Andrew Huberman interviews sleep researcher Dr Gina Poe on the science of sleep stages, memory consolidation, and emotional processing
Andrew Huberman speaks with UCLA neuroscientist Dr Gina Poe about the architecture and functions of sleep.
Summary
Andrew Huberman interviews Dr Gina Poe, a neuroscientist specializing in sleep research, in this Huberman Lab Essentials episode. Dr. Poe explains the four stages of sleep — N1, N2, slow-wave sleep, and REM — and describes how each serves distinct biological functions including memory consolidation, brain cleaning, growth hormone release, and emotional processing. She argues that consistent bedtimes are as important as consistent wake times, and that missing the first sleep cycle causes irreversible loss of a critical growth hormone bolus and slow-wave brain-cleaning activity. A central claim of the episode is that the locus coeruleus — the brain's norepinephrine center — must go completely silent during REM sleep for the brain to erase outdated memory traces and divorce emotional charge from traumatic memories, and that its failure to do so is a key mechanism underlying PTSD.
Key Takeaways
FULL TRANSCRIPT
Overview of Sleep Stages and the Architecture of a Full Night's Sleep
Andrew Huberman: Welcome to Huberman Lab Essentials, where we revisit past episodes for the most potent and actionable science-based tools for mental health, physical health, and performance. I'm Andrew Huberman and I'm a professor of neurobiology and ophthalmology at Stanford School of Medicine. And now for my discussion with Dr Gina Poe. Dr Gina Poe, welcome.
Dr Gina Poe: Thank you.
Andrew Huberman: I've really been looking forward to this conversation. I know that many people are going to be excited to learn about your work as it relates to sleep, as it relates to problem solving, creativity, and a number of other important topics. To start things off, I would love for you to educate us a bit about this thing that we are all familiar with and yet very few of us understand, which is sleep. Could you describe the various phases of sleep that exist, what distinguishes them, and perhaps frame this within the context of what a perfect night's sleep would look like?
Dr Gina Poe: Sleep is really different from wakefulness and in fact can't be replaced by any state of wakefulness that we've been able to come up with so far. Our brain chemistry is completely different, and in the different stages of sleep — of which there are two major states, non-REM and REM — those two states are entirely different from one another. And even within non-REM there are three stages. Stage one is what you slip into when you first fall asleep. It's dozing. There's a kind of interesting rhythm that goes on in the brain — a fast gamma rhythm. Then there's stage two, which is a really cool state. Sleep researchers used to sort of ignore it because it was considered a transient state between wakefulness and the deep stage three slow-wave sleep, which is the most impressively different state — that's when big slow waves sweep through our brain. We've now realized that it cleans our brain. One of the things those big slow waves do is clean our brain and do other really important things to restore us from a day of wakefulness. And then REM sleep, which is the most popular because that's where we have the most active dreams. When you wake someone out of REM sleep, they'll almost always report having dreamed something really bizarre. That's called REM sleep — rapid eye movement sleep. So those are the four states of human sleep, and we cycle through them every 90 minutes or so and then start over again. We have about four or five of those cycles per night for a perfect night's sleep. So a perfect night's sleep is seven and a half to eight hours.
Andrew Huberman: What about the sleep where we are lightly asleep and we might have a dream that has us thinking about movement, or that jolts us awake? That often happens early in the night, right?
Dr Gina Poe: Yes. That's the first stage — stage one and stage two of sleep. Stage two sleep is really cool because it has something called sleep spindles and K-complexes. Sleep spindles are bursts of activity at 10 to 15 hertz in frequency. It's a conversation between the thalamus and the cortex. The thalamus is the gateway to consciousness, and the neocortex processes all our cognition. If you wake up out of that state, you will often report a dream-like, hallucination-style experience. It won't be a long dream report like you get out of REM sleep, but it will be some kind of hallucinatory state.
Memory Consolidation Across the Night
Andrew Huberman: Are the patterns of sleep and dreaming that occur early in the night quite different from those that occur later in the night or toward morning?
Dr Gina Poe: There is some evidence that the first four hours of sleep are very important for memory processing. If you've learned something new that day or have experienced a new sensorimotor experience, your early sleep dreams will incorporate that experience much more than the later sleep dreams. Later, as that memory gets consolidated from the early structures — the hippocampus deep in the temporal lobe — to the cortex in a distributed fashion, the memory seems to move from the hippocampus to the cortex, and the dreams that incorporate that memory also move later in the night.
There was a great study by Sidarta Ribeiro who studied the consolidation of memories from the hippocampus to the cortex in a rat across the period of a full day's sleep, because rats sleep in the daytime. He found that each subsequent REM sleep period moved that memory from the hippocampus to the first area that projects to it, then the second area, then the third area. You can actually see the memory moving throughout the sleep.
Growth Hormone Release and the First Sleep Cycle
Andrew Huberman: There are a number of different hormones associated with the different stages of sleep. We know that melatonin is a hormone of nighttime that makes us sleepy. What about growth hormone release? When does that occur during sleep?
Dr Gina Poe: Growth hormone release happens all day long and all night long. But the deep slow-wave sleep that you get in the very first sleep cycle is when you get a big bolus of growth hormone release — in men and women equally. If you miss that first deep slow-wave sleep period, you also miss that big bolus of growth hormone release. You might ultimately get just as much overall growth hormone release across the day, but endocrinologists will tell you that big boluses do different things than a little bit eked out over time. There's also a big push to synthesize proteins during that time. That's when the protein synthesis that builds memories in our brain happens — in that first cycle of sleep. So you don't want to miss that, especially if you've learned something really significant that needs more synaptic space to encode it.
Andrew Huberman: How would somebody miss that first 90 minutes? By depriving themselves? Let's say I normally go to sleep at 10:00 p.m., and from 10 to 11:30 would be this first phase of sleep — that's when the big bolus of growth hormone would be released. Does that mean that if I go to sleep instead at 11:30 or midnight, I miss that first phase of sleep? Why is it not the case that I simply get that first phase of sleep starting later?
Dr Gina Poe: Every cell in our body has a clock, and all of those circadian clocks are synchronized. So our cells are ready to respond to that growth hormone release at a particular time, and it's also timed in relation to melatonin. If you miss it, you might get some growth hormone release, but it's occurring at a time when your clock has already moved to the next phase. So it's just a clock thing.
Andrew Huberman: So what this means is that we should have fairly consistent bedtimes in addition to fairly consistent wake times. Is that right?
Dr Gina Poe: Exactly. And in fact, one of the best markers of good neurological health as we get older is consistent bedtimes.
Andrew Huberman: What other things inhibit growth hormone release or other components of this first stage of sleep? Are there things I might do in the preceding hours or the preceding day — like ingest caffeine or alcohol — that can make that first stage of sleep less effective even if I'm going to sleep at the same time?
Dr Gina Poe: Alcohol definitely will do that, because alcohol is a REM sleep suppressant and it even suppresses some of that stage two transition to REM with those sleep spindles. And those sleep spindles are really important for moving memories to our cortex. It's a unique time when our hippocampus — the RAM of our brains — writes to a hard disk, which is the cortex. It's a unique time when they're connected. So if you don't want to miss that, you don't want to miss REM sleep, which is also part of the consolidation process. Alcohol before we go to sleep will affect our sleep badly until we've metabolized it and put it out of our bodies.
Andrew Huberman: What about the second and third 90-minute blocks of sleep? What is their signature besides the fact that they come second and third in the night?
Dr Gina Poe: There's more and more REM sleep the later in the night we get. There's also a change in hormones — growth hormone and melatonin levels are starting to decline, but other hormones are picking up. It is a really different stage that you also don't want to short-change yourself on. Many studies are showing that those are the times in sleep when the most creativity can happen. That's when our dreams can incorporate and put together old and new things in a new way, and our schemas are built during that time. That's when your brain is opening folders and comparing documents, seeing if there's anything the same — are these two documents very similar with just a little bit of difference? It can link those conceptually. That's probably one of the origins of creativity: finding things that are related, maybe just linked a little bit, and you can find that link and strengthen it if it makes your schema interesting and different.
Waking During the Night and Sleep Architecture
Andrew Huberman: Many people, including myself, tend to wake up maybe once during the middle of the night to use the restroom. I've tried to drink less fluid before going to sleep. I've also heard that the impulse to urinate is dictated by how quickly you drink fluid, not just the total volume. So I've switched to sipping fluids more slowly for my last beverage of the day, which seems to help. Is there any known detriment to this middle-of-the-night waking, or should we consider it a normal feature of some people's sleep architecture?
Dr Gina Poe: I think we shouldn't worry about it. Sleep is really incredibly well homeostatically regulated. Don't worry about how much you're sleeping as long as you're not intentionally depriving yourself of sleep by doing something really rewarding and exciting, because even that is stressful to your body and deprives you of a lot of the things we're talking about. It's absolutely normal to wake up at least once in the middle of the night to go to the bathroom. As long as you can get back to sleep in a reasonable amount of time — or even if it takes you an hour, don't worry about it — as long as you have a lifestyle that allows you to make up that sleep either the next morning, the next night, or by going to bed a little earlier.
Andrew Huberman: What is unique about the architecture of dreams and sleep in the last third of the night, or the second half of the night?
Dr Gina Poe: In the second half of the night, we have longer REM sleep periods, and those are considered the deepest sleep. Even though slow-wave sleep — with its big slow waves — is called deep sleep, it is deep.
Andrew Huberman: They call slow-wave sleep "deep sleep" and REM sleep "rapid eye movement sleep." But now you're telling me that REM sleep is actually the deeper sleep?
Dr Gina Poe: The reason slow-wave sleep is called deep sleep is because it's difficult to arouse people out of that state, and when you do arouse them, they're most often confused and just want to go back to sleep. If you arouse someone out of REM sleep, they're more likely to report something that was really almost like wakefulness — it was so vivid. And maybe one of the reasons why REM sleep is considered deeper, especially in adults and older people, is that deep slow-wave sleep goes away with age. The slow waves aren't as large, which is probably problematic, though we're not sure. So REM sleep becomes the deepest stage.
Paralysis, Sleepwalking, and Sleep Inertia
Andrew Huberman: We are paralyzed during REM sleep. Correct?
Dr Gina Poe: Yes, normally paralyzed, and that's really good because that's the time when we're actively having storyline dreams and we could hurt ourselves. We're actually really cut off from the outside world in terms of responding to it. This is different from sleepwalking, which occurs out of slow-wave sleep. Sleepwalking is a mixture between sleep and wakefulness — you can cook a full meal, drive your car while in deep slow-wave sleep. It's scary because you have no conscious control over it, but you can actually safely navigate some situations and even have a conversation, although it may not make much sense. In REM sleep, you're not processing the outside world, and when you act out your dreams you could be walking through a plate glass window or falling down stairs. So you really want your muscles to be inactivated during REM sleep, otherwise you will act out those dreams and really hurt yourself or your bed partner.
Andrew Huberman: As people approach morning or the time when they normally would wake up, I've heard that it's important to, if possible, complete one of these 90-minute cycles prior to waking up — that setting your alarm for halfway through one of these late-night cycles can lead to rather groggy patterns of waking.
Dr Gina Poe: It's called sleep inertia. When we wake up out of the wrong state, I liken it to a washing machine cycle. This 90-minute cycle is like a washing machine cycle — the first part is to add water, and then your clothes are soaking wet. You don't want to open the washing machine and try to put them on while they're soaking wet and full of soap. You have to wait until the cycle is through before you can put them in the dryer and actually wear them. So it's better to wait until the whole cycle is complete. That's why you want to set your alarm to a 90-minute multiple. The first cycle of sleep is actually a little longer — more like 105 to 110 minutes — but then the second and third ones get shorter as the night goes on. In the last few cycles, you're just doing the N2-REM cycle, which takes less time. If you wake up out of REM sleep, there's usually no cognitive problem.
Andrew Huberman: Are you a fan of sleep trackers?
Dr Gina Poe: Yeah. Do you use one?
I have one on. I don't live my life by them, because the best ones right now are about 70% effective at staging your sleep. So it's okay, but take it with a grain of salt.
The Brain's Cleaning Process During Slow-Wave Sleep
Andrew Huberman: Tell us a little more about the washout that occurs in the brain during sleep, and perhaps whether there are any ways to ensure that it happens — and obviously we want this to happen.
Dr Gina Poe: We talked about the circadian clock and how certain things happen at certain times. One of the things that happens when we're awake and talking to each other is that there's a lot of plasticity — there's something I'm learning from you today and you're learning from me, and that changes our synapses and the way our proteins are going to be folded and changed during sleep. This process actually uses a lot of ATP, the fuel of the brain. It also unfolds proteins while we're using them. During the first part of the night, in the first 20 minutes or so after we fall asleep, we're rebuilding that adenosine back into ATP. That's probably why power naps are called power naps — we're actually rebuilding the power. Then we're also cleaning out, through the deep slow waves of slow-wave sleep, all those misfolded proteins, unfolded proteins, and other things that get broken down and need to be rebuilt. I liken that to having a big party during wakefulness and needing all those partygoers to leave in order to do the cleanup.
What happens when a neuron is firing is that it expands — the membrane expands a little bit, becomes more translucent. That's one of the ways we know neurons expand when they fire. Every action potential, the membrane expands a little as sodium brings water into the cell. Then when they're silent, they contract. During slow waves, a good portion of neurons are firing at the same time and silent at the same time. You can think of that as contracting and expanding all at the same time — it's like a bilge pump of the brain, pumping out debris. Glia are also really important for this in terms of cleaning up debris and transferring it to where it needs to go. So I think of it as a bilge pump cleaning out our brain.
Andrew Huberman: You're talking about literally an expansion and contraction of the neurons in unison, pushing the fluid through and cleaning out any misfolded proteins or debris that might occur on the basis of these metabolic pathways. And the consequence of that is to leave the brain in a state of more pristine function for the next day. Is that right?
Dr Gina Poe: Yes. Think of it again like a party — if you don't clean up after that party and you try to hold another one the next day, it's going to get more clogged. People have a harder time moving around and enjoying themselves. If that builds up day after day, cognition — the partygoers moving around — becomes hard.
Andrew Huberman: So this bilge pump is associated with the big slow waves of slow-wave sleep, which occurs more or less in the first third of the night. Is that right?
Dr Gina Poe: That's right.
Andrew Huberman: And is this similar to the case with growth hormone, where if you go to sleep later than you normally would, you miss the washout — you don't delay it, you miss it?
Dr Gina Poe: That's right. If you go to sleep at 1 or 2 in the morning, your sleep is still going to be dominated by N2 and REM sleep, not by slow-wave sleep. So you need to get that first bit of sleep.
Andrew Huberman: Would a caveat to that be if somebody normally goes to sleep at 1 or 2 a.m. and wakes up at 10:00 a.m. — if that's their normal sleep cycle?
Dr Gina Poe: Yeah, it should be okay. Someone would want to do a sleep study with people who do that normally and see if the melatonin release is also later, and if the cortisol rise that normally happens in the morning also happens later. So if everything is shifted, good.
The Locus Coeruleus, Norepinephrine, and REM Sleep
Andrew Huberman: I'd love for you to tell us about this incredible structure in the brain — the locus coeruleus — and hopefully tell us a little bit about its relationship to epinephrine, also known as adrenaline.
Dr Gina Poe: The locus coeruleus is filled with neurons that contain norepinephrine, which is the brain's version of epinephrine or adrenaline — also called noradrenaline. Just like adrenaline in the rest of our bodies, it helps prime us to respond to our environment. When locus coeruleus neurons fire in a burst — for example, if a loud noise happens while you're concentrating on something — it fires and helps you switch your attention to that thing and then learn quickly from it. So it's really important in a stress response. It helps us do quick one-trial learning. Tonic levels are the signature of wakefulness and alertness — too much is panic, a burst is attention-switching, and sustained tonic levels are sustained constant attention.
When we go to sleep, the locus coeruleus slows from about two hertz on average to about one hertz tonically. Then when we go into REM sleep, it's the only time when it shuts off completely. That complete silence appears to be really, really important for a number of things. The main thing I think it's important for is the ability to erase and break down synapses that are no longer working for us — synapses that encode things that are now false, or that are encoding things we learned in the novelty encoding pathway of our brain that have now been consolidated to other pathways. We need to erase them from the novelty encoding pathway, and that is really, really important for being able to continue to learn things throughout our lives.
It's like erasing a thumb drive. You carry that thumb drive around all day long, and then during sleep you write that thumb drive to the cortex — to the long-term memory structures — and you need to refresh that thumb drive. That's what happens during REM sleep when the locus coeruleus is off. If we're not able to do that, we fill up that RAM very quickly and we're not able to learn new things. Even memories that are years past — if you're never able to downscale that novelty encoding structure and purge it from a traumatic memory, it will stay fresh and new and then become maladaptive.
Andrew Huberman: What approaches are you aware of that can turn down the output of the locus coeruleus during these phases of sleep and allow the late stages of sleep each night to have their maximum positive effect? Is there anything I can do besides avoiding serotonergic or noradrenergic compounds?
Dr Gina Poe: I would also avoid anything just prior to going to sleep that might excite those systems — a lot of novelty, stress-inducing video games, for example. Try to enter sleep with as much calm as you can. Deep breathing exercises are a beautiful way to calm your sympathetic fight-or-flight system. If there's a way you can make your sympathetic nervous system calm down before you go to sleep — whether that's meditation or deep breathing exercises, a warm bath, or a comforting book, nothing too exciting but also nothing too boring, just something right in the middle that makes you feel happy and calm — that's what you should do. If you instead go to sleep while you're anxious or hyped up, your sleep could become maladaptive.
Sleep Spindles, P-Waves, and the Origins of Creativity
Andrew Huberman: I'd love for you to share a little more about these spindles that have come up a few times.
Dr Gina Poe: The density of our sleep spindles — the number produced per minute — is well correlated with our intelligence in the first place. And no matter what your intelligence is and no matter what your sleep spindle density is, if you learn something during the day and increase your sleep spindle density, it's almost perfectly correlated with your ability to consolidate that information and incorporate it into the schema you already have in your brain. So if you try to learn something new, even if your sleep spindle density at baseline is great, if you don't increase your spindles that night, you're not going to use sleep to really incorporate it.
Through some great studies by Julie Seibt and Anita Lüthi, we now know that sleep spindles are accompanied by incredible plasticity out in the distal dendrites — the listening branches of our neurons that listen to other cortical areas. There are proximal dendrites that listen to the external world and are conducted through the thalamus, and then there are distal dendrites that listen to an internal conversation happening in our brains — our internal state, really. During sleep spindles, those distal dendrites are best able to learn from other cortical areas and from the hippocampus. It is during sleep spindles that the hippocampus and the cortex are best connected and when that incredible plasticity can happen. There are big surges of calcium into those distal dendrites during the N2 stage of sleep.
There's another excitatory event that comes from the brainstem and projects everywhere in our cortex, called PGO waves — or more generally P-waves, because they come from the pons and go to the thalamus and then the cortex. That is where glutamate, a major excitatory neurotransmitter involved in learning and plasticity, is being released in large amounts in those distal dendrites. P-waves and spindles work together to cause plasticity and sew our schemas together, which could be the origin of insight and creativity.
When P-waves were first discovered, they were thought to be random, because the small area that generates them projects all over the thalamus and causes P-waves all over the brain — but you don't measure them all over the brain at the same time. They seem sporadic and random. P-waves also happen even more during REM sleep. That's why people think REM dreams are so random — because these P-waves are random and could generate dreams, being an internal source of excitation that kind of replaces the outside world during our dream state. If P-waves are random, that could be the underlying reason why creativity can happen there: we're randomly co-activating different things in our brain that we can then sew together. But it might not be as random as we think — that's a caveat.
REM Sleep, Trauma, and Emotional Processing
Andrew Huberman: So the locus coeruleus is suppressed, we can't release norepinephrine, we can't act out our dreams. This almost starts to sound like a built-in, while-sleeping trauma therapy. Please, if there's anything about the locus coeruleus and dreams that can help people basically extinguish traumas or traumatic features to real-life events, we definitely want to know about it.
Dr Gina Poe: Well, one of the things that people thought might help after a trauma — like a car accident or a school shooting — is to talk about it. But in fact that ended up being counterproductive. I think one of the reasons why it was counterproductive is because it didn't bring them back down. It brought them up and continued to reactivate the emotions of it, but didn't emphasize the safety fact that it's over, or help them work through how they might avoid it again in the future to calm the sympathetic nervous system down again before they went to sleep. In none of these studies has sleep ever been considered, but to me that's the key part — bringing down your sympathetic nervous system before you go to sleep so that your sleep can be adaptive, your locus coeruleus can shut off like it normally should, and then be able to erase the novelty of the experience.
The other thing I mentioned a minute ago is that the emotional system is highly activated in REM sleep. That might seem counterproductive in terms of the nightmares and how to help REM sleep be therapeutic rather than reinforcing the emotionality of the trauma. I think the key to that again is the absence of norepinephrine. Even though the emotional system is in high gear, without norepinephrine you can actually divorce those highly activated emotions from the cognitive parts of the memory that you've just written out in that N2 stage of sleep when the sleep spindles are going. You've just consolidated the information you'll need to survive and make that experience adaptive, and now you need to divorce from that schema and from the semantic parts of memory the emotional part. Because whenever you remember something, it's fine to remember being emotional at the time, but you don't want to bring back all of the same emotional systems — the heart rate changes, the sweating, all of that. You want to be able to remember all the parts of it, and even remember that you were traumatized and that you did cry and that your heart was racing. But when you're talking about it years later, you don't want to have to relive all that. Otherwise, who would ever want to recall a traumatic memory? Because you're basically putting yourself through the same trauma — which is what people with PTSD have. They don't want to recall the traumatic memory because it's reliving it like it's just happening again.
So what we're thinking is that the emotional parts are not able to be divorced because the norepinephrine system is not downscaled during REM sleep. That REM sleep then serves to instead reinforce and in fact amplify the emotions, because your emotional system is up and the locus coeruleus is re-sewing in every night the emotionality of those memories together with the memory itself.
Andrew Huberman: I must say, you've taught us a tremendous amount in a relatively short amount of time about the architecture of sleep, the different phases — such a wealth of information, and much of it actionable for people. So I want to say thank you for taking the time to sit down and have this conversation that so many people are sure to benefit from. I know I speak for everybody when I say thank you so much.
Dr Gina Poe: Thank you so much.