Certain characteristics of dreams are modestly related to one's waking personality. Imaginativeness in REM dreams correlates positively with imaginativeness in Thematic Apperception Test stories. For both NREM and REM reports, length, vividness, emotionality, and distortion correlate positively with psychopathology scores on the Minnesota Multiphasic Personality Inventory. Evidently, emotional activation or tone during sleep parallels emotional activation or tone during wakefulness. Following a stressful presleep film, individual differences in waking anxiety correlate positively with individual differences in anxiety shown in REM reports. Over the menstrual cycle of normal women, waking mood changes correlate positively with mood changes in REM reports. Depressed patients have a depressive tone in their REM reports; schizophrenic patients have disorganized, incoherent REM reports. The findings support a continuity hypothesis that dreams "continue" certain formal characteristics of waking mental life.

Although characteristics of dreams are related to waking personality, the source of the content of any specific dream is unclear. Presleep stimuli have no substantial straightforward effect on dream content. Thus, REM and NREM reports are not substantially different after an evening of physical exercise, challenging mental work, relaxation, or an awesome TGIF party. Compared with presleep emotionally neutral films, violent films do not evoke more violent REM or NREM dreams in adults, preadolescents, or children. In fact, only about 5% of REM reports show incorporation of the filmed presleep violence. Presleep pornographic films are not followed by obviously sexual REM dreams (too bad!). Thirst, induced by 24 hour fluid restriction, is followed by isolated references to drinking rather than persistent themes of thirst and the isolated references occur in only 33% of REM reports.

As with presleep stimuli, stimuli presented during sleep have only a minimal effect on dream content. They are usually not incorporated into dreams. For example, stimulus objects in front of one's eyes which are taped open during sleep are not incorporated into REM dreams. In studies of the effects of external stimuli, tones, light flashes, cold water stimuli, and wrist shocks during REM sleep are incorporated into only 9, 24, 47, and 20% of dream reports, respectively. In virtually all instances of incorporation the stimulus was represented as a momentary event that fit into the ongoing dream narrative, and not as a determinant of the dream's theme. Like external stimuli before and during sleep, physiological events during sleep have no substantial relationship to dream content. In the male, 95% of REM periods are accompanied by erections, but the frequency of manifest sexuality of REM dreams is very low (too bad again). It has been hypothesized that mental activity during phasic REM sleep would be more perceptual, intense, bizarre, or distorted than mentation during tonic REM sleep. However, the findings are that for practical purposes, tonic and phasic REM reports are not significantly different in these qualities. Similarly, mental activity during phasic events of NREM sleep (EEG K-complexes) is not different from that reported during "tonic" NREM sleep. However, REM dreams with many phasic events (i.e., with many rapid eye movements) tend to be more emotional, vivid and better recalled than NREM dreams. Dream content is, therefore, remarkably independent of external psychological and physical stimuli both before and during sleep and remarkably independent of physiological processes during sleep. Hence, the sources of dream content, i.e., the themes and specific elements, remain a mystery.

The isolation of dream content from presleep and sleep stimuli is one aspect of the single-mindedness of dream consciousness. During dreaming, consciousness is usually focused only on the dream experience. It does not reflect on its own state (i.e., dreaming) or the dreamer's physical state (e.g., lying in bed). It has no intrusions of other thoughts or images which are then reflectively evaluated for their relevance to the dream experience. It has no intrusions from external stimuli or the organismic state (e.g., erections). The dream is simply a coherent experience without reflection on itself as a whole. The dream has no occurrences that are expressed as coming from outside the dream setting. Thus, the single-mindedness of dream consciousness is a salient difference from the multiple inputs and focuses of waking consciousness. The significance of this difference for the psychology of dreaming and for brain correlates of dreaming remains unknown.

Although the sources of specific dream content remain a mystery, something is known about the processes involved in the construction of dreams. The evidence comes from the changes in children's dreams as children develop from about age three to about age nine. At age 3 to 5, REM reports are relatively rare, with only 1/3 of the awakenings yielding a report of a dream. The reports are brief, with virtually no story, feeling, self representation, or activity by dream characters. At age 5 to 7, reports are still rare but the report length triples and contains activity. By age 7 to 9 dream reports begin to have the qualities of adult reports. They are longer, with self-portrayal and character activity set in a story or narrative. Across this age range, differences in dream reporting are more strongly related to certain cognitive skills than to psychosocial or emotional status. Waking measures of visuospatial construction and analysis are better predictors of dream reports than waking memory for visuospatial material or waking ability to describe visuospatial material. Thus, the evidence suggests that the increase in dream reports and narrative quality during development depends more on cognitive skills involved in constructing dreams than on skills involved in remembering or reporting dreams. In other words, the findings suggest that the construction of dreams depends upon the presence of cognitive abilities to analyze, abstract, manipulate and construct visuospatial images or ideas.

There are, of course, other theories of dream construction, i.e., the processes that instigate dreams and that determine their formal characteristics. Classical psychoanalytical theory postulates that emotional processes construct dreams; that dreams are instigated by a relatively infrequent, unpredictable, forbidden unconscious wish that rises during sleep. The dream is an hallucinated, disguised fulfillment of the wish. Dream distortion is a virtually universal feature of adult dreams because it serves to disguise the forbidden wish. In children, conscience has not yet developed so the instigating wish is not forbidden; hence, children's dreams are clear wish fulfillments without distortion. However, the theory cannot easily explain the frequent, predictable REM sleep dreams; the mundane undistorted quality of most adult dreams; the absence of undisguised wishes in children's dreams; the presence of abundant dreaming throughout the states of consciousness; and the differences between REM and NREM dreams.

A recent reductionist theory postulates that neurophysiological processes construct dreams. According to this theory, the dream is instigated by a periodic activation of the cerebral cortex that originates in the random firing of the pontine neurons which generate REM sleep (see Part D., NREM and REM Sleep). Dream distortion occurs regularly because the cortex cannot synthesize-make complete sense of-this random pattern of neuronal activation. However, this theory cannot easily explain the abundant occurrence of dreaming in the absence of REM sleep (i.e., the theory postulates that dreams are instigated by the activity of pontine REM generators which must be silent during the dreams outside of REM sleep); the realistic mundane quality of REM dreams (random activation should produce frequent disruptions in the flow of dream images or perhaps even kaleidoscopic images); the fact that phasic REM (the time of strongest pontine activation) is not accompanied by more bizarre content than tonic REM; the thematic coherence of most dreams; and the occurrence of repetitive dreams (whose presence would be extremely unlikely if content were determined by a random process).

A recent variation on the preceding theory postulates that an activated cortex is necessary for both waking thought and dreaming-whether at sleep onset or during REM sleep, and that continual sensory input to the mind/brain is necessary for it to produce the rational, orderly thought and perception of the waking state. A moderate reduction in sensory input, as when one is lying awake in a quiet dark room or during sleep onset, results in a stream of images that are often visually vivid and bizarre. But in REM sleep the sensory input to the mind/brain is reduced so drastically that the images are interpreted as "real" events. That is, they are hallucinations. Thus, different properties of dreaming can be attributed to different physiological and external sensory factors. During NREM sleep cortical activation is so low that the brain/mind can produce very little thought or dreaming.

In summary, sleep laboratory studies have revealed an unexpected abundance of dreaming in all states of consciousness; an unexpected thematic coherence of dream content; the isolation of dreams from other mental and physical events; and the fact that cognitive skills, rather than emotional needs are involved in dream construction. But we are still very far from understanding the sources of specific dream images and narratives, how they combine to make a dream, and the relation of specific dream content to waking mentation. We are also far from understanding how the sleeping brain invents such coherent, sensible, novel stories. Surely something basic about the nature of sleep lies hidden in the significance of this remarkable mental achievement.