Jonathan Winson is fascinated by dreams. To be specific, he is fascinated by the machinations of the brain that produce a dream’s evanescent images. But Winson hasn’t spent his professional years in an analyst’s armchair, listening to patients free-associate about their nocturnal fantasies. His domain has been a laboratory stacked to the ceiling with electronic equipment and laced with a whiff of animal odors.
Winson, at 69, is now leaving the hands-on part of his lab research to others. But just a couple of years ago he could still often be seen in his lab at Rockefeller University in New York fighting a quiet battle of wills with one of his subjects: a rat intent on crawling inside the sleeve of Winson’s tweed jacket while he gently restrained it. At those moments, Winson, with his mild gray eyes and grizzled raft of hair, seemed like a kindly uncle minding a restless child.
The rat was wearing what looked like a tiny pillbox hat. A thin ribbon of wire trailed from the hat to an overhead pulley, then ran across the ceiling and down to an electronic box a few feet away. The wire was picking up signals from a minute electrode implanted under the hat, in the rat’s brain. Every time a particular nerve cell, or neuron, fired, the electrode relayed the signal to the wire, which sent it to the electronic box, which recorded the event with a sharp click.
Using electrodes like this one to eavesdrop on the ephemeral conversations of neurons, Winson was testing an idea that had obsessed him for years--the idea that dreaming reflects a biological process by which the brain sifts through new information and incorporates it into its existing memory. The notion smacks of the obvious; we’ve all had dreams triggered by daytime events that inspire remembrances of things past. Yet what seems obvious can be notoriously hard to prove scientifically. After all, no one really knows why we sleep. As for why we dream, that’s always been the province of psychoanalysts and psychologists, not of physiologists.
By recording the activity of these rats’ neurons, however, Winson has provided the first neurological evidence that information from an animal’s waking hours is indeed reprocessed by the dreaming brain. Moreover, his findings may shed some light on why we dream periodically during sleep. Winson suspects that the brain uses these periods to carry out one of life’s more important tasks: to integrate new experiences with old ones and come up with a strategy for survival. The contents of dreams in early life, he says, reflect the building of a plan for behavior--a core plan that profoundly influences reaction to experiences in later life.
Winson is not the first to suppose there might be a fundamental link between dreaming and memory. Freud himself intuited the connection and even tried to derive a dream theory from neurobiological mechanisms. But given the paucity of knowledge about the brain at the time, the attempt fizzled, and Freud turned increasingly to psychological motives to explain dreams.
In Freud’s psychoanalytic view, dreams consist of infantile wishes and emotions that pop up in sleep. The ego, like a censor, normally represses these unacceptable feelings completely. But when the ego gets sleepy and relaxes its guard, the unconscious desires sneak through to manifest themselves in dreams. These feelings, however, are so upsetting that they would disrupt sleep if they appeared undisguised. So the censorial, if drowsy, ego cloaks them in cryptic symbols. That’s why Freud believed you could analyze the oddball contents of a person’s dreams to glimpse the workings of his unconscious mind and understand his psyche.
Yet despite its great influence, Freud’s theory has had its foundations steadily gnawed away by modern neuroscience. By the 1930s neurophysiologists were using the electroencephalogram to study electrical activity in the human cerebral cortex, the brain’s furrowed rind, which is the seat of perception and thought. Their recordings would show that most of a person’s night is spent in slow-wave sleep, marked by large, slow brain waves. The recordings of a waking person, in contrast, jitter in small, rapid waves.
In the early 1950s Eugene Aserinsky, a graduate student at the University of Chicago, pasted electrodes to his 10-year-old son’s face to record eye movements while the boy slept. Aserinsky discovered that at certain times of the night his son’s eyes skittered back and forth in unison. During such episodes of rapid eye movements, or REMs, breathing quickened and the heart beat faster. The muscles went limp and the body lay still, save for faint twitches of the extremities.
REM episodes had another oddity. Brain waves recorded at these times were small and fast, like those of a waking brain, not large and slow as was typical during sleep. Aserinsky and his adviser, Nathaniel Kleitman, suspected that the skittish eye movements coincided with dreams. Sleep labs around the world substantiated the hunch: when people were woken during REM sleep, 95 percent of the time they confirmed that they had been dreaming. Soon after this discovery another Kleitman protégé, William Dement, confirmed that REM sleep came in cycles, typically four or five times a night. Altogether, adult humans spent almost two hours a night dreaming.
Even more astounding revelations were to follow. Experiments with cats soon showed that neurons in the cerebral cortex hammered away during REM sleep, as though the animals were wide awake. The brain was clearly intensely active, yet it wasn’t getting any sensory input. Nor was it driving the body to move in response to its commands, as it would in waking animals. Why? The answer lay in the brain stem, the stalklike lower brain connecting the spinal cord to the cerebral cortex. There Michel Jouvet, a French researcher, found regions that functioned like some internal clock, periodically triggering REM sleep and its accompanying burst of brain activity; Jouvet also discovered a cluster of brain stem neurons that intercepted commands from the cortex to the spinal cord. When Jouvet destroyed those neurons, the slumbering cats got up, pounced on invisible mice, and arched their backs; they were acting out their dreams.
So Freud had it wrong. It wasn’t repressed desires but a neuronal clock in the brain stem that sent the brain into a dream state four or five times a night. Based on such findings, psychiatrist Allan Hobson of Harvard proposed in 1977 a radical new theory of dreams. It was the brain stem’s random excitement of the cortex, he said, that accounted for the hallucinatory bizarreness of dreams. Our dreams were the vivid by-products of a purely physiological process.
Hobson, now 59, is an animated man with riveting blue eyes and snow white hair that spills over his shirt collar. His office at the Massachusetts Mental Health Center in Boston is cluttered with photos, mementos, and images of brains, among them several ink drawings of brain cells resembling leafless shrubs. Hobson’s fingers caress a delicate ink drawing of nerve-cell bodies overlaid by the dark branches of their outstretched axons. Some of these may have been by Cajal himself, he says, referring to the great Spanish neuroscientist Santiago Ramón y Cajal, a contemporary of Freud’s.
Dreams are bizarre, Hobson argues, not because the Freudian ego disguises hidden wishes, but because neurons in the brain stem bombard the visual cortex with random signals as the eyes dart about during REM sleep. These signals, called PGO spikes, apparently convey information about the direction of the moving eyes, says Hobson, but the brain tries to interpret them as real visual data.
So according to our theory, explains Hobson, the visual system gets a barrage of signals, and it says, What’s this? And it goes into its memory stores, looks for a match, and says, Well, that’s sort of like this high rise I was in. What was going on there? Well, so and so was there. And there’s a story generated to go with that notion. Then a new set of signals arises, incompatible with the previous data set, and the story changes: Oh, now it’s my grandmother’s house.
It’s this barrage of brain stem signals, says Hobson, that directs the images and abrupt scene shifts in dreams. Dreams are our awareness of this organic brain activity, which probably serves a variety of purposes. Among other things, Hobson speculates, REM sleep may allow us to rev our cerebral motor and actively test all our circuits in a reliably patterned way.
Yet brain stem signals, as even Hobson agrees, are not the whole story. The REM state is generated by the brain stem and transmitted to the cortex, but the cortex feeds back, says Barbara Jones, a brain stem expert at McGill University in Montreal. If you remove the cortex, the PGO spikes become very much simplified. You can’t say information flows in just one direction. There’s a circuit.
This evidence suggests that a higher cortical process may help orchestrate whatever goes on in the brain once REM sleep begins. But what could that be? That’s precisely the question that Winson has seized on. He believes that sleep--and in particular, REM sleep--triggers the reprocessing and consolidating of daytime information into memory.
Winson thinks the key to unlocking the mystery of REM sleep and dreams lies in a part of the brain called the hippocampus (from the Greek word for sea horse, which it resembles). We have a pair of these six-inch- long structures, one buried on each side of the brain where the cortex buckles inward at the temples, and they are crucial to memory. This discovery was made inadvertently in the early 1950s when a neurosurgeon tried to subdue a patient’s intractable epilepsy by slicing out the brain tissue--including the hippocampus--where his seizures ignited. The operation left the patient unable to form new memories. After a moment’s absence, he could no longer recognize people he had conversed with for hours.
Around that same time researchers discovered that during certain activities the hippocampus produced a distinctive rhythm, which became known as theta rhythm. The activity, recorded by an electrode and displayed on a paper tracing, showed up as alternating peaks and valleys of voltage, six peaks per second. Rabbits, cats, and rats, it was found, all generated theta rhythm when exploring a strange place. Rabbits also produced it when startled by a predator, and cats when they were stalking prey. The common denominator of these behaviors seemed to be that all were important to the animals’ survival.
Then, in 1969, theta rhythm was found during another behavior that these mammals shared: REM sleep.
Back in 1969, Winson hadn’t yet embarked on his career in neuroscience. Soon after he’d received his Ph.D. in mathematics from Columbia, his father had become ill, and Winson had taken over the family manufacturing business. But he had long been intrigued by the way the brain processes memory, and this discovery linking theta rhythm to REM sleep sparked an epiphany. The hippocampus is central to memory, he recalls thinking. And here’s this theta rhythm occurring in the hippocampus during activities that are important for the animals’ survival. And now here it is again in REM sleep. I said to myself, Something’s got to be going on.
In a mid-life career switch, Winson became a guest investigator in a neuroscience lab at Rockefeller. His studies of rats in the early 1970s demonstrated that theta waves were produced in two regions of the hippocampus. Other researchers found theta waves generated in the entorhinal cortex, a staging area for information entering and leaving the hippocampus.
Winson wondered what would happen if he knocked out the neurons, located in an adjoining part of the brain, that pace the oscillations, thereby silencing the theta rhythm without harming the hippocampus itself. When he did this, he found that rats that had learned to use spatial cues to find a spot in a maze could no longer do so. Without theta rhythm, their spatial memory was obliterated.
Theta rhythm was evidently essential for hippocampal function, which in turn was essential for memory. But how was theta rhythm involved in integrating and utilizing memories? To answer that question, one needs to address a fundamental riddle: How is memory formed? How is a friend’s face, an episode of L.A. Law, or a calculus formula captured inside our heads?
Most neuroscientists think memories are encoded over a sprawling network of neurons in the brain. According to the going theory, a novel experience creates a pattern of firing in the network. Later, a reminder of that experience, or an attempt to recall it, triggers the same firing pattern. The pattern represents the memory. How is it stored? Neurobiologists believe a physical change at the synapses--the junctions where one neuron communicates with another--is responsible. As neurons fire in a pattern, their synaptic connections become strengthened to conduct the signals more readily, thereby etching the pattern more firmly into the network.
Compelling evidence for such strengthening was first found in 1972. Researchers discovered that if they buzzed a nerve pathway in the hippocampus with a rapid tattoo of electric pulses, neurons in the pathway fired more readily upon subsequent stimulation. The synaptic connections did indeed appear to have been strengthened. This change, called long-term potentiation, or LTP, has become the working model for how memories are stored.
Of course, the LTP observed in 1972 was caused by a prolonged and artificial pattern of electric pulses, quite unlike anything known to occur in the brain. But in 1986 some researchers suggested there might be a natural stimulus for LTP, at least in lower mammals like rats. That natural stimulus, they said, could be theta rhythm. In experiments on the rat hippocampus, electric pulses produced LTP most effectively when they were applied at the theta rate.
What kind of information is etched into memory by theta-rhythm- induced LTP? In rats, Winson observes, theta rhythm is perfectly synchronized with whisker movement, sniffing, and the firing of neurons in the olfactory bulb as the animal explores its environment. Could theta waves be the agency through which such sensory information is stored in long-term memory? When Winson’s group applied short electric pulses to the hippocampus at the peak of theta rhythm--simulating the way cells normally fire in the hippocampus in response to sensory stimuli--they found that they did indeed induce LTP.
Now the pieces of the puzzle were starting to fit together. As a rat explores its surroundings, theta waves wash over its hippocampus. At the same time, sensory signals from its whiskers and nose burst out in synchrony. These signals coincide with the pulsing of theta waves in the hippocampus and activate LTP, leaving a memory trace. Later, in REM sleep, theta waves turn on again and reactivate those neuronal circuits, allowing the memory trace to be reaccessed and integrated with old memories.
It was a nifty hypothesis. But the evidence was still circumstantial. To help clinch his case, Winson needed to prove that new information actually gets reprocessed during REM sleep. The problem was how to do the experiment. It may take thousands, perhaps even millions, of neurons to represent the memory of a single event, yet an experimenter can monitor only a very few using implanted electrodes. How could Winson be sure that the activity of a neuron corresponded to a specific piece of information?
About four years ago Winson and his then graduate student Constantine Pavlides came up with a solution. Rats have rather unusual neurons in the hippocampus known as place neurons, which encode a rat’s brain map of its physical space. When a rat runs around an open maze in a lab, for example, it orients itself by various landmarks in the room--a wall clock, a window--and each place neuron becomes responsive to a unique locale. Winson and Pavlides realized they could monitor the processing of spatial information in a rat by recording from just one place neuron. If the waking animal traversed a location that made the place neuron fire, the same neuron should later fire energetically during REM sleep--assuming, of course, that their theory was correct.
As a first step they carried out an exploratory test to map the location on the maze that would excite the neuron they were recording. They placed one of their pillbox-wearing rats at the center of a maze that had eight arms radiating out like the spokes of a wheel. The maze looked like a giant stationary fan upended at the top of a pole, giving the rat a panoramic view of its surroundings. The rat immediately began to explore. It pattered up and down one arm--whiskers twitching, sniffing, observing its environment--then returned to the maze’s center and pattered down another arm. Just as it crossed one spot, a barrage of clicks sounding like a toy machine gun broke out from the recording equipment. Moments later, as the rat continued onward, the sound died away. But when the rat doubled back and recrossed the spot, the machine gun noise chattered out again. The electrode had tuned into a place neuron for that spot on the maze.
Pavlides next blocked access to the spot and let the rat wander elsewhere in the maze. The monitor picked up only sporadic signals from the neuron, which was just ticking as the rat went about its business.
Now came the crucial part of the experiment. After a while the rat curled up to snooze. A second probe broadcast theta waves as the animal entered REM sleep. The place neuron stirred and broke out in a burst of signals--over and over again.
Many months and several rats later Winson and Pavlides combed through the data. Winson’s eyes twinkle with pleasure as he recalls the outcome. A consistent pattern had emerged. Each time a rat hit its trigger spot in the maze, the place neuron fired rapidly. Then the rat moved on to the rest of the maze, and the neuron quieted down. But during REM sleep the place neuron hammered away again, at the rapid rate that’s effective for inducing LTP. It was a very compelling result.
The place-neuron experiment provided the first direct evidence that the brain is reprocessing daytime information during sleep. But why does the brain go to the trouble? Why doesn’t it just process everything at once while it’s awake?
Evolution suggests a reason. REM sleep appears late in evolutionary history. Only mammals have it, and, with one documented exception, every terrestrial mammal has it. The curious exception is the echidna, or spiny anteater. This small Australian animal, which looks like an overfed hedgehog with a beak, is a monotreme, the most primitive kind of mammal--so primitive that it lays eggs, like a reptile. In addition to lacking REM sleep, the echidna is exceptional in one other respect. Its prefrontal cortex is huge, larger relative to the rest of the brain than that of any other mammal, humans included.
The brain of the echidna, Winson explains, has to perform two functions at once. It has to react to any new environmental challenge based on its previous experience, and it has to update that strategy with whatever is new in that experience. So nature provided it with an oversize prefrontal cortex, the part of the brain where, it’s believed, survival strategies are devised and stored.
But the echidna’s large prefrontal cortex amounted to an evolutionary impasse. Mammalian evolution could not proceed further, because there wasn’t enough room in the skull to accommodate more brain tissue. To get past that, higher mammals had to come up with a method of using brain space more efficiently--namely, REM sleep. REM sleep, Winson proposes, lets the brain reprocess information taken in during the day so it can get more done in a limited space. If our brain didn’t use this off- line scheme, our prefrontal cortex would have to be so big we’d need a wheelbarrow to trundle it around. Indeed, if nature hadn’t hit upon REM sleep, Winson likes to say, we would never have evolved.
So what does this brain activity imply about dreams, the nocturnal images that flit through our sleeping minds? Freud believed that dreams were driven by unconscious desires. But according to Hobson, dream images are instigated by random signals shooting up from the brain stem to the visual cortex and other parts of the forebrain. The classic flight dream, he speculates, is more likely evoked by an uprolling of the eyeballs during REM sleep than erotic impulses, as Freud would have it. If dreams reflect psychological concerns, that’s because the cortex then tries to make sense of this random neuronal activity and its interpretation may reveal something about a person’s psychological state. It’s like interpreting a physiological Rorschach inkblot. We try to read meaning into the inchoate inky smudge, but its shape (like the brain activity of dreams) is random.
Dreams, Hobson concludes, dealing his coup de grace to Freudian theory, are not a covert manifestation of hidden feelings. No disguise, no censorship, he exclaims. Trying to interpret the bizarre, incongruous elements in dreams is like attributing symbolic content to the utterings of a person with Alzheimer’s disease! You’re trying to account psychodynamically for a process that is organic.
Dreams are not disguised, Winson concurs. On that point both researchers agree. But on the question of their meaning, Winson parts company with Hobson. Winson’s experiments in the hippocampus have convinced him that the contents of our dreams are significant. I’m saying the information chosen for reprocessing is not random, he asserts. In other words, the contents of dreams are not random.
Some dream studies certainly seem to support this idea. In the late 1970s one experiment had student volunteers outfitted with goggles that made the world appear red. With each successive night, the students’ dreams became systematically more red. Even dreams of events that took place long before they ever wore goggles sometimes took on a red tint, indicating an integration of newer and older information.
Winson also believes that the contents of our dreams can be symbolic. He suspects that dream symbols arise from the associative quality of our memories. This is the quality that enables us to link the form of a rose not only to its perfume, hue, velvet texture, and spiky thorns-- information that smites our senses in one moment of time--but also to Valentine’s Day and the sayings of Gertrude Stein, information we accrue in scattered episodes. No one yet knows for sure how our brains make such associations. But presumably, since memories are linked in our minds, they must also be physically linked in the neuronal networks in our brains.
Our dreams, Winson thinks, tend to be stated in visual images and scenarios--that is, in symbolic form--because the memory mechanism involved is so ancient, inherited from lower mammals before language evolved. As a result, abstract concepts can only be expressed in images, not represented by words. What’s more, he points out, a symbol permits several concepts to be compressed into one image.
So, for example, you might dream that you are late for an exam and running down a long hall, opening one door after another, trying to find the examination room. This is the kind of dream we tend to have when we’re worried about meeting a deadline for a report or a delivery date for a shipment. According to Winson’s theory, the sleeping brain searches around in its memory networks and lands on an image that expresses several ideas--frustration, fear of failure, desire to succeed, anxiety--in a concise and undisguised manner. But there may be a bizarre component in that dream as well. Perhaps you’re wearing a funny outfit as you race down the hall, a child’s outfit reminiscent of some other time when you experienced similar feelings. So the precise meaning of any dream, even a classic anxiety dream, can only be gleaned from the context of your life and memory associations.
Winson’s view, in other words, doesn’t offer a new interpretation of dreams so much as it provides a biological and evolutionary rationale for the phenomenon of dreaming. In a similar spirit, he thinks REM sleep may offer a way to account biologically for the formation of the unconscious. Psychoanalytic studies, for example, indicate that the unconscious mind is shaped in early childhood. Winson notes that our neural circuits are quite plastic in early life, when we must rapidly build our cognitive framework--a core of knowledge that we’ll use to integrate subsequent experiences and interpret them for future reference. This process is most intensive in young children. Two-year-olds spend three hours a night in REM sleep, a third more time than adults do. In children, Winson suspects that much of this time is spent building up coping strategies to get them through life.
All of this is quite speculative, of course, he says, and deftly changes the subject. My main interest is to clarify how memory is processed in REM sleep.
Winson’s eyes regain their twinkle as he describes the plan of his latest experiment. In his 1978 experiment to gauge the importance of theta rhythm, Winson wiped out the rhythm completely. As a result, rats were deprived of theta both as they explored the maze and during REM sleep. But in 1990, he relates, Robert Vertes, of Florida Atlantic University at Boca Raton, pinpointed a part of the brain stem that drives theta rhythm exclusively during REM sleep. And that has enabled Winson and Vertes to design a maze experiment to see what happens to the memory of rats when only those neurons are knocked out.
If Winson is on target, these rats will still experience theta waves when awake and thus have no difficulty exploring a maze. But without theta in REM sleep, the processing of these daytime memories will be impossible. A series of mnemonic tests comparing these animals with control animals will then start to tease out just how this process works. Whatever the outcome, science will come one step closer to knowing whether we are, after all, such stuff as dreams are made on.
