Secondary Logo

Journal Logo


Muscle Reflexes in Motion: How, What, and Why?

Stein, Richard B.1; Thompson, Aiko K.2

Author Information
Exercise and Sport Sciences Reviews: October 2006 - Volume 34 - Issue 4 - p 145-153
doi: 10.1249/01.jes.0000240024.37996.e5
  • Free


Reflexes change dramatically, depending on the movement being performed (task dependence) and the phase of cyclic movements (phase dependence) (2,4). Yet, human reflexes have traditionally been measured at rest, whether testing tendon jerks in medical practice or with electrical or mechanical stimulation in experimental studies. This review will be concerned with muscle reflexes in the lower limbs as a part of functional movements, and we argue that results obtained at rest can be misleading with respect to the role of reflexes in movements. This finding is consistent with much of the sport science literature, where training and testing procedures are tailored closely to the activity of interest. Pierrot-Deseilligny and Burke (22) recently wrote an excellent book on reflexes in normal and some pathological states. We will refer to this comprehensive review frequently because it provides details on topics and references to articles that cannot be covered here because of strict limits on the number of pages and references.

Here, we will concentrate on major recent developments and methods to settle outstanding controversies. For example, in the next section, we show that postactivation depression of the Hoffmann reflex (H-reflex), which has been widely studied at rest, has little or no role in functional movements. We will also describe recent studies that have dramatically altered our view of excitatory reflexes in the control of basic functions such as standing and walking. Finally, we show that the most common technique for studying inhibition, by conditioning the H-reflex with a second inhibitory input, has important flaws and that the role of inhibition in modulating movements is best studied by more direct means.


There are obvious difficulties in applying tendon jerks by tapping a tendon in the same spot with the same force in a moving limb with changing stiffness of muscles and tendons. Therefore, most studies have used electrical stimulation of muscle nerves (Fig. 1A), which, near threshold, will mainly excite the largest fibers (group Ia fibers from muscle spindles and some α-motor neurons). Electrical stimuli produce two responses in the muscle (Fig. 1B): a direct motor or M-wave from stimulating motor axons and an H-reflex, which arises mainly from monosynaptic excitation of motor neurons by sensory nerve fibers from muscle spindles. The effects of the H-reflex stimulation and the tendon jerk are similar but not identical (19,22). The electrical stimulus will activate a single nerve impulse in each sensory fiber stimulated, whereas even a brief tendon tap may elicit more than one. Second, the effect of the mechanical stimulation will depend on the action of smaller γ-motor neurons (16). They change the length and velocity sensitivity of the specialized, intrafusal muscle fibers on which the sensory fibers lie and hence the sensory response to mechanical stimulation. Despite these potential differences, most studies have measured H-reflexes rather than tendon taps. The methodology is easier, although there are a number of potential pitfalls (see the next section). As will be discussed later (walking section), some studies have also applied longer-lasting stretches to a muscle during movement, which is important to confirm results with electrical stimulation.

Figure 1
Figure 1:
A weak stimulus to the tibial nerve behind the knee (A) produces impulses in motor neurons that conduct to muscles such as soleus and produce an M-wave in the EMG. Impulses in sensory (group Ia) neurons conduct to the spinal cord and produce additional excitation of motor neurons via a later, H-reflex. The peak-to-peak (pp EMG) values can be measured within the intervals where the M-wave and H-reflex occur (B) and plotted as a function of stimulus intensity (C). The M-wave increases monotonically to its maximum value (Mmax) with increasing stimulus intensity. The H-reflex increases to a peak (Hmax) but then declines. Further details are in the text.

Measuring H-Reflexes During Movement

As the stimulus intensity is increased at rest or during a movement, the M-wave increases monotonically to a maximum (Mmax), whereas the H-wave increases and then decreases (Fig. 1C). The reason is that nerve impulses are conducted in both directions when a nerve axon is stimulated along its length (not shown in Fig. 1). Thus, impulses will conduct back along motor axons, and the motor neurons will be refractory when sensory impulses activate them synaptically. At low-stimulus intensities, few motor axons are activated but neither are the many sensory fibers, so an intermediate H-reflex will be obtained. At higher intensities, most motor axons will be part of the M-wave, and these will also be refractory to generating an H-reflex. In between, the H-reflex will reach a peak (Hmax). Figure 1C shows data for the soleus H-reflex, but the same behavior is seen in other muscles that produce H-reflexes.

What level of stimulation should be used and how can this level be maintained with surface electrodes that may move with respect to the underlying nerve and muscles? We prefer a level of stimulation that produces a measurable M-wave and an H-reflex on the rising edge of the curve in Figure 1C. For example, Zehr and Stein (30) suggested a stimulus that produced 50% of the maximum H-reflex. At this level, the reflex is sensitive to both excitatory and inhibitory inputs. Others have suggested the use of a stimulus producing an H-reflex that is a constant percentage (e.g., 25%) of the Mmax (22). However, H-reflexes decrease markedly with age (9). Even some young subjects cannot produce an H-reflex that is 25% of Mmax, whereas others will produce this level with no M-wave. A measurable M-wave can help to maintain a constant stimulus level throughout a movement. If the stimulation electrodes move with respect to the nerve, then the stimulus will excite more or fewer motor neurons and presumably more or fewer sensory neurons in the nerve. In some studies on rhythmic movements, the experiment has been repeated with different stimulus levels (25). Then, H-reflexes can be selected at each phase of the cycle from trials when the M-wave has approximately the same value to compensate for the movement of the electrodes. Simonsen and Dyhre-Poulsen (24) found that the value of Mmax can vary as well with movement because the underlying muscle may move with respect to the recording electrodes with changing length, as first described more than 50 years ago (8). Thus, they urged that Mmax should be elicited at all phases of the movement and that H-waves should be normalized to the changing Mmax at each phase. However, a stimulus sufficient to generate a reliable Mmax at all phases of the step cycle may 1) require a high current to generate, 2) be painful, and 3) produce perturbations that can disrupt the movement under study. Finally, the variation is often small except for extreme movements. Thus, we recommend checking the level of M-waves and, hence, stimulus level throughout a movement but not necessarily the level of Mmax.

H-Reflexes and Postactivation Depression

Another variable is the rate of eliciting H-reflexes. Some researchers argue that intervals between eliciting H-reflexes should be 10 s or more because shorter intervals depress the H-reflex (22), probably by depression of transmission from primary muscle spindles (Ia sensory fibers) to motor neurons. Hultborn et al. (7) compared data from human reflexes and animal experiments with intracellular recording from motor neurons. In both humans and animals, the response only decreased in the pathway stimulated, so they referred to this as "postactivation depression" or "homosynaptic depression." Synaptic transmission from Ia fibers to motor neurons depends in a complex fashion on the rate of nerve impulses. As the rate is increased by up to 2-5 Hz, the excitatory postsynaptic potentials (EPSPs) recorded in cat motor neurons decrease progressively (15). However, at higher rates, EPSPs increase and reach a maximum at approximately 10-20 Hz but then decline again at frequencies of up to 100 Hz (15). Although the EPSPs may be smaller during a high-frequency train, there is a subsequent posttetanic potentiation of the reflex.

Some pathways in both experimental animals (15) and humans (22) are not depressed at long intervals, so the functional significance of postactivation depression remains uncertain. Pierrot-Deseilligny and Burke (22) argued that postactivation depression may be involved in the hyperreflexia associated with spasticity because the depression is reduced in people with spasticity after a spinal cord injury or stroke. They rated postactivation depression as the most important cause of spasticity in stroke and one of the leading causes in spinal cord injury (22, p.578). However, in Figure 2A, the post activation depression of the soleus H-reflex at rest is greatly reduced when the muscle is activated (22, p.99). The depression disappears if a subject stands in a posture that produces 15-20% of the maximum activation of the soleus EMG.

Two simple conclusions follow: 1) reflexes should not be measured at rest, if the goal is to understand their role in behavior because the results may be very different at rest during the movements of interest and 2) although postactivation depression can be demonstrated in both animal and human experiments, attributing functional significance to it or to changes after CNS lesions is dubious at best. Why are the results so different during behavior? One possible explanation is that human muscle spindles are often silent at rest (28); however, during movements, muscle spindles fire tens of impulses each second. As mentioned previously, transmitter release may be either facilitated or depressed, compared with the release when tested at rest, depending on the frequency of impulses (15). Stimuli that produce one or two extra impulses in a neuron (e.g., an Ia afferent) that is already firing tens of impulses per second will not produce significantly more depression or facilitation.

The H-reflex itself varies greatly depending on the level of subthreshold or suprathreshold activity. Figure 3 shows some examples schematically. At one membrane potential, the EPSP generated by a stimulus may not depolarize a large motor neuron to threshold (Fig. 3A) and will not contribute to the H-reflex. Even if it is not recruited to fire during weak voluntary activity, a depolarization of the membrane potential (Fig. 3B) will enable the EPSP to reach threshold and reliably generate a spike. Conversely, any action that hyperpolarizes the membrane potential will decrease the H-reflex without affecting the "rest" state of the target muscle. Thus, the "H-reflex at rest" has little meaning because we have no way of monitoring subthreshold membrane potentials that can vary with numerous influences.

The situation changes dramatically once some motor neurons are active and EMG is being generated. Subjects with very small H-reflexes at rest may increase them greatly during activity (Fig. 2B). For others with large H-reflexes at rest, the reflexes decrease or change little during activity. Figure 3 also provides a basis for these changes. Even if the EPSPs are very small in active motor neurons, they will contribute to the H-reflex on a fraction of the trials, when the neuron is already depolarized near threshold (Fig. 3). Other neurons will be depolarized, although not to threshold, and contribute because the EPSP can now reach threshold from the depolarized level. Both effects will tend to increase the small H-reflexes. With large H-reflexes, some neurons that fired all the time at rest will only fire a fraction of the time during voluntary activity because of refractoriness. Other neurons that did not fire at rest may be depolarized enough that the EPSPs can now reach threshold. Thus, for people with large H-reflexes at rest, the reflex during contraction can be larger or smaller, depending on the relative size of these two effects. In short, to be meaningful, reflexes should be measured during a level of voluntary contraction that matches the activity of interest.

Figure 2
Figure 2:
In a resting subject, the normalized H-reflex is progressively reduced (A) at shorter interstimulus intervals to 0.6 at an interval of 1 s (postactivation depression;solid line). However, the depression is considerably reduced (0.88; dashed line), if the subject contracts while seated, and disappears (1.00; dotted line), if the same voluntary activity is generated while standing. Data are the mean T SE of the 6 subjects. Voluntary activity increases the H-reflex substantially in subjects (B) who have a small H-reflex at rest. If the H-reflex is large at rest, the size declines ((Fig. 3)). Fifteen measurements were made from 10 subjects contracting while either sitting (dashed line) or standing (solid line).
Figure 3
Figure 3:
In a muscle at rest (A), a sensory stimulus (arrow) may produce an EPSP sufficient to reach threshold (horizontal line) and generate a nerve impulse (not drawn to scale) in small motor neurons (Mn.) but not in large motor neurons. After a brief interval, a second stimulus may not release enough synaptic transmitters to produce a nerve impulse even in the small neurons (postactivation depression). During voluntary activity (B), the stimulus will generate action potentials in active motor neurons on a fraction of trials, depending on how close the membrane potential is to the threshold on each trial. The stimulus may also reliably depolarize larger neurons to threshold, although they remain inactive otherwise.

There are more subtle issues based on suggestions that the reflex gain can change and new findings that the intrinsic properties of motor neurons can be modulated dramatically (12,22). Finally, reflexes change markedly, depending on the type of movement. Figure 4 shows that the H-reflex, measured at a given M-wave and level of voluntary contraction, is very different between standing and walking. We refer to this difference in reflexes between standing and walking with the same level of EMG as a "task-dependent reflex modulation." Reflexes are presumably modulated between different tasks such as standing and walking because of the different functional requirements of the two tasks. We will first consider the requirements in standing before discussing walking and running.

Figure 4
Figure 4:
The H-reflex varies at different phases of a cyclic movement, such as walking or running (phase dependence), with the level of EMG being generated in the muscle. At a given level of EMG, the H-reflex differs with the requirements of the behavior being performed (task dependence). In general, the variation is such that standing > walking > running. [Adapted from Stein, R.B., and C. Capaday. The modulation of human reflexes during functional motor tasks.Trend Neurosci. 11:328-332, 1998. Copyright © 1998 Elsevier. Used with permission.]



Traditionally, the rationale for the high reflex gain during standing was thought to be straightforward because the body can be considered as an inverted pendulum. The calf muscles typically show some activity to maintain the body at a position that is close to equilibrium, although an unstable one, because movement either forward or backward will tend to cause the body to fall. This precarious state might be stabilized, if reflexes restore the body to near the equilibrium position whenever the body sways a small distance. In this view, swaying forward stretches the calf muscles, and the stretch reflex restores the original position, if sufficiently strong to overcome the effects of gravity. Recently, the situation has become more complex. Patients who lack large sensory fibers such as the group Ia can balance to some degree using vision and other senses (22). In principle, balance could be maintained even without vision, if the stiffness of the ankle joint was sufficiently great. Then, body sway might produce a sufficient spring force to restore the body to the upright position. However, recent measurements show that the stiffness is between 10 and 35% below the value needed to stabilize the body (3).

Alternatively, reflexes from segmental receptors could effectively stabilize the ankle joints, but recent measurements of muscle length using a new ultrasound technique suggest another mechanism (14). Paradoxically, the calf muscles are shorter (not longer) when the ankle angle is smaller (the body is tilted forward, which stretches the calf muscles). In other words, the muscles are actively contracting and stretching the tendon. Furthermore, the muscle activity does not have the behavior expected of a simple reflex mechanism (i.e., responding continuously with a short delay to length and velocity changes). Rather, the EMG shows frequent, ballistic bursts of activity (approximately 2.6 times per second). These are referred to as ballistic bias impulses (13) and move the center of mass small distances (30-300 μm). Several of these bias impulses combine on average to correct sway movements, which occur at a lower peak frequency (<1 Hz). The nervous system responds to changes in the center of mass velocity but with a time delay of approximately 370 ms rather than tens of milliseconds, as expected for a reflex response.

Although the details are unknown, the data are consistent with the view that the nervous system functions not as a continuous controller of body position but as an intermittent control system that gathers information for more than several hundred milliseconds from all sensory systems. Then, probably, based on comparison with internal models of the body's orientation, a corrective, ballistic impulse is generated with the estimated, appropriate amplitude and direction. Loram et al. (13) give the following analogy: "Imagine trying to maintain a heavy ball as still as possible on a hillside. The ball is controlled by striking it with a bat at a relatively fixed rate. The motion of the ball will be caused by the blows themselves". "The batter has to judge the size of each blow. We suggest that in essence it is this never ending, trial and error process which has to be carried out in human standing". What is the role of reflexes, if any, in this process? As well as contributing to the general sensory information guiding this process, reflexes from muscle receptors may contribute to stabilize the body, if external or internal perturbations occur that are large and fast enough to disrupt the intermittent process described previously (22).


The situation is quite different during walking which generates large, purposeful movements. The calf muscles, such as soleus, are weakly active in the beginning of the stance phase; however, here, a reflex will be inappropriate for pro pulsion. The body is moving forward over the stance leg, and the movement stretches the calf muscles. A stretch reflex will impede this forward movement. The H-reflexes are small during this phase (Fig. 4) but still present. Perhaps, they help to stabilize the limb during the single-support phase of walking (10,22). Exactly how remains uncertain in the light of the changing views of balance mechanisms described previously.

Stein and Capaday (25) also described a phase dependence of the H-reflexes. Once the body has passed beyond the center of pressure of the leg (midstance), H-reflexes increase to near the values for standing. This makes functional sense because the reflex response will now have a vector that propels the body up and forward and hence aids the forward progression. In contrast, the calf muscles are again stretched during the swing phase. Any reflex response will now extend the ankle, and the foot may drag on the ground. During the swing phase, the reflex is largely suppressed in able-bodied individuals but not in some people with spasticity (22).

To be sure of the functional significance, stretches and electrical stimuli should be applied. A device was developed by Andersen and Sinkjær (1) at the University of Aalborg, Denmark that can rapidly perturb the ankle in either the flexion or extension direction. Figure 5A shows results using this device to produce gradual movements, referred to as enhancements and reductions (17). Note that the muscle is still being stretched by the forward movement of the body over the stance limb, although at a slightly faster or slower rate than would normally occur during walking. Figure 5 also shows clear changes in the soleus EMG that are proportional to velocity. The responses to stretches (enhancements) were affected by vibration and anesthetic that was selective for group Ia fibers, whereas the responses to reductions were not, suggesting a role for other muscle receptors in the reductions. Exactly which receptors are involved remains controversial, but the slow stretches certainly allow time for involving many receptors and pathways beyond the typical monosynaptic response. Nonetheless, slow perturbations can modulate the soleus EMG by up to ±40% (Fig. 5C), suggesting that reflex responses to stretch produce a substantial fraction of the normal EMG. This result does not disprove the existence of neural circuits that produce the basic walking pattern (central pattern generators). However, even if the basic pattern is produced centrally, reflexes can adapt it considerably to the environmental conditions (17). This result confirms earlier studies but uses more natural perturbations.

Figure 5
Figure 5:
A special device produces small deviations in ankle angle (A) from the trajectory that is normally followed during the stance phase of walking. The soleus EMG varies in response to the deviations (B) by up to 40% from the velocity changes (C). Thus, a substantial fraction of the EMG is probably generated by stretching the soleus muscle, as the body rolls over the foot in the stance phase of walking. [Adapted from Mrachacz-Kersting, N., B.A. Lavoie, J.B. Andersen, and T. Sinkj and r. Contribution of afferent feedback to the soleus muscle during human locomotion.J. Appl. Physiol. 93:167-177, 2005. Copyright © 2005 American Physiological Society. Used with permission.]


Stein and Capaday (25) originally studied running (Fig. 4B) and found that H-reflexes were reduced for a given level of EMG in this task as compared with walking. Simonsen and Dyhre-Poulsen (24) extended the range of speeds (up to 15 km·h−1) and made some improvements to the methodology, as described earlier. They found that the peak H-reflex actually increased from walking to high-speed running, although not in proportion to the increase in EMG. After analyzing the data in reduced gravity and eliminating points where there was little EMG, Ferris et al. (6) concluded that the main difference between walking and running was in the y-intercept (i.e., the value of the H-reflex in the absence of background EMG). Their conclusion differed from that shown in Figure 4B, where the slope of the relationship between the H-reflex and background EMG changed but not the intercept. This conclusion is rather strange because their results agreed with those shown in Figure 4B, when they included EMG data near zero, which would presumably help them estimate the y-intercept. Nonetheless, all studies agree that, for a given level of EMG, the H-reflex is highest during standing, intermediate in walking, and lowest during running.

Other Joints

Much less data are available for muscles other than soleus. The EMG pattern in quadriceps muscles differs from that in soleus. The activity is maximal during the transition from swing to stance and the early part of stance. The knee flexors are mainly coactive with the extensors. Although the knee flexes somewhat during early stance, the active contraction of the quadriceps muscles and the increased stiffness from the coactivation of flexors and extensors must prevent the knee from collapsing under the weight of the body. During the early stance phase of walking, the quadriceps H-reflex is actually larger than during standing, with matched M-waves and EMG levels, but decreases later in the stance phase and during the swing phase (22, p.547). The Aalborg device can also be used to perturb the knee joint (1). Responses were observed in all of the quadriceps muscles studied and the knee flexors. One striking result was the task dependence of the reflexes in the quadriceps muscles. The responses all increased with increased EMG activity during standing but decreased or did not change with increasing EMG activity during walking. This task dependence is probably caused by the differences in presynaptic inhibition, as will be discussed later. Mrachacz-Kersting et al. (20) explain the functional significance of the task dependence as follows: "the strongest reflex amplitudes were observed during the later part of the stance phase, at which time the background activity had already diminished. At this time the center of gravity moves over the foot in preparation for the following swing phase. A flexion of the knee joint would cause the center of gravity to shift to a position in front of the foot, making it difficult to maintain balance. In contrast, during standing the center of gravity is directly above the feet and not moving in relation to them." In conclusion, recent evidence indicates that strong excitatory reflexes control movements in both ankle and knee joint during walking.


Muscle spindles also have pathways that inhibit antagonist motor neurons (Fig. 6). These pathways are probably involved in the reciprocal organization of many of our movements. Thus, when flexor muscles are activated, the corresponding extensors are generally inhibited, and vice versa. This reciprocal organization applies to cutaneous and other pathways (e.g., the flexor reflex) and to descending voluntary control of some but not all movements (22). For example, we can coactivate antagonist muscle groups to stiffen the limb in one position. The reciprocal inhibitory pathway is strongly but not exclusively activated by group Ia fibers. The shortest latency pathway involves a single interneuron, as shown in Figure 6A, and is therefore referred to as disynaptic inhibition.

Figure 6
Figure 6:
Two methods of studying reciprocal inhibition (A). In the voluntary inhibition method (B), the common peroneal (CP) nerve is stimulated just above threshold to generate a small M-wave and H-reflex in the EMG from the tibialis anterior muscle (not shown in the diagram). Shortly after the TA H-reflex, the rectified, voluntary EMG of the antagonist, soleus muscle decreases (C). The delay is caused by the presence of at least one interneuron in the inhibitory pathway. In the conditioned H-reflex method (D), a test stimulus to the tibial nerve is given on each trial to generate an M-wave and an H-reflex in soleus (solid line), as described in Figure 1. On half of the trials, a second conditioning stimulus is given to the CP nerve a few milliseconds earlier than the test stimulus, which slightly inhibits the H-reflex (dashed line).

Short-Latency Reciprocal Inhibition

One method to study short-latency inhibition involves measuring the depression of ongoing voluntary activity (Fig. 6C). In the example shown, the common peroneal (CP) nerve was stimulated at just above the M-wave threshold. A small M-wave and an H-reflex are seen in the tibialis anterior (TA) muscle that is innervated by the CP nerve (Fig. 6B). The antagonist (soleus) muscle was contracted voluntarily at approximately 15-20% of the maximum voluntary contraction (MVC). Little change is observed in the soleus muscle EMG until just after the H-reflex in TA (EMG was rectified and averaged; 15 stimuli at 2-3 s apart). Then, the voluntary activity decreases for approximately 10 ms (dotted line in Fig. 6B). This is caused by the reciprocal inhibition, and the duration of the inhibition is what would be expected for a brief synaptic current (27). A second method involves pairing a stimulus to the CP nerve, with one to the tibial nerve (Fig. 6D). The stimulus to the tibial nerve produces an M-wave and an H-reflex in the soleus muscle, as described previously (Fig. 1). If the CP nerve is stimulated a few milliseconds before the tibial nerve, impulses will have time to conduct to the spinal cord and excite the interneurons to inhibit the soleus motor neurons (Fig. 6A). Therefore, pairing the stimuli will reduce the H-reflex.

Which method is better to measure inhibitory reflexes during voluntary activity? We tested both methods on six subjects. Figure 7A shows the effect of three different intensities of stimulation (1, 1.25, and 1.5 × motor threshold) and three different levels of voluntary contraction (10, 20, and 30% MVC; not shown). For method 1, the percent inhibition was independent of the level of voluntary contraction and was significantly related to the stimulus intensity (P < 0.001). For method 2, we tested the three different levels of CP stimulus intensity, two levels of voluntary contraction (0 and 20% MVC), and two levels of tibial nerve stimulation, which gave approximately 0.5 Hmax and Hmax. Figure 7B shows the average results for the four conditions. Only the condition with 0.5 Hmax at rest showed a significant inhibition. The other conditions showed a small mean level of inhibition, but the variability between subjects (note the large SE bars) was such that the mean level was not significant. Thus, the results are sensitive to several factors (22, p.208). The measured inhibition also increases during dorsiflexion of the ankle and almost disappears during plantar flexion (23). These voluntary contractions modify the size of the H-reflex as well, so the effects interact with one another and make it extremely difficult to interpret, when alternating dorsiflexion and plantar flexion form part of the behavior, as in walking. The conditioned H-reflex method is traditionally measured at rest, and it can only be used in the few muscles that show an H-reflex at rest. It will also be affected by subthreshold changes (Fig. 3).

Figure 7
Figure 7:
The inhibition produced by method 1 (voluntary inhibition) was directly related to the stimulus intensity (A), whereas that from method 2 (conditioned H-reflex) was smaller, relatively more variable, and not significantly dependent on stimulus intensity (mean ± SE). In fact, the inhibition was only significant in one of the four conditions studied (0.5 Hmax at rest). With larger stimuli that produced Hmax or with voluntary contraction, the inhibition was not significant. Thus, the voluntary inhibition method is preferable, particularly when studying voluntary movements (26).

In summary, our results show that the inhibition of voluntary contraction (method 1) is independent of the level of contraction and varies directly with the level of stimulation. In contrast, the conditioned H-reflex (method 2) depends on the following: 1) the level of voluntary contraction, 2) the level of CP stimulation, 3) the level of the tibial nerve stimulation, and 4) the time interval between the CP and the tibial nerve stimulation. There is only a 3-ms window in which inhibition occurs (22), despite the fact that inhibitory postsynaptic potentials typically last much longer (5). Also, because of the small size of the effect, compared with that of method 1 (Fig. 7A), more stimuli must be used. In the experiments of Figure 7A, we estimated that approximately 70 stimuli (35 test and 35 conditioned) are required to obtain statistically reliable results, compared with 15 stimuli during voluntary inhibition. Despite the methodological concerns, there is now reasonable agreement that reciprocal inhibition is modulated between standing, walking, and running and decreases with age (10,11,22). However, the changes are not as dramatic as those shown by the H-reflexes.

Other Forms of Inhibition

There are several other forms of inhibition, for example, from Golgi tendon organs, although this inhibition can reverse to excitation during walking (18). There are also nonreciprocal forms of inhibition that act at one or more joints. In general, both the excitatory and inhibitory connections across joints seem to be much more widespread in humans than in cats, where they have been studied most extensively (5,22). This may be caused by the more complex requirements of bipedal walking, but widespread connections are also seen in the human upper limb. More importantly, reflex connections in cats have traditionally been studied under anesthesia, which will particularly depress multisynaptic connections.

Pierrot-Deseilligny and Burke (22) developed an ingenious method to study the recurrent inhibition mediated through Renshaw cells, pairing an H-reflex stimulus with one 10 ms later that generates a maximum M-wave. Voluntarily activating a muscle can increase or decrease the amount of Renshaw inhibition, which lasts for tens of milliseconds. This has been interpreted in terms of a variable gain control, but other explanations are possible (29). The inhibition from Renshaw cells is widely distributed, and the inhibition from soleus to the quadriceps muscles is reduced in those tasks (e.g., standing and late stance phase of walking) when the muscles are co-contracted.

Presynaptic inhibition can reduce transmission through many sensory pathways for 100 ms or more. It serves as a gain control on the first and second synaptic relays in these pathways and is responsible for much of the task dependence of H-reflexes (25). The inhibition can be measured as a decrease in an H-reflex produced by a brief vibration of an antagonist tendon or a stimulus to the antagonist nerve (22). Thus, modulation of the H-reflex can be studied on a millisecond time scale (reciprocal inhibition), a 10-ms time scale (recurrent inhibition), or a 100-ms time scale (presynaptic inhibition). Some of the same methodological concerns apply to presynaptic inhibition as to the previous types, plus additional ones (22). Because the response is so delayed, there is time for many pathways to be involved. As a result, changes in the H-reflex may differ from those in a tendon jerk or stretch reflex (19,22). Also, responses that are secondary to muscle contractions elicited by the stimulus can occur. Thus, changes in ongoing voluntary activity (our preferred method for reciprocal inhibition) are difficult to interpret. Again, we recommend a small level of voluntary activity while ensuring that the level is maintained until the delayed H-reflex occurs. The effects tend to be larger with presynaptic inhibition, and the timing is less critical, so the conditioned H-reflex technique can yield useful results.


Finally, we return to the three questions that were raised in the title of the article. First, how should reflexes be measured during movement? Clearly, electrical stimulation with EMG recording is the most convenient method. Stimulation should be applied, as far as possible, while the subjects are performing the movements of interest, and must be used with care. The nervous system is complex and adaptive, so it is all too easy to think that you are measuring one thing but are actually measuring something else. Although they also have limitations, specialized devices that perturb movements mechanically and new methods, such as ultrasound, to detect muscle movements have produced some major and recent advances. The simplest (monosynaptic and disynaptic) reflexes from large afferents are the easiest to measure, but clever methods have been developed to measure more complex pathways, as described here and, in more detail, in another reference (22). Second, what do reflexes contribute to various movements? The evidence is now quite convincing that excitatory reflexes contribute to a substantial fraction of the force generated during human walking and help to adapt it to the terrain. Inhibitory pathways modulate the activity appropriately for the requirements of different tasks and phases within the tasks. However, reflexes surprisingly do not seem to contribute much in quiet standing, although they will respond to larger perturbations that threaten balance. Third, why should they be measured? The answer is simply that there are important task- and phase-dependent changes in reflexes in different human movements. Understanding these adaptations is essential to understanding how we control movement, and only a very few basic movements have been studied to date with suitable methods.


The authors thank Drs. Dave Bennett, Charles Capaday, Dave Collins, Monica Gorassini, and Kelvin Jones for helpful comments on the manuscript. This work was supported by grants from the Canadian Institutes of Health Research and the Christopher Reeve Paralysis Foundation.


1. Andersen, J.B., and T. Sinkjær. An actuator system for investigating electrophysiological and biomechanical features around the human ankle joint during gait. IEEE Trans. Rehabil. Eng. 3:299-306, 1994.
2. Brooke, J.D., and E.P. Zehr. Limits to fast-conducting somatosensory feedback in movement control. Exerc. Sport Sci. Rev. 34:22-28, 2006.
3. Casadio, M., P.G. Morasso, and V. Sanguineti. Direct measurement of ankle stiffness during quiet standing: implications for control modelling and clinical application. Gait Posture 21:410-424, 2005.
4. Duysens, J., F. Clarac, and H. Cruse. Load-regulating mechanisms in gait and posture: comparative aspects. Physiol. Rev. 80:83-133, 2000.
5. Eccles, J.C. The Physiology of Synapses, Berlin, Germany: Springer-Verlag, 1964.
6. Ferris, D.P., P. Aagaard, E.B. Simonsen, C.T. Farley, and P. Dyhre-Poulsen. Soleus H-reflex gain in humans walking and running under simulated reduced gravity. J. Physiol. 530:167-180, 2001.
7. Hultborn, H., M. Illert, J. Nielsen, A. Paul, M. Ballegaard, and H.Wiese. On the mechanism of the post-activation depression of the H-reflex in human subjects. Exp. Brain Res. 108:450-462, 1996.
8. Inman, V.T., H.J. Ralston, J.B. Saunders, B. Feinstein, and E.W. Wright Jr. Relation of human electromyogram to muscular tension. Electroencephalogr. Clin. Neurophysiol. Suppl. 4:187-194, 1952.
9. Kido, A., N. Tanaka, and R.B. Stein. Spinal excitation and inhibition decrease as humans age. Can. J. Physiol. Pharmacol. 82:238-248, 2004.
10. Kido, A., N. Tanaka, and R.B. Stein. Spinal reciprocal inhibition in human locomotion. J. Appl. Physiol. 96:1969-1977, 2004.
11. Lavoie, B.A., H. Devanne, and C. Capaday. Differential control of reciprocal inhibition during walking versus postural and voluntary motor tasks in humans. J. Neurophysiol. 78:429-438, 1997.
12. Li, Y., M.A. Gorassini, and D.J. Bennett. Role of persistent sodium and calcium currents in motoneuron firing and spasticity in chronic spinal rats. J. Neurophysiol. 91:767-783, 2004.
13. Loram, I.D., C.N. Maganaris, and M. Lakie. Human postural sway results from frequent, ballistic bias impulses by soleus and gastrocnemius. J. Physiol. 564:295-311, 2005.
14. Loram, I.D., C.N. Maganaris, and M. Lakie. The use of ultrasound to make non-invasive, in vivo measurement of continuous changes in human muscle contractile length. J. Appl. Physiol. 93:167-177, 2005.
15. Luscher, H., P.W. Ruenzel, and E. Henneman. Effects of impulse frequency, PTP, and temperature on responses elicited in large populations of motoneurons by impulses in single Ia-fibers. J. Neurophysiol. 50:1045-1058, 1983.
16. Matthews, P.B.C. Review lecture: evolving views on the internal operation and functional role of the muscle spindle. J. Physiol. 320:1-30, 1981.
17. Mazzaro, N., M.J. Grey, and T. Sinkjaer. Contribution of afferent feedback to the soleus muscle activity during human locomotion. J. Neurophysiol. 93:167-177, 2005.
18. McCrea, D.A., S.J. Shefchyk, M.J. Stephens, and K.G. Pearson. Disynaptic group I excitation of synergist ankle extensor motoneurons during fictive locomotion in the cat. J. Physiol. 487:527-539, 1995.
19. Morita, H., N. Petersen, N. Christensen, L.O. Sinkjaer, and T. Nielsen. Sensitivity of H-reflexes and stretch reflexes to presynaptic inhibition in humans. J. Neurophysiol. 80:610-620, 1998.
20. Mrachacz-Kersting, N., B.A. Lavoie, J.B. Andersen, and T. Sinkjaer. Characterisation of the quadriceps stretch reflex during the transition from swing to stance phase of human walking. Exp. Brain Res. 159:108-122, 2004.
21. Petersen, N., H. Morita, and J. Nielsen. Modulation of reciprocal inhibition between ankle extensors and flexors during walking in man. J. Physiol. 520:605-619, 1999.
22. Pierrot-Deseilligny, E., and D. Burke. The Circuitry of the Human Spinal Cord: Its Role in Motor Control and Movement Disorders, New York, NY: Cambridge University Press, 2005.
23. Shindo, M., H. Harayama, K. Kondo, N. Yanagisawa, and R. Tanaka. Changes in reciprocal Ia inhibition during voluntary contraction in man. Exp. Brain Res. 53:400-408, 1984.
24. Simonsen, E.B., and P. Dyhre-Poulsen. Amplitude of the human soleus H reflex during walking and running. J. Physiol. 515:929-939, 1999.
25. Stein, R.B., and C. Capaday. The modulation of human reflexes during functional motor tasks. Trends Neurosci. 11:328-332, 1988.
26. Stein, R.B., and S.C. McGie. Reciprocal inhibition and post-activation depression following electrical stimulation of human nerves. Can. Physiol. Soc., Lake Louise, 2006.
    27. Turker, K.S., and R.K. Powers. Black box revisited: a technique for estimating postsynaptic potentials in neurons. Trends Neurosci. 28:379-386, 2005.
    28. Vallbo, A.B., K.E. Hagbarth, H.E. Torebjork, and B.G. Wallin. Somatosensory, proprioceptive, and sympathetic activity in human peripheral nerves. Physiol. Rev. 59:919-957, 1979.
    29. Windhorst, U. On the role of recurrent inhibitory feedback in motor control. Prog. Neurobiol. 49:517-587, 1996.
    30. Zehr, E.P., and R.B. Stein. Interaction of the Jendrássik maneuver with segmental presynaptic inhibition. Exp. Brain Res. 124:474-480, 1999.

    movement; human; excitation; inhibition; task dependence; phase dependence

    ©2006 The American College of Sports Medicine