Several electrical and magnetic stimulation techniques are used to evoke compound muscle action potentials to study neurophysiological mechanisms (e.g., H-reflexes, F-waves, and motor-evoked potentials from transcranial magnetic stimulation of the motor cortex). The amplitudes of these evoked responses are often used to estimate the excitability status of the motoneuron pool, as these techniques recruit motoneurons via various neural circuits. However, changes in the amplitude of these evoked responses can result from both central (e.g., at the cortical level) and/or peripheral physiological processes (i.e., within the muscle). To remove the possible influence of peripheral mechanisms when interpreting these evoked responses, the amplitude from evoked responses should be normalised to the amplitude of the maximal compound muscle action potential (M_max). The M_max is a short-latency evoked response where all muscle fibres are presumably recruited via peripheral electrical stimulation of motor axons. Let’s assume that we aim to estimate changes in corticospinal excitability by recording motor-evoked potentials (MEPs) before and after an intervention. Before the intervention, we record an MEP with amplitude A_pre and an M_max with amplitude B_pre. After the intervention, we record an MEP with amplitude A_post and an M_max with amplitude B_post. We then observe that both A_post and B_post are smaller than their respective pre-intervention values. However, when comparing normalised values (i.e., A_pre/B_pre vs A_post/B_post), we find no differences between pre- and post-intervention. This suggests that the reduction in MEP amplitude was driven by a decrease in peripheral excitability rather than a change in the central mechanisms we aimed to study. Had we not normalised MEPs to M_max, we might have mistakenly concluded that corticospinal excitability decreased following the intervention. As the M_max represents the maximal synchronised activation of motor units recorded by surface EMG electrodes, this normalisation method also allows researchers to estimate the proportion of a motor unit pool recruited by a stimulus (presented as a % M_max). For example, eliciting a soleus H-reflex amplitude that is 10% of the soleus M_max suggests that 10% of the soleus motoneurons are being recruited from the entire soleus motoneuron pool.

For comparison of evoked responses amplitude within-subject and within-session (without removing the electrodes) experimental designs. Although not ideal, non-normalised evoked response data may still be analysed/interpreted, as long as analysis and interpretation occur within-subject and within-session (with no changes to electrode location) experimental design. However, the amplitude of evoked responses should be ideally normalised to the amplitude of the maximal compound muscle action potential (M_max) where possible (see method Normalized to the maximal M-wave (M_max) amplitude).

Normalisation to the maximum voluntary contraction is not appropriate to interpret evoked responses, because maximal contractions do not involve synchronised discharge of the motor unit pool like what would be observed with reflexes or by means of external stimulation.

Normalisation to a standardised isometric maximum voluntary contraction is not appropriate to interpret evoked responses, because maximal contractions do not involve synchronised discharge of the motor unit pool like what would be observed with reflexes or by means of external stimulation.

Normalisation to a standardised submaximal task is not appropriate to interpret evoked responses, because submaximal contractions do not involve the synchronised discharge of motor units like what would be observed with reflexes or by means of external stimulation.

Normalisation to the peak or mean EMG amplitude throughout a voluntary contraction task is not appropriate to interpret evoked responses, because voluntary contractions do not involve synchronised discharge of many motor units like what would be observed with reflexes or by means of external stimulation.

Interpreting non-normalised evoked response amplitudes is not appropriate to compare the evoked response amplitude across different experimental conditions or sessions. To compare evoked responses across conditions or sessions, the amplitude of evoked responses should be normalised to the amplitude of the maximal compound muscle action potential (M_max) where possible (see method Normalized to the maximal M-wave (M_max) amplitude).