H-reflex modulation during passive lengthening and shortening of the human triceps surae


  • G. J. Pinniger,

    1. Department of Biomedical Science, University of Wollongong, Wollongong, NSW 2522, Australia
    Search for more papers by this author
  • M. M. Nordlund,

    1. Department of Neuroscience, Karolinska Institute, Box 5626, SE-11486, Stockholm, Sweden
    2. Stockholm University College of Physical Education and Sports, Box 5626, SE-11486, Stockholm, Sweden
    Search for more papers by this author
  • J. R. Steele,

    1. Department of Biomedical Science, University of Wollongong, Wollongong, NSW 2522, Australia
    Search for more papers by this author
  • A. G. Cresswell

    1. Department of Neuroscience, Karolinska Institute, Box 5626, SE-11486, Stockholm, Sweden
    2. Stockholm University College of Physical Education and Sports, Box 5626, SE-11486, Stockholm, Sweden
    Search for more papers by this author

Corresponding author A. G. Cresswell: Department of Neuroscience, Karolinska Institute, Box 5626, SE-11486 Stockholm, Sweden. Email: andrew.cresswell@neuro.ki.se


  • 1The present study investigated the effects of lengthening and shortening actions on H-reflex amplitude. H-reflexes were evoked in the soleus (SOL) and medial gastrocnemius (MG) of human subjects during passive isometric, lengthening and shortening actions performed at angular velocities of 0, ±2, ±5 and ±15 deg s−1.
  • 2H-reflex amplitudes in both SOL and MG were significantly depressed during passive lengthening actions and facilitated during passive shortening actions, when compared with the isometric H-reflex amplitude.
  • 3Four experiments were performed in which the latencies from the onset of movement to delivery of the stimulus were altered. Passive H-reflex modulation during lengthening actions was found to begin at latencies of less than 60 ms suggesting that this inhibition was due to peripheral and/or spinal mechanisms.
  • 4It is postulated that the H-reflex modulation seen in the present study is related to the tonic discharge of muscle spindle afferents and the consequent effects of transmission within the Ia pathway. Inhibition of the H-reflex at less than 60 ms after the onset of muscle lengthening may be attributed to several mechanisms, which cannot be distinguished using the current protocol. These may include the inability to evoke volleys in Ia fibres that are refractory following muscle spindle discharge during rapid muscle lengthening, a reduced probability of transmitter release from the presynaptic terminal (homosynaptic post-activation depression) and presynaptic inhibition of Ia afferents from plantar flexor agonists. Short latency facilitation of the H-reflex may be attributed to temporal summation of excitatory postsynaptic potentials arising from muscle spindle afferents during rapid muscle lengthening. At longer latencies, presynaptic inhibition of Ia afferents cannot be excluded as a potential inhibitory mechanism.

It has been argued that modulation of the H-reflex during dynamic activities reflects changes in motoneurone excitability brought about to suit the functional demands of the task. Reflex modulation has been reported during various dynamic tasks including passive single joint movements (Mark et al. 1968; Gottlieb & Agarwal, 1978; Robinson et al. 1982; Etnyre & Abraham, 1986; Romano & Schieppati, 1987; Guissard et al. 1988; Voigt & Sinkjaer, 1998) and multijoint movements such as stepping and cycling (Brooke et al. 1995a; Cheng et al. 1995a; Misiaszek et al. 1995a), as well as during more functional tasks such as walking and running (Capaday & Stein, 1986, 1987a; Yang & Whelan, 1993; Faist et al. 1996; Simonsen & Dyhre-Poulsen, 1999). Factors that are known to alter H-reflex amplitude include the level of background activation (Butler et al. 1993), muscle length (Gerilovsky et al. 1989), movement velocity (Romano & Schieppati, 1987), activation history (Proske et al. 1993; Gregory et al. 1998) and muscle composition (Messina & Cotrufo, 1976). Although a number of mechanisms have been proposed to account for H-reflex modulation, such as homosynaptic post-activation depression and presynaptic inhibition of Ia afferents, the origin and functional significance of such modulation is still unclear. Much of the debate concerning this issue may hinge around methodological differences and the generalization of findings between functionally different tasks.

It is generally agreed that the H-reflex is strongly modulated throughout the gait cycle (Capaday & Stein, 1986, 1987a; Yang & Whelan, 1993; Garrett et al. 1994; Faist et al. 1996; Simonsen & Dyhre-Poulsen, 1999). For example, Capaday & Stein (1986) reported an increase in soleus (SOL) H-reflex amplitude late in the stance phase when the triceps surae is active to propel the body forward and an absence during the swing phase when the foot is dorsiflexed and SOL activation would hinder movement. However, Yang & Whelan (1993) showed the SOL H-reflex to be depressed despite voluntary activation of the SOL during the swing phase, indicating that this reflex modulation was not merely a consequence of central drive, and suggested presynaptic inhibition of primary afferents as a likely mechanism. Moreover, Capaday & Stein (1987a) reported that when gait velocity increased from walking (4 km h−1) to running (8 km h−1), a decrease in SOL H-reflex amplitude occurred despite increased underlying SOL activation.

Functional reflex modulation has also been reported during landing and hopping tasks (Moritani et al. 1990; Dyhre-Poulsen et al. 1991). In the study of Dyhre-Poulsen et al. (1991) the SOL H-reflex was strongly inhibited when landing from a downward jump with the inhibition commencing up to 100 ms prior to foot contact. During hopping, however, the SOL H-reflex increased before foot contact despite a reduction in SOL EMG and in the presence of antagonist activation. The functional significance of this modulation was interpreted in relation to muscle stiffness and the transfer of muscle from a spring to a damping unit. Moritani et al. (1990) investigated reflex modulation in two functionally different muscles, SOL and medial gastrocnemius (MG), during hopping tasks of different heights and frequencies. H-reflexes were modulated differently in the two muscles with those in the ‘fast’ MG increasing as the amplitude and frequency of jumping movements increased whereas those in SOL decreased. These findings support the notion that these peripheral reflexes are modulated in a manner to suit the functional demands of the task. However, caution must be taken when interpreting the amplitude of the reflex response in a functional sense, as Kearney et al. (1999) have reported that the reflex EMG response does not necessarily correspond to the resultant reflex torque.

Regardless of the resultant torque contribution, the fact that modulation of reflex amplitude is somewhat independent of central drive indicates that reflex magnitude is not merely a reflection of motoneurone excitability, but can also be influenced by additional neural mechanisms. In the light of model-based simulations of reflex output from a motoneurone pool (Capaday & Stein, 1987b), and experimental evidence from decerebrate cats (Capaday & Stein, 1989), it is the prevailing view of the Capaday group that the observed reflex modulation during gait was due to a centrally regulated increase in the presynaptic inhibition of Ia terminals. It was postulated that the functional significance of such a mechanism was to reduce central gain of the stretch reflex to avoid saturation of the motor output during running.

While Brooke et al. (1995b) agree that the mechanism of this reflex depression is acting at a premotoneuronal level, they suggest a peripheral origin for this inhibitory influence. During passive stepping and cycling movements modulation of the SOL H-reflex corresponded to movements of the knee and hip joints and was attributed to the influence of heteronymous Ia afferents arising from the lengthening quadriceps muscles (Brooke et al. 1995a). However, the relevance of these findings to reflex modulation during human gait was recently challenged by Schneider et al. (2000) and Garrett et al. (1999), who found a limited influence of knee motion on SOL H-reflex modulation during human gait. Unfortunately, the inhibitory influence of motion about the hip was not controlled and therefore cannot be ruled out as a potential inhibitory mechanism.

Significant H-reflex depression has been recorded within 50 ms of the onset of passive pedalling, a latency that precludes the involvement of higher motor centres (Brooke et al. 1995b). Similar rapid reflex modulation has been reported in spinal cord-injured patients during passive pedalling (Brooke et al. 1995b) and passive, sinusoidal ankle rotations (Voigt & Sinkjaer, 1998), again suggesting a peripheral origin for this reflex modulation.

Although some authors attribute inhibition of the SOL H-reflex to heteronymous Ia afferent influences resulting from hip and knee movement (Brooke et al. 1995a; Misiaszek et al. 1995a), a long-lasting reflex depression may also arise from homonymous influences (Hultborn et al. 1996; Wood et al. 1996; Kohn et al. 1997). Hultborn et al. (1996) investigated long-lasting depression of the SOL H-reflex following passive triceps surae lengthening. From experiments on both humans and cats, it was shown that the reduction in H-reflex amplitude was related to depression of Ia transmission specific to the afferent fibres related to the dynamic task (homonymous Ia afferents). Furthermore, this inhibition was seen without alteration to motoneurone excitability and, although a dorsal root potential was recorded, indicating activity of presynaptic inhibitory interneurones, the time course of this potential (< 500 ms) was considerably shorter than the long-lasting inhibition of the H-reflex (8-12 s). The authors postulated that this phenomenon was due to the reduced probability of transmitter release from the Ia fibres (homosynaptic post-activation depression) rather than traditional GABAergic presynaptic inhibition.

Similar long-lasting H-reflex depression (up to 15 s) has been reported following passive triceps surae lengthening (Wood et al. 1996), during passive cycling movements (Misiaszek et al. 1995a) and following a quadriceps tendon tap (Cheng et al. 1995b). In the study of Wood et al. (1996) equivalent experiments were performed in the cat in which muscle spindle firing was assessed during the preceding muscle stretch. Modulation of H-reflex amplitude appeared to coincide with increased muscle spindle discharge associated with the preceding change in muscle length. However, in the experiments of Misiaszek et al. (1995a) and Cheng et al. (1995b) the ankle was braced to prevent changes in SOL muscle length and, hence, prevent variations in the discharge of homonymous Ia afferents. Therefore, long-lasting H-reflex depression may occur in the absence of homonymous Ia afferent influences (Cheng et al. 1995b; Misiaszek et al. 1995a) or without variations in the discharge of heteronymous Ia afferents (Hultborn et al. 1996; Wood et al. 1996). Based on these findings, one may speculate that: (1) different peripherally mediated inhibitory mechanisms are responsible for the long-lasting H-reflex depression seen during single passive joint rotation and multijoint movements, or (2) a single, centrally modulated inhibitory mechanism is responsible for the long-lasting reflex modulation seen during both single and multijoint movements.

In an attempt to better understand the mechanisms at play during single passive joint rotations this study was conducted with two primary objectives. Firstly, H-reflex modulation was examined during lengthening and shortening actions of the triceps surae under tightly controlled experimental procedures to ensure that alterations in reflex magnitude were not due to variations in effective stimulus intensity. Secondly, although short latency (< 50 ms) reflex modulation has been recorded during passive cycling movements, presumably due to heteronymous Ia influences, to our knowledge, reflex modulation has not been investigated over a similar short latency during passive single joint movements involving homonymous Ia afferents. Therefore, reflex responses were recorded at various latencies following the onset of movement in an attempt to determine the time-course of homonymous Ia afferent influences on H-reflex amplitude. Additionally, H-reflexes were recorded from both SOL and MG during all experiments to assess whether reflex modulation was influenced by the functional characteristics of a muscle.



Data were obtained during a total of 10 experiments in eight healthy subjects (two females and six males, aged 25-52 years) with no history of neurological injury or disease. All subjects provided written informed consent before participating in the study. The experimental procedures were approved by the ethics committee of the local research institute and conformed to the Declaration of Helsinki.

Subjects lay prone on a testing bench with their left foot securely attached to a microprocessor-controlled ankle torque motor that accurately manipulated ankle displacement, velocity and acceleration (Pinniger et al. 2000). The subject's ankle joint was aligned with the axis of the torque motor and the knee of their left leg flexed approximately 20 deg from full extension. At the beginning of the experiment, subjects assumed a comfortable position in which they remained for the duration of the experiment (approximately 3 h). Subjects were required to close their eyes and wear headphones to minimize distractions from the surrounding environment. To avoid alterations in H-reflex amplitude due to changes in muscle length (Gerilovsky et al. 1989), stimulation was always delivered at an ankle angle of 90 deg. After completing each trial the foot was moved back to the starting position and the subject performed a submaximal isometric plantar flexion effort before starting the next trial to remove any thixotropic effects (Proske et al. 1993) resulting from the previous change in muscle length.

Stimulation and recording

H-reflexes and M-waves were evoked in SOL and MG by percutaneous stimulation of the tibial nerve with a single 1 ms pulse delivered from a constant current stimulator (model DS7A, Digitimer, UK). The cathode (Neuroline surface electrode, Medicotest, Denmark) was placed in the popliteal fossa and the anode (100 mm × 50 mm rubber electrode) placed proximal to the patella. The location of the stimulating electrode was carefully selected to obtain maximum amplitude reflexes from SOL and MG. Peak-to-peak amplitudes of the H-reflex and M-wave were calculated and plotted online. In some experiments the area and duration of the first negative phase of the H-reflexes and M-waves were calculated offline to confirm results obtained from amplitude calculations.

SOL and MG H-reflexes were recorded using bipolar Ag-AgCl surface electrodes (Sensor Medics, USA). Belly-tendon electrode configurations were utilized with the belly electrode placed below the gastrocnemius bifurcation for SOL and on the bulk of the muscle for MG. Tendon electrodes for both muscles were placed on the distal aspect of the Achilles’ tendon. EMG signals were amplified 200 times (NL 824, Digitimer), bandpass filtered (30-1000 Hz; Neurolog, NL 125, Digitimer) and then subjected to analog-to-digital conversion at a sample rate of 10 kHz using a 16 bit Power 1401 and signal data collection system (Cambridge Electronic Design, UK). During some experiments, electrodes were also placed over the belly of tibialis anterior to confirm that the antagonist muscle was not active during passive lengthening and shortening of the triceps surae.

Passive reflex modulation

The sensitivity of the H-reflex to facilitation and inhibition varies with respect to the size of the control H-reflex (Crone et al. 1990). Therefore, to select suitable test stimulus intensities, passive isometric recruitment curves were obtained for each subject at the start of each experiment. Recruitment curves were also recorded during passive lengthening and shortening actions at each angular velocity to determine the influence of possible variations in effective stimulus intensity, which were probably due to alterations in the spatial relationship between the stimulating electrode and the tibial nerve. Passive isometric and dynamic recruitment curves recorded from SOL and MG for one individual are presented in Fig. 1.

Figure 1.

Representative data from one subject. A-D, passive recruitment curves for SOL lengthening (A) and shortening (B) actions and MG lengthening (C) and shortening (D) actions at angular velocities of ±15 (circles), ±5 (squares) and ±2 deg s−1 (triangles). For comparison, isometric recruitment curves are plotted on each figure (diamonds). M-waves are plotted as open symbols and H-reflexes as filled symbols. E, H-reflexes at 50 %Hmax at each angular velocity for SOL (top trace) and MG (bottom trace). Note the dramatic depression of H-reflexes at each lengthening velocity.

From this data the maximum amplitude of the H-reflex and the M-wave (Hmax and Mmax, respectively) were identified and the Hmax: Mmax ratio calculated for each angular velocity. Based on the isometric recruitment curve, a stimulus intensity sufficient to elicit a response equivalent to 50 %Hmax on the ascending limb of the recruitment curve was selected as the test stimulus for the following experiments. At this stimulus intensity, the mean isometric H-reflex amplitude for SOL was 51.0 %Hmax and for MG, 38.6 %Hmax. H-reflexes and M-wave responses at this stimulus intensity were recorded during 15 trials of passive isometric, lengthening and shortening actions. Lengthening and shortening actions were performed at angular velocities of ±2, ±5 and ±15 deg s−1 through an ankle displacement of 10 deg (85-95 deg). Lengthening trials are indicated as negative angular velocities. Each trial was separated by a minimum of 20 s and the order of velocity and action type was counterbalanced between subjects.

At a stimulus intensity of 50 %Hmax the mean M-wave amplitude was a small fraction of Mmax (5.0 ± 0.4 and 5.8 ± 1.0 % of Mmax for SOL and MG, respectively; see Fig. 1). At such small amplitudes, the M-wave is not sensitive to changes in stimulus intensity and, therefore, may not be a reliable measure to assess the efficacy of the stimulus intensity. Therefore, a second series of passive isometric, lengthening and shortening actions was performed with a stimulus intensity set to obtain an M-wave with an amplitude equivalent to 50 % of the isometric maximum M-wave amplitude (50 %Mmax). At this position on the recruitment curve, the amplitude of the M-wave is sensitive to small changes in stimulation intensity; therefore, a change in the spatial relationship between the tibial nerve and the stimulating electrode can be identified easily. Mean isometric M-wave amplitudes at 50 %Mmax were 50.3 %Mmax for SOL and 32.3 %Mmax for MG. Ten trials at each velocity were tested with the stimulus intensity of 50 %Mmax under the same experimental conditions as described above.

Short latency reflex modulation

To examine whether H-reflex modulation was related to supraspinal, or spinal/peripheral mechanisms, additional experiments were performed on four subjects in whom the latency from the onset of the movement to delivery of the stimulus was altered while maintaining a constant angular displacement. The subject's ankle was moved passively through 10 deg from an initial starting angle of either 91 deg for lengthening or 89 deg for shortening actions. Again, 15 stimuli at 50 %Hmax and 10 stimuli at 50 %Mmax were delivered at a 90 deg ankle angle; that is, after having moved through 1 deg of dorsi- or plantarflexion. These movements were performed at shortening and lengthening angular velocities of ±30, ±15 and ±7 deg s−1 and, accounting for acceleration of the foot, the corresponding latencies from the start of the movement until the ankle reached 90 deg were 57, 90 and 165 ms, respectively. Allowing 15 ms for transmission of the afferent volley within the Ia fibres to the motoneurone pool, stimulations to the tibial nerve were delivered at latencies of 42, 75 and 150 ms from the onset of movement. Therefore, it was assumed that the stimulation reached the motoneurone pool when the ankle was at 90 deg for each condition.

At a latency of 57 ms it can be argued that there is insufficient time for supraspinal mechanisms to influence H-reflex amplitude; therefore, any modulation to the H-reflex amplitude at this interval must be due to peripheral/spinal mechanisms (Brooke et al. 1995b). At latencies greater than 60 ms from the onset of movement there is sufficient time for input via a transcortical loop to influence the magnitude of the reflex response, whereas at latencies greater than 120 ms the influence of voluntary activation may alter H-reflex amplitude.

Data analysis

Unless otherwise stated, results are presented as means ±s.e.m. Significant main effects of angular velocity and muscle action type on H-reflex and M-wave amplitude for SOL and MG were determined by using a repeated measures analysis of variance design. Post hoc analyses were then performed using Tukey's honestly significant difference test to identify the nature of any significant main effect. Significance was accepted at P≤ 0.05. All statistical analyses were performed using Statistica software (StatSoft, USA).


Passive reflex modulation

Passive isometric, lengthening and shortening recruitment curves for SOL and MG of one subject, whose data best represent the group data, are displayed in Fig. 1. It is evident that, despite consistent M-wave amplitudes, both SOL and MG H-reflex recruitment curves are consistently depressed in amplitude across the range of stimulus intensities and, consequently, the recruitment curves were compressed in the time domain during lengthening actions when compared to isometric and corresponding shortening actions. This finding is similar to that of Misiaszek et al. (1995b) reported during passive cyclic movements. Statistical analysis of the Hmax: Mmax ratio (Fig. 2) confirmed this depression as the lengthening ratios were significantly lower than the isometric and corresponding shortening ratios. Furthermore, the Hmax: Mmax ratios at -15 deg s−1 were significantly less than those at -2 deg s−1 for both SOL and MG. Individual recordings from SOL and MG at a stimulus intensity of 50 %Hmax for each angular velocity are presented in Fig. 1E. Despite the difference in relative stimulus intensities, SOL and MG H-reflex amplitudes exhibited a remarkably similar modulation with respect to velocity and muscle action type (mean data for all subjects are presented in Fig. 3).

Figure 2.

Mean ±s.e.m. Hmax: Mmax ratios during passive lengthening, shortening and isometric actions of SOL (▪) and MG (□). Negative values indicate lengthening actions. Although different in magnitude, SOL and MG Hmax: Mmax ratios behave in a similar manner across velocities.

Figure 3.

Mean H-reflex (•) and M-wave (○) amplitudes for all subjects recorded from SOL and MG at stimulus intensities of 50 %Hmax (A and B, respectively) and SOL and MG at stimulus intensities of 50 %Mmax (C and D, respectively). Individual s.e.m. ranged from 0.007 to 2.85 mV for SOL and from 0.005 to 2.12 mV for MG.

H-reflex and M-wave responses at 50 %Hmax

As can be seen in Fig. 3A and B, M-wave amplitudes varied little across all velocities in both SOL and MG at stimulus intensities of 50 %Hmax. There was a small decrease in mean MG M-wave amplitude in the isometric condition that probably decreased the isometric H-reflex amplitude. However, greater stimulus intensity in this condition would have increased the isometric H-reflex amplitude, but would not have changed the overall shape of the curve, particularly in relation to lengthening actions.

In accordance with the Hmax: Mmax ratio, H-reflex amplitudes at a constant stimulus intensity (50 %Hmax) decreased markedly at all lengthening velocities, whereas muscle shortening resulted in facilitation of the H-reflex compared to the isometric condition. Furthermore, during both lengthening and shortening actions, SOL H-reflex amplitudes at the two fastest velocities (±15 and ±5 deg s−1) were both significantly less than at the slowest velocity (±2 deg s−1). Similarly, MG lengthening and shortening H-reflex amplitudes at ±15 deg s−1 were significantly less than those at ±2 deg s−1. The results of H-reflex amplitude modulation were confirmed by analysis of the area under the first negative phase of the SOL H-reflex. Furthermore, there was no significant main effect of angular velocity on the SOL H-reflex duration at 50 %Hmax, confirming that variations in the area under the first negative phase were in fact due to changes in reflex amplitude.

H-reflex and M-wave responses at 50 %Mmax

H-reflexes and M-waves for SOL and MG at 50 %Mmax are presented in Fig. 3C and D. It is evident that, even when the amplitude of the M-wave was sensitive to small changes in stimulus intensity (i.e. on the steepest section of the ascending limb of the M-wave recruitment curve), the M-wave amplitudes in SOL and MG were relatively consistent across all test velocities. This finding is consistent with that of Misiaszek et al. (1995b) who reported similar reflex depression across a range of stimulus intensities from motor threshold to Mmax. The consistency of the stimulus intensity was reflected by the small s.e.m. of the M-wave for individual subjects (range: 0.1-1.05 mV and 0.1-0.72 mV for SOL and MG, respectively). At these stimulus intensities, both SOL and MG H-reflexes exhibited velocity-dependent decreases in amplitude at all lengthening velocities. Furthermore, during shortening actions, the MG H-reflex amplitude at +2 deg s−1 was significantly greater than that for the isometric and two faster shortening velocities. Similarly, the SOL H-reflex amplitude at a shortening velocity of +2 deg s−1 was significantly greater than that at +15 deg s−1. Again, identical angular velocity-specific modulation was found in SOL H-reflex area at 50 %Mmax.

Latency of reflex modulation

H-reflex amplitudes at each latency for an individual subject whose data best reflect the group mean are presented together with the isometric H-reflex amplitude in Fig. 4. Mean H-reflex amplitudes during lengthening actions were significantly less than the isometric and corresponding shortening H-reflex amplitudes for both SOL and MG. However, during lengthening actions, mean H-reflex amplitudes at the shortest latency were significantly greater than for the two longer latencies (H-reflex amplitudes at -30, -15 and -7 deg s−1, i.e. 57, 90 and 165 ms, respectively, were 3.75 ± 0.25, 1.5 ± 0.25 and 1.5 ± 0.25 mV for SOL and 4.0 ± 0.5, 1.75 ± 0.25 and 1.75 ± 0.25 mV for MG, respectively).

Figure 4.

To determine whether H-reflex modulation during dynamic passive actions was central or peripheral in origin, H-reflexes were recorded at various latencies after the onset of movement. H-reflexes from SOL and MG for one subject recorded during 1 deg dorsi- and plantarflexion at angular velocities of 30, 15 and 7 deg s−1 are shown. Starting ankle angles were 91 deg for lengthening actions and 89 deg for shortening actions and stimuli were delivered after 1 deg ankle displacement (at 90 deg ankle angle). At the selected velocities, after 1 deg displacement, the latency of stimulation from the onset of movement (indicated by the dotted lines) was 42, 75 and 150 ms, respectively. Each recording shows a consistent stimulus artefact followed by a small M-wave and an H-reflex approximately 35 ms afterwards. For comparison, isometric H-reflexes are also displayed.

The difference in H-reflex amplitude between the fastest and two slowest lengthening velocities was probably influenced by the number of action potentials travelling within the Ia afferent in the different conditions. Figure 5 displays overlaid H-reflex responses from SOL together with SOL EMG recordings from five passive lengthening trials in which no stimulation was delivered. The greater EMG activity recorded during the fastest lengthening velocity most probably reflects an increase in the Ia input from primary muscle spindle afferents that would be expected as a consequence of rapid muscle lengthening. Increases in stretch-induced reflex responses at the faster velocities may account for the facilitated H-reflexes evident in some subjects. Although H-reflex responses at the shortest latency were significantly greater than at the two longer latencies, they were still significantly depressed compared with the isometric H-reflex. Therefore, it appears that, despite the summation of Ia afferent volleys, there is still a short latency inhibitory mechanism occurring within 60 ms.

Figure 5.

Examples of SOL H-reflexes (upper traces) recorded during lengthening (continuous traces) and shortening (dotted traces) actions at each angular velocity recorded from all four subjects. The middle five traces display SOL EMG recorded during passive lengthening actions in which no stimulation was delivered. Position and velocity recordings are also presented in the lower two traces. Dashed lines indicate the latency of stimulation (42, 75 and 150 ms for angular velocities of 30, 15 and 7 deg s−1, respectively) for the trials in which H-reflexes were recorded and the time at which the reflex response would appear in the EMG recordings (35 ms later). Evidence of stretch-induced reflex responses can be seen at the faster velocities and is probably responsible for the larger H-reflexes recorded at these velocities in some subjects.


Data from the present study have revealed that SOL and MG H-reflexes are both strongly depressed during lengthening actions and facilitated during shortening actions when compared with the isometric H-reflex. Furthermore, modulation of the H-reflex during passive movements occurred at a latency of less than 60 ms and was therefore largely attributed to peripheral/spinal mechanisms. The extremely consistent M-wave amplitudes in the present study confirm that the observed H-reflex modulation was not due to alterations in the spatial relationship between the nerve and stimulating electrode.

Methodological considerations

One of the main potential sources of error when recording reflex responses during dynamic muscle actions is the spatial relationship between the nerve and the stimulating electrode. A common technique in H-reflex studies is to record a small M-wave concurrently with the H-reflex, the amplitude of which is used as a calibration to ensure consistent effective stimulus intensity. However, as can be seen in the recruitment curves of Fig. 1, at low stimulus intensities the M-wave is insensitive to small changes in stimulation. Therefore, reflexes were recorded at a stimulus intensity of 50 %Mmax where the resulting M-wave lies on the steep ascending limb of the recruitment curve and is sensitive to small changes in stimulus intensity. The consistent M-wave amplitudes in this condition indicate that passive muscle actions did not alter the spatial relationship between the stimulating electrode and the tibial nerve. Based on these findings, it is unlikely that the effective stimulus intensity was altered during trials using 50 %Hmax stimulation.

It should be noted that, at stimulus intensities sufficient to elicit an M-wave, the H-reflex response cannot be considered to arise exclusively from Ia afferent input. Furthermore, at a stimulus intensity of 50 %Mmax the resulting H-reflex had either reached a plateau or lay on the descending limb of the recruitment curve and, consequently, would probably be subjected to considerable antidromic collision. Despite the possible contamination of reflex responses, the H-reflex was modulated in a similar manner at the two stimulation intensities. Therefore, it appears as though most of the reflex modulation could be attributed to alterations in the Ia afferent pathway projecting onto the SOL motoneurone pool.

During dynamic trials there is the potential that the recording electrodes may move relative to underlying muscle, thereby influencing the amplitude of the recorded reflex response. This was not considered a problem in the present study, however, as all reflex responses were recorded at the same ankle angle (90 deg) and, presumably, at the same muscle length.

Possible inhibitory mechanisms

Just as the reflex response to a given stimulus is influenced by the composition and mechanical characteristics of the homonymous muscle, a number of neural mechanisms are also known to influence motor output in response to sensory stimuli. Mechanisms involving presynaptic inhibition (either centrally or peripherally induced), homosynaptic post-activation depression, reciprocal inhibition from antagonist activation, Ib afferents, group II afferents and cutaneous mechanoreceptors, as well as plateau potentials, have all been implicated in modulating reflex responses.

The group II afferents from lengthening plantar flexors are considered to have little influence on the reflex modulation seen in the present study. Group II afferents are predominantly regarded as indicators of static length changes; therefore, the influence of muscle spindle discharge on reflex modulation was considered to arise from predominantly Ia afferents. Likewise, the influence of Ib afferents was not considered to be the likely cause of reflex modulation as, although Ib afferents are sensitive to very small changes in muscle tension (Houk & Henneman, 1967), they are more influential during active muscle actions than when the movement is performed passively (Burg et al. 1973).

It is possible that afferent responses from cutaneous mechanoreceptors and joint receptors may influence the resultant H-reflex amplitude. However, Hultborn et al. (1996) applied a tourniquet at the ankle, thereby inducing an ischaemic block of these sensory inputs without altering the reflex depression following passive dorsiflexion. When the tourniquet was applied just distal to the knee, reflex depression was abolished and the origin of reflex inhibition was attributed to the large diameter Ia afferents located between the ankle and knee. Based on these findings, cutaneous mechanoreceptors and joint receptors were not considered to be responsible for the reflex inhibition seen in the present study.

The fact that H-reflex amplitudes during passive lengthening actions were significantly less than the isometric H-reflex amplitude when the latency from movement onset to stimulation was at its shortest (57 ms) indicates that inhibition of the H-reflex amplitude can occur in the absence of supraspinal influences. However, reflex modulation at a latency of 57 ms was not consistent, with H-reflexes being facilitated in some subjects while inhibited in others (see Fig. 5). The unpredictable nature of this modulation would not appear to reflect a centrally controlled functional reflex modulation, at least at latencies as short as 57 ms, and as such, is most probably influenced by the variability of Ia afferent activity generated from the muscle spindles.

Although some subjects displayed considerably more stretch-induced EMG activity than others, the amplitude of stretch reflexes and H-reflexes varied substantially both between and within subjects. This variability may reflect differences in muscle spindle output or the susceptibility of the motoneurone pool to Ia input. Muscle spindle output is influenced by factors such as muscle stiffness, gamma drive as well as thixotropic affects on both intrafusal and extrafusal muscle fibres. Small plantar flexor efforts were performed before the start of each passive movement in an attempt to minimize any thixotropic influences on spindle discharge. However, differences in muscle stiffness and gamma drive were not evaluated in the present study and may account for some of the differences in muscle spindle output. As this study utilized passive movements and subjects were unaware of the direction of the following movement, the excitability of the motoneurone pool was not considered to be altered by higher motor centres. Differences in the susceptibility of motoneurones to Ia input cannot, however, be excluded as an explanation for the observed variations in stretch reflex responses.

Depending on the timing of transmission within the Ia afferents, muscle spindle discharge may either inhibit or facilitate H-reflex responses. Modulation of the H-reflex by submaximal stimulation of homononymous Ia afferents has been reported by Fukushima et al. (1982). A conditioning stimulus at an intensity subthreshold for the H-reflex produced short latency inhibition of the test H-reflex for conditioning-test intervals of 1.5-3.0 ms, which was attributed to the refractoriness of Ia afferent fibres. Subsequently, the H-reflex was facilitated up to 15 ms as a result of temporal summation of excitatory postsynaptic potentials (EPSPs) at the motoneurone pool. Therefore, depending on the temporal relationship between the discharge from muscle spindle afferents and delivery of the H-reflex stimulus, the H-reflex recorded at the shortest latency during passive lengthening actions may have been inhibited due to refractory Ia afferents or facilitated as a result of EPSP summation at the motoneurone pool.

Interestingly, at latencies longer than 57 ms, H-reflexes were strongly inhibited even though the muscle spindle discharge was not sufficient to elicit stretch reflexes. It is possible that despite the weaker strength of the spindle output, the longer time frame over which the muscle was lengthening resulted in a greater total release of neurotransmitter from the presynaptic terminal and, hence, a greater influence of homosynaptic post-activation depression. However, even a latency of 165 ms seems very short for neurotransmitter depletion and one must question the functional significance of such an inhibitory mechanism. Alternatively, when assessing reflex amplitudes at longer latencies the influence of higher motor centres cannot be ruled out. It is possible that the strong inhibition seen at these latencies was a result of centrally or peripherally induced presynaptic inhibition.

As all plantar flexor muscles were lengthened simultaneously, the influence of presynaptic inhibition arising from heteronymous Ia afferents cannot be ignored (Barnes & Pompeiano, 1970; Brooke et al. 1995a). Inhibition of the MG monosynaptic reflex induced by sinusoidal changes in the length of the lateral gastrocnemius- soleus (LGS) in cats was reported by Barnes & Pompeiano (1970). This reflex depression was attributed to primary afferent depolarization as indicated by dorsal root potentials recorded in MG Ia fibres. However, depression of the monosynaptic reflex only became evident once the vibration was removed. During vibration this inhibition was overshadowed by a strong facilitation of Ia afferents also induced by LGS vibration. Therefore, although presynaptic inhibition from heteronymous Ia afferents may have occurred during lengthening actions in the present study, it is possible for this inhibition to be overshadowed by temporal facilitation of EPSPs at the motoneurone pool.

Given the complexity of neural inputs projecting onto the motoneurone pool, it is difficult to definitively identify the neural mechanism responsible for reflex modulation. It is also likely that the resultant reflex amplitude reflects the integration of a number of influential neural mechanisms and no single mechanism can fully account for the observed reflex modulation. For example, based on the time course of reflex recovery, Voigt & Sinkjaer (1998) suggest that reflex inhibition during lengthening actions is the result of the quickly recovering presynaptic inhibition (≈500 ms) and the longer lasting homosynaptic post-activation depression (≈4 s). Furthermore, it is likely that the predominant mechanism responsible for reflex modulation varies depending on the demands of the task and caution must be taken when interpreting the reflex modulation evident during different functional tasks.

In the study of Schneider et al. (2000) H-reflex inhibition was closely associated with activation of the antagonist, tibialis anterior (TA), and the authors suggested that the reflex modulation was predominantly regulated by centrally produced reciprocal inhibition. Although TA activation is known to have a strong inhibitory influence on the SOL H-reflex (Crone, 1987), Yang & Whelan (1993) have shown that depression of the SOL H-reflex during gait can occur in the absence of TA activation. In the present study reciprocal inhibition was not considered a contributing factor to the H-reflex depression during passive lengthening actions as no activity from TA was detected in any trials. However, it is possible that TA Ia afferents were activated during passive plantar flexion movements, particularly at the faster velocities. Afferent activity from TA muscle spindles, although insufficient to elicit a stretch reflex response, would have an inhibitory influence on the plantar flexor motoneurones via Ia inhibitory interneurones. The potential for such reciprocal inhibition may explain why H-reflexes at the faster shortening velocities were significantly less than those at the slower shortening velocity (see Fig. 3), although it cannot account for the facilitation of the H-reflex during shortening actions when compared with the isometric H-reflex.

Facilitation of the shortening H-reflex may, however, be related to activity within homonymous Ia afferents. It is possible that during the passive shortening actions the muscle spindles fell silent, allowing recovery of neurotransmitters that may have been depleted during the previous lengthened position. A similar scenario was reported by Wood et al. (1996) who recorded muscle spindle discharge and H-reflexes during dynamic muscle actions in the cat. Once the muscle was shortened from the lengthened position there was a large recovery of the H-reflex amplitude in the absence of spindle discharge. This recovery does not explain the observed increase in reflex amplitude above the isometric level. However, it is possible that, in the present study, isometric trials were associated with a tonic firing of muscle spindles that may have elicited homosynaptic post-activation depression and consequently reduced the isometric H-reflex amplitude. Interestingly, during the short latency experiments, shortening H-reflexes were not significantly different from the isometric H-reflex. This may indicate that at these latencies, there was insufficient time for complete recovery following neurotransmitter depletion. The consistency of shortening H-reflexes at the three latencies argues against supraspinal facilitation of the shortening H-reflexes.

Functional implications

It was interesting to note in the present study that the SOL and MG muscles, which have distinctly different fibre-type compositions, showed similar patterns of H-reflex modulation. Moritani et al. (1990) demonstrated differences in modulation of reflex responses in SOL and MG during a functional hopping movement. However, as the movements performed in the present study were not considered ‘functional’ tasks, it might be considered less likely to find significant differences in H-reflex modulation between these two muscles. During functional movements, Ia afferents can discharge at very high frequencies and the resultant EPSP has been shown to depend on the properties of the motoneurone pool (Koerber & Mendell, 1991). Changes in EPSP amplitude have also been shown to vary depending on the pattern of stimulation, which reflects a task-dependent distribution of synaptic input to the motoneurone (Koerber & Mendell, 1991). These findings may account for some of the discrepancy between SOL H-reflex amplitude and the underlying activation during phases of the gait cycle.

Differences in reflex modulation between SOL and MG may also be related to the Ia afferent motoneurone connection. It is suggested that Ia afferents predominantly depolarize slow twitch, low threshold motoneurones (Messina & Cotrufo, 1976). The predominance of these motoneurones in SOL would explain the larger Hmax: Mmax ratio recorded from SOL than from MG in the present study. Despite their different magnitudes, the Hmax: Mmax ratios of SOL and MG were modulated in the same manner during passive lengthening and shortening actions. This suggests the H-reflex modulation was not related to muscle composition or motor unit recruitment order. However, it must also be noted that, during the 50 %Hmax condition, the relative stimulus intensities were different for both SOL and MG (see ‘Passive reflex modulation’ in Methods). As indicated by Crone et al. (1990), the susceptibility to facilitation or inhibition varies depending upon the size of the control reflex, with smaller reflexes being more susceptible to inhibitory influences than larger reflexes. Therefore, differences in H-reflex modulation in each muscle due to Ia afferent- motoneurone connections may be obscured by the effect of different stimulus intensities.

It is important to note that the movements performed in this study are not considered functional tasks. However, the H-reflex modulation observed in this study probably reflects changes in the afferent pathways that project onto the SOL motoneurone pool. Passive reflex modulation provides information regarding peripheral inputs during dynamic activities in the absence of strong input from higher motor centres. These data may assist in the interpretation of H-reflex modulation during more functional tasks such as walking, running and cycling.


This research was funded by the Swedish Medical Research Council. G.J.P. was funded by the New South Wales Sporting Injuries Committee.