Sleep in the form of regularly occurring periods of quiescence and some amount of sleep rebound can be found in even the simplest of organisms from earthworms and fruit flies to nonhuman primates and human beings. We do not see evidence, however, of the emergence of distinct sleep states until we come to the reptiles. Birds and aquatic mammals also evidence distinct sleep states including the phenomenon of unihemispheric sleep, which allows these animals to sleep while flying or swimming. REM may only occur bihemispherically. The presence of high voltage slow waves as well as REM-like brain activation patterns in reptiles, birds, and mammals suggests that the biphasic, REM, and NREM sleep phases we find in humans is a very ancient adaptation indeed and that its benefits outweigh the risks associated with quiescence and reduced responsiveness to the environment.

The time it takes to fall asleep (latency) declines until midlife and then remains about the same into old age. Time spent awake after initial sleep onset (WASO) declines across the lifespan but its proportion of total sleep period increases. That is, people tend to have a greater number of awakenings as they age. REM percentages decline with age but the proportion of total sleep spent in REM remains about the same. The same is the case with N2 stage light sleep and N1 transitional sleep; these proportions remain about the same or slightly increase as people age. Finally, N3 slow wave sleep undergoes a steady decline with age until it almost completely disappears in old age. Throughout the lifespan sleep evidences intimate and possibly bidirectional causal associations with socio-emotional attachment processes between child and parent during the developmental phase and then between sexual/romantic and close friends during the adult phase. These relationships between sleep processes and attachment processes once more underline the social nature of sleep.

Dreams are cognitions that are typically dependent on sleep. However, not all forms of cognition occur during sleep. In spontaneously recalled dreams the visual sense predominates. It is rare to remember a smell or a taste from the dream. Reading and computations (arithmetic) do not frequently occur in dreams. Many dreams contain unusual amounts of emotion, and may provide greater access to older memories – -especially during late morning REM dreams. While impairment in critical self-reflective capacities may occur in dreams, it is not clear if all dreams are characterized by impairment in self-reflectiveness. The dreaming mind/brain spontaneously and automatically produces dreams in the form of narratives and likely uses cognitive operations like Freud’s dreamwork to do so.

The ultradian NREM-REM sleep cycle is embedded within the larger twenty-four- hour circadian cycle. The sleep cycle interacts with the circadian cycle, which in turn is controlled by the SCN master clock within the hypothalamus. The two-process model captures interactions between the circadian process and the sleep cycle. Disorders of biological rhythmicity such as delayed sleep phase syndrome and manic depression have significant but treatable effects on sleep.

REM's biobehavioral characteristics are paradoxical in that its physiologic correlates appear to be injurious to the health of the organism while its brain correlates suggest social-emotional functions. Unlike REM, NREM biobehavioral characteristics are slightly less paradoxical but still enigmatic. NREM’s physiologic functions may be related to immune system function while its electrophysiologic properties are clearly related to the restorative functions of sleep. Both REM and NREM sleep likely participate in memory processing but so does the waking state. The fact that NREM appears to be associated with the gradual deactivation of a select group of brain structures that are then reactivated during REM suggests that the two sleep states either work in harmony with one another to maintain optimal brain function or that NREM undoes something that REM instantiates.

Most dreams are filled with social interactions between the dreamer and familiar people in the dreamer’s life. Dream content changes significantly according to the stage in life of the dreamer but social interactions remain a constant in dream content across the lifespan. Much of the social interactions that occur in dreams can be characterized as attachment interactions; that is, interactions that reflect and help shape daytime attachment orientations (e.g., romantic attachments or familial attachments, etc.) of the dreamer. In old age and in death dreams continue to simulate social interactions, but new unfamiliar characters enter the dreams of the old and dying.

A theory of dreams must be able to explain why people think they have mutual dreams, precognitive dreams, and visitations of loved ones from beyond the grave. Simple dismissive explanations that these people are gullible won’t do as the gullibility account does not explain the similar phenomenologies and content features of these extraordinary dreams. The fact that sensorially limited individuals such as people with blindness, deaf-mute, and paraplegic conditions nevertheless have dreams where none of these impairments exist must also be explained in any decent theory of dreams.

REM dreams and NREM dreams differ in terms of content and phenomenologies, these two dream types do not exhaust dream phenomenology. Relative to REM dreams, NREM dreams tend to be less story-like, bizarre, and contain fewer aggressive social interactions and greater numbers of friendly interactions, at least where the dreamer-initiated interactions are concerned.

Theories of REM and NREM functions we have been assuming that the sleep state is doing something for the wake state; i.e., that NREM SWS restores energy for waking consciousness or that REM supports emotional memory consolidation for waking consciousness. But it is also possible that the functions of REM and NREM have more to do with the sleep states themselves rather than with waking consciousness. REM may be undoing something that NREM is doing since REM typically follows NREM in the sleep cycle. Or conversely, NREM SWS may be doing something important for the organism (e.g., immune system repair) but that function is costly, so REM functions to complete, complement, repair, or undo something that NREM had to do to accomplish its primary functions. In this scenario, SWS sleep repairs the immune system each night, but that is so onerous a job that NREM then requires REM to restore NREM’s functional capacity so that it can do its immune system repair again the following night.