may be crucial in mammalian memory processing. Important information acquired while awake may be reprocessed during TheMeaning ofDreams By Jonathan Winson

Throughout history, human beings have sought to understand the meaning of dreams. The ancient Egyptians believed dreams possessed oracular pow- state, a period of several minutes when thoughts consist of frag- er—in the Bible, for example, Joseph’s elucidation of Pharaoh’s mented images or minidramas. The hypnogogic state is followed averted seven years of famine. Other cultures have in- by slow-wave sleep, so called because at that time the terpreted dreams as inspirational, curative or alternative reality. waves of the (the convoluted outer mantle of the During the past century, scientists have offered conflicting psy- brain) are low in frequency and large in amplitude. These sig- chological and neuroscientific explanations for dreams. In nals are measured as electroencephalographic (EEG) recordings. 1900, with the publication of The Interpretation of Dreams, Researchers also discovered that a night’s sleep is punctu- Sigmund Freud proposed that dreams were the “royal road” to ated by periods in which the EEG readings are irregular in fre- the unconscious, that they revealed in disguised form the deep- quency and low in amplitude—similar to those observed in est elements of an individual’s inner life. awake individuals. These periods of mental activity are called More recently, in contrast, dreams have been characterized REM sleep. Dreaming takes place solely during these periods. as meaningless, the result of random nerve cell activity. Dream- While in REM sleep, motor are inhibited, preventing ing has also been viewed as the means by which the brain rids the body from moving freely but allowing extremities to remain itself of unnecessary information—a process of “reverse learn- slightly active. Eyes move rapidly in unison under closed lids, ing,” or unlearning. breathing becomes irregular, and heart rate increases. Based on recent findings in my own and other neuroscien- The first REM stage of the night follows 90 minutes of slow- tific laboratories, I propose that dreams are indeed meaningful. wave sleep and lasts for 10 minutes. The second and third REM Studies of the (a brain structure crucial to mem- periods follow shorter slow-wave sleep episodes but grow pro- ory), of rapid eye movement (REM) sleep and of a brain wave gressively longer themselves. The fourth and final REM inter- called theta rhythm suggest that dreaming reflects a pivotal as- val lasts 20 to 30 minutes and is followed by awakening. If a pect of the processing of memory. In particular, studies of theta dream is remembered at all, it is most often the one that oc- rhythm in subprimate animals have provided an evolutionary curred in this last phase of REM sleep. clue to the meaning of dreams. They appear to be the nightly This —alternating slow-wave and REM sleep— record of a basic mammalian memory process: the means by appears to be present in all placental and marsupial . which animals form strategies for survival and evaluate current Mammals exhibit the various REM-associated characteristics experience in light of those strategies. The existence of this pro- observed in humans, including EEG readings similar to those cess may explain the meaning of dreams in human beings. of the awake state. Animals also dream. By destroying neurons in the brain stem that inhibit movement during sleep, research- Stages of Sleep and Dreaming ers found that sleeping cats rose up and attacked or were star- THE PHYSIOLOGY OF DREAMING was first understood in tled by invisible objects—ostensibly images from dreams. 1953, when researchers characterized the human sleep cycle. By studying nonprimate animals, scientists have discovered They found that sleep in humans is initiated by the hypnogogic additional neurophysiological aspects of REM sleep. They de-

54 SCIENTIFIC AMERICAN Updated from the November 1990 issue COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. JACOB’S LADDER, painted in 1973 by Marc Chagall, depicts a biblical story. When Freud wrote The Interpretation of Dreams, the phys- Jacob dreams of angels ascending to and descending from heaven on a ladder. iology of sleep was unknown. In light of the discovery of REM sleep, certain elements of his psychoanalytic theory were mod- termined that neural control of this stage of the sleep cycle is cen- ified, and the stage was set for more neurologically based the- tered in the brain stem (the brain region closest to the spinal ories. Dreaming came to be understood as part of a biologically cord) and that during REM sleep neural signals—called pontine- determined sleep cycle. Yet the central concept of Freud’s theo- geniculate-occipital (PGO) cortex spikes—proceed from the ry—namely, the belief that dreams reveal a censored representa- brain stem to the center of visual processing, the visual cortex. tion of our innermost unconscious feelings and concerns—con- Brain stem neurons also initiate a sinusoidal wave (one resem- tinues to be used in psychoanalysis. bling a sine curve) in the hippocampus. This brain signal is called Some theorists abandoned Freud altogether following the theta rhythm. neurological discoveries. In 1977 J. Allan Hobson and Robert At least one animal experiences slow-wave but not REM McCarley of Harvard Medical School proposed the “activa- sleep—and, consequently, does not exhibit theta rhythm when tion-synthesis” hypothesis. They suggested that dreaming con- asleep. This animal is the echidna, or spiny anteater, an egg-lay- sists of associations and memories elicited from the forebrain ing (called a monotreme) that provides some insight (the neocortex and associated structures) in response to ran- into the origin of dreaming. The absence of REM sleep in the dom signals from the brain stem such as PGO spikes. Dreams echidna suggests that this stage of the sleep cycle evolved some were merely the “best fit” the forebrain could provide to this 140 million years ago, when marsupials and placentals diverged random bombardment from the brain stem. Although dreams from the monotreme line. (Monotremes were the first mammals might at times appear to have psychological content, their to develop from reptiles.) bizarreness was inherently meaningless. By all evolutionary criteria, the perpetuation of a complex The sense, or plot, of dreams resulted from order that was brain process such as REM sleep indicates that it serves an im- imposed on the chaos of neural signals, Hobson said. “That or- portant function for the survival of mammalian species. Un- der is a function of our own personal view of the world, our re-

SCALA/ART RESOURCE, ©2002 ARTISTS RIGHTS SOCIETY (ARS), NEW YORK/ADAGP, PARIS derstanding that function might reveal the meaning of dreams. mote memories,” he wrote. In other words, the individual’s

www.sciam.com THE HIDDEN MIND 55 COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. emotional vocabulary could be relevant ment by PGO waves and their own were a running record of these parasitic to dreams. In a further revision of the knowledge of the behavior of stimulated thoughts—material to be purged from original hypothesis, Hobson also sug- neural networks, Crick and Mitchison memory. “We dream to forget,” Crick gested that brain stem activation may postulated that a complex associational and Mitchison wrote. serve merely to switch from one dream neural network such as the neocortex The two researchers proposed a revi- episode to another. might become overloaded by vast amounts sion in 1986. Erasure of parasitic thoughts of incoming information. The neocortex accounted only for bizarre dream con- Reverse Learning could then develop false, or “parasitic,” tent. Nothing could be said about dream ALTHOUGH HOBSON and McCarley thoughts that would jeopardize the true narrative. Furthermore, dreaming to for- had presented an explanation of dream and orderly storage of memory. get, they said, was better expressed as content, the basic function of REM sleep According to their hypothesis, REM dreaming to reduce fantasy or obsession. remained unknown. In 1983 Francis sleep served to erase these spurious asso- None of these hypotheses seems to Crick of the Salk Institute for Biological ciations on a regular basis. Random PGO explain adequately the function of dream- Studies in San Diego and Graeme Mitchi- waves impinged on the neocortex, re- ing. On the one hand, Freud’s theory son of the University of Cambridge pro- sulting in erasure, or unlearning, of the lacked physiological evidence. (Although posed the idea of reverse learning. Work- false information. This process served an Freud had originally intended to describe ing from the Hobson-McCarley assump- essential function: it allowed the orderly the neurology of the unconscious and of tion of random neocortical bombard- processing of memory. In humans, dreams dreams in his proposed “Project for a Sci- entific Psychology,” the undertaking was premature, and he limited himself to psy- Prefrontal cortex choanalysis.) On the other hand, despite revisions to incorporate elements of psy- Septum chology, most of the later theories denied that dreams had meaning. Exploring the neuroscientific aspects of REM sleep and of memory processing seemed to me to hold the greatest poten- tial for understanding the meaning and function of dreams. The key to this re- search was theta rhythm. Theta rhythm was discovered in 1954 in awake animals by John D. Green and Arnaldo A. Arduini of the Universi- ty of California at Los Angeles. The re- searchers observed a regular sinusoidal signal of six cycles per second in the hippocampus of rabbits when the animals were apprehensive of stimuli in their en- vironment. They named the signal theta Visual rhythm after a previously discovered HIPPOCAMPUS EEG component of the same frequency. cortex Theta rhythm was subsequently re- Hippocampus corded in the tree shrew, mole, rat and Brain stem cat. Although it was consistently observed in awake animals, theta rhythm was cor- related with very different behaviors in each species. For example, in marked Spinal cord contrast to the rabbit, environmental CA1 cells CA3 stimuli did not induce theta rhythm in the cells rat. Rats demonstrated theta rhythm only during movement, typically when they explored. In 1969, however, Case H. ANATOMY OF THE BRAIN and cross section of the hippocampus show some of the regions involved in dreaming. In the hippocampus, incoming information is processed sequentially in the dentate Vanderwolf of the University of Western gyrus and the CA3 and the CA1 pyramidal cells (so named for their triangular shape). In nonprimate Ontario discovered there was one behav-

species, the theta rhythm brain wave is generated in the dentate gyrus and the CA1 cells. ior during which the animals he studied, CAROL DONNER

56 SCIENTIFIC AMERICAN THE HIDDEN MIND COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. 1 SEC

THETA RHYTHM

REM sleep REM sleep REM sleep

Exploration Apprehension

Predation

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

including the rat, showed theta rhythm: which theta rhythm was generated in the James B. Ranck, Jr., of the State Universi- REM sleep. hippocampus. Together with the neocor- ty of New York Downstate Medical Cen- In 1972 I published a commentary tex, the hippocampus is believed to pro- ter and his then co-worker Susan Mitchell pointing out that the different occurrenc- vide the neural basis for memory storage. identified a third synchronous generator es of theta rhythm could be understood The hippocampus (from the Greek word in the entorhinal cortex, and Robert in terms of animal behavior. Awake ani- for “seahorse,” which it resembles in Verdes of Wayne State University dis- mals seemed to show theta rhythm when shape) is a sequential structure composed covered the brain stem neurons that con- they were behaving in ways most crucial of three types of neurons. Information trol theta rhythm. These neurons trans- to their survival. In other words, theta from all sensory and associational areas mit signals to the septum (a forebrain rhythm appeared when they exhibited of the neocortex converges in a region structure) that activate theta rhythm in behavior that was not genetically encod- called the entorhinal cortex; from there it the hippocampus and the entorhinal cor- ed—such as feeding or sexual behavior— is transmitted to the three successive neu- tex. Thus, the brain stem activates the but rather a response to changing envi- ronal populations of the hippocampus. hippocampus and the neocortex—the ronmental information. Predatory be- The signal arrives first at the granule cells core memory system of the brain. havior in the cat, prey behavior in the of the dentate gyrus, then at the CA3 To determine the relation between rabbit, and exploration in the rat are, re- pyramidal cells (so called because of their theta rhythm and memory, I made a le- spectively, most important to their sur- triangular shape) and finally at the pyra- sion in the rat septum. Rats that had pre- vival. For example, a hungry rat will ex- midal cells of CA1. After information is viously learned, using spatial cues, to lo- plore before it eats even if food is placed processed by this trio of cells, it is re- cate a particular position in a maze were in front of it. transmitted to the entorhinal cortex and no longer able to do so after their septums then back to the neocortex. were disabled. Without theta rhythm, Role of Theta Rhythm My studies showed that theta rhythm spatial memory was destroyed. FURTHERMORE, because the hippo- was produced in two regions within the Studies of the cellular changes that campus is involved in memory process- hippocampus: the dentate gyrus and the bring about memory illustrated the role ing, the presence of theta rhythm during CA1 neurons. The rhythms in these two of theta rhythm. In particular, the dis- REM sleep in that region of the brain areas were synchronous. Subsequently, covery in 1973 of long-term potentiation might be related to that activity. I sug- gested that the theta rhythm reflected a JONATHAN WINSON started his career as an aeronautical engineer, graduating with an en- neural process whereby information es- gineering degree from the California Institute of Technology in 1946. He completed his Ph.D. sential to the survival of a species—gath- in mathematics at Columbia University and then turned to business for 15 years. Because ered during the day—was reprocessed of his long-standing interest in neuroscience, Winson then began research at the Rocke- into memory during REM sleep. feller University on memory processing. In 1979 he became associate professor there and

In 1974, by recording signals from THE AUTHOR continued his work as professor emeritus, retiring in 1996. His research was supported the hippocampus of freely moving rats by the National Institute of Mental Health, the National Science Foundation and the Harry

PATRICIA J. WYNNE and rabbits, I found the source from F. Guggenheim Foundation.

www.sciam.com THE HIDDEN MIND 57 COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. (LTP)—a change in neural behavior that Unlike other neuronal receptors, the rat is synchronized with theta rhythm, reflects previous activity—showed the NMDA possesses an additional property. as is the twitching of whiskers) and other means by which memory might be en- If a further activation of glutamate occurs sensory information converge on the en- coded. Timothy V. P. Bliss and A. R. while the granule cell is depolarized, a sec- torhinal cortex and the hippocampus. Gardner-Medwin of the National Insti- ond channel opens up, allowing an influx There they are partitioned into 200-mil- tute of Medical Research in London and of calcium. Calcium is thought to act as lisecond “bites” by theta rhythm. The Terje Lømo of the University of Oslo a second messenger, initiating a cascade NMDA receptors, acting in conjunction found changes in nerve cells that had of intracellular events that culminates in with theta rhythm, allow for long-term been intensely stimulated with electrical long-lasting synaptic changes—or LTP. storage of this information. pulses. (The description given here has been nec- A similar process occurs during REM essarily simplified. LTP is the subject of sleep. Although there is no incoming in- Long-Term Memory Storage extensive ongoing investigation.) formation or movement during REM EARLIER STUDIES had shown that if Because the tetanic impulses applied sleep, the neocortical-hippocampal net- one stimulated the pathway from the by Bliss and his colleagues did not occur work is once again paced by theta entorhinal cortex to the granule cells of naturally in the brain, the question re- rhythm. Theta rhythm might produce the hippocampus, the response of these mained as to how LTP was achieved un- long-lasting changes in memory. cells could be measured with a recording der normal circumstances. In 1986 John electrode. Using this technique, Bliss and Larson and Gary S. Lynch of the Univer- Storing Spatial Memory his colleagues measured the normal re- sity of California at Irvine and Gregory THE RESULTS OF ONE of my further sponse to a single electrical pulse. Then Rose and Thomas V. Dunwiddie of the experiments served to show that spatial they applied a long series of high- University of Colorado at Denver sug- memory was indeed being stored in the In a series of experiments, a coherent picture of MEMORY PROCESSING began to emerge.

frequency signals—called tetanic gested that the occurrence of LTP in the rat hippocampus during sleep. John pulses—to this pathway. After the train hippocampus was linked to theta rhythm. O’Keefe and Jonathan O. Dostrovsky of of tetanic stimuli, a single electrical pulse They applied a small number of electrical University College London had demon- caused much greater firing in the granule pulses to CA1 cells in the rat hippocam- strated that individual CA1 neurons in cells than had been observed prior to the pus and produced LTP, but only when the rat hippocampus fired when the experiment. The heightened effect per- the pulses were separated by the normal awake animal moved to a particular lo- sisted for as long as three days. This time that elapses between two theta cation—namely, the ’s place field. phenomenon of LTP was precisely the waves—approximately 200 milliseconds. The implication of this finding was that kind of increase in neuronal strength that Theta rhythm is apparently the natural the CA1 neuron fired to map the environ- could be capable of sustaining memory. means by which the NMDA receptor is ment, thereby committing it to memory. LTP is now considered a model for learn- activated in neurons in the hippocampus. In 1989 Pavlides and I located two ing and memory. Work in my laboratory at the Rock- CA1 neurons in the rat hippocampus that LTP is achieved by the activity of the efeller University duplicated Larson and had different place fields. We recorded NMDA (N-methyl-D-aspartate) recep- Lynch’s CA1 findings, but this time in the from both cells simultaneously. After de- tor. This molecule is embedded in the den- hippocampal granule cells. Constantine termining the normal firing rates in awake drites of the granule cells and the CA1 cells Pavlides, Yoram J. Greenstein and I then and asleep animals, we positioned a rat in of the hippocampus as well as in neurons demonstrated that LTP was dependent the place field of one of the neurons. The throughout the neocortex. Like other neu- on the presence and phase of theta neuron fired vigorously, mapping that lo- ronal receptors, the NMDA receptor is ac- rhythm. If electrical pulses were applied cation. The second cell fired only sporad- tivated by a neurotransmitter—glutamate to the cells at the peak of the , ically because it was not coding space. We in this case. Glutamate momentarily opens LTP was induced. But if the same pulse continued recording from the two pairs of a non-NMDA channel in the granule cell were applied at the trough of the waves— neurons as the rat moved about and then , allowing sodium from the extra- or when theta rhythm was absent—LTP entered several sleep cycles. Six pairs of

cellular space to flow into the neuron. This was not induced. neurons were studied in this manner. Photo Researchers, Inc. influx causes the granule cell to become A coherent picture of memory pro- We found that neurons that had cod- depolarized. If the depolarization is suffi- cessing was emerging. As a rat explores, ed space fired at a normal rate as the ani- cient, the granule cell fires, transmitting in- for example, brain stem neurons activate mal moved about prior to sleep. In sleep,

formation to other nerve cells. theta rhythm. Olfactory input (which in however, they fired at a significantly high- LABAT/JERRICAN

58 SCIENTIFIC AMERICAN THE HIDDEN MIND COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. mals that have that aspect of the sleep cy- cle and those that do not. And in fact, such differences clearly exist between the echid- na and the marsupials and placentals. The echidna has a large convoluted prefrontal cortex, larger in relation to the rest of the brain than that of any other mammal, even humans. I believe it need- ed this huge prefrontal cortex to perform Presynaptic a dual function: to react to incoming in- membrane Synaptic cleft formation in an appropriate manner Neurotransmitter Postsynaptic based on past experience and to evaluate Ion channel membrane and store new information to aid in future survival. Without theta rhythm during REM sleep, the echidna would not be able to process information while it slept. (The echidna does, however, show theta rhythm when foraging for food.) For higher capabilities to develop, the pre- Glutamate frontal cortex would have to become in- creasingly large—beyond the capacity of the skull—unless another brain mecha- nism evolved. Calcium REM sleep could have provided this new mechanism, allowing memory pro- cessing to occur “off-line.” Coincident with the apparent development of REM NMDA sleep in marsupial and placental mam- receptor mals was a remarkable neuroanatomical Sodium change: the prefrontal cortex was dra- SECOND RELEASE matically reduced in size. Far less pre- DEPOLARIZATION OF GLUTAMATE LTP frontal cortex was required to process in- formation. That area of the brain could NMDA RECEPTOR activation induces long-term potentiation (LTP), a model for memory. The release of develop to provide advanced perceptual the neurotransmitter glutamate (left panel) opens a non-NMDA (N-methyl-D-aspartate) receptor abilities in higher species. channel, allowing the influx of sodium, which depolarizes the neuron. If a further release of gluta- The nature of REM sleep supports mate occurs while the cell is depolarized (center panel), the NMDA receptor opens a second channel, which allows calcium to flow in, leading to LTP. LTP occurs as a result of increased sodium through this evolutionary argument. During the the non-NMDA channel (right panel) and the subsequent greater depolarization of the cell. day, animals gather information that in- volves locomotion and eye movement. er rate than their previous sleeping base- neurons that code space in the waking state The reprocessing of this information dur- line. There was no such increase in firing reprocess information during slow-wave ing REM sleep would not be easily sepa- rate during sleep in neurons that had not sleep and then in REM sleep. These results rated from the locomotion related to the mapped space. This experiment suggest- suggest that sleep processing of memory experience—such disassociation might be ed that the reprocessing or strengthening may have two stages—a preliminary stage expecting too great a revision of brain of information encoded when the animal in slow-wave sleep and a later phase in circuitry. So to maintain sleep, locomo- ) was awake occurred in sleep at the level REM sleep, when dreaming occurs. tion had to be suppressed by inhibiting

bottom of individual neurons. motor neurons. Suppressing eye move- Bruce L. McNaughton and his col- Evolution of REM Sleep ment was unnecessary because this ac- leagues at the University of Arizona have EVIDENCE THAT theta rhythm encodes tivity does not disturb sleep. developed a technique for simultaneously memories during REM sleep may be de- Eye movement potentials, similar to ); GABOR KISS ( recording from a large number of neurons rived not only from neuroscientific stud- PGO spikes, accompany rapid eye move- top in the hippocampus that map locations. ies but also from evolution. The emer- ment in the waking state and also during Their technique allows definitive patterns gence of a neural mechanism to process REM sleep. The function of these signals of firing to be identified. In animal studies, memory in REM sleep suggests differ- has not yet been established, but they

CAROL DONNER ( they found that ensembles of place-field ences in brain anatomy between mam- may serve to alert the visual cortex to in-

www.sciam.com THE HIDDEN MIND 59 COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. coming information when the animal is flect an individual’s strategy for survival. tions are strongly biased toward early awake and may reflect the reprocessing The subjects of dreams are broad-rang- childhood experience. of this information during REM sleep. In ing and complex, incorporating self-im- My hypothesis also offers an expla- any case, PGO spikes do not disturb age, fears, insecurities, strengths, gran- nation for the large amount of REM sleep sleep and do not have to be suppressed— diose ideas, sexual orientation, desire, in infants and children. Newborns spend unlike motor neurons. jealousy and love. eight hours a day in REM sleep. The sleep Dreams clearly have a deep psycho- cycle is disorganized at this age. Sleep oc- Strategy for Survival logical core. This observation has been curs in 50- to 60-minute bouts and begins WITH THE EVOLUTION of REM reported by psychoanalysts since Freud with REM rather than with slow-wave sleep, each species could process the in- and is strikingly illustrated by the work sleep. By the age of two, REM sleep is re- formation most important for its sur- of Rosalind Cartwright of Rush-Presby- duced to three hours a day, and the adult vival, such as the location of food or the terian–St. Luke’s Hospital in Chicago. pattern has been established. Thereafter, means of predation or escape—those ac- Cartwright studied a series of 90 subjects the time spent in REM sleep gradually di- tivities during which theta rhythm is pres- who were undergoing marital separation minishes to a little less than two hours. ent. In REM sleep this information may and divorce. All the subjects were clini- REM sleep may perform a special be accessed again and integrated with cally evaluated and psychologically test- function in infants. A leading theory pro- past experience to provide an ongoing ed to ascertain their attitudes and re- poses that it stimulates nerve growth. strategy for behavior. Although theta sponses to their personal crisis. Cart- Whatever the purpose in infants may be, rhythm has not yet been demon- wright’s subjects were also awakened I suggest that at about the age of two, Dreams MAY REFLECT a memory-processing mechanism inherited from LOWER SPECIES.

strated in primates, including hu- from REM sleep to report their dreams, when the hippocampus, which continues mans, the brain signal provides a which were then interpreted by the sub- to develop after birth, becomes function- clue to the origin of dreaming in humans. jects themselves without questions that al, REM sleep takes on its interpretive Dreams may reflect a memory-processing might have influenced their interpreta- memory function. The waking informa- mechanism inherited from lower species, tion. In 70 of the individuals studied, the tion to be integrated at this point in de- in which information important for sur- dream content conveyed the person’s un- velopment constitutes the basic cognitive vival is reprocessed during REM sleep. conscious thoughts and was strongly cor- substrate for memory—the concept of This information may constitute the core related with the manner in which he or the real world against which later expe- of the unconscious. she was coping with the crisis while riences must be compared and interpret- Because animals do not possess lan- awake. ed. The organization in memory of this guage, the information they process dur- Although the topic “chosen” for con- extensive infrastructure requires the ad- ing REM sleep is necessarily sensory. sideration during a night’s sleep is un- ditional REM sleep time. Consistent with our early mammalian predictable, certain of life’s difficulties— For reasons he could not possibly have origins, dreams in humans are sensory, as in the case of Cartwright’s subjects— known, Freud set forth a profound truth primarily visual. Dreams do not take the so engage psychological survival that in his work. There is an unconscious, and form of verbal narration. they are selected for REM sleep process- dreams are indeed the “royal road” to Also in keeping with the role REM ing. In the ordinary course of events, de- understanding it. The characteristics of sleep played in processing memories in pending on the individual’s personality, the unconscious and associated processes animals, there is no functional necessity the themes of dreams may be freewheel- of brain functioning, however, are very for this material to become conscious. ing. Moreover, when joined with the in- different from what Freud thought. Consciousness arose later in evolution in tricate associations that are an intrinsic Rather than being a cauldron of untamed humans. But neither is there any reason part of REM sleep processing, the dream’s passions and destructive wishes, I pro- for the material of dreams not to reach statement may be rather obscure. pose that the unconscious is a cohesive, consciousness. Therefore, dreams can be Nevertheless, there is every reason to continuously active mental structure that remembered—most readily if awakening believe that the cognitive process that takes note of life’s experiences and reacts occurs during or shortly after a REM took place in Cartwright’s subjects oc- according to its own scheme of interpre- sleep period. curs in every individual. Interpretation of tation. Dreams are not disguised as a Consistent with evolution and evi- the coherent statement that is being made consequence of repression. Their unusu- dence derived from neuroscience and re- depends on the individual’s tracing of rel- al character is a result of the complex as- ports of dreams, I suggest that dreams re- evant or similar events. These associa- sociations that are culled from memory.

60 SCIENTIFIC AMERICAN THE HIDDEN MIND COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC. Prefrontal EVOLUTIONARY TREE shows the divergence of cortex placentals and marsupials from monotremes. The echidna, which does not experience REM sleep, has a larger prefrontal cortex compared with the rest of its brain than does any mammal, even humans. It is larger than in similarly sized CAT OPOSSUM ECHIDNA animals, including the opossum and the cat. 0

REM processing in lower species must be Placentals Marsupials defined before the theory can be evaluat- Monotremes ed further. In addition, the Crick-Mitchi- son hypothesis as applied to the hip- 50 pocampus would suggest that neurons fire randomly during REM sleep, pro- viding reverse learning. Instead, in my ex- periment on the neurons that coded space, these neurons fired selectively, im- 100 plying an orderly processing of memory. Live- Avi Karni and his colleagues at the bearing Weizmann Institute of Science in Israel mammals were able to show that memory process- ing occurs in humans during REM sleep.

ars Ago (millions) ars Ago REM SLEEP In their experiment, individuals learned

Ye 150 to identify particular patterns on a Egg-laying screen. The memory of this skill im- mammals proved after a night with REM sleep. When the subjects were deprived of SLOW-WAVE SLEEP REM sleep, memory consolidation did 200 not occur. This study opens a promising First true mammals field for exploration. Perhaps of greatest interest is evi- dence supporting the role of REM sleep Mammallike reptiles in memory processing that has emerged 250 from molecular biology. Sidarta Ribeiro and his colleagues at the Rockefeller University have reported that the imme- Research on REM sleep suggests that dom brain stem activation has little ex- diate early gene zif-268 that is associat- there is a biologically relevant reason for planatory or predictive power. ed with learning is selectively upgraded dreaming. The revised version of the The Crick-Mitchison hypothesis pro- during REM sleep in rats exposed to ex- Hobson-McCarley activation-synthesis vides a function for REM sleep—reverse perience in a preceding waking period. hypothesis acknowledges the deep psy- learning—but it does not apply to narra- Further understanding of the role of chological core of dreams. In its present tive, only to the bizarre elements of the REM sleep may be expected from this truncated form, the hypothesis of ran- dream. What this implies with regard to area of research. SA

) MORE TO EXPLORE

chart Interspecies Differences in the Occurrence of Theta. Jonathan Winson Activity of These Cells during Subsequent Sleep Episodes. Constantine in Behavioral Biology, Vol. 7, No. 4, pages 479–487; 1972. Pavlides and Jonathan Winson in Journal of Neuroscience, Vol. 9, No. 8, Loss of Hippocampal Theta Rhythm Results in Spatial Memory Deficit pages 2907–2918; August 1989. in the Rat. Jonathan Winson in Science, Vol. 201, No. 435, pages Dependence on REM Sleep of Overnight Improvement of a Perceptual

); GABOR KISS ( 160–163; 1978. Skill. Avi Karni, David Tanne, Barton S. Rubenstein, Jean J. M. Askenasy Brain and Psyche: The Biology of the Unconscious. Jonathan Winson. and Dov Sagi in Science, Vol. 265, pages 679–682; July 29, 1994. Anchor Press, Doubleday, 1985. Reactivation of Hippocampal Ensemble Memories during Sleep. illustrations Long-Term Potentiation in the Dentate Gyrus Is Induced Preferentially Mathew A. Wilson and Bruce L. McNaughton in Science, Vol. 265, pages on the Positive Phase of Q-Rhythm. Constantine Pavlides, Yoram J. 676–679; July 29, 1994. Greenstein, Mark Grudman and Jonathan Winson in Brain Research, Brain Gene Expression during REM Sleep Depends on Prior Waking Vol. 439, pages 383–387; 1988. Experience. Sidarta Ribeiro, Vikas Goyal, Claudio V. Mello and Constantine

CAROL DONNER ( Influences of Hippocampal Firing in the Awake State on the Pavlides in Learning and Memory, Vol. 6, pages 500–508; 1999.

www.sciam.com THE HIDDEN MIND 61 COPYRIGHT 2002 SCIENTIFIC AMERICAN, INC.