The Sleeping Brain

Sorting, moving, arranging, mixing -- there's a lot of essential cognitive processing going on while you snooze.

Once upon a time, people fell asleep and that was it. They went to bed every night, and they were blank as toadstools for the next eight hours. If you saw them at breakfast and asked them what they'd been up to since yesterday, they would likely say, "Nothing--I've been asleep," and go back to munching their cornflakes. This conversation went on, virtually unchanged, for 5,000 years.

Sleep is the great unknown sea where we go diving for one third of our lives. It has generally been considered a dormant period, an essential time-out for our bodies and minds during which nothing much happens. Historically, from a medical perspective, sleep has been viewed much the same way, as a resetting of our internal clocks, or a random string of z-z-zs, or a nullity between waking hours. But that perception is starting to change. Based in part on research conducted by Jeffrey Ellenbogen, '00, a pioneering neurologist in the Boston area, sleep is being reconsidered as a time of essential brain activity that, silently and beyond our knowing, enables us to make sense of the world we inhabit.

"This is a very exciting time for the study of sleep," says Ellenbogen when tracked down in his small, neat office at Massachusetts General Hospital, where he works as an assistant in neurology and director of the Sleep Medicine Program. "In the last five or 10 years, we've seen a lot of data suggesting that the old idea that sleep is an inactive state represents an obsolete concept. We have learned that the brain is doing lots of things during sleep, including consolidating memory." Ellenbogen views sleep as one of the last big medical frontiers -- in effect, a field as tantalizing, vast and unplumbed as stars in the night sky. "We don't know why people sleep at all, let alone why they do it so long," he points out. Contrast the depth of that mystery, he continues, with the easy medical comprehension afforded other body functions like the beating of our hearts. "We understand the heart perfectly," Ellenbogen argues. "No one says, 'Why does our heart beat, or why do we have skin?' We take that knowledge for granted. That's not to say that there aren't plenty of exciting things to learn in cardiology or dermatology. But the big-picture questions are not enigmatic in the way that sleep is."

A number of obvious difficulties have hindered sleep research. The brain, where the work of sleep goes on, is virtually inaccessible to examination. It's tucked inside the cranium and susceptible to grievous damage if not handled just so. Then there's the irreducible complexity of the wrinkled mass itself. Ellenbogen notes that the human brain contains approximately 100 billion neurons, with each neuron containing tens of thousands of interconnections. As he remarks, with a look of mild wonderment, "It's a pretty complicated place."

It wasn't until the 1950s that scientists discovered that there were different kinds of sleep, including REM and non-REM phases. These were the barest outlines of the mystery of the sleeping brain.More recently, the tools of functional magnetic resonance imaging (fMRI), PET scans and electrocephalography (EEG) have been used to give researchers a glimpse into the cognitive activity that occurs whenever we hit the hay. At this point, the research is more suggestive than it is complete. Even so, enough has come to light to demand that the old notion of sleep as a mental blank period be replaced with something new -- something more akin to a cocktail party than a heavy snow.

A Dynamic Process

In a recent test at the university of Arizona that was reported in Science, a colony of rats gave the first hint of how focused and relentless the mind can be when the body is sound asleep.

The experiment was simple. With electrodes attached directly to their brains, rats were placed in a maze and closely observed. Their brains gave off a distinctive electrical firing pattern as they sought to navigate the maze. Later that day, when the rats were asleep, the hippocampus portion of their brains kept firing in exactly the same pattern, over and over again. It was as though the rats were studying the maze in their sleep in order to master it, Ellenbogen says.

One of the first signs of some comparable "off-line" activity occurring in humans came about in tests of motor skills -- what are known as procedural memories -- conducted by Matt Walker, Ph.D., a sleep researcher at Boston's Beth Israel Deaconess Medical Center. In one experiment, Walker would ask his subjects to learn a finger-tapping sequence by practicing it in segments and then combining the parts as best they could into a single rhythm. These performances were generally choppy. "But when you brought the subjects back after a night of sleep," Walker reports, "it was as though those problem points had been smoothed out." By implication, the benefits of a good night's sleep would likely accrue to pianists attempting to master a tricky passage in a concerto, baseball players struggling to hit a curveball or kids learning how to ride a bike.

Declarative memories, or memories about facts and events that can generally be put into words, pose another kind of challenge for the human brain. Ellenbogen examined this phenomenon last year, when he was a fellow in sleep medicine at Brigham and Women's Hospital. The results appeared in the July 11, 2006 issue of Current Biology. Before this study, no one had been able to show definitively that sleep promotes the strengthening of declarative memories in humans.

Ellenbogen's team created two lists of 20 randomly paired words and presented the lists to 48 men and women, who were asked to memorize them. Participants were assigned to one of five groups. One group memorized the words in the evening and was tested the next morning; one memorized the words in the morning and got tested that evening; one memorized in the evening and was interrupted with another task before being tested in the morning; one memorized in the morning and was given another memory task before being tested in the evening; and, finally, one memorized in the evening and was given another memory task before being tested the next evening.

Those who went to sleep without an interfering task did best, with 94 percent recall of the word pairs. Those who stayed awake without interference scored an average of 82 percent. A night of sleep, however, made a dramatic difference for the groups that did distracting tasks. Those in this category who stayed awake scored 32 percent, while those who had a chance to sleep before their test scored 76 percent.And even if the sleep participants went the entire next day before testing, they still scored much better than the group who learned and tested on the same day.

Sleep played an active role in determining the outcome, the researchers concluded. "Rather than being a passive state, sleep is a dynamic neurobiological process," Ellenbogen said at the time. "It turns out that the process of memory doesn't end when we stop studying, but continues during sleep."

Six Colored Ovals

What were the dimensions of this unconscious process? The question arose naturally.

One day last spring, Walker and Ellenbogen sat down and put their heads together to see if they could somehow measure the value of sleep in a mental process known as relational memory -- the ability to derive "big-picture" inferences from disparate pieces of information. Relational memory is essential to our ability to make sense of the world around us, because the world is, of course, always coming at us in scattered bits and pieces that we must assemble into patterns and meanings.

Walker gives an example from a road atlas. "Let's say I learn that you take I-95 to get from Portland, Maine, to Boston," he suggests, "and then I also learn that I-95 connects Boston to New York. When, some time later, I want to know how to get from Portland to New York, I can answer the question -- but how do I know this, exactly? Without relational memory, it's as if you get individual building blocks, but you haven't built the house." To cite another example in terms familiar to everyone from high school geometry class: A is greater than B, which is greater than C; therefore A is greater than C. The technical name for this deductive process is "transitive inference."

"People often assume that we know all of what we know because we learned it directly," Ellenbogen explains. "In fact, that's only partly true.We actually learn individual bits of information and then apply them in novel, flexible ways." Put another way, we somehow manage to draw generalizations from not-so-obviously related details.

For proof of the process, Ellenbogen and Walker needed to devise a test that included a hierarchy of relationships yet was expressed in an abstract form devoid of associations or bias for those taking the test. (If they tried car models, for example, the test results stood a chance of being skewed by a test subject's historic preference for, say, Fords over Chevrolets, or sports cars over sedans.) In a nifty move, the two men cooked up a series of six colorful ovals, each with its own subtly permutated design, capable of being displayed on a computer screen. If they were on the right track, these "fractiles"would offer about as pure a test of transitive inference as could be conducted in a lab.

They then recruited 56 college students and put them through a brief training period to teach them the ranking of the ovals. Students would learn that "red always wins over blue," or "green wins over yellow," for example. After about 20 minutes of Walker contends that "sleep not only strengthens a person's individual memories, it appears to actually knit them together and help us realize how they are associated with one another. And this may, in fact, turn out to be the primary goal of sleep -- to open the aperture of memory, to say to memories, 'Let's get in the same room and become acquainted.' "

Jeffrey Ellenbogen worked with Matt Walker on the problem of measuring transitive inference. training, the students were shown pairs of ovals in sequence, and each time asked to identify the higher-ranked of the two -- not just on the pairs they had already seen, but on never-before-seen pairs as well. This was no easy task. An intelligent person, acting consciously and deliberately, might study the ovals until the cows came home without being able to say for sure which outranked the other. Although a hidden hierarchy linked all the ovals, this was never explained to the test subjects.

The students were separated into three groups for the test. Group One was tested after a period of 20 minutes. Group Two was tested after a 12-hour period, and Group Three after 24 hours had elapsed. Approximately half of the students in Group Two slept during the 12-hour period, while the other half remained awake. All members of Group Three had a full night's sleep.

Subjects who were tested shortly after the initial learning period performed, as a group, no better than chance, getting about half correct. "Group One performed the worst," Walker recounts. "While they were able to learn and recall the component pieces (for example, Shape A is greater than Shape B, and Shape B is greater than Shape C), they could not discern the hierarchical relationships between the pieces (e.g., Shape A is greater than Shape C). They couldn't yet see the 'big picture.' "

In contrast, the groups tested after a period of at least 12 hours enjoyed a much higher success rate, identifying nearly 80 percent of the inferences correctly. Those tested after a full night's sleep were able to draw still more distant connections in the scheme of ovals.

"Even we were a little taken aback by the size of this benefit," says Ellenbogen. The findings appeared in the May 1, 2007, issue of Proceedings of the National Academy of Sciences. In commenting on the relative success shared by members of Groups Two and Three,Walker says. "These individuals were able to make leaps of inferential judgment just by letting the brain have time to unconsciously mull things over."

"This strongly implies that sleep is actively engaged in the cognitive processing of our memories," Ellenbogen notes. "Knowledge appears to expand both over time and with sleep."

Walker contends that "sleep not only strengthens a person's individual memories, it appears to actually knit them together and help us realize how they are associated with one another. And this may, in fact, turn out to be the primary goal of sleep -- to open the aperture of memory, to say to memories, 'Let's get in the same room and become acquainted. Let's have a group therapy session.' In effect, you go to bed with pieces of the memory puzzle, and awaken with the jigsaw completed."

Ripe for Investigation

As huge as the implications are for medicine and society, it's stunning to learn that sleep is being studied neurologically in such few places. Ellenbogen puts the number of labs that are examining the cognitive benefits of sleep in a manner akin to his and Walker's at a dozen or so, worldwide -- "which is a drawback," he says, "since sleep is neurology- based. It is fundamentally a brain process." Other labs are more concerned with issues like the effects of sleep deprivation on immune function and metabolic control, he says.

Medical technology is evolving to paint the neurology of sleep in more detail, but for the moment, the tools of fMRI and EEG, used together, can give only an approximate sense of what happens mentally as we doze. Through electrodes attached to a test subject's scalp, EEG is capable of tracking electrical activity within the brain second by second, providing a record of distinctive patterns of thought through its jagged peaks and valleys. EEG offers excellent temporal resolution but poor spatial resolution. It can show when something occurred in the brain, but not precisely where. In contrast, the fMRI has great spatial resolution but poor temporal resolution. For brain research, the strength of fMRI is that it can tell precisely where an electrical pattern occurs. Studies conducted at Walker's lab in California, where he is now director of UCBerkeley's Sleep and Neuroimaging Laboratory, indicate that during sleep, our memories migrate from portions of the brain involved in conscious control, such as the parietal cortex, to more non-conscious, automated regions including the motor cortex and cerebellum. "When we come back the next day, the memories are in a different place, and they have been changed in the process," says Walker. Sleep affects everything in our bodies. Accordingly, it may offer a window into countless human ailments, from epilepsy to psychiatric disorders -- which invariably display an abnormality of sleep -- to mental confusion in old age.

Dr. Beth Malow, director of the Vanderbilt Sleep Disorders Center at Vanderbilt University, has found that seizures can be reduced in her epileptic patients by providing them with a better quality of sleep."This whole area of neurological diseases and sleep is one that's ripe for investigation," she says. One of Ellenbogen's chief clinical interests these days pertains to older adults who are having trouble sleeping well. Senior citizens often suffer from faulty memories and poor-quality sleep at the same time. Could these two phenomena be directly related, and might sleep be part of the cure?

Day and night, the work goes on.