Exp Brain Res (1986) 65:135-144

Experimental BrainResearch 9 Springer-Verlag 1986

Responses of mono- and bi-articular muscles to load perturbations of the h u m a n arm F. Lacquaniti* and J.F. Soechting Laboratory of Neurophysiology, Department of Physiology, University of Minnesota, 6-255 Millard Hall, 435 Delaware Street SE, Minneapolis, MN 55455, USA

Summary. We studied the behavior of muscles acting synergistically in elbow flexion in response to load perturbations. The perturbations were applied either proximally or distally to the elbow joint and consisted of single pulses or steps of torque and of pseudorandom sequences of torque pulses. They produced changes in angular position and torque at both the shoulder and elbow joints. The electromyographic (EMG) responses evoked in biceps, brachio-radialis and brachialis muscles were different when elbow and shoulder motion was in the same direction and when the two angular motions were oppositely directed. For example, elbow extension resulted both when a downward force perturbation was applied to the forearm as well as when a posteriorly directed force applied to the upper arm was released. Elbow flexors were activated at a short latency only in the former case and not in the latter. The modulation of EMG activity in elbow flexors evoked by the perturbations was related to the global motion of the limb, including the angular motions at both the shoulder and elbow joints. The time course of the EMG responses in biceps, which acts on both joints, differed from that of brachio-radialis and brachialis muscles, which act only at the elbow. The results are discussed in the context of the possible mechanisms responsible for the muscle responses to the perturbations. Key words: Load perturbations - Multijointed limb Feedback control

Introduction In two previous papers (Lacquaniti and Soechting 1984; 1986) we have begun to describe EMG * Present address: Istituto di Fisiologia dei Centri Nervosi, CNR, 1-20100 Milano, Italy Offprint requests to: J.F. Soechting (address see above)

responses of elbow muscles to load perturbations applied either to the forearm or to the upper arm when limb motion was unconstrained. Such perturbations result in simultaneous angular motion at both~ the shoulder and elbow since these joints are dynamically coupled. For example, a force applied to the upper arm and acting in the posterior direction will produce backward extension at the shoulder and flexion at the elbow. Under such conditions, the EMG responses of biceps and triceps to the perturbations were not related to the angular motion at the elbow joint, but depended instead on both elbow and shoulder motion. Indeed, perturbations which led to elbow flexion could result in either increased activation or inactivation of biceps. Both biceps and triceps are bi-articular muscles spanning the shoulder joint as well as the elbow and thus biceps is stretched by extension at the shoulder as well as at the elbow. Based on available biomechanical data (Fick 1911) we concluded that, nevertheless, the overall length of biceps did decrease in at least some of the cases in which biceps activation was observed. Consequently, we were unable to account for our findings by postulating a negative feedback of muscle length or its time derivatives. Instead, biceps responses were found to be correlated with the net changes in torque at the elbow joint resulting from the applied perturbations. Net torque at the elbow is the resultant of all external and internal moments, the latter due to active and passive (viscoelastic) forces of muscle acting at the elbow. Elbow torque is related in a non-linear manner to the angular motion and its derivatives at the elbow and at the shoulder joint (cf. Hollerbach and Flash 1982). In this paper we address the question whether or not the observations concerning the dependence of biceps responses on angular motion at the shoulder and elbow joints generalize to muscles whose action

136

/ Anterior Deltoid

t

Angle Shoulder Torque

Flexion Extension

Biceps

Brachioradialis Angle Elbow Torque

T

Pronation Pronator Teres 0

200

400 ms

L

T

~

~

Flexion Extension

)

200

Triceps 400 ms

0

200

400 ms

Fig. 1A-C. Responses to perturbations to the forearm (A) and to the upper arm (B, C). Force perturbations were applied in the directions schematically depicted in the inserts of each panel. They consisted of a pulse 50 ms in duration starting at time 0. The traces correspond to rectified EMG activity of anterior deltoid, shoulder angle (O), shoulder torque (Ts), biceps, brachio-radialis, elbow angle (~), elbow torque (T~), wrist angle (~2), pronator teres (A, B) or triceps (C). In each panel, ensemble averages of 20 trials are shown. The data of panels A and B are from the same experiment. Forearm was initially mid-pronated in all conditions. Scales (per division) are: 10~ (0, 9 and ~), 10 N-m (T~ in A, C), 20 N-m (Ts in B), 5 N-m (T~ in A, B), 2.5 N-m (Te in C), 75 ~tV (deltoid and pronator teres in A, B), 50 ~tV (biceps in A, B and deltoid in C), 40 ~tV (brachio-radialis in A, B), 15 ~tV (biceps, brachio-radialis and triceps in C). The dashed lines mark t = 100 ms after pulse onset

is restricted to the elbow joint, such as brachioradialis and brachialis. We shall present results which suggest that even the responses of mono-articular elbow flexors depend not only on elbow motion but also on shoulder motion. Their time course, however, can differ from that of the biceps responses.

Methods The experimental setup has been described in detail elsewhere (Lacquaniti and Soechting 1984; 1986). Subjects were seated with their right upper arm approximately vertical and their forearm approximately horizontal. Flexion-extension at the elbow and prono-supination at the wrist were measured goniometrically (Lacquaniti and Soechting 1982). The angle of flexion-extension at the shoulder was derived trigonometrically from the position of two points on the upper arm, as recorded by means of an ultrasound system (Soechting 1984). Electromyographic activity was recorded by means of surface electrodes. Force perturbations were applied to the forearm or to the upper arm by means of a torque motor and a cable attached to the limb. Two different waveforms were used as input to the motor: 1) single pulses of 50 ms duration or 2) pseudo-random trains of pulses consisting of a 7th order msequence with 127 binary elements each of 40 ms duration (Davies 1970; O'Leary and Honrubia 1975; Dufresne et al. 1978). The sampling rate was 500/s for the E M G activity and 125/s for the kinematic data.

Experimental protocol In each experiment perturbations were applied to the forearm and to the upper arm. In general (14 experiments including 7 subjects), the forces applied to the forearm were directed downwards while those applied to the upper arm were directed posteriorly. Single pulses of torque were used in 11 such experiments and pseudorandom perturbations in the other three. In some instances (three experiments with three subjects) another type of perturbation was also used. The subject opposed a constant force directed posteriorly to the upper arm; the load was then released in a step-wise fashion. In 4 experiments (4 subjects) increased tonic activation of elbow muscles was achieved by asking subjects to hold a 1 kg weight in their hand or to stiffen their limb by co-contracting flexors and extensors. In all experiments, the subjects were instructed to resist the perturbations. A constant pre-load was exerted by the torque motor in the direction of application of the force perturbation. Blocks of trials from each experimental set were randomly interspersed. Ten to twenty trials entered in the ensemble average of each set, aligned with the onset of the perturbation.

Data analysis Ensemble averages were constructed from each set of data for the shoulder, elbow and wrist angles and the E M G activities after fullwave rectification. Kinematic data were smoothed and differentiated to obtain angular velocity and acceleration of shoulder and elbow and the torques acting at the shoulder and at the elbow were

137

-s

100%

J_

50%

0

i 3 [-J 40

I

t 80

I

i

I

I

Fig. 2A, B. Time histograms of the EMG responses significantly different from baseline. The time bins (4 ms) in which biceps and brachio-radialis EMG activity after pulse onset was above or below the 95% confidence limits of the baseline are plotted for all sets of data (N = 17 for A, N = 46 for B). In A the pulse perturbations were applied to the forearm, in B to the upper arm

25% 100% A

l

q

_

50%

I

0 40

i 80

I

I

I

I

12

J 4d0 ~

~ '

1120

'

m~

25%

computed according to Newtonian mechanics (Eqs. (1) and (2) of Lacquaniti and Soechting 1986). The EMG responses to single pulse perturbations were quantitatively analyzed in the following way. For each ensemble average, a baseline was defined as the mean EMG activity over the . 100 ms preceding the pulse and its 95% confidence limits were computed. Response latency was defined by the times at which EMG activity exceeded these limits. The mean amplitude of the EMG responses to the perturbations was computed over the first 100 ms after pulse onset and the baseline was subtracted from the calculated value. For the experiments in which pseudo-random perturbations were used, we calculated the average response to a single pulse of torque of 40 ms duration by utilizing a slight modification of standard cross-correlation techniques (cf. Dufresne et al. 1978; Soechting et al. 1981). Generally, when pseudo-random binary sequences are used to approximate a white noise input, the temporal resolution of the impulse .response is equal to the switching time of the sequence (40 ms in our case). In order to increase the temporal resolution of the impulse responses, we assumed that the input consisted of a two-stage process: 1) a sequence of pulses having an amplitude + 1, a duration of 4 ms (the sampling interval of the rectified EMG data) and occurring at 40 ms intervals and 2) the input stage to the torque motor whose impulse response was a pulse of 40 ms duration. Each of the parameters (EMG data, angles and torques) were cross-correlated with the first stage of the process. (Under this procedure, the auto~orrelation of the first stage is not perfectly flat: at zero lag its value is 127, at all other lags it is zero except for lags at integer multiples of 40 ms, where the auto-correlation takes on the value

-1.).

forearm at time 0. Since the force was distal to the elbow and shoulder joints, it resulted in an initial change of both shoulder and elbow torques in the extensor direction, leading to extension at both joints. (An initial supination of the hand also occurred, probably due to a torsional moment exerted by the perturbation about the forearm longitudinal axis). The EMG activities of anterior deltoid, biceps and brachio-radialis muscles all increased above baseline at about 50 ms after pulse onset. Latencies of the responses were computed as the time at which EMG activity exceeded the 95% confidence limits of the baseline. The latencies averaged over all experiments were 60 + 13 ms (N = 17) for anterior deltoid, 47 + 9 ms for biceps and 49 + 11 ms for brachioradialis 1. (The distribution in the response latencies of biceps and brachio-radialis in different experiments can also be appreciated in Fig. 2A, which shows the times at which their activity differed from baseline levels). Activity in all three muscles peaked at about 100 ms and was followed by a depression at about 160 ms. This time course agrees well with that previously reported for anterior deltoid and biceps responses under similar experimental conditions (Lacquaniti and Soechting 1986). Brachialis, when reliably recorded, gave responses qualitatively simi-

Results

Responses to single pulse perturbations Figure 1A illustrates typical results obtained when a downward force of 50 ms duration is applied to the

1 The latencies of the EMG responses were longer than those usually reported for biceps when motion is restricted to the elbow joint only (cf. Dufresne 1977). This could be due to the fact that the force actually applied to the arm did not rise instantaneously, given the time constant of the torque motor and the compliance between the motor and the limb

138

lar to those of biceps and brachio-radialis. Pronator teres, instead, was only inconsistently activated and at generally longer latencies. Figures 1B, C show results obtained when a posteriorly directed force was applied to the upper arm. Since the force was proximal to the elbow, it exerted an extensor torque at the shoulder. At the same time, the elbow flexed due to the dynamic coupling between the two joints (Lacquaniti and Soechting 1986)2. Even though no external torque was applied to the elbow, elbow torque also changed initially in the extensor direction since elbow flexion resulting from the applied perturbation stretched elbow extensors and shortened elbow flexors. The passive visco-elastic restoring forces developed by these muscles are responsible for a net change in the extensor direction of elbow torque. Both anterior deltoid and biceps were consistently activated at latencies comparable to those obtained when the force was applied to the forearm. Anterior deltoid EMG responses had a mean latency of 45 + 11 ms (N = 46) and peaked at 93 + 14 ms, while those of biceps had a latency of 58 + 14 ms and peaked at the same time. Changes in brachio-radialis activity did not parallel those in biceps when the force was applied to the upper arm, in contrast with the pattern which resulted when the force was applied to the forearm (Fig. 1A). In particular, consistent statistically significant changes in brachioradialis activity occurred much later than those in biceps, with an activation at a latency of 95 + 18 ms and a peak at 130 + 17 ms. This late activation was sometimes preceded by a smaller enhancement (Fig. 1B) or depression (Fig. 1C) of brachio-radialis activity. To estimate the overall direction and amplitude of the changes in brachio-radialis EMG activity we computed its mean amplitude over the first 100 ms following pulse onset and subtracted the baseline activity. In 80% of the trials, this measure was positive and amounted to 17% of the baseline amplitude, while in the remaining 20% it was negative ( - 6 % ) . A similar procedure for biceps gave an average value of 43%. Brachialis when reliably recorded gave responses qualitatively similar to those of brachio-radialis, whereas neither pronator teres nor triceps showed appreciable modulation. Results qualitatively comparable to those described (Fig. 1B, C) were obtained by producing shoulder extension and elbow flexion with a pertur2 In Fig. 1B one can note that a transient supination of the h a n d followed the perturbation on the u p p e r arm. This supination always lagged behind the initial flexion of the elbow by 40-50 ms. Since no external force acted on the forearm, supination could be due to the activation of elbow muscles

bation on the forearm rather than on the upper arm. In two experiments a force was applied to the forearm with a line of action which passed just below the elbow center of rotation, exerting extensor torque at the shoulder and only a negligible amount of external torque at the elbow (Lacquaniti and Soechting 1986). Since there was some variability among experiments in the amplitude and waveform of the EMG responses to the perturbations, we performed a statistical analysis to extract their significant features. To this end, the time bins over which EMG activity remained outside the 95% confidence limits of the baseline were computed for each ensemble average. The histograms of Fig. 2 were constructed using the results from all experiments in which a downward force was applied to the forearm (Fig. 2A) and for posteriorly directed forces on the upper arm (Fig. 2B). In the former case, the distribution of the changes in EMG activity was similar in biceps and brachio-radialis, their activity remaining significantly above baseline from 48 to 120 ms after pulse onset in more than 50% of the case. A similiar distribution also characterized the changes in biceps activity when the force was applied to the upper arm (Fig. 2B). In contrast, in the majority of cases brachio-radialis activity exceeded the baseline only from 96 to until after 160 ms. Prior to 80 ms, a very small percentage of the trials (less than 10%) showed a statistically significant depression in brachio-radialis activity while in a larger percentage of cases (about 25%) it increased above baseline in the interval 40 to 80 ms. The results presented so far show that when a force was applied to the forearm and angular motion at the shoulder and elbow was in the same direction, biceps and brachio-radialis activity changed in parallel with a latency of about 40 ms. When the angular motion at the two joints was in opposite directions, biceps and brachio-radialis were not strictly coactivated. In particular, brachio-radialis activity deviated only infrequently from baseline in the interval 40 to 80 ms after pulse onset, sometimes increasing and sometimes decreasing. In the following, we shall present results which indicate that the response of brachio-radialis to perturbations applied to the upper arm is most easily interpreted as consisting of two components with different latencies. 80 to 90 ms after pulse onset, brachio-radialis was consistently activated when the upper arm was extended by the perturbation (Figs. 1 and 2) and we shall term responses with latencies greater than 80 ms as "late" responses. Those with latencies of less than 60 ms will be termed "early". As stated above, when a pulse perturbation was applied to the upper arm it was difficult to obtain

139

Anterior Deltoid

_

Angle Shoulder Torque

_

------~

Flexion Extension

Biceps Brachioradialis

Fig. 3A, B. Force perturbationson the upper arm in normalconditions(A) and in the presence of highmuscletone (B). A pulse of force directedposteriorlywas appliedto the upper arm. In B, the subjectco-contractedhis elbow muscles. Data are from the same experiment. Scales (per division)are: 10~(0, alp),10 N-m (T+),2.5 N-m (Te), 40 ~tV(biceps), 30 ~V (brachio-radialis and triceps), 25 IxV (deltoid)

T

Angle Elbow Torque

Flexion Extension

Triceps 0

200

400 ms

0

clear cut early reponses from brachio-radialis. Only under conditions in which there was a high level of tonic activity in elbow flexors was it possible to obtain such early responses. Figure 3 illustrates one such example. We asked the subject to stiffen his limb by co-contracting elbow muscles. The results are shown in Fig. 3B and can be compared with results obtained from the same subject when he was asked only to resist the applied perturbation (Fig. 3A). Brachio-radialis activity was more modulated when the subject stiffened his limb (Fig. 3B), with an initial deep depression at a latency of 40 ms, followed by an increase in activity at a latency comparable to the late reponses we have already described. In three other experiments subjects were asked to hold a 1 kg weight in their hand. These experiments gave variable results. In one, brachio-radialis activity underwent the same changes as those described in Fig. 3B, while in another it did not change significantly following the perturbation. In the third, there was no depression in its activity and only an activation with a latency of 60 ms. Thus, even at higher levels of tonic activity, the reponses of brachioradialis to posteriorly directed forces on the upper arm were inconsistent from experiment to experiment. The E M G responses to perturbations which unload a muscle may not be reciprocal to the responses to stretch (cf. Crago et al. 1976; Matthews 1984). In particular, the depression in the activity of the unloaded muscle may be more labile than its activation following stretch. The results presented in Fig. 4 show that this explanation is unlikely to

200

400 ms

account for our findings and that the amplitude of the early responses of brachio-radialis depends in fact on whether the force perturbation is applied distally or proximally to the elbow, even when the muscle is stretched in both cases. We asked subjects to resist a pre-load tending to extend the upper arm which was then suddenly released. In Fig. 4B, the results of such an experiment can be compared with those obtained from the same subject when a small force perturbation is applied to the forearm (Fig. 4A). To resist the load (Fig. 4B), the subjet generated flexor torque at the shoulder by contracting shoulder muscles. (The same level of tonic activity in brachio-radialis as in 4A was obtained by asking the subject to hold a 1 kg weight in his hand.) The load was released at time 0 in a step-wise fashion. Thus, the unopposed flexor t o r q u e produced by the subject resulted in shoulder flexion and elbow extension, the latter due to dynamic coupling with shoulder motion. Note that the elbow extension induced by the perturbation was much larger and faster in Fig. 4B than in Fig. 4A. (Peak angular velocity at the elbow was 89~ in 4B and 30~ in 4A, in both cases occurring about 90 ms after pulse onset.) The initial changes in biceps activity were roughly reciprocal in the two cases, increasing in Fig. 4A and decreasing in 4B at the same latency. (As for brachio-radialis, an appreciable activation occurred in 4A (at a latency of 48 ms and with a maximum at 72 ms). When the force was applied to the upper arm, the latency of changes in brachio-radialis activity was much longer (80 ms), preceding Slightly the rebound in biceps activity. Brachialis responses were

140

L Anterior Deltoid Angle Shoulder Torque

Flexion Extension

1 Fig. 4A, B. Elbow extension induced by a force on the forearm (A) and by a force on the upper arm (B). In A, a small downwardly directed force was applied to the forearm. In B, a constant force in the backward direction was applied to the upper arm and released at time 0, resulting in shoulder flexion and elbow extension. Data are from the same experiment. The forearm was initially mid-pronated. Scales (per division) are: 10~ (O, e~ and ap), 10 N-m (T~), 5 N-m (Te), 75 ~tV (deltoid and pronator), 50 ~V (biceps) and 40 ~tV (brachio-radialis)

Biceps

Brachioradialis Angle Elbow Torque

I [

Flexion Wrist Angle

[ Pronation

Pronator Teres 0

200

400 ms

0

similar to those described for brachio-radialis but of a smaller amplitude. On the basis of results presented one can conclude that the direction of early responses in brachioradialis activity, when they are appreciable (Fig. 3B), is correlated with the direction of elbow motion caused by the perturbation: elbow extension can lead to activation of brachio-radialis and elbow flexion to its depression. However, the ease with which such responses are elicited depends on the angular motion at the shoulder. When shoulder and elbow angular motion are in the same direction (Figs. 1A, 2A, 4A), early responses are elicited consistently. However, when shoulder and elbow angular motion are oppositely directed (Figs. 1B, C, 2B, 3, 4B), early responses can be evoked only in the presence of a high tonic acitivity in elbow muscles. Responses at longer latencies (greater than 80 ms) are consistently present in brachio-radialis. Their time course tends to overlap in part that of biceps activation and appears to be poorly related to the changes in elbow angle and its derivatives when the force is applied proximally to the elbow.

Pseudo-random perturbations The results obtained using small random stimuli were qualitatively similar to those obtained using large single pulses or step perturbations. Results obtained by applying a pseudo-random sequence of pulses to the forearm are presented in Fig. 5A and for

200

400 ms

perturbations applied to the upper arm in Fig. 5B, where the average responses to a pulse of torque lasting 40 ms and beginning at the time given by the solid line (t -- 40 ms) are plotted. The EMG responses of anterior deltoid and biceps to pseudorandom perturbations applied to the forearm had a shorter latency (32 + 17 ms and 22 + 3 ms, respectively) and peaked earlier (84 + 6 ms and 68 + 6 ms) than those obtained when single pulses were applied. The responses in brachio-radialis and brachialis muscles (fig. 5A) paralleled biceps responses (brachio-radialis latency of 22 + 3 ms with a peak at 66 + 3 ms). The initial facilitation in biceps, brachio-radialis and brachialis was generally followed by a disfacilitation with a minimum at 160 ms. Pronator teres activation was instead less consistent and lagged the activation of elbow flexors. When the perturbation was applied to the upper arm (Fig. 5B), the kinematic and dynamic changes at the shoulder and elbow were also qualitatively similar to those previously described for single pulse perturbations. Anterior deltoid was activated at 30 + 7 ms, whereas biceps was activated at a slightly longer latency. The responses of brachio-radialis were more deeply modulated than those normally occuring with single pulses and resembled instead the responses sometimes obtained with the latter perturbation under conditions of increased tonic contraction of elbow muscles (Fig. 3B). They consisted of an initial depression with a latency of 31 + 8 ms and a minimum at 62 + 16 ms, followed by an activation at 80 + 23 ms peaking at 118 + 21 ms, and by a second

141

/ Anterior r-

A

Deltoid L

~

J !

Velocity

Flexion

Torque

Extension

Fig. 5A, B. Pseudo-random perturbations applied to the forearm (A) and to the upper arm (B). The traces depict the average impulse responses of the indicated variables to a 40 ms pulse of force beginning at the time indicated by the solid line and tending to extend the shoulder and elbow (A) or to extend the shoulder and flex the elbow (B). Data are from the same experiment and the forearm was mid-pronated. Scales (per division) are: 2.5 ~ (O, q> and ap), 50~ (0, @), 5 N-m (T~), 2.5 N-m (T~), 25 ~V (deltoid and pronator), 20 ~tV (biceps, brachio-radialis and triceps in A), 10 ~tV (biceps, brachioradialis and triceps in B). The dashed lines denote 100 ms after pulse onset

Biceps Brachioradialis Brachialis

Angle o Velocity w Torque

Triceps Wrist Angle I~ - I~ . i

Pronator Teres 0

200

400

ms

0

'!

~ 200

Anterior Deltoid

Angle Shoulder

-~"

" ...........

Torque Biceps ~ j

Angle

/"J~, . . . . . . . .

]

~

__.

Elbow

Torque Triceps

0

200

400

ms

Fig. 6. Variability in the early responses of brachio-radialis. Pseudo-random perturbations were applied to the upper arm with the forearm mid-pronated. Two sets of trials (each with 6 trials) were separately averaged according to whether an early depression was present (continous traces) or absent (dashed traces) in brachio-radialis responses. The stippled pattern fills the area between the two traces for E M G averages. Scales (per division) are: 2.5 ~ (O, (I)), 2 N-m (T~), 1 N-m (Te) , 50 p,V (deltoid), 5 ~V (biceps, brachio-radialis and triceps)

subsequent depression. Brachialis responses paralleled those of brachio-radialis, but were usually of a smaller amplitude. Note that for neither brachio-

400

ms

radialis or brachialis were the responses to a pulse of torque on the upper arm simply reciprocal to their responses to a pulse on the forearm. No consistent responses were found in pronator teres, nor in triceps. In one experiment we calculated the impulse responses for individual trials rather than for ensemble averages of 12 trials. On examining the impulse reponses of brachio-radialis we found some instances in which the early depression in its activity was absent. We then constructed two ensemble averages of these impulse reponses, separating the trials according to whether or not the initial depression in brachio-radialis activity was present. Figure 6 shows the results of this analysis, one set of impulse responses being denoted by the solid lines and the other by dashed lines. Note that, with the exception of the initial depression, the impulse responses of brachio-radialis are the same in the two sets of trials. In particular, the amplitude of the late activation is the same. Also, the impulse responses of the other muscles and of shoulder and elbwo angles and torques do not differ. Thus the results obtained using pseudo-random perturbations support the interpretation that brachio-radialis responses consist of distinct early and late components when the perturbation is applied to the upper arm, the early component generally geing related to the direction of elbow angular motion and the late component overlapping with biceps activity. Furthermore, they indicate that the two components can be modulated independently of each other.

142 Discussion

The results we have presented indicate that the EMG responses of elbow flexors to load perturbations of the arm are not uniquely related to the angular motion of the elbow joint, but depend instead on the angular motion at both the elbow and shoulder joints. Their responses differed when the angular motions at the two joints were in the same direction (as in the case when the force was applied to the forearm) and when the angular motions were oppositely directed (as a consequence of a force applied to the upper arm). Furthermore, the responses in mono-articular elbow flexors (such as brachioradialis and brachialis) did not always parallel those of biceps which also spans the shoulder joint. In this discussion, we shall take up the following questions concerning these observations: 1) what might be the afferent contributions underlying these responses, 2) can they be related to physical parameters of arm motion (i.e. joint angles or torques) resulting from the perturbation and 3) what might be their functional utility? With regard to the biceps responses, these questions have been taken up previously (Lacquaniti and Soechting 1986) and therefore this discussion will focus on the responses of monoarticular elbow flexors. Furthermore, it is convenient to discuss separately the "early" and "late" components of their response.

"Early" responses in brachio-radialis These were consistently evoked when elbow and shoulder angular motion were in the same direction (Figs. 1A, 2A, 5A) even when the angular velocity at the elbow was very small (Fig. 4A). Feedback from muscle spindle afferents from brachio-radialis and its synergists can account qualitatively for these responses. Elbow extension led to an increase in E M G activity, while flexion led to a decrease (unpublished observations). However, muscle spindle feedback from brachioradialis alone is unable to account also for the results obtained when elbow and shoulder angular motion are oppositely directed. Elbow extension at an angular velocity three times as great as that adequate to elicit a response to a force exerted on the forearm did not evoke any early changes in brachio-radialis activity when elbow extension was accompanied by shoulder flexion (Fig. 4B). In this context, it is known (Gottlieb and Agarwal 1979; Dufresne et al. 1979) that the EMG response to muscle stretch in which motion is restricted to a single joint increases linearly with velocity (up to 500~ for soleus and 75~

for biceps). It is also known (Hammond 1956; Evarts and Tanji 1974; Marsden et al. 1976; Gottlieb and Agarwal 1979; Dufresne et al. 1979) that the gain of the responses depends on the level of tonic EMG activity and on the instruction given to the subject (resist vs. do not resist). The tonic level of brachioradialis activity was the same in both experimental conditions depicted in Fig. 4 and the subject was instructed to resist both perturbations. Early responses in brachio-radialis were most prominent when pseudo-random perturbations were used (Figs. 5B, 6) and when there was a high level of tonic activity (Fig. 3B). Even then we were unable to evoke such responses consistently (Fig. 6). When the response was prominent, elbow flexion led to a decrease in brachio-radialis activity. By themselves, these results could be attributed to autogenetic feedback of muscle length whose gain was context dependent, specifically on the direction of motion at the shoulder. However, in about 25% of the experiments there was a small, but statistically significant increase in brachio-radialis activity following perturbations which led to elbow flexion and shoulder extension and they cannot be accounted for by this hypothesis. While it is obviously impossible to identify the afferent inputs which are responsible for the behavior of brachio-radialis from our results, muscle spindle feedback from bi-articular elbow flexors (such as biceps) and from shoulder flexors (such as anterior deltoid) onto brachio-radialis motoneurons can potentially account for these data. Given such a pattern of convergence and assuming reciprocal organization of flexors and extensors, there would be coincident excitation or inhibitation from stretch receptors of elbow and shoulder muscles when the direction of motion at both joints was the same and opposing effects when the two angular motions were oppositely directed. Note that an extensive pattern of connections from muscle spindle afferents originating from muscles acting at different joints has been demonstrated in the cat forelimb and hindlimb (Eccles et al. 1957; Eccles and Lundberg 1958; Fritz 1981). Furthermore there is no reason to discount contributions from other sources such as tendon organ afferents. The initial response of brachio-radialis thus appears to be related to angular motion (or its derivatives) at the elbow and shoulder joint. (This conclusion is only tentative since we have not systematically explored the responses to different combinations of elbow and shoulder angular motions as we did previously for biceps (Lacquaniti and Soechting 1986).

143

"Late" responses in brachio-radialis When elbow and shoulder angular motion were in the same direction the increase in brachio-radialis activity persisted for about 120 ms after the onset of the pulse perturbation. However, when elbow flexion was accompanied by shoulder extension the early depression, when present, was followed by an increase in EMG activity beginning at about 80 ms after pulse onset (Figs. 3B, 5B). This late component of the reponse was also present when there were no earlier changes in activity (Fig. 2B) and its amplitude was not correlated with the amplitude of the "early" response (Fig. 6). These observations are most easily interpreted by postulating that the response of brachio-radialis consists of two separate components having different neural substrates. Under this interpretation, these two components would have the same sign (facilitatory or disfacilitatory) when shoulder and elbow motion were in the same direction, while the "late" component would be facilitatory when elbow flexion was accompanied by shoulder extension. When motion is restricted to a single joint, a fractionation of the EMG responses to perturbations is not uncommonly found (Tatton et al. 1975; Marsden et al. 1976; Dufresne et al. 1979; Jaeger et al. 1982). The interpretation which has usually been given to those findings is similar to the one we have suggested here, namely that the distinct bursts represent responses having different latencies and neural substrates (cf. Evarts and Tanji 1974; Wiesendanger and Miles 1982). Experimental evidence in favor of (Tatton et al. 1975; Cheney and Fetz 1984; Matthews 1984) and against (Ghez and Shinoda 1978; Eklund et al. 1982) this hypothesis has been presented. The "late" component of the brachio-radialis responses always paralleled the changes in biceps activity, albeit not necessarily with the same time course. We previously showed that biceps responses to force perturbations were best related to the net change in elbow torque resulting from the applied perturbation (Lacquaniti and Soechting 1984; 1986). (Net torque is related in a nonlinear manner to the changes in angular position, velocity and acceleration at the shoulder and elbow joints.) Thus, one can tentatively suggest that similar mechanisms are responsible for both the changes in biceps activity and the "late" component of brachio-radialis activity and that they lead to a feedback control of torque at the elbow joint.

Functional utility While many questions remain concerning the neural substrates underlying the responses of mono- and bi-

articular muscles at the elbow to force perturbations and the identification of such responses with a feedback control of kinematic (angles) or dynamic (torque) parameters, it seems clear that there are nonautogenetic contributions to the motor output when a perturbation leads to motion at more than one joint of the arm. This conclusion agrees with observations of Cole et al. (1984) on the responses of thumb and index finger flexors to perturbations of the thumb during a task requiring subjects to produce a controlled pinch contact force and the responses of lower and upper lip muscles to perturbations of the lower lip during speech (Abbs and Gracco 1984). Their results are perhaps even clearer than ours since neither the thumb and index finger nor the lower and upper lips are dynamically coupled while the upper arm and the forearm are. Both of the motor tasks studied by Abbs and his coworkers involved the production of coordinated movements of different structures to achieve a required goal: finger and thumb movement to attain the desired contact force and motion of lower and upper lip to produce lip closure during speech. Close coordination of shoulder and elbow motion has been a consistent finding of studies of multi-articulate goal-directed arm movements (Soechting and Lacquanitit 1981; Lacquaniti and Soechting 1982). One can thus suggest that nonautogenetic afferent feedback is a characteristic of all motor tasks in which the coordination of motion of different structures (e.g. limb segments) is required to attain a specified goal. While the responses of brachio-radialis and biceps muscles appear to depend on motion at both the shoulder and elbow joint, the reponses of such monoand bi-articular muscle was not always the same. Therefore one can view the control system involved in the stabilization of fh~ limb as having multiple outputs as well as multiple inputs, each of the outputs (muscle responses) depending to a different degree on each input variable. In this respect the behavior would not be dissimilar from that of eye, neck and limb muscles in response to vestibular and neck afferent inputs (Robinson 1982; Baker et al. 1984; 1985; Suzuki et al. 1985). Head movement can be represented by rotations about three perpendicular axes (Robinson 1982) and the response of neck muscles to rotations in space can be characterized in terms of excitation vectors which describe the amplitude of the reponse to rotations about each of the axes. In this discription, each muscle has its own preferred vector orientation (Baker et al. 1985). The overall vestibulocollic responses result then from the coordinated action of all neck muscles excited by the stimulus. In contrast to the vestibulocollic reflex, the

144

appropriate input variables for the limb postural control system remain to be defined. Acknowledgements. This

work was supported by USPHS grant NS-15018 and by NSF grant BNS-8418539 and by the CNR (Italy).

References Abbs JH, Graco VL (1984) COntrol of complex motor gestures: orofacial muscle responses to load perturbations of lip during speech. J Neurophysiol 51:705-723 Baker J, Goldberg J, Herrmann G, Peterson B (1984) Optimal response planes and canal convergence in secondary neurons in vestibular nuclei of alert cats. Brain Res 294:133-137 Baker J, Goldberg J, Peterson B (1985) Spatial and temporal responses of the vestibulocollic reflex in decerebrate cats. J Neurophysiol 54:735-756 Cheney PD, Fetz EE (1984) Corticomotoneuronal cells contribute to long-latency stretch reflexes in the rhesus monkey. J Physiol (Lond) 349: 249-272, 1984 Cole KJ, Gracco VL, Abbs JH (1984) Autogenic and nonantogenic sensorimotor actions in the control of multiarticulate hand movements. Exp Brain Res 56:582-585 Crago PE, Houk JC, Hasan Z (1976) Regulatory actions of human stretch reflex. J Neurophysiol 39:925-935 Davies WDT (1970) System Identification for Self-adaptive Control. Wiley, London Dufresne JR, Soechting JF, Terzuolo CA (1978) Electromyographic response to pseudo-random torque disturbances of human forearm position. Neuroscience 3:1213-1226 Dufresne JR, Soechting JF, Terzuolo CA (1979) Reflex motor output to torque pulses in man: identification of short- and long-latency loops with individual feedback parameters. Neuroscience 4:1493-1500 Eccles JC, Eccles RM, Lundberg A (1957) The convergence of monosynaptic excitatory afferents onto many different species of alpha motoneurones. J Physiol (Lond) 137:22-50 Eccles RM, Lundberg A (1958) Integrative pattern of Ia synaptic actions on motoneurones of hip and knee muscles. J Physiol (Lond) 144:271-298 Eklund G, Hagbarth KE, Hagglund JV, Wallin EV (1982) The "late" reflex responses to muscle stretch: the resonance hypothesis versus the "long loop" hypothesis. J Physiol (Lond) 326:7%90 Evarts EV, Tanji J (1974) Gating of motor cortex reflexes by prior instruction. Brain Res 71:47%494 Fick R (1911) Anatomie und Mechanik der Gelenke, Vol. 31 Gustav Fischer, Jena Fritz N (1981) Ia-Synergismus an der vorderen Extremitfit der Katze. Ph.D. Thesis, Doktorarbeit Universitfit M/inchen

Ghez C, Shinoda Y (1978) Spinal mechanisms of the functional stretch reflex. Exp Brain Res 33:55-68 Gottlieb GL, Agarwal GC (1979) Response to sudden torques about ankle in man: myotatic reflex. J Neurophysiol 42: 91-106 Hammond PH (1956) The influence of prior instruction to the subject on an apparently involuntary neuromuscular response. J Physiol (Lond) 132:17-18 Hollerbach JM, Flash T (1982) Dynamic interactions between limb segments during planar arm movement. Biol Cybern 44: 67-77 Jaeger RJ, Gottlieb GL, Agarwal GC (1982) Myoelectric responses at flexors and extensors of human wrist to step torque perturbations. J Neurophysiol 48:388-401 Lacquaniti F, Soechting JF (1982) Coordination of arm and wrist motion during a reaching task. J Neurosci 2:399-408 Lacquaniti F, Soechting JF (1984) Behavior of the stretch reflex in a multi-jointed limb. Brain Res 311:161-166 Lacquaniti F, Soechting JF (1986) EMG responses to load perturbations of the upper limb: effect of dynamic coupling between shoulder and elbow motion. Exp Brain Res 61: 482-496 Marsden CD, Merton PA, Morton HB (1976) Stretch reflex and servo action in a variety of human muscles. J Physiol (London) 259:531-560 Matthews PBC (1984) Evidence from the use of vibration that the human long-latency stretch reflex depends upon spindle secondary afferents. J Physiol (London) 348:383-415 O'Leary DP, Honrubia V (1975) On-line identification of sensory systems using pseudorandom binary noise inputs. Biophys J 15:505-532 Robinson DA (1982) The use of matrices in analyzing the threedimensional behavior of the vestibulo-ocular reflex. Biol Cybern 46:53-66 Soechting JF (1984) Effect of target size on spatial and temporal characteristics of a pointing movement in man. Exp Brain Res 54:121-132 Soechting JF, Duffesne JR, Lacquaniti F (1981) Time-varying properties of myotatic response in man during some simple motor tasks. J Neurophysiol 46:1226-1243 Soechting JF, Lacquaniti F (1981) Invariant characteristics of a pointing movement in man. J Neurosci 1:710720 Suzuki I, Timerick SJB, Wilson VJ (1985) Body position with respect to the head or body position in space is coded by lumbar interneurons. J Neurophysiol 54:123-133 Tatton WG, Forner SD, Gerstein GL, Chambers WW, Liu CN (1975) The effect of postcentral cortical lesions on motor reponses to sudden upper limb displacement in monkeys. Brain Res 96:108-113 Wiesendanger M, Miles TS (1982) Ascending pathway of lowthreshold afferents to the cerebral cortex and its possible role in motor control. Physiol Rev 62:1234-1269 Received December 10, 1985 / Accepted August 20, 1986