Imagine you're cleaning off your desk. You sort some papers into folders with the relevant labels. Others you red-tag as "urgent" or yellow-tag as "semi-urgent." Quite a few go directly into the large circular file at your feet, also known as the trash basket.
Turns out, it seems that your brain does something very similar with your memories every night as you sleep.
The journal Nature Neuroscience has just published a special issue on memory, and among its authors is Dr. Robert Stickgold of Beth Israel Deaconess Medical Center and Harvard Medical School, a leading researcher on the role that sleep plays in consolidating memories.
In the past, my layperson's take-home message from his complex research might have been, "Sleep helps strengthen memories. So if it's the night before an exam and you have the option of cramming more or sleeping, better to sleep."
But there's ever so much more to know, and for researchers to find out — about sleep states beyond REM, about how our brains "tag" some memories for retention and dump others, about how sleep can improve performance overnight — even about why toddlers so desperately need naps.
Our conversation, lightly edited:
You've written a sweeping review of years of recent research on what sleep does to memories. How would you sum up for a lay audience what we now know?
I’d start by telling them a true story, which is that about 50 years ago, my father commented to me that when he was in law school studying for an exam, he would stay up late at night reading case after case, and go to sleep with a complete mishmash of cases in his mind. When he woke up the next morning, they had just all been filed away in the right spot. That was 50 years ago, and I can now say, 'Yes, and now we have an idea how.'
It really does happen while you sleep, and although some of it can happen while you're awake, especially if you're consciously working at it, sleep seems to be a time that's been set aside to make sure that filing gets done, even without your awareness or intent.
So sleep is a time of sorting and discarding memories?
Sleep is doing about five things.
The first things we knew that it could do is strengthen some memories, so you were just better at them in the morning. It could stabilize some memories so they aren't as susceptible to interference. Those are the first pieces we got, between about 2000-2007.
And then we started to get into the more subtle stuff. Sleep can also help you hold onto some of the things you learn during the day, and let the rest be forgotten. And that's the concept of selective retention.
And there is now evidence that in fact it doesn't just let the rest of it passively decay or be forgotten, but it can help it along that path of being forgotten.
Then, it turns out that the brain also does a lot of smarter stuff while you're sleeping. If you learn a list of words, it can extract the gist from them. It can make it so that you're two and a half times as likely to discover an insight into an alternative way of, say, solving math problems.
It can extract patterns. Say I'm trying to teach you how to use a deck of cards to predict the weather, which isn't very obvious. Imagine you've got four aces and in 200 trials, I show you one, two or three of the aces and then tell you it turned out to be rainy or sunny. And it turns out that it's not a simple pattern, because the cards only have a probability of being associated with sun or rain, so any given card, or two or three cards, predict sun or rain more or less, but not one or the other with certainty. Over 200 trials, people get to the point that they can predict with 70-to-80-percent accuracy what we're going to say the weather is. But after a night of sleep they’re 10 percent better. If they don't sleep, they get no better.
So somehow that information, in whatever form it’s been stored in the brain, is refined. The brain goes through it and says, 'We can clean this up and make you better at this.' Other studies have also shown this boost during sleep, and no improvement during the day.
You mention in the paper that sleep may be particularly good at helping the brain divine relations and rules — and that could even help explain why toddlers need naps, they're learning so much about the world.
Rebecca Gomez at the University of Arizona did a study in which she exposed 15-month-olds to words with a certain grammar at around noon, after which about half of them napped in the afternoon and half didn't. And then, in the late afternoon and even the next morning, the ones who napped had a good grasp of the grammar and the ones who didn't nap didn't. Sometimes I say one of my goals in life is to understand what it is about a 90-minute nap that turns a psychotic dwarf back into a charming 2-year-old.
What's funny about this effect of sleep is that we're aware of what happens; the phone call from hell is "I'm bringing Billy over for his playdate and by the way, he didn't nap." But no one stops to ask what happens in the brain of the toddler while they're napping that so alters the basic configuration of the brain so they can take in new information, roll with the punches, deal with social dissatisfaction — all these things they can do, and if they haven't had a nap, they can't do it.
So what happens while you're sleeping that so restores your brain? For one thing, you have to clean out your "Inbox" before you take in more information, and sleep seemes to be really good at that. Somehow, it’s filing, it's reorganizing, and I think what we're most impressed with is that it's doing it in a really smart way.
I was also very taken by your description of the concept that we "tag" memories, we metaphorically mark them as likely to be important for the future or not. It's like the ultimate "Will this be on the test?" Only the test is life itself.
The fundamental question our brain has to ask about every single memory is: Is this going to be on the test of life?
Dan Schacter [of Harvard] has really been pushing this concept lately, that memory is about the future and not about the past. We don't form memories so we can sit in our chairs when we're old and say, 'Do you remember how we used to go to the beach with the kids, and they were so little?' That's a freebie. The reason memory evolved is so that next year when you go to the beach and your 8-year-old asks to swim out to the raft, you remember that last year someone had to go in and drag her out because she couldn't swim that far, so no, she can't swim out to the raft — and that might double your chances of your genes being passed on. That’s what memory is really about — it's all about knowing what's worth saving, and what you'll need in the future. That's the selective piece.
And then the other half is: What form do you need it in? You don't need to remember which beach you were on, or who the nice guy was who pulled her out. You just have to remember she almost drowned trying to swim to the raft, and when she says, 'Mom, it's closer this year,' you can still say no.
This is what goes awry in PTSD: It develops because people's brains can't figure out which parts they need to save and so they save it all. The point isn’t to remember that a blue car ran through a stop sign and rammed you, but that even when there's a stop sign you should still have some sort of awareness of whether cars are coming at high speed.
So is all of this a big hint at the eternal question of what sleep is for, why we need to sleep?
What is the tongue for? Two guys were arguing, and one said the tongue was for tasting food, and the other guy said it was for articulating speech, and both have good arguments, but ultimately it's a stupid question. If you heard two guys arguing about that you'd say, 'Guys, it really does both.' But you're right, people in the sleep research field are still going around saying, 'What's its function?'
Here's what I can tell you: If you don't get a night of sleep after you get a Hepatitis B vaccination, two weeks later the amount of antibodies you've produced in response will be down 50 percent. If you go on four hours of sleep a night for five days and then take an intravenous glucose tolerance test, you look pre-diabetic. So we know that sleep is clearly doing something important for immune function, for endocrine function, and we could argue for memory function and probably for emotional function.
Sometimes I say to my students, here's what we've learned: It turns out that for humans, for every two hours during the day of taking in new information — studying, watching TV, walking through the park, you're always taking in new information — the brain needs about an hour offline to sort it out and figure out what it means and get you ready to take in more information.
Of all those functions of sleep, the memory piece is the only one that reasonably requires that you are really asleep. All the other functions people talk about seem to be more about getting the body into a relaxed state. As a child, you release something like 80 percent of your growth hormone not just during sleep but during slow-wave sleep. You can imagine during evolutionary times the organism discovered a good time to do a lot of your growing is when you’re flat out on your back. Why do they clean office buildings at night? It’s not when the staff wants to clean, it’s just that everybody else is working during the day.
The other functions of sleep would work fine if you were just lying in a meditative state for eight hours — you wouldn't need to be out of awareness of your environment, unable to smell smoke or hear breaking twigs. Only the memory piece would require that you cut yourself off from the world. Just as a VCR cannot record on one channel while playing another channel.
Most of us have heard about REM — rapid eye movement — sleep and its link to dreaming and restorative sleep, but you also talk about "slow wave sleep." What does it do?
Slow wave sleep is the deepest phase of our sleep. It comes mostly during the first half of the night, and it gets its name from the EEG patterns that are seen during this phase, which consist of large, slow waves.
We know about REM sleep because when it was first discovered it was mistakenly thought it was the only time we dream. It turns out we dream probably most of the night but REM dreams are still more vivid and emotional and complex. But we are still struggling over what each sleep stage is doing: No matter how we slice the pie, we can't quite come up with the functions of slow wave sleep vs. stage 2 vs. REM, but it's very clear there are a lot of forms of memory processing that seem to occur preferentially during slow wave sleep.
I would say most of those — though there is debate about it — are involved more with remembering things in the way the memories were initially formed. You get most of your slow wave sleep early in the night, so maybe the first thing you do when you sleep is like what you do as a writer: You start by writing things down so you won't forget them. And after you have them safely down so they won't get lost, you can start thinking about, how do we cut and paste this? How do we spin it? All that can happen later during REM sleep.
I would say slow wave sleep might turn out to be what you need to take newly formed memories and sort of file them away in a safe and stable form. It's like taking them out of your computer's memory and saving them to a hard drive. So with that metaphor, REM is probably like your word processor and Power Point and graphics programs; it's the time you take information that you've now safely stored on your hard drive and manipulate it and see what can be created out of it. A lot of times it's mundane, but sometimes it's quite spectacular.
This program aired on January 29, 2013. The audio for this program is not available.