The stuff that dreams aren't made of: why wake-state and dream-state sensory experiences differ.

Abstract

It is adaptive for individuals to be continuously alert and responsive to external stimuli (such as the sound and odor of an approaching predator or the cry of an infant), even during sleep. Natural selection thus has disfavored the occurrence during sleep of hallucinations that compromise external vigilance. In the great majority of mammalian species, including Homo sapiens, closed eyes and immobility are basic aspects of sleep. Therefore, (a) visual and movement sensory modalities (except kinesthesis) do not provide the sleeper with accurate information about the external environment or the sleeper's relationship to that environment; (b) the sleeper's forebrain "vigilance mechanism" does not monitor these modalities; hence (c) visual and movement hallucinations--similar or identical to percepts--can occur during sleep without compromising vigilance. In contrast, the other sensory modalities do provide the sleeper with a continuous flow of information about the external environment or the sleeper's relationship to that environment, and these modalities are monitored by the vigilance mechanism. Hallucinations of kinesthesis, pain, touch, warmth, cold, odor, and sound thus would compromise vigilance, and their occurrence during sleep has been disfavored by natural selection. This vigilance hypothesis generates novel predictions about dream phenomenology and REM-state neurophysiology and has implications for the general study of imagery.