Skip to main content
NCBI home page
The Journal of Physiology logoLink to The Journal of Physiology
. 1999 Apr 15;516(Pt 2):559–570. doi: 10.1111/j.1469-7793.1999.0559v.x

Task-dependent modulation of 15-30 Hz coherence between rectified EMGs from human hand and forearm muscles

J M Kilner *, S N Baker *, S Salenius *, V Jousmäki *, R Hari *, R N Lemon *
PMCID: PMC2269269  PMID: 10087353

Abstract

  1. Recent reports have shown task-related changes in oscillatory activity in the 15-30 Hz range in the sensorimotor cortex of human subjects and monkeys during skilled hand movements. In the monkey these oscillations have been shown to be coherent with oscillatory activity in the electromyographic activity of hand and forearm muscles.

  2. In this study we investigated the modulation of oscillations in the electromyogram (EMG) of human volunteers during tasks requiring precision grip of two spring-loaded levers.

  3. Two tasks were investigated: in the ‘hold’ task, subjects were required to maintain a steady grip force (ca 2·1 N or 2·6 N) for 8 s. In the ‘ramp’ task, there was an initial hold period for 3 s (force ca 2·1 N) followed by a linear increase in grip force over a 2 s period. The task ended with a further steady hold for 3 s at the higher force level (ca 2·6 N).

  4. Surface EMGs were recorded from five hand and forearm muscles in 12 subjects. The coherence of oscillatory activity was calculated between each muscle pair. Frequencies between 1 and 100 Hz were analysed.

  5. Each subject showed a peak in the coherence spectra in the 15-30 Hz bandwidth during the hold task. This coherence was absent during the initial movement of the levers. During the ramp task the coherence in the 15-30 Hz range was also significantly reduced during the movement phase, and significantly increased during the second hold period, relative to the initial hold.

  6. There was coherence between the simultaneously recorded magnetoencephalogram (MEG) and EMG during steady grip in the hold task; this coherence disappeared during the initial lever movement. Using a single equivalent current dipole source model, the coherent cortical activity was localized to the hand region of the contralateral motor cortex. This suggests that the EMG-EMG coherence was, therefore, at least in part, of cortical origin.

  7. The results are discussed in terms of a possible role for synchrony in the efficient recruitment of motor units during maintained grip.

The planning and execution of movements involve areas of the cerebral cortex that are separated in their activity both temporally and spatially. However, the mechanism by which these areas combine to produce a co-ordinated movement is unknown. Studies of the activity of neurones in the sensorimotor cortex in monkeys and humans have shown synchronous oscillatory activity in the 15-30 Hz range (Murthy & Fetz, 1992, 1996a; Sanes & Donoghue, 1993; Stancak & Pfurtscheller, 1996; Baker et al. 1997; Donoghue et al. 1998). These synchronous oscillations have been suggested to link the disparate motor signals together in a manner analogous to the perceptual ‘binding’ of stimulus attributes in the visual cortex (Singer & Gray, 1995), where the observed neuronal synchronization is attributed to neurones that participate in the encoding of related information.

Sanes & Donoghue (1993) and Murthy & Fetz (1996a) have shown that oscillations in monkey sensorimotor cortex are synchronous over large distances (up to 14 mm), suggesting the involvement of large neuronal populations. However, Murthy & Fetz (1996b) concluded that the oscillatory episodes had no consistent relationship with a variety of motor tasks, and suggested that rather than being involved directly in ‘binding’ during movement execution, the oscillations could be a neuronal correlate of attention during sensorimotor tasks.

By contrast, a number of studies have shown that 15-30 Hz oscillations recorded over the sensorimotor cortex in man disappear during a self-paced finger movement, but reappear following movement completion (Jasper & Penfield, 1949; Gastaut, 1952; Salmelin & Hari, 1994; Stancak & Pfurtscheller, 1996). Baker et al. (1997) found that 15-30 Hz local field potential oscillations recorded in monkey primary motor cortex during a precision grip task showed a marked increase during the hold phase of the task. They suggested that the synchronous oscillations could reflect a mode of processing suited to the maintenance of a maintained grip.

Oscillatory activity in motor cortex is coherent with the electromyogram (EMG) of active contralateral hand and forearm muscles (Conway et al. 1995; Salenius et al. 1996, 1997a; Baker et al. 1997). In the monkey, Baker et al. (1997) demonstrated that EMG-EMG coherence showed the same task dependence as the cortical signals, indicating that this coherence could be used to gain important insights into the temporal modulation of oscillatory activity at the cortical level.

In this study we have extended this investigation to man, searching for modulation of coherent oscillatory activity during different precision grip tasks. We confirm the disappearance of oscillations during even a slow, low force movement, and further show a ‘rebound’ phenomenon whereby coherence is increased following a movement over and above that seen before movement. Using combined magnetoencephalogram (MEG) and EMG in one of the tasks studied, we show that at least part of the coherent oscillations seen in the periphery is likely to be generated in the primary motor cortex.

A preliminary account of this work has been published previously (Kilner et al. 1998).

METHODS

Behavioural task

The results reported here were gathered from 12 healthy human subjects (4 male and 8 female) all right-handed, by self report, and aged between 19 and 33 years. All gave informed consent; the studies had local ethical committee approval and the experiments were conducted according to the Declaration of Helsinki. The subjects gripped two spring-loaded levers between the thumb and index finger of their right hand. The levers required an initial force of 1·5 N to overcome the spring tension and then a further force of 0·07 N mm−1. Visual feedback of the lever positions was provided via cursors on a computer video screen. A trial was initiated when the cursors appeared. Subjects then moved the levers so that the cursors were positioned inside target boxes. Two different tasks were tested in a given recording session. In the ‘hold’ task subjects maintained a steady grip for 8 s; this was tested at a lower force (LF) level of 2·1 N or at a higher force (HF) level of 2·6 N. In the ‘ramp’ task, after an initial 3 s hold period, also at 2·1 N force, subjects tracked a linear movement of the target boxes. This lasted 2 s, and required an increase in force to 2·6 N. The trial was completed by a further 3 s hold at the higher force level.

To determine the cortical correlate of the recorded EMG oscillations, whole-scalp MEG recordings were made in 10 other subjects (7 male and 3 female, aged between 22 and 35 years) performing a hold task similar to the one described above (see ‘Cortical source localization’ below).

Recordings

Bipolar surface EMGs were recorded from abductor pollicis brevis (AbPB), first dorsal interosseous (1DI), abductor digiti minimi (AbDM), flexor digitorum superficialis (FDS) and extensor digitorum communis (EDC) muscles. For the 1DI muscle, the hold and ramp tasks required about 9 % of the maximum voluntary contraction. EMGs were amplified (gain 1-10K), high-pass filtered at 30 Hz, and then sampled at 5 kHz by a PC-compatible computer fitted with a 1401+ interface (CED Ltd, Cambridge, UK), together with finger and thumb lever positions and markers indicating task events. The magnitude of cross-talk between EMGs in this study was small. For the muscle pair AbPB/1DI, where physical proximity would indicate the greatest likelihood of cross-talk, we calculated r2= 0·014 as an average from all cross-correlations between AbPB and 1DI EMGs (Baker et al. 1998).

In the MEG experiment cortical signals were recorded with a whole-scalp neuromagnetometer in a magnetically shielded room. The subject's head fitted snugly against the helmet-shaped bottom surface of a Neuromag-122 magnetometer (Ahonen et al. 1993; Neuromag Ltd, Helsinki, Finland). MEG and EMG signals were recorded with passbands of 0·03-300 Hz and 3-300 Hz, respectively, digitized at 600 Hz, and stored on magneto-optic disks for off-line analysis. The exact position of the head with respect to the sensor array was determined by measuring magnetic signals from three indicator coils placed on the scalp. The coil locations, with respect to three pre-determined landmarks on the skull, were identified with a 3D digitizer, and this information was used to superimpose sources for the MEG signals on individual magnetic resonance images (MRI), obtained with a 1-T Siemens Magnetom device (Siemens AG, Munich, Germany).

Analysis

Off-line, finger and thumb lever position signals were examined by eye; trials in which the subject had failed to move the levers as required were excluded. The first 50 correctly performed trials were used in subsequent analysis. EMG recordings were rectified, and then smoothed by averaging successive sets of 25 points. The resultant time series, effectively sampled at 200 Hz, was then used for spectral analysis. The maximum detectable frequency was therefore 100 Hz (Nyquist theorem; Newland, 1993).

In order to measure the correlation between the signals in the frequency domain, coherence was calculated. Coherence is a measure of the amplitude and phase correlation within a particular frequency band between two sources, and is bounded between 0 and 1. The calculations for coherence are described in Farmer et al. (1993) and Baker et al. (1997). A first analysis used EMG taken from 2 to 8 s after the task onset in the LF hold task; coherence was calculated between all available pairs of EMG recordings using an FFT (fast Fourier transform) window of 128 points, permitting a frequency resolution of 1·54 Hz. Coherence larger than S may be considered significantly different from zero with P < α where

graphic file with name tjp0516-0559-m1.jpg (1)

L is the number of disjoint sections used to calculate the coherence (Rosenberg et al. 1989). We used α= 0·05. Muscle pairs showing no significant coherence peak were excluded from further analysis; 15/120 pairs were so rejected. For coherent muscle pairs the frequency at the coherence peak, and the onset and offset of the peak, were measured by eye.

To determine the task dependence of coherence, each sweep was divided into consecutive 0·16 s blocks of 32 points aligned relative to task onset. The fast Fourier transform of each block was calculated, and the cross and power spectra found by combining successive sets of two blocks, averaged across trials. This produced a coherence versus time map. At the effective sampling rate of 200 Hz used, the map had a frequency resolution of 6·25 Hz, and a time resolution of 0·32 s. For the task-dependent coherence analysis of the ramp task, the increasing trend in rectified EMG level (see Fig. 2A) was removed using linear regression techniques.

Figure 2. Changes in coherence in EMG activity during performance of precision grip tasks.

Figure 2

Open in a new tab

A, modulation of coherence calculated for muscles AbPB and 1DI in subject 12 during the LF hold and ramp tasks. Below each plot are the rectified averaged and smoothed EMGs for both muscles and the average finger and thumb lever positions. B, task-dependant modulation of the single subject ξ-transformed coherence combined across all significant muscle pairs for the frequency bin that contained the mean peak coherence in that subject. In each plot the red line indicates the LF hold task and the black line the ramp task. The period of the ramp is shown by the dotted lines and yellow shading. The subjects have been divided into two groups, A and B, depending on the nature of their modulation (see text).

Coherence values for muscle pairs were transformed as follows:

graphic file with name tjp0516-0559-m2.jpg (2)

where C is the coherence value; the dependence of C and Z on frequency, and time relative to task onset, is suppressed for simplicity of notation. Such a value can be considered as an estimate of the ‘true’Z-transformed coherence between the two muscles, with a mean equal to this underlying value; it will be normally distributed with a standard deviation of approximately 1 (Rosenberg et al. 1989). For a given subject, values were combined across N muscle pairs to produce a composite value according to:

graphic file with name tjp0516-0559-m3.jpg (3)

If the individual Zi were independent, ξ would have a standard deviation of one. However, the Zi are likely to be correlated since they are formed from all possible pairwise combinations of a single set of recordings. In addition, the transform of eqn (2) can be shown to produce values with standard deviations which depart from the theoretical value of unity for small coherences (Benignus, 1969). For these reasons, no assumptions were made about the variance of the measure ξ when testing for significance. Instead, a Student's paired t test run across subjects was used to determine whether values of ξ during a period of interest in the task were significantly different from those during a control period.

To combine data across subjects, it was necessary to sum the single subject ξ-values. If the raw values were simply added, the result would be biased by giving undue weight to data from subjects which showed large changes in coherence with task. In order to avoid this, and to give equal weight to each subject's data, the series of ξ-values in each frequency bin as a function of time during the task were scaled to have a standard deviation of one before the summation across subjects.

For each significantly coherent muscle pair a coherence spectrum was calculated using an 128 point window during the three different phases of the ramp task; initial hold (H1), ramp (R) and final hold (H2). Any trends in rectified EMG during the ramp were removed prior to coherence analysis, as described above. These were converted to ξ-values, as above, and were normalized, such that the maximum value across the three phases of the ramp task was equal to 1. The normalized values were summed across muscle pairs and subjects. Student's paired t tests were calculated for the coherence spectra for H1 compared with R and H2 for bins between 1 and 60 Hz.

Cortical source localization

Sources of oscillatory MEG signals were modelled in the time domain as equivalent current dipoles (ECDs; Hämäläinen et al. 1993), found by a least-squares search based on the MEG signal distribution. The EMG was first converted to a series of events by thresholding; the level of this threshold was set to obtain 10-15 triggers per second. These triggers were then used to form an EMG-triggered average of the MEG signals. Source localization was restricted to signals from the 40 detectors centred over the rolandic area in each hemisphere, since other signals showed no consistent deflections in the EMG-triggered averages. The value of the mean field at a lag chosen to coincide with the peak deflection was used to produce a field map over the scalp. An ECD model was then fitted to this. Only sources which accounted for more than 85 % of the field variance were accepted.

Sources were displayed on co-registered MRI images of individual subjects as described above. Their location was compared with that of the ECD source fitted to the 20 ms somatosensory-evoked field (N20m) following median nerve stimulation at the wrist, which indicates the location of the hand representation of somatosensory cortex. MRIs and the N20m were not recorded specially for these experiments, but were taken from a laboratory database compiled from previous studies in the same subjects.

RESULTS

EMG-EMG coherence during LF hold and ramp precision grip tasks

An example of raw EMG data recorded during the LF hold task is shown in Fig. 1A. The two hand muscles (1DI and AbPB) illustrated showed bursts of activity just before the initial lever movement, followed by steady co-contraction during the hold period. Figure 1B shows the coherence spectrum for this muscle pair during this period (see bar in Fig. 1A). There was a clear peak in the 15-30 Hz range with a maximum at 22 Hz. The frequency of coherence showed some variability, both within different muscle pairs for each subject, and across subjects. This variability is illustrated by Fig. 1C, which shows the range of onset, peak and offset frequencies for each muscle pair for each subject. In order to study the modulation of coherence with the task coherence was recalculated with a higher time resolution but a lower frequency resolution, such that each frequency bin was 6·25 Hz wide. These new frequency bins are illustrated on Fig. 1C by the vertical dashed lines. At this resolution all subjects have a peak in their coherence which falls in one of three frequency bins between 12·5 and 31·25 Hz.

Figure 1. Coherence of EMG activity during the hold task.

Figure 1

Open in a new tab

A, EMG activity from two hand muscles (1DI and AbPB) and index finger and thumb lever positions recorded from subject 12 during a single trial of the LF hold task. B, the coherence spectrum calculated from the muscles shown in A for 50 trials. Coherence was calculated during the steady hold phase (horizontal bar in A). C, frequency distribution of the coherence peak for the muscle pairs showing significant coherence in each subject. For each subject the range of frequencies for the minimum, maximum and peak frequencies for different muscle pairs are indicated by the horizontal bars and the average values are shown by filled squares, filled triangles and open circles, respectively. The minimum, maximum and peak frequencies were measured by eye. The arrows in B give examples of the points measured.

Task-related modulation in EMG-EMG coherence

When coherence was calculated as a function of time following task onset (see Methods), a task dependence was observed. This is shown in Fig. 2A for the same muscle pair and subject as shown in Fig. 1A and B, in both the LF hold and ramp tasks. In these colour maps, frequency is shown on the ordinate, and time relative to task onset on the abscissa. The colour scale gives the strength of coherence for a given time and frequency bin. Both plots show a clear coherence peak in the 15-30 Hz range. This was greatest during the steady hold and was clearly decreased during movement, either at the start or end of the hold period, or during the ramp.

Figure 2B shows data combined across muscle pairs for all the subjects studied (measure ξ). For each subject a plot for the LF hold task (red), and ramp task (black), is shown, at the frequency band that contained the mean peak of coherence for that subject (see Fig. 1C). Whilst some broad patterns emerge, there is considerable inter-subject variability. The subjects have been divided into two groups (A and B), depending on the pattern of coherence during the ramp task. Those in Group A showed a significant difference between the first hold period and the ramp period at P < 0·01, calculated using Student's paired t test on the constituent Z-values whose sum ξ is shown (eqns (2) and (3)), whereas those in Group B did not. Those in Group A showed a rise in coherence during the first hold period (Fig. 2, H1), a decrease during the ramp (R) and an increase during the second hold period (H2). Those in Group B showed low coherence during the first hold and ramp periods followed by an increase during the second hold period. Across subjects, there was a wide variability in the size of the changes in the coherence. For Group A subjects the mean level of the coherence peak in the 15-30 Hz band across all muscle pairs during H1 was 0·138 (range: 0·028-0·495); for Group B subjects the mean value was 0·075 (range: 0·022-0·257). To ensure that data from each subject were given equal weight we normalized the data before combining across subjects as described in Methods.

Figure 3A shows changes in coherence in the ramp task, combined across subjects by adding the normalized ξ-values. This has been done for subjects in Group A and Group B separately, and also for all subjects together. Plots are shown for the three frequency bins that encompassed the mean peaks of coherence (12·5-18·75 Hz, 18·75-25 Hz and 25-31·25 Hz), as illustrated in Fig. 1C. For the purpose of making statistical tests, three periods, each four bins long, have been defined, corresponding to different phases of the ramp task: H1 and H2 (the two hold phases) and R (the ramp movement). These three periods are indicated by the horizontal bars below the time scale in Fig. 3A. Student's paired t test compared H1 with H2 and H1 with R.

Figure 3. Difference in task-related coherence across muscle pairs and across subjects.

Figure 3

Open in a new tab

A, the normalized ξ-values combined across muscle pairs and subjects for the ramp task for the 3 frequency bins that encompass the average peak coherence for each subject (see Fig. 1C). Data are shown separately for Group A, Group B, and both groups combined. Asterisks denote significant differences between the first hold period (H1) and the ramp period (R), and between H1 and the second hold period (H2) for the ramp task (*P < 0·01; **P < 0·001, Student's paired t test). B, the normalized ξ-values of the coherence spectra combined across muscle pairs and subjects for the 3 phases of the ramp task (H1, R and H2) for Group A, Group B, and both groups combined.

In the ramp task, all subject groups (A, B and A+B) showed a significant increase in all three frequency ranges tested between H1 and H2. Group A subjects, and the combined subject group (A+B), showed a significant decrease in the 12·5-25 Hz range from H1 to R. Group B subjects showed no significant (P > 0·01) difference from H1 to R in all three frequency ranges.

These results are mirrored in the normalized ξ-values calculated from the coherence spectra for the three phases of the ramp task summed across significant muscle pairs and across subjects (Fig. 3B). For the all-subject group (A+B) there was significantly greater coherence in the 15-30 Hz range for the H2 period than for the corresponding peaks during the H1 period (P < 0·001). During the ramp movement there was a decrease in the coherence in the 15-30 Hz bandwidth in all data groups but this decrease was significant only in Group A and the all-subject data (P < 0·001); the lack of a significant decrease for the Group B subjects was due to the low coherence during the H1 period in these subjects. No significant differences in coherence (P > 0·01) around the 10 and 40 Hz frequency bins were observed between any of the three phases of the ramp task in any of the three data groups.

EMG-EMG coherence during the hold task at different force levels

The rebound in 15-30 Hz coherence seen in Fig. 3A following the ramp movement may be a genuine after-effect of movement performance; alternatively, it is possible that it is simply caused by the higher force level which subjects had to exert during the second hold, with its altered efferent outflow and afferent sensory feedback. In order to investigate this, we compared hold tasks at each of the two force levels present in the ramp task. Of the 12 subjects investigated, 10 performed the hold task at both the lower force (LF) and higher force (HF) levels (see Methods). Figure 4A shows normalized ξ-values calculated from the muscle pairs of these 10 subjects for which a peak in the estimated coherence spectra above the 95 % confidence level was present. Both plots for the LF hold and HF hold tasks have a similar structure with a peak in the 15-30 Hz range. Figure 4B shows the difference in the normalized ξ-values for the two tasks. Student's paired t tests show there to be no significant difference (P > 0·01) in the 15·38-32·31 Hz range between the two tasks.

Figure 4. Difference in coherence between the hold task at different force levels.

Figure 4

Open in a new tab

A, the normalized ξ-values of the coherence spectra combined across muscle pairs and subjects for the steady hold period of the lower force (LF) hold and higher force (HF) hold tasks. B, the difference in ξ-values between the two hold tasks combined across muscle pairs and subjects. Student's paired t tests on the Z-values which were summed to yield ξ showed there to be no significant differences (P > 0·01) between the 2 holds at any frequency.

Coherence between cortical (MEG) and muscle (EMG) activity

To show that coherence between hand muscles during precision grip is, at least in part, due to oscillatory activity at the cortical level, coherence was also calculated between MEG and EMG during the LF hold task. Figure 5A shows the spectra for coherence between whole-scalp MEG recordings and 1DI EMG during the LF hold task in one subject. Figure 5B illustrates one of the traces at a larger scale and Fig. 5C shows the task-dependent modulation of coherence at the peak frequency for this MEG/EMG pair. These data show the same modulation and coherence bandwidth as the EMG-EMG coherence measurements: the peak lies between 15 and 30 Hz, and was seen only during the steady hold period and not during the periods where the levers were being moved into target.

Figure 5. Coherence between magnetoencephalographic and EMG recordings during the hold task.

Figure 5

Open in a new tab

A, coherence spectra calculated separately for each of the 122 MEG channels and the EMG signal from the right 1DI during the hold task for one subject. Coherence was calculated during the steady hold phase; the head is viewed from above. The upper traces of each pair of records correspond to the latitudinal and the lower traces to the longitudinal derivatives of the magnetic field perpendicular to the head, as illustrated by the schematic diagrams at the top which also give the head orientation. The channel indicated by a box is enlarged in B. B, coherence spectrum from a channel above the left sensorimotor cortex. The 95 % confidence interval for this spectrum is displayed as the dotted line. C, MEG-EMG coherence values for the 18·75-25 Hz range as a function of time during the hold task for the channel whose coherence spectrum is shown in B. The average forces of the finger and thumb levers for the task are shown above. D, sources corresponding to the MEG activity coherent with the 1DI EMG activity, and to the N20m component evoked by electrical stimulation of the right median nerve, indicating the location of the somatosensory hand area. The sources are superimposed on the surface-rendered MRI of the subject's brain. E, magnetic field pattern corresponding to the EMG-triggered MEG deflection preceding the onset of EMG complexes by 25 ms obtained by averaging MEG relative to EMG activity (source shown in D). The field map is displayed on the helmet-shaped sensor array with the head viewed from above. Continuous lines indicate outward and dashed lines inward magnetic flow (contour step 4 fT cm−1). The arrow indicates the location of the current dipole source shown in D.

Figure 5D shows the location of an equivalent current dipole (ECD) for the coherent cortical activity (open circle), superimposed on an MRI surface-rendered image of this subject. The source of the somatosensory-evoked field following median nerve stimulation (N20m) is also shown (filled circle). The coherent cortical activity was just anterior to the N20m source, suggesting generation in the hand representation of the precentral gyrus. The source of coherent activity was in a similar position for all subjects.

Figure 5E gives a map above the scalp surface of the EMG-triggered MEG pattern at a 25 ms lag between MEG and EMG. The pattern is typical of that produced by a single ECD, and thus agrees with the single dipole model, whose location is indicated by the arrow.

DISCUSSION

We have demonstrated a marked task-related modulation in the oscillatory activity of human EMG recorded during a precision grip task. This result supports and extends our recent observations in the monkey (Baker et al. 1997) showing that 20-30 Hz coherence between pairs of hand muscle EMGs was only present during the hold phase of the precision grip task. The MEG recordings show that the cortical activity coherent with that in the muscles originates from the motor cortex, and has the same task-related modulation observed in the monkey by Baker et al. (1997), i.e. coherence is present during the hold period and not during periods in which the levers are moved into the target boxes. In addition, we have shown here the presence of a ‘rebound’ phenomenon, by which the degree of coherence of these oscillations is increased in a second hold period following a period of fine, controlled movement.

As in previous work (Farmer et al. 1993; Baker et al. 1997), the coherence peak seen in the present recordings was centred around 20 Hz. However, Fig. 1C indicates considerable variability in frequency. Single subjects showed broad (ca 10 Hz wide) peaks; over the 12 subjects tested, the coherence maximum varied over a 10 Hz range. Hence overall, coherence was seen from below 10 Hz to nearly 40 Hz. Care must therefore be taken in attempting to identify the frequencies involved in the task-related changes described here with those reported in earlier studies.

In both electroencephalogram (EEG) and MEG recordings from sensorimotor cortex, three main bands of oscillatory activity have been recognized. Oscillations at 10 Hz decrease before and during movement (Chatrian et al. 1959; Salmelin & Hari, 1994; Stancak & Pfurtscheller, 1996). Oscillations in the 20-30 Hz bandwidth also decline during movements, and show a rebound phenomenon afterwards with transient power increase (Salmelin & Hari, 1994; Pfurtscheller et al. 1996; Feige et al. 1996). Finally, ‘gamma’ range oscillations (ca 40 Hz) may show a brief synchronization with movement (Pfurtscheller & Neuper, 1992; Salenius et al. 1996), although power in this range is not always seen (Nashmi et al. 1994; Salmelin & Hari, 1994) or if seen, may not show movement-related synchronization (Feige et al. 1996). Oscillations in EMG activity at these frequencies, the Piper rhythm, have also been recorded during strong muscle contractions (Piper, 1907; Brown, 1997; Brown et al. 1998). In MEG data which did show power between 10 and 40 Hz, only the central frequencies from 16 to 32 Hz showed coherence with the EMG of a contracting muscle (Conway et al. 1995). Although our MEG-EMG coherence calculations did occasionally rise above significance for higher and lower frequencies (e.g. Fig. 5B), the levels of coherence were much lower than that seen around 25 Hz. The EMG-EMG synchronization seen here probably reflects, at least in part, a cortical input at 15-30 Hz. The wide spread of frequencies observed is therefore unlikely to reflect the intrusion of multiple cortical carrier waves into the EMG.

The EMG-EMG coherence illustrated in Fig. 3 shows a decrease during the ramp phase of the task, but increases following the cessation of movement to reach a higher level than that seen before the ramp movement began. This rebound is clearly a result of the preceding movement phase, as a there was no significant difference between the EMG-EMG coherence for steady hold tasks at the corresponding force levels for the two hold periods in the ramp task (Fig. 4). The task modulations observed may correspond to the suppression of the 20 Hz EEG and MEG power before and rebound after voluntary movement (Salmelin & Hari, 1994; Pfurtscheller et al. 1997). There was a rebound of coherence after the movement across all three frequency bands investigated (Fig. 3A), and comparison of the coherence measure versus frequency for the pre- and post-movement periods reveals no obvious shift to higher or lower frequencies during the rebound effect. It is important to remember that our tests for coherence at the EMG level select out that part of cortical activity which is capable of influencing motor output. Feige et al. (1996) proposed that the pre-movement 20 Hz rhythm and the post-movement rebound result from different cortical generators. If this is the case, they clearly both encompass the output cells in the motor cortex hand representation.

It has been shown that the sensorimotor cortex 10 Hz μ-rhythm has a non-sinusoidal ‘arch-shaped’ waveform (Adrian & Matthews, 1934; Jasper & Penfield, 1949; Tiihonen et al. 1989), which in a Fourier analysis must lead to power at the 20 Hz first harmonic. This 20 Hz harmonic shows a different relationship to movement from the independent 20 Hz rhythm. According to Pfurtscheller et al. (1997) the 10 Hz rhythm and its harmonic decline several seconds prior to movement, and remain desynchronized afterwards, whereas the 20 Hz non-harmonic rhythm desynchronizes only a few hundred milliseconds before movement and then exhibits a rapid post-movement rebound. Our findings at the EMG level are consistent with such a view. We did not see any distinct EMG-EMG coherence peak at 10 Hz, suggesting that the 10 Hz μ-rhythm component does not have a large influence on EMG (Conway et al. 1995; Baker et al. 1997; Salenius et al. 1997a) in agreement with the data of Fig. 5A and B. Further, the changes in the level of the 20 Hz EMG coherence during the hold task are like those seen for the cortical 20 Hz, not 10 Hz oscillations. The very small or absent synchronization of EMG by the cortical 10 Hz rhythm may be explained by studies which have suggested a more posterior source localization for the 10 Hz than for the 20 Hz rhythm, placing it in postcentral cortex (Salmelin & Hari, 1994; Salenius et al. 1997b). Invasive recordings of local field potentials in monkey motor cortex have not shown 10 Hz oscillations (Murthy & Fetz, 1992, 1996a, b; Sanes & Donoghue, 1993; Baker et al. 1997; Donoghue et al. 1998).

A number of previous reports have shown oscillatory activity in the motor output, either in position or force records (tremor) or in the EMG (Farmer et al. 1993; Wessberg & Vallbo, 1995; Brown, 1996; McAuley et al. 1997), although none to date have investigated the modulation with performance of a motor task. Peaks can be seen in the power spectra of peripheral measures at 10, 20 and 40 Hz (McAuley et al. 1997); coherence between motor units appears mainly limited to the 20 Hz band (Farmer et al. 1993), in agreement with the present findings. However, power peaks at all three frequencies were unaltered by peripheral loading and partial peripheral anaesthesia of the hand, suggesting central generation (McAuley et al. 1997). As already noted above, our data showed broad coherence peaks, which across subjects encompassed all three of these frequency bands. The data of Fig. 3B indicate that the apparent single peak may have been composed of three separate peaks at the frequencies found by McAuley et al. (1997). Each of the traces in Fig. 3B for the ramp period demonstrates a profound suppression of 20 Hz coherence relative to the hold, but there is a hint (strongest for the Group B subjects) that coherence at 10 and 40 Hz is preserved. If so, this would indicate a separate, probably non-cortical central source of these rhythms. Using a bicoherence analysis (not shown) (Brillinger 1975), we have verified that power in the 10, 20 and 40 Hz bands in our EMG data shows no inter-relationship, suggesting that the frequency components are not simply harmonics of a common oscillator. However, whatever their central source, the 10 and 40 Hz rhythms exhibit no striking modulation during the task we have tested.

The function of the 20 Hz rhythmicity remains unclear. The disappearance of coherence at this frequency during movement, and the rapid rebound following it, supports the notion that this is an ‘idling’ rhythm, representing cortex at rest (Adrian & Matthews, 1934; Buser, 1987; Lopes da Silva, 1991; Salmelin & Hari, 1994). However, in contrast to many EEG-based studies, in our experiments the subjects were not at rest during the period of maximum coherence, but were actively maintaining a steady grip (cf. Baker et al. 1997). The variation in the magnitude of coherence between subjects (Fig. 2B) indicates that 20 Hz oscillations are not essential for task performance, but could reflect a form of control used by the motor system to varying extents in different subjects. A number of subjects showed a slow decline in the coherence level during the hold task (red lines in Fig. 2B), despite a constant level of EMG and unchanging lever kinematics. Such a pattern can be clearly seen for subject 12 illustrated in Fig. 2A (LF hold task), whose coherence plot (red trace arrowed in Fig. 2B) shows a slow decline. Such a pattern is inconsistent with a purely idling rhythm.

One possibility is that the cortical oscillations act as an efficient means of recruiting motoneurones, whilst maintaining as low a corticospinal firing rate as possible. Theoretical studies show that a synchronous input can produce more output firing than an asynchronous one (Murthy & Fetz, 1994; Baker, 1997). During steady grip some primary motor cortex neurones, including cortico-motoneuronal (CM) cells facilitating the gripping muscles, show a negative correlation with grip force level (Wannier et al. 1991; Maier et al. 1993). During such a task, the motor cortex could be operating in a mode of increased oscillatory activity, with efficient motoneurone recruitment. Here the dominant pattern of muscular activity would be co-contraction.

A quite distinct mode of operation would be required for skilled movement. The highly predictable nature of oscillatory firing from one cycle to the next could limit the amount of information processing of which the cortex is capable. The cortical output cells are deeply embedded in the network responsible for cortical oscillations (Baker et al. 1997; Pinches et al. 1997), such that their activity cannot be decoupled from the global oscillations. Thus for movement it might be necessary to suppress oscillatory activity. The dominant muscle pattern is now one of fractionation and this is supported by smaller groups of CM cells acting in an asynchronous manner, firing phasically at high rates to recruit motoneurones without the benefit of synchrony (Maier et al. 1993; Bennett & Lemon, 1996). Thus the modulation of oscillations in our ramp task could reflect these two different modes of operation. Whilst speculative, the hypothesis that these two different modes of operation exist, and are mutually exclusive, can explain the present experimental findings on the task dependence of these rhythms.

In conclusion, we have confirmed the presence of coherent oscillatory activity in human hand and forearm muscles and have shown that the 15-30 Hz component has task-dependent modulation; being present during steady grips, decreased during movements and greatest during hold periods following movement. We have further confirmed that these oscillations are coherent with MEG signals from the contralateral hand area of the primary motor cortex. These results are consistent with the hypothesis that motor cortical oscillations are used as a mechanism of more efficient motor unit recruitment.

Acknowledgments

This study was supported by The Wellcome Trust, the Medical Research Council, the Academy of Finland, and the EU's Large-Scale Facility BIRCH at the Low Temperature Laboratory, Espoo.

References

  1. Adrian ED, Matthews BH. The Berger rhythm: potential changes from the occipital lobes in man. Brain. 1934;57:355–385. doi: 10.1093/brain/awp324. [DOI] [PubMed] [Google Scholar]
  2. Ahonen AI, Hämäläinen MS, Kajola MJ, Knuutila JET, Laine PP, Lounasmaa OV, Parkkonen LT, Simola JT, Tesche CD. 122-channel SQUID instrument for investigating the magnetic signals from the human brain. Physica Scripta. 1993;49:198–205. [Google Scholar]
  3. Baker SN. Quantification of the relative efficacies of asynchronous and oscillating inputs to a motoneurone pool using a computer model. The Journal of Physiology. 1997;504.P:116. P. [Google Scholar]
  4. Baker SN, Olivier E, Lemon RN. Coherent oscillations in monkey motor cortex and hand muscle EMG show task-dependent modulation. The Journal of Physiology. 1997;501:225–241. doi: 10.1111/j.1469-7793.1997.225bo.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Baker SN, Olivier E, Lemon RN. An investigation of the intrinsic circuitry of the motor cortex of the monkey using intra-cortical microstimulation. Experimental Brain Research. 1998;123:397–411. doi: 10.1007/s002210050585. [DOI] [PubMed] [Google Scholar]
  6. Benignus VA. Estimation of the coherence spectrum and its confidence interval using the fast fourier transform. IEEE Transactions in Acoustics. 1969;17:145–150. [Google Scholar]
  7. Bennett KMB, Lemon RN. Corticomotoneuronal contribution to the fractionation of muscle activity during precision grip in the monkey. Journal of Neurophysiology. 1996;75:1826–1842. doi: 10.1152/jn.1996.75.5.1826. [DOI] [PubMed] [Google Scholar]
  8. Brillinger DR. Time Series - Data Analysis and Theory. New York: Holt, Rinehart & Winston; 1975. [Google Scholar]
  9. Brown P. Muscle sound during human wrist movements. The Journal of Physiology. 1996;494.P:68. P. [Google Scholar]
  10. Brown P. Muscle sound in Parkinson's disease. Lancet. 1997;349:533–535. doi: 10.1016/S0140-6736(97)80086-4. [DOI] [PubMed] [Google Scholar]
  11. Brown P, Salenius S, Rothwell JC, Hari R. The cortical correlate of the Piper rhythm in man. The Journal of Physiology. 1998;507.P:34. doi: 10.1152/jn.1998.80.6.2911. P. [DOI] [PubMed] [Google Scholar]
  12. Buser P. Thalamocortical mechanisms underlying synchronised EEG activity. In: Halliday AM, Butler SR, Paul R, editors. A Textbook of Clinical Neurophysiology. Chichester: John Wiley; 1987. pp. 595–621. [Google Scholar]
  13. Chatrian GE, Petersen MC, Lazarte JA. The blocking of the rolandic wicket rhythm and some central changes related to movement. Electroencephalography and Clinical Neurophysiology. 1959;11:497–510. doi: 10.1016/0013-4694(59)90048-3. [DOI] [PubMed] [Google Scholar]
  14. Conway BA, Halliday DM, Shahani U, Maas P, Weir AL, Rosenberg JR. Synchronization between motor cortex and spinal motoneuronal pool during the performance of a maintained motor task in man. The Journal of Physiology. 1995;489:917–924. doi: 10.1113/jphysiol.1995.sp021104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Donoghue JP, Sanes JN, Hatsopoulos NG, Gaal G. Neural discharge and local field potential oscillations in primate motor cortex during voluntary movements. Journal of Neurophysiology. 1998;79:159–173. doi: 10.1152/jn.1998.79.1.159. [DOI] [PubMed] [Google Scholar]
  16. Farmer SF, Bremner FD, Halliday DM, Rosenberg JR, Stephens JA. The frequency content of common synaptic inputs to motoneurones studied during voluntary isometric contraction in man. The Journal of Physiology. 1993;470:127–155. doi: 10.1113/jphysiol.1993.sp019851. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Feige B, Kristeva-Feige R, Rossi S, Pizzella V, Rossini PM. Neuromagnetic study of movement-related changes in rhythmic brain activity. Brain Research. 1996;734:252–260. [PubMed] [Google Scholar]
  18. Gastaut H. Étude electrocorticographique de la reactivité des rythmes rolandiques. Review of Neurology (Paris) 1952;87:176–182. [PubMed] [Google Scholar]
  19. Hämäläinen M, Hari R, Ilmoniemi RJ, Knuutila J, Lounasmaa OV. Magnetoencephalography - theory, instrumentation, and applications to noninvasive studies of the working human brain. Reviews of Modern Physics. 1993;65:413–498. 10.1103/RevModPhys.65.413. [Google Scholar]
  20. Jasper H, Penfield W. Electrocorticograms in man: Effect of voluntary movement upon the electrical activity of the precentral gyrus. Archiv für Psychiatrie und Zeitschrift. 1949;183:163–174. [Google Scholar]
  21. Kilner JM, Baker SN, Salenius S, Hari R, Lemon RN. 15–30 Hz coherence between rectified EMGs from human hand muscles has task-related modulation. The Journal of Physiology. 1998;509.P:175. doi: 10.1111/j.1469-7793.1999.0559v.x. P. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lopes da Silva F. Neural mechanisms underlying brain waves: from neural membranes to networks. Electroencephalography and Clinical Neurophysiology. 1991;79:81–93. doi: 10.1016/0013-4694(91)90044-5. [DOI] [PubMed] [Google Scholar]
  23. McAuley JH, Rothwell JC, Marsden CD. Frequency peaks of tremor, muscle vibration and electromyographic activity at 10 Hz, 20 Hz and 40 Hz during human finger muscle contraction may reflect rhythmicities of central neural firing. Experimental Brain Research. 1997;114:525–541. doi: 10.1007/pl00005662. [DOI] [PubMed] [Google Scholar]
  24. Maier M, Bennett KMB, Hepp-Reymond M-C, Lemon RN. Contribution of the monkey cortico-motoneuronal system to the control of force in precision grip. Journal of Neurophysiology. 1993;69:772–785. doi: 10.1152/jn.1993.69.3.772. [DOI] [PubMed] [Google Scholar]
  25. Murthy VN, Fetz EE. Coherent 25-hz to 35-hz oscillations in the sensorimotor cortex of awake behaving monkeys. Proceedings of the National Academy of Sciences of the USA. 1992;89:5670–5674. doi: 10.1073/pnas.89.12.5670. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Murthy VN, Fetz EE. Effects of input synchrony on the firing rate of a 3-conductance cortical neuron model. Neural Computation. 1994;6:1111–1126. [Google Scholar]
  27. Murthy VN, Fetz EE. Oscillatory activity in sensorimotor cortex of awake monkeys: synchronization of local field potentials and relation to behavior. Journal of Neurophysiology. 1996a;76:3949–3967. doi: 10.1152/jn.1996.76.6.3949. [DOI] [PubMed] [Google Scholar]
  28. Murthy VN, Fetz EE. Sychronization of neurons during local field potential oscillations in sensorimotor cortex of awake monkeys. Journal of Neurophysiology. 1996b;76:3968–3982. doi: 10.1152/jn.1996.76.6.3968. [DOI] [PubMed] [Google Scholar]
  29. Nashmi R, Mendonca AJ, MacKay WA. EEG rhythms of the sensorimotor region during hand movements. Electroencephalography and Clinical Neurophysiology. 1994;91:456–467. doi: 10.1016/0013-4694(94)90166-x. [DOI] [PubMed] [Google Scholar]
  30. Newland DE. An Introduction to Random Vibrations, Spectral and Wavelet Analysis. 3. Harlow, UK: Longman; 1993. [Google Scholar]
  31. Pfurtscheller G, Neuper C. Simultaneous EEG 10 Hz desynchronization and 40 Hz synchronization during finger movements. NeuroReport. 1992;3:1057–1060. doi: 10.1097/00001756-199212000-00006. [DOI] [PubMed] [Google Scholar]
  32. Pfurtscheller G, Stancak A, Edlinger G. On the existence of different types of central beta rhythms below 30 Hz. Electroencephalography and Clinical Neurophysiology. 1997;102:316–325. doi: 10.1016/s0013-4694(96)96612-2. [DOI] [PubMed] [Google Scholar]
  33. Pfurtscheller G, Stancak A, Neuper C. Post-movement beta synchronization. A correlate of an idling motor area? Electroencephalography and Clinical Neurophysiology. 1996;98:281–293. doi: 10.1016/0013-4694(95)00258-8. [DOI] [PubMed] [Google Scholar]
  34. Pinches EM, Baker SN, Lemon RN. Quantitative assessment of phase locking in discharge of identified pyramidal tract neurones during 25 Hz oscillations in monkey motor cortex. The Journal of Physiology. 1997;501.P:36. P. [Google Scholar]
  35. Piper H. Uber den willkurlichen Muskeltetanus. Pflügers Archiv. 1907;119:301–338. [Google Scholar]
  36. Rosenberg JR, Amjad AM, Breeze P, Brillinger DR, Halliday DM. The fourier approach to the identification of functional coupling between neuronal spike trains. Progress in Biophysics and Molecular Biology. 1989;53:1–31. doi: 10.1016/0079-6107(89)90004-7. [DOI] [PubMed] [Google Scholar]
  37. Salenius S, Portin K, Kajola M, Salmelin R, Hari R. Cortical control of human motoneuron firing during isometric contraction. Journal of Neurophysiology. 1997a;77:3401–3405. doi: 10.1152/jn.1997.77.6.3401. [DOI] [PubMed] [Google Scholar]
  38. Salenius S, Salmelin R, Neuper C, Pfurtscheller G, Hari R. Human cortical 40 Hz rhythm is closely related to EMG rhythmicity. Neuroscience Letters. 1996;213:75–78. doi: 10.1016/0304-3940(96)12796-8. [DOI] [PubMed] [Google Scholar]
  39. Salenius S, Schnitzler AK, Salmelin R, Jousmäki V, Hari R. Modulation of human cortical rolandic rhythms during natural sensorimotor tasks. Neuroimage. 1997b;5:221–228. doi: 10.1006/nimg.1997.0261. [DOI] [PubMed] [Google Scholar]
  40. Salmelin R, Hari R. Spatiotemporal characteristics of sensorimotor neuromagnetic rhythms related to thumb movement. Neuroscience. 1994;60:537–550. doi: 10.1016/0306-4522(94)90263-1. [DOI] [PubMed] [Google Scholar]
  41. Sanes JN, Donoghue JP. Oscillations in local field potentials of the primate motor cortex during voluntary movement. Proceedings of the National Academy of Sciences of the USA. 1993;90:4470–4474. doi: 10.1073/pnas.90.10.4470. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Singer W, Gray CM. Visual feature integration and the temporal correlation hypothesis. Annual Review of Neuroscience. 1995;18:555–586. doi: 10.1146/annurev.ne.18.030195.003011. [DOI] [PubMed] [Google Scholar]
  43. Stancak A, Pfurtscheller G. Mu-rhythm changes in brisk and slow self-paced finger movements. NeuroReport. 1996;7:1161–1164. doi: 10.1097/00001756-199604260-00013. [DOI] [PubMed] [Google Scholar]
  44. Tiihonen J, Kajola M, Hari R. Magnetic mu rhythm in man. Neuroscience. 1989;32:793–800. doi: 10.1016/0306-4522(89)90299-6. [DOI] [PubMed] [Google Scholar]
  45. Wannier TMJ, Maier MA, Hepp-Reymond M-C. Contrasting properties of monkey somatosensory and motor cortex neurons activated during the control of force in precision grip. Journal of Neurophysiology. 1991;65:572–589. doi: 10.1152/jn.1991.65.3.572. [DOI] [PubMed] [Google Scholar]
  46. Wessberg J, Vallbo Å B. Coding of pulsatile motor output by human muscle afferents during slow finger movements. The Journal of Physiology. 1995;485:271–282. doi: 10.1113/jphysiol.1995.sp020729. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from The Journal of Physiology are provided here courtesy of The Physiological Society

RESOURCES