__ __ l!B
AJ2A EISEVIER
HUMAN Human Movement Science 14 (1995) 351-369
!EiiNT
Velocity-dependent activation of postural muscles in a simple two-joint synergy Mark B. Shapiro a, Alexander S. Aruin b, Mark L. Latash ‘3* aDepartment of Electrical Engineering and Computer Science, University of Illinois, Chicago, IL 60680, USA b Department of Physical Medicine and Rehabilitation, Rush-Presbyterian St. Luke’s Medical Center, Chicago, IL 60612, USA ’ Department of Exercise and Sport Science, The Pennsylvania State University, 200 Biomechanics Laboratory, University Park, PA 16802, USA
Abstract
A simple two-joint synergy was studied over a range of movement velocities. We hypothesized that focal and postural components of the synergy are consequences of a single control process and, as such, will demonstrate similar scaling with movement velocity. Healthy subjects performed discrete elbow or wrist flexion or extension movements in a sag&al plane under the instruction to move one of the joints at different speed in different trials of a series. Joint angles and electromyographic (EMG) signals from two flexor and two extensor muscles were recorded and analyzed. Irrespective whether the focal movement took place in the elbow or in the wrist joint, and irrespective of the movement direction and velocity, the elbow flexor and the wrist flexor tended to demonstrate simultaneous EMG bursts, while the elbow extensor and the wrist extensor also showed similar patterns of activation. During flexion (extension) movements in either joint, the latencies of both elbow and wrist extensors (flexors) decreased with velocity of the focal movement. Integrals of the EMG bursts in all the muscles increased with movement speed in all the series. Typically, there was a close to linear relation between the integral EMG indices for the elbow and wrist flexors as well as for the elbow and wrist extensors. We conclude that there exists a simple, scalable synergy which is used by the central nervous system in a wide range of movement velocities to simplify control of the postural component of a motor task.
* Corresponding
author. E-mail:
[email protected],
Fax: + 1 814 865-2440, Tel.: + 1 814 86.5-3445.
0167-9457/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDI 0167-9457(95)00016-X
352
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
1. Introduction
Fast voluntary limb movement is always a source of postural perturbation because of the reaction forces and joint coupling. When a person performs a movement in a joint of a multi-joint limb, his/her central nervous system probably “wants” to preserve control over the endpoint of the limb. To do so, it needs to be able to predict the postural perturbations that would occur in other joints of the limb, and to try to compensate for them in advance (in a feed-forward manner) by changing the level of activation of muscles acting at the other joints. This mechanism has been termed “anticipatory postural adjustments” (Hugon et al., 1982; Burbaud et al., 1988; Aoki, 1991; Koshland et al., 1991; Massion, 1992). On the other hand, Bernstein (1967) introduced a notion of synergy as a combination of commands to a number of muscles acting at a number of joints united by a common functional goal. Synergies were assumed to be tunable and used as blocks for construction of complex, multi-joint movements. As such, synergies involve anticipatory postural adjustments at a number of joints. Anticipatory postural reactions have been known to depend on movement velocity. This dependence seems to be nonmonotonic and involve a threshold effect. In particular, slow movements may be performed in the absence of visible anticipatory changes in the activity of postural muscles (Lee et al., 1987; Horak and Nashner, 1986; Horak et al., 1990; Crenna et al., 1987; Oddsson, 1989). These studies were performed, however, in standing persons and involved the task of maintaining the vertical posture. It is not clear whether the same rules apply to anticipatory adjustments in the activity of muscles acting at non-focal joints of a multi-joint limb. Recently, we studied muscle activation patterns during single-joint movement tasks in a two-joint (wrist and elbow) system (Latash et al., 1995a). One of the joints performed a primary movement “as fast as possible” in a sagittal plane, while the other carried out an apparently postural component of the task. This task may be compared to moving an inverted pendulum with an additional joint on its axis. We observed a robust synergy consisting of closely coupled electromyographic (EMG) patterns in the elbow and wrist flexors as well as in the elbow and wrist extensors. This synergy has later been shown to be preserved in elderly subjects, in patients with Parkinson’s disease (Latash et al., 1995b), and in persons with Down syndrome (Almeida et al., 1994). These observations led us to a hypothesis that anticipatory postural reactions during movements of a multi-joint system are not an addition to a primary motor command but a separate
MB. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
353
peripheral pattern of a single control process. This hypothesis predicts, in particular, a tight coupling between EMG patterns acting at a focal and at a postural joint preserved over a range of parameters of the focal movement, such as movement velocity. In this study, we analyzed the dependence of adjustments in the activity of muscles acting at an apparently postural (non-focal) joint upon movement velocity in the focal joint. The following major question is in the focus of the study: Does the observed synergy scale as a whole with movement velocity? Two components of this major question were analyzed: (1) How is the postural component of the synergy modified during slower movements when expected postural perturbations are small? and (2) Will the flexorflexor and extensor-extensor coupling observed during fast movements be preserved over the whole range of movement velocities, or will the central nervous system switch to an alternative strategy at a certain movement speed?
2. Methods 2.1. Subjects
Five without subjects Human
neurologically healthy, right-handed male volunteers, aged 30-56, any history of motor disorders participated in the study. All the gave informed consent according to the procedure approved by the Investigation Committee of the Medical Center.
2.2. Apparatus
The subjects sat comfortably in a chair and placed their right upper arm on an adjustable horizontal surface so that the shoulder joint was flexed 90”. The elbow joint was 90” into flexion, so that the forearm was vertical. The forearm was supinated, so that the subject faced his palm. All the fingers were extended and oriented vertically. A vertical thin metal rod showed the subjects the required initial position; two more adjustable rods were used to show the required final positions. A Mac-IIci computer controlled the experiment, digitized, and recorded joint angles and EMGs. Two goniometers (Penny and Giles) were taped on the subject’s arm and measured wrist and elbow joint angle changes in a
354
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
sagittal plane. Velocity was derived from the angle signals by differentiation after filtering at 100 Hz with a second-order Butterworth filter. EMGs of biceps, lateral head of triceps, flexor carpi radialis and extensor carpi ulnaris longus were recorded. Pediatric electrocardiographic self-adhesive, disposable electrodes were taped over the muscle bellies and used for a bipolar EMG recording. The distance between the electrodes was 5 cm. The EMGs were amplified (1600X), band-pass filtered (60-500 Hz), and digitized with 12 bit resolution. All the signals were sampled at 500 Hz. More details can be found in a previous publication (Latash et al., 1995a). 2.3. Procedure The experimental procedure involved 4 series. Prior to each series, the subjects were given 5 practice trials. Within a series, the subjects were instructed to perform unidirectional movements at speeds varying from “as fast as possible” to “slow”, in a subject selected, random order, not to bother about accuracy, and not to correct the final position. During the first series, the subjects performed 40 elbow flexion movements over a nominal distance of 35”. During the second series, they performed 40 elbow extension movements over 35”. The third series involved 40 wrist flexion movements over 45”, and the fourth series involved 40 wrist extension movements over 45”. However, accuracy requirements were not stressed, and the subjects could move over a preferred distance as long as they did so reproducibly in all the trials of a series. The intervals between the trials within each series were 8 s. There were intervals of about 2 min between the series. During each series, the subjects were instructed to occupy an initial position (900 in the elbow joint and 180” in the wrist joint, i.e. both the forearm and the hand vertical), to wait for a tone signal (a beep), to make a smooth movement to a final position, to stop at the final position, to wait for a second beep, and to return back. According to our convention, the initial position corresponded to 0” in both joints; extension in a joint corresponded to positive values of joint angle and velocity. 2.4. Data processing
The joint whose movement was explicitly required by the instruction (the focal joint) will be addressed as M-joint, and the other, non-focal joint will be addressed as postural joint (P-joint>. During flexion movements in
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
355
M-joint, we will address the flexor of M-joint as M-agonist and the flexor of P-joint as P-agonist, while the extensor of M-joint will be called Mantagonist and the extensor of P-joint as P-antagonist. Similarly, during extensions in M-joint, the extensor muscles will be addressed as M-agonist and P-agonist, while the flexor muscles will be addressed as M-antagonist and P-antagonist. After the EMG signals were rectified and filtered with a second-order Butterworth 100 Hz filter, all the trials were viewed on the monitor screen, and the time of the beginning of the first agonist burst (biceps for elbow flexion; triceps for elbow extension, flexor carpi radialis for wrist flexion, and extensor carpi ulnaris for wrist extension) was defined visually. This time was used for alignment and for all further data processing as “time zero”. The following integral measures of muscle activity were used: 1. Background EMG integrated over 50 ms for each muscle and each trial in each series (an integral of the EMG from - 150 ms to - 100 ms with respect to time zero); 2. M-agonist burst (EMG integral from 0 to the time of peak velocity in the focal joint); 3. M-antagonist burst (EMG integral from 0 to the end of the movement in the focal joint defined as time when velocity returns to zero for the first time); 4. P-agonist and P-antagonist bursts (defined as EMGs integrated over the same periods as M-agonist and M-antagonist); All the integral measures were corrected for the background activity using the following formula: jE, = /E - jEb * (A t/50), where /E is an integral measure of EMG, 1~5, is the corrected EMG integral, /E, is the background EMG integral over 50 ms, and At is the time of integration for /I!? in ms. We further normalized EMG integrals by dividing them by the integral of the M-agonist burst, averaged across 6 fastest trials in a series when this particular muscle was the M-agonist. The following kinematic parameters were used: 1. Maximal deviation (API from the initial position in the P-joint; 2. Peak velocity in the M-joint. The cross-correlation functions between pairs of the EMG signals were calculated for each trial and for each subject separately after filtering the rectified EMGs with a 40 Hz second-order Butter-worth filter. Additional filtering was used because we were interested in the relations between the
356
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
EMG envelopes. Peak values of the cross-correlation functions and time shifts of the peaks were measured. Statistical procedures included Z-transformation of the coefficients of linear correlation and single-group, two-tailed, Student’s t-tests.
3. Results Typically, EMG patterns in both muscle pairs demonstrated commonly observed “tri-phasic” patterns (for a review see Gottlieb et al., 1989a,b) in a wide range of movement speeds. During movements in either joint and in either direction, the elbow flexor (biceps) and the wrist flexor (flexor carpi radialis) tended to demonstrate similar patterns of activity, while the elbow extensor (triceps) and the wrist extensor (extensor carpi ulnaris) also showed simultaneous EMG patterns. In particular, during both elbow flexion and wrist flexion, the movements were initiated by an EMG burst in the prime mover (biceps or wrist flexor), accompanied by a simultaneous EMG burst in the other flexor muscle, and followed by EMG bursts in triceps and wrist extensor. During both elbow extension and wrist extension, the first EMG events were bursts in triceps and wrist extensor, followed by EMG bursts in biceps and wrist flexor. Because of the similarity of the EMG patterns observed in the two flexors and in the two extensors, we address postural muscles as P-agonist and P-antagonist with respect to the focal movement (cf. the Methods). To illustrate the typical kinematic and EMG patterns, 24 trials were selected from one series for one subject and were divided into three groups: (1) Eight fastest movements; (2) Eight movements at medium speeds; and (3) Eight slowest movements. The averages across the trials in each group for an elbow extension series are presented in Fig. 1. The arrows show the time of alignment (the beginning of the first triceps EMG burst) which, in this figure corresponds to 0.15 s. The wrist joint exhibits rather complex trajectories, with peak deviations from the initial position scaling with velocity of the primary movement. However, final deviation of the wrist from the initial position is not increasing with velocity. For better visualization, we inverted the EMGs of both flexor muscles, biceps (M-antagonist) and wrist flexor (P-antagonist). Note the similarity of the changes in the EMG bursts in both flexors with velocity of the focal movement. In particular, the latency of the EMG bursts in both biceps and wrist flexor has a general tendency to decrease with an increase in move-
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
357
ELBOW EXTENSION ELBOW ANGLE 357
WRIST ANGLE
0
01
02
0.3
0.4
0.5
0.6
5ooWRIST VELOCITY r
ml
EMG r
Fig. 1. Kinematic and EMG patterns during elbow extensions in one of the subjects (averages across 8 fastest, slowest and medium speed trials are shown). Note a typical tri-phasic pattern in the wrist flexors and extensors. M-agonist (triceps) and P-agonist (wrist extensor) EMGs are shown upwards; Mantagonist (biceps) and P-antagonist (wrist flexor) EMGs are inverted for better visualization. Time scale is in s; EMG scales are in arbitrary units; trajectory scales are in deg; velocity scales are in deg/s. Extension corresponds to positive angle values. According to our convention, the beginning of the M-agonist burst (time zero) corresponds to 0.15 s.
ment velocity. In this particular set of data, the dependence of the antagonist latency on focal joint velocity is seen more clearly in P-antagonist (wrist flexor). Early activation of P-agonist (wrist extensor in Fig. 1) during faster movements was seen in 6 out of 20 series (4 directions x 5 subjects).
M.B. Shapiro et al. /Human
358
Elbow
AP 16
r
-
y = 4.0561
Flexion
Movement Science 14 (1995) 351-369 Wrist
AP
+ ~044793~
R= 0.8378
__
1
Flexion
y = 0.093255
+,O.O0197\6x .
R= 0.30585
.
0
-1-
.
.
.
AP
“r
-350
__ y =
-300
-250
Peak
Velocity
Elbow 3.1125
-200
-150
-1500
Extension
Wrist
AP R= 0.83 118
1
I
-1000 Peak
+ ~.022823x
.
I
-5 -400
.
.
r __
-500
Velocity
Extension
y = -1.4819
+ O.OCCQ4617x R= 0.036984 . . .
. -1
.
.
l
.
.
.
.
,’
-2
t
250
300
350 Peak
400 Velocity
450
500
550
400
.
500
I
t
600
700
Peak
Velocity
.
. ”
.
-
I
200
. .
.
.
.
.
.
” .
=-•-. I
800
1
900
Fig. 2. Maximal deviations (AP) from the initial position of the P-joints for separate trials in each series for one subject. Note that AP in the wrist linearly depends on peak velocity for elbow movements with high correlation. AP scales are in deg, peak velocity scales are in deg/s. Negative values of velocity correspond to joint flexion.
More commonly, however, there was velocity-independent simultaneous activation of M- and P-agonist muscles. Fig. 2 shows a typical relation between peak movement velocity (V,,) and maximal deviation of the postural joint (AP) from the initial position measured in individual trials for each series. Note that wrist AP during elbow movements linearly depends upon V,, in the elbow with high coefficients of linear correlation. This is not true, however, for AP in the
MB. Shapiro et al. /Human
Movement
Science 14 (1995) 351-369
359
elbow joint during wrist movements. Such graphs were plotted for each subject. Z-transform was applied to the correlation coefficients R.Z-scores were calculated for each subject and each movement direction and averaged across subjects. In particular, for elbow flexions, mean Z = 1.004 (se f 0.237; p < 0.01; single group, two-tailed Student’s t-test); for elbow extensions, mean Z = 1.097 (se f 0.119; p < 0.01; single group, two-tailed Student’s t-test). During wrist movements, only in two subjects during wrist flexions, Z > 0.5; for all other wrist movements, Z-values were small. Because of the similarity of the behavior of the M- and P-agonist and antagonist muscles, we integrated their EMGs over the same time intervals. For individual trials, EMGs of both M-agonist and P-agonist were integrated over acceleration time (from the time of alignment to the time of peak velocity in the focal joint). EMG of both M-antagonist and Pantagonist were integrated over movement time (cf. Gottlieb et al., 1989b). EMG integrals for both M- and P-muscles increased with V,, (typical examples for one of the subjects are shown in Figs. 3 and 4). This dependence was linear, with high values of the coefficients of linear correlation (RIin most cases. Low values of R were commonly seen only during wrist extension movements. We would like to note that the values of the coefficients of linear correlation between V,, and EMG integrals were not smaller for the EMGs in P-muscles than for the EMGs in M-muscles (e.g., the right lower panel in Fig. 3 and the right upper panel in Fig. 4; see also Table 1). Z-transform was applied to the coefficients of linear correlation (R), and Z-scores for the dependence of the EMG integrals upon V,,, are presented in the upper part of Table 1 (mean values and standard errors in parentheses). Note the high mean values of Z for all the movements except wrist extensions. In a few cases, integrals of the EMGs demonstrated a pronounced non-linear dependence upon velocity in the focal joint with a threshold effect (Fig. 5). Note, however, that similar non-linear relations were seen for the EMG integrals (/EMG) in both M- and P-muscles. We assessed the relative timing of pairs of the EMGs with the method of cross-correlation. The time shift corresponding to the peak of a cross-correlation function was assumed to reflect the “average delay” (latency) between a pair of EMGs. Peak values of the coefficients of correlation were commonly over 0.7. Averages across the subjects of the Z-scores for the linear relations between V,, and latency values are presented in the lower part of Table 1. These values are rather low, although in some cases they reach the level of statistical significance. Note, that significant relations
M.B. Shapiro et al. /Human
360
Elbow IEMC
~
Movement Science I4 (1995) 351-369
Flexion
y = -0.50283
Wrist
+ -0.0026286x
3427 + -0.0015567~
R= 0.8703 I R= 0.88155
IEMC
y=
-
1.4
-
-
Flexion
0.076153
+ -0.0006932x
y = -0.55656
. 1.2
R= 0.84009
+ -0.OC~37515x R= 0.80515
M-ag - * - f P-zig -
.
0.8 0.8 0.6 0.6 0.4 0.4 0.2
02
n “Coo
-550
-500
450
-400
Peak
Elbow IEMG 2 r
-y -
-300
-250
0 -1600
-200
-1400
Wrist
Extension +0.0018025x
-y = 0.029317
-loo0
Peak
+ 0.00086028x
R= 0.77621 R= 0.6965
IEMG
__
2
-o-
i
1S
-1200
Velocity
= ~.I8598 -
-350
-
-800
600
400
Velocity
Extension
y -_ -0.05766
+0.0013415x
-y = -1.9981
+0.0037441x
R= 0.42404 R=0.80611
M-ag
-f P-ag
O,/’
IS
I
0.5
0.5
0
I
0 200
300
400
500 Peak
600 Velocity
700
800
900
500
600
700 Peak
800
900
1000
1100
Velocity
Fig. 3. EMGs of M-agonist C/M-ag) and P-agonist C/P-ag), integrated from time zero to time of peak velocity (V,,,), in one subject. Note that EMG integrals for both M- and P-muscles increase with an increase of L&,x. EMG integrals scales are in arbitrary units, peak velocity scales are in deg/s. Negative values of velocity correspond to joint flexion.
were more commonly seen for P-antagonist latency. Fig. 6 presents an illustration of two series in one subject (wrist flexions and wrist extensions) with rather high coefficients of linear correlation (RI between peak velocity and P-antagonist latency. Note that M-antagonist latency shows a strong linear relation to peak velocity for the wrist flexion but not for the wrist extension movements.
M.B. Shapiro et al. /Human
Elbow JEMC 2.5
__ _ _
r
Movement Science 14 (1995) 351-369
Flexion
y = -0.51738 y = -054052
361
Wrist Flexion
+ ~.00;2115Olx
R= 0.86055
JEMG
+ -0.0033596x
R= 0.85809
I .4
-
y=
-
-
--+-_ 8
I .2
0.01196
+ -0.00016895x
-y = -0.9063
R= 0.43687
+ -0.0012151x
R= 0.86702
M-ant f P-ant
1 0.8 0.6 0.4 0.2
0.5
0 0 -600
iI2 -550
-500
-450
-400
Peak
Elbow JEMG 0.8
-
-
-350
-300
-250
-200
-1600
-1400
-1200
Velocity
-loo0
Peak
-800
-600
400
Velocity
Wrist Extension
Extension
y = -0.090112
+ 0.00059942x
y = 0.071819
+ 0.00037413x
R= 0.69388
JEMG
__ -
R= 049789
--+8
0.7
y = dI.0439 -
+ 0.0001915x
-7 = -0.12594
-
R= 056632
+ O.OCK!Sl283r
R= 0.4574
M-ant f M-ant 0
0.4
-
0
/
/ 0.3
-
1:
-OS
0 -200
300
400
500 Peak
600 Velocity
Fig. 4. EMGs of M-antagonist one subject. Note that EMG velocity. EMG integrals scales of velocity correspond to joint
700
800
900
500
,-b
/
/
I
9
I
t
1
600
700
800
900
1000
Peak
, 1100
Velocity
(1 M-ant) and P-antagonist (/P-ant), integrated over movement time, in integrals for both M-and P-muscles increase with an increase of peak are in arbitrary units, peak velocity scales are in deg/s. Negative values flexion.
We also analyzed the relation between the EMG integrals in M-agonist and P-agonist as well as between M-antagonist and P-antagonist across all movement velocities. Fig. 7 illustrates examples of such relations for one of the subjects while Table 2 presents the averaged across the subjects values of Z-scores. Statistically significant relations were observed for all movements except wrist extensions.
362
M.B. Shapiro et al. /Human
Table 1 Z-scores for the linear correlations
between
Elbow flexion \ M-agonist /P-agonist 1 M-antagonist j P-antagonist Latency (P-ag) Latency (M-ant) Latency (P-ant)
1.038 (0.102) 1.098 (0.141) 1.037 (0.123) 0.938 (0.116) 0.335 (0.099) 0.280 (0.152) 0.461 (0.112)
Averaged across five subjects * p < 0.05; * * p < 0.01.
** ** ** **
’
Movement Science 14 (1995) 351-369
peak velocity
and EMG indices
Elbow extension
Wrist flexion
Wrist extension
1.034 * ’ (0.136) 0.728 * * (0.174) 0.763 * * (0.107) 0.783 * * (0.106) 0.305 (0.136) 0.545 * * (0.084) 0.58 * * CO.1151
0.886 * * (0.167) 0.844 * * (0.121) 0.789 * * (0.141) 0.908 * * (0.152) 0.216 (0.056) 0.373 CO.2181 0.455 * (0.188)
0.278 (0.105) 0.745 * * (0.171) 0.319 CO.1591 0.459 (0.203) 0.386 CO.0571 0.252 (0.131) 0.397 CO.1941
values are presented.
Wrist
Standard
errors
are shown in parentheses.
Flexion
~ y = 0.0014239 * ~‘(-0.0040594~) R= 0.73969 ~EMC_ ~. y _ - 0.0017249 * e”(-0.0043269x) R= 0.86792 1.4 -\ b 1.2 - 800 O\ l\ \ do
;
f;;;g$)
-1600 -1400 -1200 -1000 -800 Peak Velocity
-600
-400
Fig. 5. An example of the threshold effect, observed in a number of wrist flexion series. Note that the threshold effect was observed in both P-agonist (I bidO- VP) , biceps EMG integrated from 0 to the time of peak velocity) and P-antagonist (/tric(MT) , triceps EMG integrated over movement time). EMG integral scales are in arbitrary units, peak velocity scales are in deg/s. Negative values of velocity correspond to joint flexion.
M. B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
Wrist Flexion y = 0.20458 + 0.00013174x - y = 0.21886 +0.00017435x
LZileOCy - 03
F
-0.05
1
I
025
L
-1200
-1400
Latency
R= 0.76776 R= 0.79286
-M-antagonist - e- -P-antagonist
0.25
I
-1000 -800 600 Peak Velocity
363
l
I
I
400
-200
Wrist Extension y = 0.15106 + ~.00010653x
-
- y = 0.25929
- 0
0.2
R= 0.3592 + -0.0002607 Ix R= 0.75 178 --c M-antagonist - 8 -P-antagonist
0
0.15 0.1 0.05 0 Xl.05 1 300 Fig. 6. Linear dependence one out of five subjects series. In wrist exiension
’
400
of the latencies
in wrist flexion
’
500
’
’
’
600 700 800 Peak Velocity
of M-and
P-antagonists
and wrist extension
series, however,
P-antagonist
velocity than M-antagonist latency. Latency scales values of velocity correspond to joint flexion.
’
’
1
900 1000 1100 upon peak velocity,
series. Note high correlation
latency shows stronger
are in s, peak velocity
observed in only in wrist flexion
dependence
scales are in deg/s.
upon
peak
Negative
4. Discussion A movement in a joint of a multi-joint limb creates reactive and inertial torques in other joints of the limb that perturb posture in these joints. Faster movements are apparently accompanied by larger torques at the focal joint and, correspondingly, larger torques at other joints of the limb.
364
M.B. Shapiro et al. /Human Elbow
Jr-a.2 _-
0.8
Movement Science 14 (1995) 351-369
Flexion
y = 0.015944
Elbow
IP-ag
+ 0.4746%
R= 0.81186
-
12
Extension
y = 0.098049
+ 0.42281~
R= 0.79492
. 0.7
. .
0
02
0.4
0.6
1
0.8
1.2
IM-ag
JP-allt
JP-ant __
25
y = 0.27942
+ 1.1165x
R= 0.81192
__
0.5
y = 0.15978 .
+
0.45
.
F 0.4
-
0.35
-
0.3
-
0.25
-
0.2 0.15
-
/
- l*@ .
l
l0
01
I
0
02
0
0.4
0.6
0.8
I
,
1
1.2
0.1.’ 0
0.1
’
’
’
’
’
’
’
0.2
0.3
0.4
0.5
0.6
0.7
0.8
[M-ant
jM-ant
Fig. 7. Relations between integrated EMG in agonists of two joints, / M-ag and 1 P-ag, and antagonists of two joints, /M-ant and /P-ant. The dependencies are linear with high coefficients of linear correlation. EMG integrals scales are in arbitrary units. Table 2 Z-scores for the linear correlations pairs
/M-agonist vs. 1 P-agonist / M-antag vs. 1 P-antag Averaged
between
EMG integrals
for M-and
P-agonist
and antagonist
Elbow flexion
Elbow extension
Wrist flexion
Wrist extension
0.905 (0.116) 0.994 CO.0761
0.767 (0.162) 0.749 CO.1131
0.733 (0.281) 0.911 CO.1211
0.324 CO.0841 0.367 (0.163)
across five subjects
values are presented.
Standard
errors
are shown in parentheses.
muscle
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
365
This may be the main reason for the strong positive dependence of the postural joint deviation upon movement velocity observed in our experiments during elbow movements. Such a relation was not seen during wrist movements. This asymmetry is likely to reflect the asymmetry of the biomechanics, in particular the difference in the mass-inertial properties of the segments and peak torque values at the focal joint (e.g., Zatsiorsky et al., 1984). Let us assume that the central nervous system wishes to exercise control over the trajectory of the endpoint of the limb. This trajectory will apparently depend upon movements in all the joints of the limb including non-focal joints. The central nervous system has three basic options to deal with these perturbations: First, it may ignore them, and allow a non-focal joint flap. Eventually the flapping will be damped by the viscoelastic properties of the muscles and tendons (cf. Gottlieb and Agarwal, 1988; Kearney and Hunter, 1990). Second, it may co-contract the muscles acting at a non-focal joint thus increasing joint stiffness and viscosity (Davis and Kelso, 1982; Kearney and Hunter, 1990; Latash, 1992). This will not eliminate flapping but will apparently decrease the deviation of the joint from the original equilibrium position. Third, it may generate a non-trivial phasic pattern of activation of the muscles acting at a non-focal joint that would counteract the perturbing forces and, in an ideal world, might be able to keep the non-focal joint motionless in the original position. Apparently, the first strategy is the easiest one to follow. However, it is likely to lead to relatively large deviations in the non-focal joints and, therefore, to large deviations of the endpoint from the desired trajectory. The second strategy is not very challenging either, and is likely to be effective in decreasing the deviations of the endpoint. However, it will never be able to completely eliminate the effects of the perturbing torques at non-focal joints. The third strategy looks most challenging, but it has a potential of eliminating (or nearly eliminating) deviations in the non-focal joints thus keeping the actual endpoint trajectory close to the planned one. The question emerges whether the task of generating the exact patterns of hypothetical motor commands to muscles controlling non-focal joints is heavy enough to prevent the central nervous system from using the optimal, third strategy. Our earlier study (Latash et al., 1995a) has shown that the central nervous system takes this challenge, at least for movements performed “as fast as possible”, and controls non-focal joints with phasic
366
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
bursts of muscle activity closely resembling the tri-phasic EMG pattern seen in the muscles controlling the focal joint (see also Aoki, 1991; Koshland et al., 1991). The present study expands this conclusion to movements at intermediate speeds. Moreover, it suggests that the patterns of muscle activation at non-focal joints may be related to the patterns of muscle activation at the focal joint by a simple scaling procedure. We think that our observations provide support for a hypothesis that the patterns of muscle activation at focal and non-focal joints are consequences of one and the same central control process (Latash et al., 1995a; cf. Massion, 1992). Then, a time scaling of the hypothetical control process will result in similar scaling of the activation patterns in the muscles controlling all the joints involved in the motor task. Even if a task is formulated explicitly as a requirement to move a focal joint but not other joints of the limb, the central controller may reinterpret it as a more natural task of moving the endpoint of the limb and, therefore, use similar scaling of muscle activation patterns at all the joints of the limb. Phasic patterns of the EMG activity in postural muscles during fast voluntary movements have been described by Friedli et al. (1984) and Crenna et al. (1987). They resemble the well-known tri-phasic EMG pattern observed in a focal agonist-antagonist pair during fast voluntary movements (Hallett et al., 1975; Lestienne, 1979; Mustard and Lee, 1987; Gottlieb et al., 1989a). In the present study, we observed tri-phasic EMG patterns in muscles acting at the focal joint and in postural muscles with a tight coupling between the EMG bursts in the two flexors as well as in the two extensors that was preserved over a wide range of movement speeds. So, the earlier finding is not an artifact of movements performed “as fast as possible” and is not related to any ceiling or saturation effect (cf. commentaries to Gottlieb et al., 1989a). Major characteristics of the EMG patterns, such as EMG integrals and latency of the antagonist burst, showed similar dependence upon movement velocity for both flexor-extensor pairs acting at the focal and at the postural joint. Typical changes in these characteristics with speed in both M- and P-muscles are similar to those described earlier for single-joint movements (Corcos et al., 1989). In particular, an increase in focal movement speed was accompanied by an increase in the EMG integrals of both M- and P-agonists and both M- and P-antagonists as well as by a decrease in the latencies of both M- and P-antagonists. In some cases, we observed a non-linear relation between integrated EMG indices and movement speed or atypical threshold effects (cf. Lee, 1980); however, even in these cases,
M.B. Shapiro et al. /Human
Movement Science 14 (I 995) 351-369
367
similar atypical behavior was seen in both M- and P-muscle pairs. So, the basic rule of scaling the activity of muscles acting at a focal and non-focal joints holds through the investigated range of movement velocities. There were certain differences in the behavior of the M- and P-muscles although they were overshadowed by the similarities. In particular, there was a tendency for an earlier activation of P-agonist for faster movements with respect to the time of activation of M-agonist. This lead to an earlier and bigger displacement of the postural joint in the direction opposite to the later displacement caused by inertial forces. These findings are in a good correspondence with the ideas and observations by Bouisset and Zattara (1987, 1990). Most of our findings were well reproducible both over subjects and over different movements. Wrist extension was an exception. During these series, all the subjects demonstrated much lower coefficients of correlation between peak velocity in the focal movements and different EMG indices. We do not have a good explanation for these observations. The subjects reported that wrist extension was the least comfortable movement and that they felt clumsy performing it. Maybe, the low reproducibility of the findings during these movements is a correlate of clumsiness. Finally, we suggest that the central nervous system is simplifying the task of postural control and using the same synergy across movement speeds rather than solving explicitly the inverse dynamics equations and calculating each time which muscle force (and activation) patterns are required to maintain control over the trajectory of the limb endpoint. That is, the central nervous system uses modulation of one or a few parameters to adjust a motor command to modifications in commonly changing characteristics of motor task such as planned movement velocity (or movement time). This strategy eventually leads to the generation of appropriate patterns of muscle forces and, thus, may be considered an example of an approximate inverse dynamics which does not prescribe exact force patterns but rather allows them to develop based on actual movement course (joint kinematics). Note that our earlier biomechanical analysis (Latash et al., 1995a) is valid across movement speeds. Therefore, the described synergy is a universal and efficient way of minimizing postural deviations at different speeds. Acknowledgements
This study was in part supported by a grant HD-30128 from the National Center for Medical Rehabilitation Research, NIH.
368
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
References Almeida G.L., AS. Aruin and M.L. Latash, 1994. ‘Organization of a simple, two-joint synergy in individuals with Down syndrome’. In: M.L. Latash (ed.), Motor control in Down syndrome-II (pp. 16-23). Chicago, IL: Rush University. Aoki, F., 1991. Activity patterns of upper arm muscles in relation to direction of rapid wrist movement in man. Experimental Brain Research 83, 679-682. Bernstein, N.A., 1967. The co-ordination and regulation of movements. Oxford: Pergamon Press. Bouisset, S. and M. Zattara, 1987. Biomechanical study of the programming of anticipatory postural adjustments associated with voluntary movement. Journal of Biomechanics 20, 735-742. Bouisset, S. and M. Zattara, 1990. ‘Segmental movement as a perturbation to balance? Facts and concepts’. In: J.M. Winters and S.L.-Y. Woo teds.), Multiple muscle systems. Biomechanics and movement organization (pp. 498-506). New York: Springer-Verlag. Burbaud, P., F. Garanx, Ch. Gross and B. Bioulac, 1988. Postural adjustments in the monkey: Effects of velocity on EMG sequence. Neuroscience Letters 84, 51-56. Corcos, D.M., G.L. Gottlieb and G.C. Agarwal, 1989. Organizing principles for single joint movements. II. A speed-sensitive strategy. Journal of Neurophysiology 62, 358-368. Crenna, P., C. Frigo, J. Massion and A. Pedotti, 1987. Forward and backward axial synergies in man. Experimental Brain Research 65, 538-548. Davis, W.R. and J.A.S. Kelso, 1982. Analysis of “invariant characteristics” in the motor control of Down’s syndrome and normal subjects. Journal of Motor Behavior 14, 194-212. Friedli, W.G., M. Hallett and S.R. Simon, 1984. Postural adjustments associated with rapid voluntary arm movements. I. Electromyographic data. Journal of Neurology, Neurosurgery and Psychiatry 47, 611-622. Gottlieb, G.L. and G.C. Agarwal, 1988. Compliance of single joints: Elastic and plastic characteristics. Journal of Neurophysiology 59, 937-951. Gottlieb, G.L., D.M. Corcos and G.C. Agarwal, 1989a. Strategies for the control of voluntary movements with one mechanical degree of freedom. Behavioral and Brain Sciences 12, 189-250. Gottlieb, G.L., D.M. Corcos and G.C. Agarwal, 1989b. Organizing principles for single joint movements. I: A speed-insensitive strategy. Journal of Neurophysiology 62, 342-357. Hallett, M., B. Shahani and R. Young, 1975. EMG analysis of stereotyped voluntary movements in man. Journal of Neurology, Neurosurgery and Psychiatry 38, 1154-1162. Horak, F.B. and L.M. Nashner, 1986. Central program of postural movements: Adaptation to altered support-surface configurations. Journal of Neurophysiology 55, 1369-1381. Horak, F.B., L.M. Nashner and H.C. Diener, 1990. Postural strategies associated with somatosensory and vestibular loss. Experimental Brain Research 82, 167-177. Hugon, M., J. Massion and M. Wiesendanger, 1982. Anticipatory postural changes induced by active unloading and comparison with passive unloading in man. Pflugers Archive 393, 292-296. Kearney, R. and J.A. Hunter, 1990 System identification of human joint dynamics. CRC Critical Reviews in Biomedical Engineering 18, 55-87 Koshland, G.F., L. Gerilovsky and Z. Hasan, 1991. Activity of wrist muscles elicited during imposed or voluntary movements about the elbow joint. Journal of Motor Behavior 23, 91-100. Latash, M.L., 1992. Independent control of joint stiffness in the framework of the equilibrium-point hypothesis. Biological Cybernetics 67, 377-384. Latash M.L., A.S. Aruin and M.B. Shapiro, 1995a. The relation between posture and movement: A study of a simple synergy in a two-joint task. Human Movement Science 14, 79-107. Latash M.L., A.S. Aruin, I. Neyman, J.J. Nicholas and M.B. Shapiro, 1995b. Feedforward postural reactions in patients with Parkinson’s disease in a two-joint motor task. Electroencephalography and Clinical Neurophysiology 97, 77-89.
M.B. Shapiro et al. /Human
Movement Science 14 (1995) 351-369
369
Lee, W.A., 1980. Anticipatory control of posture and task muscles during rapid arm flexion. Journal of Motor Behavior 12, 185-196. Lee, W.A., A. Boughton and W.Z. Rymer, 1987. Absence of stretch reflex gain enhancement in voluntary activated spastic muscle. Experimental Neurology 98, 317-355. Lestienne, F., 1979. Effects of inertial loads and velocity on the braking process of voluntary limb movements. Experimental Brain Research 35, 407-418. Massion, J., 1992. Movement, posture and equilibrium: Interaction and coordination. Progress in Neurobiology 38, 35-56. Mustard, B.E. and R.G. Lee, 1987. Relationship between EMG patterns and kinematic properties for flexion movements at the human wrist. Experimental Brain Research 66, 247-256. Oddsson, L., 1989. Motor patterns of a fast voluntary postural task in man: trunk extension in standing. Acta Physiologica Scandinavica 136, 47-58. Zatsiorsky, V.M., A.S. Aruin and V.N. Selujanov, 1984. Biomechanik des Menschlichen Bewegungsapparates. Berlin: Sportverlag.
