MATTHEW WALKE t's a pleasure to be here. And I want to start with a standard disclaimer, which is that when most speakers look to their audience and they see people who are falling asleep or nodding off, it can be profoundly disheartening. However, based on the topic of today's presentation, I'm almost going to actively encourage that kind of behavior from you. In fact, knowing what I know particularly about the relationship between sleep and memory, it's actually the greatest form of flattery for me to see people like you not being able to resist the urge to strengthen what I'm telling you by falling asleep. So feel free just to sort of ebb and flow in and out of consciousness throughout the entire talk. I'll take absolutely no offense. And the talk itself is really going to come in at four main acts, so to speak. Firstly, I want to spend some time telling you about what sleep actually is, the different types, it's characteristics, its structure. And then after that, I'll tell you about the variety of different functions, plural, that we're starting to understand that sleep serves. So I'll tell you about the role of sleep in promoting learning and also memory. But I'll also then tell you how sleep can go beyond simply manipulating individual memories. Sleep seems to be intelligent in that it can cross-link new pieces of information together so you can come up with creative, novel insights the next day. And then finally, I'll describe a role of sleep beyond information processing into your mental health and how sleep seems to be critical for emotional regulation, preparing specific brain circuits for next day social and emotional interactions. So that's the basic overview. Coming on to what sleep is, and I do love this picture. You can just kind of get a sense of the quality and the depth of sleep that's happening there. If we're going on that whole savanna grasslands kind of side street by the way, I just want to come onto this, the giraffe. Firstly, what a strange morphology for a creature. Have you ever wondered how something that looks like that sleeps? Would you like to know how a giraffe sleeps? That's how a giraffe sleeps. Isn't that remarkable. And it tells us at least two things. Firstly, despite such bizarre anatomy, sleep will still find a way to be obtained by the brain. Second, and more generally, in every species that we've studied to date, sleep, or something that looks very much like it, has been observed. What that means is that sleep has fought its way through vehemently every step along the evolutionary pathway. If that's true, sleep must be essential at some of the most basic of biological levels. And that's exactly what we're starting to discover. And sleep in terms of mammalian species at least has been broadly separated into two main types, as some of you may know. On the one hand, we have non-rapid eye movement sleep or non-REM sleep for short. And non-REM sleep has been further subdivided into four separate stages, unimaginatively called stages 1 through 4-- increasing in their depth of sleep-- or a creative bunch of sleep researchers. So increasing in the depth of sleep, stages 3 and 4 are those really deep stages of dreamless sleep. And they're often grouped together under the term slow-wave sleep. Why? Because of these slow, lazy brain waves that happen during the stage of sleep that we measure with electrodes on the head. But don't be fooled. That's not that your brain is dormant by any stretch of the imagination. What it means is that vast portions of your brain, hundreds of thousands of neurons, have all decided to synchronize together and sing together in time. It's a phenomena like no other brain state that we know of. It doesn't happen whilst you're awake. It's a strange phenomena and we still don't truly understand why. On the other hand, we have rapid eye movement sleep or REM sleep named, not after the popular Michael Stipe pop band, but because of these bizarre, horizontal, shuttling eye movements that occur during this stage of sleep. And again, we don't truly understand why your eyes move during that stage of sleep. And it turns out that these two types of sleep, REM and non-REM, will play out in a battle for brain domination throughout the night. And that sort of cerebral war is going to be won and lost every 90 minutes and replayed every 90 minutes. And what that creates is a standard architecture of sleep, what we call a sleep cycle. So I'll just unpack this for you here. We've got the different stages of sleep on the vertical axis. And then time of night along the horizontal axis. And I'll speed this up for you. But what you can see is that upon falling asleep, your brain goes on this delightful roller coaster ride in and out of these different stages of sleep. So you'll quickly descend down into the deep stages of non-REM sleep, 3 and 4. And you'll stay there for a while. And then after about 70 or 80 minutes, you'll start to rise back up and you'll pop up and have a short REM sleep period, here in red. And then back down you go again, down into non-REM sleep and then up into REM. As I said, this cycle is 90 minutes, non-REM through REM. And that's stable across the night. However, what changes is the ratio of non-REM to REM within that 90-minute window as you move across the night, such that in the first half of the night the majority of those 90-minute cycles are comprised of deep non-REM sleep, slow-wave sleep. Whereas as you push through to the second half of the night, now that ratio balance shifts across. And instead, they're dominated much more by rapid eye movement sleep, as well as that lighter form of non-dreaming sleep, stage 2 non-REM sleep. And just to come back to REM sleep, REM sleep is the principal stage during which your brain dreams. And REM sleep is a case of essentially how your brain goes completely out of its mind. Because every one here, as long as you slept last night, you became flagrantly psychotic. Now before you reject my diagnosis of a nightly psychosis, let me give you five good reasons. Because last night when you were in REM sleep and you were dreaming, you started to see things which were not there. So you were hallucinating. Secondly, you believed things that couldn't possibly be true. So you were delusional. Third, you became confused about time, place, and person. So you're suffering from disorientation. Fourth, you had wildly fluctuating emotions. Something that psychiatrists call being affectively labile. And then how wonderful, you woke up this morning and you forgot most, if not all, of that dream experience. So you're suffering from amnesia. If you were to experience any one of those five symptoms whilst you're awake, you would be seeking psychiatric treatment. Yet for reasons again that we don't fully understand, it seems to be both a normal biological and psychological process. One of the other fascinating features of REM sleep is this, paralysis. All of you, when you went into REM sleep last night, were paralyzed. It turns out that there's mechanism deep down here in your brain stem-- so here we have the brain, which as Woody Allen suggested, was his second most favorite organ of the body. And so here's the front of the brain, back of the brain, brain stem down here. Now, this war of REM and non-REM sleep essentially plays out down here. And then is beamed up to the top of the wrinkled mass, atop of the brain, called the cortex. But there's also another signal that goes south, down into the spinal cord. And this signal during REM sleep that goes south essentially inhibits what we call the alpha motor neurons in your spinal cord. They control all of your voluntary skeletal muscles. So ensuring REM sleep, your brain paralyzes your body so your mind can dream safely. It's a bad evolutionary design when you're not perceiving your outside world to start acting out all of those dream commands. And there are plenty of them. Just as a quick aside on this process note by the way, sometimes this persists despite you waking up. Some of you may have experienced this, this persistence of sleep paralysis on awakening. It's quite unusual-- well, it's not unusual proportional wise. About 25% of the population will experience this. It's about as common as hiccups. And what seems to happen is that your brain starts to wake up, but the paralysis isn't released from the body. So you start to become aware. But you can't lift your eyelids, voluntary muscles. You can't move. You can't say anything. It is often associated with a sense of sort of anxiety, a sense of someone else being there in the room. It turns out that this sleep paralysis, this persistence, accurately explains most, if not all, of so-called alien adoptions. I mean when was the last time you ever heard of someone being abducted during the day, in the middle of a meeting? You know. I mean-- whoosh, what was that? Well I believe Jimmy just got abducted by aliens. No, it never happens like that. It's usually at night, when you're in bed. People describe a sense of a presence in the room, that you were paralyzed by these other agents.1 You couldn't move.1 You couldn't fight back.1 You couldn't talk.1 It's a strange interesting feature.1 That's a little bit about what sleep is, together1 with some odd sides.1 I don't quite know I threw that in there.1 But anyway, let's now come onto what sleep is doing.1 And it is serving a whole broad array of functions.1 Firstly, let me tell you why it's essential to sleep before1 learning, to prepare your brain, almost like a dry1 sponge, ready to soak up new information the next day.1 And to sort of test this question, we're going to run1 an experiment.1 Essentially, is pulling the all-nighter a good idea?1 Here's how you do this.1 You take two groups of participants.1 You assign them to a sleep group or a1 sleep deprivation group.1 Both are awake across the first day.1 But then across the following night, those in the1 deprivation group, we keep them awake in the laboratory1 under full supervision.1 They can't fall asleep.1 The sleep group they get a full eight hours.1 Both of them are awake across the second day.1 And then we have them try and cram a whole bunch of facts1 into their brain.1 And then we're going to test them to see how efficient that1 learning has been.1 But instead of testing them immediately after learning, we1 actually wait until two full recovery nights of sleep1 before we test them.1 So that any measure of memory that we get is not confounded1 by them simply being too sleepy or inattentive to1 recollect what they've learned.1 And that's what you're looking at here on the vertical axis,1 the efficiency of learning.1 So the higher up you are, the better you are.1 And if you put those two groups head to head, what you1 find is that under conditions of sleep deprivation, there is1 a quite profound 40% deficit in the capacity of your brain1 to make new memories, to be able to create new1 experiences.1 And this should perhaps be little concerning considering1 what we know is happening to sleep in our educational1 populations.1 If you want to put this in context, it's simply the1 difference between acing the exam and failing it miserably.1 Now, of course these are just performance data.1 We don't know what's going on inside the brain.1 So to answer that question, we've repeated these1 experiments.1 But now, during that attempted learning, subjects are1 actually inside an MRI scanner as we're taking snapshots of1 brain activity to see which parts of the brain are1 switching on or not switching on.1 So you get these attempted maps of learning in the sleep1 group and in the sleep deprivation group.1 And then you simply subtract one from the other to see what1 the difference is.1 And when you do that subtraction, you find a highly1 selective, but highly significant impairment in this1 part of the brain here.1 It's a structure called the hippocampus that I'm1 circling for you.1 So just to orient you for those not familiar with MRI1 images, it's as if I've sliced through the1 brain from ear to ear.1 And you're looking in from the front, top of the brain,1 bottom of the brain, left and right side.1 And I'm circling for you the hippocampus here.1 You have one on the left and one on the right.1 And these cool, blue blobs demonstrate that this part of1 the brain was significantly impaired in those people who1 were sleep deprived compared to a nice, strong signal1 coming from that part of the brain in those people who had1 had a good night of sleep.1 Why is this important?1 Well, it turns out that this structure, the hippocampus, is1 the quintessential reservoir for where your brain creates1 new memories.1 In fact if you want to know what life is like without a1 functioning hippocampus, just watch the movie "Memento." I'm1 sure many of you have seen this film.1 If you haven't, watch it, it's a great film.1 And I won't spoil it for you.1 But essentially, this gentleman1 has some brain damage.1 And from that point forward, he can no longer1 make any new memories.1 He is densely amnesic.1 The part of his brain was damaged was this structure,1 the hippocampus.1 It is the very same structure that sleep deprivation seems1 to selectively attack and block your brain's capacity1 for efficient learning.1 Let me just go back to these data because there's an1 unresolved question here.1 That was the bad that happens when you don't get sleep.1 What's going on in those people who are getting sleep?1 In other words, what is it about the sleep that they're1 getting, the physiology of their sleep, that seems to be1 promoting the restoration of memory?1 And what we've been finding is that there are specific1 electrical brainwave patterns that are promoting this memory1 restoration.1 And they're coming from non-rapid eye movement sleep.1 And they're these delightful little chaps.1 They're called sleep spindles.1 These are short, synchronous bursts of electrical activity1 in the EEG, the electroencephalogram.1 They last for about one second of time.1 So you're going along, brrrrrrr, that's1 the burst of activity.1 Your brain doesn't make that sound, of course.1 That would just be strange.1 But they're these sort of champagne cork, synchronous1 bursts of activity.1 And we believe that they form part of a broad network that1 promotes the transformation or the translocation of memories1 from one location in the brain to another.1 And you can think of the USB, since I'm at Google, in a1 crass analogy, like a USB hippocampus stick.1 It's very good grabbing information somewhat quickly,1 but it has a limited storage capacity.1 And we believe that these spindles are helping promote1 the transfer from that hippocampus USB stick, up into1 that folded mass, the cortex, essentially, in terms of the1 analogy, the hard drive, the mass storage1 capacity of the system.1 And by promoting that real estate transaction, that1 shifting of geography of information within the brain,1 not only do you take previous memories and make them safe,1 put them onto the hard drive, you clear out the USB stick in1 terms of its memory capacity.1 So when you wake up the next day, you're freely able to1 start loading up new information again.1 Because what we find is that the more of these sleep1 spindles that you have, the greater the degree of1 restoration of your learning capacity the next day.1 So each of these dots represents an individual1 participant.1 The more of those spindles that you have, the greater the1 degree of memory return in terms of capacity for learning1 that you get the next day.1 So we're starting to understand not just the bad,1 when you don't get sleep, but exactly what it is in terms of1 the good, when you do get sleep, that promotes these1 cognitive benefits.1 It turns out that it's not just sufficient for you to1 sleep before learning.1 You also need to sleep after learning to essentially cement1 that new information into the neural architecture of the1 brain and make it less vulnerable to being forgotten.1 So it's essentially like hitting the save button.1 It just takes a lot longer organically within the brain1 to do that.1 And there's now good evidence that following that type of a1 learning scenario, you do need sleep to hit that save button1 so that you get that improved recollection1 the following day.1 And for fact-based memories, what you would think of as1 textbook-like memory, that seems to require, in terms of1 sleep, deep sleep, stages 3 and 4, or that slow-wave sleep1 that I described.1 So there's lots of good evidence of the past sort of1 15 or 20 years that this is the case,1 correlational evidence.1 But of course, what you tend to want in science is a causal1 demonstration.1 So the question is if you can increase the amount or the1 quality of your deep slow-wave sleep, presumably you could1 boost the amount of memory benefit that1 that sleep is providing.1 The question of course becomes how do you boost the quality1 of your slow-wave sleep?1 Well, there are a variety of different ways.1 But of course, your favorite and my favorite that would be1 this, direct current brain stimulation.1 1 Have you seen those adverts late night on television where1 they say don't try this at home?1 This is one of those.1 This is not car battery and a couple of electrodes, OK.1 Although that would be an interesting experiment.1 Just imagine--1 I'm just picturing someone tucking themselves into bed at1 night, with a bed partner.1 Good night, honey.1 And you're playing these electrodes.1 She says, what are you doing?1 Don't worry about me.1 I'm just boosting my sleep.1 So you can inject essentially a small amount of voltage.1 And I'll just show you.1 They're clinically approved.1 This is what it looks like.1 You inject a small amount of voltage into the brain.1 Now, it's so small that you don't even feel it.1 That's how tiny it is.1 But it is physiologically efficacious.1 And the idea here is that you've going to try and pulse1 in time with the brain during those slow brain waves, OK.1 And you're going to try and boost the amplitude, the size1 of those slow waves, on the sea of your brain's cortex.1 And by boosting that quality of that deep sleep, what1 happens to memory?1 So you're going to be applying it during that1 deep, slow-wave sleep.1 You've sort of singing in time with the brain.1 And there are two groups in this experiment.1 Both groups get all of the equipment1 applied to their head.1 One of them doesn't get any stimulation1 during sleep, however.1 The other does get simulation.1 And here's how the experiment works.1 Here's the mock stimulation group, so the1 placebo as it were.1 They're going to study a whole list of facts1 before going to bed.1 Then you can briefly test them to see what their1 retention is like.1 Then after a night of sleep, the next morning you test them1 again to see how well their brain has retained the1 information following sleep.1 In the other group, the experimental group, this is1 where we're going to stimulate the brain activity.1 We're going to juice it up and see if you can1 sort of enhance it.1 This is great study done by a German group a few years ago.1 The question is what happens in terms1 of the memory benefit?1 Well, if you look at the group that slept but didn't get1 simulation, we see the nice, normal memory retention1 benefit across sleep, replicating what we've seen1 many times before.1 In the group that gets the stimulation, you almost double1 the amount of memory benefit that you get by way of sleep,1 a causal demonstration that when you manipulate sleep,1 your manipulate memory.1 One of the depressing things, however, unfortunately, is1 some evidence that we recently published just a few months1 ago, looking at the interaction between sleep and1 memory as you're getting older.1 Which for me, seems to be rather rapid.1 And what we know certainly, and of course everyone knows,1 is that as you get older, your capacity for learning and2 memory starts to deteriorate.2 But one of the quintessential physiological hallmarks of2 aging is that your sleep starts to deteriorate.2 And it's not all types of sleep homogeneously.2 Some types of sleep get hit by the aging process far more2 severely than others.2 The type that gets hit most severely is that deep,2 slow-wave sleep.2 And so the question was whether or not these factors2 are simply co-occuring or actually closely related?2 In fact, we demonstrated that they are significantly2 interrelated.2 And this pernicious drop in deep sleep by over about 70%2 accurately accounts for about 50% of the forgetting that2 happens with age.2 These are huge numbers.2 So there's a suggestion here that disrupted sleep is an2 underappreciated factor that may contribute to what we2 called cognitive decline in aging.2 The exciting silver lining part to that cloud, however,2 is that it's a potentially treatable target.2 So we're now trying to see if we can use these types of2 methods to restore some quality of sleep in aging and2 see if as a consequence, we can give2 back some memory function.2 2 As it happens, it's not just sleep after learning to2 strengthen individual memories.2 Because we've been recently finding that sleep can go far2 beyond individual memories.2 Sleep can actually seemingly cross-link vast sets of2 information, and from that abstract understanding, and2 even develop creative insights and ideas from that2 information processing en mass.2 Let me show you an example of this.2 Here, in this study, you're going to be, as the subject,2 performing what's called the numeric number reduction task.2 It's the type of test that psychologists love to2 administer and participants hate to perform.2 What you're going to do is see lots and lots of2 these number strings.2 And you're going to have to work through them to come up2 with a final end solution.2 Now, one way that you can work through these problems is by2 using some rules that I'll give you.2 The first thing you can see is that there are only three2 numbers here that make up this string, 1, 4, and 9.2 And this is common.2 Thought that the numbers are the same.2 But this notion that there's only ever three numbers in a2 string set.2 That's common.2 And here's what you're going to do.2 You're going to take the first number, compare it to the next2 number, and the first rule is this.2 If this number is the same as the next number, write down2 the very same number, which it is in this case, a 1,2 Now, you've got the 1.2 Compare it to the next number in the line.2 Is it the same number?2 If it is, write down the same number.2 Well, it's not.2 And here's the second rule.2 If it's a different number, write down the only other2 remaining number in the string, which would be a 9.2 So let's repeat that again.2 You can take the 9, compare it to a 4.2 Same or different?2 It's different.2 Write down the only other remaining number, a 1.2 1 to a 9, different.2 Write down 4.2 4 to 4, it's the same number.2 So write down the same number, 4 to a 9, 1.2 1 to a 9.2 Oh, my goodness, is it boring and laborious.2 Now it turns out, and this is exactly how the experiment2 works, if you paid attention to what I said, I told you one2 way to solve these problems is by using those rules.2 Because it turns out there's another way.2 There is a hidden rule.2 There is a shortcut.2 There's a cheat.2 And if you figure it out, you can blow through many more of2 these problems.2 And here's the cheat.2 The second number that you produce in the string is2 always the final answer.2 And so whilst this is different across all the2 problems in terms of the number, the overarching rule,2 the commonality across this2 informational set, is the same.2 So here's what we're going to do.2 We're going to expose a whole collection of participants to2 these problems.2 Then 12 hours later, you're going to bring them back and2 expose them to some more problems.2 And at that 12 hour delay point, you're then going to2 see what proportion of those participants have developed2 insight into that hidden rule.2 Half of those participants are going to remain2 awake across the day.2 Expose them to problems in the morning,2 reexpose them in the evening.2 The other half, they're exposed in the evening.2 They reexpose in the morning to the problems.2 And therefore, they've had a full eight-hour night of sleep2 in between.2 So the brain has had equal amounts of opportunity time to2 distill that informational set and see if it can2 find out the solution.2 The only difference is that one group has had sleep.2 The other hasn't.2 And we're going to put sort of wake and sleep, head to head2 in this Coke-Pepsi challenge to see which one wins out.2 And so here's our outcome metric, the proportion of2 participants in each of those two groups that gained that2 knowledge, that creative insight.2 In the group that remained awake across the day, less2 than 25% of those participants developed that2 hidden insight knowledge.2 What about the sleep group, worse, the same, better?2 Well, of course they were better.2 But what was shocking was how much better.2 This was how much better after sleep.2 2 Over 60% of participants, having slept, developed2 insight into that hidden rule.2 And what we've been finding--2 what I should say is it's almost as though sleep, there2 is an algorithm in sleep that takes vast informational sets2 and starts to try and understand the statistical2 regularities and the rules of those mass data sets.2 It's a huge distillation.2 It's a collision of information, creative2 information processing.2 And we're finding that some, not all, but some of these2 types of associative memory processing occurs during rapid2 eye movement sleep, dreaming sleep.2 And I believe that it's probably not a coincidence2 that this is the stage from which we dream.2 If dreaming is a reflection of whatever information2 processing is going on with the brain, then it may be this2 hypersensitive, hypercreative creative, hyperassociative2 processing that's going on, that leads to2 these creative insights.2 As an aside, many people, when I present this evidence to2 them, will say well, aren't there those sort of creative2 genius types in history who were supposed not2 to sleep very much?2 One of them that's often quoted2 to me is this gentleman.2 Does anybody know who this is?2 AUDIENC dison.2 MATTHEW WALKE dison, exactly.2 What he's holding is a bit of a giveaway.2 A brilliant man of course, supposed2 to be a short sleeper.2 Now, of course, we'll never truly know if he was a short2 sleeper or not.2 But even if he was a short sleeper, it turns out that2 Thomas Edison was a habitual napper during the day.2 Here he is after a pretty good garden party it looks like.2 Here he is on his inventor's bench taking a nap.2 In fact, Edison understood the creative brilliance of sleep2 and he used it as a tool.2 Here's what he would do.2 He would take a metal saucepan,2 like this behind him.2 He would turn it upside down and rest it underneath the2 armrest of his chair.2 Then he would take two steel ball bearings in his hand,2 rest the back of his arm on the chair.2 Take a pad of paper and a pencil, put it next2 to him on his desk.2 And then slowly relax back and fall asleep.2 And so he didn't sleep too long.2 What would happen is that his muscle tone would relax.2 He would release the steel ball bearings.2 They would crash on the saucepan underneath2 him, wake him up.2 And then he would write down all of the ideas that he was2 having from his sleep.2 Isn't that brilliant?2 What a guy.2 So no wonder you're never told you should really stay awake2 on a problem.2 Nobody tells you that.2 Instead that they tell you to sleep on a problem.2 And we're starting to find scientific evidence that2 rigorously backs that up.2 It turns out, and a friend and a colleague told me this, that2 this phrase of "sleeping on a problem" seems to be common in2 most all languages that he's explored to date.2 What that means is that this phenomenon seems to transcend2 cultural boundaries.2 And I should also note that it probably says a lot about the2 difference between me as a British gentleman and our arch2 rivals, the French.2 Because the French translation it turns out of this,2 essentially is not sleeping on a problem.2 It's that you sleep with a problem.2 British, you sleep on a problem.2 French you sleep with a problem.2 And it turns out that the politics, the people in2 politics, reflect this.2 If you look at the past president Mr. Sarkozy and Mrs.2 Sarkozy, these are the press release2 pictures that they offer.2 She's draped on a bed.2 He's looking forlorn at her.2 Whereas the people in British politics, who did we have?2 Well, we had Margaret Thatcher.2 We had sort of Tony Blair.2 You sleep with a problem.2 You sleep on a problem.2 I'll say no more.2 Before--2 I'm probably never going to be able to go back to the UK now2 after that.2 Beyond information processing, of which now there is good2 evidence for in terms of sleep dependency, we're now starting2 to realize there's another brain function of sleep.2 And that is in preparing the emotional circuits of the2 brain, offering you stable mental health.2 Now, I think many of us have a sense that these two factors2 of sleep and emotion interact in some2 meaningful kind of way.2 An example would be a parent holding a child,2 the child is crying.2 And they look at you and they say well, you just didn't2 sleep well last night.2 As if there's some universal parental knowledge that bad2 sleep the night before equals bad mood and emotion2 reactivity the next day.2 We also know clinically that these factors interact in that2 nearly all psychiatric mood disorders display co-occurring2 abnormalities of sleep.2 In fact, these sleep abnormalities are so prominent2 they form part of the diagnostic criteria for those2 psychiatric disorders.2 But despite that suggested interplay, we've known2 remarkably little about the basic brain dynamics of this2 relationship.2 And that's something that we've also been testing.2 When you think you've got two factors that are interacting,3 one way to test that interaction is to manipulate3 one of the factors and then observe what3 happens to the other.3 So here we're going to manipulate sleep and dial it3 down again and block it with deprivation and see if as a3 consequence, we can trigger an amplified emotional brain3 reaction as a consequence.3 So a very similar design to one I showed you before, a3 sleep group and a deprivation group.3 The deprivation group, we keep awake.30:2