A week in the life of the human brain: stable states punctuated by chaotic-like transitions

Many important neurocognitive states, such as our natural behavior or our level of fatigue, shift over minutes-to-hours in the real-world. To understand how neural states form and change on these timescales outside of experimental conditions, we analyzed 3-12 days of continuous intracranial recordings in twenty human participants that were confined to the hospital but freely socialized, read, used digital devices, slept, etc. while under simultaneous neural and video monitoring. We found that brain networks formed stable states that predicted both behavior and physiology. Changes in behavior corresponded to bursts of rapid neural fluctuations where the brain chaotically explored many configurations before settling into new states. Using deep neural networks and Koopman operators, we found that these dynamics divided into a slow, quasiperiodic cycle that tracked different stages of consciousness, primarily through the default mode network, and rapid, chaotic fluctuations off this cycle that tracked behavior.

Pre-print at: https://www.researchsquare.com/article/rs-2752903/v1

Portion of work profiled in MIT Technology Review: https://www.technologyreview.com/2023/02/07/1067951/brains-week-order-chaos/