Secondary Logo

Journal Logo


A Principle of Neuromechanical Matching for Motor Unit Recruitment in Human Movement

Hudson, Anna L.; Gandevia, Simon C.; Butler, Jane E.

Author Information
Exercise and Sport Sciences Reviews: July 2019 - Volume 47 - Issue 3 - p 157-168
doi: 10.1249/JES.0000000000000191
  • Free

Key Points

  • In the respiratory muscles, motor units are recruited according to their inspiratory mechanical advantage, which provides an efficient strategy to ventilate the lungs.
  • This recruitment is adaptable, not related to the intrinsic properties of motor units, and inconsistent with motor unit recruitment according to the classic size principle.
  • We propose a principle of motor unit recruitment by neuromechanical matching.
  • The mechanisms are likely to involve premotoneuronal neurons in the spinal cord that sculpt descending drive to the motoneurons depending on the motor task.
  • There is evidence of motor unit recruitment by neuromechanical matching in nonrespiratory muscles, and, thus, this principle may extend to many movements.


Motor tasks involve the recruitment and discharge of motor units, both within and across muscles, which work in synergy to develop force. A motor unit is defined as a motoneuron, its axon and the muscle fibers it innervates. The firing pattern of a motoneuron is determined by the conversion of synaptic inputs into action potentials, which depends on the intrinsic properties of motoneurons, neuromodulation of the biophysical properties of motoneurons, and the summation of multiple excitatory and inhibitory synaptic inputs to the motoneuron (1). Given the one-to-one relationship between motoneuron discharge and a motor unit action potential in the muscle, recordings of motor unit activity are a surrogate for direct recordings from the motoneurons in the ventral column of the spinal cord, with the major advantage of being able to make motor unit recordings in vivo in human participants. The discharge of action potentials in the muscle, or single motor unit (SMU) activity, can be recorded with an intramuscular electrode, including from the respiratory muscles during the motor tasks of breathing and nonrespiratory contractions with the appropriate safety precautions (see hereinafter). By sampling from large populations of motor units during motor tasks, the output of motoneuron pools across synergistic muscles can be compared. Here, the term motor units will be used in reference to a muscle and motoneurons in reference to a motor nucleus or motoneuron pool.

The presiding principle of motor unit recruitment is Henneman’s size principle, informed mostly by recruitment of limb motor units (2,3). In this review, we briefly summarize the evidence for the size principle. Then, based on our data, from recordings in the axial or respiratory muscles, we hypothesize that the mechanics of muscles also may govern motor unit recruitment, in other words, motor unit recruitment by a principle of neuromechanical matching. The potential mechanisms by which neural activity is matched to muscle mechanics are discussed. Finally, we explore how motor unit recruitment by neuromechanical matching may be integral to many movements, including nonrespiratory movements. This principle of motor unit recruitment by neuromechanical matching may explain some previously observed limitations and apparent deviations from the size principle, especially changes in motor unit behavior across different motor tasks.

Here, we define a motor task as the resultant movement or intended movement (in an isometric contraction) due to activation of motor units in one or more muscles. On the gross anatomical level, a muscle is defined as the collection of muscle fibers contained between the anatomical origin and insertions (tendons or aponeuroses). However, all muscle fibers contained in a muscle typically are not activated simultaneously, nor is activation of muscle fibers necessarily evenly distributed throughout the muscle. There is regional activity of subpopulations of muscle fibers due to activation of subpopulations of motor units (see “Neuromechanical Matching in the Human Respiratory Muscles” and “Neuromechanical Matching in Nonrespiratory Muscles” sections). Therefore, the use of the term “muscle” here does not always imply activation of all fibers and in these circumstances we have clarified this detail. The mechanics of muscle action of an individual muscle can differ between tasks. By extension, differential activation of groups of motor units within and across synergistic muscles can result in drastically different movements. The amount of force that a muscle and its motor units can produce in a motor task depends on key biomechanical factors such as its length-tension relation, velocity of shortening, the muscle’s physiological cross-sectional area and specific tension, the pennation angle of the muscle fibers, and the moment arm of the muscle-tendon unit (see (4)). For the respiratory muscles discussed here, the measure of their “mechanical advantage” (see “Neuromechanical Matching in the Human Respiratory Muscles” section) incorporates the length-tension, pennation angle, and moment arm elements of these biomechanical factors and respiratory effect also captures the muscle’s physiological cross-sectional area and specific tension. Respiratory effect evaluates the contribution to pressure (i.e., force) a respiratory muscle can produce in breathing.


Henneman and colleagues (2) discovered the relation between the threshold at which motoneurons discharge and their size and the orderly recruitment of motoneurons according to their size (or threshold that reflects the input resistance of motoneurons) and termed it the size principle of motor unit recruitment (for review see (3)). According to this principle, if all motoneurons within a pool receive the same excitatory and inhibitory synaptic inputs, motoneurons are recruited in a systematic fashion that is dependent on their intrinsic properties. It has since been shown that many properties of motoneurons or motor units correlate with recruitment threshold such as soma size, membrane resistance, intrinsic excitability, axonal size, axonal conduction velocity, spike amplitude, twitch contraction time, force, and fatigability (for review see (3,5)). Therefore, during a contraction, the motoneurons of small motor units, which produce low forces and are fatigue resistant, require less depolarizing current and are recruited before motoneurons of large motor units that can produce more force, but fatigue more quickly due to glycolytic metabolism. This ensures both control of fine movements and rapid production of force for stronger movements and minimizes fatigue in everyday tasks requiring low-level contractions.

Stable motor unit recruitment according to size has been demonstrated in many circumstances (for review see (3,5,6)). As most physiological systems have an inherent level of noise, stable is taken to mean the same motor unit recruitment order for at least 80% of motor units in a contraction or the same motor unit recruitment order in at least 80% of contractions. For example, in cats, for motoneuron pairs recorded from L7 and S1 roots in response to stimulation of supraspinal centers, the smallest cell is recruited before the larger cell ~ 90% of the time (7). In voluntary isometric contractions of first dorsal interosseous (FDI) muscle in humans, the linear correlation coefficient between twitch tension (i.e., motor unit size) and threshold force at which motor units are recruited is greater than 0.8, but not all data points fall on the line of identity (8). It was subsequently noted that although orderly size-related recruitment occurs in isometric contractions for intrinsic hand muscles FDI and abductor pollicis brevis, only about a third of motor unit pairs were always recruited in order (as predicted by their relative twitch amplitudes) in dynamic contractions (rhythmic scissoring movement) (9,10).

Given such examples of altered motor unit recruitment in different types of contractions or tasks, within the context of the anatomical and functional complexities of skeletal muscles, Loeb and colleagues (11,12) conceived the idea of the task group, that is a set of motor units recruited in an ordered manner for a particular task or motor behavior. It was thought that the exact motor units recruited may vary between different tasks, but there is still orderly recruitment of the motor units according to their size in a particular task group. For example, task groups of motor units are observed in the biceps femoris in the cat for hip extension and knee flexion (12), in the biceps brachii in humans for flexion and supination (13,14), in extensor digitorum to extend different digits (15), and in the abdominal muscles for respiratory and nonrespiratory tasks (16).

Just as one muscle produces force in more than one direction, many muscles can contribute to force production in a given direction. Cope and Sokoloff (6) proposed that the idea of motor unit task groups should be extended such that they contain motor units from more than one muscle. Therefore, when more than one muscle can contribute to force generation in a particular direction, motor units recruitment is coordinated across muscles, and such that there is an ensemble-based scheme of motor unit recruitment rather than a “muscle-based” scheme (6). This seems to be the case in the decerebrate cat as, for motor unit pairs in medial and lateral gastrocnemii, motoneurons with slower conduction velocities are recruited before those with higher conduction velocities (17). Thus, as for task groups within a muscle, recruitment by Henneman’s size principle across muscles would ensure coordinated recruitment of motor units that would result in smooth, graded increases in force.

Relevant to this review, motor units in the diaphragm are recruited in a relatively stable order within repeated voluntary breaths of different sizes and rates (18). Furthermore, based on the size of motor unit action potentials at a given recording site using a monopolar needle electrode (exposed tip of 0.15 mm2), small diaphragmatic motor units are recruited before larger units in a voluntary breath (see Fig. 1 in (18)). However, during voluntary breaths at fast flows (i.e., double the flow in quiet breathing) or when the input to the motoneurons changes, such as from the motor cortex in slow voluntary breaths to the medullary respiratory centers in automatic breaths, motor unit recruitment order is altered for two in every four motor units in the population (18), suggesting some nonuniformity in the distribution of descending drives to the phrenic motoneurons. As for the diaphragm, the recruitment order in the human parasternal intercostal muscles also is relatively stable in quiet breathing, but altered for one to two in every four motor units during voluntary hyperventilation (19), although only 31 units were sampled in the parasternal intercostals compared with 60 motor units in the diaphragm (18).

Figure 1
Figure 1:
Differential activity in the human inspiratory muscles during quiet breathing. A. Schematic of the human inspiratory muscles. The majority of the muscles shown are obligatory, active in every breath, but the sternomastoid is active only in large breaths. Pale colors are used for more superficial muscles and darker colors for deeper muscles. Panels (B), (C), and (D) depict the graded patterns of inspiratory muscle activity that have been demonstrated in humans. Relative muscle activation during breathing is indicated by the reference color scale, with greatest activation in warm colors and less activity in cooler colors. For example, the muscles shown in purple (e.g., third ventral external intercostal and sternomastoid) are only active in larger breaths. There are rostrocaudal gradients in the parasternal intercostal muscles from the first-to-fifth interspaces (panel B) and the external intercostal muscles from the third, fifth, and seventh interspaces (panel C). There also is a dostroventral gradient of activity in the external intercostal muscles in the third interspace (panel D). For the inspiratory muscles in the neck (panels B and D), the scalene muscles are active in every breath, but the sternomastoid is only active in larger breaths.

One assumption of the size principle is that synaptic input or drive is distributed equally to all the motoneurons within a particular pool. However, modeling experiments have highlighted the role of synaptic input in differential recruitment of motoneurons. Kernell and Hultborn (20) showed that changes in recruitment gain could be achieved only if the distribution of synaptic input was differential across the motoneuron pool. Recruitment gain is defined as the number of motoneurons recruited for a given synaptic input (20). An increase in recruitment gain reduces the range of activation thresholds for motoneurons as more motoneurons are recruited by the same synaptic input. When the model assumed that synaptic input was equally distributed, it was possible to change motoneuron recruitment thresholds but not recruitment gain (20).

Uneven distribution of synaptic input or drive across motoneuron pools may account for some of the variability in motor unit recruitment order that has been described when the muscle contracts in a different manner (e.g., dynamically) or the contraction requires selective activation of units, in other words, a task group. The uneven distribution of synaptic inputs may result from descending, afferent or propriospinal inputs to motoneurons, and here, we argue these inputs may be organized to ensure recruitment of motor units depending on their relative mechanics for the motor task.


Motor units from many muscles are recruited during inspiration to generate negative pleural pressure (see Fig. 1; for review see (21)). Compared with other skeletal muscles, the respiratory motoneurons are automatically and rhythmically depolarized by descending neural drive that originates in the respiratory centers in the ponto-medullary region of the brain (22). However, respiratory muscles also can be activated volitionally, in both respiratory (i.e., large breaths) and nonrespiratory (i.e., trunk movements) tasks, and consequently their motoneurons also receive neural drive from higher brain areas including the motor cortex (for review see (22)). These drives depolarize the same motoneurons, and current evidence suggests that the integration of drives occurs at the spinal cord, rather than at the brain stem (22). Irrespective of whether volitional drive bypasses the medulla or not, the output of the final common path, the motoneurons, can be assessed by SMU discharge.

For the motor task of breathing, SMU recordings have revealed that the output of the various human inspiratory synergist muscles is not uniform with differential motor unit activity across five pump muscles in terms of the timing and amplitude of inspiratory activity (for review see (22,23)). Within the intercostal muscles, the timing and amplitude of motor unit activity varies across interspaces, but also between different regions of a muscle in an interspace ((24,25); see Fig. 1). For example, motor units in the dorsal region of the external intercostal muscle in the third interspace contract earlier in inspiration and discharge at a faster rate compared with motor units in the ventral region of the same interspace innervated by the same motoneurons pool (25). Anatomically, the dorsal and ventral regions of the external intercostal muscle may be considered a single muscle, based on data from dogs that demonstrate motoneurons from different regions of an intercostal space are colocated in the ventral horn (26). Despite this, the delay in recruitment between the dorsal and ventral regions is ~1 s, far longer than any difference due to transmission of neural signals along the intercostal nerve. Inspiratory motor unit activity is similar in the medial and lateral portions of the second human parasternal intercostal muscle, suggesting no regional difference in activation along an interspace for this muscle (24).

These rostrocaudal and dorsoventral gradients of motor unit output in the human intercostal muscles are remarkable because they correlate with the mechanics of these muscles for the motor task of inspiration, in other words with inspiratory mechanical advantage (Fig. 2). Stemming from theoretical and experimental work in dogs and humans by De Troyer and colleagues, the inspiratory mechanical advantage of many human respiratory muscles has been computed (for review see (28)). In brief, for the respiratory muscles, mechanical advantage is computed as the relative change in muscle length during passive inflation of the lungs (i.e., %/l). It is a surrogate measure for the respiratory effect of a muscle, which is defined as the change in airway pressure generated by the muscle if it were active in isolation. Given isolated muscle contraction is not possible in human participants, the inspiratory mechanical advantage and respiratory effect of the human respiratory muscles is estimated using computed tomography and measurements from cadavers (28). With passive lung inflation, a muscle that shortens more has a greater inspiratory mechanical advantage, also meaning it has a greater respiratory effect; in other words, it can evoke a greater change in airway opening pressure.

Figure 2
Figure 2:
Neuromechanical matching in the human intercostal muscles. A. Time and frequency plots reveal a rostrocaudal gradient of activity in the parasternal intercostal muscles during quiet breathing. Each horizontal line represents data from one motor unit recorded in the first, third, and fifth interspaces in five healthy male participants. The onset and end of inspiration are shown by vertical dashed lines (from 0% to 100%) of inspiratory time (horizontal axis). Thin horizontal lines at the bottom of each plot indicate those units that discharged tonically. The thick horizontal lines show the time that the discharge frequency increased during inspiration. Discharge frequency is coded by color (shown in inset). The color of thick lines shows the peak firing frequency, and the smaller circles at the start and end show the frequencies at recruitment and derecruitment, respectively. The large black circles show the time at which the peak frequency occurred. The ascending unbroken white line in each panel represents the onset time of the multiunit electromyography (EMG) ordered by onset time (see (27)). The onset of activity is earlier and the peak discharge rate of the motor units is higher in the rostral first interspace compared with caudal third and fifth spaces. B. Using single motor unit (SMU) recordings to estimate neural drive to the parasternal intercostal muscles in different interspaces, we first demonstrated that the rostrocaudal gradient of activity across the five interspaces (black circles) is tightly coupled to the inspiratory mechanical advantage of the same regions of muscles (24). We have replicated the results that there is neuromechanical matching in the human parasternal intercostal muscles during quiet breathing (gray circles) and demonstrated that the matching is maintained in targeted voluntary breaths (white circles) (data from (27)). The lines are colored to indicate the rostrocaudal gradient, with warm-to-cold gradient representing a decrease in both neural activity and inspiratory mechanical advantage. All correlations have r-squared values of 0.99 or greater.

By comparison of measures of inspiratory mechanical advantage for different regions of the human intercostal muscles with an estimate of neural drive (i.e., number of motor unit spikes during inspiration from our SMU data) to the pools of motoneurons that innervate these regions of muscle, we discovered a strong linear relationship between neural drive and mechanical advantage, in other words neuromechanical matching (see also (29)). This occurs for both the external intercostal and parasternal intercostal muscles during quiet breathing (24,25) and the parasternal intercostal muscles during voluntary breaths (27). Motor unit recruitment by neuromechanical matching reduces the metabolic cost of muscle activation (28), thus providing an efficient strategy for ventilation. Figure 2 highlights replication of our finding that the neural drive to the human parasternal intercostal muscles during quiet breathing is tightly coupled to inspiratory mechanical advantage ((24,27) see also (30)). In addition to the intercostal muscles, we also have demonstrated that the human scalene and sternocleidomastoid muscles in the neck, obligatory and accessory muscles of inspiration, respectively (see Fig. 1), also are active according to their relative mechanical advantages for inspiration in large inspiratory efforts (31).

Respiratory muscles are multifunctional and are active in other motor tasks that have different mechanical requirements. For example, the parasternal intercostals elevate the ribs vertically in quiet breathing, but translate the ribs laterally in trunk rotation. By direct comparison of the motor unit activity, we have recently shown that the differential activation of the parasternal intercostals across interspaces for inspiration is effectively reversed for the different motor task of trunk rotation ((30); Fig. 3). These changes in motor unit behavior do not reflect noise in motoneuron recruitment by a size-related principle of recruitment. Based on data for muscle shortening in dogs and rib and muscle fiber angulation in humans, there is a strong case that these changes in motor unit activity match the mechanical advantage of the muscles in the different tasks (30). Thus, neuromechanical matching in the human respiratory muscles adapts across tasks and, as proposed for the graded output in breathing, may be due to neural organization in the spinal cord ((30); see “Mechanisms of Neuromechanical Matching in the Respiratory Muscles” section). Altered neural drive to the inspiratory muscles in rotation compared with breathing does not occur for all obligatory inspiratory muscles. The parasternal intercostal muscles show strong activity in rotations to the ipsilateral side (30,32), whereas the diaphragm shows minimal activity in similar rotations (see (23)). This suggests that the observed behavior of motor unit output to the parasternal intercostal muscles in our recent study (30) is related to task-specific muscle function, rather than a widespread phenomenon among the inspiratory muscles. Although the observed changes in motor unit output between tasks occurred across muscles that are anatomically distinct, this is evidence for a change in the timing of excitatory inputs to these motoneurons and not just recruitment according to motoneuron size.

Figure 3
Figure 3:
Adaptability of parasternal intercostal motor unit recruitment in different motor tasks. Recruitment of parasternal intercostal motor units in the first (black bars), second (gray), and fourth (white) spaces during quiet breathing (left panel) and rotation (right panel). Median values for each interspace are shown on each histogram by arrows above the histograms. For quiet breathing (panel A), motor unit recruitment is indicated relative to inspiratory time during recordings, and for rotation (panel B), motor unit recruitment is indicated relative to torque of maximal voluntary rotation contractions (% MVC). In quiet breathing, motor units from the first and second interspaces are recruited early in the breath, before those from the fourth space. The graded neural activity of human parasternal intercostal muscles during quiet breathing is matched to the known mechanics of these muscles for inspiration. This pattern is effectively reversed in rotation as motor units in the first interspace are recruited late in rotation, compared with the second and fourth spaces. PSIC, parasternal intercostal.

Is neuromechanical matching in the respiratory muscles during breathing adaptable? Given their position in the quasi-rigid three-dimensional ribcage, most respiratory muscles are relatively impervious to acute changes in their mechanics as it is difficult to manipulate the bones onto which they originate or insert. Thus, we used an acute change in body posture, to an upside-down posture, to elicit changes in respiratory muscle mechanics (33). A change in the gravitational vector in the upside-down posture changes the mechanics of the respiratory system (i.e., chest wall, lung, and respiratory muscles) compared with that in standing. The inspiratory activity in two inspiratory muscles, scalenes and diaphragm, was affected differentially during quiet breathing in an upside-down posture. There was diminished activity in the scalenes compared with standing, but activity in the diaphragm was unaltered (33). Although the mechanics of the muscles were not quantified, we interpreted the decrease in scalene activity as an adaptation to the change in the mechanical advantage of the muscle given the reduced requirement of the scalenes to elevate the ribcage in an upside-down posture (33). Thus, motor unit recruitment in the respiratory muscles is adaptable across tasks when the mechanics of the muscles change (i.e., breathing vs rotation), or within a task when the mechanics of the muscles change (i.e., breathing in the upright vs upside-down postures).


The mechanism responsible for the differential distribution of neural drive to the inspiratory muscles has been investigated mainly in the canine intercostal muscles where similar gradients of inspiratory mechanical advantage and matched neural activity are observed (for review see (28)). However, complementary observations from other animal preparations are discussed where relevant or where data in humans or dogs are lacking.

As for other skeletal muscles, if we assume that respiratory motor units are recruited in a systematic fashion due to the “size principle” only (see “Motor Unit Recruitment and the Size Principle” section), then the differential activity observed in the intercostal muscles will be reflected by differences in the intrinsic properties of the motoneurons that innervate these muscles. Fiber-type composition is a good indicator of motoneuron properties as slow-twitch oxidative (SO) fibers are generally innervated by small, low-threshold motoneurons, and fast-twitch oxidative-glycolytic (FOG) and fast-twitch glycolytic fibers are supplied by large higher-threshold motoneurons (for review see (3)). However, the proportions of muscle fiber type do not differ between human external intercostal muscles for SO fibers (60%–65% SO fibers) (see (25)), or either between or within parasternal intercostals in dogs (~60% SO and 40% FOG fibers) (for review see (28)). In addition, the somal surface area and volume of motoneurons innervating the canine parasternal intercostal muscles does not differ between different interspaces, or between medial and lateral portions of the parasternal intercostal muscles (26). The motoneurons that innervate medial and lateral motor units are colocated in the ventral horn in dogs (26), as are motoneurons that innervate the different layers of intercostal muscle in any given interspace in cats (34), suggesting that there is not task-dependent topographical organization in intercostal motoneuron pools. In the cat, the proportion of fiber types does vary along a rostrocaudal gradient in the external intercostal muscles, with a greater proportion of SO fibers in the rostral interspaces (in the regions where the muscles have greater inspiratory activity), but again, there is no difference in the distribution of fiber type in different parasternal intercostals (35). Therefore, the distribution of activity in the intercostal muscles is not likely to be solely determined by an intrinsic, size-related property of the motoneurons. Although the quiet breathing and trunk rotation tasks where neuromechanical matching has been observed (see “Neuromechanical Matching in the Human Respiratory Muscles” section) are low-force tasks, reversed recruitment across spinal levels of the same motor units (30) provides strong evidence against motor unit recruitment exclusively by motoneuron size.

The role of afferent input has been assessed by section of either the phrenic nerves or thoracic dorsal roots in spontaneously breathing dogs. Removal of feedback does not alter the rostrocaudal pattern of activity in the parasternal intercostal muscles (for review see (23)). With support from our finding in humans that the typical activation of the scalenes and sternocleidomastoid muscles according to their relative inspiratory mechanical advantages was not altered with changes in lung volume (i.e., when feedback from the lungs and inspiratory muscles would differ) (31), on-going peripheral feedback plays a minimal role for the usual differential recruitment of inspiratory muscles in breathing. Although not implicated in the pattern of activation of inspiratory muscles in breathing, afferent feedback may be critical to changes in inspiratory muscle activity when the task or mechanics of the muscles change. In nonrespiratory muscles, for example, cutaneous afferent stimulation can affect differentially slow- and fast-twitch motor units in the cat (36) or modify motor unit behavior in humans compared with size-related recruitment in voluntary contractions without cutaneous stimulation (37,38). Regarding descending drive to the respiratory motoneurons, the strength of bulbospinal connections to different intercostal motoneurons at different spinal levels does not differ (28,39). Thus, the differential motoneuron output along a rostrocaudal gradient in breathing does not seem to depend on differential input from respiratory centers. Of course, motoneuron gain may vary across the intercostal spaces, but as discussed hereinafter, evidence of neuromodulation for inspiratory intercostal motoneurons output is lacking.

Alternatively, a mechanism at the spinal cord that sculpts the descending drive to the motoneurons may generate the topographic distribution of drive to intercostal muscles. This would provide motor unit recruitment by neuromechanical matching during both automatic and voluntary breaths with descending drive from bulbospinal and corticospinal projections, respectively, but also allow adaptability of motor unit output in different motor tasks (see Fig. 3). DiMarco and Kowalski (40) have shown that the usual pattern of output from canine intercostal motoneurons (i.e., graded activity between and within an interspace during spontaneous breathing) can be reproduced using high-frequency spinal cord stimulation over the ventral spinal cord at thoracic level T2. As these dogs also had a complete spinal section at the cervical level C2 (albeit in an acute preparation, see (41)), these data suggest that mechanisms within the spinal cord can produce the usual pattern of inspiratory output from the intercostal motoneurons according to the relative mechanics of the muscles (28), at least in the dog. Consistent with this is the preservation of the rostrocaudal gradient of the parasternal intercostal motor unit output in quiet breaths and matched voluntary breaths in human participants (27).

What might this spinal mechanism be? We previously postulated a role for “spinal distribution networks” of propriospinal, segmental, and intersegmental neurons that distribute neural drive to the different inspiratory motoneuron pools in different motor tasks (22; see Fig. 4). In animals, there is increasing evidence for spinal interneurons that contribute to respiratory motoneuron output (for review see (42)). These data also show that respiratory interneurons project between spinal segments and to the contralateral ventral horn. Thus, it is likely that there is a type of propriospinal system for phrenic and intercostal motoneurons (22,30,40) as there is for forelimb motoneurons in tasks such as reaching that require coordinated activity across multiple muscles.

Figure 4
Figure 4:
Recruitment of motoneurons by neuromechanical matching. A. Diagrammatic representation of the proposed spinal distribution networks that distribute neural drive to the intercostal motoneurons depending on task and the mechanical advantage of the muscles for those tasks. Motoneuronal (MN) pools in the first, second, and fourth thoracic spinal levels and the corresponding muscle fibers in the first, second, and fourth parasternal intercostal muscles are depicted. The angulation of the muscle fibers varies along a rostrocaudal gradient (see (30)). Spinal distribution networks of propriospinal, segmental, and intersegmental neurons are depicted as a column of cells ipsilateral to the motoneurons, but some of these neurons may have their cell bodies in the contralateral side of the spinal cord. As for the motoneurons, the spinal distribution network neurons vary in their intrinsic properties (e.g., somal size) and also may be influenced by changes in their biophysical properties by neuromodulation (see text). Given the size-related intrinsic properties of motor units in the parasternal intercostal muscles is similar between interspaces (see text), if motor unit recruitment was governed by the size principle alone, the onset and amplitude of inspiratory activity during quiet breathing or trunk rotation would be similar in the first, second, and fourth parasternal intercostal muscles (depicted by the same red color of the muscles in the ribcage, panel B), but this is not the case. For panels B, C, and D, the reference color scale for relative muscle activation is shown on the bottom left of the figure. Shown in purple (panel A) is the spinal distribution network for inspiration that sculpts descending drive from the bulbospinal pathways during automatic breathing and from the corticospinal (CS) pathways during voluntary breaths. Spinal distribution networks for inspiration on both sides of the spinal cord receive similar input as bulbospinal projections are largely bilateral. When activated via the inspiratory spinal distribution networks, the relative input to the parasternal intercostal motoneurons is distributed along a rostrocaudal gradient with earlier and greater activity in the first and second interspaces compared with the fourth space (24,30). This graded activity (depicted by the color of the muscles in the ribcage and purple arrow, panel C) is matched to the relative mechanical advantage of these interspaces for inspiration, that is, neuromechanical matching. The graded output persists during voluntary breaths in humans (27) and during high-frequency stimulation of the thoracic spinal cord in dogs with an acute C2 spinal section (indicated by the dashed line) (40). Shown in blue (panel A) is the spinal distribution network for another motor task, voluntary trunk rotation, which sculpts descending drive via CS pathways. During rotation, the relative input to the same parasternal intercostal motoneurons by the rotation spinal distribution network is depicted to be different to that distributed by the inspiratory spinal distribution network as the output of the motor units is effectively reversed compared with that in inspiration (30). During trunk rotation, there is delayed recruitment of motor units in the first interspace compared with the second and fourth spaces (depicted by the color of the muscles and blue arrow, panel D). The output is matched to the predicted mechanical advantage of these muscles (see (30)). Given neuromechanical matching is adaptable (e.g., altered inspiratory muscle output with changes in posture (33)), the proposed spinal distribution networks are responsive to afferent inputs and other descending drive (e.g., from vestibular system), indicated by the pink arrows (panel A). BS, bulbospinal; CS, corticospinal; PSIC, parasternal intercostal.

Synaptic inputs to respiratory motoneurons from central respiratory centers or spinal interneurons also may be modulated by mechanisms that alter the gain of the motoneurons. Active dendritic mechanisms involving persistent inward currents or ligand-gated channel-mediated synaptic inputs may amplify synaptic signals in some motoneurons (for review see (1)). For phrenic motoneurons, the amplification of synaptic excitation from central respiratory drive potentials occurs via ligand-mediated mechanisms (NMDA), but expiratory motoneurons in the thoracic region (that presumably innervate the expiratory intercostal muscles) show persistent inward currents in decerebrate cats (43). It is not known if inspiratory motoneurons in the thoracic region (i.e., that innervate the inspiratory intercostal muscles where regional differences in electromyography (EMG) are observed) have persistent inward currents like their expiratory counterparts. Regional differences in the density of serotonergic inputs to canine parasternal intercostal motoneurons occur mediolaterally, but not rostrocaudally (26), and the topographic distribution of inspiratory activity in this muscle occurs along both gradients (for review see (28)). Replication in the varied density of serotonergic input to motoneurons that innervate the medial and lateral canine parasternal intercostal muscles only would confirm that other excitatory inputs or other mechanisms may generate the regional differences in EMG, at least along the rostrocaudal gradient.

If descending neuromodulation is important for intercostal motoneurons output, this would be another spinal mechanism, in addition to spinal distribution networks of interneurons, by which inspiratory output of different “task groups” of motor units is sculpted to match the inspiratory mechanical advantages of the muscles for breathing (Fig. 4). Furthermore, it may provide a mechanism that adapts neuromechanical matching in the respiratory muscles across motor tasks, as it has been postulated that neuromodulation allows motoneurons to have different states for different functions or tasks (e.g., postural, voluntary, and oscillatory behaviors) (1). However, without evidence for neuromodulation in inspiratory intercostal motoneurons, this potential mechanism for neuromechanical matching remains unproven.


The respiratory muscles are skeletal muscles like those of the limbs and trunk. Therefore, perhaps unsurprisingly, neuromechanical matching also may operate in other skeletal muscles. Limb muscles also provide a model in which to investigate further the adaptability of neuromechanical matching.

In the human triceps surae muscle group, recruitment of motor units in medial gastrocnemius (MG) is delayed when the knee is flexed, such that the plantar flexion torque and level of soleus EMG at which they are recruited are significantly greater compared with when the knee is extended (44). This was proposed to reflect a reduction in the force-producing capability of MG to plantar flexion torque with the knee flexed, although the mechanical properties of the muscles were not measured. However, based on the methodology to assess the mechanical advantage of the respiratory muscles (see “Neuromechanical Matching in the Human Respiratory Muscles” section), the relative contribution of a limb muscle to torque around a joint could be assessed by the fractional change in muscle length/joint angle for a passive movement around a joint. Kawakami and colleagues (45) measured the change in muscle fascicle length of the soleus and gastrocnemii muscles over a range of knee and ankle angles in the passive condition. For MG, the mechanical advantage for plantar flexion is 0.42 mm per degree when the knee is fully extended, but only 0.13 mm per degree with the knee flexed at 90 degrees. In contrast, with a change in knee angle from 0 to 90 degrees, the change in mechanical advantage of soleus is unchanged. These differences are consistent with the relative change in force contribution and activity of gastrocnemius and soleus during plantar flexion with a change in knee angle (44) and thus fit with motor unit recruitment by neuromechanical matching (Fig. 5).

Figure 5
Figure 5:
Example of neuromechanical matching in the triceps surae muscles for plantar flexion contractions. A. Participants performed plantar flexion torque contractions (ankle angle, 90 degrees) in two positions: knee extended (knee angle, 0 degrees) and knee flexed (knee angle, 90 degrees) designed to alter medial gastrocnemius (MG) muscle length but not soleus muscle length. Using data from Kawakami et al. (45) to estimate the mechanical advantage (mech adv) of these muscles during plantar flexion contractions (see text), the change in knee angle from 0 (extended) to 90 degrees (flexed) would decrease the mechanical advantage of the MG significantly (by 0.29 mm per degree), but have little effect on that of soleus (0.02 mm per degree). B. Motor units were recorded from MG (bottom panel) during ramped plantar flexion contractions in the knee extended and knee flexed positions. Intramuscular soleus electromyography (EMG) (top panel) and plantar flexion torque (middle panel) also are shown. Compared with the onset of torque (indicated by “A”), the level of soleus EMG and torque at which an MG motor unit was recruited (indicated by “B”) was delayed in the knee flexed position compared with knee extended position (indicated by the greater A-to-B delay shaded in gray in the right panel). C. Group data for soleus EMG and plantar flexion torque at which MG motor units were recruited in the knee extended (black circles) and knee flexed (white circles) positions. With the knee flexed, the soleus EMG and torque at which MG units were recruited was significantly greater, likely due to the reduced mechanical advantage of MG in this flexed position. rec, recruitment; SOL, soleus. [Adapted from (44). Copyright © 2000 Springer Nature. Used with permission. Adaptations are themselves works protected by copyright. In order to publish this adaptation, authorization must be obtained both from the owner of the copyright in the original work and from the owner of copyright in the translation or adaptation.]

For the upper limb, we have demonstrated coupling between the neural and mechanical behavior for synergist muscles (FDI and long finger flexors) for flexion of the index finger (4). Furthermore, this matching is adaptable. If the moment arm of FDI for index finger flexion is acutely increased (indicated by ultrasound) thereby increasing flexion twitch force (evoked by ulnar nerve stimulation), the proportion of neural drive to this muscle increases (4).

Although the changes in motor unit recruitment are not specifically linked to known changes in the contribution of a muscle at a joint (i.e., its mechanical advantage or mechanical effectiveness), many studies have suggested that some motor units or muscles are recruited in relation to their direction of pull or contribution to force for limb (e.g., (13,46,47)) and axial (e.g., (48–50)) muscles. Therefore, recruitment of motor units by neuromechanical matching may explain divergence from recruitment purely based on the size-principle in some nonrespiratory muscles, particularly in studies that observed altered motor unit recruitment across motor tasks. Our proposed principle of motor unit recruitment complements that of the size principle and provides a central mechanism to produce efficient voluntary movements.


From its inception, Henneman’s size principle provided a hypothesis by which the recruitment of motor units in most contractions could be explained, even for task groups of motor units within a muscle (11) or ensemble-based task groups across muscles (6). Although orderly recruitment of motor units across muscles by the size principle would result in smooth gradation of force, it has been noted previously that a fixed order of motor unit recruitment would limit the muscle to perform only tasks that depended on graded motor unit force (51). As such, Cope and Sokoloff (6) proposed a selection process, by which the appropriate motor units, in other words, those that generate force in the desired direction, have to be selected for use and are then recruited by their size-related properties. We are not the first to observe changes in the motor unit behavior with different tasks, but our work in human respiratory muscles provides direct evidence for a mechanism that matches neural output to the relative mechanics of muscles (or regions of muscles) for different movements.

The size principle relies on the assumption that synaptic input is distributed equally to all the motoneurons within a particular pool. As described previously, historical data and our data from respiratory muscles that demonstrate divergence from motoneuron recruitment according to their size-related properties (e.g., in different contraction types or motor tasks) highlight the importance of this assumption when testing the size principle because inputs to motoneurons may change in these circumstances. Our recent finding of varied recruitment behavior in the same parasternal intercostal motor units when the task changed (30) shows that the differential synaptic input is more important than the size-related properties of the motor units for recruitment according to the principle of neuromechanical matching. This mechanism is likely to occur at the spinal cord and involves changes in synaptic input that select the units based on their mechanical advantage for the task. How might this occur and how widespread is it?

The spinal mechanisms postulated for the respiratory muscles (i.e., spinal distribution network of interneurons) are applicable to any skeletal muscle where motor unit recruitment by neuromechanical matching may occur. These interneurons may include afferent connections between the muscle and spinal cord, which are localized to a particular segment or neuromuscular compartment of the muscle (i.e., akin to task groups for efferent control). This was termed reflex partitioning and provides a potential mechanism for fine control of activity within a task group (see (52)). However, it does not operate in the human tibialis anterior muscle at least for low-threshold motoneurons (52). McKeon and colleagues (52) suggested reflex partitioning may be related to “muscle architecture and function” and not applicable to “simple muscles with single physiological functions” (cf. (53) in vastus medialis). The presence of any reflex partitioning may correspond to the presence of motor unit compartments or small motor unit territories for regional activation of muscle fibers. Motor unit compartments are absent in the MG, a muscle with single distal and proximal tendons (54), but present in muscles with more complex anatomy and broad muscle origins such as axial and trunk muscles (e.g., (25,48)). In this respect, muscle architecture and function may dictate the balance between the neuromechanical matching and size principles of motor unit recruitment. Muscles with large motor unit territories and muscles that transmit their force through discrete origin and insertions (i.e., single, long tendons) are instances where motor unit recruitment by neuromechanical matching may not occur. As described by Heroux and colleagues (54) for MG:

“In the absence of a separate bony attachment that would confer a mechanical advantage to muscle fibres in the proximal (knee flexion) or distal (ankle plantar flexion) portion of the MG muscle, there is no benefit for [motor units] with small spatial territories to be grouped in close proximity to either joint given that [motor unit] activity will generate tension at the proximal and distal muscle insertions.”


A principle of neuromechanical matching for motor unit recruitment in human movement is significant to the fields of exercise science and sports medicine. For example, physiological studies that measure motor unit output in movements (e.g., locomotion) must now consider mechanisms beyond the size principle. This new principle will improve our understanding of the mechanisms behind altered movement control (e.g., with motor impairments or injury) and the improvements that can be realized with training. Furthermore, uptake of this concept by the exercise science and sports medicine fields will expedite answers to outstanding research and clinical questions, some of which are noted hereinafter.

The adaptability of neuromechanical matching described previously is for acute changes in the mechanics of muscles. How the coupling of neural and mechanical behavior is altered by chronic changes to muscle mechanics in motor impairments is not known. If damage to spinal distribution networks occurs, then neuromechanical matching may be impaired. Similarly, if neuromechanical matching is disrupted because of injury or diseases that alter biomechanics or skeletal muscle function, then optimization of neuromechanical matching may provide a target for physiotherapy or therapeutic biofeedback.

To test neuromechanical matching in all skeletal muscles, the methodology to measure muscle mechanics is required. Given the amount of force a muscle can produce in any motor task depends on several key factors; studies that use physiological cross-sectional area alone to infer the mechanics of a muscle should be interpreted with caution as the mechanics of a muscle may vary independently to physiological cross-sectional area. For example, the moment arm of a muscle can change with different postures of the structure into which the muscle inserts (see (4)). Although the techniques to measure mechanical advantage or mechanical effectiveness for the respiratory muscles are established, current methodology uses computed tomography scans (28). There is a need for a method to make comparable measures of respiratory mechanical advantage using safer and cheaper technologies such as ultrasound. Ideally, these methods can then be applied to assess mechanical advantage in the respiratory muscles during tasks other than breathing and in postures other than supine, and in various motor tasks for nonrespiratory muscles. The typical contraction of the respiratory muscles (i.e., during quiet breathing) is orchestrated — a cyclical contraction from the same muscle length (at end expiratory lung volume) at a comparable extent and contraction velocity. Therefore, factors such as the velocity of the contraction that has been shown to affect motor unit recruitment (e.g., (18,55)) and contraction history (e.g., (51)) that may affect the force a muscle can generate in a contraction will need to be considered and explored in the application of the principle of neuromechanical matching to nonrespiratory muscles.


A.L.H. was supported by a Lung Foundation Australia/Boehringer Ingelheim COPD Research Fellowship; S.C.G. and J.E.B. were supported by the NHMRC (Australia).


1. Johnson MD, Thompson CK, Tysseling VM, Powers RK, Heckman CJ. The potential for understanding the synaptic organization of human motor commands via the firing patterns of motoneurons. J. Neurophysiol. 2017; 118:520–31.
2. Henneman E. Relation between size of neurons and their susceptibility to discharge. Science. 1957; 126:1345–7.
3. Henneman E, Mendell LM. Functional organisation of motoneuron pool and its inputs. In: Brookhart JM, Mountcastle VB, editors. Handbook of Physiology, Section 1, The Nervous System. Volume 2, Motor Control, part 1. Volume 2. Bethesda (MD): American Physiological Society; 1981. p. 423–508.
4. Hudson AL, Taylor JL, Gandevia SC, Butler JE. Coupling between mechanical and neural behaviour in the human first dorsal interosseous muscle. J. Physiol. 2009; 587:917–25.
5. Binder MD, Heckman CJ, Powers RK. The physiological control of motoneuron activity. In: Rowell LB, Shepherd JT, editors. Handbook of Physiology, Section 12, Exercise: Regulation and Integration of Multiple Systems. Bethesda (MD): American Physiological Society; 1996. p. 3–53.
6. Cope TC, Sokoloff AJ. Orderly recruitment among motoneurons supplying different muscles. J. Physiol. Paris. 1999; 93:81–5.
7. Somjen G, Carpenter DO, Henneman E. Responses of motoneurons of different sizes to graded stimulation of supraspinal centers of the brain. J. Neurophysiol. 1965; 28:958–65.
8. Milner-Brown HS, Stein RB, Yemm R. The orderly recruitment of human motor units during voluntary isometric contractions. J. Physiol. 1973; 230:359–70.
9. Thomas CK, Ross BH, Calancie B. Human motor-unit recruitment during isometric contractions and repeated dynamic movements. J. Neurophysiol. 1987; 57:311–24.
10. Thomas CK, Ross BH, Stein RB. Motor-unit recruitment in human first dorsal interosseous muscle for static contractions in three different directions. J. Neurophysiol. 1986; 55:1017–29.
11. Loeb GE. Motoneurone task groups: coping with kinematic heterogeneity. J. Exp. Biol. 1985; 115:137–46.
12. Chanaud CM, Pratt CA, Loeb GE. Functionally complex muscles of the cat hindlimb. V. The roles of histochemical fiber-type regionalization and mechanical heterogeneity in differential muscle activation. Exp. Brain Res. 1991; 85:300–13.
13. ter Haar Romeny BM, Denier van der Gon JJ, Gielen CC. Changes in recruitment order of motor units in the human biceps muscle. Exp. Neurol. 1982; 78:360–8.
14. van Zuylen EJ, Gielen CC, Denier van der Gon JJ. Coordination and inhomogeneous activation of human arm muscles during isometric torques. J. Neurophysiol. 1988; 60:1523–48.
15. Riek S, Bawa P. Recruitment of motor units in human forearm extensors. J. Neurophysiol. 1992; 68:100–8.
16. Puckree T, Cerny F, Bishop B. Abdominal motor unit activity during respiratory and nonrespiratory tasks. J. Appl. Physiol. 1998; 84:1707–15.
17. Sokoloff AJ, Siegel SG, Cope TC. Recruitment order among motoneurons from different motor nuclei. J. Neurophysiol. 1999; 81:2485–92.
18. Butler JE, McKenzie DK, Gandevia SC. Discharge properties and recruitment of human diaphragmatic motor units during voluntary inspiratory tasks. J. Physiol. 1999; 518(Pt 3):907–20.
19. Watson TW, Whitelaw WA. Voluntary hyperventilation changes recruitment order of parasternal intercostal motor units. J. Appl. Physiol. 1987; 62:187–93.
20. Kernell D, Hultborn H. Synaptic effects on recruitment gain: a mechanism of importance for the input–output relations of motoneurone pools? Brain Res. 1990; 507:176–9.
21. De Troyer A, Boriek AM. Mechanics of the respiratory muscles. Compr. Physiol. 2011; 1:1273–300.
22. Hudson AL, Gandevia SC, Butler JE. Control of human inspiratory motoneurones during voluntary and involuntary contractions. Respir. Physiol. Neurobiol. 2011; 179:23–33.
23. Butler JE, Hudson AL, Gandevia SC. The neural control of human inspiratory muscles. Prog. Brain Res. 2014; 209:295–308.
24. Gandevia SC, Hudson AL, Gorman RB, Butler JE, De Troyer A. Spatial distribution of inspiratory drive to the parasternal intercostal muscles in humans. J. Physiol. 2006; 573:263–75.
25. De Troyer A, Gorman RB, Gandevia SC. Distribution of inspiratory drive to the external intercostal muscles in humans. J. Physiol. 2003; 546:943–54.
26. Zhan WZ, Mantilla CB, Zhan P, et al. Regional differences in serotonergic input to canine parasternal intercostal motoneurons. J. Appl. Physiol. 2000; 88:1581–9.
27. Hudson AL, Gandevia SC, Butler JE. Common rostrocaudal gradient of output from human intercostal motoneurones during voluntary and automatic breathing. Respir. Physiol. Neurobiol. 2011; 175:20–8.
28. De Troyer A, Kirkwood PA, Wilson TA. Respiratory action of the intercostal muscles. Physiol. Rev. 2005; 85:717–56.
29. Butler JE, Gandevia SC. The output from human inspiratory motoneurone pools. J. Physiol. 2008; 586:1257–64.
30. Hudson AL, Gandevia SC, Butler JE. Task-dependent output of human parasternal intercostal motor units across spinal levels. J. Physiol. 2017; 595:7081–92.
31. Hudson AL, Gandevia SC, Butler JE. The effect of lung volume on the co-ordinated recruitment of scalene and sternomastoid muscles in humans. J. Physiol. 2007; 584:261–70.
32. Hudson AL, Butler JE, Gandevia SC, De Troyer A. Interplay between the inspiratory and postural functions of the human parasternal intercostal muscles. J. Neurophysiol. 2010; 103:1622–9.
33. Hudson AL, Joulia F, Butler AA, Fitzpatrick RC, Gandevia SC, Butler JE. Activation of human inspiratory muscles in an upside-down posture. Respir. Physiol. Neurobiol. 2016; 226:152–9.
34. Larnicol N, Rose D, Marlot D, Duron B. Spinal localization of the intercostal motoneurones innervating the upper thoracic spaces. Neurosci. Lett. 1982; 31:13–8.
35. Greer JJ, Martin TP. Distribution of muscle fibre types and EMG activity in cat intercostal muscles. J. Appl. Physiol. 1990; 69:1208–11.
36. Burke RE, Jankowska E, ten Bruggencate G. A comparison of peripheral and rubrospinal synaptic input to slow and fast twitch motor units of triceps surae. J. Physiol. 1970; 207:709–32.
37. Garnett R, Stephens JA. The reflex responses of single motor units in human first dorsal interosseous muscle following cutaneous afferent stimulation. J. Physiol. 1980; 303:351–64.
38. Kanda K, Desmedt JE. Cutaneous facilitation of large motor units and motor control of human fingers in precision grip. In: Desmedt JE, editor. Motor Control Mechanisms in Health and Disease. New York: Raven Press; 1983. p. 253–61.
39. Davies JG, Kirkwood PA, Sears TA. The distribution of monosynaptic connexions from inspiratory bulbospinal neurones to inspiratory motoneurones in the cat. J. Physiol. 1985; 368:63–87.
40. DiMarco AF, Kowalski KE. Activation of inspiratory muscles via spinal cord stimulation. Respir. Physiol. Neurobiol. 2013; 189:438–49.
41. Gandevia SC, Kirkwood PA. Spinal breathing: stimulation and surprises. J. Physiol. 2011; 589:2661–2.
42. Lane MA. Spinal respiratory motoneurons and interneurons. Respir. Physiol. Neurobiol. 2011; 179:3–13.
43. Enriquez Denton M, Wienecke J, Zhang M, Hultborn H, Kirkwood PA. Voltage-dependent amplification of synaptic inputs in respiratory motoneurones. J. Physiol. 2012; 590:3067–90.
44. Kennedy PM, Cresswell AG. The effect of muscle length on motor-unit recruitment during isometric plantar flexion in humans. Exp. Brain Res. 2001; 137:58–64.
45. Kawakami Y, Ichinose Y, Fukunaga T. Architectural and functional features of human triceps surae muscles during contraction. J. Appl. Physiol. 1998; 85:398–404.
46. Desmedt JE, Godaux E. Spinal motoneuron recruitment in man: rank deordering with direction but not with speed of voluntary movement. Science. 1981; 214:933–6.
47. Heroux ME, Dakin CJ, Luu BL, Inglis JT, Blouin JS. Absence of lateral gastrocnemius activity and differential motor unit behavior in soleus and medial gastrocnemius during standing balance. J. Appl. Physiol. (1985). 2014; 116:140–8.
48. Falla D, Farina D. Motor units in cranial and caudal regions of the upper trapezius muscle have different discharge rates during brief static contractions. Acta Physiol. 2008; 192:551–8.
49. Bhutada MK, Phanachet I, Whittle T, Peck CC, Murray GM. Regional properties of the superior head of human lateral pterygoid muscle. Eur. J. Oral Sci. 2008; 116:518–24.
50. Park RJ, Tsao H, Cresswell AG, Hodges PW. Anticipatory postural activity of the deep trunk muscles differs between anatomical regions based on their mechanical advantage. Neuroscience. 2014; 261:161–72.
51. Herzog W, Schappacher G, DuVall M, Leonard TR, Herzog JA. Residual force enhancement following eccentric contractions: a new mechanism involving titin. Physiology (Bethesda). 2016; 31:300–12.
52. McKeon B, Gandevia S, Burke D. Absence of somatotopic projection of muscle afferents onto motoneurons of same muscle. J. Neurophysiol. 1984; 51:185–94.
53. Gallina A, Blouin JS, Ivanova TD, Garland SJ. Regionalization of the stretch reflex in the human vastus medialis. J. Physiol. 2017; 595:4991–5001.
54. Heroux ME, Brown HJ, Inglis JT, Siegmund GP, Blouin JS. Motor units in the human medial gastrocnemius muscle are not spatially localized or functionally grouped. J. Physiol. 2015; 593:3711–26.
55. Hodson-Tole EF, Wakeling JM. Motor unit recruitment patterns 1: responses to changes in locomotor velocity and incline. J. Exp. Biol. 2008; 211:1882–92.

neuromechanical matching; respiratory muscles; motor control; motor unit; motoneurons; spinal distribution network

Copyright © 2019 by the American College of Sports Medicine