Ever since the introduction of rTMS (Pascual-Leone et al. 1994), it has been well recognized that the response to rTMS is highly variable. This was firstly reported in a small number of subjects with conventional rTMS (Maeda et al. 2000). Subsequent studies in larger numbers of healthy subjects have confirmed that there is a considerable variability in response to any rTMS protocol (Table 1.1) (Müller-Dahlhaus et al. 2008; Hamada et al. 2013; López-Alonso et al. 2014; Wiethoff et al. 2014). In general, the probability of producing the “expected” response may be as low as 50 %, at least as measured by effects on the MEP based on the recent studies with relatively large number of subjects (Figs. 1.1c, d, and 1.2) (see also (Horvath et al. 2014)). A number of factors have been identified to explain this variability, such as age, gender, time of day, physical activity, prior history of synaptic activity, state of cortex, interneuron networks, or even genetics (Ridding and Ziemann 2010). However, none of them accounts for a large proportion of the variation which thus must be regarded as multifactorial. It may be possible to simplify the sources of variability into two groups: intrinsic and extrinsic. Intrinsic variability may relate to factors that are impossible to modify, such as age, gender and genetics. Extrinsic variability is potentially controllable and includes factors such as state of cortex, prior history of synaptic activity, time of day, physical activity, detection of the motor hotspot, the attention level of subjects in a long experiment, etc. For example, some evidence suggests target muscle activity prior to or during rTMS intervention affects response variability. It might be possible to minimize this by a short period of complete EMG silence in target muscle prior to delivering rTMS. However, it is difficult to define a true “rest” condition. Even though participants may maintain complete silence in a target muscle, this does not guarantee that this is true of the whole motor system. In fact, even in a target muscle at rest, motor threshold can be modified when subjects change the focus of their attention (Gandevia and Rothwell 1987). This implies that the resting condition may vary depending on unavoidable fluctuations of neuronal states including attention, and thus any measure related to rest (e.g. resting motor threshold or MEP at rest) may be ill defined. Finally, it should be remembered that variation in response to rTMS may be due to variation in the ability of the test stimulus to pick up the effects. This could reflect, for example, interindividual variability in interneuron networks involved in the MEP.

Although there are problems in using MEP measurements to detect effects of rTMS, the advantage is that they provide an objective and useful way to measure cortical excitability. Apart from MEP, EEG responses to TMS (transcranial evoked potential, TEP) are a second objective read out of TMS (Massimini et al. 2005; Premoli et al. 2014). The advantage of TEP is that it is available, in principle, to any area of the brain, in contrast to the MEP which can only obtained by TMS over the primary motor cortex. However, there are no studies of range of variation in TEP measures after rTMS in different individuals.

There is a good evidence that rTMS improves or facilitates the function of certain areas after brain damage or dysfunction. In fact, many clinical trials have reported favourable effects on symptoms in various neurological and psychiatric diseases, such as stroke, depression, Parkinson’s disease, pain, etc (Lefaucheur et al. 2014). However, the beneficial effects of rTMS are variable, and the results of these trials are inconsistent. The question is why does this happen?

As already mentioned, we know that the effects of rTMS on MEP excitability are highly variable, but it is not yet clear whether variability in MEPs translates directly into variability in behavioural effect. It is often tacitly accepted that this relationship exists since we select for therapy those protocols that have the “desired” effect on MEPs. However, it may be too simplistic assumption, and therefore, it is worthwhile to know whether the response to rTMS measured using MEPs predicts either (a) a person’s intrinsic ability to learn a certain task and/or (b) the effectiveness of an rTMS protocol to enhance a person’s performance in a task. For the first point, there is some evidence that MEP changes produced by rTMS do not correlate motor learning rate (Li Voti et al. 2011). The answer may be more positive for the second point. Kang et al (2011) found a negative correlation between rTMS effects on MEPs and the effects of the same rTMS protocol on motor learning (Kang et al. 2011). However, the number of subjects was small, and more information is required to answer the question with certainty. Finally, it may be important to note that the MEP only reflects activity in the large diameter axons of the pyramidal tract. These represent only about 2 % of the total tract. Thus, it is possible that at least some effects of rTMS on behaviour result from activity in other components of the tract or even activity in other tracts such as the rubulospinal, reticulospinal, cortico-cortical and cortico-subcortical pathways (Lemon 2008). In this context, it is interesting to note that MEP changes in the corticospinal system may not correlate with changes in other pathways. Thus, application of an inhibitory rTMS protocol (QPS) over left primary motor cortex (M1) reduced MEPs evoked from left M1, but did not change interhemispheric (cortico-cortical) inhibition from left to right M1 (Tsutsumi et al. 2014), suggesting that effects on cortico-cortical and corticospinal pathways differ. Future studies are required in order to predict the effects of rTMS in a clinical setting.