Neural integration and segregation revealed by a joint time-vertex connectome spectral analysis

Abstract

Brain oscillations are produced by the coordinated activity of large groups of neurons and different rhythms are thought to reflect different modes of information processing. These modes, in turn, are known to occur at different spatial scales. Nevertheless, how these rhythms support different modes of information processing at the brain scale is not yet fully understood. Here we present "Joint Time-Vertex Connectome Spectral Analysis", a framework for characterizing the spectral content of brain activity both in time (temporal frequencies) and in space (spatial connectome harmonics). This method allows us to estimate the contribution of integration (global communication) and segregation (functional specialization) mechanisms at different temporal frequency bands in source-reconstructed M/EEG signals, thus providing a better understanding of the complex interplay between different information processing modes. We validated our method on two different datasets, an auditory steady-state response (ASSR) and a visual grating task. Our results suggest that different information processing mechanisms are carried out at different frequency channels: while integration seems to be a specific mechanism occurring at low temporal frequencies (alpha and theta), segregation is only observed at higher temporal frequencies (high and low gamma). Crucially, the estimated contribution of the integration and segregation mechanisms predicts performance in a behavioral task, demonstrating the neurophysiological relevance of this new framework.