Brain Research 751 Ž1997. 239–246

Research report

Intermuscular co-ordination during fast contact control leg tasks in man Caroline A.M. Doorenbosch ) , Tom G. Welter, Gerrit Jan van Ingen Schenau Institute of Fundamental and Clinical MoÕement Sciences, Vrije UniÕersiteit, Õan de Boechorststraat 9, 1081 BT Amsterdam, Netherlands Accepted 29 October 1996

Abstract From previous inverse dynamic analyses of human leg extensions, it was hypothesized that the underlying processes for the activation of mono- and biarticular muscles are different; the mono-articular muscles being activated when they shortened, whereas the biarticular muscles appeared responsible for the control of the external force direction. In the present study, experiments were performed on a dynamometer which was especially developed to test this hypothesis. Subjects had to exert different prescribed force vectors on a moving force-plate during leg extension, which they had intensively practised prior to the actual experiments. Of each trial, position, force and EMG activity were recorded. Net joint torques were calculated by the method of inverse dynamics and related to the EMG-patterns of the mono- and biarticular upper leg muscles to reveal whether the previously observed different roles in contact tasks might constitute a general principle in motor control. The results showed that although the action of the biarticular m. rectus femoris and hamstrings muscles was consistent in controlling the direction of the external force, the actions of the mono-articular muscles did not agree with their hypothesized role as simple work generators. The generalizability of a different control for mono-and biarticular muscles could thus not be confirmed for these tasks. They might rather reflect one out of more available strategies the CNS can use to control different contact control tasks. Keywords: Biarticular leg muscle; Force control; Motor control; Electromyography

1. Introduction Most goal-directed human and animal motions involve multiple joints in which specific combinations of joint torques are necessary to achieve the required segmental accelerations and the appropriate direction and magnitude of the force on the environment. From a mathematical point of view, there is a considerable number of ways in which muscles may interact to generate a specific movement. However, results from experimental studies using electromyography ŽEMG. or direct muscle force measurements indicate that even the most complex, i.e. multi-joint, movements are performed in a stereotyped way Že.g., w9,12,28x.. To understand the backgrounds of these stereotyped patterns of intermuscular coordination, a number of modelling and experimental approaches have been used to study the distribution of the forces between muscles. The general purpose of these studies was to reveal Žbiomecha-

) Corresponding author. Present address: Dept. of Rehabilitation, Academic Hospital Vrije Universiteit, Postbus 7057, Amsterdam. Fax: q31 Ž20. 4440787; e-mail: [email protected]

nical, physiological or neural. constraints that determine or restrict the number of possible muscle activation patterns Žfor an overview, see w5x.. In an inverse dynamic analysis of cycling, muscle activation patterns were interpreted as being useful to control the external force direction w13x. In that study, the activation of biarticular muscles appeared primarily associated with the distribution of the net torques about the joints, i.e., the direction of the external force whereas the activation of the mono-articular muscles related to the amount of muscle shortening. Comparable results were found for running w15x. Based on these findings, it was hypothesized that specific Žmechanical. task constraints are present, that may indicate different modes of control for mono- and biarticular muscles. Several neurophysiological studies appear to support this idea of a different organization in control of mono- and biarticular muscles Že.g., w2,22–24x; see w14x for more references.. For example, Buford and Smith w2x analyzed forward and backward walking of intact cats and showed that the mono-articular muscles demonstrated a similar pattern during both types of walking, whereas only the activity of the biarticular muscles differed during forward compared to backward walking.

0006-8993r97r$17.00 Copyright q 1997 Elsevier Science B.V. All rights reserved. PII S 0 0 0 6 - 8 9 9 3 Ž 9 6 . 0 1 3 2 7 - 3

240

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

To test the hypothesized different roles for mono- and biarticular muscles experiments were done in which force direction had to be controlled in Žquasi-.static leg extensions w6,16x. However, in comparison to cycling, the role of the mono-articular muscles was less clear in these studies, possibly due to the Žquasi-.static character of those experiments. Experiments under dynamic conditions might reveal more clearly the proposed differences in control for mono- and biarticular muscles. For this purpose, a dynamometer ŽFig. 1. is used to test the hypothesis whether a general organizational principle underlies a more or less independent control of mono- and biarticular muscles. With the tasks on the dynamometer Žcontact control tasks., in which the subjects had to exert a force in prescribed directions, it is expected that the different force directions would be realized by changes in activation of the biarticular muscles only. The tasks were executed cyclically at relatively high speed. The aim of the present study was to examine the roles of the mono- and biarticular muscles during these so-called fast contact control leg tasks ŽCCTs..

Fig. 2. Illustration of the standard position of the subject Ž1058 knee angle and 1158 hip angle. in the three positions of the dynamometer as were used for the experiments. Vertical position is indicated 08, forward tilted y158 and backward titled 158. Subjects were instructed to exert the force in seven different directions, ranging from 458 with intervals of 158 to 1358 relative to the force plate. The solid arrows indicate the direction of the reaction force.

gauges built in the force-plate were used to measure the exerted force in the sagittal plane.

2. Materials and methods

2.2. Subjects and experimental procedure

2.1. Dynamometer

Three male and two female subjects Žage: 25.4 " 3.6 years; height: 1.79 " 0.1 m; weight 71.1 " 6.5 kg. participated in the experiments. Informed consent was obtained from all subjects. During two months prior to the actual experiments, all subjects participated twice a week, in intensive training sessions of 1 h, in which they practiced the required tasks. Subjects were first placed in a reference position, while sitting on the bike-saddle with their right foot on the force-plate ŽFig. 1.. In this reference position the hip joint was flexed 115 degrees and the knee joint was set at 100 degrees of flexion ŽFig. 2.. During the CCTs, subjects had to exert a force in a prescribed direction with a constant magnitude, while the force-plate moved five times downand upwards with a velocity of 40 cm sy1 , starting at 17 cm above until 17 cm below the reference position and vice versa. The subjects were instructed to exert the prescribed force vector while the force-plate moved downwards and hold the force magnitude and direction as constant as possible during the whole movement. The magnitude of the force differed for the male and female subjects and was set at 300 and 200 N respectively. On-line feedback of the direction and magnitude of the required and the actual reaction force was presented on a computer screen which was placed in front of the subjects. No movements of the foot relative to the force-plate were allowed. Force directions ranged from 45 up to 135 degrees with intervals of 15 degrees ŽFig. 2.. All seven force directions were executed in each of three different positions of the rail and force-plate, indicated as 0 Žperpendicular to the ground., y15 degrees Žforward tilted. and 15 degrees Žbackwards tilted. ŽFig. 2..

In the dynamometer used for the experiments, a forceplate is moved in a straight line along a rail of about 1 m. The rail could be mechanically rotated about an axis at the lower side of the rail, as well as translated over a double bar at the bottom of the frame ŽFig. 1.. This allowed for the adjustments of different movement directions of the force-plate. The force-plate runs over a ball-rotation spindle via a metal mounting fixture. This spindle is attached by cogwheels to a servo-motor ŽSEM, MT3OU4-36, London, UK. which is controlled by a motion control software package ŽGALIL DMC-710, Sunnyvale, CA, USA.. Strain

Fig. 1. Illustration of the dynamometer. The right foot of the subject is placed on a force plate which can be moved in up- and downward direction along a rail. The rail can be rotated and translated Žsee arrows. such that different movement directions of the force plate can be obtained.

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

For each protocol, computer generated audible signals were used to indicate the start of the movement period and the end of the trial. These signals were simultaneously recorded as block pulses and were used for synchronization. During each trial, joint position and reaction force were recorded and muscle activation was registered by means of EMG.

241

position data, the center of application was related to the position of the foot. Newtonian equations of motion were applied to a linked segment model of the human body in order to calculate the net torques about the hip, knee and ankle joints w7x. Extension torques were defined as positive. 2.4. Electromyography

2.3. Position and force Retro-reflective markers Ždiameter 13 mm. placed on the skin corresponded with the fifth metatarsophalangeal joint, heel, lateral malleolus, knee joint Žon the lateral collateral ligament at the height of the knee cleft., greater trochanter, top of iliac crest and neck Žat the height of fifth cervical vertebra.. Two landmarks were placed on the force-plate and three on the dynamometer frame as a reference orthogonal coordinate system. Joint positions of the leg in the sagittal plane were collected at 60 Hz using a motion analysis software system ŽVICON, Oxford Metrics, Oxford, UK.. Four camera positions were applied in order to get accurate cartesian coordinates of the landmarks. Special attention was paid to minimize reflections, which could give rise to false marker positions, using black curtains and paint. With the VICON-system, 3-D coordinates are calculated. After proper scaling, the absolute coordinates of the anatomical landmarks in the sagittal plane were obtained, which were subsequently low pass filtered ŽButterworth 2nd order, bi-directional of a 5-points approximation, cutoff frequency 2 Hz.. This low cutoff frequency was chosen in order to minimize quantisation noise and is appropriate in these isokinetic movements since the joint angles and their derivatives appear not to contain higher frequencies than 1 Hz. The coordinates of the landmarks defined four body segments: foot, lower leg, upper leg and the upper body. Joint angles as well as joint angular velocities were calculated ŽLanczos, 5-point differentiation filter, w18x.. To determine the positions of the mass centers of the different body segments, the coordinates were combined with anthropometric measurements of each subject and data from Clauser et al. w4x and Winter w31x. Vertical and fore–aft components of the reaction force of the right foot were recorded by means of strain gauges built in a force-plate. The analog force signals were amplified, low-pass filtered Ž70 Hz, 4th order, 24 dBroct. and sampled Ž300 Hz, 12 bits. by MONSTIM. Simultaneously, the force signals were processed by a computer to generate visual feedback of the force vector for the subject. The force direction perpendicular on the force-plate was indicated as 90 degrees and plotted as a vertical line. Of the required force vector, both magnitude and direction were indicated with deviation margins of "6%. From the distribution of the separated force components measured by the force-plate, the center of pressure of the force vector was calculated. Combined with the points of reference of the

Electromyographic activity was recorded ŽDISAS type 15C01, Skovlunde, Denmark. from the following leg muscles: m. gluteus maximus, m. rectus femoris, m. semitendinosus, m. biceps femoris c.longum., m. vastus medialis., m. vastus lateralis. EMG was obtained using bipolar Ag– AgCl surface electrodes ŽMediTrace type: ECE 1081; leadoff area 0.81 cm2 , center-to-center electrode distance 2.5 cm. placed on the muscle bellies after standard skin preparations w1x. EMG signals were amplified and sampled Ž600 Hz. on-line by the VICON-system after band-pass filtering Ž20–200 Hz.. The EMG signals were plotted on a writer ŽGould ES 1000, France. for visual inspection during the experiment. Off-line analysis of the EMG signals included rectifying, low-pass filtering ŽButterworth 2nd order, zero phase lag through bi-directional application of a fifth approximation, cutoff frequency 2 Hz. and normalizing to 100% of the SIC values. Values of these standard isometric contractions ŽSIC. for each muscle group were obtained prior to the actual experiments by the methods described in Jacobs and Ingen Schenau w16x. SIC values were taken as the mean value of the rectified curve for a period of 1 s in which the activation level was stable. The SIC-tests were also used as a functional test to identify possible cross-talk as was recommended by Winter et al. w32x. 2.5. Muscle length changes Joint angle changes during a foot displacement of 20 cm Ž"10 cm relative to the reference position. in the direction of the movement and information of total muscle length Ž Loi . as a function of joint angleŽs. were used to calculate the length changes Ž D L oi . for each muscle for each trial. Positive values indicate a muscle shortening. In order to determine the muscle length as a percentage of segment length, a database of coefficients, derived from various studies, was used. Details can be found in Jacobs and Ingen Schenau w16x. 2.6. Treatment of data EMG, position and force data were synchronized by the pulses, which also generated the audible Žcue. signals. For each task, the mean value of net torques and EMG, calculated over a distinct period for each of the two last cycles, were used for further analysis. The mean data were derived from the period for which the knee angle was

242

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

tion and the actual direction of the force as exerted by all subjects is shown in Fig. 3 for each subject. The mean values with the standard deviation Žbetween parentheses. of the each force direction are shown at the inset. 3.2. Net torques and electromyography

Fig. 3. The prescribed force direction Žrange 45–1358 with intervals of 158. on the horizontal axis versus the executed force direction of each subject. The inset showed the mean value Ž ns 5. and the standard deviation between parentheses for all subjects.

within 5 degrees of the reference position during the downward movement. These datapoints were subsequently averaged for all subjects at corresponding tasks. 3. Results 3.1. Task performance The tasks were performed with high accuracy by all subjects. A comparison between the prescribed force direc-

The mean EMG-values of all subjects, expressed as a percentage of the SIC value are shown in Fig. 4. The EMG-data are plotted as a function of the net torques about the joints that are crossed by the muscle. Fig. 4 shows that the activity of the gluteus maximus is low at negative Žflexion. hip torques, whereas at higher hip extending net torques, the activity of the gluteus maximus increases until about 50% of SIC. Also for the mono-articular vasti muscles an increase of about 60% is seen when they can contribute to the extending knee torques. It is noteworthy that the mono-articular knee extensors are activated Žuntil about 20% of SIC. at negative knee torques as well. Since the biarticular muscles cross two joints over which they have a simultaneous effect, the EMG data are plotted against the combination of these torques, defined as the net torque difference between knee and hip ŽT k y Th .. As a consequence, a combination of hip extension and knee flexion results in a negative T k y Th difference, whereas a positive difference originates from the opposite

Fig. 4. For each task, the mean EMG values expressed in % of the SIC-level of four mono- and biarticular muscles of the upper leg are plotted as a function of net torque ŽT k or Th . or net torque difference of knee and hip ŽT k y Th .. Positive values are defined as net extending torques.

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

243

A combination of the datapoints of the biarticular muscles in Fig. 4 is made by subtracting the hamstrings activity from that of the rectus, denoted as RF-HAM. This activity difference is shown in Fig. 5 as a function of net torque difference ŽT k y Th .. A remarkably strong relationship between the activity difference RF-HAM and torque difference is found. The mean Ž"S.D.. correlation coefficient was 0.986 Ž".004.. 3.3. Net torques, EMG and muscle length changes

Fig. 5. The activity difference between the hamstrings and rectus femoris for all tasks as a function of the net torque difference ŽT k yT h .. Mean EMG data of all subjects Ž ns 5. are shown with the fitted line. The inset shows the correlation coefficient for the two variables.

combination, i.e. hip flexion and knee extension. Comparing the patterns of the rectus femoris and hamstrings, it is clear that these are reversed. The activity of the hamstrings is very low at positive values of T k y T h , whereas the rectus is silent in the negative range. At higher negative values of the net torque difference, the activity of the hamstrings increases, while the opposite is found for the rectus femoris: an increase at a higher positive difference.

Three different movement directions were used in the experiments. During the movement, the length of the muscles changes as a result of the angular changes in the joints. Since the trajectory of the foot is fixed for each movement direction, the angular changes of the joints are similar irrespective of the executed force directions. For all muscles, three different combinations of muscle length changes Ž D Loi . are present. In Fig. 6, the mean EMG data of each muscle are plotted as a function of the muscle length changes Ž D L oi . for each movement direction. In these plots, positive values indicate muscle shortening. The radius of the circles around each datapoint corresponded to the magnitude of required net torque ŽT h or T k . or net torque difference ŽT k y Th or Th y T k . for that condition.

Fig. 6. Mean EMG data as a function of the amount of muscle shortening Ž D Loi . in each movement direction of the force plate. Positive values of dLoi indicate muscle shortening. The magnitude of the corresponding net torque ŽT k , Th or T k y Th or Th y T k . is shown as the radius of the circle around each datapoint. Filled circles indicate a negative net torque ŽT k or Th . difference ŽT k y Th or Th y T k ..

244

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

Filled circles indicate negative torque values. Fitted regression lines demonstrate the relation between the activity level and the amount of muscle shortening. For the mono-articular muscles, m. gluteus maximus and mm. vasti and for the biarticular m. rectus femoris, an increase of EMG activity is found with increasing net extending torques and increasing muscle shortening. For the imposed movements, only the biarticular hamstrings are lengthening, which occurred in direction 158 of the movement. With larger negative values of D L oi , the EMG activity of this muscle group decreases. However, when comparing circles with equal radius at different D Loi values, it is clear that there is only a weak relationship between the length changes and the EMG-activity for both the biarticular as well as the mono-articular muscles.

4. Discussion 4.1. Variability In order to test the generalizability of the hypothesis that mono-articular muscles are mainly activated when they can contribute to work output and biarticular muscles are uniquely responsible for the direction of the force, indicating different roles in the control of force and position, subjects were asked to perform fast CCTs on a dynamometer. First, it is discussed whether the subjects performed the tasks in a similar way. According to Fig. 3, all subjects performed the experimental tasks quite accurately. The control of force direction results directly from a combination of net torques about the joints, which are a function of muscle activation. The EMG data of the upper leg muscles ŽFig. 4. showed that all subjects performed the task in a remarkably similar manner, despite the redundant set of muscles. This most likely points at constraints which lead the subjects to a stereotyped intermuscular coordination to accomplish these tasks. On the basis of the present results, it is not clear what type of constraining mechanism may have led to this highly stereotyped pattern of muscle coordination. Obviously, the results show that the activation of mono- and biarticular muscles is similar in the present CCTs. The question thus arises to what extent this can be explained as an exception of a more general principle or as evidence against the hypothesized neural constraints suggested in previous studies w13–15x. 4.2. Is the control of mono- and biarticular muscles based on different organizational processes? From previous analyses of cycling, running and static leg tasks, it was hypothesized that the biarticular muscles, rectus femoris and hamstrings are uniquely responsible for the control of the force direction on the pedal, allowing this force to do work on the pedal w13,15,16x. For the fast CCTs in the present study, the biarticular upper leg mus-

cles again showed a strong reciprocal pattern of activity as a function of the required net torques. Given the largely noisy character of the electrophysiological data, the mean correlation coefficient of 0.986, describing the relationship of between net torque difference of knee and hip ŽT k y Th . and the RF-HAM activity difference is even remarkably high. Apparently, this reciprocal behavior played an important role in the distribution of the net torques about hip and knee, hence in the control of the direction of the external force. However, if the control of the force direction were to be uniquely tuned by the biarticular muscles, the activity of the mono-articular muscles should remain constant, irrespective of the required net torques. This was not confirmed in the present study. Fig. 4 shows that the activity of both the mono- and biarticular muscles increase with increasing net torque. This means that changes in the activity of the mono-articular muscles also have their effect on the magnitude of the net torques and hence on the direction of the external force vector. Clearly, the high correlation coefficient mentioned for the biarticular muscles, combined with the strong relationships between the net joint torques and the activity of the mono-articular muscles can only be explained by assuming a common input of the motoneurons of both types of muscles in this task. This appears to be in strong contrast to what was found previously w13,15x. Moreover, in the stereotyped performance of the dynamometer tasks, the activation of the mono-articular muscles did not solely coincide with the amount of muscle shortening, as would be expected according to the proposed hypothesis. Both the biarticular as well as the monoarticular muscle activity related to the amount of muscle length changes Žsee Fig. 6.. In comparison to the observations in cycling and running, the different roles between mono- and biarticular muscles are thus considerably less pronounced. The results of this study can be judged as evidence against the presence of a neural constraint of distinctly different control of mono- and biarticular muscles as a general organizational principle. Whether one should propose an alternative general control principle or constraint, which can explain the activation patterns of all CCTs, or that the observed coordination of the present tasks should be judged as an additional available strategy of the CNS is not clear. To refer to the first point, an alternative candidate might be found in the implications of the proposed hypothesis, namely that energy wasting eccentric contractions are minimized w12,25,29x. According to this criterion of minimization of negative muscle power, muscles are not classified to the number of joints they cross, since both mono-articular as well as biarticular muscles are uniquely suitable to perform a specific task with high efficiency w17x. The second option mentioned above is to assume that the CNS has the possibility to use a different strategy

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

where not only the biarticular muscles but also the monoarticular muscles control the distribution of net torque about the joints. This is for example supported by the present observation that for the mono-articular muscles only a weak relationship exists between muscle activation and the extent to which these muscles can contribute to external work; i.e., the amount of muscle shortening, which is in sharp contrast to the actions found in previous studies Že.g. w15x.. Also the finding that the relation between muscle activation and joint torque is already present at relatively low joint torques Žsee Fig. 4., whereas such a relation can be completely absent in comparable joint positions and external forces in cycling and running, directs to the use of different strategies rather than understanding muscle activation patterns of CCTs by assuming some umbrella organizational principle or constraint. As argued before w14x, the results of several neurophysiological studies about innate automatic, centrally driven rhythmical movements as Žfictive. locomotion, scratching or paw shaking and as well as observations from analyses on reflex control did seem to point to differential neural constraints which realize a distinct control for mono- or biarticular muscles w2,3,8,10,11,21–24,30x. The differences in actions of mono- and biarticular muscles during these innate automatic movements might be organized on the basis of circuitries, such as propriospinal networks located in phylogenetically older parts of the CNS or by younger structures influenced by genetically coded ‘ value systems’ as advocated by Sporns and Edelman w26x. Comparable arguments can be formulated to explain different organizations in the control of mono- and biarticular muscles in fast responses to sudden disturbances w19,20x. The present results however, demonstrate that, at least in man, other strategies can be used as well. It might well be that this different strategy is mainly induced by the instruction to realize a particular external force vector which is prescribed and controlled by visual information. This visual force feedback is distinctly different from the type of feedback Že.g. fatigue. and goal of the movement as for example in locomotion. These considerations are also in line with the findings of Tax et al. w27x that the motor unit behavior of arm muscles depended on the type of instruction, which determines the initial goal Žmovement or force.. However, finally it should be noted that at present the results of this and previous studies to human CCTs are not sufficient to formulate decisive arguments against Kuo’s statement w17x that the CNS does not have any a priori knowledge of the existence of mono- and biarticular muscles at all.

Acknowledgements The authors gratefully acknowledge dr DirkJan Veeger for his helpful comments on earlier versions of the

245

manuscript. This study was supported by the Life Sciences Foundation, which is funded by the Netherlands Organization for Scientific Research ŽNWO..

References w1x Basmajian, J.V. and DeLuca, C.J., Muscles AliÕe; Their Functions ReÕealed by Electromyography, Williams and Wilkins, Baltimore, MD, USA, 1985. w2x Buford, J.A. and Smith, J.L., Adaptive control for backward quadrupedal walking. II. Hindlimb muscle synergies, J. Neurophysiol., 64 Ž1990. 756–766. w3x Carlson Kuhta, P.C. and Smith, J.L., Scratch responses in normal cats: hindlimb kinematics and muscle synergies, J. Neurophysiol., 64 Ž1990. 1653–1667. w4x Clauser, C.E., McConville, T.J. and Young, J.W., Weight, volume and center of mass of segments of the human body, Wright-Patterson Air Force Base, AMRL-TR-69-70, Ohio, USA, 1969, pp. 339– 356,. w5x Collins, J.J., The redundant nature of locomotor optimization laws, J. Biomech., 28 Ž1995. 251–267. w6x Doorenbosch, C.A.M. and Ingen Schenau, G.J. van, The role of mono- and biarticular muscles during contact control leg tasks in man, Hum. MoÕem. Sci., 14 Ž1995. 279–300. w7x Elftmann, H., Forces and energy changes in the leg during walking, Am. J. Physiol., 125 Ž1939. 339–356. w8x Grillner, S. and Zangger, P., On the central generation of locomotion in the low spinal cat, Exp. Brain Res., 34 Ž1979. 241–261. w9x Herzog, W., Leonard, T.R. and Guimaraes, A.C.S., Forces in gastrocnemius, soleus and plantaris muscles of the freely moving cat, J. Biomech., 26 Ž1993. 945–953. w10x Hoffer, J.A., Loeb, G.E., Sugano, N., Marks, W.B., O’Donovan, M.J. and Pratt, C.A., Cat hindlimb motorneurons during locomotion. III. Functional segregation in sartorius, J. Neurophysiol., 57 Ž1987. 554–562. w11x Hoy, M.G., Zernicke, R.F. and Smith, J.L., Contrasting roles of inertial and muscular moments at the ankle and knee during pawshake responses, J. Neurophysiol., 54 Ž1985. 1282–1294. w12x Ingen Schenau, G.J. van, From rotation to translation: constraints on multi-joint movements and the unique action of mono-and biarticular muscles, Hum. MoÕem. Sci., 8 Ž1989. 301–337. w13x Ingen Schenau, G.J. van, Boots, P.J.M., Groot, G. de, Snackers, R.J. and Woensel, W.W.L.M. van, The constrained control of force and position in multi-joint movements, Neuroscience, 46 Ž1992. 197– 207. w14x Ingen Schenau, G.J. van, Pratt, C.A. and Macpherson, J.M., Differential use and control of mono- and biarticular muscles, Hum. MoÕem. Sci., 13 Ž1994. 495–517. w15x Ingen Schenau, G.J. van, Welter, T.G., Dorssers, W.M.M., Beelen, A., Groot, G. de and Jacobs, R., The control of mono-articular muscles in multijoint leg extensions in man, J. Physiol., 484.1 Ž1995. 247–254. w16x Jacobs, R. and Ingen Schenau, G.J. van, Control of an external force in leg extensions in humans, J. Physiol., 457 Ž1992. 611–626. w17x Kuo, A.D., A mechanical analysis of force distribution between redundant, multiple degree-of-freedom actuators in the human: Implications for the central nervous system, Hum. MoÕem. Sci., 13 Ž1994. 635–663. w18x Lees, A., An optimised film analysis method based on finite differences techniques, J. Hum. MoÕem. Stud., 6 Ž1980. 165–180. w19x Macpherson, J.M., Strategies that simplify the control of quadrupedal stance. I. Forces at the ground, J. Neurophysiol., 60 Ž1988. 204–217. w20x Macpherson, J.M., Strategies that simplify the control of quadrupedal stance. II. Electromyographic activity, J. Neurophysiol., 60 Ž1988. 218–231.

246

C.A.M. Doorenbosch et al.r Brain Research 751 (1997) 239–246

w21x Perell, K.L., Gregor, R.J., Buford, J.A. and Smith, J.L., Adaptive control for backward quadrupedal walking. IV. Hindlimb kinetics during stance and swing, J. Neurophysiol., 70 Ž1993. 2226–2240. w22x Perret, C. and Cabelguen, J.-M., Central and reflex participation in the timing of locomotor activations of a bifunctional muscle, the semi-tendinosus, in the cat, Brain Res., 106 Ž1976. 390–395. w23x Perret, C. and Cabelguen, J.-M., Main characteristics of the hindlimb locomotor cycle in the decorticate cat with special reference to bifunctional muscles, Brain Res., 187 Ž1980. 333–352. w24x Pratt, C.A., Chanaud, C.M. and Loeb, G.E., Functionally complex muscles of the cat hindlimb. IV. Intramuscular distribution of movement command signals and cutaneous reflexes in broad, bifunctional muscles, Exp. Brain Res., 85 Ž1991. 281–299. w25x Roeleveld, K., Lute, E.M.C., Veeger, H.E.J., Gwinn, T. and Woude, L.H.V. van der, Power output and technique of wheelchair athletes, Adapt. Phys. Act. Q., 11 Ž1994. 71–85. w26x Sporns, O. and Edelman, G.E., Solving Bernstein’s problem: a proposal for the development of coordinated movement by selection,

Child DeÕ., 64 Ž1993. 960–981. w27x Tax, A.A.M., Denier van der Gon, J.J., Gielen, C.C.A.M. and Kleyne, M., Differences in the central control of m.biceps brachii in movement and force tasks, Exp. Brain Res., 79 Ž1990. 138–142. w28x Veeger, H.E.J., Woude, L.H.V. van der and Rozendal, R.H., Load on the upper extremity in manual wheelchair propulsion. J. Electromyogr. Kinesiol., 1 Ž1991. 270–280. w29x Veeger, H.E.J. and Woude, L.H.V. van der, Force generation in manual wheelchair propulsion. In J. Vossoughi ŽEd.., Proceedings of the XIIIth Southern Biomedical Engineering Conference, pp. 779–782, Washington, DC, USA, 1994. w30x Vidal, C., Viala, D. and Buser, P., Central locomotor programming in the rabbit, Brain Res., 168 Ž1979. 57–73. w31x Winter, D.A., Biomechanics of Human MoÕements, John Wiley and Sons, New York, USA, 1979. w32x Winter, D.A., Fuglevand, A.J. and Archer, S.E., Crosstalk in surface electromyography: theoretical and practical estimates, J. Electromyogr. Kinesiol., 4 Ž1994. 15–26.