Research Links Sleep, Dreams, and Learning
By William J. Cromie

In his office, alongside photographs of his wife and two sons, Jeffrey Sutton has a picture of his brain taken while he was dreaming. The assistant professor of psychiatry uses such cerebral images to answer questions about what our brain does while we sleep and dream.

For example, there is evidence that we learn while we sleep. Experiments have associated intense periods of daytime learning with longer periods of sleep that night, and particularly with dreaming. People awakened repeatedly from their dreams don't retain much of what they learned the day before.

"We see changes in the brain that may be caused by sleep-related learning," Sutton said. He referred to studies done by him and others in which people sleep in a magnetic resonance imaging (MRI) machine that takes pictures of their brain activity. At the same time, electrodes on their scalp and eyelids record eye movements that indicate dreaming.

"You scan people's brains before learning, then after sleep," Sutton explains. "The images let you look at how the brain reorganizes itself."

In other words, with the right technology it should be possible to see the brain learning.

Sutton's studies form part of a larger research effort in which computer models of the brain are tested by watching the brain at work, then using the resulting images to correct the models. "We expect this technique will reveal not only what happens in our brains when we sleep and dream, but what brain abnormalities correlate with disorders such as Alzheimer's disease, stroke, and depression."

Sleep or Die

No one knows all the purposes of sleeping and dreaming, although lack of sleep can be lethal. Sleep controls heat regulation and appetite. If you're cold and hungry, you won't dream much, if at all.

Sleep-deprived rats do okay for a week or two, then their appetites increase dramatically. Even when they get all they want to eat, their weights decrease, their body temperatures become unstable, and they die. Humans deprived of sleep hallucinate and behave abnormally.

Sleep rests the body but not the mind. MRI pictures show furious activity from the base of the brain to its wrinkled covering, the cortex, or thinking dome.

One theory holds that this excitement involves consolidation of information learned during the day. The process could include discarding what the brain considers junk mail, as well as making new connections between brain cells. Called unsupervised learning, the latter produces novel associations and thoughts. You often hear people say, "It came to me in a dream."

Sutton has watched the sleeping brains of about 15 people. During dreaming, he saw waves of activity starting in the brain stem, moving up through areas concerned with emotion and memory, then spreading over the cortex.

Nerve cells in the brain stem drive sleeping and dreaming by altering the balance of chemicals used to send and receive messages in the brain. The changes quickly travel to other parts of your head.

"The amygdala, an almond-shaped gland responsible for emotion, goes ballistic during dreams," Sutton says.

Nerve impulses also crackle in cerebral areas concerned with vision, memory, attention, and thought. All this activity is associated with anxiety, joy, anger, sadness, guilt, eroticism, time distortion, bizarre scenes, sudden shifts in subject, and incongruities.

Humans try to make sense of it all by constructing stories that string all these things together, albeit in wacky and weird ways. Sutton thinks such narratives may just be side effects of chemical changes that represent the real purposes of this nervous activity, such as learning and consolidating memories.

"Sleep deprivation impairs learning in humans and animals," Sutton says. Not just learning after sleep-lack, but before it. Rats make smarter moves when running a maze after a good night's sleep.

In one series of experiments, people tried to identify the position of objects that they saw quickly displayed on a screen. Researchers thought this skill would be learned immediately by repetition. But, in fact, subjects did better after a restful sleep. To investigate this surprise finding further, the researchers trained people in a repetitive task in the evening before they went to sleep. They then awoke some of them every time sensors on their eyelids showed them to be dreaming. These people retained little. In contrast, other subjects awakened during nondreaming sleep improved overnight.

How come? Studies by Sutton and others pin part of it on a powerful brain chemical called acetylcholine, which passes messages between brain cells. Acetylcholine promotes dreaming and has been implicated in memory consolidation during sleep. Allan Hobson, professor of psychiatry at the Medical School, found a substantial increase in the dreaming of cats when he injected the chemical into their brain stems.

Sutton and Hobson built a computer model that mimics brain changes during sleep and dream. Such a dream machine guides experimenters to pressing questions that need to be answered. The experiments, in turn, feed back new knowledge into the electronic brain.

Dreams To Diagnose Disease

Research on the dreamy role of acetylcholine may lead to a better understanding of Alzheimer's disease, which involves a disabling loss of memory and the ability to learn. Brain cells that produce this chemical are among the first to degenerate in Alzheimer's victims.

Michael Hasselmo, an associate professor in the Department of Psychology, has built a computer model to simulate Alzheimer's. Its learning and memory circuits change with variations in the availability of acetylcholine.

Sutton thinks that by integrating computer models and experimental results on such senility-simulating circuits, it might be possible to see changes that would predict who will get Alzheimer's. There's also the hope that such understanding will lead to better treatments for the disease.

Although such possibilities probably lie a long way in the future, they are not totally off-the-wall. Depression, for example, is linked with sleep disturbances. People suffering from it start to dream more quickly than those who do not. "The difference is likely due to an imbalance in brain chemicals, including too much acetylcholine and too little adrenaline," Sutton explains. Antidepressant drugs are designed to correct the imbalance.

Sutton believes that feedback between his brain machine and MRI pictures of the brain at work will provide more insight not only of depression and Alzheimer's, but of stroke, multiple sclerosis, and other disorders that affect large areas of the brain.

In one experiment, he and his colleagues looked at pictures of brains while their owners did simple motor tasks, such as tapping their fingers in simple and complex patterns. As expected, they saw activity in small networks of cells located in brain areas that control movements. Interestingly, the same type of brain arousal takes place whether people actually do finger tapping or only imagine it.

What surprised Sutton most, however, was detection of remarkably similar activity in much larger networks spanning areas of the cortex dealing with both input from the senses and output signals to the muscles.

"Patterns of activity in small, more primitive areas of the brain are recapitulated in larger, more advanced parts," Sutton says. "This means that nature did not have to develop new rules of operation for different levels of the brain from small clusters of cells to large systems."

In other words, as the brain evolved from a thimbleful of cells in a worm's head to the billions of cells with trillions of connections in humans, many of the same principles of organization were retained.

Those similarities make it infinitely easier to make computer models of the brain. "We already have built models which allow us to understand what is going on more quickly," Sutton notes. "Many types of mental illness may result from disorders of this organization. Understanding the details of what is happening will allow us to help real people with real suffering."

On a philosophical level, Sutton sees what he and others are doing as "using technology that works like our minds and brains to probe our minds and brains. We cannot get outside this loop, and that will always limit our understanding of ourselves. Our brains may never truly understand our minds."