PREFACE

Top Previous Next References

The science of physiology, which, during the last hundred years has exclusively concerned itself with the study of animals, has been occupied during the last decade with problems which are more specifically concerned with the physiology of man. From the physiology of rest and of basal function we now increasingly turn to the study of man in an active role ... in situations in which enormous demands are made upon the subjects' attention, will and adaptability.

N. Bernstein (1967)

The Co-ordination and Regulation of Movements

	I. INTRODUCTION

Top Previous Next References

If muscle is regarded as a motor, then the way it behaves depends not only on its intrinsic properties but also on the way that it is driven and the way feedback systems maintain its output. Feedback may operate locally at the level of the spinal motoneuron or at supraspinal levels. Just as many sites within the muscle cell control force, so many sites within the central nervous system (CNS) can modify the output of motoneurons. Modern reviews of muscle physiology often proceed on the premise that the limit to production of force by volition is at or within the muscle cell itself, or that if this is not so, deficiencies in drive to motoneurons are quantitatively small (e.g., Refs. 8, 90, 231, 235, 636, 665). As indicated in section IA, there is a history of controversy about central and peripheral factors in muscle fatigue.

Several factors have contributed to the delay in establishing the role of "central" factors in human muscle fatigue. First, it has simply been convenient to assume that the limits to muscle force established in reduced preparations of muscle devoid of effective neural input apply to a conscious human subject. Second, the methods to gauge central drive to muscle have not always been technically rigorous, so that findings obtained with them have been easily criticized or ignored. Third, although changes in the CNS during exercise can be measured, it has been more demanding to show that they cause a deficit in force production.

This review covers some of the changes that occur at the motoneuron pool and at supraspinal sites during human muscle fatigue. It highlights the occurrence and measurement of such changes and attempts to determine their functional importance.

Table 1 outlines major developments about "central" factors in human muscle fatigue. By the late 19th century it was clear that muscles were adapted for different tasks (e.g., red vs. white muscle) and that muscle performance could be limited by the muscle and also by the neural machinery that drove it (216). Physiologists, including Fick, Fechner, Mosso, and Waller, recognized the steps that could define the extent to which the limitation was muscular. For example, in his influential book Fatigue, Alessandro Mosso (540) knew it would be cogent to compare voluntary performance with that reproduced by external electrical stimulation of the muscles. Unfortunately, his methods of stimulation were not sufficient for the task. Mosso and others adopted additional approaches, one of which relied on the variability of performance of a voluntary task requiring repeated submaximal isotonic contractions. If performance deviated from that expected, then Mosso inferred that the change, usually a deterioration, represented an influence of the CNS. Not only did prior physical exercise diminish performance but so did excessive mental "work" (usually measured in professorial colleagues who had lectured or examined medical students) (Fig. 1A). Mental excitement or agitation could improve voluntary endurance (463). The conclusion was that performance variations reflected central factors, which Mosso believed somehow directly altered peripheral function of the muscle. In analogous fashion, Waller, Lombard, and others noted that the excitability of muscles was sometimes preserved after apparent exhaustion: "that it (fatigue) is in part central is proved by the fact that when cerebral action has ceased to be effective ... electrical stimulation of nerve or muscle is still provocative of contraction" (746) (for similar views see Refs. 463, 607). These propositions were dramatically revisited much later by Ikai and Steinhaus when they showed that the maximal voluntary strength of the elbow flexors could be increased by local cues such as firing a gun before maximal efforts (354) (Fig. 2). Hypnosis also altered performance, and epinephrine injection or ingestion of amphetamine enhanced strength (354).

View larger version (43K): [in this window] [in a new window]   Fig. 1. Early studies of maximal voluntary activation and central fatigue. A: data obtained by Mosso (540) using his ergograph. Trace indicates the distance moved by the middle finger when lifting a weight of 3 kg each 2 s. Normal "fatigue tracing" and one obtained 24 h later "after lecturing" are shown. [Redrawn from Mosso (540).] B: Reid's comparison of a maximal voluntary contraction (MVC) of finger flexion and force produced by stimulation of the median nerve at the elbow (stimulus rate not specified) under isometric conditions. Artificial stimulation produced "as strong contraction as voluntary effort," although this is not clear in all the records (see horizontal dashed line), and Reid often found that large weights could be lifted by artificial electrical stimulation and not volition (see Fig. 4). [Redrawn from Reid (607).]

View larger version (13K): [in this window] [in a new window]   Fig. 2. Classical study of factors affecting maximal voluntary strength. Mean value for brief MVCs of elbow flexors made at 1-min intervals by 10 subjects. Control contractions (open circles), contractions preceded by an unexpected gunshot (solid circles), and a final contraction (star) in which the subject shouted are shown. [Redrawn from Ikai and Steinhaus (354).]

A. V. Hill, in his book Muscular Activity (332), adopted the view subsequently accepted widely by exercise physiologists. Athletes were best for studies of the limits of voluntary performance because "with young athletic people one may be sure that they really have gone `all out,' moderately certain of not killing them, and practically certain that their stoppage is due to oxygen-want and to lactic acid in their muscles. Quantitatively the phenomena of exhaustion may be widely different, qualitatively they are the same, in our athlete, in your normal man, in your dyspnoeic patient." Altough no direct evidence was presented to support this view about going "all out," Hill appreciated the difficulties if this assumption were not justifiable. "If one took a patient from the hospital and made him work till he could barely move, one could never be sure (a) that he had really driven himself to the limitit requires an athlete to know how to exhaust himself; (b) that one would not kill him; and (c) what the cause of his stopping was."

If subjects did go all out, there was the worry derived from clinical cases, that muscles would be torn, tendons ruptured, and bones broken. [We now know that unless there has been a pathological change in the tendon or bone, the strength of bone and tendon exceeds that of muscles (775).] Thus the common view was that the maximal contractile force was so high that "strength is kept in bounds by the inability of the higher centres to activate the muscles to the full" (528). This quandary prompted a milestone in 1954 when two reports appeared. Merton (524) and Bigland and Lippold (59) used electrical stimulation of the motor nerve to compare directly maximal force in a stimulated tetanus and during a maximal voluntary contraction (MVC). While supramaximal "artificial" stimulation of the nerve drives all motor units synchronously, unlike voluntary muscle contractions, this difference is not critical for the comparison provided that the stimulation frequency is high enough to produce a maximal tetanus.

Does the force produced by tetanic electrical stimulation exceed that produced by voluntary action? Both reports claimed that the two forces were similar, although the errors associated with the tests were not quantified and are not trivial (see sect. II). An earlier study by Reid (1928) had also reached this conclusion, although the raw records suggest that voluntary contractions did not quite match the force of an isometric tetanus (Fig. 1B). Using a special myograph that was said to measure the force produced only by adduction of the thumb, Merton compared the response to stimulation of the ulnar nerve above the wrist to the force of voluntary thumb adduction. Difficulties are that small changes in thumb position alter the forces produced by stimulation of the ulnar nerve, and the forces produced by maximal voluntary efforts involve both intrinsic and unstimulated extrinsic hand muscles. In later studies of intrinsic hand muscles Ikai et al. (355) found that tetanic force produced by stimulation exceeded that produced voluntarily, but others have subsequently found that voluntary forces exceeded tetanic force (e.g., Refs. 164, 198, 328, 470) (see sect. II). Mindful of the many experimental difficulties, Merton later commented that his findings came about due to a "happy cancellation of errors" (484). It is unfortunate for muscle physiologists that few accessible nerve-muscle combinations allow isolated supramaximal electrical stimulation of all the motor axons that produce a simple force that is easily measured and mimicked perfectly by maximal voluntary effort.

Merton (524) provided additional insight when he showed that during a voluntary effort the force increment added by a stimulus to the ulnar nerve was inversely proportional to the level of initial force. No force increment appeared when voluntary force approached maximal values (Fig. 3A). Two conclusions were drawn: first, during a maximal effort "those muscle fibres whose motor nerve fibres are excited by the shock are contracting maximally," and second, the relation between voluntary force and the size of the interpolated twitch meant that absolute maximal force could be predicted by linear extrapolation. A corollary to the first conclusion was that during maximal effort, voluntary drive to the motoneurons was sufficient not only to recruit all stimulated motoneurons capable of exerting force, but also to drive them to frequencies that achieve full force. This contribution not only provided a method to investigate maximal voluntary performance, but it placed the limits for voluntary force back where physiologists thought they should be, within the muscle. The technique of twitch interpolation lay unused until the 1980s when Belanger and McComas (45) reassessed its assumptions and applicability for use in the plantar- and dorsiflexor muscles of the ankle. Independently, Grimby and colleagues (300, 301) used brief tetani superimposed on an isometric contraction of extensor digitorum brevis when they estimated optimal firing frequencies to minimize fatigue. Later a special amplifier was developed that improved the resolution of Merton's method 10-fold (312). Submaximal electrical stimulation superimposed during maximal voluntary eccentric contractions of knee extensors was later shown to increase ongoing torque by ~20%, thereby establishing the usefulness of interpolated stimulation for demonstration of suboptimal drive during muscle lengthening (750).

View larger version (11K): [in this window] [in a new window]   Fig. 3. Twitch interpolation of adductor pollicis muscle. A: traces from Merton's original study with vertical gain enhanced. Control twitch produced by supramaximal stimulation of the ulnar nerve and superimposed responses during attempted MVCs. Twitches aligned so that the time of the stimulus (arrow) is coincident, and baseline force before interpolated twitch has been subtracted. [Redrawn from Merton (524).] B: control twitch produced by supramaximal stimulation of the ulnar nerve (control twitch: b) and superimposed responses during attempted MVC (interpolated twitch: a). Voluntary activation (in percent) = 100  100(a/b). The reduction in force after the stimulus-induced force increment is due to nerve and muscle refractoriness, antidromic collision in motor axons, and spinal reflex effects of the stimulus. C: typical thumb twitches elicited by a transcranial magnetic shock to the motor cortex during a weak contraction at 5% MVC or during MVC. [B and C from Herbert et al. (326).]

Finally, Merton applied his twitch interpolation technique in fatigue to establish that "when strength fails electrical stimulation of the motor nerve cannot restore it" (524). He concluded that voluntary effort produced maximal force from muscles and continued to produce it even when peripheral fatigue developed. If the blood supply to the muscle was cut off with a blood pressure cuff after the MVC, then the twitch force did not recover. That is "there is no recovery of (voluntary) strength until the blood flow to the muscles is restored. If fatigue were a phenomenon of the central nervous system, it seems most improbable that recovery should be delayed by occluding the circulation to the arm" (525). The weight of these arguments seemed sufficient for his pronouncement that "anyone with a sphygmomanometer and an open mind can readily convince himself that the site of this fatigue is in the muscles themselves" (525).

If the muscle were the only significant limit to voluntary performance, at least for intrinsic muscles of the hand, then it was natural to check how the muscle was driven by the CNS. The surface electromyogram (EMG) provided an enticing way to examine the role of the CNS in muscle fatigue. Bigland and Lippold (59) had established in 1954 that the integrated surface EMG increased linearly with force (59). There is an obvious increase in the low-frequency content of the signal during fatigue, but this can be fully explained by changes in the compound muscle action potential (419, 534), and thus the ongoing signal provided no certain clue about central changes in motoneuron firing frequency. Changes in the frequency spectrum of the EMG accompany muscle fatigue, but they do not definitively cause it at a peripheral level, nor do they necessarily signify altered neural drive.

The amplitude of the EMG also seemed simple to interpret in terms of "neural drive" in fatigue, but the size and propagation velocity of the intracellular muscle fiber action potential, and possible compromise at the neuromuscular junction, affects the signal. Although the consensus is that blocking at the neuromuscular junction does not occur significantly with natural rates of motor unit firing (e.g., Refs. 61, 70, 247, 704), activity-induced changes in the single fiber potential, and activation of the muscle's electrogenic Na⁺/K⁺ pump to degrees which vary between fiber types and the degree of local ischemia seriously limit the surface EMG as a measure of voluntary activation of motoneurons.

The alternative to measures of global EMG during fatigue was to record the discharge of single motor units in voluntary contractions. Although the principle of motoneuron recruitment and frequency modulation had been known for decades (4), few had succeeded in recording unitary activity during sustained strong contractions due to the interference pattern caused by the discharge of many muscle fibers. Bigland and Lippold (59) recorded from adductor pollicis and abductor digiti minimi with a selective electrode based on two thin flexible wires (insulated to the tips) inserted into the muscle via a hypodermic needle. This method was subsequently adopted widely.

Motor units were believed to be recruited in a relatively stable order according to Henneman's size principle, from those with slow conduction velocity producing small forces to those with fast conduction velocity producing large forces (325) (for review see Refs. 78, 324). This principle links motoneuron properties (i.e., small size, long afterhyperpolarization, and low axonal conduction velocity) with properties of muscle fibers (small twitch force, long contraction time, slow fiber conduction velocity, and low fatiguability). Although the distinctions between the various motor unit types may be more blurred in humans than experimental animals, this "size" principle of orderly recruitment appears to hold in humans, under most circumstances with isometric contractions (e.g., Refs. 181, 242, 537) (cf. Ref. 710), although some exceptions appear to occur during nonisometric contractions (e.g., Refs. 125, 342, 545) and fatigue (228), and for muscles with complex actions and different "task groups" of motor units (e.g., Refs. 615, 737) (for review see Refs. 153, 460). Functionally significant exceptions to the dominant principle of orderly recruitment have been hard to find. However, it must be conceded that different inputs to motoneuron pools, be they reflex (such as group Ia inputs, reciprocal inhibition, cutaneous inputs) or central (such as corticospinal inputs, recurrent inhibition, reticulospinal inputs), do not change the firing frequency of all motoneurons in the pool equally (e.g., Ref. 79) (for review see Refs. 78, 122, 324). For example, the corticospinal input in the cat will increase the firing of high-threshold motoneurons but decrease that of low-threshold motoneurons, thus altering the "gain" of the whole pool (402) and contributing to disrupt the order of motoneuron recruitment. An additional effect of the distribution of motor unit size and fatiguability across the pool is that during a sustained maximal contraction the decline in force will be dominated by fatigue in the large motor units, those recruited late in voluntary contractions. Furthermore, there is probably increased "spacing" between the thresholds of the motoneurons recruited close to maximal voluntary force (28), so that greater "effort" will be needed to generate the final part of the force in maximal efforts.

In 1971, Merton and colleagues (483) made another key observation on motor unit behavior: during a sustained maximal voluntary effort the firing rate of a single motor unit in first dorsal interosseous declined from ~60-80 Hz at the start to ~20 Hz after 30 s (Fig. 4) (483). The shortest initial interspike interval in their studies corresponded to a peak instantaneous frequency of ~150 Hz. They blocked the proximal ulnar nerve, a procedure which removed the major innervation of the muscle and made it easy to isolate the few motor units innervated by the median nerve. (The block also removes much homonymous and heteronymous muscle afferent feedback.) The decline in firing frequency was termed "muscular wisdom" because it matched the firing of the motoneuron to the altered contractile properties of the muscles (see sect. IIIB). It was already known that muscle relaxation slowed in repetitive contractions both in human muscles during strong voluntary efforts and in isolated mammalian muscles, a phenomenon now well characterized (e.g., Ref. 749), so that with the lower fusion frequencies lower firing rates might provide the same activation of the muscle. A direct association between the two phenomena was proposed by Bigland-Ritchie and co-workers (65, 76).

View larger version (50K): [in this window] [in a new window]   Fig. 4. A: records from a single motor unit in adductor pollicis during MVC lasting 1 min. This unit was found after blocking the ulnar nerve at the elbow. Four 2-s sections were removed from the continuous record, covering the intervals 0-2, 8-10, 38-40, and 58-60 s. B: discharge frequency of the same unit during the MVC is shown as the reciprocal of the period occupied by 10 consecutive spikes. Points were plotted at the end of the 10-spike period. A curve has been fitted to the decline in rate. The highest initial peak firing rate was 150 Hz. Arrow and horizontal dashed lines mark a period when discharge rate almost doubled, evidence of prior submaximal voluntary drive. [Redrawn from Marsden et al. (484).]

The decline in maximal voluntary firing rates with fatigue was subsequently confirmed (51, 66, 264, 289, 484, 707). The concept of muscular "wisdom" developed further because the greatest output from muscle (measured as force integrated over time) was generated by stimulation with a declining frequency which began at 50-100 Hz and leveled off at 15-20 Hz after about a minute (e.g., Refs. 301, 375, 483, 484). When rates were too high, failure at the neuromuscular junction and/or sarcolemma developed (e.g., Ref. 69). Meanwhile, other techniques were developed to record the firing of motor units during strong efforts with solid monopolar electrodes (66), fine-wire electrodes (313), and branched bipolar electrodes (227). Improved techniques arose to extract the firing of individual units from the multiunit EMG (e.g., Ref. 173). The major debate concerned the possibility that a reflex arising in the contracting muscle causes the fall in motor unit firing rate.

To those familiar with stimulus rates required for tetanic fusion of mixed muscles of small laboratory animals, the firing rate of motor units in maximal voluntary efforts in humans appeared too low to produce maximal force (224). At least two factors are important: first, as shown elegantly by Rack and colleagues (450, 602), asynchronous stimulation of several groups of motor units produces higher forces than if they all discharge synchronously. However, the magnitude of this effect is not well established. Second, some of the natural variation in firing rate during voluntary contractions produces force more efficiently than regular trains of stimuli, and this may minimize fatigue, due to a series of "nonlinear" intrinsic muscle properties including catchlike behavior (in the cat, see for example Refs. 57, 123, 648, 776; in humans, see for example Refs. 80, 475, 711), twitch potentiation (e.g., Refs. 195, 280, 591), and nonlinear force summation (141, 175, 592). These factors produce hysteresis in the force-frequency relationship (e.g., Refs. 81, 284, 475, 571). Once units are recruited at a high firing rate, near-maximal force can be held with frequencies well below those initially required for fusion. Hence, a lower-than-expected firing rate in sustained voluntary efforts is not, on its own, sufficient to indicate that the muscle is not producing optimal force. The critical factor for muscle fatigue is whether the firing rates of motor units fall too much in voluntary efforts so that submaximal force is produced. Much evidence now supports this view.

This review emphasizes data obtained with isometric contractions because the neural mechanisms can be delineated more easily for them. Many of these are also relevant for concentric and eccentric contractions. After the introductory section, evidence is presented that neural drive is often insufficient to generate true maximal strength. Then, the development of central fatigue is considered along with the spinal and supraspinal mechanisms that are involved. Recent work using noninvasive stimulation of the motor cortex and corticospinal tracts is reviewed. The final section considers briefly other "central" aspects of muscular fatigue including possible neurotransmitter systems and task-related changes in performance.

With a seemingly simple word like "fatigue" it is tempting to assume that, while its meaning is not necessarily understood in the same way by the lay public and physiologists, there is at least agreement on its meaning among physiologists and clinicians. Not so. When applied to muscular exercise, fatigue can refer to "failure to maintain the required or expected force" (217) or failure to "continue working at a given exercise intensity" (90). On this view fatigue would occur suddenly, hence, the phrases to reach "the point" of fatigue or exhaustion. Although the absurdity of this position is obvious, it would follow that fatigue in muscles (and its CNS accompaniments) begins only at the point of task failure when a subject exercises at a set rate to "exhaustion." This sort of definition was emphasized at the influential Ciba Foundation symposium on human muscle fatigue held in London in 1980 (140). In fact, the maximal force-generating capacity of muscles starts to decline once exercise commences so that fatigue really begins almost at the onset of the exercise and develops progressively before the muscles fail to perform the required task. Hence, a more realistic definition of fatigue is "any exercise-induced reduction in the ability to exert muscle force or power, regardless of whether or not the task can be sustained" (74). Because of the potential clinical significance of fatigue of respiratory muscles, a meeting of physicians formally defined muscle fatigue as "a loss in the capacity for developing force and/or velocity of a muscle, resulting from muscle activity under load and which is reversible by rest" (554). Finally, fatigue refers not only to a physiological or pathological state in which muscles perform below their expected maximum, but to a symptom reported by subjects in whom there may be no obvious defect in muscle performance. Indeed, it is the most common symptom in medical and psychiatric practice (331, 486).

Table 2 presents definitions. Because peripheral force-generating capacity usually declines early in exercise and because CNS changes occur before muscles fail to perform a task, the most useful definition of muscle fatigue must encompass, as given above, "any exercise-induced reduction in force generating capacity." Rest reverses it. This definition ignores competing intramuscular mechanisms that potentiate force during "fatiguing" exercise (e.g., Refs. 280, 591, 605, 732) and focuses on the net reduction in performance that ultimately develops. Fatigue can be assessed by measurement of maximal voluntary force, maximal voluntary shortening velocity, or power. Specific tests are required to determine the extent to which any reduction in voluntary capacity is centrally mediated (see sect. II). "Voluntary activation" refers to a notional level of "drive" to muscle fibers and motoneurons. This term is used loosely, often without distinction between drive to the motoneuron pool and that to the muscle. These drives are not the same: one recruits motoneurons and increases their firing, and the other relies on muscle fibers to translate the motoneuron firing into force. As applied to motoneurons, the term voluntary activation is sloppy because it does not specify the source of their excitation (from descending motor paths, reflex inputs, and from associated spinal circuitry). The "maximal evocable force" is that produced when the muscle is fully activated by volition or appropriate electrical stimulation and is the formal term for true maximal force.

During exercise the failure to maintain the initial maximal force depends on "peripheral" fatigue occurring distal to the point of nerve stimulation and on "central" fatigue resulting from a failure to activate the muscle voluntarily. This arbitrarily includes branch-point failure and failure at the neuromuscular junction in the "peripheral" component. That component of overall muscle fatigue dependent on a progressive failure to drive motoneurons (and muscle fibers) voluntarily is termed "central fatigue." It is a progressive reduction in voluntary activation during exercise. Part of this central fatigue is "supraspinal fatigue" because motor cortical output becomes less than optimal (see sect. IV). Eventually, there is "task failure" when the exercise can no longer be continued. This point is often termed "exhaustion" by exercise physiologists. Surprisingly, the neural mechanisms underlying task failure have received little attention from electrophysiologists who are capable of stimulating the neuromuscular apparatus to check the validity of Waller's original claim that exercise stops when effective muscle contraction is still possible (607). Some recent studies have confirmed this (466, 467, 511). An early and a recent example of central fatigue with premature task failure are shown in Figure 5.

View larger version (42K): [in this window] [in a new window]   Fig. 5. Examples of task failure in repeated isotonic contractions and in a sustained submaximal MVC. A: data for maximal isotonic contractions of the flexors of the middle finger (3 kg at 24/min). When the weight could barely be lifted voluntarily (i.e., task failure), it was lifted during median nerve stimulation. Complete rest (open bar) for a minute allowed recovery from both peripheral and central fatigue. It restored voluntary force more readily than repeated tetani (solid bar) that allowed only central fatigue to recover. [Redrawn from Reid (607).] B: torque during sustained contraction of ankle plantar flexors at 30% MVC in which the subject tried to maintain the target force as long as possible. After the first voluntary contraction (at 410 s), the triceps surae muscles were electrically stimulated for 1 min, and the subject was then able to maintain the target torque voluntarily. Top traces show the resting control twitch response and superimposed twitch responses at 7, 396, 473, and 573 s. Arrows indicate timing of interpolated stimuli. Note the increment in force produced with the stimulus interpolated before the end of the main voluntary contraction. At this stage, voluntary activation was not maximal. [Redrawn from Loscher et al. (466).]

It is not possible to specify all the sites within the CNS at which contributions to voluntary activation, central fatigue, and supraspinal fatigue occur. The traditional model for considering muscle performance traces a causative "chain" from high levels within the CNS via descending paths to the motoneuron and then via motor axons to the neuromuscular junction, the sarcolemma, t tubules, and ultimately the actin and myosin interactions (Fig. 6A). A common, but unintended (and illogical), assumption in such a model is that any change at a link in the chain will affect force production. To continue this analogy, the chain is as "weak" as any of its links so that, theoretically, evolution might have ensured that all were equally strong. This does not hold because, for example, in normal subjects during voluntary tasks, events at the muscle cell and at supraspinal levels provide definite limits, while conduction block in (say) motor axons and failure at the neuromuscular junction do not. Of course, diseases and lesions (e.g., myesthenia gravis and spinal cord injury) damage preferentially particular links in the chain.

View larger version (28K): [in this window] [in a new window]   Fig. 6. Steps involved in voluntary force production and factors acting at motoneuronal level. A: diagrammatic representation of the "chain" involved in voluntary contractions. A major source of feedback, that from the muscle, is shown acting at three levels in the central nervous system. Other sources of feedback that also act at these levels are not shown. B: summary of inputs to - and -motoneurons for an agonist muscle. Cells with solid circles are inhibitory. Dotted curved region at premotoneuronal terminals denotes presynaptic inhibition acting selectively on the afferent paths to motoneurons.

Just as it has been important to determine whether activity-induced changes in processes within muscle cells cause fatigue, one must be equally careful to determine whether more "proximal" changes cause, or merely, accompany fatigue. For example, the surface EMG, a direct result of motoneuronal activity, changes during and after exercise, but the changes do not necessarily alter force production (see sect. III). The critical links in the "chain" of Figure 6A operate across the full range of exercise, from when many muscle groups contract, as in cycling or running, to "laboratory" exercise of one muscle or muscle group. The latter form of exercise makes it easier to measure accurately supraspinal and motoneuronal drives with electrophysiological techniques.

	II. EVIDENCE THAT VOLUNTARY ACTIVATION IS SUBMAXIMAL IN "MAXIMAL" EFFORTS

Top Previous Next References

If, in maximal voluntary efforts, the CNS fails to generate maximal evocable force, then a "reserve" exists that could theoretically be called on in exceptional circumstances. Maximal voluntary force would usually be limited by the subject's capacity to activate motor units. The term limit here does not imply that there are no other limits; the force generated by the recruited motor units is also a limit. Two possibilities exist if there is no reserve at the start of maximal exercise: voluntary activation remains maximal during exercise (with a purely peripheral limit to performance), or central fatigue develops. Three possibilities exist if voluntary activation is initially submaximal: it may increase (and thus draw on the reserve), stay the same, or decrease with exercise. Of the five possible outcomes in maximal tasks, in only one (full initial activation with no central fatigue) is the performance limited solely by the muscle.

The isometric force trace itself suggests that maximal voluntary force is less than the maximal evocable force or true maximum. Many myographic recordings show fluctuations in force that are not (or would not be) present in contractions produced by nerve stimulation. Provided the fluctuations are not dominated by variable input from antagonist muscles (which is unlikely, see Refs. 9, 161), or the action of remote muscles stabilizing the proximal skeleton, they reflect changes in output from the agonist motoneuron pool. The magnitude of the fluctuations would then set a lower boundary to the variation in voluntary drive during a contraction.

Many other observations allow inferences about whether muscles are activated maximally by volition. These range from claims of feats of superhuman strength and endurance to the beneficial effects of various ergogenic substances. Claims of the former type require extraordinary documentation and remain unsubstantiated, whereas the latter are beyond the current scope. Some of the influential observations and techniques are considered below. Included are observations made during training, comparison of unilateral and bilateral contractions, and the technique of twitch interpolation.

Many approaches rely on measurements of MVCs, and unless steps are taken to maximize motivation, the contribution of "supraspinal" factors will be magnified. Therefore, when interpreting experimental findings, one should check the following six experimental details. Failure to disclose these methodological details is as unhelpful as failure to perform them.

1)	All maximal efforts should be accompanied by some instruction and practice.
2)	Feedback of performance should be given during the efforts (e.g., clear visual display) rather than delayed until after them.
3)	Appropriate standardized verbal encouragement should be given (e.g., Refs. 58, 514), preferably by the investigators and others, rather than an audio tape.
4)	Subjects must be allowed to reject efforts that they do not regard as "maximal," although with care, this occurs rarely.
5)	In studies that involve repeated testing within a session, or studies in patient groups additional precautions are needed. The gain of any real-time visual feedback should be varied so that the subject or patient is not necessarily aware of the magnitude of any decline in performance, the aim being to maximize performance without necessarily providing a calibrated indicator of it.
6)	With repeated testing over many sessions, weeks, or months, provision of rewards should be considered (326).

These various critical procedures are often ignored. Without attention to such details it is inevitable that voluntary activation will be variable and that it will be submaximal from the outset of the exercise.

Voluntary muscle strength increases with training, but the mechanisms are controversial (for review, see Refs. 90, 223, 225; see also Refs. 503, 507, 539, 637, 638), and there is much intersubject variation (330). Strength may be assessed under controlled conditions such as isometric or concentric isokinetic contractions, or less controlled conditions such as measurement of the largest weights that can be lifted. If the rate of increase in voluntary strength exceeds that attributable to a change in the muscle, then some of the improvement must have occurred in the CNS, either through learning or altered patterns of muscle and motor unit recruitment. This means that in the untrained state, voluntary activation must have been insufficient to produce maximal evocable force, with training improving voluntary activation, at least in the tested task. Thus any increase in force with training can be split into peripheral and central adaptations, with the latter often termed the "neural training effect."

An early study by Scripture et al. in the 1890s (661) illustrates the issues: subjects squeezed a mercury sphygmomanometer "as strongly as possible." The height attained by the mercury was noted. Over the first 2 wk, maximal performance increased steadily by ~70% for the right hand, but it also increased ~40% on the left hand which was tested on only the first and last day. The increase in strength on the side contralateral to training involved learning a useful strategy that could be applied on either side (see also Ref. 161). The earliest phase of strength training involves learning the right pattern of muscle activation, and once learned it can be applied, for example, on the contralateral side. Such learning has a degree of specificity for the precise voluntary task, for example, the angle at which it was undertaken (e.g., Refs. 273, 409, 451) or its velocity (42). The methodological requirements for measurement of maximal voluntary force may explain why not all studies find rapid and highly specific training effects.

To attempt accurate conclusions about the extent of peripheral and central contributions to increased voluntary muscle strength after training, many investigators have relied on measures that might parallel either central drive (e.g., the EMG) or peripheral force (e.g., muscle cross-sectional area and tetanic force). While application of these approaches has often revealed some neural training effect, the arguments are not always straightforward nor the methods watertight. Some relevant arguments are considered below. The final position is that while results from some methods have been overinterpreted, much evidence favors the view that voluntary activation is a key limiting factor in force production before training.

The EMG increases within days or weeks of training. This occurs for various training protocols using isometric, concentric, and other forms of contraction, and it may precede purely peripheral adaptations in the muscles (e.g., Refs. 164, 308-311, 415, 539, 547). The simple explanation is that with training more motor units are recruited or are firing faster. However, local peripheral factors also change the surface-recorded EMG. These include, first, a change in the amplitude of the single fiber action potential (due to a change in fiber size, membrane potential, or sarcolemmal function). Second, the recorded potential may change due to altered electrical conductivity between the electrodes and around the muscle fibers. These factors can be partially controlled by measurement of the compound muscle action potential produced by supramaximal nerve stimulation (M_max). Increased voluntary EMG after dynamic training of a hand muscle was observed (194) in the absence of a change in the maximal muscle action potential, but others found a parallel increase in voluntary EMG and M_max after quadriceps training (611). Many factors including changes in fiber type, intramuscular ionic concentrations, and sodium-potassium pump content will alter the muscle fiber action potential (e.g., Ref. 665). Third, the contribution of far-field EMG sources to the signal may change due to altered drive to synergist or antagonist muscles. Intramuscular recordings from these muscles would be instructive here.

If the above factors are controlled, then a real change in the surface EMG after training will reflect a change in the neural drive to the muscle. However, the translation of a real increase in surface EMG into force also needs scrutiny. For example, an increase in the number of doublet discharges (i.e., discharges with interspike intervals of ~5 ms) (38, 178) during the sustained phase of a maximal task may not increase force, whereas doublets during the initial concentric phase of a task will increase the shortening velocity and rate of force production. Initial doublets may minimize later force declines (75) (cf. Ref. 475). The propensity to fire doublets is probably largely regulated at the motoneuron. The contraction times of the motor units may shorten due to training, and thus fusion would need a higher rate of discharge. Then, increased surface EMG would not necessarily signify an enhanced ability to extract muscle force. Thus, in terms of both the peripheral EMG signal and the force-EMG relationship, an increase in the surface EMG is not an invariant index of an increase in voluntary activation after training. This consideration should be extended to measurements of reflexes after training (e.g., Refs. 536, 639).

In a few studies single motor units have been studied after training. With months of practice, the ability to sustain the discharge of high-threshold motor units in the toe extensors improved and central fatigue declined (301). Both observations point to an ability to improve the neural drive to muscle in a strong contraction. After 12 wk of dynamic training of ankle dorsiflexor muscles, whole muscle contraction time was unchanged, but the MVC and speed of ballistic contractions increased (729). The percentage of doublet discharges increased sixfold (from 5 to 30% of the units), and the maximal firing rates at the onset of ballistic efforts increased. Less direct evidence suggests that the maximal rate of force development is usually limited by the ability to deliver asynchronous motor unit discharges with short initial interspike intervals (see Refs. 484, 532).

Increases in maximal voluntary force after training have been compared with increases in cross-sectional area of muscle. The anatomical cross-sectional area of muscles apparently does not increase as much as maximal voluntary strength (e.g., Refs. 353, 369, 376, 472, 547). In a review, Booth and Thomason (90) estimated that cross-sectional area increases by 0.1%/day with strength training, whereas an estimate across studies reveals that voluntary force typically increases by ~1%/day. If the training period is brief, this suggests that voluntary activation was incomplete before training (90). However, not all studies confirm this disproportionate change, and the argument is indirect (e.g., Refs. 376, 494, 547). It assumes that measures of anatomical cross-sectional area are a valid measure of the force-generating capacity, a property which depends on physiological cross-sectional area and specific tension. Physiological cross-sectional areas are two to eight times the anatomical cross-sectional areas in leg muscles (252). Ideally, corrections should be included for changes in muscle fibre type, specific tension, changes in architecture of both muscle fibers and tendon, changes in length-tension properties of muscle fibers, and changes in nonmyofibrillar components of the muscle cross section (e.g., Refs. 390, 493, 494; cf. Ref. 166). It is difficult to estimate the cumulative effects of the necessary corrections. Any could cause large errors in the assumed proportionality between a muscle's apparent cross-sectional area and the force at the tendon.

Because the response to nerve stimulation bypasses volition, such responses have been checked before and after training. Maximal voluntary force has been reported to rise more than the increase in tetanic force in a number of studies (e.g., Refs. 164, 506, 547, 772). Data from a typical study are shown in Figure 7. Force evoked by artificial stimulation is not the same as that produced volitionally in more ways than just the lack of intervention of the will. As is critically important during sustained efforts, voluntary contractions use natural motor unit firing patterns that may enhance force rises and reduce force declines. However, voluntary contraction also allows full use of synergist muscles and those responsible for stabilization of proximal joints. These are not usually included in the artificial stimulation. If trick movements require attention in evaluation of patients with nerve lesions, the same caution must apply to those neurologically intact performing strength tests.

View larger version (13K): [in this window] [in a new window]   Fig. 7. Training can induce a greater increase in maximal voluntary isometric force than force produced by electrical stimulation. MVC of first dorsal interosseous (solid circles) is compared with the tetanic force produced by electrical stimulation over the motor point (open circles). Maximal tetanic force increased by 11% after 8 wk of strength training (80 MVCs/day), but maximal voluntary force increased more, by 33%. Forces have been normalized to their initial pretraining values. Values are group mean data ± SE. [Redrawn from Davies et al. (164).]

Factors needing careful control for evoked force measures include the following: the degree of potentiation (especially of the twitch), the duration of the tetanus (peak forces may not be reached in short tetani), the maximal frequency of stimulation, and the unintended stimulation of antagonists. For example, ulnar stimulation activates both abductors and adductors of the fingers, and common peroneal nerve stimulation activates ankle plantarflexors as well as dorsiflexors. In many studies, the frequency and level of stimulation are limited by the tolerance of the subject. Mosso (540) long ago reported his unwillingness to inflict a painful current and he "had not the heart to use it, in spite of the devotion" of his subjects. Supramaximal stimuli that avoid antagonists are best. Tetani at submaximal intensities will not necessarily test the motor units used in training because weak stimuli preferentially excite large motor axons (with high thresholds for voluntary activation). Furthermore, if submaximal intensities are used, different motor units will be activated in different trials due to intrinsic and extrinsic variations in axonal threshold. Fortunately, the limit imposed by painful stimulation is less when stimuli are delivered during strong contractions, as with twitch interpolation, presumably because sensory "gating" at cortical and subcortical levels makes the same stimulus much more bearable.

If imagination of exercise improves voluntary strength, then voluntary activation had to be submaximal before "training," and it was improved by mental rehearsal. Yue and Cole (773) evaluated the effect of 4 wk of imagined training on the MVC of abductor digiti minimi, an intrinsic hand muscle. This muscle was selected because it is rarely used in large sustained efforts. Voluntary strength increased by 22% in those undertaking imagined training and 30% in those training with real contractions (statistically equivalent increases), while there was no increase in a control group. Twitch force of abduction produced by ulnar nerve stimulation and the force of voluntary toe extension were unchanged in all groups. Voluntary strength increased with both real and imagined training on the contralateral nontrained side by ~10%. Voluntary EMG (normalized to the maximal compound muscle action potential M_max) increased with real training, but not imagined training, and it did not increase on the contralateral side. Because the increases in voluntary abduction force were poorly correlated with the force of flexion of the little finger, they are unlikely to reflect hypertrophy of abductor digiti minimi produced by imagined training. This study provides unequivocal evidence that voluntary efforts of the tested muscle did not produce maximal evocable force before training.

An attempt to reproduce this effect for the elbow flexor muscles failed (326). Imagined training for 8 wk produced only a small increase in force. Maximal voluntary activation measured with twitch interpolation was high before the training (~96%) and did not increase after it. This finding has been noted for quadriceps (316; see also Ref. 376), although the sensitivity of the measures was lower. The findings suggest that any increase in force that accompanies training for these muscle groups reflects changes in the muscle. Subjects undergoing real training increased strength significantly more (18%) than those performing no training (6.5%) or imagined training (6.8%). There is no obvious technical reason for the discrepancy in the two studies. However, a likely explanation is that maximal voluntary activation is lower for some intrinsic hand muscles than for elbow flexors (327). Hence, a greater central reserve, accessible by real or imagined training, exists for intrinsic hand muscles.

Support for the view that voluntary activation is limited also comes from studies of unilateral and bilateral contractions. If, as commonly reported, the MVC during bilateral contractions is less than the sum of the forces produced in unilateral MVCs, then voluntary activation in the bilateral task is deficient (e.g., Refs. 341, 423, 663, 733, 734). Training, familiarization, and the actual task may reduce this bilateral "deficit" (341, 629, 663; cf. Refs. 632, 653). The ratio of the bilateral maximal force to the summed forces from each side may be as low as 75%, but is usually ~90%. Because measurements of force and EMG do not necessarily give concordant results for the size of the deficit (e.g., Refs. 341, 653), biomechanical constraints in the bilateral task (especially contraction of remote stabilizing muscles) must be relevant. However, neural factors are involved. For example, direct interhemispheric connections and subcortical pathways can contribute to mediate an inhibitory interaction among homologous muscles (e.g., Refs. 188, 282).

Two recent studies have examined in detail the behavior of muscles activated in bilateral MVCs. In one, the voluntary activation of adductor pollicis was studied during simultaneous contralateral contractions of the homologous hand muscle or of the contralateral elbow flexors (327). Bilateral maximal contractions of the elbow flexors were not performed due to the difficulty in stabilization of the subject (cf. Ref. 565). Voluntary activation of the thumb adductors was ~90% (based on twitch interpolation) and unchanged when the subject produced maximal force by elbow flexion on the contralateral side. However, it diminished slightly, but significantly (by ~1.5%), when both thumb adductors contracted simultaneously, and, as would be expected, maximal force showed a similar difference. Furthermore, voluntary activation changed in parallel for both muscles in simultaneous maximal efforts (Fig. 8). This positive correlation between voluntary activation on the two sides did not occur for simultaneous efforts of the thumb adductor and elbow flexors. It is as if supraspinal drive changes together for the muscle pair, rather than activation on one side being at the expense of that on the other side. When unilateral and bilateral contractions of the quadriceps were performed with measures of force, surface EMG, motor unit firing rates, and voluntary activation, there was no support for a significant deficit in bilateral performance (361). There were no confounding effects due to cocontraction of antagonist muscles.

View larger version (18K): [in this window] [in a new window]   Fig. 8. Bilateral contraction of adductor pollicis assessed with twitch interpolation. Voluntary activation was measured in MVCs of thumb adduction on the left and right sides using twitch interpolation with supramaximal stimulation of the ulnar nerve at the wrist. The higher measure of maximal voluntary activation is plotted against the lower one achieved concurrently in opposite limb. The level of voluntary drive was highly positively correlated on the two sides when subjects simultaneously contracted the thumb adductors bilaterally in MVCs. Data are from contractions in all subjects with each symbol representing one subject. [From Herbert et al. (326).]

In summary, although a deficit in force production may exist for bilateral contractions of some homologous muscles, the effect can be relatively small. Moment-to-moment performance of homologous muscles may be correlated due to variations in supraspinal drive.

As indicated in section IA, the technique of twitch interpolation seemed to measure voluntary activation of muscle. Merton (524) reported that the increment in force produced by a supramaximal stimulus to the ulnar nerve supplying adductor pollicis diminished linearly as voluntary force increased and that at maximal voluntary effort no additional force was evoked (524). The proportional decline with increasing central drive has been confirmed in animal studies with twitches interpolated to the diaphragm during increasing central respiratory drive, but the linear decline was only examined over a small range because the highest neural activation that could be obtained was well submaximal (186, 233). Merton's original claims can now be assessed in detail.

Usually voluntary activation is derived by the formula: voluntary activation = 100(1  T_interpolated/T_control), where T_interpolated is the size of the interpolated twitch and T_control is the size of a control twitch produced by identical nerve stimulation in a relaxed potentiated muscle. If evoked forces are measured at high gain with the voluntary force offset using a special amplifier (312), then small force increments can frequently be measured in response to single stimuli delivered during attempted MVCs. With this conventional formula, voluntary activation could exceed 100% if interpolated twitches had negative values. This is physiologically unreasonable, and MVCs with a rapidly declining baseline are best discounted (for discussion, see Ref. 328).

With Merton's original system, force increments of ~5% of the resting twitch could not be easily resolved, whereas modern systems can resolve below 1% of the twitch (261), which may be below 0.1% of the MVC. Figure 3 enhances the resolution of Merton's system and contrasts it with that in a recent study of the same muscle (327). Voluntary activation was not usually complete for MVCs of thumb adduction tested with supramaximal stimulation of the ulnar nerve. A similar conclusion was reached using motor cortical stimulation, although the measures of voluntary activation with the two forms of stimulation are not comparable (see sect. IV). Twitch interpolation is ideally suited to estimate voluntary activation when all muscles contributing force are stimulated. As indicated below this situation occurs rarely (if ever). Thus it does not apply for thumb adduction, although the adductor pollicis does provide the majority of the torque (681).

Twitch interpolation has now been applied in various ways for examination of maximal voluntary activation in a range of muscles, including abdominal muscles (269, 697), abductor digiti minimi (266), adductor pollicis (51, 65, 66, 69, 72, 215, 327, 524, 672), ankle plantarflexors (45), biceps brachii (9, 10, 13, 14, 22, 51, 179, 192, 260, 266, 326, 327, 459, 514, 551, 564, 633, 702), brachioradialis (13), diaphragm (14, 48, 49, 201, 267-269, 431, 511, 514, 516, 530, 674), extensor digitorum brevis (300), first dorsal interosseous (712), masseter (471), quadriceps femoris (e.g., Refs. 103, 104, 283, 316, 349, 361, 376, 438, 553, 562, 610, 616, 621, 632-634, 690, 695, 722, 743), soleus (51), tibialis anterior (44-46, 266, 397-399, 712), and triceps brachii (709). In many studies insensitive forms of twitch interpolation have propped up the view that voluntary activation is "maximal" and therefore that measures of voluntary force (motor unit firing rates) are also roughly "maximal." However, other studies have found definite evidence that additional stimulation produces more force than can be achieved with volition. Furthermore, a realistic model incorporating measured muscle properties and motor unit firing rates confirms even at maximal voluntary force small superimposed twitches should be evident (Fig. 9A) (328).

View larger version (26K): [in this window] [in a new window]   Fig. 9. Real and simulated twitch interpolation for adductor pollicis. A: simulated and experimental interpolated twitches of adductor pollicis. Twitches obtained at rest and during contractions to ~34, ~82, and 100% MVC. Twitches are aligned with onset of force (shown by arrow) and baseline force preceding the stimulus subtracted. Thick line shows simulated data, and thin lines show experimental data. Note that the twitch was nearly but not fully extinguished during the "maximal" contractions. B: simulated force responses to stimuli interpolated at varying levels of "excitation" of the motoneuron pool. Force responses of the whole muscle when excited from 0 to 100% of maximal excitation are shown. Stimulus is delivered at the arrow. Note the small force increments at excitations which correspond to maximal observed motor unit firing rates. C: amplitude of interpolated twitches (expressed as a percentage of resting twitch amplitude) at contraction intensities of 0-100% MVC. Experimental data are shown, as well as simulations with single and paired interpolated stimuli. D: force response of the 10th, 40th, 80th, and 110th motor units (with units numbered in order of size) at rest and at 34, 85, and 100% MVC. Stimuli were delivered at the arrows. Force scale differs between panels. Note the importance of high-threshold motor units in generation of the interpolated twitches. [From Herbert and Gandevia (328).]

Few studies using twitch interpolation have compared voluntary activation during maximal efforts of different muscles. Belanger and McComas (45) found that ankle plantarflexors could not be activated as fully as ankle dorsiflexors in attempted MVCs. This difficulty in voluntary activation of plantarflexors (particularly soleus) has been confirmed by others using twitch interpolation (63) and is familiar to electromyographers who coax activity from this muscle in patients. Thus voluntary activation varies between muscles. Corticospinal connections may offer a partial explanation for the muscles acting around the ankle because the predominant initial effect of cortical stimulation on soleus motoneurons is inhibition (e.g., Refs. 98, 154). Maximal voluntary activation of an intrinsic hand muscle (adductor pollicis) is lower than that for the elbow flexors when tested in the same session in the same subjects (329). Similarly, voluntary activation of the diaphragm in maximal inspiratory tasks is also slightly lower than that of the elbow flexors (14). Maximal activation of elbow flexor muscles with radial innervation is less than that of those with musculocutaneous innervation with the forearm supinated (13). Twitch interpolation has revealed that voluntary activation of some muscles is often well submaximal in attempted maximal voluntary efforts, for example, in the jaw closers (471) and the abdominal muscles (269, 697). No muscle appears blessed with truly optimal drive during maximal isometric efforts. Given the variety of muscles tested (proximal, distal, truncal, upper limb, and lower limb), this conclusion suggests that no central or peripheral specialization of a muscle protects it from some failure of voluntary activation.

How should voluntary activation in MVCs be reported? The median of trials in a subject provides the best measure of central tendency as the data are usually not distributed normally, with voluntary activation being constrained to values at or below 100%. Results from subjects tested repeatedly are shown in Figure 10. An alternative is to average the evoked responses from acceptable trials, but the average is unduly influenced by single trials with poor activation (266).

View larger version (15K): [in this window] [in a new window]   Fig. 10. Reliability of measures of maximal voluntary activation measured with twitch interpolation for elbow flexor muscles. Data shown for each MVC were performed by 5 subjects studied on 5 separate days. Voluntary drive was reproducible but differed significantly between subjects. [From Allen et al. (9). Copyright 1995 John Wiley & Sons, Inc.]

Another way in which data obtained from twitch interpolation have been used to infer that voluntary activation produces submaximal force is to extrapolate the relationship between the size of the superimposed twitch and the voluntary force at which it was obtained. The use of twitch interpolation to predict maximal voluntary force was first noted by Merton (524) and has been applied by others for the diaphragm (e.g., Refs. 48, 49, 269), quadriceps (e.g., Refs. 43, 562, 633), elbow flexors (e.g., Refs. 179, 192), and jaw closers (471). Unfortunately, many experimental errors and physiological realities conspire to make it difficult to define a simple linear function between a force increment to superimposed nerve stimulation and the absolute level of "drive" to the muscle or its motoneurons. These are summarized in Figure 11. They include failure to use a fully potentiated twitch, a compliant myograph (e.g., Ref. 465), antidromic collision and axonal refractoriness (328), nonlinearities between firing frequency and force, recruitment of synergists (13), and use of a stimulus that inadvertently activates antagonist muscles (22, 108) or is submaximal.

View larger version (24K): [in this window] [in a new window]   Fig. 11. Factors affecting the relationship between the size of an interpolated muscle twitch to a single stimulus and the level of voluntary isometric force. The inverse relationship between voluntary force and the size of the superimposed twitch can be markedly distorted by experimental factors. It is drawn as a simple linear relation (see Ref. 328), with the twitch contraction of the muscle being 10% of the maximal voluntary force. Thus the initial twitch force will be lower than it should if the muscle twitch is not potentiated and if the myograph and connections to it are not stiff at low forces. Additional factors that reduce the twitch force during stronger contractions include antidromic collision (and muscle/nerve refractoriness). The size of superimposed twitches will also be reduced if any antagonist muscles are inadvertently stimulated. Other factors that can influence the precise shape of the relationship include the recruitment of additional synergists (not in direct proportion to the agonist) and failure to prevent small changes in length of the tested muscles. The dotted line shows the increase in slope of the relation when paired stimuli are used. The distortions can be accentuated when many muscles evoke the voluntary force and the test stimulus is submaximal. The arrow indicates the direction of change and not its magnitude. Most of the distortions occur across the full range of voluntary forces. The figure emphasizes that many factors can distort the relationship between voluntary force and the superimposed twitch (i.e., voluntary activation).

The common equation for voluntary activation has some predictable consequences. First, the value of activation depends on the size of the control twitch. Given that a motor unit recruited in an MVC will be potentiated, the control twitch should also be potentiated. In practice, the control twitch is produced a short time (<5 s) after the maximal effort. The assumption that the degree of potentiation (combined with any fatigue from the MVC) is equivalent for the interpolated and the control twitch has not been formally tested. Second, when fatigue develops, it preferentially drops the forces produced with isolated single twitches (and at low frequencies, so-called "low-frequency" fatigue), an effect ascribed to impaired excitation-contraction coupling (e.g., Refs. 152, 215). This means that voluntary activation may erroneously appear to deteriorate; a practical solution is to use paired stimuli that show a decline in force with fatigue that parallels the decline in maximal voluntary force. Third, if it is to be used quantitatively, the technique relies on stimulation of the same motor axons in the twitch and during the contraction. This cannot be reliably ensured unless supramaximal stimuli are delivered. With repeated contractions and fatigue, motor axons are hyperpolarized so that fewer are recruited with the same stimulus intensity (55, 723).

A natural modification of "twitch" interpolation was to increase the size of the "signal" by delivery of more than one stimulus (either a pair, or a brief tetanus, Fig. 9C) (267, 301, 328, 552, 692; see also Ref. 329; cf. Ref. 43). Superficially this is attractive and, at least for paired stimuli (10-ms interval), the induced errors are trivial when analyzed in a detailed model (328). However, as the number of stimuli and duration of stimulation increase, there are unwanted consequences due to antidromic activation of motoneurons and Renshaw cells, plus reflex effects on synergists and remote muscles. When applied to some elbow flexor muscles at high voluntary force (>80% MVC), superimposed pulse trains do not always produce a predictably larger force, perhaps due to an inhibitory reflex involving synergists. For quadriceps, superimposed submaximal tetanic stimulation increased the signal, but this was more than offset by an increase in the "noise" of the background force (552). Even though submaximal tetanic stimulation fails to activate all relevant axons for quantitative twitch interpolation, an evoked force increment is nonetheless unequivocal evidence of a failure of voluntary drive.

At least for some muscles, the relationship between the response to the interpolated stimulus and voluntary force seems nonlinear at high forces. Maximal voluntary forces are produced that are higher than predicted from linear extrapolation of data obtained at lower forces. An inference is that maximal evocable force greatly exceeds that observed or predicted by linear extrapolation if twitch interpolation becomes insensitive as a measure of voluntary activation at high voluntary forces. Recent experimental and modeling studies have addressed this issue.

Single, pairs, and sets of four stimuli were applied to biceps brachii during graded voluntary contraction of the elbow flexors in a protocol that minimized the effects of potentiation (13). The size of the evoked twitch(es) in biceps/brachialis decreased in proportion to the voluntary force until ~80% MVC where the slope became less steep. At least two factors were shown to contribute to this nonlinearity. First, based on twitch interpolation of radially innervated elbow flexors, these synergists were not activated proportionately, being less well activated at high voluntary forces than biceps brachii. Second, despite a near rigid myograph and attempts to fix the shoulder, it was impossible to eliminate some initial shortening and then lengthening of the elbow flexor muscles. The adjective isometric, frequently used to describe such contractions, is somewhat unsatisfactory; not only do sarcomeres shorten and thereby lengthen tendons, but whole end-to-end length is not maintained. It was concluded that factors other than voluntary activation of biceps contributed to the nonlinear behavior at high voluntary forces. A third technical factor is that if excessive currents are used to stimulate biceps and brachialis, the weaker antagonist elbow extensor muscles inadvertently contract (22, 108). This would truncate the diminishing twitches of the elbow flexors and result in abolition of a force increment at submaximal voluntary forces.

One study that shows a markedly concave relationship between interpolated force increment and voluntary force used a large