Spindle oscillations are commonly observed during stage two of non-REM sleep. During sleep spindles, the cerebral cortex and thalamus interact through feedback connections. Both initiation and termination of spindle oscillations are thought to originate in the thalamus, based on thalamic recordings and computational models, although some in vivo results suggest otherwise. Here, we have used computer modeling and in vivo multisite recordings from the cortex and the thalamus in cats to examine the involvement of the cortex in spindle oscillations. We found that although the propagation of spindles depended on synaptic interaction within the thalamus, the initiation and termination of spindle sequences critically involved corticothalamic influences.

Humans are less responsive to the surrounding environment during sleep. However, the extent to which the human brain responds to external stimuli during sleep is uncertain. We used simultaneous EEG and functional MRI to characterize brain responses to tones during wakefulness and non-rapid eye movement (NREM) sleep. Sounds during wakefulness elicited responses in the thalamus and primary auditory cortex. These responses persisted in NREM sleep, except throughout spindles, during which they became less consistent. When sounds induced a K complex, activity in the auditory cortex was enhanced and responses in distant frontal areas were elicited, similar to the stereotypical pattern associated with slow oscillations. These data show that sound processing during NREM sleep is constrained by fundamental brain oscillatory modes (slow oscillations and spindles), which result in a complex interplay between spontaneous and induced brain activity. The distortion of sensory information at the thalamic level, especially during spindles, functionally isolates the cortex from the environment and might provide unique conditions favorable for off-line memory processing. I t is commonly believed that during non-rapid eye movement (NREM) sleep, the brain is isolated from the environment, based on an increase in the sensory threshold observed in sleeping individuals, and indirectly from the inability to remember the content of external stimuli delivered during sleep (1). However, a number of studies, ranging from single unit recordings in animals (2, 3) to human EEG (4) and neuroimaging (5) experiments, demonstrate the persistence of brain responses to sensory stimulations during NREM sleep, suggesting that the brain can still process external stimuli during NREM sleep.In fact, the transmission of sensory information during NREM sleep is thought to be selectively reduced during sleep spindles (6, 7). Sleep spindles-a hallmark of light NREM sleep [mostly stage 2 (S2) sleep]-are characterized in humans by waxing and waning 11-to 15-Hz oscillations lasting 0.5-3 s (8). They are generated by the thalamus, which acts as a pacemaker (9, 10), and result from reciprocal rhythmic interactions between reticular (nRT) and thalamo-cortical (TC) cells. Induced by a recurrent inhibition from nRT cells, postinhibitory rebound spike bursts in TC cells entrain cortical populations in spindle oscillations (11). In turn, a cortico-thalamic feedback synchronizes spindle oscillations in widespread thalamic territories (12). It has been suggested that synaptic blockade in the thalamus filters out sensory transmissions to the forebrain during sleep spindles (13). Accordingly, in humans, modifications of auditory event-related potentials (ERPs) recorded on the scalp have been interpreted as reflecting the thalamic inhibition of information processing during spindles, although direct evidence supporting this hypothesis is still lacking (14, 15).To assess this hypothesis, we used simultaneous EEG/functional MRI (fMRI) recordings to better characterize regio...