In the present proof-of-concept study, we raised the general question whether the faciliatory influence of paired-pulse TMS on intracortical excitability in the human motor cortex can be applied to classical conditioning. To this end, we utilized faciliatory paired-pulse TMS with default parameters18. The first TMS pulse was applied at subthreshold intensity of 95% resting motor threshold15, which paves the way for the second pulse applied at suprathreshold intensity of 130%. While a short interval between both pulses ( < ~ 6ms) leads to an inhibition of the MEP, a longer interval, such as the 12ms interval we used, provokes a facilitated MEP amplitude17. The amount of amplitude amplification after faciliatory paired-pulse TMS is an indirect marker of intracortical facilitation18. For conditioning, we paired the paving TMS effect of the first TMS pulse with one out of two auditory stimuli. In the final test phase, 20 min after conditioning, we used single instead of paired TMS pulses, delivered at the same suprathreshold intensity as the second pulse during conditioning (i.e., 130%). Half of those TMS pulses were paired with the tone applied during conditioning, the other half with the control tone.
Comparing MEPs provoked by single-pulse TMS paired with the conditioned tone to MEPs provoked by the same single-pulse TMS paired with the control tone we observed significantly enhanced amplitudes, suggesting that TMS-induced intracortical facilitation can be classically conditioned to a paired auditory stimulus. The effect of conditioning across all participants, however, was weak (p = 0.03 for signed paired t-test and p = 0.05 for the unsigned t-test, 55% responder). This weak effectiveness of conditioning appeared to be related to the variability of MEP amplitudes after paired-pulse TMS, as suggested by the inverse correlation between paired-pulse MEP amplitudes during conditioning and the effect of conditioning in the test phase. This is in line with the finding that the comparison of MEP amplitudes paired with the conditioned tone vs. control tone in those individuals with a median paired-pulse MEP amplitude below 2 mV in the conditioning phase revealed stronger effects (p = 0.01) than for the whole group (p = 0.03). The responder rate also increased from 55% for the whole group to 61%, and even further to 79% (p = 0.005), if we considered only those individuals with a median paired-pulse MEP amplitude below 1 mV. We found these increasingly stronger conditioning effects for smaller paired-pulse MEP amplitudes although the sample size, and hence statistical power, decreased from n = 75 to n = 38 in the “< 2 mV group” and to even n = 14 in the “< 1 mV group”. Together these findings suggest that the effect of conditioning depends on the amplitude of the paired-pulse TMS response during conditioning.
Designed as a proof-of-concept study, we here focused on the conditioning effect of paired-pulse TMS, but we did not assess the individual amount of intracortical facilitation induced by paired-pulse TMS. This would have required to add interleaved single-pulse TMS applications (i.e., applied at the same 130% MT as the second stimulus in the paired-pulse paradigm) to the conditioning phase und to compare the MEP responses due to paired-pulse TMS to MEP responses due to interleaved single-pulse TMS. It therefore remains unclear whether high paired-pulse MEP amplitudes during conditioning correspond to very weak or very strong facilitatory effects of paired-pulse TMS.
In the former case (i.e., very weak facilitatory effect), MEP responses to interleaved single-pulse TMS would have been in a similar range as the MEP amplitudes due to paired-pulse TMS. In this scenario, transsynaptic activation of most pyramidal cells in the motor cortex by the 1st conditioning TMS stimulus alone might have prevented a relevant facilitatory effect. Such a ceiling effect is well known for intracortical facilitation, and has been described by several authors15, 17.
A very strong facilitatory effect, as an alternative explanation for weak conditioning effects, could instead rely on an inverse interaction between the brainstem and the motor cortex. The combination of auditory stimuli, activating the startle system, and TMS, applied to the motor cortex, is a well-established paradigm to investigate interactions between the brainstem motor system and the cerebral cortex. An auditory stimulus preceding motor cortex TMS by 30 to 60 ms was shown to inhibit TMS-induced MEPs in upper limb muscles19, 20. The anatomical origin of this interaction seems to depend on intracortical projections within the motor cortex21, and inhibitory projections between the caudal brainstem reticular formation and motor cortex19, 20. If paired-pulse TMS induces very strong facilitation of the pyramidal cells in the motor cortex, it may, at the same time, weaken the influence of auditory inputs in activating cross-modal connections required to link the auditory input to the motor output. Studies, combining TMS conditioning with brain mapping techniques such as EEG, MEG or fMRI, are required to deepen our understanding of brainstem-cortex interactions involved in auditory-TMS conditioning.
TMS at suprathreshold intensities provokes clear, partly even uncomfortable pulsations at the skull that remain neither unconscious, nor unattended. To minimize the confounding influence of attention in our study, we jittered the intervals between single TMS applications to prevent prediction of their onsets. We also tested for a decline of paired-pulse MEP amplitudes over the conditioning phase and indeed found that paired-pulse MEP amplitudes declined continuously and significantly over trials (see Fig. 2). This decline may reflect a decline in attention, or, alternatively, simple habituation to the paired-pulse TMS applications22. Irrespectively of their origin, we found large individual variability in MEP amplitudes over the conditioning phase. That is why we applied the median instead of the mean as a representative measure for each participant. To control for habitual influences in future studies, a TMS habituation session should precede the TMS conditioning phase. To control for attentional influences, attention should be shifted to external cues (e.g., counting specific items in a movie) or internal cues (e.g., directly counting TMS applications). The latter method may even amplify the effect of conditioning since attention is shifted towards the intervention that causes conditioning, whereas in the former approach attention may be shifted away from conditioning.
In two previous TMS conditioning studies13, 14, auditory-visual stimuli were paired with single TMS pulses over the motor cortex to condition TMS provoked MEPs. Without TMS, Luber et al. found that the auditory-visual stimuli alone failed to provoke any MEP responses even with repeated conditioning sessions over days14. Johnson et al., however, found such MEP responses provoked by auditory-visual stimuli alone, which, however, were much smaller than TMS-induced MEPs and identifiable only in a percentage of tests and participants13. TMS preceded by the auditory-visual stimuli, nevertheless, provoked attenuated MEP responses after conditioning in both studies, suggesting conditioned cortical excitability which paves the way for a modified TMS response. Our findings generally support these previous findings, however, due to the paired-pulse TMS protocol, we can assign the conditioning effect to a distinct neurophysiological mechanism, namely the facilitation of the excitability of cortico-cortically projecting pyramidal cells17.
There are many open questions, such as, whether the conditioning effect depends on the individual level of paired-pulse TMS induced intracortical facilitation and how long the conditioned effect lasts before extinction. These questions, as well as the effects of repeated conditioning and pharmacological interventions must be explored in future studies. Whether such conditioning effects mimic the recently described beneficial TMS effects in distinct patient groups at all1–4, remains another challenging question for future research.
Pavlov’s pioneering work inspired decades of conditioning research7–12. More recent findings even suggest that conscious awareness of the neutral stimulus is not required for successful conditioning10. Present findings extend the scope of classical conditioning to artificially induced facilitation of intracortical excitability through paired-pulse TMS. These findings may open-up the avenue for further research on conditioning effects of non-invasive brain stimulation of the human brain.