Recent studies of sleep that bring closer the development of a cognitive neuroscience of conscious states are reviewed to reveal a unique integration of mechanistic and functional concepts. The electrophysiology of non-rapid eye movement (NREM) sleep has been detailed to reveal the effects of brainstem reticular and neuromodulatory deactivation of the thalamocortical system. But instead of viewing these changes simply as inactivation, it has been suggested that the slow waves and spindles of NREM might reflect a differential mode of information processing by the brain in sleep. In particular, these findings are compatible with a two-stage hypothesis of sleep enhancement of plasticity — a model that is also supported by studies of human cognitive enhancement in sleep. The transformational role of two methodological innovations is emphasized. Home-based studies of sleep allow investigators to obtain vast amounts of data about the kind of consciousness that is associated with active waking, quiet waking, sleep onset, NREM and REM sleep. At the same time, positron emission tomography (PET) imaging is used to reveal the relative blood flow in different regions of the human brain in waking, NREM and REM sleep. The data from these two sources are complementary and informative. For example, in REM sleep (compared with waking) there is more blood flow directed towards the brain stem, the limbic forebrain and the parietal operculum; these findings map onto the marked increase in hallucinatory experience during REM. By contrast, the low levels of thinking during REM map onto the decrease in dorsolateral prefrontal cortical blood flow in that state. The relevance of these findings to the brain-based activation–synthesis theory of dreaming is stressed by a systematic review of the differential increases in activation of forebrain structures revealed by PET imaging in: ascending arousal systems; thalamocortical and thalamic subcortical structures; limbic and paralimbic structures; motor initiation and control areas; visual association cortex; and the inferior parietal lobe. Plasticity is now thought to be a major functional process that is generated by sleep. This idea is supported by recent developmental and human memory studies. The human cognitive studies reveal enhancement of learning by both NREM and REM sleep, indicating a two-stage process. Such a model shares assumptions and structures with Buzsáki's electrophysiological model of a hippocampal-neocortical dialogue. Buzsáki posits a transfer of data from neocortex to hippocampus in active waking, and consolidation of information within the hippocampus along with its transfer back to the neocortex for longer-term storage during quiet waking and NREM. Recent experimental and theoretical work further indicates that intracortical processing occurs during REM, at which time new associative connections might be formed within the neocortex.
