Concurrent multimodal estimation of cortical excitation and inhibition shows that EEG-based estimates are linked to the ratio of glutamate to GABA

Neural excitation/inhibition (E/I) balance is dynamically regulated on multiple timescales. Adaptive changes in E/I balance changes can support healthy development, learning, and cognition, while disordered E/I balance is implicated in neurodevelopmental disorders, neurodegenerative disorders, and states of impaired vigilance. There has been growing interest in inferring E/I balance from efficient and noninvasive measurements such as electroencephalography (EEG), and several algorithms have been proposed to estimate E/I balance from EEG recordings. Despite promising results, there has been a lack of validation studies testing the underlying neurochemical changes leading to increased or decreased EEG-based E/I. Here we assess E/I balance with concurrent EEG and magnetic resonance spectroscopy (MRS) in humans. We estimate a standard measure of E/I balance, glutamate concentration divided by GABA concentration, and assess the correspondence between each candidate EEG-based E/I algorithm to this MRS-based estimate. Due to the methodological interest in both between-subjects research (e.g., comparing disordered to healthy E/I) as well as within-subjects research (e.g., comparing pre-intervention to post-intervention E/I), we quantify the associations between EEG-based E/I and MRS-based E/I separately for between-subjects and within-subjects comparisons. We find that each EEG-based E/I algorithm shows reliable and positive associations with MRS-based E/I. While these associations are evident for between-subjects comparisons, they are quite weak for within-subjects comparisons. Candidate EEG-based E/I algorithms are thus likely to be reflecting, at least in part, relative concentrations of cortical glutamate and GABA, although poor signal quality appears to limit these methods when attempting to increase temporal precision and identify within-subjects variations.