Exploring the Relationship Between Sleep, Dreams, and Memory

MINNEAPOLIS—Sleep benefits the most difficult learning tasks and helps extract the most important information from the brain, according to Robert Stickgold, PhD, who presented his view on the role that sleep plays in processing memories at the 21st Annual Meeting of the Associated Professional Sleep Societies.

“There are many, many different systems within the brain that are used for the storage and processing of information,” said Dr. Stickgold, Associate Professor of Psychiatry at Harvard Medical School in Boston. “Sleep acts to consolidate and strengthen specific memories by creating new ways of identification and creation of associations among them,” including specific details. This requires physiologic processing using a variety of distinct brain states, which may correspond with stages of sleep. “A strong hypothesis would be that varied stages of sleep originally evolved to meet the varying demands of the memory system and enhance understanding,” he added.

Within the past decade, an explosive growth in knowledge of the relationship between sleep and memory has occurred. Research conducted by Dr. Stickgold and colleagues has focused on slow-wave sleep, which precedes and follows REM sleep and is believed to be critical for memory consolidation. Slow-wave sleep has four distinct stages, from drowsiness to progressively deeper sleep. These different stages are believed to be associated with procedural learning tasks, such as learning how to play checkers or whether to take an umbrella if it looks likely to rain.

To test the theory that sleep is involved in consolidating procedural memory, Dr. Stickgold and Matthew P. Walker, PhD, a former colleague at Harvard who is now the Principal Investigator of the Sleep and Neuroimaging Laboratory in the Department of Psychology at the University of California, Berkeley, conducted experiments in which participants were trained to tap nine-digit sequences—the second-to-last digit of which was always the same—and then allowed to sleep. After a night of sleep, the volunteers were two-and-a-half times more likely to recognize the sequence pattern.

“What’s getting better on this test is a complicated question,” Dr. Stickgold said. “While we sleep, the brain sifts through and integrates all that information,” extracting information not available before we go to sleep. “[Participants] get better, but they get better differently,” he added, noting that those who recognized the sequence did not necessarily complete the sequence any faster. “It’s as if the brain has decided to focus on the details … how to take that sequence and step back and nail down exactly what I’ve learned to try to look for—a pattern or a shortcut or a meta-description. You don’t get faster, but you figure out the shortcut. It’s not time that’s important for this improvement in performance; it’s sleep itself.”

Recent data suggest that people who discovered this insight had less slow-wave REM sleep than those who did not. Some research suggests that sleep works to extract rules from specific information. “The rules don’t have to be correct,” he said; rather, the brain finds a pattern, “and this is the best it can do at trying to figure out what [that pattern] means.”

According to another study conducted with thematic word lists, “it turns out that when you see a list of words, your brain does two things,” said Dr. Stickgold. “It tries to create a memory of each and every one of those words that it saw, and it tries to create a memory of the gist of what that list was about. It has both of those memories, and it gets them confused,” in that the brain “remembers the gist long after any of the words on the list.”

One subcategory of procedural learning is perceptual memory, or how one learns to carry out various perceptual tasks. “It could be visual; it could be auditory; it could be kinesthetic; it can be the ability to reach into your pocket and [make the distinction necessary to] pull out a nickel instead of a quarter, unlike when you’re in [in a foreign country] and you have take the whole handful out,” said Dr. Stickgold. “This is a skill that’s developed slowly, mostly unconsciously, and it is consolidated across a night’s sleep.”

Dr. Stickgold noted that one unexpected source of evidence for sleep-dependent memory consolidation comes from the study of dreaming. “Dream reports, in common with microarray analyses of gene expression, single-cell and neural network recordings, and human brain images, can provide a window into the activity of the sleeping brain, albeit a subjective one,” he wrote in Nature in 2005.

Two studies suggest that episodic memories are not reactivated during dreaming, as sleep more likely protects and consolidates episodic memories, in addition to discriminating among them. In one study, volunteers who played the computer game Tetris before going to sleep reported that as they drifted off to sleep, they saw specific shapes from the game floating downward. When a group of patients with mild amnesia and bilateral damage to the hippocampus and medial temporal lobes who could no longer form new memories—to the extent that they had no memory of playing the game despite having done so for hours—were asked about their dreams, three of five reported seeing images “turned on their sides” or as “shapes on a screen,” although they could not remember what they were.

“So although those people cannot form a declarative memory that they can [recall], they still form traces of this memory in nonhippocampal regions of their brain, which holds the information in a way that they don’t have conscious access to it,” said Dr. Stickgold. “This is the path to the unconscious. They are displaying memories that, due to brain lesions, they are unable to recall during the day,” suggesting that it is not the act of playing the game that creates the Tetris images, or procedural learning, but some other feature.

In another study, 14 of 16 volunteers who played an “Alpine racer” computer game before going to sleep “reported images of skiing across three nights of sleep-onset dreaming,” Dr. Stickgold said. “Like Tetris, it has a profound impact on sleep-onset dreams,” with 42% of participants noting a skiing image the first night.

“So what happens to those images?” Dr. Stickgold asked. “Well, most of them disappear, but something even more interesting happens. We start getting reports like this: ‘I felt as if I was falling downhill.’ The brain, now having had two hours of sleep to process this information, is no longer playing it the way it happened; it’s asking different questions now: What’s that like? What other information do I have that might be relevant to this? What should link this up with? What association should I connect to this?”

He noted that one person reported, “‘I was moving forward very stiffly, my entire upper body incredibly straight; I felt like I was moving forward on a conveyor belt.’ So whatever it is that made those dreams, those images, so intrusive into our sleep-onset dreams, that’s been altered across the first two hours of sleep, so that it no longer had that intrusive quality. This is really important clinically, because that’s what fails in [patients with] PTSD [posttraumatic stress disorder]. We’ve all had those sleep-onset dreams after September 11. What should happen over time is that those memories get processed; they get integrated into our sense of self and to our understanding of the world and the trauma disappears. In those individuals for whom this doesn’t happen, the replay … continues. I would argue that this is a failure of sleep-dependent memory processing…. It’s as if the brain is trying to create meaning, and the way we create meaning is by taking information and embedding it in our general knowledge. After September 11, people said, ‘I don’t understand what happened.’ It’s not that they didn’t understand [a plane] crashing into a building. They understood that; they didn’t understand what that meant about their life. They didn’t know if they should go to work [the next day]. They didn’t know if they should take public transportation. They didn’t know if they should get on that plane with ‘that guy who looks a little different to me.’ It’s that kind of understanding that is probably the most sophisticated conditioning the brain does.”