BEYOND MACRO-STATES: CHARACTERIZING MICRO-STATE DYNAMICS OF NEURONAL ENSEMBLES IN THE PRIMARY VISUAL CORTEX OF MICE ACROSS THE WAKE-SLEEP CYCLE

Traditional characterizations of brain states across the wake-sleep cycle rely on electrophysiological indicators, most often evinced from local field potentials (LFP) and EEG recordings, which are used to classify neural activity into discrete, temporally homogeneous stages. However, these aggregate signals may obscure the rich, high-dimensional dynamics occurring at the level of neuronal ensembles. In this study, we investigated whether the architecture of neuronal ensembles adheres to canonical wake-sleep boundaries or is instead governed by fine-grained arousal dynamics.
Using large-field-of-view two-photon calcium imaging in primary visual cortex of head-fixed mice, we recorded population activity at single-cell resolution throughout the wake-sleep cycle, while monitoring pupil size and orofacial motion. We applied a Hidden Markov Model (HMM) on neuronal population activity to extract latent micro-states, independent of traditional state labels or arousal markers.
Our results reveal that the architecture of neuronal ensembles structures is organized along latent micro-states that do not align with traditional macro-scale wake-sleep classifications. Instead, we found that the activity profiles and transition dynamics of these neural micro-states were better explained by considering a continuous spectrum of movement and arousal factors. These findings suggest that neuronal ensembles, representing a fundamental organizational unit of brain dynamics, do not reflect behavioral macro-states, but rather a micro-state trajectory shaped by arousal and motion. This shift in perspective provides a novel, more granular understanding of which factors determine microcircuit-levels patterns of neural activity, underlying spontaneous fluctuations in brain function and responses.