Delayed, Reduced, and Redundant: Information Processing of Prediction Errors during Human Sleep.

Area of Science:

• Neuroscience
• Cognitive Science
• Computational Neuroscience

Background:

• The human brain maintains environmental responsiveness during sleep through 'sentinel processing mode'.
• Understanding neural information processing across sleep stages is crucial for comprehending consciousness fluctuations.

Purpose of the Study:

• To characterize auditory prediction error processing during wakefulness and sleep using advanced information-theoretic and electrophysiological methods.
• To investigate how neural codes change with varying levels of consciousness across sleep stages.

Main Methods:

• Employed mutual information (MI), co-information (co-I), event-related potentials (ERP), and temporal generalization analyses (TGA).
• Utilized an auditory 'local-global' oddball paradigm during wakefulness and polysomnography-monitored sleep.
• Analyzed mismatch responses to deviant auditory stimuli across sleep stages (N1-N3, REM).

Main Results:

• Auditory prediction error processing persists across all sleep stages, with increased ERP amplitudes in deeper NREM sleep.
• Significant reduction in encoded prediction error information observed in N3 and REM sleep.
• Neural information processing became more delayed and redundant with increasing sleep depth, with a shared neural code between N2 and N3 sleep.

Conclusions:

• The neural code of sentinel processing mode shifts from wakefulness to deep sleep, characterized by delayed processing and reduced information richness.
• Increasing sleep depth is associated with less informative and more redundant neural dynamics, correlating with waning consciousness.
• These findings offer insights into the neural mechanisms underlying the loss of consciousness during sleep.