Emergence of Brain Rhythms: Model Interpretation of EEG Data

TL;DR

This paper presents a neural network model that reproduces various EEG brain rhythms, providing insights into brain activity, disorders, and potential applications in deep learning for phase transition detection.

Contribution

The study introduces a first-principles network model that generates EEG-like rhythms and explores their relation to brain phenomena and disorders.

Findings

Reproduces alpha, beta, gamma rhythms in the model

Identifies complex phenomena underlying EEG rhythms

Suggests deep learning applications for brain phase transitions

Abstract

Electroencephalography (EEG) monitors ---by either intrusive or noninvasive electrodes--- time and frequency variations and spectral content of voltage fluctuations or waves, known as brain rhythms, which in some way uncover activity during both rest periods and specific events in which the subject is under stimulus. This is a useful tool to explore brain behavior, as it complements imaging techniques that have a poorer temporal resolution. We here approach the understanding of EEG data from first principles by studying a networked model of excitatory and inhibitory neurons which generates a variety of comparable waves. In fact, we thus reproduce $\alpha$, $\beta,$ $\gamma$ and other rhythms as observed by EEG, and identify the details of the respectively involved complex phenomena, including a precise relationship between an input and the collective response to it. It ensues the potentiality of our model to better understand actual mind mechanisms and its possible disorders, and we also describe kind of stochastic resonance phenomena which locate main qualitative changes of mental behavior in (e.g.) humans. We also discuss the plausible use of these findings to design deep learning algorithms to detect the occurence of phase transitions in the brain and to analyse its consequences.