Peripheral control of the antagonist muscle during unexpectedly loaded arm movements

260

Bra#i Research, 585 (1992) 260-266 ~O 1992 Elsevier Science Publishers B.V. All rights reserved 0006-8993/92/$05.00

BRES !~097

Short Communications

Peripheral control of the antagonist muscle during unexpectedly loaded arm movements R o g e r W. S i m m o n s a n d C h a r l e s R i c h a r d s o n San Diego State Unirersity, San Diego, CA 92182 (USA) (Accepted 10 December 1991)

KO' words: Arm flexion: Electromyograph; Mechanical impedance: Impedance hypothesis

Ten subjects completed a series of goal-directed arm flex!on movements unexpectedly perturbed by three different types of mechanical load. Examination of ¢lcctromyognlph (EMG) waveform,~ and kinematic information collected during randomly distributed test trials facilitated investigation into the interaction between loading conditions and the response-associated EMG innervation patterns. Results of the EMG waveform analysis revealed that inertial and spring loads produced cocontraction and triphasic activation patterns, respectively. Unexpected application of a stretched-spring load, which produced a change in initial torque values without changing the rate of loading, also resulted in the use of a triphasic activation pattern. These different EMG patterns were observed while movement displacement for all three loads fell within the limits of the target area.

In a previous experiment ~ it was demonstrated that control of externally loaded limb movements is based on the central nervous system (CNS) regulating limb stiffness by modifying the associated muscle activation pattern. Specifically, when an exterr,tl load provides little or no inherent stiffness (e.g. a, inertial load) the stiffness required to control the movement is provided by cocontraction of the agonist and antagonist muscles. If the applied load provides a high degree of stiffness (e.g. a spring load) the contribution of the associated musculature in providing stiffness to control the response is considerably reduced. Under these conditions motor control would be maintained by the adoption of a triphasic activation pattern. These results were consistent with formal descriptions of stiffness regulation such as the mechanical impedance hypothesis TM. Central to the impedance hypothesis is the notion that the CNS 'sets' the appropriate neuromuscular pattern after the subject has assessed the task requiremcnts prior to the response. By unexpectedly applying different types of external loads to the limb of blindfolded subjects we eliminated the opportunity to engage in this pre-response assessment strategy. How-

ever, even under these restricted conditions, subjects altered their neuromuscular activation patterns to pro. duce intrinsic muscle stiffness which synergistically complemented external load stiffness. The present experiment investigated the nature of the interaction between electromyograph (EMG) patterns and external loading by using three different types of external load: an inertial load, a spring load, and a third load consisting of a stretched-spring load. Application of an inertial or spring load was predicted to result in cocontraction and triphasic EMG patterns respectively. Of central interest was the type of EMG pattern associated with the stretched-spring load, which altered the initial torque requirements without changing subsequent loading characteristics. A schematic diagram of the apparatus is presented in Fig, I. Subjects placed their elbow over the axis of rotation of a light-weight lever (48,5 cm long) which freely rotated horizontally on a near-frictionless metal bearing. Subjects gripped a small contoured block placed at the distal end of the lever arm with the hand in a pronated position, The lever rested against a mechanical stop which defined the movement start

Corr('slxmdence: R.W, Simmons, Department of Physical Education, San Diego State University, San Diego, CA 92182, USA.

261 position at an angle of 0.8727 radians relative to the subject's frontal plane. The subject was required to move the lever arm as rapidly and accurately as possible in a counterclockwise direction toward a target area positioned 0.8013 radians to the left of the start position. The target area was marked with upper and lower boundaries of +0.065 radians. Three loads were separately applied to the apparatus. An inertial load, consisting of a small lead block weighing 1.1 kg, was placed in an aluminum cradle mounted at the distal end of the lever. The two spring loads were applied via a thin wire attached to the lever arm at a point 25.5 cm distal to the center of rotation. The wire passed around a freely rotating pulley (4.5 cm radius) positioned 20.3 cm to the right of the lever arm, and attached to a linear slide mechanism aligned with the wire. The 35 cm load spring could be hooked at one end to the slide and at the other end to a metal block which could be moved to one of two positions on the apparatus base board. In the first position (A), the spring was at resting length with zero tension. The spring set in position A was referred to as the spring load. in the second position (B), the metal block was moved 9 cm behind hook position A. Stretching the spring to position B increased the initial spring load to half the spring load when the lever arm was located at the center of the target area and defined the stretched-spring load. The tension developed on the spring at position B was 0.57 kg and 1.7 kg at the movement start position and target center respectively. Movement of the lever through 0.8(}13 radians to the target center generated a linear spring extension of 18

cm. The spring was constructed of piano wire and calibrated to require a force of 1.1 kg to stabilize the spring at 18 cm of extension. The stiffness coefficient of the spring was 0.11 kg/cm. An electrical recording of lever displacement and instantaneous tangential acceleration data were smoothed using a dual low-pass filter with a 20 Hz cut-off frequency. EMG activity was monitored using two pairs of disposable surface electrodes positioned parallel to the long axis of the biceps brachaii (agonist) and triceps brachaii (antagonist) muscles. Following standard preparation procedures ~ electrodes were positioned approximately 2 cm apart, and inter-electrode resistance was less than 10 k/L The electrodes were attached to four small transmitters tuned using a 1 Hz signal prior to testing. Transmitter signals were processed through four independent channels of a Transkinetics telemetry receiver unit with a signal-to-noise ratio of 65 dB at 1 kHz, plus 100/zV RF input, and low and high frequency ¢.at-off values of 10 Hz and 1500 Hz respectively. Six channels of analog information (displacement, acceleration and four channels of EMG) were sampled at a l kHz rate and digitized using a 12-bit A - D converter and stored using a 286 PC. Ten right-handed male subjects were blindfolded and seated with the apparatus lever arm positioned level with the lower end of the subject's sternum. Subjects completed a set of thirty training trials with a 1.1 kg inertial load. Quantitative and qualitative knowledge of results was provided after each trial (e.g. plus two units and speed up the movement). This

SLIDE PULLEY

~

I~

[~

INERTIAL

/

~

K

SPRING ~

METAL BLOCK ATPOSITION A

[

~

METAL BEARING

/

%

TARGET

Fig. 1. A schematicdiagramof the apparatus.

i

POSITION B

I

I

262 initial training session was followed by a further set of training trials that continued until the subject completed five consecutive trials with each response terminated in the target zone while responding as fast as possible. The fifth successful training trial was followed by a test trial in which the 1.1 kg inertial load was removed from the apparatus and without the subject knowing, immediately re-applied to the lever arm or replaced by the spring load in position A or B. This manipulation resulted in an inertial (1.1 kg) test load, a spring (1.1 kg) test load and a spring-stretched (1.7 kg) test load

.900,700¢/} z -< .500-

.300•1 0 0 I

'

100.

I

'

300.

I

500.

respectively. K i n e m a t i c and E M G data were recorded

fiw each test trial and were completed without provision of knowledge of results. This sequence of training trials, followed by a test trial, was repeated fifteen times for all subjects with each of the three loads being randomly applied five times across trials. The mean displacement for each load conditien was 0.863, 0,8013 and 0.7463 radians for the inertial, spring and stretched-spring conditions respectively. These values were within the target area, Movement time was defined as the time from movement initiation until the time when the deceleration curve crossed zero and was analyzed using a 3(load)x 5(trials) design '~ with repeated measures on both factors, The analysis revealed a significant load effect, F,,I M~ 117,9, P < 0,05. Post hoc analysis (Tukey's test) indicated that mean movement time for the inertial load condition (~ - 462,7 ms) was significantly greater than for either the spring (~ ~ 333,8 ms) or spring. stretched (~ - 328,9 ms) conditions. Each EMG record was rectified and smoothed (cutoff frequency = 100 Hzl~)), examined by two raters on two separate occasions (at least 24 h apart) and classi. fled as being either triphasic or cocontractive in nature, A triphasic pattern was defined as two discrete bursts of agonist activity separated by a silent period during which a discrete pulse of antagonist activity was observed, When antagonist activity was initiated during the silent period and then sustained to occur simultaneously with the second agonist pulse, the EMG record was classified as a cocontraction pattern. Typical displacement records and associated cocontraction and triphasic patterns (produced by subject 2) representing the EMG record for a single trial during inertial, spring and stretched-spring conditions are presented in Figs. 2, 3 and 4 respectively. The illustrated EMG patterns a~,: typical of those produced by all 10 subjects. A Wilcoxon matched-pairs signed ranks test ~ was used to test the hypothesis that the distributions of

z

'

i

"

.

100.

300.

100.

300,

500,

3 z,-j

w

i

I

500,

TIME (ms)

Fig, 2, EMO activity and displacement record for a sin~lu trial completed by subject 2 during a, inertial load condition, The top trace is th~ displacementotime record in radians-millisecond~,The middle and bottom traces are full.wave rectified biceps and triceps EMO records respectively, and illustrate a cocontraction activ,lion patter., The three traces are time-locked and the arrow represents the point of movement initiation,

EMG patterns were identical for the inertial, spring, and stretched.spring load conditions. With T,,o,.l~, - 23, significant differences were determined when the inertial load EMG patterns were compared with spring and stretched-spring load conditions. No significant difference was determined when comparing the EMG patterns of the two spring load groups. Table I contains percentage values for each type of innervation pattern as a function of load type. For inertial load trials 83.3% of the records were judged to be cocontractive, while for the spring and stretched-spring conditions 91.7% and 80.8% of the records, respectively, were classified as triphasic. Inter-rater reliability was calculated at phi--0.865. When transformed into a chi-squared statistic 6, this reliability coefficient was found to be significant, X~ = 15.5, P < 0.01.

263 .900-

.900-

,700-

.700Z

z <_ .500D

< .500-

.300-

,300-

.100-

.100" I

I

I

100.

300.

100.

300,

i

I

I

500.

100.

I

300.

500.

5~o.

100.

300.

s~o.

I

'

z

i

Z

LU

I

100,

'

I

300. TIME (ms)

'

'

I

100.

500.

'

300. TIME (ms)

I

500,

Fig, 3, EMG activity and displacement record for a single trial completed by subject 2 during a spring-load movement, Traces are the same as in Fig. 2 and the EMG records represent a triphasic activation p~ttern,

Fig. 4. EMG activity and displacement record for a single trial completed by subject 2 during a spring.stretched movement. Traces are the same as in Fig. 2 and the EMG pattern illustrates a triphusic activation pattern.

As an illustration of how the individual waveform pattern is maintained across load conditions the averaged EMG waveforms produced by subject 2 under the three load conditions are presented in Figs. 5, 6 and 7. The slight overlap of the agonist-antagonist during the spring and spring-stretched conditions is attributable to averaging of individual records which are not temporally scaled. This has the effect of 'smoothing out' the waveform such that the peak EMG value is underestimated and the onset and termination time of the agonist-antagonist trace is over-estimated. Nevertheless, the concontraction and triphasic patterns are distinguishable and occur only as predicted by type of load. The accuracy of the subjective classification of EMG waveforms wus checked by grouping test trials according to waveform (either cocontraction or triphasic) for each subject regardless of load type, and determining the time of the end of the antagonist pulse and the onset of the second agonist pulse.

To be consistent with the results produced by the subjective classification, cocontraction should manifest a temporal relationship in which the end of the antagonist pulse occurs later in time than the onset of the second agonist pulse. The data presented in Table II are consistent with this predicted temporal relationship in all subjects. Independent t-tests of each pair of means for each subject re~ealed significant differences in eight of ten subjects. The data for the two remaining subjects were in the predicted direction.

TABLE ! Percentages of observed triphasic and cocontraction neuromuscular activation patterns during three loading conditions

Load condition Inertial Spring Stretched-spring

Neuromuscular activation pattern Cocontraction Triphasic 16.7 83.3 91.7 8.3 80.8 19.2

2o4 For triphasic classification the end of the antagonist pulse should occur at the same time as, or earlier than, the onset of the second agonist pulse. The data of Table !! are consistent with this prediction for eight of ten subjects, Subjects 5 and 6 produced results which were typical of the cocontraction activation pattern. Individual inspection of trial data for these two subjects revealed that in each case the end of the antagonist activity for two trials out of ten ~as considerably delayed relative to the onset of the second agonist pulse. This had the effect of producing a mean value which appeared more aligned with concontraction than triphasic activity. The remaining trials produced times consistent with the prediction of triphasic patterning. Using the time of movement initiation as a standard reference, the following time intervals were also determined for each trial: time of initiation of antagonist activity, time of maximal antagonist activity and the duration of the antagonist pulse. Data for each time interval were pooled across trials and analyzed using a

.900.700•500 t t~

•300 t .100~ I 100.

,

I 300.

i

|

|

I

500.

I

lOO.

300.

500.

I

I

I

t/) Z O =[ LU

=

900 =

~

700 100.

300, 500, TIME (ms) Fig, 6, Average F,MG |1¢livily and disph1~¢nlu11I data pruduccd by suhje¢l 2 during a spring.load 111OVClUeUl,'rl1~: 1'CCil~1'ucal activation of the ugonisl.anlugonisI pair is lypical of 11Iriphasic pattern, Traces ar,J II1~ salllC as iu Fig, 2,

=500 300 = =

100 = I

J

100.

300,

,

500,

1

-' £ ~ - - ~ ' '

I

I00.

I

100.

1

300,

'

I

300, TIME (ms)

/

500.

'

I

500,

Fig..s. Average EMG activity and displacement records for subject 2, Trials are averaged over five trials for an inertial load condition, The EMG pattern is cocontraction, Traces are the same as in Fig, 2,

one-way ANOVA'~. The moan temporal values for oach load and time interval arc presented in Table ill. The analyses revealed that load type did not affect onset time of antagonist activity, ~,a7 -- 2,54, P > 0.05, but did significantly change the time of maximal antagonist activity, F:,,~-- i 1.0, P < 0.01, and the duration of antagonist activity ~,27 = 22.5, P < 0.01. Post hoc analyses (Tukey test, P < 0.05) revealed that the time of maximal antagonist activity and the time of the entire autagonist burst were greater for the inertial load than tbr the two spring loads. The main result of the c~:pcriment indicated that application of a stretched-spring load produced a triphasic EMG pattern similar to that observed under normal spring load conditions (Figs. 3 and 4). As predicted, the inertial load condition produced a predominantly cocontraction activation pattern (Fig. 2). Within the context of this result, movements were terminated within the target area, while movement time and the amount of stiffness generated was load

265 T A B L E !!! .900-

Mean tbne (in milliseconds) of onset, maxinml, and chmttion of antagonist actirity under three loading conditions

.700-

t/) z t-,,, <-- . 5 0 0 ,,~ ,,1,, .300-

Load condition

inertial Spring Stretched-spring

.100I

'

100.

I

l

300.

500.

Time bffen,al Onset

Maximal

Duration

49.6 42. i 31.2

169.5 130.1 108.4

395.2 205.4 195.6

m i

~

z

_j.. l

100.

'

'-

I

'

100.

3(~0.

'

I

'

300. TIME (ms}

500.

I

500,

Fig, 7. Average E M G activity and displacemt~nt data produced by subj¢c! 2 during a spring-stretched mtwement. Similar to the activalion pattern produced under a spring-load condition EMG activity is triphasic, Traces ar~ Iht~ same as in Fil;, 2.

dependent. These results were produced under conditions in which vision was not allowed, loads were unexpectedly applied, and subjects were unaware that T A B L E il

Comparison of averaged times Jbr the end of antagoni,~t acticity ¢Antag) aml the onset of the second agonist ptdse (Ag2) for subjects as a ftmction of type of EMG pattern "~ubject

I 2 3 4 5 6 7 8 9 10

EMG Pattena Cocontraction

Triphasic

Antag

Ag 2

Antag

Ag 2

271.0 384.0 41)3.9 360.6 573,4 390.2 281.8 345.4 244.2 213.6

250,7 221.5 185.0 216.9 337.8 199.2 240.4 194,8 154.2 160,5

179.8 216.5 21 ! .1) 230,8 246,6 232,0 191 ,I 192.0 133.8 145.3

182,3 198,0 2(19. I 211),5 196,3 * 182,1 * 218,2 191.4 127.2 146,8

* * * * * * * *

Independent t-tests were used to determined significant differences between means where the highest critical value is t 3 = 3.18, P < 0.05 *

the type of load could change on any given test trial. Without pre-response information about the type of external load, the CNS should not have been able to consistently preset the neuromuscular activation pattern. However, subjects did regulate activity of the antagonist according to the stiffness hypothesis s, and it was concluded that this regulatory mechanism was based on peripheral feedback. Given the relatively fast speed of response, it is necessary to consider whether sufficient time was available during the response for feedback information to traverse the transcortical loop. Onset time of antagonist activity was unaffected by load type and considered to be centrally pre-programmed. This result extends earlier reports in which this time marker was uninfluenced by surplus inertial loading 2.1°, but is at odds with those results indicating onset time is related to the size of inertial load ~?, the type of instruction given to the subject and movement amplitude 4. in contrast to the invariance of antagonist onset time, the time of maximal antagonist activity varied with load type (Table !), with the shortest time interval between movement initiation and maximal activity being approximately 108 ms. This time interval exceeds the minimum time of 60-80 ms required to traverse the transcortical loop during 'intense displacements, and also exceeds the time of 37 ms required to detect a difference in angular velocity of 0.6 rad/s during variable inertial loading conditions 13, Therefore, sufficient time was available in the present study for feedback to alter pre-programmed commands in response to changes in the external load. When movement time and displacement data are considered together with EMG records, two levels of interrelated control emerge. At one level neuromuscular activity and movement time are varied in accordance with the dynamic characteristics of the external load. This produces a second level of control in which displacement is held constant (i.e. within the target area) across loading conditions "~.The regulation of the neuromuscular activity is facilitated through feedback which, after some critical time delay, may supplement centrally generated commands tl, or be present at the

266 neuromuscular level throughout the response s. Thus, the response is not based on a single set of motor commands that are invariantly applied across movement conditions, but is comprised of a regulated interaction between central commands and peripheral feedback, which takes into account the mechanical qualities of the limb-load system in order to produce the desired result. In summary, the experiment d e m o n s t r a t e d that neuromuscular stiffness is regulated according to the type of load applied to the limb. The regulation is manifested through altered antagonist activity via a feedback loop connection. This regulatory mechanism serves to produce invariant displacement responses as required by the task demands of the experiment.

! Basmajian. J.V.. Muscles Afire. 2nd ed,.. Williams and Wilkens, Baltimore, 1967. 2 Ben¢ck¢. R., Meinck. H.M. and Conrad, B. Rapid goal-directed elbow flexion movements: limitations of the speed control system due to neural constraints. E.ff~. Brain Res., 59 (1985)47(~-477. 3 Bizzi, E., Dev, P.. Morasso. P. and Polit, A., The effect of load

disturbances during centrally initiated movements, Z Neurophysiol., 41 (1978) 542-556. 4 Brown, S.H. and Cooke, D., Amplitude- and instruction-dependent modulation of movement-related electromyogram activity in humans, J. Physiol., 316 (1981) 97-107. 5 Conrad, B., Matsunami, K., Meyer-Lohmann, J., Weisendanger, M. and Brooks, V.B., Cortical load compensation during voluntary elbow movements, Brain Res., 71 (1974) 507-514. 6 Hays, W.L., Statistics for the Social Sciences 2nd edn., Holt, Rinehart, Winston, New York, 1973. 7 Hogan, N., Adaptive control of mechanical impedance by coactivation of antagonist muscles, IEEE Trans. Autom. Control, AC-29 (1984) 681-690. 8 Hogan, N., Bizzi, E., Mussa-lvaldi, F.A. and Flash, T., Controlling multijoint motor behavior. In K.B. Pandolf (Ed.), Exercise and Sport Science ReL'iews, MacMillan, New York, 1987. 9 Kirk, R.E., Experimental Design: Procedures for the Behavioral Sciences, Brooks/Cole, Belmont, CA, 1968. 10 Lestienne, F., Effects of inertial load and velocity on the braking process of voluntary limb movements. Exp. Brain Res., 35 (1979) 407-418. I 1 Paillard, J., The multichanneling of visual cues and the organization of a visually guided response, in G.E. Stelmach and J. Requin (Eds), Tutorials in Motor Behat'ior, North-Holland, Amsterdam, 1980. 12 Simmons, R.W. and Richardson, C., Peripheral regulation of stiffness during arm movements by coactivation of the antagonist muscles, Brain Res., 473 (1988) 134-140. 13 Smeets, J.B.J., Erkelens, CJ. and Denier van der Gon, JJ., Adjustments of fast goal-directed movements in response to an unexpected inertial load, EXp. Brain Res., 81 (1990) 303-312.