 1992;89:649±654. Buchtal F, Madsen A. Synchronous activity in normal and atrophic muscle. Electroenceph clin Neurophysiol 1950;2:425±444. Clough JFM, Kernell D, Phillips CG. The distribution of monosynaptic excitation from the pyramidal tract and from primary spindle afferents to motoneurones of the baboon's hand and forearm. J Physiol 1968;198:145±166. Conway BA, Halliday DM, Farmer SF, Shahani U, Maas P, Weir AI, Rosenberg JR. Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man. J Physiol 1995;489(3):917±924. Datta AK, Farmer SF, Stephens JA. Central nervous pathways underlying synchronization of human motor unit ®ring studied during voluntary contractions. J Physiol 1991;432:401±425. De Luca CJ, Roy AM, Erim Z. Synchronization of motor-unit ®rings in several human muscles. J Neurophysiol. 1993;70(5):2010±2023. Dietz V, Bischofberger E, Wita G, Freund HJ. Correlation between the discharge of two simultaneously recorded motor units and physiologic tremor. Electroenceph clin Neurophysiol 1976;40:97± 105. Dumitru D. Electrodiagnostic medicine. Philadelphia, PA: Henley and Belfus Inc, 1995. Ellaway PH. Cumulative sum technique and its application to the analysis of peristimulus time histograms. Electroenceph clin Neurophysiol 1978;45:302±304. Farmer SF, Bremner FD, Halliday DM, Rosenberg JR, Stephens JA. The frequency content of common synaptic inputs to motoneurons studied
1249
during voluntary isometric contraction in man. J Physiol 1993;470:127± 155. Kirkwood PA, Sears TA. Synaptic connections to intercostal motoneurons as revealed by the common excitation potential. J Physiol 1978;275:102±134. McKiernan BJ, Marcario JK, Karrer JH, Cheney PD. Corticomotoneuronal postspike effects in shoulder, elbow, wrist, digit, and intrinsic hand muscles during a reach and prehension task. J Neurophysiol 1998;80:1961±1980. Milner-Brown HS, Stein RB, Yemm R. The contractile properties of human motor units during voluntary isometric contractions. J Physiol 1975;228:285±306. Moore GP, Segundo JP, Perkel DL, Levitan H. Statistical signs of synaptic interaction in neurons. Biophy J 1970;10:876±900. Nordstrom AM, Fuglevand AJ, Enoka RM. Estimating the strength of common input to human motoneurons from the cross correlogram. J Physiol 1992;453:547±574. Palmer E, Ashby P. Corticospinal projections to upper limb motoneurones in humans. J Physiol 1992;448:397±412. Rosenberg JR, Amjad AM, Breeze P, Brillinger DR, Halliday DM. The Fourier approach to the identi®cation of functional coupling between neuronal spike trains. Prog Biophys Mol Biol 1989;53:1±31. Sears TA, Stagg D. Short-term synchronization of intercostal motoneuron activity. J Physiol 1976;263:357±381. Toro C, Friehs G, Ojakangas C, Maxwell R, Gates JR, Gumnit RJ, Ebner TJ. 8-12 Hz rhythmic oscillations in human motor cortex during twodimensional arm movements: evidence for representation of kinematic parameters. Electroenceph clin Neurophysiol 1994;93:390±403.
www.elsevier.com/locate/clinph
Synchronization of single motor units during voluntary contractions in the upper and lower extremities Myung-Shin Kim a,b, Yoshihisa Masakado a,*, Yutaka Tomita a, Naoichi Chino a, Young Sook Pae b, KyungEun Lee b a
Department of Rehabilitation Medicine, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan b Department of Pharmacology, Ewha Womans University School of Medicine, Seoul, South Korea Accepted 23 March 2001
Abstract Objective: To investigate motor unit synchronization in the time and frequency domains and compare the amount and nature of this synchronization between upper and lower extremity muscles in human subjects. Methods: A total of 120 motor unit pairs from biceps brachii (BB), ®rst dorsal interosseous (1DI), vastus medialis (VM), and tibialis anterior (TA) on the dominant side were analyzed and compared. Pairs of motor unit spike trains were recorded from two concentric needle electrodes inserted within these muscles in healthy volunteers. Subjects were instructed to maintain a weak isometric contraction of these muscles so that an individual motor unit recorded from each concentric needle discharged at a steady rate of approximately 10 impulses/s. Pairs of motor unit spike trains were cross-correlated in the time domain, and coherence analysis in the frequency domain was performed on the same spike train data. Results: Synchronization was seen in all the muscles studied. Strength of motor unit synchronization, expressed as synchronization index (SI), was greater in 1DI muscles compared to other muscles (P , 0:01). Coherence analysis revealed signi®cant association between motor unit ®rings in the 1-5 and 25±30 Hz frequency ranges in all the muscles studied. The incidence of 25±30 Hz coherence peaks were found to be greater for 1DI muscles compared to other muscles. Conclusion: The above results suggest a possible role for corticospinal projections in producing pre-synaptic inputs responsible for synchronization of motor unit ®rings and 25±30 Hz coherence peaks. q 2001 Elsevier Science Ireland Ltd. All rights reserved. Keywords: Motor unit synchronization; Cross-correlation analysis; Coherence analysis; Upper extremity muscles; Lower extremity muscles; Human subjects
1. Introduction Motor unit synchronization has been widely studied as a means of investigating synaptic connectivity. Motor units ®re asynchronously under normal conditions such as smooth muscle contractions, but they can also ®re in synchrony (Buchtal and Madsen, 1950; Milner-Brown et al., 1975). Motor unit synchronization is de®ned as the tendency for two motor units to ®re with dependent latencies relative to each other, more often than would be expected if the motor units were to ®re randomly, but independently. Motor unit synchronization can be revealed by cross-correlating the motoneuron discharges of the two spike trains when a narrow central peak is seen in the cross-correlation histogram (Sears and Stagg, 1976). The occurrences of the peaks may be separated into two categories; those that occur * Corresponding author. Tel.: 181-3-3353-1211; fax: 181-3-3225-6014. E-mail address: [email protected] (Y. Masakado).
within a few milliseconds of zero latency are referred to as short-term synchronization, and those that occur at greater latencies as long-term synchronization (De Luca et al., 1993; Kirkwood and Sears, 1978; Sears and Stagg, 1976). The origin of synchronization is unclear: primary afferent ®bers, segmental neurons, propriospinal neurons, and supraspinal structures could be postulated as sources of motor unit synchronization. However, several lines of evidence indicate that the most signi®cant contributor to motor unit synchronization is the last order common stem pre-synaptic input from descending neurons to motoneurons (Bremner et al., 1991; Datta et al., 1991; Farmer et al., 1993). The corticospinal system is known to exert a greater excitatory in¯uence over distal than over proximal upper extremity muscles (Palmer and Ashby, 1992). In man, as in monkeys, the density of corticospinal projections is generally greater to motor nuclei innervating distal muscles than
1388-2457/01/$ - see front matter q 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 1388-245 7(01)00549-1
CLINPH 2000161
1244
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
proximal muscles (Brouwer and Ashby, 1992). Transcranial magnetic stimulation studies revealed that ®rst dorsal interosseous (1DI) muscles had larger representation showing higher amplitude motor evoked potentials and lower thresholds. However, synchronization behavior in upper and lower extremity muscles has not yet been extensively studied. Cross-correlation methods in the time domain can yield useful information about the presence and strength of synchronization between motor unit ®rings, which in turn can be used to make inferences about the pre-synaptic organization, strength, and synaptic time course of common inputs to motoneurons. A useful extension of cross-correlation analysis of motor unit ®ring in man would be to perform the frequency domain equivalent, coherence analysis (Brillinger, 1983). Signi®cant coherence between a pair of motor unit spike trains would imply some common periodicity of the pre-synaptic input and that the periodic component of the pre-synaptic input common to the motoneurons can be identi®ed. No consensus currently exists on which synchronization index derived from the cross-correlation histogram is most appropriate for expressing the strength of the correlated activity in the two motor neurons. Although it is recognized that the synchronous counts must be normalized to baseline discharge to allow for comparison with other cross-correlation histograms, a variety of normalization procedures have been proposed (Ellaway, 1978; Sears and Stagg, 1976; Nordstrom et al., 1992). In the present study, we used a new method of analyzing the synchronization behavior in the time domain to assess the presence and strength of synchronization. We also used coherence analysis to monitor the rhythmic behavior of these motor unit ®rings, and compared these responses in the upper and lower extremity muscles in humans. 2. Methods 2.1. Subjects
mity muscles, 30 tibialis anterior (TA) motor unit pairs and 30 vastus medialis (VM) motor unit pairs were examined. Subjects were instructed to maintain a weak isometric contraction of each muscle so that both motor units recording from each needle discharged at a steady rate of approximately 10 impulses/s, and discharges of these motor units were recorded for 3 min. The subjects were aided by auditory and visual feedback of unit activity and a ratemeter display. The needle positions were adjusted so that the activity of an individual low threshold motor unit could be recorded in each channel. 2.3. Data analysis 2.3.1. Cross-correlation histogram construction Cross-correlation histograms were used to analyze the interdependence between discharge intervals of simultaneously observed motor unit ®ring trains. These histograms were constructed from pairs of motor unit spike trains by measuring only the ®rst order forward and backward recurrence intervals (the nearest forward and backward ®ring times) of the alternate motor unit with respect to each ®ring of a reference motor unit. The preparation for statistical analysis of the spike trains involved converting the raw electromyography (EMG) signals into sequences of standard pulses by passing the EMG signal through a simple level-detection circuit Discovery Version 5.0 (DATAWAVE Technologies). Fig. 1 shows a schematic representation of how these measurements were made. The recurrence times were then used to increment the bins representing the corresponding latencies in the cross-correlation histogram. A bin width of 1 ms and a pre- and post-trigger latency of 200 ms were used. The middle panel of Fig. 1 shows an example of a cross-correlation histogram using this technique. Peaks in the cross-correlation histogram indicate interdependence between two discharging motor units, one ®ring forward or backward relative to the other motor unit.
Six healthy volunteers (5 men and one woman) aged between 27 and 42 years participated in the study. All participants were free of any known neuromuscular, musculoskeletal, or cardiopulmonary disorders and all showed a preference using their right hand for writing and activities of daily living, and their right leg for kicking a ball. All the subjects were informed of the purpose and procedures of the experiments, and their consent to be tested was obtained prior to invitation of the study.
2.3.2. Synchronization in time domain Strength of synchronization, expressed as synchronization index (SI) was calculated from the normalization procedure as depicted in the bottom panel of Fig. 1, was chosen because of its greater reliability than the single-bin measure used in earlier studies. The raw cross-correlation histogram on the middle panel of Fig. 1 is normalized, so that the total number of data points equals 1000. The moving average of the normalized histogram was calculated using the following equation:
2.2. Motor unit recordings
Fi 1=27 Ci-3 1 3Ci-2 1 6Ci-1 1 7Ci 1 6Ci11 1 3Ci12 1 Ci13
Two concentric needle electrodes were inserted into the muscles under study. A total of 120 motor unit pairs on the dominant side were examined. For the upper extremity muscles, we examined 30 1DI motor unit pairs and 30 biceps brachii (BB) motor unit pairs. For the lower extre-
where Fi is the number in the i-th bin after the moving average is calculated, and Ci is the number in the i-th bin before the moving average. SI Fi 2 7.64; where 7.64 is the average (n 5:00) and 3 times standard deviation (SD) of the sum of randomized
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
1245
A Mann±Whitney test was used to compare the magnitude of synchronization among these muscles. 2.3.3. Synchronization in frequency domain Coherence is a measurement showing the correlation strength between the two motor unit spike trains in the frequency domain, which varies between 0 and 1 (Rosenberg et al., 1989). Calculation of coherence provides a bounded measure of the association between the two processes. Coherence may be interpreted as a representation of the frequency component of common synaptic inputs. The coherence spectra were calculated of¯ine from the same data ®les used to calculate the cross intensity estimates using MATLAB Version 5.2 (Mathworks, Natick, MA). 3. Results 3.1. Synchronization in time domain
Fig. 1. Cross-correlation histogram construction from pairs of motor unit action potential trains and synchronization index. (A) Pairs of motor unit spike trains. Schematic description of the forward and backward ®rst order recurrence times. Firing times of each individual motor unit are depicted by vertical bars. Train 1 represents the ®rings of the reference (triggering) motor unit. Train 2 represents the ®rings of the alternate unit. Tb, forward ®rst order recurrence time; Tf, backward ®rst order recurrence time. (B) Example of a cross-correlation histogram that displays short term synchronization in ®rst dorsal interosseous in one subject. Lower red horizontal line shows the average value of occurrences in a bin, whereas the upper red horizontal line represents the 3 SD con®dence level. Synchronization index (SI) is the area belonging to above 3 SD con®dence level in positive Y-axis. X-axis represents the latency differences between two different motor unit action potential trains. (C). (a) Synchronization index (SI). (b) Duration of synchronization.
motor unit ®rings. If the number of occurrences in a bin was found to be greater than the mean value at a 3 SD con®dence level, then the peak was designated as a signi®cant peak, and peak width, synchronization Duration, was determined by ®nding all adjacent bins above the mean at a 3 SD con®dence level. This can be easily detected by using the newly constructed histogram on the bottom panel of Fig. 1. The Mean of synchronization is measured as the median location in which consecutive bins exceeded the 3 SD con®dence level on the bottom panel of Fig. 1. The Mode of synchronization is measured as the location of the highest peak on the bottom panel of Fig. 1.
Motor unit ®ring of the upper and lower extremity muscles were studied during weak voluntary isometric contraction using pairs of needle electrodes in normal human subjects. Cross-correlation histogram of the ®ring time of one event unit before and after the time of ®ring of another reference unit, showed a clear central peak, indicative of synchronization. Synchronization was seen in all the muscles studied. Percentage of motor unit synchronization was more prevalent in muscles that were located distally (1DI and TA, 100%) than those located proximally (BB: 93%, VM: 90%) (Fig. 2). The strength of synchronization, expressed as the synchronization index (SI), was 33.8 ^ 24.7 for 1DI, 12.7 ^ 16.2 for TA, 8.4 ^ 14.5 for BB, and 2.7 ^ 2.2 for VM. Motor unit short-term synchronization was more prominent in the distally located muscles compared to the proximally located muscles in both upper and lower extremities (P , 0:01) (Table 1). SI compared by subjects showed that synchronization was more pronounced in 1DI muscles than any other muscles. The rank order of motor unit synchronization in
Fig. 2. Detection percentage of signi®cant central peaks in cross-correlation histogram. 1DI, ®rst dorsal interosseous; BB, biceps brachii; TA, tibialis anterior; VM, vastus medialis.
1246
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
Table 1 Synchronization in time domain a Muscles
Synchronization index
Checked First dorsal interosseous Biceps brachii Tibialis anterior Vastus medialis
33.8 ^ 24.7* 8.4 ^ 14.5 12.7 ^ 16.2* 2.7 ^ 2.5
Variables Duration (ms) b
Mean (ms) c
Mode (ms) d
10.9 ^ 2.8 5.4 ^ 3.9 7.6 ^ 3.7 4.0 ^ 1.9
2 1.6 ^ 2.3 2 0.5 ^ 3.8 1.0 ^ 5.3 0.7 ^ 4.9
2 0.6 ^ 1.8 2 0.5 ^ 4.0 2.3 ^ 5.0 1.6 ^ 1.8
a *P , 0:01. (A Mann±Whitney test was employed for comparisons of motor unit synchronization index between proximal and the distal muscles. In upper extremity; biceps brachii vs. ®rst dorsal interossei. In lower extremity; vastus medialis vs. tibialis anterior.) b Width of each synchronous event that exceeded the mean at a 3SD con®dence level. c The median location of each synchronous event that exceeded the mean at a 3 SD con®dence level. d The location of the highest signi®cant peak that exceeded the mean at a 3 SD con®dence level. Values are mean ^ SD.
different subjects was found to be the same in all muscle pairings tested, with 1DI being the largest and VM being the smallest. Certain subjects consistently showed more motor unit synchronization than others in equivalent recordings (Fig. 3). The Duration of synchronization was longer in the distally located muscles compared to the proximally located muscles in the following order: 1DI, TA, BB, and VM. The mean and mode of synchronization was centered around 0 ms, indicative of short-term synchronization (Table 1). 3.2. Synchronization in frequency domain Coherence analysis revealed signi®cant association between motor unit ®rings in the 1±5 and 25±30 Hz frequency ranges in all muscles studied. The signi®cant coherence peaks in the 1±5 Hz range were detected in 100% of 1DI and VM, and 97% of BB and TA. The percentage of 25-30 Hz coherence peaks were 50% for 1DI, 20% for BB, 13% for TA, and 3% for VM (Fig. 4). The ranges of the lower frequency coherence peaks were 1±7 Hz for 1DI, 1±6 Hz for BB, 1±5 Hz for TA, and 1±4 Hz for VM. The ranges of the higher frequency coherence peaks were 18±23 Hz for 1DI, 18±21 Hz for BB, 27±34
Fig. 3. Synchronization index for each subject. ; ®rst dorsal interosseous; B, biceps brachii; ; tibialis anterior; ; vastus medialis.
Hz for TA, and 37±39 Hz for VM. The most frequently occurring coherence value in the lower frequency band was 2 Hz for 1DI, 1 Hz for BB,1 Hz for TA, and 1 Hz for VM. The most frequently occurring coherence value in the higher frequency band was 20 Hz for 1DI, 20 Hz for BB, 30 Hz for TA, and 38 Hz for VM (Table 2). 4. Discussion Synchronization analysis using EMG recording and cross-correlation histogram has an advantage over studying pre-synaptic organization without disturbing the body system in human subjects. During isometric contraction, motor units of the same muscle tend to ®re in synchrony more often than expected by chance (Sears and Stagg, 1976). Early applications of cross-correlation analysis involved the recording of extracellular spike activity from invertebrate neurons in which synaptic interaction is mediated by large post-synaptic potentials (Moore et al., 1970). The times of spike occurrence from two neurons were treated as stochastic point processes and used to construct a histogram in which the times of the reference spikes, de®ned as time zero, were correlated with those of
Fig. 4. Detection percentage of signi®cant coherence peaks in each muscle. 1DI, ®rst dorsal interosseous; BB, biceps brachii; TA, tibialis anterior; VM, vastus medialis.
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249 Table 2 Synchronization in frequency domain a Muscles
Coherence peaks
Checked
Lower frequency (Hz) Higher frequency (Hz)
First dorsal interosseous Biceps brachii Tibialis anterior Vastus medialis
1±7 (2) 1±6 (1) 1±5 (1) 1±4 (1)
18±23 (20) 18±21 (20) 27±34 (30) 37±39 (38)
a All the values are mean frequency ranges of coherence peaks and the most frequently occurring frequency values of coherence peaks are in the parentheses.
another spike train, the response. In the absence of synaptic interaction, the neuronal discharges are temporally unrelated and the histogram is ¯at. The appearance of peaks or troughs in the histogram indicates a raising or lowering of response neuron ®ring probability brought about either by direct synaptic communication between the neurons or through the in¯uence of a pre-synaptic input that is common to the neurons. In the situation where one of the neurons acts as an excitatory input to the other, thus modulating its ®ring, a peak of increased ®ring probability appears at a delay corresponding to the time taken for the discharges of the ®rst neuron to in¯uence the second and for this to be recorded. In the case where recordings are made from a pair of neurons that share a common pre-synaptic input, the primary effect is recognized as a peak of increased ®ring probability centered around time zero (Sears and Stagg, 1976). In the present study, we examined a total of 120 motor unit pairs in 4 different upper and lower extremity muscles during gentle isometric contraction in an effort to see the differences in synchronization behavior in time and frequency domain. The discharge characteristics of the motor units differ in every muscle. In 1DI, most motor units are recruited at relatively low force thresholds compared to larger muscles that are composed of slow twitch ®bers such as vastus medialis (Dumitru, 1995). We studied at minimal force of contraction so that each motor unit studied would discharge roughly 10 impulses/s. We compared motor unit synchronization and the periodic component of pre-synaptic input common to the motor unit pairs between upper and lower extremity muscles. Our data demonstrated that motor unit synchronization was a more prominent feature in the 1DI muscles when time domain analysis in a cross-correlation histogram was applied (Fig. 5). 1DI muscles showed the highest synchronization index (SI) and 100% synchronization compared to other muscles. Our data showed a greater amount of synchronization compared to other studies (De Luca et al., 1993; Dietz et al., 1976; Nordstrom et al., 1992). This discrepancy may be due to the analytic procedure previously employed to collect the data, different muscle contraction levels, and different synchronization indexes used. Debate continues about the origin of this motor unit
1247
synchronization and whether it is generated by the peripheral or spinal and supraspinal pathways. Proximal and distal muscles differ in a number of neuromuscular system parameters. Compared to distal muscles, segmental pathways are more in¯uential in producing motor unit ®ring of proximal muscles in which the synchronization index in the cross-correlation histogram and the incidence of 25±30 Hz coherence peaks were found to be smaller. These results suggest that a peripheral afferent origin of synchronization is very unlikely. In contrast to a peripheral afferent origin of this shortterm synchronization, the corticospinal tract is known to monosynaptically activate intrinsic hand muscles powerfully (Clough et al., 1968). Individual corticomotoneuronal cells make more frequent and more potent terminations in the motoneuron pools of the distal muscles than proximal muscles (McKiernan et al., 1998). Supporting evidence for a corticospinal origin of short-term synchronization in man includes loss of short-term synchronization following lesions of the corticospinal pathway, whereas normal motor unit synchronization in a deafferented patient supports a central origin (Datta et al., 1991; Farmer et al., 1993). In our experiment, the duration of synchronization was longer in 1DI muscles. This longer duration of synchronization might suggest increased synchronized activity of interneurons projecting to the distal rather than proximal motoneuron pools. However, this should be reviewed more carefully. From studies of primates and man, one can assume that oscillatory activity exists in the motor system. In humans, individual motor unit discharges within and between muscles and EMGs are commonly modulated by 1±12 and 16±32 Hz drives (Farmer et al., 1993). The magnetoencephalogram recorded from a localized region of the sensorimotor cortex indicated coherence with the EMG from the contralateral 1DI within the 13±35 Hz frequency range, suggesting a possible role for the central pathway in producing motor output (Conway et al., 1995). In our study, signi®cant coherence has been observed between motor unit pairs in the 1±5 and 25±30 Hz ranges, suggesting a common rhythmic input to the motoneuron pools. Coherence peaks in the 25±30 Hz frequency range were found to be more prevalent in the distally located muscles compared to the proximally located muscles. This may indicate a central origin, and supports the hypothesis that stronger corticospinal projections to the distal muscles generate such local rhythmic activity. Toro et al. (1994) suggested a particular role for 20±30 Hz oscillations in the control of distal movements. In the present study, slightly different ranges of higher frequency peaks were observed for each group of muscles. Different frequency ranges for different muscles (18±23 Hz for 1DI, 18±21 Hz for BB, 27±34 Hz for TA, and 37±39 Hz for VM) might suggest different individual oscillating activities in the brain.
1248
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249
Fig. 5. Cross-correlation histogram and matched coherence spectra. (A) Cross-correlation histogram constructed between the discharges of two motor units recorded from within ®rst dorsal interosseous (1), biceps brachii (2), tibialis anterior (3), and vastus medialis (4). (B) Coherence spectra for the data used to construct the cross-correlation histogram in (A).
The observations made in the present study indicate that synchronization is more prominent in the distally located muscles, and higher frequency (25±30 Hz) coherence peaks are also more prevalent in the distally located muscles in both upper and lower extremities. These results may be related to a greater number of strong corticospinal projections in distal motoneurons.
References Bremner FD, Baker JR, Cole JD, Stephens JA. Correlation between the discharges of motor units recorded from the same and from different ®nger muscles in man. J Physiol 1991;432:355±380. Brillinger DR. The ®nite Fourier transformation of a stationary process. In: Brillinger DR, editor. Handbook of statistics, Amsterdam: Elsevier, 1983.
M.-S. Kim et al. / Clinical Neurophysiology 112 (2001) 1243±1249 Brouwer B, Ashby P. Corticospinal projections to lower limb motoneurons in man. Exp Brain Res 1992;89:649±654. Buchtal F, Madsen A. Synchronous activity in normal and atrophic muscle. Electroenceph clin Neurophysiol 1950;2:425±444. Clough JFM, Kernell D, Phillips CG. The distribution of monosynaptic excitation from the pyramidal tract and from primary spindle afferents to motoneurones of the baboon's hand and forearm. J Physiol 1968;198:145±166. Conway BA, Halliday DM, Farmer SF, Shahani U, Maas P, Weir AI, Rosenberg JR. Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man. J Physiol 1995;489(3):917±924. Datta AK, Farmer SF, Stephens JA. Central nervous pathways underlying synchronization of human motor unit ®ring studied during voluntary contractions. J Physiol 1991;432:401±425. De Luca CJ, Roy AM, Erim Z. Synchronization of motor-unit ®rings in several human muscles. J Neurophysiol. 1993;70(5):2010±2023. Dietz V, Bischofberger E, Wita G, Freund HJ. Correlation between the discharge of two simultaneously recorded motor units and physiologic tremor. Electroenceph clin Neurophysiol 1976;40:97± 105. Dumitru D. Electrodiagnostic medicine. Philadelphia, PA: Henley and Belfus Inc, 1995. Ellaway PH. Cumulative sum technique and its application to the analysis of peristimulus time histograms. Electroenceph clin Neurophysiol 1978;45:302±304. Farmer SF, Bremner FD, Halliday DM, Rosenberg JR, Stephens JA. The frequency content of common synaptic inputs to motoneurons studied
1249
during voluntary isometric contraction in man. J Physiol 1993;470:127± 155. Kirkwood PA, Sears TA. Synaptic connections to intercostal motoneurons as revealed by the common excitation potential. J Physiol 1978;275:102±134. McKiernan BJ, Marcario JK, Karrer JH, Cheney PD. Corticomotoneuronal postspike effects in shoulder, elbow, wrist, digit, and intrinsic hand muscles during a reach and prehension task. J Neurophysiol 1998;80:1961±1980. Milner-Brown HS, Stein RB, Yemm R. The contractile properties of human motor units during voluntary isometric contractions. J Physiol 1975;228:285±306. Moore GP, Segundo JP, Perkel DL, Levitan H. Statistical signs of synaptic interaction in neurons. Biophy J 1970;10:876±900. Nordstrom AM, Fuglevand AJ, Enoka RM. Estimating the strength of common input to human motoneurons from the cross correlogram. J Physiol 1992;453:547±574. Palmer E, Ashby P. Corticospinal projections to upper limb motoneurones in humans. J Physiol 1992;448:397±412. Rosenberg JR, Amjad AM, Breeze P, Brillinger DR, Halliday DM. The Fourier approach to the identi®cation of functional coupling between neuronal spike trains. Prog Biophys Mol Biol 1989;53:1±31. Sears TA, Stagg D. Short-term synchronization of intercostal motoneuron activity. J Physiol 1976;263:357±381. Toro C, Friehs G, Ojakangas C, Maxwell R, Gates JR, Gumnit RJ, Ebner TJ. 8-12 Hz rhythmic oscillations in human motor cortex during twodimensional arm movements: evidence for representation of kinematic parameters. Electroenceph clin Neurophysiol 1994;93:390±403.