Cover Story

Robert Stickgold, PhD, wants you to know that your mother was right-you really should be getting more sleep.

Stickgold, a professor of psychiatry at Harvard Medical School, believes that sleep allows us to process, consolidate and retain new memories and skills. In his lab, he investigates sleep's effects on students who learn new tasks, such as complex finger-tapping sequences. He's found that depriving the students of a night's sleep afterward significantly cuts back on how well they remember the new skill as much as three days later.

The idea that people need sleep to consolidate memories has waxed and waned in popularity since psychologists John Jenkins and Karl Dallenbach first proposed that sleep benefits memory more than 80 years ago. In the last 10 years, evidence from memory studies in humans and rats, as well as research on the cellular and molecular workings of the brain, have provided increasing evidence for the link.

Now, researchers are investigating which sorts of memories might be consolidated during sleep and which stages of sleep are important-after all, sleep is not a monolithic entity but is divided into stages that cycle back and forth, from the most active (rapid eye movement, or REM) to the least active (slow wave sleep).

Still, the theory remains controversial. And the details remain elusive-partially because scientists don't entirely understand how the brain retains memories at all. Without that knowledge, it's difficult to be certain of how-or even if-sleep contributes to the process.

"It would have been nice if the memory researchers had worked everything out," says Stickgold, "but it turns out we may be able to help them some."

An old idea, new interest

Several researchers studied sleep and memory in the 1960s and 70s, but the number of studies in the area tapered off during the 1980s. The most recent spate of interest in memory and learning took off in 1994, after an influential paper published in Science (Vol. 265, No. 5,172, pages 679-682) helped revive the field, according to Stickgold.

In that study, neuroscientist Avi Karni, MD, PhD, trained participants to report whether they'd seen horizontal or vertical lines appear over a background pattern on a computer screen while they focused elsewhere on the screen. Most participants took about 100 milliseconds to identify the lines' orientation. But when the participants returned the next day after a normal night's sleep, they were able to do the task about 15 milliseconds faster-and the improvement lasted for a year. When researchers kept the participants in the lab overnight and woke them every time they fell into REM sleep, they didn't improve at all.

Stickgold and others have followed up on this study, examining such questions as whether sleep stages other than REM may be involved. In a 2000 study, for example, Stickgold found that participants' improvement on the line-identifying task depended on how much slow wave sleep they got during the first quarter of the night and how much REM sleep they got during the last quarter of the night.

"The more you sleep, the more improvement you show-it's really a nice linear relationship," Stickgold says.

In a follow-up study, he showed that sleep deprivation the night after training blocked improvement on the task even after participants had two nights of normal sleep to recover.

In another study, Stickgold taught participants to tap out a number sequence on a keyboard. Students improved their speed on the task with practice for a few minutes, but then quickly reached a peak performance level. A control group of students who were tested 12 hours later on the same day showed no more improvement on the task, but those who went home and slept overnight improved 20 percent by the next day, although they had not practiced any more.

Of course, these studies tested procedural rather than declarative memory. Procedural memories are memories for motor and perceptual skills like riding a bike or hitting a baseball-or tapping numbers on a keyboard. Declarative memories, on the other hand, are memories of facts like the capital of North Dakota or a friend's birthday.

Evidence for sleep's effect on declarative memory is much weaker than for its effect on procedural memory. In fact, few studies have shown any link between the two. Still, says Stickgold, even this distinction is less sharp than it might seem-sleep may effect "complex procedural" memories, like the ability to synthesize new information with old and develop new ideas.

In fact, a 2004 Nature study by German neuroscientist Jan Born, MD, of the University of Lübeck, found evidence for just that (Vol. 427, No. 6,972, pages 352-355). Born gave participants a math test that required them to use a set of complex rules to convert eight-number strings of digits into new, seven-number strings, and then to identify the last number in the new string. But there was a trick to make this task easier: The second number the participants calculated was always the same as the last number of the new string.

When the participants first took the test, none recognized the trick. After a night of sleep, though, 13 out of 22 participants caught on. Only 5 out of 22 participants who were retested after the equivalent amount of daytime wakefulness got it.

"If you're going to be tested on 72 irregular French verbs tomorrow, you might as well stay up late and cram," Stickgold says. "But if they're going to throw a curveball at you and ask you to explain the differences between the French Revolution and the Industrial Revolution, you're better off having gotten some sleep."

What's going on inside the brain

Studies of human learning provide tantalizing evidence that sleep helps us retain new memories, but they don't provide information about how it does so. For that, most researchers turn to animal studies. Scientists are exploring a few lines of evidence, most of which, by necessity, approach the issue of memory obliquely. University of Pennsylvania neuroscientist Marcos Frank, PhD, for example, is studying how sleep influences brain cell plasticity-the ability to form new synapses and connections between brain cells. Researchers know that brain cell plasticity is key to creating and retaining memories, but because scientists don't know precisely where memories are stored in the brain, they can't look for "new-memory" synapses directly.

Instead, Frank is examining how sleep affects plasticity in the visual cortex of cats. He blocks the cats' vision in one eye for a few hours when the cats are young, which causes cells in the visual cortex to remodel themselves to respond less strongly to the deprived eye. In a 2001 study in Neuron (Vol. 30, No. 1, pages 275-287), Frank blocked cats' vision for six hours while allowing them to explore their environments in a lighted room. Then he tested the synaptic remodeling that resulted. In the next phase of the experiment, he allowed one group of cats to sleep for six hours and kept another group awake in the dark. He found that the synaptic remodeling continued for the extra six hours in the sleep group, but not in the waking group.

"We're not studying sleep and memory per se, but we are studying a model of how the brain can change its structure with experience," Frank says. "We found that sleep does seem to play a role in this plasticity and enhance changes that are already taking place."

Now, Frank and his colleagues are working to isolate the particular sleep stages-REM or non-REM-that contribute most to this plasticity.

Other researchers, meanwhile, are looking at a smaller level than cells-they're investigating genes.

Neuroscientist Sidarta Ribeiro, PhD, for example, has studied a gene called zif-268, which is related to brain plasticity. When activated, the gene activates a protein called synapsin II, which promotes the formation of new synapses. In a 1999 study, Ribeiro allowed rats to explore a new environment-thus creating new memories-and then to sleep for several hours. He found that during the REM sleep that followed, the zif-268 gene was more active in the rats' cerebral cortexes and hippocampi-the areas thought to be important for memory.

"The hippocampus is the entrance gate for memories, and then they migrate to the cortex," Ribeiro explains.

He, like Frank, is now working to sort out the contribution that slow wave sleep makes to the process as well.

The other side

Studies like these, and dozens of others, have convinced most sleep researchers that sleep plays an important and unique role in human memory. But a few scientists vehemently disagree with their findings. One of the most vocal detractors is psychologist and neuroscientist Robert Vertes, PhD, who studies memory at Florida Atlantic University.

One of his central arguments comes from the widespread use of antidepressants. Older antidepressants, such as monoamine oxidase inhibitors (MAOIs), completely eliminate REM sleep in users, and even the newer selective serotonin reuptake inhibitor (SSRI) antidepressants can significantly reduce REM sleep for months or years. Given the huge number of people who've taken these drugs, Vertes asks, why haven't we seen widespread reports of memory impairment?

"They'd pull these drugs off the market if people were saying 'They're making me senile,'" he says.

Supporters of the memory consolidation theory point out that it's possible that functions associated with REM sleep could migrate to other sleep stages in a person deprived of REM sleep for a long time. And, they say, just because people are not self-reporting memory problems doesn't mean that those problems aren't occurring.

"Bill Gates has been quoted saying that his programmers can program for 72 hours straight," Stickgold says. "And I say-yeah, but their product is Windows."

But Vertes and some others-including, perhaps most notably, psychologist Jerome Siegel, PhD (see "Brain, heal thyself")-remain unconvinced. There are too many questions that the theory fails to answer, they say. For example, if our brains consolidate memories during REM sleep, why is it that our dreams-our only window into the brain's activity during that sleep stage-so rarely relate to anything we want to remember?

And what about new memories that a person acquires during his or her first waking hours? Where are they held during the 12 to 16 hours until that person goes back to sleep?

Overall, Vertes says, he just doesn't find the memory consolidation studies convincing. "I believe that memory is online, not offline," he says.

Stickgold agrees that his critics raise points researchers need to address.

"I think every issue Siegel and Vertes raise is important," he says. "But they're not death knells for the field-they're the questions that arise as a field matures."

Further Reading

  • Stickgold, R. (2005). Sleep-dependent memory consolidation. Nature, 437(7063), 1272-1278.

  • Vertes, R.P. (2004). Memory consolidation in sleep: Dream or reality. Neuron, 44(1), 135-148.