The secrets of sleep
It's a mystery, but it clearly makes us smarter and healthier
Before long, some labs noticed that amounts of REM sleep increased as animals learned various tasks but went back to normal after tasks got mastered. What's more, experiments that deprived animals of REM sleep by disrupting them during this sleep stage found that they didn't learn as well as animals that got in plenty of dreaming. But the idea that sleep might aid smarts didn't catch fire. "Sleep wasn't supposed to be for that. It was supposed to be for restoration," recalls Carlyle Smith of Trent University in Ontario, who has studied sleep and learning for over 30 years.
Slowly, though, the idea has attracted more interest, especially since the early 1990s. Part of the change is that scientists have redesigned experiments to counter critics' early objections, such as the possibility that the sleepless learn less because of stress and fatigue, not the loss of sleep-specific brain work. They've also realized that certain kinds of learning seem more linked to sleep than others. Memorizing lists of words or facts--what's called "declarative" memory--doesn't seem all that dependent on sleep. But scientists have lately gathered compelling evidence that people's "how to" learning, or "procedural memory," gets a boost from a bout of sleep.
Last October, for example, Kimberly Fenn and colleagues at the University of Chicago showed sleep's benefits for this type of learning with the help of an annoying, outdated speech synthesizer. When Fenn's computer says "smart," the word comes out sounding more like "smote," and it's hard to figure out what word it's actually saying. "When people first come into the lab, they're really bad," Fenn says, but 30 minutes of training vastly improves their understanding of the garbled speech. What happens over the next 12 hours depends on whether they are allowed to sleep or not. In a series of carefully controlled studies, Fenn showed that while people's learned ability seems to fade away over the course of the day, a night's sleep brings it back. Test-takers who slept showed nearly twice as much improvement as those who did not.
Timing. Similar effects have appeared in other laboratory learning tests of "how to" motor skills (typing a string of numbers as fast as possible) and visual perception (spotting rare diagonal lines in a sea of horizontal and vertical ones). For example, Harvard Medical School's Robert Stickgold trained people to quickly find diagonal lines; then he either let them sleep that night or kept them awake. Both groups got to sleep normally for two nights before being retested, to make sure they weren't fatigued. Only those who got sleep soon after their training showed improvement over their pretraining test scores. "You absolutely have to sleep the first night," says Stickgold.
Other studies hint that in addition to solidifying learned skills, sleep lets the brain forge brand-new strategies that it might never develop while awake. Earlier this year, scientists in Germany described an experiment in which they taught volunteers how to do a math problem. But there was also a shortcut solution to the problem, which wasn't mentioned. After either letting people sleep for eight hours or keeping them up, the researchers found that the sleepers were more than twice as likely to discover the mathematical trick on their own. Jan Born of the University of Lubeck, who led the study, thinks that the sleeping brain may process and repackage learned information so that, upon awaking, the brain can suddenly "see" it in new ways.
advertisement
