The Rest-Task Relationship of Intrinsic Neural Timescales

Abstract

The human brain processes information across multiple temporal scales simultaneously. Understanding how the brain integrate information across diverse timescales is a recent problem in neuroscience. In order to address this problem, the concept of intrinsic neural timescales (INTs) was developed and defined as the time window during which prior information can influence the processing of new stimuli. The aim of our thesis is to investigate the rest-task relationship of INTs. We address three specific aims: First, we examine whether INTs are modulated by continuous behavioral tasks. We analyze wide-field calcium imaging in mice during spontaneous locomotion and EEG data from humans performing self-evaluation tasks. We demonstrate that behavioral states can be accurately classified from INT topographies, and active states show prolonged INTs compared to rest. Computational modeling reveals that these changes can depend on the strength of recurrent neural connections. Second, we investigate how resting-state INTs relate to event-related brain activity in discontinuous tasks. To address this question, we use magnetoencephalogram (MEG) data of an emotional face recognition paradigm and resting state scan obtained from humans. Drawing on the fluctuation-dissipation theorem from statistical physics, we predict and confirm a positive correlation between resting-state INTs and the magnitude of event-related fields. Neurobiological modeling using the Jansen-Rit framework suggests that the empirically observed relationship may be mediated by intracolumnar connection strengths. Third, we present IntrinsicTimescales.jl, an open-source Julia software package implementing established and novel methods for INT estimation and is designed to handle the complexity of modern neuroimaging datasets with high performance and accuracy. Our findings provide converging evidence that resting-state and task-state brain dynamics are fundamentally interconnected, with INTs serving as a bridge between spontaneous neural fluctuations and task-evoked responses. This work advances our understanding of how the brain's intrinsic dynamics relate to the information processing across behavioral states.