Eleven subjects, eight men and three women aged between 25 and 43 years, participated in the study. All gave informed consent. The procedures conformed to the guidelines laid out in the Declaration of Helsinki and were approved by the local ethics committee.
Subjects sat in a rigid chair with the forearm firmly strapped to a rigid plate at the wrist (Fig. 1A). The upper arm remained approximately vertical. The plate was attached to an aluminium beam which rotated on low-friction bearings about the axis of the elbow, and could be loaded with weights that were firmly screwed into place to prevent movement. The plate also rotated about the elbow axis, but was attached at its distal end to the beam with a strain-gauge type load cell (XTran 2 KN, Applied Measurement, VIC, Australia).
| Figure 1 Experimental apparatus and protocol |
For measurement of voluntary activation during maximal concentric contractions, subjects were asked to lift the weights ‘as fast as possible’. To encourage maximal effort throughout, the subjects were instructed to keep pulling as fast as possible from a starting position with the bar about 70 deg below horizontal until the beam was stopped by a brake at about 70 deg above the horizontal. A light switch embedded in the device's axle triggered as the beam passed through the horizontal. After a delay of 25 ms, the switch triggered a constant-current stimulator which delivered a single, supramaximal stimulus to the elbow flexor muscles (0.1 ms duration; typically 100–300 mA) through 2 cm diameter electrodes firmly adhered to the skin over the mid-belly and distal tendon of the biceps muscle. The same switch triggered a purpose-built amplifier which subtracted the voltage of the amplified force signal at the time of the stimulus from the subsequent signal, before amplifying the signal a further 10 times (Hales & Gandevia, 1988; Allen et al. 1995). This permitted a high level of amplification of the analog force signal prior to sampling. The angular velocity of the beam (and hence also of the arm, plate and weights) was measured with an optical encoder built into the axle. The digital signal from the encoder was converted to an analog signal which was then integrated to give angular displacement and smoothed with a ten point running average and differentiated to give angular acceleration. Force and angular velocity signals were sampled at 1000 Hz with 12-bit resolution.
Force measured with the load cell was multiplied by the load cell's lever arm to give torque produced by the plate on the load cell (Tlc). The force seen by the load cell reflects the torque produced by muscles (and perhaps other soft tissues) at the elbow, as well as torques due to the weight and inertia of the plate and arm. Muscle torque (Tmm; positive when the torque acts to flex the arm) was thus calculated from the relation:
where
Tw is the weight torque of the plate and the arm (the value of this parameter is always negative because the weight always acts in the extension direction),
I is the moment of inertia of the plate and arm, and α is the angular acceleration of the plate and arm (which are assumed to be equal). The weight torque of the plate was measured directly with the centre of mass of the plate horizontal to its axis, and thereafter was calculated by multiplying by the cosine of the angular displacement from that position. The moment of inertia of the plate was determined from its measured weight torque when horizontal and its period of oscillation (
Resnick & Halliday, 1977). Weight torque and moment of inertia of the arm were estimated from body mass and height using the tables of
Zatskiorsky et al. (1990) as modified by
de Leva (1996), taking into account the flexed position of the fingers. The weight torque of the arm was subsequently multiplied by the cosine of the angle the beam made with the horizontal.
Measurement of interpolated twitch amplitude
During maximal concentric contractions, muscle torque was changing at the time the stimulus was delivered, presumably because the force-generating capacity of muscles is length and velocity dependent. This was readily observed in trials in which the muscle was not stimulated; muscle torque was then usually decreasing (or occasionally increasing) even though no twitch was evoked. The decrease or increase in torque was almost always monotonic for several hundred milliseconds and could not be mistaken for the twitch response to a single stimulus. Nevertheless, estimates of interpolated twitch amplitude obtained by subtracting torque immediately preceding the twitch from the peak torque of the twitch would have been biased if they did not take account of the changes in torque that occur throughout the twitch. We estimated the muscle torque that would have occurred in the absence of an interpolated twitch by assuming that the rate of change in torque over 25 ms preceding the stimulus was sustained for the brief duration of the twitch. Interpolated twitch amplitude was determined by subtracting the torque predicted in this way from the measured torque. To enable comparison of voluntary activation during concentric and isometric contractions, the amplitude of interpolated twitches obtained under isometric conditions was obtained in the same way. This method, based on extrapolation, has previously been used for isometric contractions and does not alter the level of voluntary activation measured from a series of contractions in individual subjects (
Allen et al. 1995).
Measurement of resting twitch amplitude
To calculate voluntary activation, the amplitude of interpolated twitches was normalized to the amplitude of twitches evoked from resting muscle (‘control’ twitches). Control twitches for dynamic contractions were obtained in the following way. First, the beam was fixed in the horizontal position and the subject performed a maximal isometric contraction (to potentiate the muscle). Five seconds later, with the subject completely relaxed, the beam was swung through the horizontal at a range of angular velocities, triggering the stimulator to evoke a control twitch. The EMG obtained from surface electrodes placed over the belly of the triceps and brachioradialis muscles was displayed on an oscilloscope at high gain to confirm that subjects were relaxed at the time the stimulus was delivered. Resting twitches were evoked at a range of angular velocities and the relationship between control twitch amplitude and angular velocity was determined for each subject.
Calculation of voluntary activation
Voluntary activation was calculated as 100 × (1 -
a/b) where
a is interpolated twitch amplitude and
b is the amplitude of the control twitch (e.g.
Bigland-Ritchie et al. 1983;
Allen et al. 1995). When calculated in this way, voluntary activation is relatively insensitive to the amplitude of the resting twitch. For example, underestimation of resting twitch amplitude by 20 % would produce an error in voluntary activation of only 1.2 % if voluntary activation was truly 95 %. To calculate voluntary activation for a particular concentric contraction, the amplitude of the control twitch at the velocity of that contraction was estimated from the regression of control twitch amplitude on angular velocity (see above). Voluntary activation was not calculated for trials in which angular velocity exceeded 300 deg s
−1, because the resolution of the method was limited at high velocities (see Discussion).
Protocol
First, interpolated twitches and control twitches were evoked during and after five maximal voluntary isometric contractions (
Fig. 1B). Then the beam was unclamped and loaded with a moderate weight, typically to a torque of 36 N m with the beam horizontal. This compares with a median maximal isometric torque for the present group of subjects of 81.4 N m. Subjects performed five maximal lifts with this weight, during which interpolated twitches were obtained, and after which control twitches were obtained as described above. The procedure was repeated with a heavy weight (typically 58 N m) and a light weight (typically 23 N m), before five final isometric contractions. Subjects were asked to perform ‘maximal’ efforts regardless of the weight being lifted, so the effect of varying the weight was to vary the velocity of contraction. At least 1 min of rest was allowed between contractions. For three subjects, the stimulus was sometimes delivered at angles 30 deg above the horizontal to see if there was any effect of joint angle on voluntary activation.
Effect of fatigue
To investigate the level of voluntary drive during fatiguing dynamic contractions, five subjects repeatedly lifted and lowered a near-maximal weight and a stimulus was delivered to the contracting muscles during each lift. Subjects were asked to perform the lifts at a regular cadence (rather than at the maximal possible speed as for other measurements reported here), so initially the lifts involved submaximal efforts. The weights lifted were set so that between the sixth and twelfth concentric contractions subjects became unable to continue lifting without assistance. At that stage, one of the investigators supplied as little assistance as was necessary to help the subject complete the lift, and the subject continued lifting in this way, with increasing levels of assistance, for another ten to twelve repetitions. No assistance was provided as the load was lowered. The cadence and angular velocity of the lifts was kept as constant as possible across repetitions. As it was not possible to determine resting twitch amplitude for each level of fatigue, voluntary activation scores were not formally calculated for fatigue experiments.
Statistics
Because levels of voluntary activation are not normally distributed, non-parametric statistical tests were used and details are given in the Results. Some correlations were examined with linear regression. Statistical significance was set at the 5 % level.