The peak-to-peak amplitude of the H reflex and M wave was computed from the raw EMG signal. The amplitude of the three responses recorded for each current intensity was averaged before being normalised to the M_MAX. The H-reflex RC [up to H_MAX(ascending part of the RC)] and M-wave RC were fitted with a Boltzmann sigmoid function (Klimstra & Zehr, 2008; Penzer et al., 2015) to determine relevant model-based variables (Figure 1): 1/ H_MAX and M_MAX; 2/ the intensities evoking a response of half H_MAX (I_H50) and M_MAX (I_M50); 3/ H_M5%; and 4/ M_Hmax (Lagerquist & Collins, 2008; Scaglioni et al., 2003).

the latency was measured from the moment of the stimulation to the first peak of the respective response. The latency was then expressed relative to the length of the arm (ms.cm^-1) as the latency of the H reflex in the FCR has been shown to be a function of the length of the arm of the participants (Burke, 2016; Khosrawi et al., 2015; Schimsheimer et al., 1985). Even though this variable may present some limitations (exact location of the recording and stimulating electrodes), such a normalisation can be used to compare the response latency between young and old adults (Scaglioni et al., 2003). The arm length was measured as the distance between the acromion and the styloid process of the radius.

We also estimated the SDTC and response threshold (reflecting the rheobase) for the H reflex and M wave using Weiss’ law (Kiernan & Kaji, 2013; Lin et al., 2002; Mogyoros et al., 1996). The I_H50and I_M50 were converted into threshold charges (current intensity × duration, expressed in mA.ms) and plotted against the corresponding duration (0.05 ms, 0.2 ms, and 1 ms) (Lin et al., 2002). The SDTC was then determined by the negative intercept on the duration axis, whereas the response threshold was given by the slope of the regression line (Figure 1E). Although this method is commonly performed with more than three pulse durations, using three durations should not have influenced our results as no difference in the SDTC calculated from two, five, 12 or 50 data points was observed (Mogyoros et al., 1996).

An example of the fitted RC for one young and one old adult is illustrated in Figure 2. The ability to fit the data with a Boltzmann function for each pulse duration significantly differed between age groups [chi² (3, 40) = 8.4, p = 0.037]. We could elicit H reflexes with a pulse duration of 1 ms in 37 participants (19 young and 18 old adults). Among these 37 participants, data from 1 young and 6 old adults had to be removed due to H-reflex amplitude being too small (<3% M_max) or inconsistent to be fitted by the Boltzmann function for briefer pulse duration. Therefore, the RC of only 18 young and 12 old adults has been retained for analysis. To have similar sample sizes for statistical analysis, we matched the 12 old participants [height: 1.67 (0.08) m; weight: 65.6 (8.8) kg, BMI: 23.7 (2.6)] with 12 young adults [height: 1.72 (0.10) m; weight: 66.6 (7.3) kg; BMI: 22.5 (2.1)] based on gender (8 women and 4 men), BMI and the kind of physical activity regularly practiced at the time of their participation (endurance, strength or mobility activities). Matching participants based on BMI contributed to controlling for large differences in body composition that could influence EMG recordings (Petrofsky, 2008) or induce long-term neurophysiological adaptations to excessive body weight (Maffiuletti et al., 2021). As a sedentary lifestyle can speed up the effects of ageing (Shadyab et al., 2017), physical activity was also considered as a matching factor. In addition, the kind of physical activity practiced by participants was considered to limit possible activity-dependent influences on the H reflex and M wave (Tøien et al., 2023).