Preclinical and EEG analysis links autism neurosubtypes to opposing electrophysiological profiles

Neurophysiological excitation versus inhibition (E:I) imbalance has long been theorized as one of the primary neurobiological explanations behind autism. However, progress applying this theory to most autistic individuals has been impeded by limitations in understanding of how non- invasive electrophysiological data (e.g., EEG) can be translationally used to pinpoint underlying E:I mechanisms. Using in-silico modeling alongside in-vivo animal validations, we show that a fractal component measured by the Hurst exponent (H) tracks with single neuron excitability, while {gamma} oscillations tracks with the ratio of E versus I synaptic conductances. Both H and {gamma} are complementary for predicting overall network excitability. In human EEG data we find two autism neurosubtypes characterized by opposing profiles of H and {gamma} and different large-scale brain- behavioral relationships affecting language, cognition, and other co-occurring neuropsychiatric issues. This work establishes that non-invasively measured electrophysiological signals can track with different aspects of E:I balance and that autism is composed of opposing E:I neurosubtypes.