MEP

amplitude at time T after cTBS was defined as the averaged peak-to-peak amplitude of the MEPs recorded during the corresponding batch; this value was then expressed as the change in MEPs compared with pre-cTBS, i.e. [MEPs(T) – MEPs(pre-cTBS)]/MEPs(pre-cTBS). Thus, negative values reflect suppression after cTBS. Student’s t-tests were run to determine if MEP amplitudes were significantly different from zero after cTBS. Bonferroni was applied to correct for multiple comparisons. To account for the variance of the baseline, Student’s t-tests were also run on raw, non-normalized, data. Electroencephalography E7080 in vivo data recorded during batches of single-pulse TMS (Fig. 1C) were processed offline using the EEGlab toolbox (Delorme & Makeig, 2004) running in a MATLAB environment (Mathworks). The EEG signals were analysed with the common reference, as recorded. They were first high-pass filtered above 1 Hz. Continuous data were epoched from 200 ms before the TMS pulse to 600 ms after. Baseline correction was applied based on a pre-TMS interval of 200 ms. Disconnected channels were removed and

recomputed (spherical interpolation) after cleaning (see below). Independent component analysis was performed to separate residual electrical from physiological responses to a TMS pulse. Components related to electrical artifacts were identified by their activity strongly peaking at the vicinity of the stimulation sites during the first tens of milliseconds after a pulse, and by their spectrum covering a restricted frequency range with strong harmonics. Components this website clearly reflecting other artifacts, such as muscle contamination or eye blinks, were also removed. On average, 9.6 ± 4.1 (range 3–17) components were removed, most of the artifacts being identified in the first few components. We cannot exclude that true brain response to TMS was also partly removed with components identified as artifacts. However, as the same components were removed for all conditions within a subject, we

expect changes in EEG response to TMS after cTBS to be related to cTBS-induced changes in brain excitability. Grand-average of TMS-induced for EEG responses were then calculated for the group. For pre-cTBS and for each time batch after cTBS, we calculated the grand-average time-domain response at the C3 electrode (over M1). For each of the pre-cTBS and post-cTBS conditions, we identified the amplitude of four TMS-evoked potentials (TEPs) that are commonly reported in the literature (Paus et al., 2001; Komssi et al., 2004; Bonato et al., 2006; Komssi & Kahkonen, 2006; Van Der Werf & Paus, 2006; Fitzgerald, 2010), i.e. P30, N45, P55 and N100. Then, changes in amplitude compared with pre-cTBS were calculated for each TEP as [TEP(T) – TEP(pre-cTBS)]/TEP(pre-cTBS).