Motor unit recruitment and fatigue: slow versus fast twitch, the size principle, and graded force

Intuition Beginner

Your muscles are not all the same. Inside every skeletal muscle lie two main kinds of muscle fibres, mixed together like threads in a rope. Slow-twitch fibres (also called Type I) contract slowly but can keep going for a very long time. They are packed with mitochondria and blood vessels, giving them a red colour. Marathon runners rely heavily on these fibres.

Fast-twitch fibres (Type II) contract quickly and generate a lot of force, but they tire rapidly. They are paler because they have fewer mitochondria and less blood supply. Sprinters depend on these fibres for explosive power. Most people have a mix of both kinds throughout their muscles.

When your brain tells a muscle to produce a small amount of force, only a few small motor units activate. A motor unit is one nerve cell (motor neuron) and all the muscle fibres it controls. Small motor units tend to contain slow-twitch fibres. As you need more force, the nervous system recruits larger and larger motor units, adding fast-twitch fibres on top of the already-active slow ones.

This ordering rule is called the size principle: small motor units are recruited first, large ones last. It is why you can pick up a coffee cup smoothly and then instantly switch to heaving a heavy suitcase — the brain adjusts how many motor units it calls on, starting small and scaling up.

Muscles also fatigue. Slow-twitch fibres resist fatigue because they produce ATP aerobically (using oxygen). Fast-twitch fibres burn through their energy reserves quickly and accumulate metabolic by-products that impair contraction. When a muscle fatigues, you feel it as a loss of force even though you are trying just as hard.

Visual Beginner

The diagram above shows three motor units of increasing size within a single muscle. The small unit fires first during gentle movements. The medium unit joins for moderate effort. The large unit activates only when maximal force is needed.

Worked example Beginner

A muscle has 200 motor units. The smallest unit produces 0.5 mN of force; the largest produces 100 mN. You want to hold a 2 N (2000 mN) weight steady.

The nervous system recruits motor units in order from smallest to largest. If we assume forces increase roughly linearly from smallest to largest, the average unit contributes about $(0.5+100)/2\approx 50$ mN. To reach 2000 mN you need roughly $2000/50=40$ motor units active — about 20% of the total pool. The remaining 160 units (including the most powerful fast-twitch units) stay in reserve, ready to fire if the load increases.

This illustrates two key ideas: the size principle ensures smooth, fine-grained force control at low forces (because only small units are active), and large force reserves are available for emergencies.

Check your understanding Beginner

Formal definition Intermediate+

Motor unit anatomy

A motor unit consists of a single alpha-motoneuron (cell body in the ventral horn of the spinal cord, axon projecting through the peripheral nerve to the muscle) and all the muscle fibres innervated by the terminal branches of that axon. When the motoneuron fires an action potential, every fibre in the motor unit contracts nearly simultaneously. The motor unit is the irreducible quantum of voluntary force production.

Motor unit size varies dramatically across muscles. The extraocular muscles have motor units innervating as few as 5-10 fibres (fine angular control of the eye). The gastrocnemius has units innervating 1000-2000 fibres (coarse but powerful force). Within a single muscle, motor units span a range of sizes, with smaller units producing less force and larger units producing more.

All fibres within a single motor unit are of the same histochemical type. This was established by the cross-reinnervation experiments of Buller, Eccles, and Eccles (1960): switching a fast nerve to a slow muscle (and vice versa) converts the fibre type over weeks, demonstrating that the motoneuron's firing pattern and trophic signals determine fibre-type identity.

Fibre-type classification

Mammalian skeletal muscle fibres are classified into three principal categories based on myosin heavy-chain (MyHC) isoform, ATPase activity, and metabolic profile:

• Type I (slow oxidative): MyHC-I isoform, low ATPase rate, slow contraction speed ($v_{\max }\approx 1-2$ fibre-lengths/s), high mitochondrial density (10-15% of cell volume), high capillary supply, high myoglobin content, fatigue-resistant. Dominant in postural muscles and endurance-trained muscle.

• Type IIa (fast oxidative-glycolytic): MyHC-IIa isoform, intermediate ATPase rate, moderately fast contraction speed ($v_{\max }\approx 3-5$ fibre-lengths/s), moderate mitochondrial and capillary density, moderately fatigue-resistant. Functions as a bridge between pure endurance and pure power.

• Type IIx (fast glycolytic): MyHC-IIx isoform (MyHC-IIb in rodents, largely absent in human limb muscle), high ATPase rate, fast contraction speed ($v_{\max }\approx 5-8$ fibre-lengths/s), low mitochondrial density (2-4% of cell volume), sparse capillary supply, fatigues rapidly. Dominant in muscles used for brief, powerful bursts.

Many human fibres co-express two adjacent MyHC isoforms (e.g., I/IIa or IIa/IIx hybrids), creating a continuous phenotypic spectrum rather than three discrete categories.

The size principle (Henneman 1957)

Henneman's size principle ^{[Henneman 1957]} states that motor units are recruited in order of increasing motoneuron size (and therefore increasing force output) as net excitatory drive to the motoneuron pool increases. The principle rests on the biophysics of the motoneuron: smaller somata have higher input resistance ($R_{\mathrm{in}}\propto 1/\mathrm{surface}\mathrm{area}$), so a given synaptic current produces a larger depolarisation in a small motoneuron than in a large one. The small motoneuron reaches threshold first.

Because motoneuron size correlates strongly with fibre type (small motoneurons $\to$ Type I units; large motoneurons $\to$ Type IIx units), the size principle enforces a recruitment order: slow, fatigue-resistant units fire first during low-intensity activity; fast, fatiguable units fire last, only when high force is demanded. Derecruitment follows the reverse order (large units drop out first).

The size principle holds across muscle groups, across reflex and voluntary drives, and across species. Exceptions exist (selective recruitment of fast units during rapid perturbations, task-dependent reordering in some reflex arcs) but these are modulatory overlays on a principle that governs the overwhelming majority of voluntary and postural contractions.

Rate coding and graded force

Once a motor unit is recruited, force can be increased by raising its firing rate (rate coding). A single action potential produces a twitch lasting 20-150 ms (fibre-type dependent). At low firing rates (6-10 Hz), successive twitches are separated by partial relaxation. As rate increases, twitches summate (temporal summation), and force rises toward the fused tetanus plateau at 40-60 Hz for slow units and 80-120 Hz for fast units. Tetanic force may be 3-5 times the single-twitch amplitude.

Whole-muscle force during voluntary contraction is the sum over all active motor units:

$$F(t)=\sum _{i=1}^N{r}_i(t)\cdot f_i(L,v)$$

where $r_i(t)$ is the instantaneous firing rate of unit $i$ and $f_i(L,v)$ is the force-length-velocity function of unit $i$ at the current fibre length $L$ and shortening velocity $v$. Force is graded by two independent mechanisms: recruitment (bringing new units online) and rate coding (increasing the firing rate of already-recruited units).

The relative contribution of recruitment versus rate coding differs across muscles. In small distal muscles (e.g., intrinsic hand muscles), recruitment is complete by roughly 50% of maximal voluntary contraction (MVC), and further force is generated by rate coding alone. In large proximal muscles (e.g., biceps, quadriceps), recruitment continues up to 80-85% MVC.

Contraction types

• Isometric contraction: the muscle generates force without changing length. Force depends on activation level and the length-tension relationship.
• Isotonic contraction: the muscle shortens (concentric) or lengthens (eccentric) against a constant load. Shortening velocity depends on the load via the Hill force-velocity relation (see 18.04.02).
• Isokinetic contraction: the muscle shortens at a constant velocity regardless of force, typically imposed by a dynamometer in laboratory settings.

Muscle fatigue mechanisms

Fatigue is the activity-induced decline in force production despite sustained neural command. It is not a single mechanism but a cascade across multiple sites:

Peripheral fatigue (within the muscle):

1. Metabolic depletion: phosphocreatine stores are depleted within seconds of maximal effort, reducing the immediate ATP-buffering capacity. Glycogen depletion limits prolonged submaximal effort.
2. Accumulation of metabolites: inorganic phosphate ($P_i$) accumulates and shifts the cross-bridge equilibrium toward pre-power-stroke states, reducing force per cross-bridge. Hydrogen ions ($H^{+}$) from glycolytic ATP production lower intracellular pH, inhibiting myosin ATPase and sarcoplasmic reticulum calcium release. ADP accumulation slows cross-bridge detachment kinetics.
3. Impaired calcium handling: repeated contractions impair sarcoplasmic reticulum calcium release (possibly via RyR modification and reduced SR calcium content) and reuptake (SERCA inhibition), reducing the calcium transient that triggers each contraction.

Central fatigue (neural origin):

4. Descending drive reduction: supraspinal centres reduce voluntary drive during sustained maximal effort, measurable as an increase in force produced by supramaximal twitch interpolation (the muscle can produce more force than the voluntary command elicits).
5. Reflex inhibition: group III/IV muscle afferents activated by metabolites inhibit alpha-motoneurons via spinal interneurons, reducing net excitatory drive.
6. Neuromuscular transmission failure: at very high firing rates, presynaptic acetylcholine release may fail to keep pace with demand, though this is rare in healthy muscle.

The time course of fatigue ranges from seconds (high-intensity, phosphocreatine-dominated) to minutes (glycolytic, metabolite-accumulation) to hours (glycogen-depletion, ultramarathon). Recovery follows an inverse pattern: phosphocreatine resynthesis takes 2-5 minutes; metabolite clearance takes 10-30 minutes; glycogen repletion takes 24-48 hours.

Key mechanism Intermediate+

The force-frequency relationship

The relationship between stimulation frequency and force in a single motor unit is sigmoidal. At low frequencies (1-5 Hz), each stimulus produces a discrete twitch with full relaxation between twitches. Force equals the peak twitch force. As frequency rises into the unfused-tetanus range (10-30 Hz for slow units, 20-50 Hz for fast units), successive twitches overlap and force rises steeply on the rising limb of the sigmoid. Above the fusion frequency (roughly 30-60 Hz for slow, 60-100 Hz for fast units), twitches fully fuse and force plateaus at the tetanic maximum.

The sigmoidal shape arises from the convolution of the stimulus train with the twitch impulse response. If the twitch kernel is $h(t)$, a stimulus at rate $r$ produces steady-state force:

$$F_{\mathrm{ss}}(r)=\sum _{k=0}^{\infty }h(k/r)$$

which grows monotonically with $r$ and saturates when successive twitches fully overlap. The steep portion of the sigmoid corresponds to the transition from discrete twitches to partial fusion — the operating range where small changes in firing rate produce large changes in force. This is the range used most during voluntary contraction.

The force-frequency curve shifts left (lower frequencies needed for a given fraction of tetanic force) in fatigued muscle because the twitch duration prolongs (slower calcium reuptake and slower cross-bridge kinetics), causing twitches to fuse at lower frequencies. This leftward shift partially compensates for the reduction in peak tetanic force during fatigue.

Recruitment-threshold scaling and the Henneman model

Henneman's original 1957 observation was phenomenological: in decerebrate cat preparations, progressively increasing afferent input to the ventral horn produced a stereotyped recruitment order across motoneurons. The 1965 quantitative follow-up with Somjen and Carpenter showed that the recruitment threshold current was monotonically related to motoneuron soma size, measured by antidromic latency and input resistance.

The model treats the motoneuron pool as a set of integrate-and-fire elements with distributed thresholds ${\theta }_1<{\theta }_2<\cdots <{\theta }_N$, where ${\theta }_i$ increases with soma size. Net synaptic input $I(t)$ drives the pool, and unit $i$ is recruited when $I(t)$ first exceeds ${\theta }_i$. The total force is:

$$F(t)=\sum _{i:{\theta }_i\le I(t)}{f}_i(r_i(t),L,v)$$

The monotonicity of the recruitment order is preserved because ${\theta }_i$ is a deterministic function of soma size. Modulatory inputs (descending monoaminergic pathways, Renshaw cell recurrent inhibition, presynaptic inhibition of Ia afferents) can shift individual thresholds but do not reverse the ordering in normal conditions.

Fatigue as a multi-site cascade

The temporal sequence of fatigue mechanisms during a sustained maximal voluntary contraction (MVC) of a mixed muscle proceeds as follows:

1. 0-30 s: phosphocreatine depletion and $P_i$ accumulation reduce force per cross-bridge. Central fatigue begins (voluntary drive declines). This is the dominant phase for high-intensity exercise.
2. 30 s - 5 min: intracellular pH drops (from 7.0 to ~6.4 in maximally exercising muscle), inhibiting cross-bridge cycling and SR calcium release. Glycogenolysis rate peaks then declines as glycogen is depleted.
3. 5-60 min: glycogen depletion becomes rate-limiting for ATP supply. Blood glucose uptake increases but cannot fully compensate. Fatigue is now primarily metabolic rather than metabolite-accumulation.
4. >60 min: muscle glycogen approaches exhaustion. Thermoregulatory strain and cardiovascular drift compound the peripheral fatigue. Central fatigue increases.

The relative contribution of each mechanism depends on the initial fibre-type composition of the active units: Type II-rich muscles fatigue faster and via metabolic accumulation; Type I-rich muscles fatigue slower and via glycogen depletion.

Exercises Intermediate+

Motor unit analysis and neuromuscular disease Master

Electromyography (EMG)

EMG records the electrical activity of contracting motor units. Two recording approaches are used:

Surface EMG uses electrodes on the skin overlying the muscle. It records a summated signal from many motor units and is useful for assessing overall muscle activation patterns, timing, and amplitude. The rectified and integrated surface EMG signal increases approximately linearly with force at low contraction intensities but underestimates force above 50% MVC because of phase cancellation (action potentials of opposite polarity sum destructively in the volume conductor).

Intramuscular (fine-wire or needle) EMG inserts electrodes directly into the muscle, recording from a small volume of tissue. This allows decomposition of the composite signal into individual motor unit action potential (MUAP) trains. Modern decomposition algorithms can identify 5-15 concurrently active motor units from a single needle recording, providing direct experimental access to individual recruitment thresholds, firing rates, and derecruitment patterns.

High-density surface EMG uses arrays of 64-128 surface electrodes to record spatial maps of muscle electrical activity. Blind-source separation techniques (e.g., fastICA, convolutive ICA) decompose these signals into single-motor-unit contributions without needle insertion. This approach has confirmed Henneman's size principle in human voluntary contractions across many laboratories and muscle groups.

Motor unit number estimation (MUNE)

MUNE estimates the total number of functioning motor units in a muscle. The standard technique applies incremental electrical stimulation to the motor nerve and counts the number of discrete force or EMG amplitude steps, each step corresponding to the recruitment of one additional motor unit. The ratio of the maximal CMAP (compound muscle action potential) amplitude to the mean single-unit amplitude gives the motor unit count.

Normal values: ~100-300 motor units in distal hand muscles, ~300-600 in proximal limb muscles, declining with age (approximately 1-2% per year after age 60, reflecting motoneuron loss). In amyotrophic lateral sclerosis (ALS), MUNE declines rapidly (50-80% within months), providing a sensitive biomarker for disease progression. In myasthenia gravis, motor unit count is preserved but the amplitude of individual MUAPs fluctuates because of variable neuromuscular transmission failure.

Neuromuscular diseases

Amyotrophic lateral sclerosis (ALS): progressive degeneration of alpha-motoneurons in the ventral horn and corticospinal tract. Motor unit loss follows a dying-forward (corticospinal excitotoxicity) and dying-back (distal axonopathy) pattern. As motoneurons die, surviving motoneurons sprout collaterals to reinnervate orphaned fibres, producing enlarged motor units detectable as increased MUAP duration and amplitude on needle EMG. The compensatory reinnervation temporarily preserves strength but eventually fails when the sprouting capacity is exhausted. MUNE declines while single-unit amplitude increases — a characteristic dissociation.

Myasthenia gravis: autoimmune attack on postsynaptic acetylcholine receptors (AChR) at the neuromuscular junction. Antibody-mediated receptor loss reduces the safety factor for neuromuscular transmission. At low firing rates, each endplate potential reaches threshold and the muscle contracts normally. At higher rates or during sustained activity, the progressive depletion of presynaptic acetylcholine vesicles (normally compensated by the safety factor) causes transmission failure in a fraction of junctions. Clinically, this manifests as fatigable weakness: strength is normal at rest but declines with repeated use. Repetitive nerve stimulation (RNS) at 3 Hz shows a characteristic decremental response (>10% amplitude decline by the fourth stimulus). Single-fibre EMG shows increased jitter (variability in interpotential interval between two fibres of the same motor unit), the most sensitive electrodiagnostic test.

Muscular dystrophies (Duchenne/Becker): mutations in the dystrophin gene compromise the dystrophin-glycoprotein complex that links the cytoskeleton to the extracellular matrix. Membrane fragility leads to calcium influx, proteolytic damage, fibre necrosis, and chronic regeneration. Motor unit architecture is disrupted not by neural pathology but by the loss and regeneration of constituent fibres. EMG shows myopathic changes: short-duration, polyphasic MUAPs (because fibres are split and heterogeneous) and early recruitment (many units activate at low force levels because individual fibres produce less force).

Muscle plasticity

Fibre-type shifting with training: Chronic endurance training converts Type IIx fibres toward the IIa phenotype within 4-6 weeks, driven by PGC-1alpha-mediated transcriptional reprogramming (see 18.04.01). The IIx-to-IIa shift is robust and reproducible; the IIa-to-I shift is slower and more limited (5-10% of fibre count over months). Chronic low-frequency electrical stimulation (10 Hz continuous, mimicking slow-motoneuron firing patterns) drives near-complete fast-to-slow conversion in animal models, demonstrating that the molecular plasticity is present but normally accessed only by extreme stimuli.

Sprint or resistance training drives the opposite shift (IIa toward IIx) and increases SR density, glycolytic enzyme activity, and myofibrillar protein content via the Akt/mTOR pathway. The direction and magnitude of fibre-type shifting are determined by the pattern of neural activation (rate and duration) and the mechanical load, integrated through the AMPK-versus-mTOR signalling balance.

Immobilization atrophy: Limb immobilization (casting, bed rest, microgravity) produces rapid atrophy — 0.5% fibre cross-sectional area per day in the first 2-3 weeks. The molecular mechanism includes reduced mechanical loading (disabling the Akt/mTOR hypertrophy signal), elevated MuRF1 and atrogin-1 expression (ubiquitin-proteasome-mediated protein degradation), and suppression of satellite cell activity. Slow-twitch fibres atrophy faster than fast-twitch fibres during immobilization, possibly because the baseline mechanical loading that maintains their phenotype is removed. Reinnervation and fibre-type grouping occur with aging: as motoneurons die and surviving neurons sprout to cover orphaned fibres, the normal checkerboard pattern of fibre types on muscle biopsy is replaced by large clusters of uniform fibre type (fibre-type grouping), a histological marker of chronic denervation-reinnervation.

Satellite cells and hypertrophy

Satellite cells (Pax7-positive muscle stem cells, identified by Mauro 1961) lie between the basal lamina and sarcolemma of mature muscle fibres. During hypertrophy, satellite cells activate, proliferate, and fuse with existing fibres, donating their nuclei to maintain the myonuclear domain (the cytoplasmic volume per nucleus) at a roughly constant ratio. Modest hypertrophy can proceed via increased protein synthesis per existing nucleus, but gains exceeding ~25% fibre cross-sectional area require satellite-cell-mediated nuclear addition.

This mechanism underpins the "muscle memory" phenomenon: previously trained muscle retains elevated myonuclear number even after detraining-induced atrophy, enabling faster re-hypertrophy upon retraining. Myonuclear number does not decline during atrophy (nuclei are not expelled), creating a ratchet-like hysteresis in the hypertrophy-atrophy cycle.

Eccentric vs concentric muscle damage

Eccentric contractions (muscle lengthening under tension) produce more force per cross-bridge than concentric contractions but cause disproportionate structural damage. The mechanism involves non-uniform sarcomere lengthening on the descending limb of the length-tension curve: weaker sarcomeres lengthen rapidly ("sarcomere popping"), stretching passive elastic elements and disrupting the sarcomere architecture. This triggers a cascade of sarcolemmal disruption, calcium influx, proteolytic activation (calpains), inflammation, and delayed-onset muscle soreness (DOMS) peaking 24-72 hours post-exercise.

DOMS is not caused by lactate accumulation (lactate clears within 1-2 hours post-exercise). It is caused by mechanical disruption of sarcomeres and the subsequent inflammatory response (neutrophil infiltration, macrophage phagocytosis, prostaglandin-mediated nociceptor sensitisation). The repeated-bout effect — a single bout of eccentric exercise confers protection against DOMS from subsequent bouts for weeks — involves neural, mechanical, and cellular adaptations including increased sarcomere uniformity, strengthened extracellular matrix, and altered motor unit recruitment strategies.

Fatigue-induced changes in EMG power spectrum

During sustained submaximal contractions, the EMG power spectrum shifts toward lower frequencies (the EMG spectral shift). This shift reflects two concurrent processes: (1) motor unit firing rates decline as central fatigue reduces descending drive, and (2) MUAP duration increases as fibre conduction velocity slows (a consequence of intracellular acidosis and extracellular potassium accumulation during repeated firing). The median frequency of the EMG power spectrum declines monotonically during sustained contraction and has been proposed as a noninvasive fatigue index, though inter-individual variability and electrode-placement sensitivity limit its clinical utility.

Connections Master

Motor unit recruitment bridges cellular muscle physiology (18.04.01, 18.04.02) to neural control of movement (18.05.01). The size principle is a consequence of motoneuron biophysics (input resistance scaling with soma size) married to fibre-type specialisation (slow oxidative vs fast glycolytic). The result is a control system that is both energy-efficient (slow units handle routine loads aerobically) and powerful (fast units are available on demand).

The graded-force problem — how a tissue made of all-or-nothing elements (each fibre either contracts or does not) produces smoothly varying force — is solved by the combination of recruitment (adding discrete quanta of force) and rate coding (modulating force per quantum continuously). This solution is analogous to pulse-width modulation in engineering: the motor unit is the pulse, the firing rate is the width.

Fatigue connects to metabolic biochemistry (17.04.01, 17.04.02) through the ATP supply-demand balance that differs across fibre types. The fibre-type trade-off between power and endurance is a Pareto optimisation: a fibre can be adapted for speed or for efficiency, but not both simultaneously. Athletic training pushes fibres toward one end of this spectrum; aging and disease push them toward inefficiency and loss.

The clinical dimension — ALS, myasthenia gravis, muscular dystrophy, sarcopenia — demonstrates what happens when individual components of the motor unit system fail. Each disease has a characteristic signature detectable by EMG: motoneuron loss (declining MUNE), transmission failure (jitter and decrement), or myofibre destruction (myopathic MUAP changes).

Historical notes Master

The modern understanding of motor unit recruitment began with Sherrington's concept of the "final common pathway" (1906): the motoneuron as the last neural element through which all motor commands must pass. The term "motor unit" was coined by Liddell and Sherrington in 1925 to describe the motoneuron and its innervated fibres as a functional unit.

Henneman's 1957 Science paper established the size principle through a simple observation: in decerebrate cat preparations, progressively increasing afferent input produced stereotyped recruitment and derecruitment patterns across identified motoneurons. The 1965 papers with Somjen and Carpenter quantified the relationship between soma size and recruitment threshold using intracellular recordings. The biophysical explanation — smaller cells have higher input resistance and depolarise more for a given synaptic current — followed from cable theory.

The fibre-type classification emerged from the histochemical era of the 1960s-70s. Brooke and Kaiser (1970) developed the ATPase staining method that distinguished fibre types by their myosin ATPase pH sensitivity, revealing the three-type classification (I, IIa, IIb/IIx) still used today. The molecular basis — distinct MyHC isoform genes — was established by Schiaffino, Reggiani, and colleagues in the 1990s-2000s.

The force-frequency relationship was quantified by Cooper and Eccles (1930) and refined by Rack and Westbury (1969) in cat soleus. The concept of unfused tetanus as the operational range of voluntary motor unit firing was developed by Bigland and Lippold (1954), who showed that surface EMG amplitude increases linearly with force during graded voluntary contractions.

EMG decomposition — the extraction of individual motor unit firing patterns from composite recordings — was pioneered by LeFever and De Luca (1982) using precision wire electrodes and template-matching algorithms. Modern high-density surface EMG with blind-source separation (Holobar and Zazula 2007, Farina et al. 2008) has made noninvasive single-motor-unit identification practical.

The distinction between central and peripheral fatigue was established by Merton (1954), who showed that a maximal voluntary contraction of the adductor pollicis could not be augmented by supramaximal nerve stimulation, concluding that all fatigue was peripheral. Bigland-Ritchie and colleagues (1978-1986) challenged this by demonstrating central fatigue in larger muscles and longer contractions, establishing the modern two-component framework.

Bibliography Master

1. Henneman, E. — Relation between size of neurons and their susceptibility to discharge. Science 126, 1345-1347 (1957).

2. Henneman, E., Somjen, G. & Carpenter, D. O. — Functional significance of cell size in spinal motoneurons. J. Neurophysiol. 28, 560-580 (1965).

3. Burke, R. E. — Motor units: anatomy, physiology, and functional organization. In Handbook of Physiology, Section 1: The Nervous System, Vol. II, Motor Control, Part 1, pp. 345-422. American Physiological Society (1981).

4. Enoka, R. M. — Neuromechanics of Human Movement, 5th ed. Human Kinetics (2015).

5. Schiaffino, S. & Reggiani, C. — Fiber types in mammalian skeletal muscles. Physiol. Rev. 91, 1447-1531 (2011).

6. Bigland-Ritchie, B., Jones, D. A., Hosking, G. P. & Edwards, R. H. T. — Central and peripheral fatigue in sustained maximum voluntary contractions of human quadriceps. Clin. Sci. Mol. Med. 54, 609-614 (1978).

7. Farina, D., Holobar, A., Merletti, R. & Enoka, R. M. — Decoding the neural drive to muscles from the surface electromyogram. Clin. Neurophysiol. 121, 1616-1623 (2010).

8. Merton, P. A. — Voluntary strength and fatigue. J. Physiol. 123, 553-564 (1954).

9. Liddell, E. G. T. & Sherrington, C. S. — Recruitment and some other features of reflex inhibition. Proc. R. Soc. Lond. B 97, 488-518 (1925).

10. Mauro, A. — Satellite cell of skeletal muscle fibers. J. Biophys. Biochem. Cytol. 9, 493-495 (1961).