Examining the effective connectivity of the ventral occipito-temporal cortex during visual word processing with combined TMS-EEG

International audienceWritten language processing involves rapid extraction of different kinds of information, from low level visual features to high-level linguistic information. Neuroimaging studies have shown that a specific area, the left ventral occipito-temporal cortex (vOT), plays a crucial role in this complex activity. It is co-activated with several perceptive and linguistic-related areas during reading tasks. It remains unclear, however, if such correlations of neuronal activity reflect effective connectivity between vOT and the co-activated areas. Here, we used a combination of TMS and EEG to firstly investigate the temporal dynamics of EEG activity during word-reading and secondly to examine the effective connectivity between vOT and other areas recruited by the reading network at different functional states: states being modulated by the characteristic of a stimulus (colored dots vs. words) and the task performed by the participants (visual vs. linguistic). The rational is that stimulation of the neurons in the vOT should induce a current spread of neural activity toward anatomically connected areas, modulated by the stimuli and the tasks.Four Go/NoGo tasks (visual task on dots, visual, phonological and semantic tasks on words) were performed during EEG recording. In half of the trials, a TMS pulse was applied at the scalp position corresponding to the left vOT, 100ms after stimulus onset (based on a previous TMS study). vOT was identified in individual anatomical MRI of each subject using functional data (i.e. word-reading contrast in an fMRI experiment). Analyses of the stimulus-evoked response (only “NoTMS” trials) revealed a pronounced negative peak in the 150-200ms time-window at bilateral occipital electrodes during the three “word” conditions, which was largely reduced in the “color” condition. The amplitude of this N170 was not modulated by the task being performed (visual, phonologic or semantic), suggesting a mainly bottom-up response. Specificity to the linguistic tasks (vs. visual) seemed to emerge at right frontal-lateral electrodes in a later time-window (200-300ms), previously associated with pre-semantic orthographic or phonologic processing.In a second set of analyses, TMS-evoked potentials were computed for each subject and task by subtracting NoTMS responses from the TMS condition. An initial modulation of TMS-EPs, related to stimulus type (words vs. colored-dots), was found in the 200-250ms time-window at right lateral-posterior electrodes (i.e. inferior parietal region). A second effect on TMS-EPs, related to the task being performed (visual vs. linguistic), was found in the 350-400ms interval at central-posterior electrodes.In summary, we observed that TMS stimulation of the vOT induced currents that varied, firstly, in the parietal cortex of the controlateral hemisphere as a function of the stimulus being processed, and ,slightly later, in a bilateral posterior region as a function of the information being monitored by the subject (i.e. visual or linguistic). While further analyses of the data should help us interpret the observed task- vs. stimulus dependent TMS effects (e.g. current sources densities analyses), our results so far demonstrate that TMS can be used to probe the networks engaged during reading tasks and to show task-dependent changes in connectivity