Experimental Brain Research

, Volume 218, Issue 4, pp 601–607

The stability of Mmax and Hmax amplitude over time


    • Neuroscience Research AustraliaUniversity of New South Wales
  • Christine T. Shiner
    • Neuroscience Research AustraliaUniversity of New South Wales
  • Ganesha K. Thayaparan
    • Neuroscience Research AustraliaUniversity of New South Wales
  • David Burke
    • Institute of Clinical NeurosciencesThe University of Sydney and Royal Prince Alfred Hospital
Research Article

DOI: 10.1007/s00221-012-3053-4

Cite this article as:
McNulty, P.A., Shiner, C.T., Thayaparan, G.K. et al. Exp Brain Res (2012) 218: 601. doi:10.1007/s00221-012-3053-4


The stability of the maximal muscle response (Mmax) is critical to H reflex methodology. It has previously been reported that the amplitude of Mmax declines over time. If reproducible, this finding would have implications for all experimental studies that normalise the output of the motoneurone pool against the M wave. We investigated the effect of time on changes in Mmax and the maximal H reflex (Hmax) evoked at 4-s intervals over 60 min. To identify an influence of homosynaptic depression, we extended the interstimulus interval to 10 s and the time to 100 min. Two recording montages over soleus were used to ensure that interelectrode distance was not a critical factor. The soleus Mmax and H reflex were evoked by stimulation of the tibial nerve in the popliteal fossa in 7 subjects who sat with the knee flexed to 30° and the ankle plantar flexed by ~30°. We found no change in the pooled data for Mmax, Hmax, a reflex 50% of maximal, or the current required to produce it. However, one subject had a statistically significant increase in Mmax and a concurrent decrease in Hmax regardless of the interstimulus interval. On average, there was no change in the Hmax/Mmax ratio over time. While both Mmax and Hmax may change in response to many factors, these results suggest that, typically, time is not one of them.


Maximal muscle responseH reflexSpinal circuitry excitabilityHuman soleus muscleMotor control


The H reflex is often used as a tool to probe the excitability of spinal circuitry and its role in human motor control (for reviews see Pierrot-Deseilligny and Mazevet 2000; Pierrot-Deseilligny and Burke 2005). The amplitude of the H reflex is inherently variable both trial to trial (Funase et al. 1999) and in response to external influences including posture (Goulart et al. 2000), arousal (Brunia 1971; Bathien and Morin 1972), task (Schneider et al. 2000), and muscle length (Garland et al. 1994). The reflex also varies as a percentage of the maximal output of the motoneurone pool (Mmax) (Meinck 1980; Crone et al. 1990; McNulty et al. 2008) making comparisons between subjects difficult without normalisation. For this reason, the H reflex is typically investigated and expressed as a percentage of Mmax (Crone et al. 1990). This relationship provides a generally accepted measure of the excitability of the motoneurone pool (Angel and Hofmann 1963; Pierrot-Deseilligny and Burke 2005). Thus, the stability of Mmax is critically important to H reflex methodology and, indeed, to any measure that normalises the output of the motoneurone pool to Mmax.

Crone et al. (1999) reported that the amplitude of both Mmax and Hmax (the maximal H reflex) decreased over the time course of an experiment. Both techniques, Mmax and Hmax, are well validated and incorporated into routine procedures in both research and clinical applications. If this finding is reproducible, it raises a cautionary note for reflex studies lasting ≥45 min and for other techniques where data are normalised to Mmax, such as twitch interpolation studies of voluntary activation (e.g. Allen et al. 1995) and fatigue (Søgaard et al. 2006).

The day-to-day stability of the Mmax has been well established (Christie et al. 2005; Calder et al. 2005; Jaberzadeh et al. 2004). However, the amplitude of Mmax is known to change under certain conditions including post-tetanic potentiation (Bigland-Ritchie et al. 1979; Cupido et al. 1996; Hicks et al. 1989); the release of ischaemia (Cupido et al. 1996); changes in muscle length (Ismail and Ranatunga 1978), and the velocity of shortening (Katz 1939); after acute stretching (Weir et al. 2005); and during lengthening contractions (Katz 1939). The effect of fatigue on the amplitude of Mmax is less consistent; studies have reported both the presence (Stephens and Taylor 1972; Fuglevand et al. 1993) and absence (Bigland-Ritchie et al. 1986) of changes with fatigue.

An important methodological influence on the amplitude of Hmax is the rate of stimulation. With repetitive stimulation at frequencies below ~10 Hz, the amplitude of the reflex is reduced (Hoehler et al. 1981). This can also be measured as an increase in the current required to produce a reflex of predetermined amplitude (Lin et al. 2002; McNulty et al. 2008). Such changes are known as post-activation or homosynaptic depression (Eccles and Rall 1951).

In this study, we have extended the work of Crone and colleagues (Crone et al. 1990) examining the soleus Mmax and H reflex over 60–100 min with stimuli at 0.25 Hz and again at 0.1 Hz to minimise the potential effect of homosynaptic depression on reflex amplitude. We also compared changes in the output of the muscle Mmax and Hmax with changes in the input to the muscle required to evoke a reflex that was 50% of Hmax using threshold tracking protocols.



Data were collected in 2 series of experiments with different interstimulus intervals on 7 neurologically healthy subjects, 3 women and 4 men, mean age 27.6 years, range 21–40 years. A power analysis was undertaken using the published data of Crone et al. (1999) for the decrease in the maximal M wave in 17 subjects (mean decrease in amplitude 20.5%, SEM 2.2% [i.e. SD 9.07%]). A sample size of 7 would have a power of 98.1% to detect a significant change with a probability of 0.01 (and power of 99.8% for a probability of 0.05) in a two-tailed t test. The test side was randomly chosen, with 4 studies on the right leg and 6 on the left. Three subjects including one of the authors participated in both series of experiments. All subjects gave signed, informed consent to the studies that had been approved by the Human Research Ethics Committee, University of New South Wales.

The methods and subject positioning have been described in detail elsewhere (McNulty et al. 2008) and are outlined briefly below. Subjects sat with the test leg firmly fixed to an isometric foot plate with straps around the ankle and instep. To optimise H reflex responses, the knee was flexed ~30° from full extension and the ankle was plantar flexed ~30° from a right angle (Delwaide 1973; Hoehler et al. 1981) (see Fig. 1a). Subjects were seated in a comfortable chair in which they could relax fully to maintain a constant level of alertness. Distractions in the laboratory were minimised throughout the recording session. The duration of the experiment was 60 min (series 1) or 100 min (series 2, see below), adequate to detect changes of the extent described by Crone et al. (1999), in which, on average, the maximal decrease in amplitudes occurred at 44.2 min.
Fig. 1

Experimental set-up and protocol. a Experimental set-up. b Stimulus–response curve from a single experiment showing the steady increase in the stimulus current from below motor threshold to Mmax (upper trace). The lower trace shows the response curves for the M wave (filled circles) and the H reflex (open circles). This was used to determine the current intensity required to evoke a: Mmax, b: Hmax, and c: an H reflex that was 50% Hmax. This last intensity was used as a starting point for threshold tracking of the 50%Hmax reflex response. c and d Ten sets of stimuli were delivered on separate channels using a constant interstimulus interval, 4 s in c and 10 s in d. Sets began every 3 min for 20 sets (c) or every 10 min for 10 sets (d). Filled squares: supramaximal stimulus set to 120% of that required for Mmax; filled circles: Mmax; open circles: Hmax; patterned circles: current tracked to produce an H reflex that was 50% Hmax (d only)

Soleus EMG was recorded using disposable 10-mm Ag/AgCl electrodes in two simultaneous recordings. In the first configuration, electrodes were placed 40 mm apart, positioned in the midline over the soleus muscle immediately below the division of the two heads of the gastrocnemii. To investigate whether interelectrode distance could explain a change in Mmax, we also recorded soleus EMG using surface electrodes spaced ~200 mm apart, using the more proximal of the closely spaced electrodes and a distal electrode positioned over the lateral Achilles tendon at the level of the fibular malleolus. EMG was amplified (200–500×), filtered (1 Hz–1 kHz), digitised at 10 kHz, and recorded using the QTRAC program (©Prof Hugh Bostock, Institute of Neurology, London, UK) and a National Instruments data acquisition card (NI BNC-2110, National Instruments, USA).


A computer-controlled, isolated constant-current source (DS5, Digitimer, UK) under control of the QTRAC threshold tracking program was used to stimulate the tibial nerve in the popliteal fossa near the midline of the knee flexion crease. A 4-mm Ag/AgCl surface probe was used to deliver square-wave depolarising pulses 1 ms wide at a site optimised to produce a soleus H reflex. The anode was fixed to the patella ligament. The current was either set at a given stimulus intensity (fixed) or constantly adjusted to generate a test reflex of a predetermined amplitude, reflecting changes in central excitability of the soleus motoneurone pool (tracked).

A stimulus–response curve (Fig. 1b) was completed for each subject to define the current required to generate: Mmax; a muscle response that was half maximum (50% Mmax); Hmax; and an H reflex that was 50% of maximum (50% Hmax). The supramaximal stimulus intensity was set to be 120% of that required to generate Mmax. Three supramaximal stimuli were delivered before and after the test stimuli to ensure that any observed changes were not due to fatigue or movement of the electrodes. The experiment was conducted in two series, the first using an interstimulus interval of 4 s (0.25 Hz) and the second using 10 s (0.1 Hz).

In the first series of experiments (n = 6), a cycle of stimuli with equal 4-s (0.25 Hz) interstimulus intervals produced: (1) Mmax using a fixed stimulus and (2) Hmax using a fixed stimulus (Fig. 1c). Ten cycles were repeated every 3 min for 60 min. Stimuli to produce the Mmax and Hmax were delivered alternately, so that each was evoked 8 s apart. In the second series of experiments (n = 4), the interstimulus interval was increased to 10 s to minimise homosynaptic depression (McNulty et al. 2008). The stimuli were (1) Mmax using a fixed stimulus, (2) Hmax using a fixed stimulus, and (3) 50% Hmax for which the stimulus intensity was adjusted to track the threshold of the reflex depending on the amplitude of the previous 50% Hmax response (Fig. 1d). Thus, each specific stimulus recurred every 30 s. Ten cycles were repeated every 10 min for 100 min. Three subjects completed both protocols.

Data analysis

The amplitude of the M and H responses was measured peak to peak. The threshold for the 50% Hmax was measured as the current required to produce the reflex. To enable comparisons between subjects, the threshold for the H reflex was normalised to the current required to elicit an M wave 50% of Mmax, based on the stimulus–response curve. The responses during each set of stimuli for each response type were averaged with comparisons made between the first set of responses and the last for each experiment. The response amplitudes recorded using the two EMG electrode configurations (200 and 40 mm) were compared. Data were tested for normality with a Kolmogorov–Smirnov test, then analysed using paired t tests for normally distributed data and Wilcoxon signed rank test for non-parametric data, using SigmaStat v3.11 (Systat Software Inc, USA). Data are presented as mean ± SEM or median and interquartile range [IQR], and differences were considered significant when p < 0.05.


Individual responses

There were no significant differences in the amplitude of any response between the first set of stimuli and the last with the exception of one subject illustrated in Fig. 2a. For this subject, Mmax significantly increased in amplitude from 4.6 to 5.2 mV (p < 0.001) while Hmax concurrently decreased in amplitude from 2.4 to 1.5 mV (p < 0.001). This pattern is different to that reported by Crone et al. (1999) and was also recorded when the interstimulus interval was increased to 10 s in an experiment conducted 2 weeks later (Fig. 3a). When the threshold for a 50% Hmax reflex was tracked over 100 min in this subject, the difference in the current required to generate the target reflex was significant (p = 0.035), being lower in the final set compared to the first (decreasing from 58.2 to 49.8% of the current required to produce a 50% Mmax). However, there was no change in the amplitude of the response, that is, the required current was successfully tracked to produce a reflex of the predetermined amplitude. In another subject, a non-significant decrease in Mmax was observed while the amplitude of Hmax did not change (Fig. 2b), a pattern repeated when the interstimulus interval was increased to 10 s. In this subject, neither the current required nor the amplitude of the 50% Hmax varied. In another individual subject, we noticed a ‘sag’ in the amplitude of Mmax during the 4-s interstimulus interval series, a change that was not significant (Fig. 2d). In most studies, there was no difference for any measure between the first and the last step of responses at either the 4-s or the 10-s interstimulus interval (Figs. 2c, 3b).
Fig. 2

Mmax and Hmax over time. In each panel, each set of data points contains the peak response to 10 stimuli. The population analysis showed no significant change with time, but the single subject data illustrated in a did, the only one to do so. For this subject, Mmax increased over time (p < 0.001) and Hmax decreased with time (p < 0.001). The pattern of responses was similar during the longer protocol with a less marked increase in Mmax but a greater decrease in Hmax (Fig. 3a). b Mmax for another subject showed a slight, non-significant decrease in Mmax with no change in Hmax. c Five of the 7 subjects showed no change in either Mmax or Hmax over time. One such example is shown here. d The amplitude of Mmax did not change in these data when the first and last sets are compared. However, there was a small, non-significant but obvious sag in Mmax amplitude that was maximal in the 11th set. There was no change in Hmax

Fig. 3

10-s interstimulus intervals with threshold tracking of the H reflex (a) and Hmax/Mmax relationship for the first and last data sets. a When the interstimulus interval was extended to preclude any potential for post-activation depression, the response of the same subject as illustrated in Fig. 2a followed the same pattern. Neither the current required to evoke an H reflex that was 50% of Hmax nor the amplitude of the tracked 50% reflex changed over time for this subject. b There was no change over time in the amplitude of Mmax, Hmax, 50% Hmax, or the current required to generate it for all other subjects, one of who is illustrated here. c Pooled data showing stability of all parameters measured for both the short 4-s interstimulus interval over 60 min (filled bars) or the longer 10-s interstimulus interval over 100 min (empty bars). d The ratio of Hmax to Mmax did not change when the first and last data sets were compared for individual subjects. There was, however, a difference for the first subject in both the short and the long protocol (see Figs. 2a, 3a)

Pooled responses

When the data for all subjects in each series were pooled, there was no significant change in the amplitude of Mmax or Hmax when the mean of the first set of responses was compared to the last set, regardless of whether the interstimulus interval was 4 s or 10 s (Fig. 3c). Similarly, there was no change when the mean current required to generate an H reflex 50% of maximum was compared for the first and the last set of responses at the 10-s (0.1 Hz) interstimulus interval (Fig. 3c). Nor was there a change in the peak of the H reflex evoked in this manner. The Hmax/Mmax ratio did not change with time from the first to the last set of stimuli, at either interstimulus interval, or between intervals for 2 of the 3 subjects who completed both series (Fig. 3d). There was a significant difference in the ratio of Hmax/Mmax as a function of time for one subject between the two interstimulus intervals tested, with the ratio being 1.8 for the 4-s interstimulus interval and 3.7 for the 10-s interstimulus interval. However, for this subject, there was also a significant difference in the Hmax/Mmax ratio as tested during the stimulus–response curves at the beginning of each experiment. This is the same subject whose data are illustrated in Figs. 2a and 3a. The two different experiments were conducted 2 weeks apart for the 3 subjects tested. For all subjects, soleus EMG was measured on two separate channels with either a 40 mm or ~200 mm interelectrode distance. There was no effect of electrode spacing and no significant difference between the two recordings for any parameter or any subject.


In this study, we have investigated whether the maximal response of the muscle (Mmax) or the amplitude of the maximal H reflex (Hmax) changes as a function of time. We found that there was no change in the average amplitude of Hmax between the first and last set of responses measured over 60 min. Similarly, there was no change in Mmax as a function of time. These findings fail to confirm those of a previous report (Crone et al. 1999). When the elapsed time was increased to 100 min and the interstimulus interval to 10 s to detect whether homosynaptic depression may have influenced the reflex response, there was no change in either Mmax or Hmax. Furthermore, there was no effect of EMG interelectrode distance. We also found no change over 100 min in the current required to produce a test reflex 50% of maximum using a 10-s interstimulus interval. We cannot comment on whether changes might have occurred if the experiments were longer than 60–100 min. Significant changes in the H and M waves were found in one subject, but these were not the same as those described by Crone et al. (1999), in that the Mmax increased significantly, rather than decreased. In any case, there was no mean change in the population responses in either series of experiments, whether they lasted 60 min (series 1) or 100 min (series 2). The stability of the H and M responses supports our observations made in a previous H reflex study using a variety of interstimulus intervals, during voluntary contractions and when the reflex response was set to be 10% Mmax (McNulty et al. 2008).

The stability of Mmax is crucial for the interpretation and validation of the H reflex. Mmax not only allows an estimate of the proportion of the motoneurone pool recruited into the reflex, but serves to highlight changes in muscle geometry that might occur with movement and alter reflex amplitude (Pierrot-Deseilligny and Mazevet 2000). Normalising the H reflex against Mmax can be used to control for non-linearities in the input–output relationship of the motoneurone pool with the amplitude of the target reflex typically around 10–20% of Mmax (Crone et al. 1990). Threshold tracking has recently been used as a means of overcoming such ‘pool problems’. By tracking the current required to produce a test reflex of a predetermined proportion of Mmax, the population of α-motoneurones contributing to the reflex can be kept relatively constant (Lin et al. 2002; McNulty et al. 2008). The stability of Mmax is essential to this technique and was tested throughout the experiment to ensure that conditions, including the muscle position, had not changed.

Having been unable to reproduce the findings of Crone et al. (1999), we cannot propose adequate experimental reasons to explain a decrease in Mmax over time. It is conceivable that a decrease in the amplitude of Mmax might be the consequence of gradual displacement (or drying) of stimulating electrodes, producing a gradual decrease in activation of triceps surae muscles other than soleus, (i.e. one or both gastrocnemius muscles). To investigate this and the potential influence of electrode placement (Bromberg and Spiegelberg 1997; Nandedkar and Barkhaus 2007), we recorded EMG using two configurations, the first with our usual 40 mm interelectrode distance and the second with a broad ~200 mm spacing. There was no difference between the two recordings.

Repetitive stimuli at higher frequencies and voluntary activity are known to increase the amplitude of single muscle fibre potentials due to activity-induced hyperpolarisation of the muscle fibre membrane associated with increased Na+/K+ pump activity (Hicks et al. 1989; McComas et al. 1994; Cupido et al. 1996). Reductions in the compound muscle action potential amplitude may be due to a reduction in the availability of neurotransmitter following localised ischaemia (Cupido et al. 1996). Although the number of stimuli might have a crucial and cumulative effect on Mmax, changes with repetitive stimuli should produce an increase in the amplitude of Mmax, not a decrease (Hicks et al. 1989; McComas et al. 1994; Cupido et al. 1996). It is unlikely that the pattern of stimulation in the current study was sufficient to induce muscle fibre hyperpolarisation or ischaemia. It is conceivable that changes could result from variations in temperature although we did not formally document this, and Crone et al. (1999) do not report it. If temperature was associated with a decline in the amplitude of Mmax, ambient temperature would have to fall noticeably to cool the muscle sufficiently (Rutkove 2000). We note that any cooling should have had peripheral effects on both the H and M waves, such that the H/M ratio would probably not change. Alternatively, limb immobilisation for the duration of the experiment may have induced muscle cooling.

Changes in Hmax were noted both by Crone et al. (1999) and in the current study for a single subject. However as noted in the Introduction, the amplitude of the H reflex is inherently variable. The stability of Hmax at both 0.25 and 0.1 Hz in the present study argues against homosynaptic depression explaining the reduction seen by Crone et al. (1999). Moreover, we saw no change in the current required to produce an H reflex 50% of maximum, having chosen this reflex size to provide the maximal scope for the reflex to either increase or decrease in amplitude but still remain on the ascending limb of the reflex recruitment curve, (see McNulty et al. 2008).

Although Crone et al. (1999) chose to emphasise the decrease in Mmax over time, it is perhaps more important to note the variability of the responses evident in their results. The long interval between measurements could have resulted in minor differences in posture or stimulus stability, thereby altering the response. In the present study, we found 100 min was the limit of subjects’ ability to maintain a constant arousal level without distraction or falling asleep. The H reflex remains an accessible, non-invasive tool for estimating synaptic efficacy within the spinal cord, and it is important to express the amplitude of the reflex relative to the maximal output of the motoneurone pool (Mmax). Thus, the stability and reliability of this measure is fundamental to H reflex studies. Mmax may change in response to many factors, but the present study suggests that time, typically, is not one of them.


This study was supported by the National Health and Medical Research Council of Australia and the Office of Science and Medical Research, New South Wales.

Conflict of interest

The authors declare that they have no conflict of interest.

Copyright information

© Springer-Verlag 2012