Is sleep good for everything? Scientists hate giving unqualified answers. But the more sleep researchers look, the more the answer seems to be tending toward a resounding affirmative.

The slumbering brain plays an essential role in learning and memory, one of the findings that sleep researchers have reinforced repeatedly in recent years. But that’s not all. There’s a growing recognition that sleep appears to be involved in regulating basic metabolic processes and even in mental health. Robert Stickgold, a leading sleep researcher based at Harvard Medical School, gives a prcis here of the current state of sommeil as it relates to memory, schizophrenia, depression, diabetes—and he even explains what naps are good for.

How far have we come in understanding sleep?

Although we understood the function of every other basic drive 2,000 years ago, we are still struggling to figure out what the biological functions of sleep are. One of the clearest messages now is that for every two hours humans spend awake during the day, the brain needs an hour offline to process the information it takes in and figure out what to save and what to dump and how to file and what it all means.

So what is sleep for?

Memories are processed during sleep. But sleep doesn’t have just one function. It’s a little bit like listening to tongue researchers arguing about whether the function of the tongue has to do with taste or speech. And you want to say: ‘Guys, c’mon, it’s both.’ There’s very good evidence now that sleep, besides helping memory, has a role in immune and endocrine functions. There’s a lot of talk about to what extent the obesity epidemic is actually a consequence of too little sleep.

How did sleep come to play so many diverse roles?

It’s a little bit of a tricky game because it looks like sleep has been identified evolutionarily as a good time to get stuff done. If you work in a big office, the cleaning staff comes in at night. It’s not because that’s when they like to or that’s when they’re most efficient but that’s when the people are out of the office and the office is shut down and it’s easiest to get housekeeping chores done. I imagine that some functions over evolutionary time have glommed onto sleep because that’s when the system is most shut down. Something like two-thirds of your growth hormone is secreted during deep, slow-wave sleep and the presumption is not so much that that this is a critical function of sleep, but that this is the best time to pause and spend energy on growth.

What is sleep’s relationship to memory?

Sleep plays a clear and often critical role in the stabilization of memories and in their strengthening and retention, allowing some memories to be retained and others to be let go. It’s also critical in the extraction of gist from large collections of information and the discovery of rules that control that information so if you give various stimuli, sleep can help you determine what the rules are as to what the stimuli produce and can even help with the development of insight in patterns and rules that you didn’t even know were there.

What is the importance of dreams?

Much less, much less. A decade back we had a paper in The Journal of Cognitive Neuroscience in which we showed that episodic memories of events in our lives aren’t replayed in dreams. So the old concept that you dream about what happens, you don’t dream what actually happened, is actually well validated. The dreaming brain seems to have as its goal to construct hallucinatory scenarios that bear resemblance to recent events but are different from them.

Erin Wamsley and I have shown with a computerized maze learning task that if you take a nap after learning you improve on that task more than if you stay awake. Also, if you collect dream reports during that nap, you find it’s the people who report that they’ve been dreaming about the maze task who show the preponderance of improvement during re-testing. Interestingly, if you look at the dream content, it’s very clear that the dream content per se is not helpful. It has a sort of wacky bizarre relationship to the task, but not in a way that would help you perform better, which has led us to describe it as a sort of biomarker of a memory process.

What we think we’re seeing is that at the same time that the hippocampal region of the brain, which we know encodes spatial maps, is replaying and strengthening its memory for the layout of the maze, other regions of the brain involved in dreaming are constructing these ‘what if’ situations, imaginary future scenarios, sort of getting at the question of what is the usefulness of this information.

One subject reported dreaming of the maze and then remembering being in a bat cave a few years ago, and you can imagine his brain saying: 'OK, how do these two fit together? Is there something about this bat cave that might help me do this task better? Or is there something about this maze task that I should not just file under experiments that I participated in, but file in a way that if I go exploring in a cave again then maybe something I learned about exploring a maze from this game I just played might be useful?’ And I think we’re getting a sense that that’s what dreaming’s about—how might this information be useful to me in the future? It’s exploring associative networks and trying to find associations that look promising.

What are implications of sleep for psychiatric disorders?

If you take an adult who has both sleep apnea and depression, you’ll find that they are very tightly linked. If you have depression, there’s a fourfold increase in your likelihood of apnea and if you have apnea, there’s a fivefold increase risk of depression. If you take someone with both depression and apnea, and treat the apnea with CPAP [continuous positive airway pressure], you can get their depression scores to drop below clinical levels.

If you take kids comorbid for sleep apnea and ADHD—in the case of children the apnea is usually caused by enlarged adenoids and tonsils—if you remove the tonsils and adenoids you’ll get a larger reduction in the ADHD symptoms than if put them on Ritalin.

If you take people with bipolar disorder and sleep deprive them, you’ll flip them into the manic state.

If you look at depressed people, REM sleep comes much earlier in night. When treatments for depression fail to reverse this effect, the likelihood of recurrence of the depression is much higher. And depriving depressed patients selectively of REM sleep can produce a dramatic reduction in their symptoms, although they return as soon as the deprivation is stopped .

If you take chronic, medicated schizophrenic patients, they have this huge deficit in what are called sleep spindles, which are these one-second long EEG oscillations. Chronic medicated patients have a 40 percent reduction in sleep spindles compared to controls; Giulio Tononi’s group [at the University of Wisconsin] and our group have found the larger the reduction in spindles, the more positive symptoms [hallucinations and delusions, for instance] the patient shows. And the reduction in spindles correlates with a reduction in sleep-dependent memory enhancement. We just had a paper out in the journal Sleep that shows that if you give these schizophrenic patients, Lunesta (eszopiclone), you can increase their spindles and sleep-dependent memory processing now looks normal.

Are we at a point that we can make recommendations on how much sleep someone needs each night?

Asking how much sleep you need at night is like asking how asking how much food you need. The average person would say 2000 calories. But you also need a certain amount of protein and vitamins, for example. The calories are just to support energy demands of the body.

When you ask me how much sleep someone needs I ask for what? It might be that the amount of sleep needed to consolidate episodic memory is different than the amount of sleep you need to extract gist from information, and both may be both different from what you need to optimize glucose regulation and antibody production.

If you take college students and put them on four hours of sleep a night for five days and then do a glucose tolerance test, they’re looking pre-diabetic. That’s not enough sleep for maintaining normal insulin regulation, but for some memory tests four hours may be more than enough.

And we have to ask how much sleep when? For some memory processes, a 90-minute nap gives you as much benefit as a whole night of sleep and in some cases more benefit than six hours or sleep or four hours of sleep at night. There’s something unusually efficient about naps in terms of these memory processes.

On the other hand, if you feel like the amount of sleep you’re getting is enough, if you don’t sleep longer on weekends, if you don’t think it would be a joke to not set an alarm clock, if you don’t drink coffee within the first two or three hours of getting up, you’re probably doing fine.

If you drink two cups of coffee to get going in the morning, you don’t have enough sleep. If you sleep two hours later on weekends, you are not getting enough sleep. I think the amount varies from person to person. There isn’t an absolute amount. I tell people to turn off the alarm clock for a week and see what happens. If you discover you’re waking an hour and a half after you’re supposed to be at work, you’re probably not going to bed early enough. I still think eight hours looks like the best bet if I were to guess.

Image Source: Albert Londe/Wikimedia Commons