Slide Slam G10
Brain oscillations and microstates: a new way of analyzing ERP data. An example applied to a verbal Stroop task
Eric Ménétré1, Tomas Ros2, Marina Laganaro1; 1University of Geneva, Faculty of Psychology and educational sciences, NeuroPsychoLinguistic lab, 2University of Geneva, Center for Biomedical Imaging, Clinical and Translational Neuroimaging
Introduction: investigating neural activity in cognitive studies can be achieved by several different methods. Microstate analyses have been proven to be very helpful at disentangling the different brain network underlying a specific cognitive process and their time course (Murray et al., 2008) and have been applied in many studies on language. Microstates are obtained by segmenting the ERP signal in different periods of topographical stability. It is well established that a single topography can reflect a brain network composed of different structures, however the relationship between the topographies and the brain oscillations underlying these structures remained to be clarified. To investigate this link, a previous study performed a microstates analysis on EEG signal filtered in different narrow frequency bands (Férat et al., 2020). The results showed a good reliability of the microstates over the different frequency bands even though quantitative differences were observed. The present study aims at clarifying if the method is applicable to ERP signal and to estimate the reliability of the microstates in the different frequency bands. To obtain results comparable with a furnished literature, these analyses were performed on a verbal Stroop task. Method: 31 young healthy subjects performed a verbal Stroop task including 180 trials (60 congruent, 60 incongruent and 60 neutral trials) while undergoing a continuous EEG recording with 128 electrodes. To investigate the relationship between microstates and frequency bands, the raw signal was filtered to cover different frequency bands, namely: delta, theta, alpha, beta and gamma and the subjects’ average ERPs were segmented using a TAAHC algorithm for each of the frequency band as well as for the broadband signal. Since the verbal response generates an artifact, analyses were carried on two alignment points: locked to the stimulus and locked to the response onset (backwards). Results: The microstates analysis on the broadband signal suggests that four topographies best explain the signal. Regarding the segmentation performed on data filtered in narrow frequency bands, the higher the frequencies, the more different the segmentation gets from the broadband signal. Delta and theta bands show globally comparable microstates (among conditions and frequency bands) while the signals filtered in the alpha to gamma bands are each best explained by two topographic maps characterized by an inversion of polarity. Discussion: these encouraging preliminary results tend to show that microstates on broadband signal is only a simplified version of the results merging several generators firing at different frequencies. By filtering the signal in frequency bands, subtle changes in the microstates could be highlighted, opening new perspectives in the field of neuroimaging. Moreover, by appreciating the duration or global explained variance of the microstates present in each of the conditions and frequency bands, it might become possible to single out brain networks related to the interference processing from those related to word production.