Spatial-Temporal Analysis of Neural Desynchronization in Sleeplike States Reveals Critical Dynamics*

Sleep is traditionally divided into two global brain states: non-rapid eye movement (non-REM) sleep, characterized by synchronized neuronal activity, and rapid eye movement (REM) sleep, marked by desynchronized, wake-like dynamics. With some exceptions, these two states were long considered mutually exclusive global brain states -- the brain periodically switched between one and the other during sleep. While recent research has begun to suggest otherwise, the technological trade-offs between high spatial and high temporal resolution neurophysiological recordings has made studying the spatial properties of REM challenging. However, advances in wide-field voltage-sensitive dye imaging in mice now offer both high spatial and temporal resolution, allowing us to study the spatio-temporal properties of REM and non-REM. 

Using a urethane-anesthetized mouse model of sleep, we apply time-frequency analysis to identify how REM- and non-REM-like activity patterns evolve over space and time on the cortical surface. We find that if spatial information is ignored (e.g., using single-point probes as is traditionally used to study REM) we reproduce the conventional picture of global REM/non-REM transitions. However, when spatial information is incorporated, we observe transient REM occurs heterogeneously across the cortical surface with pockets of desychronized activity in a background of synchronized neural activity. These desynchronized REM clusters are distributed in a scale-free manner across the cortex, consistent with critical dynamics and suggestive of a phase transition at the boundary between synchronization and desynchronization. Understanding these spatio-temporal patterns not only offers new insight into brain-state transitions but may also shed light on disorders such as Alzheimer’s disease, which disrupt sleep architecture (e.g., reduced REM time) and quality of sleep for patients.