The Meaning of Dreams may reflect a fundamental aspect of mammalian memory processing. Crucial information acquired during the waking state may be reprocessed during

by Jonathan Winson

The Author hroughout history human beings have sought to understand the meaning of dreams. The ancient Egyptians believed dreams JONATHAN WINSON possessed oracular power—in the Bible, for example, Joseph’s

started his career as an elucidation of Pharaoh’s averted seven years of famine. SCALA/ART RESOURCE aeronautical engineer, grad- TOther cultures have interpreted dreams as inspirational, curative or alterna- uating with an engineering degree from the California tive reality. During the past century, scientists have offered conflicting psy- Institute of Technology in chological and neuroscientific explanations for dreams. In 1900, with the 1946. He completed his publication of The Interpretation of Dreams, Sigmund Freud proposed that Ph.D. in mathematics at dreams were the “royal road” to the unconscious; that they revealed in dis- Columbia University and guised form the deepest elements of an individual’s inner life. then turned to business for More recently, in contrast, dreams have been characterized as meaningless, 15 years. Because of his keen interest in neurosci- the result of random nerve cell activity. Dreaming has also been viewed as the ence, however, Winson means by which the rids itself of unnecessary information—a process started to do research at the of “reverse learning,” or unlearning. Rockefeller University on Based on recent findings in my own and other neuroscientific laboratories, memory processing during I propose that dreams are indeed meaningful. Studies of the hippocampus (a waking and sleeping states. brain structure crucial to memory), of rapid eye movement (REM) sleep and In 1979 he became an asso- of a brain wave called theta rhythm suggest that dreaming reflects a pivotal ciate professor there and as professor emeritus contin- aspect of the processing of memory. In particular, studies of theta rhythm in ues his work on memory subprimate animals have provided an evolutionary clue to the meaning of and dreaming. His research dreams. They appear to be the nightly record of a basic mammalian memo- has been supported by the ry process: the means by which animals form strategies for survival and National Institute of Men- evaluate current experience in light of those strategies. The existence of this tal Health, the National Sci- process may explain the meaning of dreams in human beings. ence Foundation and the Harry F. Guggenheim Foundation. Stages of Sleep and Dreaming

he physiology of dreaming was first understood in 1953, when re- Tsearchers characterized the human sleep cycle. They found that sleep in humans is initiated by the hypnogogic state, a period of several minutes Dreaming occurs solely when thoughts consist of fragmented images or minidramas. The hyp- during REM sleep. nogogic state is followed by slow-wave sleep, so called because at that time the brain waves of the neocortex (the convoluted outer mantle of the brain) are low in frequency and large in amplitude. These signals are measured as electroencephalographic (EEG) recordings. Researchers also discovered that a night’s sleep is punctuated by periods in which the EEG readings are irregular in frequency and low in amplitude— similar to those observed in awake individuals. These periods of mental ac- tivity are called REM sleep. Dreaming takes place solely during these peri- ods. While in REM sleep, motor neurons are inhibited, preventing the body from moving freely but allowing extremities to remain slightly active. Eyes Photo Researchers, Inc. Photo Researchers,

JACOB’S LADDER, painted in 1973 by Marc Chagall, depicts a biblical story. Jacob dreams of angels ascend-

LABAT/JERRICAN LABAT/JERRICAN ing to and descending from heaven on a ladder.

58 Mysteries of the Mind Reprinted from the November 1990 issue Copyright 1990 Scientific American, Inc. move rapidly in unison under closed This sleep cycle—alternating slow- entists have discovered additional neu- lids, breathing becomes irregular and wave and REM sleep—appears to be rophysiological aspects of REM sleep. heart rate increases. present in all placental and marsupial They determined that neural control of The first REM stage of the night fol- mammals. Mammals exhibit the vari- this stage of the sleep cycle is centered lows 90 minutes of slow-wave sleep and ous REM-associated characteristics ob- in the brain stem (the brain region clos- lasts for 10 minutes. The second and served in humans, including EEG read- est to the spinal cord) and that during third REM periods follow shorter slow- ings similar to those of the awake state. REM sleep neural signals—called pon- wave sleep episodes but grow progres- Animals also dream. By destroying neu- tine-geniculate-occipital (PGO) cortex sively longer themselves. The fourth and rons in the brain stem that inhibit move- spikes—proceed from the brain stem to final REM interval lasts 20 to 30 min- ment during sleep, researchers found the center of visual processing, the visu- utes and is followed by awakening. If a that sleeping cats rose up and attacked al cortex. Brain stem neurons also initi- dream is remembered at all, it is most or were startled by invisible objects—os- ate a sinusoidal wave (one resembling a often the one that occurred in this last tensibly images from dreams. sine curve) in the hippocampus. This phase of REM sleep. By studying subprimate animals, sci- brain signal is called theta rhythm.

Copyright 1990 Scientific American, Inc. PREFRONTAL CORTEX were merely the “best fit” the forebrain could provide to this random bombard- ment from the brain stem. Although SEPTUM dreams might at times appear to have psychological content, their bizarreness was inherently meaningless. The sense, or plot, of dreams resulted from order that was imposed on the chaos of neural signals, Hobson said. “That order is a function of our own personal view of the world, our remote memories,” Hobson wrote. In other words, the individual’s emotional vo- cabulary could be relevant to dreams. In a further revision of the original hy- pothesis, Hobson also suggested that brain stem activation may merely serve to switch from one dream episode to another.

Reverse Learning

VISUAL lthough Hobson and McCarley had HIPPOCAMPUS ENTORHINAL CORTEX CORTEX A presented an explanation of dream HIPPOCAMPUS content, the basic function of REM sleep admittedly remained unknown. In 1983 BRAIN STEM Francis Crick of the Salk Institute in La Jolla, Calif., and Graeme Mitchison of the University of Cambridge in England DENTATE GYRUS proposed the idea of reverse learning. SPINAL CORD Working from the Hobson-McCarley CA3 CA1 CELLS assumption of random neocortical bom-

CELLS OL DONNER bardment by PGO waves and their own AR C knowledge of the behavior of stimulated ANATOMY OF THE BRAIN and cross section of the hippocampus show some of the neural networks, they postulated that a regions involved in dreaming. In the hippocampus, incoming information is processed complex associational neural network sequentially in the dentate gyrus and the CA3 and the CA1 pyramidal cells. In subpri- such as the neocortex might become mate species, theta rhythm is generated in the dentate gyrus and the CA1 cells. overloaded by vast amounts of incoming information. The neocortex could then develop false, or “parasitic,” thoughts At least one animal experiences slow- unknown. In light of the discovery of that would jeopardize the true and or- wave but not REM sleep—and, conse- REM sleep, certain elements of his psy- derly storage of memory. quently, does not exhibit theta rhythm choanalytic theory were modified, and According to Crick and Mitchison’s when asleep. This animal is the echidna, the stage was set for more neurological- hypothesis, REM sleep served to erase or spiny anteater, an egg-laying mam- ly based theories. Dreaming came to be these spurious associations on a regular mal (called a monotreme) that provides understood as part of a biologically de- basis. Random PGO waves impinged on some insight into the origin of dream- termined sleep cycle. Yet the central the neocortex, resulting in erasure, or un- ing. The absence of REM sleep in the concept of Freud’s theory—namely, the learning, of the false information. This echidna suggests that this stage of the belief that dreams reveal a censored process served an essential function: it sleep cycle evolved some 140 million representation of our innermost uncon- allowed the orderly processing of mem- years ago, when marsupials and placen- scious feelings and concerns—continues ory. In humans, dreams were a running tals diverged from the monotreme line. to be used in psychoanalysis. record of these parasitic thoughts—ma- (Monotremes were the first mammals Some theorists abandoned Freud al- terial to be purged from memory. “We to develop from reptiles.) together following the neurological dis- dream to forget,” Crick and Mitchison By all evolutionary criteria, the perpet- coveries. In 1977 J. Allan Hobson and wrote. uation of a complex brain process such Robert McCarley of Harvard Medical The two researchers proposed a revi- as REM sleep indicates that it serves an School proposed the “activation-syn- sion in 1986. Erasure of parasitic important function for the survival of thesis” hypothesis. They suggested that thoughts accounted only for bizarre mammalian species. Understanding that dreaming consists of associations and dream content. Nothing could be said function might reveal the meaning of memories elicited from the forebrain (the about dream narrative. Furthermore, dreams. neocortex and associated structures) in dreaming to forget, they said, was bet- When Freud wrote The Interpretation response to random signals from the ter expressed as dreaming to reduce of Dreams, the physiology of sleep was brain stem such as PGO spikes. Dreams fantasy or obsession.

60 Mysteries of the Mind The Meaning of Dreams Copyright 1990 Scientific American, Inc. None of these hypotheses seems to ex- ta rhythm in the rat. Rats demonstrated In 1974, by recording signals from the plain adequately the function of dream- theta rhythm only during movement, hippocampus of freely moving rats and ing. On the one hand, Freud’s theory typically when they explored. In 1969, rabbits, I found the source from which lacked physiological evidence. (Although however, Case H. Vanderwolf of the Uni- theta rhythm was generated in the hip- Freud had originally intended to de- versity of Western Ontario discovered pocampus. Together with the neocortex, scribe the neurology of the unconscious there was one behavior during which the the hippocampus is believed to provide and of dreams in his proposed “Project animals he studied, including the rat, the neural basis for memory storage. The for a Scientific Psychology,” the under- showed theta rhythm: REM sleep. hippocampus (from the Greek word for taking was premature, and he limited In 1972 I published a commentary “seahorse,” which it resembles in shape) himself to psychoanalysis.) On the oth- pointing out that the different occurrenc- is a sequential structure composed of er hand, despite revisions to incorpo- es of theta rhythm could be understood three types of neurons. Information from rate elements of psychology, most of in terms of animal behavior. Awake an- all sensory and associational areas of the the later theories denied that dreams imals seemed to show theta rhythm neocortex converges in a region called had meaning. when they were behaving in ways most the entorhinal cortex; from there it is Exploring the neuroscientific aspects crucial to their survival. In other words, transmitted to the three successive neu- of REM sleep and of memory process- theta rhythm appeared when they exhib- ronal populations of the hippocampus. ing seemed to me to hold the greatest ited behavior that was not genetically The signal arrives first at the granule potential for understanding the mean- encoded—such as feeding or sexual be- cells of the dentate gyrus, then at the ing and function of dreams. The key to havior—but rather a response to chang- CA3 pyramidal cells (so called because this research was theta rhythm. ing environmental information. Preda- of their triangular shape) and finally at Theta rhythm was discovered in 1954 tory behavior in the cat, prey behavior the pyramidal cells of CA1. After infor- in awake animals by John D. Green and in the rabbit, and exploration in the rat mation is processed by this trio of cells, Arnaldo A. Arduini of the University of are, respectively, most important to their it is retransmitted to the entorhinal cor- California at Los Angeles. The research- survival. (For example, a hungry rat will tex and then back to the neocortex. ers observed a regular sinusoidal signal explore before it eats even if food is My studies showed that theta rhythm of six cycles per second in the hippocam- placed in front of it.) was produced in two regions within the pus of rabbits when the animals were hippocampus: the dentate gyrus and the apprehensive of stimuli in their envi- Role of Theta Rhythm CA1 neurons. The rhythms in these two ronment. They named the signal theta areas were synchronous. Subsequently, rhythm after a previously discovered urthermore, because the hippocam- Susan Mitchell and James B. Ranck, Jr., EEG component of the same frequency. Fpus is involved in memory process- of the State University of New York Theta rhythm was subsequently re- ing, the presence of theta rhythm dur- Downstate Medical Center identified a corded in the tree shrew, mole, rat and ing REM sleep in that region of the third synchronous generator in the en- cat. Although it was consistently ob- brain might be related to that activity. I torhinal cortex, and Robert Verdes of served in awake animals, theta rhythm suggested that the theta rhythm reflect- Wayne State University discovered the was correlated with very different be- ed a neural process whereby information brain stem neurons that control theta haviors in each species. For example, essential to the survival of a species— rhythm. These neurons transmit signals in marked contrast to the rabbit, envi- gathered during the day—was repro- to the septum (a forebrain structure) ronmental stimuli did not induce the- cessed into memory during REM sleep. that activate theta rhythm in the hippo-

1 SEC

THETA RHYTHM REM SLEEP REM SLEEP REM SLEEP

PREDATION APPREHENSION EXPLORATION TRICIA J. WYNNE A P

THETA RHYTHM is present during different waking behav- animal’s survival. In placental and marsupial animals, theta iors in different species. Each of these behaviors is pivotal to the rhythm is present during rapid eye movement (REM) sleep.

The Meaning of Dreams Mysteries of the Mind 61 Copyright 1990 Scientific American, Inc. ing electrode. Using this technique, Bliss and his colleagues measured the normal

OL DONNER response to a single electrical pulse. Then AR C they applied a long series of high-fre- quency signals—called tetanic pulses— to this pathway. After the train of tetanic stimuli, a single electrical pulse caused much greater firing in the granule cells than had been observed prior to the ex- periment. The heightened effect persisted for as long as three days. This phenom- enon of LTP was precisely the kind of PRESYNAPTIC increase in neuronal strength that could MEMBRANE SYNAPTIC CLEFT be capable of sustaining memory. LTP NEUROTRANSMITTER POSTSYNAPTIC is now considered a model for learning ION CHANNEL MEMBRANE and memory. LTP is achieved by the activity of the NMDA (N-methyl-D-aspartate) recep- tor. This molecule is embedded in the dendrites of the granule cells and the CA1 cells of the hippocampus as well as in neurons throughout the neocor- GLUTAMATE tex. Like other neuronal receptors, the NMDA receptor is activated by a neu- rotransmitter—glutamate in this case. Glutamate momentarily opens a non- NMDA channel in the granule cell den- CALCIUM drite, allowing sodium from the extra- cellular space to flow into the neuron. This influx causes the granule cell to be- come depolarized. If the depolarization is sufficient, the granule cell fires, trans- NMDA RECEPTOR mitting information to other nerve cells. Unlike other neuronal receptors, SODIUM NMDA possesses an additional proper- SECOND RELEASE ty. If a further activation of glutamate DEPOLARIZATION OF GLUTAMATE LTP occurs while the granule cell is depolar- GABOR KISS ized, a second channel opens up, allow- NMDA RECEPTOR activation induces long-term potentiation (LTP), a model for ing an influx of calcium. Calcium is memory. The release of the neurotransmitter glutamate (left panel) opens a non– thought to act as a second messenger, ini- NMDA receptor channel, allowing the influx of sodium, which depolarizes the neuron. tiating a cascade of intracellular events If a further release of glutamate occurs while the cell is depolarized (center panel), the that culminates in long-lasting synaptic NMDA receptor opens a second channel, which allows calcium to flow in, leading to LTP. LTP occurs as a result of increased sodium through the non-NMDA channel changes—or LTP. (The description giv- (right panel) and the subsequent greater depolarization of the cell. en here has been necessarily simplified. LTP is the subject of extensive ongoing investigation.) campus and the entorhinal cortex. Thus, that reflects previous activity—showed Because the tetanic impulses applied the brain stem activates the hippocam- the means by which memory might be by Bliss and his colleagues did not oc- pus and the neocortex—the core memo- encoded. Timothy V. P. Bliss and A. R. cur naturally in the brain, the question ry system of the brain. Gardner-Medwin of the National Insti- remained as to how LTP was achieved To determine the relation between tute of Medical Research in London under normal circumstances. In 1986 theta rhythm and memory, I made a le- and Terje Lømo of the University of John Larson and Gary S. Lynch of the sion in the rat septum. Rats that had Oslo found changes in nerve cells that University of California at Irvine and previously learned, using spatial cues, had been intensely stimulated with elec- Gregory Rose and Thomas V. Dunwid- to locate a particular position in a maze trical pulses. die of the University of Colorado at were no longer able to do so after their Denver suggested that the occurrence of septums were disabled. Without theta Long-Term Memory Storage LTP in the hippocampus was linked to rhythm, spatial memory was destroyed. theta rhythm. They applied a small num- Studies of the cellular changes that arlier studies had shown that if one ber of electrical pulses to CA1 cells in bring about memory illustrated the role Estimulated the pathway from the the rat hippocampus and produced LTP, of theta rhythm. In particular, the dis- entorhinal cortex to the granule cells of but only when the pulses were separat- covery in 1973 of long-term potentia- the hippocampus, the response of these ed by the normal time that elapses be- tion (LTP)—a change in neural behavior cells could be measured with a record- tween two theta waves—approximately

62 Mysteries of the Mind The Meaning of Dreams Copyright 1990 Scientific American, Inc. 200 milliseconds. Theta rhythm is ap- versity College London had demonstrat- in the waking state reprocess informa- parently the natural means by which the ed that individual CA1 neurons in the tion during slow-wave sleep and then in NMDA receptor is activated in neurons rat hippocampus fired when the awake REM sleep. These results suggest that in the hippocampus. animal moved to a particular location— sleep processing of memory may have Work in my laboratory at the Rocke- namely, the neuron’s place field. The two stages—a preliminary stage in slow- feller University duplicated Larson and implication of this finding was that the wave sleep and a later phase in REM Lynch’s CA1 findings, but this time in CA1 neuron fired to map the environ- sleep, when dreaming occurs. the hippocampal granule cells. Constan- ment, thereby committing it to memory. tine Pavlides, Yoram J. Greenstein and I In 1989 Pavlides and I located two Evolution of REM Sleep then demonstrated that LTP was depen- CA1 neurons in the rat hippocampus dent on the presence and phase of theta that had different place fields. We re- vidence that theta rhythm encodes rhythm. If electrical pulses were applied corded from both cells simultaneously. Ememories during REM sleep may to the cells at the peak of the theta After determining the normal firing rates be derived not only from neuroscientif- wave, LTP was induced. But if the same in awake and asleep animals, we posi- ic studies but also from evolution. The pulse were applied at the trough of the tioned a rat in the place field of one of emergence of a neural mechanism to waves—or when theta rhythm was ab- the neurons. The neuron fired vigorous- process memory in REM sleep suggests sent—LTP was not induced. ly, mapping that location. The second differences in brain anatomy between A coherent picture of memory process- cell fired only sporadically because it mammals that have that aspect of the ing was emerging. As a rat explores, for was not coding space. We continued re- sleep cycle and those that do not. And example, brain stem neurons activate cording from the two pairs of neurons in fact, such differences clearly exist be- theta rhythm. Olfactory input (which as the rat moved about and then entered tween the echidna and the marsupials in the rat is synchronized with theta several sleep cycles. Six pairs of neurons and placentals. rhythm, as is the twitching of whiskers) were studied in this manner. The echidna has a large convoluted and other sensory information converge We found that neurons that had cod- prefrontal cortex, larger in relation to on the entorhinal cortex and the hippo- ed space fired at a normal rate as the the rest of the brain than that of any campus. There they are partitioned into animal moved about prior to sleep. In other mammal, even humans. I believe 200-millisecond “bites” by theta rhythm. sleep, however, they fired at a signifi- it needed this huge prefrontal cortex to The NMDA receptors, acting in con- cantly higher rate than their previous perform a dual function: to react to in- junction with theta rhythm, allow for sleeping baseline. There was no such in- coming information in an appropriate long-term storage of this information. crease in firing rate during sleep in neu- manner based on past experience and to A similar process occurs during REM rons that had not mapped space. This evaluate and store new information to sleep. Although there is no incoming in- experiment suggested that the reprocess- aid in future survival. Without theta formation or movement during REM ing or strengthening of information en- rhythm during REM sleep, the echidna sleep, the neocortical-hippocampal net- coded when the animal was awake oc- would not be able to process informa- work is once again paced by theta curred in sleep at the level of individual tion while it slept. (The echidna does, rhythm. Theta rhythm might produce neurons. however, show theta rhythm when for- long-lasting changes in memory. Bruce L. McNaughton and his col- aging for food.) For higher capabilities leagues at the University of Arizona have to develop, the prefrontal cortex would Storing Spatial Memory developed a technique for simultaneous- have to become increasingly large—be- ly recording from a large number of neu- yond the capacity of the skull—unless he results of one of my further ex- rons in the hippocampus that map loca- another brain mechanism evolved. Tperiments served to show that spa- tions. Their technique allows definitive REM sleep could have provided this tial memory was indeed being stored in patterns of firing to be identified. In an- new mechanism, allowing memory pro- the rat hippocampus during sleep. John imal studies, they found that ensembles cessing to occur “off-line.” Coincident O’Keefe and J. Dostrovsky of the Uni- of place-field neurons that code space with the apparent development of REM

S 6

T OL 3 LTP

MILLIV 0 200 MILLISECONDS PULSES AT PEAK 0 10 20 MILLISECONDS

ELECTRICAL PULSE THETA RHYTHM

S 6 T

OL 3 NO LTP

MILLIV 0

GABOR KISS PULSES AT TROUGH 0 10 20 MILLISECONDS LTP in the granule cells of the hippocampus is achieved by theta milliseconds (the time between the peaks of two theta waves), rhythm. Electrical pulses, which have been separated by 200 result in LTP when applied at the peak of theta rhythm.

The Meaning of Dreams Mysteries of the Mind 63 Copyright 1990 Scientific American, Inc. PREFRONTAL EVOLUTIONARY TREE shows the di- CORTEX vergence of placentals and marsupials from monotremes. The echidna, which

OL DONNER does not possess REM sleep, has a larger AR

C prefrontal cortex compared with the rest of its brain than does any mammal, even CAT OPOSSUM ECHIDNA 0 humans. It is larger than in similarly sized animals, including the opossum and cat.

PLACENTALS MARSUPIALS MONOTREMES going strategy for behavior. Although theta rhythm has not yet been demon- 50 strated in primates, including humans, the brain signal provides a clue to the origin of dreaming in humans. Dreams may reflect a memory-processing mech- anism inherited from lower species, in which information important for sur- 100 vival is reprocessed during REM sleep. LIVE- BEARING This information may constitute the core MAMMALS of the unconscious. Because animals do not possess lan- guage, the information they process dur- REM SLEEP ing REM sleep is necessarily sensory. 150

YEARS AGO (MILLIONS) YEARS AGO Consistent with our early mammalian origins, dreams in humans are sensory, EGG-LAYING MAMMALS primarily visual. Dreams do not take the form of verbal narration. Also in keeping with the role REM SLOW-WAVE SLEEP sleep played in processing memories in 200 animals, there is no functional necessity FIRST TRUE MAMMALS for this material to become conscious. Consciousness arose later in evolution in humans. But neither is there any rea- MAMMALLIKE REPTILES son for the material of dreams not to

GABOR KISS 250 reach consciousness. Therefore, dreams can be remembered—most readily if awakening occurs during or shortly af- ter a REM sleep period. sleep in marsupial and placental mam- PGO spikes, accompany rapid eye move- Consistent with evolution and evi- mals was a remarkable neuroanatomical ment in the waking state and also dur- dence derived from neuroscience and change: the prefrontal cortex was dra- ing REM sleep. The function of these reports of dreams, I suggest that dreams matically reduced in size. Far less pre- signals has not yet been established, but reflect an individual’s strategy for sur- frontal cortex was required to process they may serve to alert the visual cortex vival. The subjects of dreams are broad- information. That area of the brain could to incoming information when the ani- ranging and complex, incorporating develop to provide advanced perceptu- mal is awake and may reflect the repro- self-image, fears, insecurities, strengths, al abilities in higher species. cessing of this information during REM grandiose ideas, sexual orientation, de- The nature of REM sleep supports sleep. In any case, PGO spikes do not sire, jealousy and love. this evolutionary argument. During the disturb sleep and do not have to be sup- Dreams clearly have a deep psycho- day, animals gather information that in- pressed—unlike motor neurons. logical core. This observation has been volves locomotion and eye movement. reported by psychoanalysts since Freud The reprocessing of this information Strategy for Survival and is strikingly illustrated by the work during REM sleep would not be easily of Rosalind Cartwright of Rush-Presby- separated from the locomotion related ith the evolution of REM sleep, terian–St. Luke’s Hospital in Chicago. to the experience—such disassociation Weach species could process the in- Cartwright is studying a series of 90 sub- might be expecting too great a revision formation most important for its sur- jects who are undergoing marital sepa- of brain circuitry. So to maintain sleep, vival, such as the location of food or the ration and divorce. All the subjects are locomotion had to be suppressed by in- means of predation or escape—those clinically evaluated and psychologically hibiting motor neurons. Suppressing eye activities during which theta rhythm is tested to ascertain their attitudes and movement was unnecessary because this present. In REM sleep this information responses to their personal crisis. Cart- activity does not disturb sleep. may be accessed again and integrated wright’s subjects are also awakened Eye movement potentials, similar to with past experience to provide an on- from REM sleep to report their dreams,

64 Mysteries of the Mind The Meaning of Dreams

Copyright 1990 Scientific American, Inc. which are then interpreted by the sub- in his work. There is an unconscious, and Mitchison hypothesis as applied to the jects themselves without questions that dreams are indeed the “royal road” to hippocampus would suggest that neu- might influence their interpretation. In its understanding. The characteristics of rons fire randomly during REM sleep, the 70 individuals studied to date, the the unconscious and associated pro- providing reverse learning. Instead, in my dream content conveys the person’s un- cesses of brain functioning, however, are experiment on the neurons that coded conscious thoughts and is strongly cor- very different from what Freud thought. space, these neurons fired selectively, im- related with the manner in which he or Rather than being a cauldron of un- plying an orderly processing of memory. she is coping with the crisis while awake. tamed passions and destructive wishes, Recently Avi Karni and his colleagues Although the topic “chosen” for con- I propose that the unconscious is a co- at the Weizmann Institute of Science in sideration during a night’s sleep is un- hesive, continually active mental struc- Israel were able to show that memory predictable, certain of life’s difficulties— ture that takes note of life’s experiences processing occurs in humans during as in the case of Cartwright’s subjects— and reacts according to its own scheme REM sleep. In their experiment, indi- so engage psychological survival that of interpretation. Dreams are not dis- viduals learned to identify particular pat- they are selected for REM sleep process- guised as a consequence of repression. terns on a screen. The memory of this ing. In the ordinary course of events, de- Their unusual character is a result of skill improved after a night with REM pending on the individual’s personality, the complex associations that are culled sleep. When the subjects were deprived the themes of dreams may be freewheel- from memory. of REM sleep, memory consolidation ing. Moreover, when joined with the in- did not occur. This study is an impor- tricate associations that are an intrinsic Memory Consolidation tant breakthrough and opens a particu- part of REM sleep processing, the larly promising field for exploration. dream’s statement may be rather obscure. esearch on REM sleep suggests Further study will continue to eluci- Nevertheless, there is every reason to Rthat there is a biologically relevant date the meaning of dreams. In particu- believe that the cognitive process taking reason for dreaming. The revised version lar, an experiment is needed to deter- place in Cartwright’s subjects occurs in of the Hobson-McCarley activation- mine whether eliminating theta rhythm every individual. Interpretation of the synthesis hypothesis acknowledges the during REM sleep alone results in a coherent statement that is being made deep psychological core of dreams. In memory deficit. Because theta rhythm depends on the individual’s tracing of its present truncated form, the hypothe- has not been demonstrated in primates, relevant or similar events. These associ- sis of random brain stem activation has it may have disappeared as vision re- ations are strongly biased toward early little explanatory or predictive power. placed olfaction as the dominant sense. childhood experience. The Crick-Mitchison hypothesis pro- An equivalent neural mechanism may My hypothesis also offers an expla- vides a function for REM sleep—reverse exist in the hippocampus that periodi- nation for the large amount of REM learning—but it does not apply to narra- cally activates the NMDA receptor. sleep in infants and children. Newborns tive, only to the bizarre elements of the These studies in animals and others to spend eight hours a day in REM sleep. dream. What this implies with regard to come in humans will probe fundamen- The sleep cycle is disorganized at this REM processing in lower species must tal aspects of memory processing and age. Sleep occurs in 50- to 60-minute be defined before the theory can be eval- the neuroscientific basis of human psy- bouts and begins with REM rather uated further. In addition, the Crick- chological structure. SA than with slow-wave sleep. By the age of two, REM sleep is reduced to three hours a day, and the adult pattern has been established. Thereafter, the time Further Reading spent in REM sleep gradually diminish- es to a little less than two hours. Interspecies Differences in the Occurrence of Theta. Jonathan Winson in Behav- ioral Biology, Vol. 7, No. 4, pages 479–487; 1972. REM sleep may perform a special Loss of Hippocampal Theta Rhythm Results in Spatial Memory Deficit in the function in infants. A leading theory Rat. Jonathan Winson in Science, Vol. 201, No. 435, pages 160–163; 1978. proposes that it stimulates nerve growth. Brain and Psyche: The Biology of the Unconscious. Jonathan Winson. Anchor Press, Whatever the purpose in infants may be, Doubleday, 1985. I suggest that at about the age of two, Long-Term Potentiation in the Dentate Gyrus Is Induced Preferentially on the when the hippocampus, which contin- Positive Phase of Q-Rhythm. Constantine Pavlides, Yoram J. Greenstein, Mark Grud- ues to develop after birth, becomes func- man and Jonathan Winson in Brain Research, Vol. 439, pages 383–387; 1988. Influences of Hippocampal Place Cell Firing in the Awake State on the Activity tional, REM sleep takes on its interpre- of These Cells during Subsequent Sleep Episodes. Constantine Pavlides and tive memory function. The waking in- Jonathan Winson in Journal of Neuroscience, Vol. 9, No. 8, pages 2907–2918; August, formation to be integrated at this point 1989. in development constitutes the basic cog- Dependence on REM Sleep of Overnight Improvement of a Perceptual Skill. Avi nitive substrate for memory—the con- Karni, David Tanne, Barton S. Rubenstein, Jean J. M. Askenasy and Dov Sagi in Science, cept of the real world against which lat- Vol. 265, pages 679–682; July 29, 1994. Reactivation of Hippocampal Ensemble Memories during Sleep. Mathew A. Wilson er experiences must be compared and and Bruce L. McNaughton in Science, Vol. 265, pages 676–679; July 29, 1994. interpreted. The organization in memory REM Sleep and the Reactivation of Recent Correlation Patterns in Hippocam- of this extensive infrastructure requires pal Neuronal Ensembles. H. S. Kudrimoto, W. F. Skaggs, C. A. Barnes, B. L. Mc- the additional REM sleep time. Naughton, J. L. Gerrard, M. S. Suster and K. L. Weaver in Society for Neuroscience Ab- For reasons he could not possibly have stracts, page 1871; 1996. known, Freud set forth a profound truth

The Meaning of Dreams Mysteries of the Mind 67

Copyright 1990 Scientific American, Inc.