Electroencephalography and clinical Neurophysiology, 1983, 55:687-698 687 Elsevier Scientific Publishers Ireland, Ltd.

S T R E T C H REFLEX M O D U L A T I O N D U R I N G A CYCLIC ELBOW M O V E M E N T I

W.A. MACKAY, H.C. KWAN, J.T. MURPHY and Y.C. WONG Department of Physiology, University of Toronto, Toronto, Ont. M5S 1A8 (Canada)

(Accepted for publication: January 19, 1983)

Significant modulation of the stretch reflex dur- ing voluntary movements has been demonstrated in several studies on human subjects. For tracking movements of either the foot or arm, reflex re- sponsiveness has been shown to be maximal near the onset of voluntary muscle activity (Dufresne et al. 1980; Gottl ieb and Agarwal 1980; Soechting et al. 1981), and to follow a pattern which is quite distinct from that of the voluntary EMG. None of these studies, however, have specifically looked for, or demonstrated increases in reflex responsive- ness prior to the onset of voluntary E M G dis- charge, such as has been reported for the cat soleus muscle during the locomotor cycle (Akazawa et al. 1982). It appears that such increases do not occur prior to movements of maximum velocity (Brown and Cooke 1981; Hallett et al. 1981). The velocity of a movement has a critical bearing on reflex behavior. The myotatic reflexes are de- pressed at the onset of a very vigorous contraction (Desmedt and Godaux 1978; Soechting et al. 1981). Nonetheless, Bonnet (1981) and Bonnet and Re- quin (1982) have reported small but significant increases in the myotatic reflexes during the pre- paratory period before the cue to move at maxi- mum speed. According to Hallett et al. (1981), such modulation is inconsistent and the result of subtle changes in motoneuron excitability. The overwhelming majority of movements, however, are not performed as fast as possible. It is far more meaningful to examine the behavior of the stretch reflex prior to and during movements for which it is designed, rather than during extreme conditions

A preliminary report of this work was presented at the Soc. for Neurosci. annual meeting in Cincinnati, 1980 (Soc. Neuro- sci. Abstr, 6 (1980) 217).

which depress it. The pattern of stretch reflex modulation during movement is important to establish in detail because it provides important clues as to the function of the reflex. The reflex should be most responsive at those phases of a movement where it serves some definite function.

For arm muscles, movement-related modulation is more marked in the late stretch reflex, occurring 50-100 msec after the torque pulse, than in the early myotatic reflex (Wieneke and Denier van der Gon 1974; Dufresne et al. 1980). 'Late ' reflex responses can be caused by delayed short latency reflexes restimulated by continuing mechanical oscil lat ions in the musculoskele ta l system (Hagbarth et al. 1981; Eklund et al. 1982). To avoid this problem, we have taken care to apply test pulses which produced a smooth, non-oscillat- ing increment in joint torque.

The purpose of the experiments reported here was to establish details of the pattern of modula- tion of the stretch reflex during a normally paced movement, and to compare that pattern with the dynamic load encountered by the muscles during the movement. It was initially hypothesized that reflex responsiveness would be most elevated at those phases of a movement when the intrinsic load of the forearm (inertial, viscous and elastic forces generated during the motion itself) was the greatest. In fact, reflex responsiveness was elevated just before the load due to joint impedance was increasing.

Methods

Experiments were carried out on 11 normal subjects (7 male, 4 female) ranging in age from 20

0013-4649/83/0000-0000/$03.00 © 1983 Elsevier Scientific Publishers Ireland, Ltd.

688 W.A. MacKAY ET AL.

to 37. With the right arm snugly fitted into a manipulandum (Fig. 1), each continuously per- formed a rhythmic flexion-stop-extension-stop movement of the elbow in the horizontal plane, paced by a metronome. The cycle period was 2 or 4 sec and the movement amplitude 15 °. Visual feedback on a videomonitor allowed the subjects to see target end-points, although amplitude accu- racy was not stressed. Subjects were instructed to maintain the rhythm as strictly as possible, and not to correct for missing target points.

A torque motor coupled to the manipulandum delivered ramp test pulses of 50 or 70 msec dura- tion at different phases of the movement. A test pulse was given in alternate cycle periods, such that the pulse occurred in different phases on successive trials. For each phase tested, ramp dis- placements were delivered in either the flexion or extension direction to stretch the elbow extensor or flexor muscles respectively.

Two types of test pulse were used, either a feedback controlled, fixed change in forearm angu- lar velocity (Gottlieb and Agarwal 1980) or a constant ramp increment of elbow torque (Fig. 2). In the first case, the current angular velocity of the elbow was computed over the segment preceding that receiving the test pulse, and the stimulus ramp added to it. For the second type, a computer generated ramp and plateau signal (Fig. 8) directly drove the torque motor. The magnitude of the test pulse was kept small (about 3 N-m) in order that the overall performance not be seriously affected by it, and to avoid generating significant mechani- cal oscillations in the limb. The test pulses were of

9 1 I

TORQUE MOTOR

\

stra in gauges

Fig. 1. Diagram of manipulandum and torque motor assembly. Strain gauges are mounted on metal bar supporting the forearm cast. Elbow rotation was about the same axis as that of the

manipulandum.

A

D I S P L A C E M E N T

100 ms

B TORQUE

D I S P L A C E M E N T

<"-7"_

I I t I - - t . I - 4

-188 -58 0 50 100 msec 15fl

Fig. 2. Examples of stimulus test pulses delivered in different phases of the movement cycle. A: feedback-controlled change in angular velocity. Both directions of perturbation are il- lustrated. B: ramp and plateau increment of joint torque, causing displacement of forearm in extension direction. The rise in torque at the end of the plateau is partly due to the reflex response, and partly to the passive elastic resistance of the elbow to the continuing displacement.

the same order of magnitude as the dynamic load encountered by the muscles in performing the task. Subjects were instructed to totally ignore the test pulses and continue with the movement cycle. Sufficient time elapsed between perturbations such that depressive or facilitative effects of one re- sponse to the next (Aldridge and Stein 1982) were not present.

Being intent on maintaining the rhythmic accu- racy of the movement, subjects did not anticipate test pulses or react to them. Reflex responses to pulses delivered at randomly selected cycle seg- ments were similar to their counterparts delivered sequentially. Most subjects were unaware of the cessation of test pulses at the end of data collec- tion.

REFLEX MODULATION DURING MOVEMENT 689

Brachialis and triceps E M G signals were re- corded with surface electrodes placed over the distal belly of the brachialis and the lateral head of the triceps, respectively. In all subjects, virtually no activity was observed in the biceps for the task, partly because the forearm was fully or semi-pro- nated. E M G signals were fullwave rectified and partially integrated (time constant of 20 msec). Elbow torque was monitored by means of strain gauges mounted on the forearm support of the manipulandum (Fig. 1), and angular displacement by means of a potentiometer mounted on the torque motor. Torque, displacement and in- tegrated E M G signals were all digitized at a sam- piing rate of 100 Hz. Data were collected in seg- ments of 2.5 sec, initiated at a specific velocity of movement. In addition, a 250 msec segment of the E M G was digitized at 2 msec intervals following the onset of each test pulse.

In numerous experiments, inertial load was in- creased by attaching weights to the manipulan- dum. The weights added a moment of inertia about the elbow of 0.20 kg-m 2. The normal mo- ment of inertia of the forearm is approximately 0.06-0.07 kg-m 2 (Allum and Young 1976). In a few experiments an elastic load was added by attaching compliant tubing to the manipulandum.

Data analysis The magnitude of the late reflex was computed

by integrating the processed E M G signal over the poststimulus interval 60-115 msec. All latencies were measured from the onset of the torque incre- ment at the forearm which was at least 5 msec after the computer initiation of the test pulse (time 0 in Fig. 2B). Non-reflex E M G activity was then subtracted out. The latter was estimated by in- tegrating the mean EMG, measured in the absence of test pulses, over the identical interval with respect to the movement cycle. Because the input test pulses were maintained constant throughout the movement cycle, the reflex output provided a relative measure of reflex responsiveness (as dis- tinct from 'gain' which is dimensionless).

Estimation of intrinsic load of forearm Subjects were asked to relax their arm muscles

as the investigator moved the manipulandum in

the same flexion-extension cycle as the subject had actively performed. Active and passive trajectories were superimposed to ensure that they were as identical as possible. Since the experimenter moved the manipulandum at the torque motor coupling (see Fig. 1), the strain gauges monitored the resis- tance offered by the forearm to the movement pattern. Inertial and elastic forces were obvious in all subjects. With a reversal of sign, these forces were proportional to the same ones which the muscles would encounter in moving the limb.

The angular excursion at the elbow was too small (90 + 7.5 °) to introduce trigonometric non- linearities. In the active movement, it is possible that antagonist muscles would offer more elastic resistance than in the relaxed, passive situation. But in this paradigm there was little co-contrac- tion, so that muscles were generally working against a relaxed antagonist.

Results

Only very rarely did the ramp displacement of the forearm, used as a test stimulus, elicit an observable early reflex in the brachialis or triceps muscles, at a latency comparable to the tendon jerk, which is about 17 msec for elbow muscles (Chan et al. 1979). A late reflex, with a latency of generally 50-60 msec, was regularly elicited in all subjects, except when the arm was totally relaxed. Another component with a latency of 30-40 msec was sometimes seen at the onset of voluntary muscle activity. It may be identical with the ' M I ' of Lee and Tatton (1975), but it was usually of low amplitude, and could be largely eliminated by using a smooth torque stimulus and by asking the subject to perform the task with minimal effort. Only the late reflex was examined for modulation over the movement cycle.

In the experiments, the subjects continuously performed a flexion-stop-extension-stop elbow movement with a cycle period of 2 sec (or in some cases 4 sec). During this performance test pulses displaced the forearm in either direction to test the stretch reflexes in the extensor or flexor muscles. Every 50 msec phase of the cycle was tested in sequence. Reflex amplitudes are plotted in all

690 W.A. MacKAY ET AL.

h i s t og rams wi th re fe rence to the t ime of onse t of the to rque s t imulus .

The la te s t re tch reflex regular ly showed maxi -

m a l r e spons iveness at the onse t of a c o n t r a c t i o n of a muscle , whe t h e r the musc le was ac t ing to accel- e ra te the fo rea rm or dece lera te it (Figs. 3 a n d 4).

Th i s was obse rved in all subjects . O n c e v o l u n t a r y musc le ac t iv i ty s ta r ted to rise, however , the ampl i -

t ude of the evoked reflex decreased, qu ick ly fa l l ing

to base l ine levels (Figs. 3, 4 a n d 5). In later s tages of musc le shor t en ing , the reflex ga in cou ld in- crease aga in somewha t , b u t this was subjec t to grea t va r i a t i on b e t w e e n ind iv idua l s . The pa t t e rn of reflex m o d u l a t i o n para l l e led the sequence of v o l u n t a r y E M G act iv i ty except tha t the peaks of reflex r e spons iveness were a d v a n c e d in t ime on

r- "1

Movement Cycle: f7OF

7°E

2 sec I r- n ~ A 1 mV

r i i I i I i 1 200 ms

Fig. 3. Example of responses in brachialis muscle of one subject, to a complete series of test pulses delivered in each 50 msec phase of the movement cycle. The onset of the test pulse is marked by a vertical line. The trials were collected consecutively as displayed, except for intervening trials with test pulses in the opposite direction. The inset at the bottom right displays the 200 msec period after the test pulse on an expanded time scale for 2 trials. Note that voluntary brachialis activity in the deceleration phase of elbow extension was sporadic, appearing as occasional non-reflex peaks at the beginning or end of individual traces.

REFLEX M O D U L A T I O N D U R I N G M OVE M E NT 691

A

1 s e c

E M G : ext

"

REFLEX: ext 11

,~uOmO 10 °Uoo,

B

EMG: fl fl I

Fig. 4. Modulation of the late stretch reflex during the movement cycle for 2 subjects. The upper trace is 10 repetitions of the movement cycle without test pulses delivered. The corresponding velocity (V), acceleration (A) and torque (T) functions are displayed in B. The histograms of reflex magni tude give the mean of 3 trials for each 50 msec interval and are plotted with respect to the time of initiation of the test pulse. Maximum size of the integrated reflex response was 1 mV-msec for subject AW, 2 mV-msec for DP.

those of voluntary activity. Furthermore, the reflex was sometimes of relatively large amplitude in phases of the cycle when voluntary activity could potentially be present but in fact was not. For example, subject AW in Fig. 4 showed a peak in triceps stretch responsiveness during the decelera- tion phase of elbow flexion, but virtually no volun- tary activity in triceps muscle. If the viscoelastic properties of the arm had been insufficient to brake the movement, triceps activity would have been observed here. The same phenomenon occurred with brachialis muscle in subject DP (Fig. 4) during the deceleration phase of elbow extension.

When the movement cycle was tested at 40 individual phases, it was not possible to obtain more than 5 trials for each phase before subjects

displayed extreme signs of boredom. To more rigorously verify the main features of the modula- tion pattern, in some experiments only 5 phases were selected for testing. In each phase the reflex response was averaged over 20 trials. Again the same modulation pattern was observed. In Fig. 5 the modulation of the triceps reflex prior to and during elbow extension is illustrated in a subject with a strong late reflex (visible to some degree throughout most of the movement cycle). The amplitude of the voluntary discharge for the exten- sion was very small relative to the reflex response. Maximum reflex amplitude occurred when the re- flex coincided with the onset of voluntary dis- charge (Fig. 5c) and declined thereafter, although voluntary muscle activity continued. The identical modulation pattern was observed for both types of

692 W.A. MacKAY ET AL.

TORQUE

f!

a

b

C

~ ' ~ - - ~ - - e x t

d

e

I - I I - 1 8 8 - 5 8 8

I I I 58 IE8 i nse¢ 15E

a b c d e I I I I I

f M O V E M E N T CYCLE

8 1 2 BeG

Fig. 5. Modulation of the late reflex in triceps muscle of one subject, prior to and during the extension phase of the move- ment cycle. Both the brachialis (fl) and triceps (ext) EMG records are shown for each of the 5 tested phases a - e of the movement• Each record is the mean of 20 trials• The interval of voluntary triceps activity is indicated by a horizontal line under the trace of the movement cycle. The standard interval for late reflex integration is indicated by dotted lines for comparison with the histograms in other figures• Arrow: voluntary EMG activity.

stimulus employed, a controlled change in angular velocity or smooth increment in joint torque.

Onset of modulation The timing of the increase in reflex responsive-

ness relative to the onset of voluntary muscle activity was highly dependent both on the subject and the muscle tested. Of the 22 muscles examined (2 in each subject), 8 showed no advance of increased responsiveness on voluntary EMG. The other 14, however, showed a clear advance of 50-200 msec (measured to the nearest 50 msec). No consistent increase was observed for reflexes occurring more than 200 msec before the volun- tary discharge. The assessment of modulation onset was made from an analysis of series of single trials such as the records in Fig. 3, where the premature occurrence of voluntary activity (generating the false impression of early modulation) could be most clearly seen. The mode advance for brachialis was 150 msec (8 subjects) and for triceps, 100 msec (6 subjects).

Relation to intrinsic loading The total impedance opposing elbow motion

was complex, involving both the mechanical prop- erties of the forelimb itself, plus those of the manipulandum-torque motor apparatus• The chief restrictive force generated by the torque motor was a mild viscosity (seen in the torque trace, T, in Fig. 4B) monitored by strain gauges mounted on the manipulandum (Fig. 1). The intrinsic load of the forelimb was inertial and viscoelastic. This intrin- sic load was not monitored by the strain gauges as the task was performed.

The load of the forearm itself was measured by having subjects relax their arm in the manipulan- dum while the experimenter moved the manipu- landum asssembly from the torque motor end, in the same cycle as the subject had performed ac- tively. The torque monitored by the strain gauges was a measure of the intrinsic load of the forearm, and this was added to the viscous load of the motor (Fig. 6). The total load was then differenti- ated with respect to time. As can be seen in Fig. 6, reflex responsiveness bore more of a resemblance to the time derivative of total load than to the load itself.

The reflex generally increased prior to an in- crease in dynamic load with the peak in respon- siveness advanced about 100-150 msec on the

REFLEX MODULATION DURING MOVEMENT 693

T : m o t o r

forearm ..... ~ '

DP BM

I 15 Nm/s

[i ! i ,o,o , , ,,i,,,l,,,Bo,,,,, .... lt, i i ~ i : ~ ~! OHm, . . . . . . . . . . . . . . . o0ooo ] o . . : ~ . ! : . . . . . . , . . . . . , . - , , - . . o ~ , , , ~ u

v . . . . . . . . . . . . . . ,

e

Fig. 6. Modulation of the late reflex in triceps and brachialis muscles for 2 subjects, shown with the coincident variation in joint load and its derivative with respect to time. Total joint torque (T) was estimated from 2 components, the load of the torque motor-manipulandum and that of the forearm itself, measured separately. Histograms of the late reflex magnitude (brachialis up, triceps down) give the mean of 5 trials for each 50 msec interval.

peak rate of rise of the load. By the time the load was increasing the fastest, reflex responsiveness was declining or had even reached baseline levels. These features are evident in the flexor stretch reflex amplitudes of both subjects in Fig. 6, in relation to increasing loads at the time of flexion acceleration and extension deceleration.

The tonic and unusua l ly s t rong tr iceps responsiveness in subject DP (Fig. 6) seemed to have obviated any need for an increase in reflex gain prior to extension acceleration. But it high- lighted the sharp drop in reflex gain at the mo- ments when extensor muscle load was increasing the fastest.

Altered loading conditions Reflex responsiveness was also tested under two

addit ional conditions, an increased moment of in- ertia (by adding mass to the manipulandum) or an increased elastic load (compliant rubber tubing).

The aim of this experiment was to check if the pat tern of reflex modula t ion would still more closely resemble the time derivative of joint load than the load itself. In each case, the external load outweighed the intrinsic load of the forearm so that the moni tored torque gave a reasonable mea- sure of the joint load. In general, reflex responsive- ness was elevated prior to and during periods of increasing joint load (Fig. 7). This elevated re- sponsiveness was not maintained when joint load s topped increasing even if the latter remained high.

Latency shifts Changes in reflex magni tude over the move-

ment cycle could be caused by systematic fluctua- tions of threshold rather than by gain changes in the reflex pathways. Threshold changes by them- selves would be manifested as changes in reflex latency, latency decreasing as reflex ampli tude in- creases. In some cases reflex latency did change,

694 W.A. MacKAY ET AL.

INERTIAL LOAD L ~ SPRING L O A D ~ . //'/~ ' ~ ,X~

E M G : e x t

fl

REFLEX: e x t

I . . . . . l i , .

" I , l i , , [ i l l i . , , , l l l I,I I , , , , , , , , , , , . ,.,il

o,$ mV

ti lalt '

l ~V-ms

1 sec

~ ' ~ 2 ~ - ~ ~ , . ~ . . ~ \ .

Fig. 7. Modulation of the late reflex in 2 subjects when either a spring load was applied to oppose elbow flexion (A) or the inertial load was increased (B and C). Histograms of late reflex magnitude give mean of 2 trials for each 50 msec interval of the movement cycle.

but the changes in latency were not consistently related to reflex magnitude. Moreover, they usu-

Motor Input

L2L ~ _ = , Brachia l is ~._ - ~ , , 7 ~ • ~ i

Tr iceps ~ " ~ [ ~ ' ~ ' " ~ - - - J I -- 1

1 O0 msec

V

Fig. 8. Modulation of late reflex in triceps of one subject, prior to and during elbow extension. The 5 superimposed oscillo- grams were collected in sequence, with the largest reflex re- sponse occurring at the onset of voluntary activity. The 2 smallest responses were elicited 400 msec before and 500 msec after the onset of voluntary discharge. The 2 middle responses were elicited 200 msec before and 200 msec after the onset of voluntary discharge. (Analogous to Fig. 5 but different subject).

ally seemed to be due to the appearance of an early component of the reflex at the time of volun- tary EMG activity. The major, late, component maintained a remarkably constant latency, al- though its rise time and rate of increase changed dramatically from one phase of the cycle to another (Fig. 8). Fig. 5 shows a latency shift of the late reflex, decreasing by about 7 msec between its elicitation at 400 and at 200 msec prior to the voluntary discharge (phases 'a' and 'b'). Thereafter, the reflex latency remained constant up to and throughout the period of active movement, al- though reflex magnitude altered considerably.

Discussion

In this study very pronounced modulation of the late stretch reflex was observed. The general pattern of modulation is similar to that reported by Dufresne et al. (1980) for elbow movement, but

REFLEX MODULATION DURING MOVEMENT 695

is demonstrated here in more detail. In addition we have shown that in some subjects the increase in reflex responsiveness precedes voluntary E M G onset by up to 200 msec. Once voluntary muscle activity rises, reflex responsiveness drops markedly, as shown by Gottlieb and Agarwal (1980) in soleus muscle.

Although audiospinal potentiation can be ob- served in the E M G (Rossignol and Melvill Jones 1976), the metronome clicks used here were too quiet to have a direct influence on reflex gain. Moreover, subjects invariably paced themselves so that metronome clicks occurred during the move- ments, near peak velocity, and not at movement onsets where reflex responses were most potenti- ated.

Mechanism of gain modulation Changes in fusimotor tone are the most obvious

mechanism of gain modulation of the stretch re- flex. Lund et al. (1979) have reported an increase of gamma efferent discharge starting about 200 msec before the onset of muscle tension (masseter muscle) and reaching a maximum at the start of the dynamic phase of contraction. This pattern resembles that of the reflex modulation observed here, except for the rapid decrease which follows the rise. Fusimotor discharge does not stop until the decline of alpha-motoneuron discharge (Lund et al. 1979). Initial drops in reflex latency prior to the initiation of movement (Fig. 5) could have been due to the onset of fusimotor discharge, for the latency then remained fixed until the end of the movement. But Burke et al. (1980) have dem- onstrated that no increase in spindle responsive- ness occurs under conditions which greatly aug- ment reflex responsiveness in the pretibial muscles. The fusimotor system is, therefore, unlikely to be the principle source of the modulation pattern which we observed.

Alpha-motoneuron excitability, or net level of membrane depolarization, would also modulate reflex responsiveness roughly in proportion to E M G amplitude. This factor again is probably only a partial cause of the modulation pattern for the following reasons. First, in several subjects most of the recorded motor unit population failed to discharge over the movement cycle (Figs. 5 and

8) and yet modulation of the elicited reflex was as strong as in subjects with prominent voluntary discharge (Fig. 3). Secondly, the pattern of modu- lation of the reflex differed significantly as a func- tion of time from the voluntary E M G activity. In keeping with this difference, the effect of motoneuron excitability on the transmission of afferent volleys is known to vary from the modu- lation of the late reflex reported here. Both tendon jerk and H-reflexes increase in amplitude prior to a prompt voluntary movement (Kots 1977; Pier- rot-Deseilligny et al. 1971). The increase starts abruptly, at 60 msec before the voluntary EMG burst and augments continuously as the interval between the reflex and voluntary burst narrows. The reflex is maximal when it is elicited during the voluntary E M G activity, i.e., when motoneuron excitability is at its peak (Kots 1977). In this study, the increase in late stretch reflex responsive- ness started when the reflex occurred anywhere from 0 to 200 msec prior to the onset of the voluntary E M G burst, depending on the subject and the muscle tested. Invariably reflex respon- siveness decreased during voluntary E M G activity. It appears that the changes in reflex responsive- ness are at least in part due to gain changes within the reflex pathways, somewhere prior to the motoneuron.

The relative decline in the reflex as voluntary E M G rose in amplitude cannot be ascribed to occlusion or saturation within the motor nuclei. The E M G levels were always in a low range that facilitated the late reflex in isometric contractions. The drop in reflex responsiveness was not accom- panied by an increase in latency, and in some subjects the decline in responsiveness preceded and shortening of the muscle. While mechanical 'unloading ' of the muscle spindles may contribute to the observed decrease in the reflex, central gain changes probably play a role as well.

Function of the stretch reflex Whether or not the reflex pathways are actually

used during the period of increased responsiveness is a question which these experiments did not address, although the test pulse was of a magni- tude comparable to the actual dynamic loads expe- rienced by the elbow muscles during the task. The

696 W.A. MacKAY ET AL.

data of Burke et al. (1978) indicate that muscle spindle discharge is increased at the very onset of muscle contraction, a time when reflex responsive- ness is high. Spindle discharge slows down as the muscle starts shortening rapidly. At the onset of movement, muscle contraction is isometric due to inertial loading. An isometric contraction elicits increased spindle discharge in proportion to the torque produced (Vallbo 1974). Because of rela- tively high reflex responsiveness during the isomet- ric phase, significant reinforcing tension would be generated in the agonist muscle to help overcome the inertial load.

That the reflex does make a significant contri- bution to dynamic load compensation is further implicated by the drop in responsiveness which occurs immediately after the rise. With a con- tinued high gain, the reflex would produce increas- ingly higher muscle tensions as the dynamic load continued to increase. The effect would be to continuously amplify the acceleration phase of contraction, resulting in overshooting. The tem- poral advance of the reflex modulation pattern on that of the derivative of total joint load (Fig. 7) also indicates a strong potential contribution of the reflex to reinforcing agonist muscle tension at the time of the peak rate of rise of impedance. The 100-150 msec advance in time is required for reflex latency and the build-up of resultant con- tractile force.

The pattern of late reflex modulation is not consistent with the 'stiffness servo' hypothesis which predicts that stretch reflex gain should re- main relatively constant over the course of a movement (Houk 1979). But this hypothesis was formulated with reference to the early myotatic reflex, and therefore was not directly tested in these experiments. If the late reflex serves any mechanical function, it must be in relation to the compensation for expected increases in limb dy- namic load because it is immediately prior to a dynamic load that reflex responsiveness is most elevated.

Our results imply that the nervous system con- tains an internal model of the intrinsic loads, which will be routinely encountered in a specific movement, and regulates the gain of the stretch reflex on the basis of that model. Reflex output

may normally serve to produce a step increase in muscle stiffness (Kwan et al. 1979; Cooker et al. 1980; Akazawa et al. 1982) at the onset of a movement, so that dynamic loads are effectively resisted, and the movement is promptly initiated. Smith et al. (1972) have shown that the rapid initiation of an elbow extension is significantly impaired by fusimotor blockade of the agonist causing a depression of its stretch reflex.

The late reflex discussed here may be identical to the 'action tonic stretch reflex' of Neilson and Neilson (1978) which appears during voluntary contractions and serves to damp transient oscil- lations associated with the natural resonant fre- quencies of a limb. In other words, the reflex counterbalances forces imposed by the elastic and inertial properties of the limb. Similarly, Ghez and Martin (1982) have concluded that the antagonist burst associated with movement deceleration is essentially a stretch reflex preventing terminal oscillation of the limb mass-spring. The pattern of reflex modulation shown here supports this hy- pothesis in that antagonist reflex responsiveness is enhanced at about the time of peak velocity.

With regard to the initial agonist burst, even in very fast movements, the last half of the burst can be modified by spindle afferent input (Brown and Cooke 1981). In normal movements, however, our results show that it is the initial phase of agonist activity that is most adjustable by the stretch reflex. Because of the relatively high gain of the stretch reflex at the very onset of movement, agonist activity can be adjusted to the evolving dynamic load, which in compound joint move- ments will include interaction torques (Hollerbach and Flash 1982). It should be noted, however, that the late reflex may deliver a more 'quantal' than finely graded output (Chan and Kearney 1982).

Summaff

Small torque pulses were delivered to the fore- arm in order to test the stretch reflex of the brachialis and triceps arm muscles in 11 normal subjects performing a cyclic movement about the elbow in the horizontal plane. The flexion-exten- sion movement was paced by a metronome and

REFLEX MODULATION DURING MOVEMENT 697

pe r fo rmed under var ious load ing condi t ions . Re- flexes for each muscle were tested ei ther in each 50 msec segment of the 2 sec cycle per iod, or in a smal ler number of selected phases. A late reflex, appea r ing at a l a tency of about 60 msec (measured f rom the onset of the torque increment) , was mod- u la ted extensively dur ing the movemen t cycle. The amp l i t ude of the late reflex increased marked ly at the onset of a muscle cont rac t ion . In many of the subjects reflex responsiveness began to increase as ear ly as 200 msec pr ior to the onset of vo lun ta ry muscle activity. Peak reflex responses were el ici ted by s t imuli de l ivered 100-150 msec pr io r to the peak rate of increase of dynamic load ( composed of inert ial , viscous and elast ic forces). The increase in responsiveness was fol lowed by a d rop which was general ly co inc ident in t ime with the peak rate of increase of the load oppos ing muscle cont rac- t ion. The modu la t i on of the late reflex is ap- p ropr i a t e ly t imed for ref lex-genera ted tension to help counterac t dynamic loads, intr insic to the movement .

R~sum~

Modulation du rOflexe d'Otirement pendant un mouvement cyclique de l'avant-bras

De pet i tes tors ions (d 'une dur6e de 50 ~ 120 msec) ont 6t6 utilis6es pour 6tudier le r6flexe d '6 t i r ement des muscles brachia l i s et t r iceps du bras, sur 11 sujets no rmaux qui ex6cutaient un m o u v e m e n t ry thmique de l ' avan t -b ras dans le p lan hor izonta l . Le cycle f lexion-extension 6tait r6g16 p a r un m6tronome, et ex6cut6 sous diff6rentes charges. Pour chaque muscle, les r6flexes 6taient test6s soit pour chaque in terval le de 50 msec de la p6r iode du cycle de 2 sec, soit pour un n o m b r e plus res t re int de p6r iodes s6lectionn6es. U n r6flexe tardif , d 'une la tence d ' env i ron 60 msec (mesur6e pa r t i r du d6but de l ' augmen ta t i on du couple moteur) , subissai t une modu la t i on impor t an t e pen- dan t le cycle. L ' a m p l i t u d e du r6flexe ta rd i f augmen ta i t no t ab l emen t au d6but d ' une cont rac- t ion musculaire . Sur de n o m b r e u x sujets, la modu- la t ion du r6flexe commenqai t avec une la tence aussi br6ve que 200 msec avant le d6but de I ' E M G

volontaire . Les r6ponses r6flexes maximales 6taient 6voqu6es p a r des p e r t u r b a t i o n s d6clench6es 100-150 msec avant que l ' augmen ta t ion de la charge d y n a m i q u e (compos6e des forces d ' iner t ie , de viscosit6 et d'61asticit6) n 'a t te igne son maxi- mum. L ' augmen ta t i on des r6ponses r6flexes 6tait suivie d ' une chute qui cdincidai t g6n6ralement avec le m a x i m u m d ' a u g m e n t a t i o n de la force oppos6e la con t rac t ion musculaire . La modu la t i on du r6flexe ta rd i f se situe dans le t emps de telle sorte que la force engendr6e pa r ce r6flexe puisse con- t r eba lance r les forces dynamiques intr ins6ques au mouvement .

This study would never have been completed without the able computer programming support of Mr. Hoi Nguyen-Huu. Mr. Donald Crammond and Mr. Doug Pon also assisted in the experiments.

The project was supported by the Medical Research Coun- cil of Canada, Grant MA-7092.

References

Akazawa, K., Aldridge, J.W., Steeves, J.D. and Stein, R.B. Modulation of stretch reflexes during locomotion in the mesencephalic cat. J. Physiol. (Lond.), 1982, 329: 553-567.

Aldridge, J.W. and Stein, R.B. Nonlinear properties of stretch reflex studied in the decerebrate cat. J. Neurophysiol., 1982, 47: 179-192.

Allum, J.H.J. and Young, L.R. The relaxed oscillation tech- nique for the determination of the moment of inertia of limb segments. Biomechanics, 1976, 9: 21-25.

Bonnet, M. Comparison of monosynaptic tendon reflexes dur- ing preparation for ballistic or ramp movement. Electroen- ceph. clin. Neurophysiol., 1981, 51 : 353- 362.

Bonnet, M. and Requin, J. Long loop and spinal reflexes in man during preparation for intended directional hand movements. J. Neurosci., 1982, 2: 90-96.

Brown, S.H. and Cooke, J.D. Responses to force perturbations preceding voluntary human arm movements. Brain Res., 1981, 220: 350-355.

Burke, D., Hagbarth, K.E. and L6fstedt, L. Muscle spindle activity in man during shortening and lengthening contrac- tions. J. Physiol. (Lond.), 1978, 277: 131-142.

Burke, D., McKean, B., Skuse, N.F. and Westerman, R.A. Anticipation and fusimotor activity in preparation for a voluntary contraction. J. Physiol. (Lond.), 1980, 306: 337-348.

Chan, C.W.Y. and Kearney, R.E. Is the functional stretch response servo controlled or preprogrammed? Electroen- ceph. clin. Neurophysiol., 1982, 53: 310-324.

Chan, C,W.Y., Melvill Jones, G., Kearney, R.E. and Watt, D.G.D. The 'late' electromyographic response to limb dis-

698 W.A. MacKAY ET AL.

placement in man. I. Evidence for supraspinal contribution. Electroenceph. clin. Neurophysiol., 1979, 46: 173-181.

Cooker, H.S., Larson, C.R. and Luschei, E.S. Evidence that the human jaw stretch reflex increases the resistance of the mandible to small displacements. J. Physiol. (Lond.), 1980, 308: 61-78.

Desmedt, J.E. and Godaux, E. Ballistic skilled movements: load compensation and patterning of the motor commands. In: J.E. Desmedt (Ed.), Cerebral Motor Control in Man: Long Loop Mechanisms. Prog. clin. Neurophysiol., Vol. 4. Karger, Basel, 1978: 21-55.

Dufresne, J.R., Soechting, J.F. and Terzuolo, C.A. Modulation of the myotatic reflex gain in man during intentional move- ments. Brain Res., 1980, 193: 67-84.

Eklund, G., Hagbarth, K.E., Hagglund, J.V. and Wallin, E.U. Mechanical oscillations contributing to the segmentation of the reflex electromyogram response to stretching human muscles. J. Physiol. (Lond.), 1982, 326: 65-77.

Ghez, C. and Martin, J.H. The control of rapid limb movement in the cat. III. Agonist-antagonist coupling. Exp. Brain Res., 1982, 45: 115-125.

Gottlieb, G.L. and Agarwal, G.C. Response to sudden torques about ankle in man. III. Suppression of stretch-evoked responses during phasic contraction. J. Neurophysiol., 1980, 44: 233-246.

Hagbarth, K.E., Hagglund, J.V., Wallin, E.U. and Young, R.R. Grouped spindle and electromyographic responses to abrupt wrist extension movements in man. J. Physiol. (Lond.), 1981, 312: 81-96.

Hallett, M., Bielawski, M. and Marsden, C,D. Behavior of the long-latency stretch reflex prior to voluntary movement. Brain Res., 1981, 219: 178-185.

Hollerbach, J.M. and Flash, T. Dynamic interactions between limb segments during planar arm movement. Biol. Cyber- net., 1982, 44: 67-77.

Houk, J.C. Regulation of stiffness by skeletomotor reflexes. Ann. Rev. Physiol., 1979, 41: 99-114.

Kots, Y.M. The Organization of Voluntary Movement. Plenum, New York, 1977.

Kwan, H.C., Murphy, J.T. and Repeck, M.W. Control of stiffness by the medium latency electromyographic response to limb perturbation. Canad. J. Physiol. Pharmacol., 1979, 57: 277-285.

Lee, R.G. and Tatton, W.G. Motor responses to sudden limb displacements in primates with specific CNS lesions and in human patients with motor system disorders. Canad. J. neurol. Sci., 1975, 2: 285-293.

Lund, J.P., Smith, A.M., Sessle, B.J. and Murakami, T. Activity of trigeminal a- and y-motoneurons and muscle afferents during performance of a biting task. J. Neurophysiol., 1979, 42: 710-725.

Neilson, P.D. and Neilson, M.D. The role of action reflexes in the damping of mechanical oscillations. Brain Res., 1978, 142: 439-453.

Pierrot-Deseilligny, E., Lacert, P. et Cathala, H.P. Amplitude et variabilit6 des r6flexes monosynaptiques avant un mouve- ment volontaire. Physiol. Behav., 1971, 7: 495-508.

Rossignol, S. and Melvill Jones, G. Audio-spinal influence in man studied by the H-reflex and its possible role on rhythmic movements synchronized to sound. Electroenceph. clin. Neurophysiol., 1976, 41: 83-92.

Smith, J.L., Roberts, E.M. and Atkins, E. Fusimotor neuron block and voluntary arm movement in man. Amer. J. phys. Med., 1972, 51: 225-239.

Soechting, J.F., Dufresne, J.R. and Lacquaniti, F. Time-varying properties of myotatic response in man during some simple motor tasks. J. Neurophysiol., 1981, 46:1226 1243.

Vallbo, A.B. Human muscle spindle discharge during isometric voluntary contractions. Amplitude relations between spin- dle frequency and torque. Acta physiol, scan& 1974, 90: 319-336.

Wieneke, G.H. and Denier van der Gon, J.J. Variations in the output impedance of the human motor system. Kybernetik, 1974, 15: 159-178.