Brain Research Bulletin, Vol. 34, No. 6, pp. 519-531, 1994 Copyright 0 1994 Elsevier Science Ltd
Printed in the USA. All rights reserved 0361-9230/94 $6.00 + .OO
0361-9230(94)EOO46-3
Pulse Control During Rapid Isometric Contractions of the Elbow Joint YOSHIHIKO
YAMAZAKI,*’
MASATAKA
SUZUKI?
AND
TADAAKI
MANO+
*Department of Health and Physical Education, Nagoya Institute of Technology, Nagoya 466, Japan fDepartment of General Education, Kinjo Gakuin University, Nagoya 463, Japan $Department of Higher Nervous Control, Research Institute of Environmental Medicine, Nagoya University, Nagoya 464-01, Japan Received
12 April 1993; Accepted
15 February
1994
YAMAZAKI, Y., M. SUZUKI AND T. MANO. Pulse control during rapid isometric contractions of the elbow joint. BRAIN RES BULL 34(6) 5 19-531, 1994.-Normal subjects exerted isometric torque at a rapid rate of contraction with different amplitude targets, and at different rates with a constant target. In the fastest contractions, the agonist triceps and antagonist biceps muscles showed an electromyographic volley characterized by a slow wave with an initial negative and subsequent positive phases. The duration of each phase was relatively constant for different target amplitudes. In the agonist, the area comprised by the negative
or positive waveform increased with increasing target amplitude and correlated with the peak of the first time derivative of torque. The relationship of the antagonist volley to the target amplitude varied among subjects both in the area and the interval of the agonist-antagonist volley. With a decreasing contraction rate, the agonist volley decreased in the area in correlation with the peak of the torque derivative, retaining a constant duration, while the antagonist volley disappeared. These results suggest that rapid isometric contractions are controlled by an amplitude-modulated pulse of agonist activation, which is associated with a pulse of antagonist activation, while the pulse of the antagonist activation is variably utilized among subjects. Isometric
contraction
Monopolar
surface EMG
Elbow joint
IN our daily lives, movements are frequently directed at a target. At the turn of the century, Woodworth (3 1) suggested that targetdirected movements are produced by an initial impulse of force that mainly drives the limb to the target, and is followed by corrective impulses. In this particular line of research, Freund and Budingen (10) suggested that in rapid isometric contractions a motor system operates by adjusting the speed of contraction while maintaining a constant contraction time for different target amplitudes. They considered that the amplitude of the agonist activation ruled the contraction speed because the duration of the initial agonist activation was approximately constant. However, discrepancies in the agonist duration constancy observed in later studies (6,11) weakened the theory of speed control operation being solely dependent on the amplitude of the agonist activation. Ghez et al. (11,12,14,15) proposed a pulse height control based on a similar observation in the force trajectory of rapid isometric contractions. They suggest that the rate of force rise is modulated, whereas the force rise time remains relatively constant. They demonstrated that pulse height control was achieved by interplay between the agonist and antagonist muscles. Pulse height control, therefore, does not require duration constancy and amplitude modulation of agonist activation. In fact, agonist activation duration increases with the rate of force rise (1 l), and agonist activation quantity is only moderately correlated with the rate of
Human
force rise (14). However, the high predictability of peak force with an early trajectory measure, i.e., the peak value of the second time derivative of force (14,15), suggests that an initial part of the agonist activation contains the amplitude-modulated pulse of excitation originally addressed by Freund and Budingen (10). Recently, we found that in rapid isometric extensions of the elbow, EMGs of the agonist muscle, recorded by a monopolar electrode, initiated from an EMG volley showing large amplitude and constant duration with an initial negative potential followed by a positive one (33). This result prompted us to reinvestigate the question of whether the agonist muscle was activated as an amplitude-modulated pulse of excitation. This study investigated the EMG volley in isometric contractions of the elbow under different task conditions, and determined the time course of agonist and antagonist activation. Our results suggest that rapid isometric contractions of the elbow are controlled by an amplitude-modulated pulse of agonist activation in association with a pulse of antagonist activation. This pulse was used in different ways in different subjects. METHOD
Subjects, Apparatus, and
Tasks
Five subjects with no known history of neurological disorders (ages: 18-43) participated in this study after giving informed
’ To whom requests for reprints should be addressed.
519
520
consent. Two of the 5 subjects were repeatedly subjected to pilot studies and had participated in previous similar experiments. Subjects were seated comfortably with the right arm (dominant arm for all subjects) abducted 90”. The forearm and hand were placed in a firm moulded splint with the forearm in a neutral position at a right angle to the upper arm. The splint was fixed to an aluminum plate that was connected to a vertical axis supported by ball bearings. The elbow joint was positioned exactly above the vertical axis. The aluminum plate was immobilized at the distal end by a shaft on which strain gauges were mounted for measuring joint torque. The subject’s body was loosely strapped to the chair, and the left side of the body was against a wall. After setting up the experiment, subjects produced an isometric torque extending the elbow with a maximum voluntary contraction (MVC) that was used to determine target torque levels for each subject. Subjects viewed a computer monitor displaying three vertical lines: one stationary line showed the torque level of the relaxed state, another stationary line defined the target, and a movable line showed the subject’s joint torque. These were always displayed on the computer monitor, and subjects initially learned the elbow extension torque and the resultant motion of the movable line. Subjects performed isometric c~~~act~ons to extend the elbow from a relaxed state. A beep produced by the computer signaled subjects to initiate contractions. Subjects were told to respond when ready and not to attempt to produce short reaction times. The path of the movable line and how close it came to the target were always visible to subjects, who were, thus, given knowledge of results. In the first series of experiments, subjects made pulse contractions as fast and accurately as possible from a relaxed state to visually defined targets at 9, 18, 36, and 54% MVC torques. Target steps were similar to those used in recently published papers (6,33). Return from target was passive. Each target block consisted of 15 trials, and targets within each block were randomized. In the second series of experiments, subjects performed contractions from relaxation to a given target (27%, 36%, or 54% MVC) with different rates of torque rise. The target was the same for each subject: 27% MVC for one subject, 36% MVC for three subjects, and 54% MVC for one subject. First, subjects made a block of contractions under the instruction to perform as fast and accurately as possible. Secondly, subjects were asked to perform a contraction against the same target in the first block, but at a slightly reduced rate. During practice, a person monitoring the rate of contraction instructed subjects to increase the rate if the contraction was too slow in comparison to that of the first block. If the rate was the same as that of the first block, the subject was instructed to decrease it. Lastly, subjects were instructed to contract at a slower rate than in the second block against the same target as in the previous two blocks. They had to reach the target within I s, and to consistently reproduce contractions having a torque curve with a smooth rise and a single peak. Each block of contractions consisted of 20 trials. In both series of experiments, subjects practiced 10 times to become familiar with a new target or contraction rate before the first block and before each new block. Several minutes of rest were allowed between blocks.
EMG activity was recorded from the agonist triceps lateralis and longus muscles, and from the antagonist biceps brachii mus-
YAMAZAKI,
SUZUKI AND MAN0
I
lrntt
1oorns+ FIG. I. Tarque, first derivative of torque, and EMCs of rapid pulse contractions targeted at 9%, 18%, 36%, and 54% of MVC torque. All recordings were averaged (n = 10) by EMG onset of triceps lateralis (vertical line). Top and following traces are joint torque and first derivative of the torque (dTorque/dt), respectively. EMGs of the biceps brachii (BB) and triceps lateralis (TL) averaged after fullwave rectification or without rectification are superimposed. EMGs without rectification are shown as dotted area. Numbers within torque recordings correspond to numbered EMGs shown below.
cle with Ag-AgCl surface electrodes (0.8 cm in diameter). The surface EMG activity was recorded with a monopolar electrode arrangement: a detecting electrode was positioned over the muscle motor point region with a reference electrode over the tendinous part around the olecranon for the triceps lateralis and longus, and the cubital fossa for the biceps brachii. This electrode arrangement was the most appropriate for recording the EMG volley (33). The motor point region was estimated by surface electrical stimulation. The EMG signals were preamplified near the electrodes (lead length: lo-20 cm) and arnp~i~~ again with overall band ranges between 7-500 Hz. Joint torque and EMG signals were digitized at 1000 samples/s with a I2-bit A/D converter, and stored on a hard disk.
CONTROL OF RAPID ISOMETRIC CONTRACTIONS
521
1200 Nmfs
800
0
30
lVn 40
lo Peakibrque FIG. 2. Relationship between peak torque and peak dTorque/dt in rapid pulse contractions for different target torque. Peak dTorque/dt increased with increasing target torques for ail subjects. Each symbol denotes plots from a singfe subject. Each letter inserted close to the symbol shows subject number. Both symbol and subject number remain the same for each subject throughout the following figures. All plots are means + SD.
Data Analysis Data were reproduced on a computer monitor. The onset of the agonist triceps lateralis EMG was visually detected on the monitor by indicating the onset using a mouse with resolution of 1 ms. EMG onset was defined by the first clear separation from baseline to negativity because, in our monopolar EMG recordings, the agonist EMG always initiated from a clear negative potential. Each onset was written to a file header and served to align and average EMG and torque recordings. In the first series of experiments in which subjects made contractions as fast and accurately as possible to four target amplitudes, 5 of 15 trials were rejected from each block. Rejected trials were individually selected as being the maximal overshooting or undershooting of a given target. This rejection was applied to increase disparity of mechanical variables between different target blocks. Such disparity was required because different target blocks were compared on the basis of averaged data. In the second series of experiments, 10 of 20 trials were rejected. The mean of peak torque was calculated for all contractions at three different rates. Ten rejected trials were selected in each block from those showing the greater deviation from the mean of peak torque (4-6 trials), and then from those showing shorter or longer extremes in torque rise time (4-6 trials). The first rejection process was employed to reduce variability in peak torque across blocks, which allowed us to compare EMGs of different contraction rates under the same condition in peak torque. The second rejection process served to decrease variability in torque rise time. Mechanical recordings and EMGs were digitally lowpass-filtered at 50 Hz and bandpass-filtered at 7-500 Hz (liner-phase FIR filter, 5th order). The first time, derivative of torque (dTorqueldt) was calculated in each trial by digital diffe~ntiation, and then the peak value of the derivative was obtained by digitally searching for maxima. The second time, derivative of torque (d’Torque/dt’) was successively calculated from the dTorque/dt,
and then a peak value was also obtained. The d’Torque/dt’ was not calculated for different rates of contractions in the second series of experiments because of low signal-to-noise ratios. The EMGs of the agonist triceps and antagonist biceps muscles were averaged without rectification and aligned at the onset of the triceps lateralis EMG. The averaged EMGs of the triceps and biceps muscles showed the EMG volley that was clearly characterized by an initial negative and then a positive wave. The duration of the EMG volley was separately measured in the negative and positive phases of the averaged EMGs. The negative phase duration was defined by the time interval between the volley onset and the point where the negative potential first crossed an EMG baseline. The positive phase duration was from the end of the negative phase to the point where the positive potential returned to an arbitrary base line, which was determined by superimposing the averaged EMGs of four different amplitude recordings. The EMG volley was quantified by integrating the averaged EMGs separately for the negative and positive phases. The EMG volley quantity was also calculated in each trial using the duration obtained from the averaged recordings. Integrated EMGs with a negative sign were inverted. EMGs were also averaged after ratification, but no measurements were done for the rectified and averaged EMGs. Correlation coefficients between variables were calculated and shown by a range and median for all subjects.
RESULTS
Different Amplitudes Torque. Figure 1 shows recordings of exerted torque, the first derivative of torque, and EMGs of the antagonist biceps and agonist triceps lateralis muscles. The subject performed isometric pulse contractions whose peak values targeted at 9, 18, 36, and
YAMAZAKI,
522
A
SUZUKI
AND MAN0
120
ms
-
sl
80
Target torque
B
%MVC
120 ,
0
9
18
45
Tar-i&t tor3q6Ue
54
%MVC
FIG. 3. (A,B) Durations of negative and positive phases of agonist EMG volley for different target torques. In nonrectified and averaged EMGs, durations of the first negative and subsequent positive phases of agonist EMG volley were separately measured. AI1 durations except for positive phase of s5 were relatively constant at different target torques.
54% of MVC torque, being instructed to perform as fast and accurately as possible. All were averaged recordings, aligned at the onset of the triceps lateralis EMG. In Fig. 1, torque recordings show relatively stable trajectories over different target amplitudes; the time to each peak of dTorque/dt from torque onset was approximately constant, although time to reach peak torque increased slightly with torque amplitudes. A systematic change was seen between the peak torque and the peak dTorque/dt. Figure 2 shows the relationship between peak torque and peak dTorque/ dt for different blocks of target ~plitudes. Peak dTorqu~dt increased with increasing peak torque. This linear relationship was also seen in the analysis of each trial. Peak dTorque/dt highly
correlated with peak torque for all subjects (I = 0.979-0.996; median: 0.992). Peak d’Torque/dt’ also correlated with the peak torque (I = 0.942-0.986; median: 0.974). In each subject, both the peak torque and the peak dTorqueldt showed distinct separation among different blocks of target amplitudes, which allowed comparison of each block with different torque levels. Two subjects who had prior experience showed higher performance than other inexperienced subjects (sl and s2 in Fig. 2). Performance was assessed by the magnitude of peak dTorque/dt plotted against peak torque. Al~ough subjects were different in both experience and performance, results on EMGs, which will be stated below, were approximately the same. This shows that
CONTROL OF RAPID ISOMETRIC CONTRACTIONS
A
s W
523
80 , 1
mV*ms
I
od’d’lil”“““’
45
54
45
54
Tar&t to&e
%MJC
mV’ms
W
0
9
18
27
36
Target torque
?LMvc
FIG. 4. (A,B) EMG quantities of negative and positive phases of agonist EMG volley for dif-
ferent target torques. EMG quantities were obtained by separately integrating negative and positive phases of the EMG volley over each duration. EMG quantities of both phases increased with target torque for ah subjects.
subject differences in experience and performance do not affect results. EMGs of the agonist and antagonist. Because the triceps longus EMG was almost the same as the triceps lateralis EMG, we will address only the latter. In Fig. 1, two superimposed EMG traces show averaged recordings with and without ~ctifi~ation. In rectified EMGs, the agonist triceps lateralis was activated, followed by antagonist biceps activation. In nonrectified EMGs, the EMG volley, both in the agonist and antagonist, was composed of negative and subsequent positive phases. The EMG volleys of the agonist and antagonist had larger amplitudes in the rectified EMGs. In our previous reports, the same EMG volley was found
in both isometric and anisometric contractions of the elbow performed as fast and accurately as possible to a target (32,33). Hereafter, we term the agonist volley AGv, and the antagonist volley ANTv.
Agonist EMG volley. Figure 3 shows the durations of AGv in all subjects. Duration was measured separately in the negative and positive phases of the nonrectified and averaged EMGs. The negative duration did not vary with target amplitudes and remained approximately constant across subjects (Fig. 3A). The positive phase duration did not vary with target amplitudes except in one subject, but was more variable across subjects than the negative phase duration (Fig. 3B).
YAMAZAKI,
524
A
SUZUKI
AND MAN0
120
ms
-s> 8%
80
=a
0
9
18
45
Tar&t to&
54 %MV(
. 3
FIG. 5. (A,B) Durations of negative and positive phases of antagonist EMG volley at different target torques. Negative durations are relatively constant except for plots of 36% MVC target torque. Positive duration tends to increase with target torque increase.
Figure 4 shows quantities of the AGv in which negative (Fig. 4A) and positive (Fig. 4B) phases in the nonrectified and averaged EMGs were separately integrated. Quantities of the AGv in both phases increased with increasing target amplitudes (see also Fig. 1). These increases in quantity were not derived from elongation of the volley duration, because both phases showed a constant duration over target amplitudes except for the positive duration in one subject (Fig. 3). The AGv quantities were also calculated for individual subjects by integrating each trial over the duration. The negative phase quantity correlated with the peak torque (r = 0.910-0.953; median: 0.935), the peak dTorqueldt (0.923-0.970; median: 0.962), and the peak d’Torque/ dt2 (0.927-0.963; median: 0.956). The positive phase quantity
also correlated with the peak torque (r = 0.899-0.977; median: 0.953) the peak dTorque/dt (I = 0.899-0.976; median: 0.952), and the peak d’Torque/dt’ (1. = 0.885-0.976; median: 0.926). Furthermore, the negative phase quantity correlated with the positive phase quantity (r = 0.889-0.988; median: 0.927). Antagonist EMG volley. Figure 5 shows durations of the ANTv for the negative and positive phases. These durations were measured from nonrectified and averaged EMGs. In each subject, the negative phase duration of the ANTv was approximately constant across target amplitudes, but the relationship was weak compared to that in the AGv (Fig. 5A). The positive phase duration of the ANTv increased slightly with target amplitudes (Fig. 5B). The positive phase duration differed considerably among
CONTROL
RAPID ISOMETRIC CONTRACTIONS
of 0
I
525
f.
,
9
18
.
,
27
.
,
.
toy&e
,
.
,
45
54
45
54
Target
LLl
0
9
18
27
36
Target torque
.I
%MvC
%MVC
FIG. 6. (A,B) EMG quantities of negative and positive phases of antagonist EMG volley for different target torques. EMG quantities of both phases increased with target amplitude in three subjects, while remaining unchanged in other two.
subjects. Figure 6 shows ANTv quantities separately integrated for the negative and positive phases. The ANTv quantities showed different changes to target amplitudes among subjects: in two subjects, the negative and positive phase quantities were relatively stable, although in three subjects, they increased with target amplitudes. Onset interval between the agonist and antagonist Eh4G volleys. Figure 7 shows the onset interval between the AGv
and ANTv measured in the nonrectified and averaged ings. The onset interval varied with target amplitudes subjects. Analysis including rejected trials. In this series of ments, 5 of 15 trials in each block were rejected. We
recordacross experistudied
whether the rejection biased the results or not. When peak torques of rejected trials along with accepted ones were plotted for all target amplitudes against each of the trajectory variables (peak dTorque/dt, peak d’Torqu~d~, time to peak torque, and time to peak dTorque/dt), plots of both the rejected and accepted trials showed an identical relationship in every graph. Moreover, when all data, including rejected trials, were analyzed for durations and quantities of AGv and ANTv, and intervals between AGv and ANTv, the results were mainly the same as those based on selected data. Thus, the rejection did not bias the results. However, without the rejection process, the discrepancy in trajectory variables between different target blocks decreased, especially between 9% and 18% MVC targets.
YAMAZAKI.
526
SUZUKI
AND MAN0
120
ms
-
-
sl
L
OL 0
9
45
18
T&t
54
t&e
FIG. 7. Onset time interval between agonist and antagonist EMG volleys for different torques. Subjects show different changes in the interval for target amplitudes.
Three Contraction Rates Torque. Figure 8 shows features of the contractions, times to reach peak torque, and peak dTorque/dt, when subjects contracted at three different rates of torque rise. The target amplitude was constant for each subject. The contraction features differed among subjects, because changes in the contraction rate were dependent on the choice by each subject. Torque variables shown in Fig. 8 differed among blocks at each contraction rate, which allowed comparison between blocks as different sets of samples of contraction rates. EMGs of the agonist and antagonist. Figure 9 shows recordings of torque, dTorque/dt, and EMGs when the rate of torque rise changed to maintain a constant target amplitude: from upper to lower, exerted torque, dTorque/dt, and EMGs of the antagonist biceps and agonist triceps lateralis muscles. All are averaged recordings, aligned at the onset of the triceps lateralis EMG. The EMGs averaged with and without rectification are superimposed. In the fastest contraction (Fig. 9a), torque recordings were characterized by a rapid rise with a high peak of dTorque/dt, and the nonrectified EMGs clearly showed the AGv and ANTv. When the contraction rate was reduced (Fig. 9b,c), the rectified EMGs in the initial agonist activation decreased in amplitude and increased in duration. The AGv magnitude also decreased with the decreasing contraction rate, but the AGv was still seen at the lowest rate, and the negative phase of the AGv duration remained relatively constant. The rectified EMGs of the antagonist greatly decreased in the two slower contractions. The ANTv was not observed at slower contraction rates. Figure 10 shows both the negative phase duration and quantity of the AGv, which were plotted against peak dTorque/dt. The negative duration was relatively constant to changes in peak dTorque/dt, but the quantity increased with increasing peak dTorque/dt for all subjects. The quantity was also calculated in each trial by integrating the EMGs over the duration. The AGv quantity correlated with peak dTorque/dt (I = 0.923-0.970; median: 0.962). The quantity did not correlate with torque amplitude because the target amplitude was kept constant. Because it be-
%MVC target
came difficult to identify the end of the positive phase when the rate of torque rise decreased, the positive phase duration was not measured, which prevented calculation of the positive phase quantity. There was no difference in results between the two experienced and the other inexperienced subjects. This indicates that our results hold true, irrespective of subject experience and performance. Analysis including rejected trials. In this second series of experiments, 10 of 20 trials from each block were rejected. To verify whether or not the high proportion of trial rejection biased our results, the results obtained from selected trials were compared with those analyzed without the selection process. The results were essentially the same with and without selection; the negative phase duration of the AGv was relatively constant over different contraction rates, and the quantity of the AGv increased with increasing contraction rate. Therefore, trial selection did not bias our results. DISCUSSION
Agonist EMG When subjects made targeted isometric contractions as rapidly and accurately as possible, the AGv showed a constant duration over different target amplitudes, and the AGv quantity correlated with the peak and time derivatives of torque. When the rate of torque rise was changed while the target amplitude remained constant, the AGv varied with the rate of torque rise, preserving the duration constancy. Because EMGs, recorded by surface electrodes, indicate gross activation of the motoneuron pool innervating a muscle, the abrupt increase of activation with a constant duration in the AGv indicates that the gross activation of the agonist motoneuron pool has the aspect of a pulse with a constant duration. The systematic change in the AGv quantity with torque variables indicates that pulse activation undergoes amplitude modulation. Therefore, our results suggest, on agonist organization, that rapid isometric contractions are controlled by
CONTROL
OF RAPID
ISOMETRIC
527
CONTRACTIONS
Time to peak torque ***.*..*.**.**** Time to peak dTorque/dl
400 ms
Subject
B
800 Nmls
r
rapid
+-e
200
sl
0 15
20
Peak torque FIG. 8. (A,B) Torque variables in pulse contractions performed at three contraction rates for same target amplitude. Both in A and B, target torques differed for each subject, but were constant for a single subject. A shows times to peak torque and peak dTorque/dt. B shows peak dTorque/dt. All plots are means k SD.
a constant duration pulse of activation, and that the contraction rate is modulated by the amplitude of the pulse. One might think that the constant duration pulse of agonist activation with amplitude modulation was established in rapid isometric contractions. Ghez et al. (11,12,14,15) proposed a pulse height control theory, which states that the rate of force rise is modulated with a relatively constant force rise time. Pulse height control, however, does not conditionally require the agonist to be subjected to the constant duration pulse of activation, because the control, mainly based on force recordings, depends on pushpull forces of mutually antagonistic muscles. In their experiments, the duration of the agonist activation lengthened with increases in the ratio of force rise (1 I), and the EMG quantity of
the agonist activation showed only a moderate correlation with the rate of force rise (14). Fteund and Budingen (10) reported that both the time to peak force and the agonist burst duration shared a constant duration over different force amplitudes in the fastest isometric contractions. This implies that the agonist muscle is activated as the constant duration pulse of excitation. However, later studies failed to confirm EMG duration constancy in the agonist activation (6,ll). Since then, it has not been determined whether agonist activation has a facet of the constant duration pulse. In the fastest contractions, the AGv quantity of the negative phase highly correlated with trajectory variables: peak torque (r = 0.910-0.953; median: 0.935), peak dTorque/dt (0.923-
YAMAZAKI,
528
a
b
C
0 250 Nm/s
-
a b
SUZUKI
AND MAN0
function of plural antagonistic muscle forces, and EMGs precede the d2Torque/dt2. A motor program may change under various conditions such as instructions, the knowledge of results, fatigue, and progress in the learning process. In future studies, the negative phase AGv could serve as a reliable measure of motor program variability. Because the averaging procedure used in the present experiments removes variability in the motor program, such analysis should be based on single trials. In Fig. 1, torque curves smoothly increased and then decreased up to the 36% MVC target, whereas in the 54% MVC target the curve increased close to the peak but lingered around the peak or somewhat increased. This irregularity in the torque response, which was observed in three of five subjects, was not seen in other studies. The irregularity may derive from any difference in scaling of rapid torque responses between lower and higher torque targets. It is likely that a motor control system has a limited capacity to activate a muscle within a short period, which may cause changes in torque scaling at higher targets. Further studies are required to determine the changes in torque scaling, because there is no information on scaling and muscle activation during rapid contractions close to MVC targets. The irregularity might reflect any corrective response that subjects made around peak torque. Because our subjects showed a single peak in torque responses for variable target amplitudes, they did not make the conspicuous corrections usually associated with multiple peaks. However, it cannot be ruled out that subjects corrected a response once initiated because of the absence of our instructions to systematically avoid corrective responses, thus resulting in the irregularity of the torque curve.
[Antagonist
BB
.. . . Cl
..
TL
1mV + 1OOms
FIG. 9. Torque, first derivative of torque, and EMGs at three contraction rates for 54% MVC target. All are averaged recordings (n = 10) aligned by EMG onset in triceps lateralis (vertical line). From top down, torque, first derivative of torque (dTorque/dt), EMGs of the biceps brachii (BB), and triceps lateralis (TL) are shown. EMGs were averaged after fullwave rectification or without rectification, and differently processed EMGs are superimposed. Dotted area shows EMGs averaged without
rectification. Torque recordings indicated by characters correspond to EMGs shown by the same character.
0.970; median: 0.962), and peak d*Torque/dt* (0.927-0.963; median: 0.956). Considering that the negative phase of the AGv lasted approximately 50 ms from the EMG onset, it is surprising that such an early phase of the agonist EMG can predict torque trajectories. This predictability indicates that the contractions have a feature of ballistic movement (27), where the consequence of a complete movement are determined at an early stage of that movement (23). The ballistic movement is considered to be largely governed by a motor program (9,14,20,23). Although d2Torque/dt2 has been utilized to measure the motor program ( 14,15), the negative phase AGv can be a more direct and earlier measure of the motor program, because joint torque is a complex
EMG
In the fastest contractions, there was marked antagonist activation. The initial part of the antagonist activation also showed an EMG volley (Fig. 9) which suggests that the antagonist activation is also organized under a pulse control. Thus, to control the fastest isometric contractions, the agonist activation with pulse amplitude modulation cooperates with the pulse activation of the antagonist. However, there were different manifestations of antagonist cooperation across subjects. In comparison with the systematic relationship between the AGv quantity and target torques, the ANTv quantity varied across subjects at different target torques (Fig. 6). The interval of the AGv-ANTv to target torques also varied across subjects (Fig. 7). The variability of the ANTv is partly due to the method of averaging: the antagonist EMGs were averaged by aligning the agonist EMG onset. The ANTv variability was also caused by the complex role of the antagonist muscle that controls the rising phase of torque (11,30). In some subjects, this antagonist role was altered at greater torque targets by delaying the ANTv onset (Fig. 7), which probably increased exerting torque and the rate of torque rise by eliminating the limiting torque of the antagonist. Furthermore, the antagonist frequently shows a weak tonic activation before the burst activation (6,l l), the role of which remains unclear. In anisometric contractions, the same antagonist contraction occurs; the contraction may contribute to increased joint stiffness to improve termination accuracy (13), stability (21) effort reduction (22) and effectiveness of the following pulse activation (32), and to counteract centrifugal force (22). In contrast to the distinct role of the agonist in generating driving torque, the antagonist has a complex role, allowing subjects to utilize the antagonist in different ways for different tasks. Furthermore, variability in the ANTv may reflect a subject’s movement strategy for the task. Antagonist activity is sensitive to the way the subject tries to
CONTROL
OF RAPID
ISOMETRIC
529
CONTRACTIONS
120
A
ms
Sl
0
200
400
600
Nm
800
Nmls
800
Peak dTorque/dt
B
80 mV’ms
._ 3
60-
$> cc3 +a
200
400
600
Peak dTorque/dt FIG. 10. (A,B) Duration and EMG quantity of negative phase of agonist EMG volley at three contraction rates. Duration and EMG quantity are plotted against peak dTorque/dt. In A, negative duration of agonist EMG volley decreased slightly with increasing contraction rates. In B, EMG quantity of the negative phase of agonist EMG volley increased with increasing contraction rates. perform the task in isometric contractions of the elbow (6), and in anisometric contractions of the elbow (3), the wrist (29), and the finger (25). In simulated elbow movements, nonlinearity of antagonist activation is related to some movement parameters: peak displacement and time to reach peak displacement (30). In addition, we cannot rule out the possibility that the variability of the ANTv is due to some corrective response that might have caused the irregularity of the torque curve in the 54% MVC target. EMG Volley The EMG volley was extremely long during both the negative and positive phases compared to that of ordinary EMGs. Many
researchers suspect that the EMG volley is a movement artifact. However, in a previous paper (33), we demonstrated that the EMG volley observed in rapid isometric contractions of the elbow was not a movement artifact because it was seen even when monopolar electrode and lead movements were eliminated. The EMG volley agreed with motor unit firing condensed at the onset of the rapid contractions. Our view was further supported by other indirect observations: the EMG volley preceded movement of the monopolar electrode and did not correlate in amplitude with electrode movements, and the long negative potential that characterized the EMG volley was synthesized in the triceps surae muscle by repetitive electrical stimulation to the tibial nerve. Rapid contractions require the subject to recruit many motor units with a high firing rate within a limited time. Motor units
530
YAMAZAKI,
are recruited at a lower force threshold in faster contractions than in slower ones (8,28). Firing frequency of motor units at the onset of the fastest contractions is initially high, and then rapidly decreases in subsequent firings (8,17). These recruitment and firing characteristics result in dense firing of many motor units within a relatively limited time. Because each motor unit potential constitutes a depolarization zone that is primarily negative (18,19), dense firing at the onset of the fastest contractions would build up a collective depolarization zone. This registers as a negative potential when recorded by a monopolar electrode placed exactly on the depolarization zone, which occurs around a motor point and propagates to both ends of the muscle fiber. With propagation to the muscle fiber, a positive potential due to volume conduction is recorded on the monopolar electrode, and the positive potential appears when the depolarization zone reaches the reference electrode, as in the case of a single motor unit (18,19). Hence, we speculate that the negative phase of the EMG volley is a manifestation of dense motor unit firing in rapid contractions. In previous experiments (32,33), where EMGs of each trial were analyzed without averaging in both isometric and anisometric contractions, the same EMG volley was recorded during rapid contractions; the negative phase of the EMG volley consistently demonstrated a constant duration, whereas the duration of the positive phase varied and was frequently difficult to identify. In the present experiments based on averaged recordings, the positive phase was consistently observed and correlated with the negative phase in quantity, but the positive phase duration had greater variability than that of the negative phase (Figs. 3 and 5). The variability of the positive phase was due to several factors such as differences in the distribution of motor end plates, conduction velocity among muscle fibers, placement of the reference electrode, length of each muscle fiber, and intervention of
SUZUKI
AND MAN0
motor units recruited during the positive phase. Considering the variability and instability of the positive phase, the negative phase is a more appropriate parameter of the EMG volley than the positive phase. Durations of the negative and positive phases in the EMG volley were measured from nonrectified and averaged EMGs. Such a duration measurement may include some errors derived from artifact. Although a movement artifact is particularly annoying, it is unavoidable in surface EMG recordings. Although, in previous studies, we demonstrated that the EMG volley is not a movement artifact, a small error in the duration measurement due to such an artifact could not be prevented, because the artifact usually assumes a greater significance with averaging. Estimating muscular force from surface EMGs has been attempted (1). Studies concerning the EMGs and force (2,5,7,16,24,26) have shown that although they are fairly parallel in a steady contraction (24) their parallel relationship deteriorates during dynamic changes in force ($16). One cause of estimation error is that the surface EMGs, usually recorded by bipolar electrodes, do not reliably measure motor unit activation during dynamic contractions: the biphasic action potential of each motor unit was thought to cancel the motor unit potential when recorded by surface electrodes during the fastest and strong contractions (4). Because, as already noted, the EMG volley recorded by monopolar electrode is likely to reflect the dense firing of many motor units, using the EMG volley could improve the estimation of muscle force from surface EMGs. ACKNOWLEDGEMENT This work was supported by Grant-in-Aid for Scientific Research No. 04680123 from the Ministry of Education, Science and Culture, Japan.
REFERENCES 1. Agarwal, G. C.; Gottlieb, G. L. Mathematical modeling and simulation of the postural control loop: Part I. CRC Crit. Rev. BME 8:93134; 1982. 2. Bigland, B.; Lippold, 0. C. J. The relation between force, velocity and integrated electrical activity in human muscles. J. Physiol. (Land.) 123:214-224; 1954. 3. Brown, S. H.; Cooke, J. D. Amplitude- and instruction-dependent modulation of movement-related electromyogram activity in humans. J. Physiol. (Lond.) 316:97-107; 1981. 4. Calve& T. W.; Chapman, A. E. The relationship between the surface EMG and force transients in muscle: Simulation and experimental studies. Proc. IEEE 65:682-689: 1977. 5. Coggshall, J. C.; Bekey, G. A. EMG-force dynamics in human skeletal muscle. Med. Biol. Eng. 8:265-270; 1970. 6. Corcos, D. M.; Agarwal, G. C.; Flaherty, B. P.; Gottlieb, G. L. Organizing principles for single-joint movements. IV. Implications for isometric contractions. J. Neurophysiol. 64: 1033- 1042; 1990. 7. Crosby, P. A. Use of surface electromyogram as a measure of dynamic force in human limb muscles. Med. Biol. Enn.1 Comout. 1 16519-524; 1978. 8. Desmedt, J. E.; Godaux, E. Ballistic contractions in man: Characteristic recruitment pattern of single motor units of the tibialis anterior muscle. J. Physiol. (Lond.) 264:673-693; 1977. 9. Desmedt, J. E.; Godaux, E. Ballistic skilled movements: Load compensation and patterning the motor commands. In: Desmedt, J. E., ed. Progress in clinical neurophysiology. Cerebral motor control in man: Long loop mechanisms. Basel: Karger; 1978:21-55. 10. Freund, H.; Bundingen, H. J. The relationship between speed and amplitude of the fastest voluntary contractions of human arm muscles. Exp. Brain Res. 31:1-12; 1978. 11. Ghez, C.; Gordon, J. Trajectory control in targeted force impulses. I. Role of opposing muscles. Exp. Brain Res. 67:225-240; 1987.
12. Ghez, C.; Vicario, D. The control of rapid limb movement in the cat. II. Scaling of rapid force adjustments. Exp. Brain Res. 33: 191203; 1978. 13. Ghez, C.; Vicario, D.; Martin, J. H.; Yumiya, H. Sensory motor processing of targeted movements in motor cortex. In: Desmedt, J. E., ed. Motor control mechanisms in health and disease. New York: Raven Press; 1983:61-92. 14. Gordon, J.; Ghez, C. Trajectory control in targeted force impulses. II. Pulse height control. Exp. Brain Res. 67:241-252; 1987. 15. Gordon, J.; Ghez, C. Trajectory control in targeted force impulses. III. Compensatory adjustments for initial errors. Exp. Brain Res. 67:253-269; 1987. 16. Gottlieb, G. L.; Agarwal, G. C. Dynamic relationship between isometric muscle tension and the electromyogram in man. J. Appl. Physiol. 30:345-351; 1971. 17. Grimby, L.; Hannerz, J. Firing rate and recruitment order of toe extensor motor units in different modes of voluntary contraction. J. Physiol. (Lond.) 264:865-879; 1977. 18. Gydikov, A. The potential of the motor unit. In: Shahani, M., ed. The motor system: Neurophysiology and muscle mechanisms. Amsterdam: Elsevier Scientific Publishing Co.; 1974:49 -71. 19. Gydikov, A.; Kosarov, D. Extraterritorial potential field of impulses from separate motor units in human muscles. Electromyography 12:283-305; 1972. 20. Hallett, M.; Shahani, B. T.; Young, R. R. EMG analysis of stereotyped voluntary movements in man. I. Neurol. Neurosurg. Psychiatry 38:1154-1162; 1975. 21. Hogan, N. Adaptive control of mechanical impedance by coactivation of antagonist muscles. IEEE Trans. Auto. Control AC-29:681690; 1984.
CONTROL OF RAPID ISOMETRIC CONTRACTIONS
22. Karst, G. M.; Hasan, 2. Antagonist muscle activity during human forearm movements under varying kinematic and loading conditions. Exp. Brain Res. 67:391-401; 1987. 23. Kornbuber, H. H. Motor functions of cerebellum and basal ganglia: The cere~lloco~ic~ saccadic (b~listi~) clock, the ~~~llonucle~ hold regulator, and the basal ganglia ramp (voluntary speed smooth movement) generator. Kybemetic 8:157- 162; 1971. 24. Lippold, 0. C. J. The relation between integrated action potentials in a human muscle and its isometric tension. J. Physiol. (Lond.) 117:492-499,1952. 25. Meinck, H.; Benecke, R.; Meyer, W.; Hohne, J.; Conrad, B. Human ballistic finger flexion: Uncoupling of the three-burst pattern. Exp. Brain Res. 55127-133; 1984. 26. Soechting, J. F.; Roberts, W. J. Transfer characteristics between EMC activity and muscle tension under isometric conditions in man. J. Physiol. (Paris) 70~779-793; 1975.
531
27. Stetson, R. H. A motor theory of rhythm and discrete succession. Psychol. Rev. 12:250-271; 1905. 28. Tanji, J.; Kato, M. Recruitment of motor units in voluntary contraction of a finger muscle in man. Exp. Neural. 40:759-770; 1973. 29. Waters, P.; St&k, P. L. Influence of ‘strategy’ on muscle activity during ballistic movements. Brain Res. 207:189-194; 1981. 30. Wierzbicka, M. M.; Wieger, A. W.; Shahani, B. T. Role of agonist and antagonist muscles in fast arm movements in man. Exp. Brain Res. 63:331-340; 1986. 31. Woodworth, R. S. The accuracy of voluntary movement. Psychol. Rev. 3(2, Suppl 13):1-114; 1899. 32. Yamazaki, Y.; Suzuki, M.; Mano, T. Control of rapid elbow extension movement. Brain Res. Bull. 30: 1 l- 19; 1993. 33. Yamazaki, Y.; Suzuki, M.; Mano, T. An electromyographic volley at initia~on of rapid isometric con~actions of the elbow. Brain Res. Bull. 30:181-187; 1993.
