Matt Walker: Sleep | Lex Fridman Podcast #210

Lex Fridman 0:00 The following is a conversation with Matt Walker, sleep scientist, professor of neuroscience and psychology at Berkeley, author of Why We Sleep, and the host of a new podcast called The Matt Walker Podcast. It's ten-minute episodes a couple of times a month, covering sleep and other health and science topics. I love it and recommend it highly. It's up there with the greats, like the Huberman Lab Podcast with Andrew Huberman, and I think, uh, David Sinclair is putting out a, an audio series soon too. I can't wait to listen to it. I'm really excited by the future of science in the podcasting world. To support this podcast, please check out our sponsors, Stamps.com, Squarespace, Athletic Greens, BetterHelp, and Onnit. Their links are in the description. As a side note, let me say that, to me, a healthy life is one in which you fall in love with the world around you, with ideas, with people, with small goals and big goals, no matter how difficult, with dreams you hold onto and chase for years. Life should be lived fully. That, to me, is the priority. That, to me, is a healthy life. Second to that is the understanding and the utilization of the best available science on diet, exercise, supplements, sleep, and other lifestyle choices. To me, science in the realm of health is a guide for what we should try, not the absolute truth of how to live life. The goal is to learn to listen to your body and figure out what works best for you. All that said, a good night's sleep can be a great tool in making life awesome and productive, and Matt is a great advocate of the how and the why of sleep. We agree on some things and disagree on others, but he's a great human being, a great scientist, and as of recently, a friend with whom I enjoy having these wide-ranging conversations. This is the Lex Fridman Podcast, and here is my conversation with Matt Walker.

Lex Fridman 2:05 You should try these shades on and see what you look like.

Matt Walker 2:11 So they are now your shades?

Lex Fridman 2:12 (laughs)

Matt Walker 2:13 And that's not a question.

Lex Fridman 2:14 It's the same thing as, uh, Putin took the Super Bowl ring and it's now his ring.

Matt Walker 2:19 (laughs)

Lex Fridman 2:20 (laughs)

Matt Walker 2:20 Uh, yeah, one wonders if he was offered it, but, um, they are yours.

Lex Fridman 2:24 (laughs)

Lex Fridman 2:26 When did you first fall in love with the dream of understanding sleep? Like-

Matt Walker 2:33 Mm-hmm.

Lex Fridman 2:33 ... where did the fascination with sleep begin?

Matt Walker 2:38 So back in the United Kingdom, you can sort of start doing medicine at age 18, and it's a five-year program, and I was at, uh, the Queen's Medical Center in the UK, and I remember just being fascinated by states of consciousness, and particularly anesthesia. I was thinking, "Isn't that inc- Within seconds, I can take a perfectly conscious human being, and I can remove all existence of the mentality and their awareness within seconds," and that stunned me. So I started to get really interested in conscious states. I even started to read a lot about hypnosis, (laughs) um, and all of these things, hypnosis, even sleep and dreams at the time, they were very esoteric. It was sort of charlatan science at that stage, and I think almost all of my colleagues and I are accidental sleep researchers.

Lex Fridman 3:36 (laughs)

Matt Walker 3:36 You know, (laughs) no one, as I recall, in the classroom, when you're sort of five years old and the teacher says, "What would you like to be when you grow up?" You know, no one's putting their hand up and saying, "I would love to be a sleep researcher." (laughs) Um, and so when I was doing my PhD, I was trying to identify different forms of dementia very early on in the course, and I was using electrical brain wave recordings to do that, and I was failing miserably. It was a disaster, just no result after no result. And I used to go home to the doctor's residence with this sort of little igloo of journals that, at the weekend, I would sort of sit in and, uh, and read, and, (laughs) which I'm now thinking, "Do I really want to admit this?" 'Cause it sounds-

Lex Fridman 4:18 (laughs)

Matt Walker 4:18 ... like I had no social life, which I didn't.

Lex Fridman 4:20 Yeah.

Matt Walker 4:20 I'm a social leper, but, and I started to realize that some parts of the brain were, um, sleep-related areas and some dementias were eating away those sleep-related areas. Other dementias would leave them untouched. And I thought, "Well, I'm doing this all wrong. I'm measuring my patients while they're awake." Instead, I should be measuring them while they're asleep. Started doing that, got some amazing results, and then I wanted to ask the question, "Is that sleep, um, disruption that my patients are experiencing as they go into dementia, maybe it's not a symptom of the dementia. I wonder if it's a cause of the dementia." And at that point, which was, (clears throat) (laughs) cough, cough, 20 years ago, um, no one could answer a very simple fundamental question, "Why do we sleep?" And I, at that time, didn't realize that some of the most brilliant minds in scientific history had tried to answer that question and failed, and at that point, I just thought, "Well, I'm going to go and do a couple of years of sleep research, and I'll figure out why we sleep, and then I'll come back to my patients and this question of dementia." And as I said, that was 20 years ago, and what I realized is that hard questions care very little about who asks them. They will meter out their lessons of difficulty all the same, and I was schooled (laughs) in the difficulty of the question, "Why do we sleep?" But in truth, 20 years later, we've had to upend the question. Rather than saying, "Why do we sleep?" And by the way, the, the answer then was that we sleep to cure sleepiness. (laughs) Which is like saying-

Lex Fridman 6:04 Right.

Matt Walker 6:05 ... "W- you know, we eat to cure hunger."

Lex Fridman 6:06 Yeah.

Matt Walker 6:07 That tells you nothing about the physiological benefits of, of food. Same with sleep. Now, we've, uh, actually have to ask the question, is there any physiological system in the body, or any major operation of the mind that isn't wonderfully enhanced when we get sleep or demonstrably impaired when we don't get enough? And so far, for the most part, the answer seems to be no.

Lex Fridman 6:30 So far the answer seems to be no, so th- why,

Lex Fridman 6:35 why does the body and the mind crave sleep then? Why do we sleep? D- how, um, can we begin to answer that question then?

Matt Walker 6:45 So I think one of the ways that I think about this, or one of the answers that came to me is the following. The reason that we implode so quickly and so thoroughly with insufficient sleep is because human beings seem to be one of the few species that will deliberately deprive themselves of sleep for no apparent good reason, biological.

Lex Fridman 7:06 Mm-hmm.

Matt Walker 7:07 And what that led me then to was the following. Mother Nature as a consequence ... So no other species does what w- we do in that context. There are a few species that do undergo sleep deprivation but for very obvious, clear biological reasons. One is when they're in a condition of severe starvation. The second is when they're caring for their newborn. So, for example, killer whales will often deprive themselves. The female will go away from the pod, give birth, and then bring the calf back, and during that time, the mother will undergo sleep deprivation. And then the, the third one is during migration when birds are flying transoceanographic, 2-3,000 miles. But for the most part, it's never seen in the animal kingdom, which brings me back to the point, therefore, Mother Nature in the course of evolution has never had to face the challenge of this thing called sleep deprivation, and therefore she has never created a safety net in place to circumnavigate this common influence.

Lex Fridman 8:15 Mm-hmm.

Matt Walker 8:15 And there is a good example where we have, which is called the adipose cell, the fat cell, because during our evolutionary past, we had famine and we had feast, and Mother Nature came up with a very clever recipe, which is, "How can I store caloric credit so that I can spend it when I go into debt?" And the fat cell was born, brilliant idea. Where is the fat cell for sleep? (laughs) Where is that sort of banking chip for sleep? And unfortunately, we don't seem to have one because she's never had to face that challenge.

Lex Fridman 8:49 So even if there's not some kinda physics fundamental need for sleep, that, uh, physiologically or psychologically, the f- the fact is most organisms are built such that they need it, and then Mother Nature never built an extra mechanism for sleep deprivation. So it's interesting that why we sleep might not have a good answer- (laughs)

Matt Walker 9:14 (laughs)

Lex Fridman 9:15 ... but we need to sleep to be healthy is nevertheless true.

Matt Walker 9:19 Yeah, and we have many answers right now. In some ways, the question of why we sleep was the wrong question too. It's, you know, what are the pluripotent many reasons we sleep? We don't just sleep for one reason because from an evolutionary perspective it is the most idiotic thing that you could imagine.

Lex Fridman 9:39 Yeah.

Matt Walker 9:39 You know, when you're sleeping, you're not finding a mate, you're not reproducing, you're not caring for your young, you're not foraging for food, and worse still, you're vulnerable to predation. So on any one of those grounds (laughs) , but especially as a collective, sleep should've been strongly selected against in the course of evolution. But in every species that we've studied carefully to date, sleep is present.

Lex Fridman 10:05 Yeah, so it is important. So, like, you're, you're right. I, I think I've heard arguments from an evolutionary biology perspective that sleep is actually advantageous, you know, maybe like some kinda predator-prey relationships.

Matt Walker 10:18 Yeah.

Lex Fridman 10:18 But you're saying, and it actually makes way more sense what you're saying, is it should've been selected against. Like, why close your eyes? (laughs)

Matt Walker 10:27 Yeah. Why? Because ... And s- you know, there was an energy conservation hypothesis-

Lex Fridman 10:31 Yeah, right.

Matt Walker 10:31 ... for a while, which is that we need to essentially go into low battery mode, you know, power down because it's unsustainable, but in fact that actually has been blasted out the water because sleep is an incredibly active process. In fact, the difference between you just lying on the couch but remaining conscious versus you lying on the couch and falling asleep, it's only a savings of about 140, 150 calories. In other words, you know, you just go out and club another baby seal or whatever it was.

Lex Fridman 11:01 Yeah. (laughs)

Matt Walker 11:01 And you wouldn't worry, you know? So there has to be much more to it than energy conservation, much more to it than sharing, you know, ecosystem space and time, much more to it than simply predator-prey relationships. If sleep really did ... And, you know, looking back, even very old evolutionary organisms like earthworms, millions of years old, they have periods where they're active and periods where they're passively asleep. It's called lethargicus. (laughs) And so what that in some ways suggested to me was sleep evolved with life itself on i- this planet, and then it has fought its way through heroically every step along the evolutionary pathway, which then leads to the sort of famous, um, sleep statement from a researcher, that if sleep doesn't serve an absolutely vital function or functions, then it's the biggest mistake the (laughs) evolutionary process has ever, ever made. And we've now realized Mother Nature didn't make a spectacular blunder with sleep.

Lex Fridman 11:59 Hmm. You've mentioned the idea of conscious states.Do you think of sleep as a fundamentally different conscious state than awakeness?

Matt Walker 12:11 Mm-hmm.

Lex Fridman 12:11 And how many conscious states are there? So when you're into it, you're understanding of what the mind can do, do you think awake state, sleep state, or is there some kind of continuum, there's a complicated state transition diagram? Like, how do you think about this whole space?

Matt Walker 12:28 I think about it as a state space diagram, and I think it's probably more of a continuum than we have believed it to be or suggested it to be. So we used to think absent of anesthesia that there were really three main, main states of consciousness. There was being awake, being in non-rapid eye movement sleep or non-dream sleep, and then being in rapid eye movement sleep or dream sleep, and those were the three states within which your brain could percolate and be conscious. Uh, you know, conscious during non-REM sleep is maybe a stretch to say, but I still believe there is plenty of consciousness there. I don't believe that, though, anymore, and the reason is because we can have daydreams, and we are in a very different wakeful state in those daydreams than we are when we are as we are now together, present and exteroceptively focused-

Lex Fridman 13:27 Mm-hmm.

Matt Walker 13:27 ... rather than introceptively focused. And then we also know that as you are sort of progressing into those different stages of sleep, during non-REM sleep, you can also still dream. Depends on your definition of dreaming, but we seem to have some degree of dreaming in almost all-

Lex Fridman 13:46 Mm-hmm.

Matt Walker 13:46 ... stages of sleep. We've also then found that when you are sleep-deprived, there, even individual brain cells will fall asleep-

Lex Fridman 13:56 Hm.

Matt Walker 13:56 ... despite the animal being, you know, behaviorally, from best we can tell, awake, individual brain cells and clusters of brain cells will go into a sleep-like state, and humans do this too. When we are sleep deprived, we have what are called micro sleeps, where the eyelid will partially close and the brain essentially falls, lapses into a state of sleep, but behaviorally, you seem to be awake, and the danger here is road traffic accidents. So these are the, what we call these sort of micro sleep, uh, events at the wheel. Now, if you're traveling at 65 miles an hour in a two-ton vehicle, you know, it takes probably around one second to drift from one lane to the next, and it takes two seconds to go completely off the road. So if you have one of these micro sleeps at the wheel, you know, it could be the last micro sleep that you ever have. Um, but I don't now see it as a set of, you know, very binary, distinct, you know, step function states. It's not a one or a zero. I see it more of a, as a

Matt Walker 15:06 continuum, yeah.

Lex Fridman 15:07 S- so, I've for, for, uh, five, six years at MIT really focused on this human side of driving question, and one of the big concerns is the, uh, micro sleeps, drowsiness, these kinds of ideas. And one of the open questions was, is it possible through computer vision to detect ... or any kind of sensors. The nice thing about computer vision is you don't have to have direct contact to the person. Is it possible to detect increases in, uh, drowsiness? Is it possible to detect these kind of micro sleeps or actually just sleep in general?

Matt Walker 15:46 Yeah.

Lex Fridman 15:46 Um, uh, among other things, like distraction, th- these are all words that have so many meanings, there's so many debates like-

Matt Walker 15:53 Mm-hmm.

Lex Fridman 15:53 ... like attention is a, is a whole nother one. Just because you're looking at something doesn't mean you're loading in the information. Just because you're looking away doesn't mean your peripheral vision can't pick up the important information. There's so many complicated vision science things there. Um, so I, I wonder if you could say something to ... Uh, you know, they say the eyes are the windows to the soul. Do you think, um, the eyes can reveal something about, uh, sleepiness, uh, through, uh, through computer vision, through, through just looking at the video of the face? And Andrew Huberman and I, your friend-

Matt Walker 16:30 Oh.

Lex Fridman 16:30 ... have talked about this.

Matt Walker 16:31 He's a brilliant man. Brilliant man.

Lex Fridman 16:31 So I would love to work on this, uh, together. It's a-

Matt Walker 16:33 You should do.

Lex Fridman 16:34 It's a fascinating problem. But drowsiness is a tricky one. So there's what kind of information? There's, uh, blinking and there's eye movement, and those are the ones that can be picked, uh, up with computer vision. Do you think those are signals that could be used to say something about where we are in this continuum?

Matt Walker 16:52 Yeah, I do, and I think there are a number of other features too. I think, um, you know, aperture of eye, so in other words, partial closures, full closures, um, duration of those closures, duration of those partial closures of the eyelid. Um, I think there may be some information in the pupil as well, because as we're transitioning between those states ch- there are changes in what's called the automatic nervous system, or technically it's called the autonomic nervous system, part of which will control your pupillary size.

Lex Fridman 17:26 Mm-hmm.

Matt Walker 17:27 So I actually think that there is probably a, a wealth of information. When you combine that probably with aspects of steering angle, steering maneuver-

Lex Fridman 17:39 Mm-hmm.

Matt Walker 17:39 ... and if you can sense the pressure on the pedals as well-

Lex Fridman 17:43 Mm-hmm.

Matt Walker 17:44 ... my guess is that there is some combinatorial feature that creates a phenotype of you are starting (laughs) to fall asleep. And as the autonomous controls develop, the, it's time for them to kick in. Some manufacturers, auto manufacturers sort of have something-... beta version, maybe an alpha version of (laughs) this already starting to come online, where they have a little camera in the wheel that I think tries to look at some features.

Lex Fridman 18:14 Almost everybody doing this, and it's very alpha.

Matt Walker 18:18 (laughs)

Lex Fridman 18:19 (laughs) . Uh, so, uh, you know, the thing that you currently have is when people have that in their car, there's a coffee cup or something that comes up that you might be sleepy. The, the primary signal that they're comfortable using is, uh, steering wheel reversals. So, so basically using your interaction with the steering wheel and how much you're interacting with it as a sign of sleepiness. So if you have to constantly correct the car-

Matt Walker 18:43 Yeah.

Lex Fridman 18:43 ... that's a sign of, like, you starting to drift into microsleep. I think that's a very, very crude signal. It's probably a powerful one. There's a whole nother component to this, which is it seems like it's so driver and subject dependent. Um, the, the, how our behavior changes as we get sleepy and drowsy seems to be-

Matt Walker 19:05 Right.

Lex Fridman 19:05 ... different in complicated, fascinating ways, where you can't just use one signal. It's kind of like what you were saying, there has to be a lot of different signals that you should then be able to combine. The hope is there's, uh, the search is for, like, universal signals that are pretty damn good for, like, 90% of people.

Matt Walker 19:23 Yeah. But I don't think we need to take necessarily quite that approach. I think what we could do in some clever fashion is using the individual. So what you and I are perhaps suggesting here is that there is a, an array of features that we know provide information that is sensitive to whether or not you're falling asleep at the wheel. Some of those, let's say that there are ten of them. You know, for me, seven of them are the cardinal features.

Lex Fridman 19:51 Mm-hmm.

Matt Walker 19:52 For you, however, you know, six of them, and they're not all the same, sort of overlapping, are those for you. I think what we need is algorithms that can firstly understand when you are well-slept. So let's say that people have sleep trackers at night, and then your car integrates that information and-

Lex Fridman 20:09 That would be amazing.

Matt Walker 20:10 ... understands when you are well-slept.

Lex Fridman 20:12 Yeah.

Matt Walker 20:12 And then you've got the data of the individual behavior unique to that individual snowflake, like (laughs) -

Lex Fridman 20:19 Mm-hmm.

Matt Walker 20:20 ... when they are well-slept. This is the signature of well-rested driving. Then you can look at deviations from that and pattern match it with the sleep history of that individual, and then I don't need to find the sort of, you know, the one-size-fits-all approach for 99% of the people. I can create a very bespoke, tailor-like set of features, a Savile Row suit of sleepiness features. You know, that would be my... If you want to ask me about moon shots and crazy ideas, that's where I would go, but to start with, I think your approaches is a great one. Let's find something that covers 99% of the people, because the worrying thing about microsleeps, of course, unlike, you know, drugs or alcohol, which, you know, certainly is a terrible thing to be behind the wheel. With those, often you, you react too late, and that's the reason you get into an accident. When you fall asleep behind the wheel, you don't react at all. The, you know, at that point, there is a 2-ton missile driving down the street and no one's in control.

Lex Fridman 21:29 Yeah.

Matt Walker 21:30 That's why those accidents can often be more dangerous.

Lex Fridman 21:34 Yeah, and, uh, the fascinating thing is, in the case of semi-autonomous vehicles, like Tesla Autopilot, this is where I've had disagreements with Mr. Elon Musk-

Matt Walker 21:44 (laughs)

Lex Fridman 21:45 ... is, uh, and, and, uh, the human factors community, which is this community that one of the big things they study is, uh, human supervision over automation. So you have, like, pilots, you know, supervising an airplane that's mostly flying autonomously. The question is, when we're actually doing the driving, how do microsleeps or general, how does dra- drowsiness progress and how does it affect our driving? That question becomes more fascinating, more complicated when your task is not driving but supervising the driving. So your task is to take over when stuff goes wrong, and that is, is complicated, but the basic conclusions from many decades is that humans are really crappy at supervising-

Matt Walker 22:34 Mm-hmm.

Lex Fridman 22:34 ... because they get, uh, they get drowsy and, uh, lose vigilance much, much faster. The really surprising thing with Tesla Autopilot, it was surprising to me, s- surprising to, uh, the human factors community, and in fact they still argue with me about it, is, uh, it seems that humans in Teslas with, with Autopilot and other similar systems are not becoming less vigilant, at least the, with, with the studies we've done. So there's something about the urgency of driving. I, I can't, I'm not sure the why, but there's something about the risk, I think, the fact that you might die (laughs) , uh, is still keeping people awake. The question is, as Tesla Autopilot or similar systems get better and better and better, how does that affect increasing drowsiness? And that's when you need to have... That's where the big disagreement was, you need to have driver sensing, meaning driver-facing camera, uh, that tracks some kind of information about the face that can tell you, uh, drowsiness. So that you can tell the car if you're drowsy so that the car can be like, "You should be probably driving or pull to the side."

Matt Walker 23:44 Right. Or I need to do some of the heavy lifting here.

Lex Fridman 23:47 Yeah.

Matt Walker 23:48 Um-

Lex Fridman 23:48 So there needs to be that dance of inter- interaction-

Matt Walker 23:50 Exactly. Yeah.

Lex Fridman 23:51 ... of, of, uh, human and machine, but currently it's mostly, uh, steering wheel based. So, you know-

Matt Walker 23:58 Yeah.

Lex Fridman 23:58 ... this, this idea that your hands should be on the, on the steering wheel, that's a, uh, a sign that you're paying attention.... is, um, is an outdated and a very crude metric.

Matt Walker 24:09 I agree. Yeah. I think there are far more sophisticated ways that we can solve that problem, uh, if we invest.

Lex Fridman 24:17 Can I ask you a-

Matt Walker 24:18 (laughs)

Lex Fridman 24:18 ... big philosophical question, before we get into fun details? Um, on the topic of conscious states,

Lex Fridman 24:28 how fundamental do you think is consciousness to the human mind? I ask this from almost like a robotics perspective. So, in your study of sleep, do you think the, the hard question of consciousness, that it feels like something to be us, is that like a nice little feature, like a, like a, like a quirk of our mind, or is it somehow fundamental? Because sleep feels like we take, we take a s- step out of that consciousness a little bit. So, from all your study of sleep, do you think consciousness is like deeply part of who we are, or is it just a nice trick?

Matt Walker 25:07 I think it's a deeply embedded feature that I can imagine has a whole panoply of biological benefits. But to your point about sleep, what is interesting if you do a lot of dream research, and we've done some, it's very, very rare, at all in fact, for you to end up becoming someone other than who you are in your dreams. Now, you can have third-person perspective dreams where you can see yourself in the dream as if you're sort of, you know, you've risen above your, y- your physical being. But for the most part, it's very rare that we lose our sense of conscious self. And maybe I'm sort of doing a sleight of hand because it's really, what I'm saying is it's very rare that we lose our sense of who we are-

Lex Fridman 25:57 Mm-hmm.

Matt Walker 25:58 ... in dreams. We never do. Now, that's not to suggest that dreams aren't utterly bizarre and, uh, I mean, you know, when you slept last night, which I know, um, may have been-

Lex Fridman 26:10 (laughs)

Matt Walker 26:10 ... p- perhaps a little less than, than me. But when you went into dreaming, you know, you became flagrantly psychotic.

Lex Fridman 26:17 Mm-hmm.

Matt Walker 26:18 And there are five essentially good reasons. Firstly, you started to see things which were not there, so you were hallucinating. Second, you believed things that couldn't possibly be true, so you were delusional. Third, you became confused about time and place and person, so you're suffering from what we would call disorientation. Fourth, you have wildly fluctuating emotions, something that, um, psychiatrists will call being affectively labile.

Lex Fridman 26:46 (laughs)

Matt Walker 26:46 And then how wonderful that you woke up this morning and you forgot most, if not all, of that dream experience, so you're suffering from amnesia.

Lex Fridman 26:52 (laughs)

Matt Walker 26:52 If you were to experience any one of those five things while you were awake-

Lex Fridman 26:56 Yeah.

Matt Walker 26:57 ... you would probably be seeking psychological-

Lex Fridman 26:59 Yeah.

Matt Walker 27:00 ... help. But what ... So, I place that as a backdrop against your astute question-

Lex Fridman 27:05 Mm-hmm.

Matt Walker 27:05 ... because despite all of that psychosis, there is still a present self-

Lex Fridman 27:13 Mm-hmm.

Matt Walker 27:14 ... nested at the heart of it. Meaning that I think it's very difficult for us to abandon our conscious sense of self. And if it's that hard, you know, it's the old adage in some ways that you can't outrun your shadow.

Lex Fridman 27:29 Mm-hmm.

Matt Walker 27:29 But here it's more of a philosophical question which is about the conscious mind and what the state of consciousness actually means in a human being.

Lex Fridman 27:37 Mm-hmm.

Matt Walker 27:38 So, I think that that, to me, you can ... You become so dislocated from so many other rational ways of waking consciousness, but one thing that won't go away, that won't get perturbed or sort of, m- you know, manacled, is this: your sense of conscious self.

Lex Fridman 27:58 Yeah, that's a strong sign that consciousness is fundamental to the human mind, um, or we're just creatures of habit, we've gotten used to having consciousness.

Matt Walker 28:06 (laughs)

Lex Fridman 28:06 Maybe it just takes a lot of, uh, either chemical substances or a lot of, uh, like mental work to escape that.

Matt Walker 28:16 I mean, it's like trying to launch a rocket. You know, the energy that has to be put in to create escape velocity from the gravitational pull of this thing called Planet Earth is immense.

Lex Fridman 28:28 Yeah.

Matt Walker 28:29 Well, the same thing is true for, uh, for us to abandon our sense of conscious self.

Lex Fridman 28:36 Yeah.

Matt Walker 28:36 The amount of biological, the amount of substances (laughs) , the amount of wacky stuff that you have to do to truly get escape velocity from your conscious self. What does that tell us about then the fundamental state of our conscious self?

Lex Fridman 28:52 Yeah, it also probably says that it, uh, it's quite useful to have consciousness for, uh, for survival and for just operation in this world, and perhaps for intelligence. I'm one of the, on the AI side, people that think that, uh, intelligence requires consciousness. So, like high levels of general intelligence requires consciousness. Most people in, in the AI field think like consciousness and intelligence are fundamentally different. You could build a computer that's super intelligent, it doesn't have to be conscious. I think that if you define super intelligence by being good at chess, yes.

Matt Walker 29:28 (laughs)

Lex Fridman 29:28 But if you define super intelligence as being able to operate in this living world of humans and be able to perform all kinds of different tasks, consciousness, it seems to be somehow fundamental to, uh, like, to, to, to richly integrate yourself into the human experience, into society, it feels like you have to be a conscious being. But then we don't even know what consciousness is, and we certainly don't know how to engineer it in our machines.

Matt Walker 29:56 I love the fact that there are still questions that are so embryonic because, you know, I suspect it's the same with you. Answers to me are simply ways to get to more questions.

Lex Fridman 30:09 (laughs)

Matt Walker 30:09 You know, it's questions were ...... you know, questions turn me on, answers less so. And I love the fact that we are still embryonic in our sense of arguing about even what the definition of consciousness is. But I also find it fascina- uh, I, I think it's thoroughly delightful to absorb yourself in the thought, uh, think about the brain, and we can move back across the complexity of phylogeny from, you know, humans, to mammals, to sort of birds, to reptiles, amphibians, fish, and s- you can, bacteria, whatever you want, and you can go through this and say, "Okay, where is the hard line of, you know, what we would define as consciousness?" And, and I'm sure it's got something to do with the complexity of the neural system. Of that, I'm fairly certain. But to me, it's always been fascinating. So, what is it then? You know, is it that I just keep adding neurons to a Petri dish, and I just keep adding them and adding them and adding them, and at some point when I hit a critical mass of interconnected neurons, that is the mass of the, you know, the interconnected human brain, then bingo. (laughs)

Lex Fridman 31:16 Yeah. (laughs)

Matt Walker 31:17 All of a sudden it kicks into gear and we have consciousness.

Lex Fridman 31:21 Like a phase shift, phase transition of some kind.

Matt Walker 31:23 Correct, yeah. But there is something about the complexity of the nervous system that I think is fundamental to consciousness, and the reason I bring that up is because when we're trying to then think about creating it in an artificial way, does that inform us as to the complexity that we should be looking at in terms of development? I, I also think that it's a missed opportunity in the sort of digital space for us to try to recreate human consciousness. We've already got human consciousness. What if we were to think about creating some other form of c- why do we have to think that the ultimate in the creation of, you know, an artificial intelligence is the replication, you know, of a human state of consciousness? Can we not think outside of our own (laughs) consciousness and believe that there is something even more incredible or more complementary, more orthogonal?

Lex Fridman 32:22 Mm-hmm.

Matt Walker 32:23 Um, so I'm sometimes perplexed that people are trying to mimic human consciousness rather than think about creating something that's different.

Matt Walker 32:34 So...

Lex Fridman 32:34 Yeah. I think of human consciousness, or consciousness in general, as this magic, um, superpower that allows us to deeply experience the world, and just as you're saying, I don't think that superpower has to take the exact flavor as humans have. That's my love for robots. I would love to add the ability to robots that can experience the world and other humans, uh, deeply. I'm humbled by the fact that that idea does not necessarily need to look anything like how humans experience the world, but there's a dance of, um, human to robot connection, the same way human to dog or human to cat connection, that there's a, there's a, there's a magic there to that interaction, and I'm not sure how to create that magic, but it's a worthy effort. I also love, just exactly as you said, on the question of consciousness or engineering consciousness, the fun thing about this problem is it s- seems obvious to me that 100 years from now, no matter what we do today, uh, people, if we're still here, will laugh at how silly our notions were. So like, it's almost impossible for me to imagine that we will truly solve this problem fully, uh, in my lifetime. (laughs)

Matt Walker 33:56 (laughs)

Lex Fridman 33:56 And, and more than that, everything we'll do will be silly 100 years from now, but it's still, uh, a w- that makes it fun to me, because it's like you have the full freedom to not even be right. (laughs) Just to try.

Matt Walker 34:11 Yeah.

Lex Fridman 34:11 Just to try is freedom. And, uh, and I, that's how I see...

Matt Walker 34:15 Can you get me that T-shirt, please? (laughs)

Lex Fridman 34:17 (laughs)

Matt Walker 34:17 I love that.

Lex Fridman 34:18 So, a- and, you know, human-robot interaction is fascinating because it's like, it's like watching dancing. I've been, uh, dancing tango recently, and just it's like there is no goal. The goal is to create something magical.

Matt Walker 34:33 Mm-hmm.

Lex Fridman 34:33 And, uh, whether consciousness, or emotion, or elegance of movement, all of those things, uh, aid in the creation of the magic, and it's a free, it's art, it's an art form to explore how, how to make that, um, how to create that in a way that's compelling.

Matt Walker 34:51 Yeah, I love the, the line in Scent of a Woman with Al Pacino, where he's speaking about the tango and he says, "Really, it's just freedom that if you get tangled up, you just keep tangoing on."

Lex Fridman 35:01 I still to this day, I think, uh, well, first or second time I talked to Joe Rogan on his podcast, I said, we got into this heated argument about whether, uh, um, Scent of a Woman is a better movie than John Wick, (laughs) because it's, it's one of my favorite movies for many reasons. One is-

Matt Walker 35:20 Which one? Scent of a Woman?

Lex Fridman 35:21 Se- Scent of a Woman. Uh, partially-

Matt Walker 35:24 I didn't know that, by the way. I was just gonna... (laughs)

Lex Fridman 35:26 (laughs) You just, you tossed that out there.

Matt Walker 35:26 Yeah, I didn't know if you would actually know of the movie.

Lex Fridman 35:29 Oh, awesome. Awesome.

Matt Walker 35:29 Yeah.

Lex Fridman 35:29 No, yeah, I said, "I love the tango scene. I love Al Pacino's performance. It's a wonderful movie." (laughs) Then Joe, Joe was saying John Wick is better, so we, to this day, argue about this.

Matt Walker 35:41 I think it depends in, on what conscious state you're in...

Lex Fridman 35:44 Yes, sir.

Matt Walker 35:45 ... that you would be ready and receptive to, but, um, Scent of a Woman, eh, I think it has one of the best monologues at the end of the movie that has ever been written or at least performed, uh...

Lex Fridman 35:57 When Al Pacino def- defends the-

Matt Walker 36:00 That's correct.

Lex Fridman 36:00 ... the young- the younger... Yeah, I, uh, I often think about that. There's been times in my life, uh, I don't know about you, where I wish I had an Al Pacino in my life.

Matt Walker 36:12 (laughs)

Lex Fridman 36:12 Where, um...... integrity is really important in this life.

Matt Walker 36:18 It is.

Lex Fridman 36:18 And sometimes, you find yourself in places where there's pressure to sacrifice that integrity, and y- you want, uh, what is it, Lieutenant Colonel, or whatever he was-

Matt Walker 36:30 (laughs) Slade.

Lex Fridman 36:32 ... to come in (laughs) -

Matt Walker 36:32 (laughs)

Lex Fridman 36:32 ... to come in, uh, on your side and, and scream at everyone and say, "What the hell are we doing here?"

Matt Walker 36:38 Being, you know, unfortunately, British, and sort of having that slightly, um, awkward sort of Hugh Grant gene, it's-

Lex Fridman 36:45 (laughs)

Matt Walker 36:45 ... it's very, very (laughs) ... we're at the opposite end of the spectrum of the remarkable feat of, uh, Al Pacino at the- the end of that scene, but, um, and yeah, integrity is, um, it's a, it's a challenging thing, and I value it much, and I think, um, it can take 20 years to build a reputation and two minutes to lose it.

Lex Fridman 37:06 Yeah.

Matt Walker 37:06 And there is nothing more that I value than, than integrity, and, you know, if I'm ever wrong about anything, I truly don't want to be wrong for any longer than I have to be. Um, you know, that's what being, in some ways, a scientist is. You're- you're just driven by truth, and the irony, relative to something like mathematics, is that, in science, you never find truth. What- all you do in science is you discount the things that are likely to be untrue, leaving only the possibility of what could be true.

Lex Fridman 37:42 Yeah.

Matt Walker 37:42 But in math, you know, when you create, you know, a proof, it's a proof for, you know, f- from that point forward, there is truth in mathematics, and there's, I think there's a beauty in that. But I kind of like the messiness of, of science because, again, to me, it's less about the truth of the answer, and it is more about the pursuit of questions.

Lex Fridman 38:07 But there, integrity becomes more and more important, and it becomes more difficult. There's a lot of pressures, just like in the rest of the world, but there's a lot of pressures on a scientist. Uh, one is like funding sources.

Matt Walker 38:19 Mm, yeah.

Lex Fridman 38:19 I've noticed this, that, um, you know, money affects everyone's mind, I think. Um, I've been always somebody that I believe money can't- you can't buy my opinion. Uh, I don't care how much money, billions or trillions, the- to- but that pressure is there, and you have to be very cognizant of it and make sure that your opinion is not defined by the funding sources. And then the other is just your own success of, uh, you know, for a couple of decades, publishing an idea, um, and then realizing at some point that that idea was wrong all along.

Matt Walker 38:57 Right.

Lex Fridman 38:58 And that- that's a tough thing for people to do, but that's also integrity, is to walk away, is to sh- say that you were wrong. Um, th- i- that doesn't have to be in a- some big, dramatic way. It could be in a bunch of tiny ways along the way.

Matt Walker 39:11 Right.

Lex Fridman 39:12 Like, uh, reconfigure your intuition about a- a particular, uh, problem. That's- and all of that is integrity. When everybody in the room, uh, w- you know, believes a certain thing, everybody in the community believes a certain thing, to, uh, to be able to still be open-minded in the face of that.

Matt Walker 39:31 Yeah, and I think it comes down, in some ways, to the issue of ego-

Lex Fridman 39:35 Yeah.

Matt Walker 39:35 ... that you bond your, you know, correctness or your rightness, your scientific theory with your sense of ego.

Lex Fridman 39:45 Yeah.

Matt Walker 39:45 You know, I've never found it that difficult to let go of theories in the face of counter-evidence, in part because I have such low self-esteem.

Lex Fridman 39:56 Well, I- I kinda like that. I- I've always liked that combination. I have the same- I'm like very self-critical, imposter syndrome, all those things, uh, putting yourself below the podium, but at the same time, having the ego that drives the ambition to work your ass off, like some kind of weird drive, maybe (laughs) like, to- drive to be better, like thinking you- yourself is not that great and always driving to be better-

Matt Walker 40:20 Mm.

Lex Fridman 40:21 ... and then at the same time, because that's- that can be paralyzing and exhausting and so on, at the same time, just being grateful to be alive. But in the sciences, in the actual effort, never be satisfied, never think of yourself highly. That seems to be a nice combination. (laughs)

Matt Walker 40:35 I very much hope that that is part of who I am and I remain very quietly motivated and driven, and I, like you, love the idea of perfection, and I know I will never achieve it, but I will never stop trying to.

Lex Fridman 40:52 So, s- similar to you, which sounds weird because there's all these videos of, uh, of me on the internet (laughs) . So, I think- I think I just naturally lean into the things I'm afraid of and I'm uncomfortable doing.

Matt Walker 41:07 Yeah.

Lex Fridman 41:07 Like I'm very afraid of talking to people and, you know, just even, uh, to- before talking to you today, just a lot of anxiet- anxiety and all those kinds of things.

Matt Walker 41:16 About talking to, to me?

Lex Fridman 41:17 Yeah, yeah.

Matt Walker 41:18 Oh, nice.

Lex Fridman 41:18 It's like some nervousness. (laughs)

Matt Walker 41:20 (laughs)

Lex Fridman 41:20 Uh, fear, in some cases, uh, self-doubt, and all those kinds of things, but I do it anyway. So-

Matt Walker 41:26 Mm-hmm.

Lex Fridman 41:26 ... the reason I bring that up is, um, you've, uh, launched a podcast.

Matt Walker 41:32 I have. (laughs)

Lex Fridman 41:34 (laughs) Allow me to say, I think you're a great science communicator. So this challenge of being, um, afraid or cautious of being in the public eye, and yet, having a longing to communicate some of the things you're excited about in the space of sleep and beyond, what's your vision with this project?

Matt Walker 41:59 I think firstly to your- to that question, like you, I am always more afraid of not trying than trying.

Lex Fridman 42:07 (laughs) Yeah. That's true.

Matt Walker 42:09 That, to me, frightens me more. But with the podcast, I think, really, I have two very simple goals.I want to try and democratize the science of sleep. And in doing so, my goal would be to try and reunite humanity with the sleep that it is so desperately bereft of. And if I can do that through a number of different means, um, the podcast is a little bit different than this format. It's, uh, going to be short form monologues, um, (laughs) from, from yours truly, uh, that will last usually less than just 10 minutes, and I see it as simply a little slice of sleep goodness that can accompany your waking day.

Lex Fridman 42:53 It's hard to know what is the right way to do science communication. Like, uh, your friend of mine, Andrew Huberman, is- does... He- he's an incredible human being.

Matt Walker 43:05 Oh, gosh, yes.

Lex Fridman 43:06 So he does, like, two hours of... I wonder how many takes he does, I don't know, but it looks like he doesn't do any...

Matt Walker 43:12 Yeah, I suspect... He's that magnificent-

Lex Fridman 43:15 Yeah.

Matt Walker 43:15 ... of a human being.

Lex Fridman 43:16 When I talk to him in, like, in person, he always generates intelligent words, well-cited, non-stop for hours. So I don't...

Matt Walker 43:25 He's a Gatling gun of information-

Lex Fridman 43:27 (laughs) Yeah.

Matt Walker 43:28 ... and it's pristine.

Lex Fridman 43:29 And passion, and all those kinds of things. So that- that's an interesting medium. I, I wouldn't have, um... It's funny, 'cause I wouldn't have done it the way he's doing it, I wouldn't advise him to do it the way he's doing it, 'cause I thought there's no way you could do what you're doing. (laughs)

Matt Walker 43:43 (laughs)

Lex Fridman 43:43 'Cause it's a lot of work. Uh, but he is, like, doing an incredible job of it. I just think... It's the same with, like, Dan Carlin and Hardcore History. I thought, I thought that the way Andrew's doing it would crush him the way it crushes Dan Carlin. So Dan has so much pressure on him to do a good job that he ends up publishing, like, two episodes a year.

Matt Walker 44:07 Yeah.

Lex Fridman 44:08 So that pressure can be paralyzing. The pressure of, like, putting out, like, strong scientific statements, th- that can be overwhelming. Now, Andrew seems to be (laughs) just plowing through anyway. If there's mistakes, he'll, he'll say there's corrections, and so on.

Matt Walker 44:24 Yeah.

Lex Fridman 44:25 I just, I wonder, I actually haven't talked to him too much about it, like, psychologically, how difficult is it to put yourself out there for an hour or two a week of just non-stop dropping knowledge, any one sentence of which could be totally wrong, it could be a mistake.

Matt Walker 44:42 Right. And there will be-

Lex Fridman 44:43 Yeah.

Matt Walker 44:43 ... mistakes, you know. And I, you know, in the first edition of my book, there were errors that, you know, we corrected in the second edition too.

Lex Fridman 44:51 Mm-hmm.

Matt Walker 44:52 But there will be, probabilistically, you know-

Lex Fridman 44:56 Yeah.

Matt Walker 44:56 ... if you've got, you know, 10 facts per page of a book, and you've got 350 pages, odds are it's probably not going to be utter perfection out the gate, and it will be the same way for Andrew too. But having the, the reverence of, um, a humble mind and simply accepting the things that are wrong and correcting them and doing the right thing, I know that that's his mentality.

Lex Fridman 45:26 I do wanna say that I'm just kinda honored to be, uh, th- it's like, it's a cool group of, like, scientific people that, uh, I'm fortunate enough to now be interacting with, is- is you, and Andrew, and, um, David Sinclair has been thinking about throwing his hat in the ring.

Matt Walker 45:43 Oh, I hope so. David is another one of those very special people in the world.

Lex Fridman 45:47 So it's cool because podcasts are... It's cool, it's a, it's such a powerful medium of communication, it's much freer than more constrained, like, publications and so on.

Matt Walker 45:58 Yeah.

Lex Fridman 45:58 Or it's much more accessible and inspiring than, like, I don't know, conference presentations or lectures, and so it's, it's a really exciting medium to me, and it's cool that there's this, like, group of people that, uh, are becoming friends and are, and putting stuff out there and supporting each other, so it- it is fun to also watch how that's going to evolve in your case. 'Cause I wonder, it'll be two a month-

Matt Walker 46:20 Or, or devolve-

Lex Fridman 46:21 (laughs)

Matt Walker 46:22 ... I think is the (laughs) -

Lex Fridman 46:22 Devolve.

Matt Walker 46:23 ... is the answer to that, like...

Lex Fridman 46:25 Well, I mean, some of it is persistence through the challenges that we've been talking about, which is like-

Matt Walker 46:31 I think I've got a lot to learn.

Lex Fridman 46:32 Yeah.

Matt Walker 46:33 But I will persist.

Lex Fridman 46:35 L- can I ask you some detail stuff? You mentioned I would.

Matt Walker 46:38 Oh my goodness, go anywhere you wish with, with sleep.

Lex Fridman 46:41 (laughs) Uh, so

Lex Fridman 46:42 I'm a big fan of coffee and caffeine, and I've been, especially in the last few days, consuming a very large amount, and I'm cognizant of the fact that, um, my body is affected by caffeine different than the anecdotal information that other people tell me. I seem to be not at all affected by it. It's almost, um... It- it feels like more like a ritual than it is a, a chemical boost to my performance. Like, I can drink several cups of coffee right before bed and just knock out anyway. I'm not sure if it's, uh, biological, chemical, or it has to do with just the fact that I'm cos- consuming huge amounts of caffeine. All that to say, uh, what do you think is the relationship between coffee and sleep? Caffeine and sleep? If there's an interesting distinction there.

Matt Walker 47:34 There is a distinction. So, I think the first thing to say, which is going to sound strange coming from me, is drink coffee. (laughs) Um, the health benefits associated with drinking coffee are really quite well-established now, um, but I think the, the counterpoint to that... Well, firstly, the dose and the timing make the poison, and I'll perhaps come back to that-

Lex Fridman 47:59 Yes.

Matt Walker 48:00 ... in just a second. But for coffee, it's actually not the caffeine.

Lex Fridman 48:06 Mm-hmm.

Matt Walker 48:07 So, you know, a lot of people ha- have asked me about this rightful paradox between the fact that sleep provides all of these incredible health benefits, and then coffee, which can have a deleterious-... impact on your sleep, has (laughs) a whole collection of health benefits, many of them Venn diagram overlapping with those that sleep provides. How on earth can you reconcile those two? And the answer is that... Well, the answer is very simple, it's called antioxidants. That it turns out that for most people in Western civilization, because of diet not being quite what it should be, the major source through which they obtain antioxidants is the coffee bean. So, the, the humble coffee bean has now been asked to carry the astronomical weight of serving up the large majority of people's antioxidant needs.

Lex Fridman 49:02 Mm-hmm.

Matt Walker 49:02 And you can see this if, for example, you look at the health benefits of decaffeinated coffee. It has a whole constellation of really great health benefits too. So, it's not the caffeine. And that's why I liked what you said-

Lex Fridman 49:16 Mm-hmm.

Matt Walker 49:16 ... this sort of separation of church and state between coffee-

Lex Fridman 49:19 (laughs)

Matt Walker 49:19 ... and caffeine. It's not the caffeine, it's the coffee bean itself that provides those health benefits. But coming back to how it impacts sleep, it impacts sleep in probably at least three different ways. The first is that, for most people, caffeine can make it, obviously, a little harder to fall asleep. Caffeine can make it harder to stay asleep. But let's say that you are one of those individuals, and I think you are, and you can say, "Look, I can have three or four espressos with dinner and I fall asleep just fine, and I stay asleep soundly across the night, so there's no problem." Th- the downside there is that, even if that is true, the amount of deep sleep that you get will not be as deep, and so you will actually lose somewhere between 10 to 30% of your deep sleep if you drink caffeine in the evening.

Lex Fridman 50:10 Mm-hmm.

Matt Walker 50:11 So, to give you some context, to, to drop your deep sleep by, let's say, 20%, I'd probably have to age you by 15 years, or you could do it every night with a cup of coffee. I think the fourth component that is perhaps less well-understood about coffee is its timing, and that's why I was saying the timing and the dose make the poison. The dose, by the way, once you get past about three cups of coffee a day, the health benefits actually start to turn down in the opposite direction.

Lex Fridman 50:39 Mm-hmm.

Matt Walker 50:40 So, there is a U-shaped function. It's sort of a, you know, the Goldilocks syndrome, not too-

Lex Fridman 50:45 Mm-hmm.

Matt Walker 50:45 ... little, not too much, just the right amount. The second component is the timing, though. Caffeine has a half-life of about, um, five to six hours, meaning that after five to six hours, 50% of that, on average, for the average adult, is still in the system, which means that it has a quarter life of 10 to 12 hours. So, in other words, if you have a coffee at noon, a quarter of that caffeine is still circulating in your brain at midnight. So, having a cup of coffee at noon, one could argue is the equivalent of tucking yourself into bed at midnight, and before you turn the light out-

Lex Fridman 51:19 Mm-hmm.

Matt Walker 51:19 ... you swig a quarter of a cup of coffee. But that doesn't still answer your question as to why are you so immune. So, I'm someone who is actually, unfortunately, very sensitive to caffeine, and if I have, you know, even two cups of coffee in the morning, um, I, I don't sleep as well that night. And I find it miserable because I love the smell of coffee, I love the routine, I love the ritual. I think I would love to be invested in it. It's just terrible for my sleep, so I switch to decaf. There is a difference from one individual to the next, and it's controlled by a set of liver enzymes called cytochrome P450 enzymes. And there is a particular gene that, if you have a different sort of version of this gene, it's called CYP1A2, that gene will determine the speed of the clearance of caffeine from your system.

Lex Fridman 52:17 Mm-hmm.

Matt Walker 52:18 Some people will have a version of that gene that is very effective and efficient at clearing that caffeine, and so their half-life could be as short as two hours rather than five to six hours. Other people, uh, hands up, Matt Walker, um, have a version of that gene that is not very effective at (laughs) clearing out the, uh, the caffeine, and therefore, their half-life sort of sensitivity could be somewhere between, you know, eight to nine hours.

Lex Fridman 52:47 Mm-hmm.

Matt Walker 52:48 So, we understand that there are individual differences, but overall, I guess, the, the top line here is drink coffee, um, and understand that it's not the caffeine, it's the coffee that's the benefit, and the dose makes the poison.

Lex Fridman 53:01 Is there some aspect to it that's it's like a muscle in terms of the... or the combination of, of letters and numbers-

Matt Walker 53:08 Mm-hmm.

Lex Fridman 53:08 ... that you just said? Is there some aspect that if, um, I can improve the quarter life, the half-life c- could decrease that number if I just practice?

Matt Walker 53:19 (laughs)

Lex Fridman 53:20 Like, I drink a lot of coffee, so, so, like, habit, uh, uh, alters how your body's able to get rid of the caffeine.

Matt Walker 53:27 Not how the body is able to get rid of the caffeine, but, uh, it does alter how sensitive the body is to the caffeine. And it's not at the level of the enzyme degrading the caffeine. It's at the level of the receptors that caffeine will act upon.

Lex Fridman 53:43 Mm-hmm.

Matt Walker 53:44 Now, it turns out that those are called adenosine receptors, and maybe we can speak about what adenosine is, and sleep pressure, and all of that good stuff. But as you start to drink more and more coffee, um, the body tries to fight back, and it happens with many different drugs, by the way, and it's called tolerance.

Lex Fridman 54:02 Mm-hmm.

Matt Walker 54:03 And so one of the ways that your body becomes tolerant to a drug is that the receptors that the drug is binding to, these sort of welcome sites, these sort of, you know, pitcher mitts, um, as it were, that receive the drug, those start to get...... taken away from the surface of the cell, and it's what we call receptor internalization. So the cell starts to think, "Gee-whiz, you know, there's- there's a lot of stimulation going on. This is too much. So I'm just going to... When normally I would, you know, coat my cell with, let's just say, five of these receptors, for argument's sake. Things are going a little bit too ballistic right now. I'm going to take away at least two of those receptors and downscale it to just having three of those." And now, you need two cups of coffee to get the same effect that one cup of coffee got you before. And that's why then when you go cold turkey on coffee, all of a sudden the system has equilibrated itself to expecting X amount of stimulation, and now all of that stimulation is gone, so it's now got too few receptors-

Lex Fridman 55:12 Mm-hmm.

Matt Walker 55:13 ... and you have a caffeine withdrawal syndrome. And that's why, for example, with, you know, drugs of abuse, things like heroin, when people go into abstinence, you know, as they're sort of moving into their addiction, they will build up a- a progressive tolerance to that drug, so they need to take more of it to get the same high.

Lex Fridman 55:33 Mm-hmm.

Matt Walker 55:34 But then if they go cold turkey for some p- period of time, the system goes back to being more sensitive again. It starts to repopulate the surface of the cell with these receptors. But now, when they reuse and they fall off the wagon, if they go back to the same dose that they were using before, you know, 10 weeks ago or three months ago, that dose can kill them. They can have an overdose.

Lex Fridman 55:57 Mm-hmm.

Matt Walker 55:58 Even though they were using the same amount at those two different times, the difference is that it's not the dose of the drug, it's the sensitivity of the system.

Lex Fridman 56:08 Mm-hmm.

Matt Walker 56:08 And that's the same thing that we see with caffeine, in terms of training the muscle, as it were, is the system becomes less sensitive, can calibrate.

Lex Fridman 56:17 Is there a time, the number of hours before bed, that's a- a safe bet to s- to most people to recommend you shouldn't drink caffeine this many hours? Like is there an average half-life that you should be aiming at?

Matt Walker 56:35 Yeah.

Lex Fridman 56:35 Uh, or is this advice kind of impossible because there's so much variability?

Matt Walker 56:39 There is huge variability, and I think everyone themselves, you know, to a degree knows it, although I'll put a caveat on that too because it's a slightly dangerous point. So the recommendation for the average adult and who... Where is the average adult in society? There is no such thing. But for the (laughs) average adult, it would be probably cutting yourself off maybe 10 hours, you know, before. So assuming a normative bedtime in society, I would say try to stop drinking caffeine, you know, before 2:00 PM and, um, just keep an eye out, you know. And if you're struggling with sleep, dial down the caffeine and see if it makes a difference.

Lex Fridman 57:18 Can I ask you about sleep and, uh, learning? So how does sleep affect learning? Sleep before learning, sleep after learning, which are both fascinating kind of dynamics-

Matt Walker 57:33 Yeah.

Lex Fridman 57:33 ... of the mind's interaction with this extra conscious state?

Matt Walker 57:38 Yeah. Sleep is profoundly and very intimately related to your memory systems and your informational systems. Um, the first, as you just mentioned, is that sleep before learning will essentially prepare your brain, um, almost like a dry sponge ready to sort of, (sucking noise) you know, initially soak up new information. In other words, you need sleep before learning to effectively imprint information into the brain, to lay down fresh memory traces. Um, and without sleep, the memory circuits of the brain, and we know, we've studied these memory circuits, will... You know, they essentially become waterlogged, as it were, for the sponge analogy, and you can't absorb the information as effectively. So you need sleep before learning, but you also need sleep, unfortunately, after learning too, to then take those freshly minted memories and effectively hit the save button on them. But it's nowhere near as quick as a digital system. It takes hours because it's a physical, biological change that happens at the level of brain cells. But sleep after learning will cement and solidify that new memory into the neural architecture of the brain, therefore making it less likely to be forgotten. So, uh, you know, I often think of sleep in that way as, um, it's almost sort of future-proofing information (laughs) in-

Lex Fridman 59:06 I- in what way?

Matt Walker 59:07 ... your brain. Well, it means that it gives it a higher degree of assurance f- to be remembered in the future-

Lex Fridman 59:15 Mm-hmm.

Matt Walker 59:15 ... rather than go through the sort of degradation that we think of as forgetting. So the brain has, in some ways by default, you know, i- there is forget... And- and actually I would love to... I was going to say sleep is relevant for memory in three different ways, but I'm going to amend that and say there's four different ways, which is learning, maintaining, memorizing, abstraction assimilation, association, then forgetting, which o- the last one sounds oxymoronic-

Lex Fridman 59:50 Mm-hmm.

Matt Walker 59:50 ... based on the former three, but I'll see if I can explain. So sleep after learning then sort of, you know, sets th- that information like amber (laughs) um, in, you know, in- in solidification. The third benefit, however, is that sleep, we've learned more recently, is much more intelligent than we ever gave it credit for. Sleep doesn't simply just take individual memories and strengthen them. Sleep will then intelligently integrate and cross-link and associate that information together.... and it's almost like informational alchemy-

Lex Fridman 1:00:29 (laughs)

Matt Walker 1:00:29 ... so that w- you wake up the next morning with a revised mind wide web-

Lex Fridman 1:00:33 Mm-hmm.

Matt Walker 1:00:33 ... of associations. And that's probably the reason that, you know, you've never been told to stay awake on a problem.

Lex Fridman 1:00:41 (laughs)

Matt Walker 1:00:42 You know, y- and f- in every language that I've inquired about, that phrase or very, something very similar seems to exist, which means to me that this creative associative benefit of sleep transcends cultural boundaries. It is a common experience across humanity. Um, (laughs) now I should note that I think the French translation of that is much closer to, um, I think, "You sleep with a problem," whereas the British- (laughs)

Lex Fridman 1:01:10 (laughs)

Matt Walker 1:01:10 ... you know, "You sleep on a problem." The French, "You sleep with a problem." I think it says so much about the romantic difference-

Lex Fridman 1:01:15 (laughs)

Matt Walker 1:01:15 ... between the, the British and the French, but let's, let's not go there. Um... (laughs)

Lex Fridman 1:01:19 That's brilliant.

Matt Walker 1:01:20 So-

Lex Fridman 1:01:20 Such a subtle, but such a fundamental difference, yeah-

Matt Walker 1:01:23 Oh, cultures.

Lex Fridman 1:01:24 ... between cultures.

Matt Walker 1:01:24 Yeah, goodness me.

Lex Fridman 1:01:25 Sleep with the problem. (laughs)

Matt Walker 1:01:27 Uh, yes, exactly, right on it.

Lex Fridman 1:01:28 That's why I love the French. (laughs)

Matt Walker 1:01:31 Um, so, um, we can sort of double-click on any one of these, uh, and go in- into detail, but the fourth, I became really enchanted by about eight years ago in our research, which was this idea of forgetting. And I started to think that forgetting may be the price that we pay for remembering.

Lex Fridman 1:01:55 Hmm.

Matt Walker 1:01:56 And in that sense, there is an enormous benefit to letting go. And you may be thinking, "That sounds ridiculous. I don't want to forget. In fact, my biggest problem is I keep forgetting things." But the brain i- has a fi- well, we believe has a finite storage capacity. We can't prove it yet, but my suspicion is that that's probably true. It doesn't have an infinite storage capacity. It has constraints.

Lex Fridman 1:02:26 Mm-hmm.

Matt Walker 1:02:27 If that's the case, we can't simply go through life being, you know, constantly informational aggregators unless, you know, we are programmed to say we've got a hard drive space of about 85 to 90 years, and we're good, and we can do that. Maybe that's true. I don't think that's true. I think forgetting is an incredibly good and useful thing. So for example, you know, it's not beneficial from an evolutionary perspective for me to remember where I parked my car three years ago. So, it's important that I can remember today's parking spot, but I don't want to have the junk kinda DNA, uh, from a memory perspective of, you know, where my, I parked my car, you know, two years ago. Um, now, I actually have a, in some ways, a problem with forgetting. I'm, uh... And again, I'm not trying to sort of be laudatory, but, you know, I, I, I tend not to forget too many things, and I don't think that that's a good thing. And, um, the, there's a wonderful, um, neurologist, Luria, who wrote a book called The Mind of the Mnemonicist, and it was a brilliant book, both because it was written exquisitely, but he was studying these sort of memory savants who basically could remember everything that he gave them, and he tried to find a chink in their armor. And the first half of the book is essentially about him f- seeing how far he can push them before they fail, and he never found that place. He could never find a place where they stopped remembering.

Lex Fridman 1:04:08 Mm-hmm.

Matt Walker 1:04:09 And then, in his brilliance, he turned the question on its head. He said, not, "What is the benefit of constantly remembering?" But instead, "What is the detriment to never forgetting?" And when you start to realize his descriptions of those individuals, it's probably a life that you would not want.

Lex Fridman 1:04:28 Mm-hmm. But it's, uh, fascinating both from a human perspective, but also AI perspective. There's, uh, there's a big challenge in the machine learning community of how to build systems that are able to remember for prolonged periods of time, lifelong continuous learning, so where you build up information over time. So, memory is one of the biggest open problems in, uh, in AI and machine learning, but at the same time, the right way to formulate memory is actually forgetting, because you have to be exceptionally selective at which kind of stuff you remember, and that's where the step of assimilation and integration that you're referring to is really important. I mean, we forget most of the things, and, and the question is exactly the cost of forgetting at the very edge of stuff that could be important or could not be.

Matt Walker 1:05:21 Mm-hmm.

Lex Fridman 1:05:21 How do we remember or not those things? Like, for example, I've, uh, you know, doing a podcast, I've become cognizant of one feature of my forgetting that's been problematic, which is I forget names and titles of books and so on. So when I read, I remember ideas. I remember quotes.

Matt Walker 1:05:45 Mm-hmm.

Lex Fridman 1:05:45 I remember statements and, like, that's the space in which I'm thinking. But when you communicate to others, you have to say, "This person in this book said that."

Matt Walker 1:05:56 Mm-hmm. Mm-hmm.

Lex Fridman 1:05:56 So it's the same thing with, uh, with like... Andrew Huberman is, uh, is masterful at this. This is important in academia, remembering the authors of a paper and the title of the paper as part of remembering the idea. And I've been feeling the cost of not re- not being able to naturally remember those things. And so that, that's something I need to sort of work on, but that's an example-

Matt Walker 1:06:18 Are you good with faces?

Lex Fridman 1:06:20 Yes, very good at faces.

Matt Walker 1:06:21 But not good with names. So, I'm-

Lex Fridman 1:06:23 Not good with names.

Matt Walker 1:06:23 ... exactly like you, and there is, you know, an understanding of that in-... the brain, too. We understand that there is partitioning of those, in terms of the territory of the brain that takes care of faces and facts and places, and, uh, they can be separate. So, I will never forget a face, but, you know, and as I said, I usually forget very little. But for some reason, names are a struggle. I think in some ways because I'm probably just a slightly anxious person. So, when you first meet someone, which is usually the time when a name is introduced.

Lex Fridman 1:06:55 Mm-hmm.

Matt Walker 1:06:55 You know, you were saying you were sort of anxious maybe about sort of sitting down with me, um, but I- I find that a little bit, you know, activating. And so, it's not as though there's anything wrong with my memory. It's just the emotional state I'm in when I'm first meeting someone-

Lex Fridman 1:07:10 Yeah.

Matt Walker 1:07:10 ... uh, you know, it's a little bit perturbing. But I will never forget their- their face, but, uh...

Lex Fridman 1:07:15 I completely relate to that because I almost don't hear, uh, people's names when they tell me because I'm so anxious. (laughs)

Matt Walker 1:07:21 Yeah, yeah.

Lex Fridman 1:07:22 But I- I- I think there's certain quirks of social interaction that, uh, show that you care about the person, that you remember that person, that they matter to you. They- they had an impact on you. And one of the ways to show that is you remember their name.

Matt Walker 1:07:38 Yeah.

Lex Fridman 1:07:38 And, but that's a quirk to me because there's, um, m- a lot of people I meet have a- a deep impact on me, but they- I can't communicate that unless I know their name. Unless I know some of the, um, details that- that, uh, we humans seem to use to communicate that we remember each other. What I- I remember well is the feeling we shared.

Matt Walker 1:08:04 Yeah.

Lex Fridman 1:08:05 Is the experience we shared. What I don't remember well is the detailed labels of- of those experiences, and I need to certainly work on that.

Matt Walker 1:08:14 Well, I- I don't know. I think it's, you know, just allowing yourself to be innate and who you are, is- is also a beautiful thing, too. I- I'm not suggesting it's not important to try and better oneself, and but also sometimes worry about the misery that that puts us in, but-

Lex Fridman 1:08:31 (laughs)

Matt Walker 1:08:31 ... like you-

Lex Fridman 1:08:32 Yeah.

Matt Walker 1:08:32 ... I will, I- I do struggle with name. But I know for the first time when we met in the lobby-

Lex Fridman 1:08:39 Mm-hmm.

Matt Walker 1:08:40 ... um, I know exactly what you look like. I know that you were wearing headphones. I know the shape and the size of those headphones. You didn't have your black jacket on. I know exactly what the weave of your shirt looked like.

Lex Fridman 1:08:51 Yeah.

Matt Walker 1:08:52 I knew what your shoes looked like. And I knew exactly the height of your-

Lex Fridman 1:08:55 Yes.

Matt Walker 1:08:56 ... the end of your pants from the top-

Lex Fridman 1:08:58 Yeah.

Matt Walker 1:08:58 ... of your shoes.

Lex Fridman 1:08:59 Yeah.

Matt Walker 1:08:59 And so those things I don't forget, you know?

Lex Fridman 1:09:02 Yeah.

Matt Walker 1:09:02 And I can-

Lex Fridman 1:09:02 That's fascinating. I'm the same.

Matt Walker 1:09:03 ... remember when people, I met people, you know, two years ago, and I'll say, "Oh yes, we met there, and, um, I remember you had those fantastic, you know, boots on. I thought they were bloody great pair of boots," you know?

Lex Fridman 1:09:14 Yeah.

Matt Walker 1:09:14 And they're like, "How do you..." I didn't even remember what I was wearing that day.

Lex Fridman 1:09:17 Yeah, yeah. It's fascinating. Yeah, I'm- I'm the exactly the same way. But you can't un- until we have Neuralink or something like that, we can't communicate that you remember all those things.

Matt Walker 1:09:26 I know. That's what I want too. (laughs)

Lex Fridman 1:09:28 So you have to be able to use, uh, tricks of human communication for that.

Matt Walker 1:09:31 Yeah.

Lex Fridman 1:09:31 But, so that, I mean, that- that's the- it ultimately is a trick of like which to remember, which to forget.

Matt Walker 1:09:37 Right.

Lex Fridman 1:09:38 And the forgetting is so, it's- it's so fascinating you say this. I mean, it seems to be deeply connected to that assimilation process. So forgetting, you try to fit all the new stuff into this big web of the old stuff-

Matt Walker 1:09:55 Yeah.

Lex Fridman 1:09:55 ... and the things that don't fit, you throw out.

Matt Walker 1:09:58 I think the assimilation, the way I've been thinking about it with sleep, and it's particularly sort of dream sleep that we think can help with this assimilation, is that during wake, we have one version of associative processing. And what I mean by that is we see the most obvious connections. So I think of wakefulness as a Google search gone right.

Lex Fridman 1:10:23 Mm-hmm.

Matt Walker 1:10:24 Whereas I see dream sleep as doing something very different. I think dream sleep is a little bit like group therapy for memories, that everyone gets a name badge-

Lex Fridman 1:10:35 Mm-hmm.

Matt Walker 1:10:36 ... and sleep gathers in all of the individual pieces of the day, and it sort of starts to get you to, forces you, in fact, to speak to the people not at the front of the room that you think you've got the most obvious connection with, but to speak with the people all the way at the back of the room that at first you think, "I've got no obvious connection with them at all." But once you get chatting with them, you learn that you do have a very distant, non-obvious connection, but it's still a connection none the same. And it's almost as though you're doing a Google search where, you know, I input, you know, Lex Fridman and it doesn't take me to the first page of your home site.

Lex Fridman 1:11:12 Page 20. (laughs)

Matt Walker 1:11:12 It takes me to page 20-

Lex Fridman 1:11:13 (laughs)

Matt Walker 1:11:13 ... which is about some like field hockey game-

Lex Fridman 1:11:16 Yeah.

Matt Walker 1:11:16 ... in Utah.

Lex Fridman 1:11:17 Yeah, exactly.

Matt Walker 1:11:17 Now it turns out that there actually is a link if I look at it.

Lex Fridman 1:11:20 Yeah.

Matt Walker 1:11:20 It's a distant, non-obvious one. And to me, I find that exciting because when you fuse things together that shouldn't normally go together, but when they do they cause marked advances in evolutionary fitness, it sounds like the biological basis of creativity. And that's exactly what I think dream sleep and the algorithm of dream sleep is designed to do.

Lex Fridman 1:11:40 Mm-hmm.

Matt Walker 1:11:40 You know, it's not a Boolean-like system where you have, you know, the sort of assumptions of true and false. You know, maybe it's more fuzzy logic system. And I think REM sleep is a perfect environment within which we do... You know, it's- it's almost like memory pinball. You know, you get the information that you've learned during the day and then you pull the lever back and you shoot it up into the attic of your brain. You know, this cortex filled with all of your past historical knowledge. And you start to bounce it around and see where one of those things lights up and you build a new connection there and you build another one there too. You're developing schemas. And so in that way, I think you could argue, you know, we dream, therefore we are. (laughs)

Lex Fridman 1:12:27 (laughs) Yeah, so in terms of...... this line between learning and thinking through a new thing, uh, that seems to be deeply, uh, connected. There's this, um, legendary engineer, named Jim Keller, who keeps yelling at me about this, he says it's very effective, he likes to f- for difficult problems, before bed, think about that difficult problem. We're not talking about like drama at work or all that kind of stuff, no, like a scientific, for him, engineering problem. He likes to, like, intensely think about it as to prime his mind-

Matt Walker 1:13:02 Yes.

Lex Fridman 1:13:03 ... before sleeping, then go to sleep, and then, uh-

Matt Walker 1:13:06 Yeah.

Lex Fridman 1:13:06 ... he finds that, uh, the next day, he's, he's able to think much clearer, and there's new ideas that come, but also just, I guess, it's more well-integrated. Um, and sometimes during the process of, uh, like he's able to, like, wake up and, like, see, see new insights-

Matt Walker 1:13:27 That's right.

Lex Fridman 1:13:27 ... if he's deeply, sort of, aggressively thinking through a problem. (laughs) So, he-

Matt Walker 1:13:30 And there's, there's many scientific, you know, demonstrations of this, you know, the- Mendeleev with the periodic table of elements, you know, he was trying for months to understand... I mean talk about a, an ecumenical problem of epic proportions. Here's your question today, (laughs) you have to understand how all of the known elements in the universe fit together-

Lex Fridman 1:13:54 Yeah.

Matt Walker 1:13:54 ... in a logical way. Good luck, take care. It was non-trivial at the time, and he would try and try. He was so obsessed with it, he created playing cards with all of the different elements on, and then he would go on these long train journeys around Europe, and he would just f- sort of, deal these cards in front of them and he would shuffle them, shuffling and shuffling, and he would just try to see if he could find what the answer was.

Lex Fridman 1:14:20 Mm-hmm.

Matt Walker 1:14:20 And then, so the story goes, you know, he fell asleep and he had a dream, and in that dream, you know, all of these elements started to dance and play around, and they snapped into a logical grid-

Lex Fridman 1:14:32 Mm-hmm.

Matt Walker 1:14:32 ... you know, atomic weight, et cetera, et cetera. And it wasn't his waking brain that solved the problem, it was his sleeping brain that solved the impenetrable problem that his waking brain could not. And there's been count- you know, even in the arts and in music, some wonderful dreams, you know, Frankenstein, Mary Shelley's epic Gothic novel, came to her in a dream at Lord Byron's home. Um, and then we've got, you know, Paul McCartney, um, yesterday the song came to him in a dream. He was in, uh, filming, um - gosh what was the movie? Um, I don't recall it. I should be shocked because I'm from Liverpool myself.

Lex Fridman 1:15:16 (laughing)

Matt Walker 1:15:17 Um, and, but he was on Wimpole Street, in London-

Lex Fridman 1:15:20 Yes.

Matt Walker 1:15:20 ... um, and filming, and they, he came up with that song, the melody, um, in his sleep. Not to be outdone by The Beatles, and by the way, Let It Be, um, also came from a dream-

Lex Fridman 1:15:33 Mm-hmm.

Matt Walker 1:15:33 ... that McCartney had. People usually give it, you know, religious overtones, you know, "Mother Mary comes to me speaking words of wisdom, let it be." E- if you've ever asked who Mother Mary is, it's not the, you know, the biblical content, it's his, his mother, it's, it's, it's Mary McCartney, (laughs) and she came to him in a dream, and gifted him the song. But the best story I've heard is, um, not to be outdone by The Beatles, The Stones-

Lex Fridman 1:16:05 (laughs)

Matt Walker 1:16:05 ... um, Keith Richards-

Lex Fridman 1:16:06 Yeah.

Matt Walker 1:16:07 ... uh, who I think once was suggested that... Who was it? It was a comedian who was saying that in an interview with Rolling Stone, Keith Richards suggested or inferred that young kids should not do drugs. And they said-

Lex Fridman 1:16:22 (laughing)

Matt Walker 1:16:22 ... "Well, look-"

Lex Fridman 1:16:24 Yeah.

Matt Walker 1:16:24 "... young kids can't do drugs because you've done all of the drugs."

Lex Fridman 1:16:27 All of them, yeah. (laughing)

Matt Walker 1:16:28 (laughing) And I always thought that... But, um, Keith Richards described, uh, w- he would always go to bed with his, um, guitar-

Lex Fridman 1:16:36 Mm-hmm.

Matt Walker 1:16:36 ... and a tape recorder, um, and then probably he would have a whole set of other things in the bed with him, and who knows how many other people.

Lex Fridman 1:16:45 Yeah. (laughs)

Matt Walker 1:16:46 But anyway, uh, and then he said in his autobiography, and I'm paraphrasing here but, "One morning, I woke up, and I realized that the tape had recorded all the way to the end, so I rewound the tape, and I hit play, and there in some kind of ghostly form were the opening chords to Satisfaction."

Lex Fridman 1:17:08 Mm.

Matt Walker 1:17:09 The most famous successful Rolling Stones song-

Lex Fridman 1:17:11 Yeah.

Matt Walker 1:17:11 ... of all of, all time.

Lex Fridman 1:17:13 Yeah.

Matt Walker 1:17:13 Followed by then 43 minutes of snoring. (laughing)

Lex Fridman 1:17:18 (laughing) That's awesome.

Matt Walker 1:17:19 But that riff came to him, one, one of the most famous riffs in all of rock and roll came to him by way of a dream-inspired insight. So, I think, you know, there is too many of those anecdote- and we've now got the sci- you know, I don't rely on anecdotes-

Lex Fridman 1:17:33 Mm-hmm.

Matt Walker 1:17:33 ... as science. We've now done the studies in the laboratory, and we can reliably demonstrate that sleep inspires creativity, inspires problem-solving capacity.

Lex Fridman 1:17:41 Well, the interesting thing is, is it possible to some of the ideas that you talk about, to turn them into a protocol that could be practiced rigorously? So, what Jim Keller espouses is saying not just the fact that sleep helps you, uh, i- increase the creativity, but turn it into a process, like, literally, like don't do it accidentally, you know, like, (laughs) uh, like an athlete does certain things to optimize their performance, they have a training routine, they have a regimen of, uh, of like cycling and, and then, and, uh, sprints and, and long distance stuff. In the same way, thinking about your job as an idea generator-

Matt Walker 1:18:24 Mm-hmm.

Lex Fridman 1:18:24 ... in the engineering space as like this is good for my performance, so like for an hour before bed, think through a problem, like every night-... and then use sleep to, uh, work through that problem. I mean, that, he's the first person that I heard, like, of the people I really respect that do, like, what I do, which is, like, programming engineering-type work. Like, using sleep, not accidentally, but with a purpose.

Matt Walker 1:18:50 Mm-hmm.

Lex Fridman 1:18:50 Like, using sleep.

Matt Walker 1:18:52 You know, that's just basically the difference between, as you said, a, a passive approach to it versus-

Lex Fridman 1:18:58 Yeah.

Matt Walker 1:18:58 ... you know, an, an active deterministic or hope for a deterministic approach to it. In other words, that you are actually trying to harness the power of, of sleep in a deliberate way, rather than an unthoughtful way. I still think that, you know, Mother Nature through, you know, 3.6 million years of evolution has probably got it mostly figured out, in terms of-

Lex Fridman 1:19:23 Yeah.

Matt Walker 1:19:23 ... what information should be uploaded at night and worked through. I think her algorithm is probably pretty good at this stage. It's not to suggest though that, you know, we can't try to tweak it and nudge it. You know, it's a very light hand on the tiller is, is what he's doing. I, I don't think there's anything wrong with that.

Lex Fridman 1:19:41 You know, just, like, for example, for me, fasting has improved my ability to focus deeply, and productivity significantly, and in that same way, and it's possible that playing with these ideas of thinking before bed or some hours before bed, or some ... Playing with different protocols will have a significant leap over what Mother Nature naturally does. So, if you let your body do what it naturally does, you may not achieve the same level of performance 'cause Mother Nature has not designed us to think deeply about, uh, chip design or, uh, programming artificial intelligence systems. So, (laughs) like-

Matt Walker 1:20:20 Well, she's gifted us the architecture and the capacity to do that.

Lex Fridman 1:20:25 To do that.

Matt Walker 1:20:25 What we do with that is, you know, is what life's experience dictates.

Lex Fridman 1:20:31 Sure.

Matt Walker 1:20:31 She gives us the blueprint to do many, but, you know.

Lex Fridman 1:20:34 Well, if I were to sort of s- introspect and self-analyze what Mother Nature wants me to do, I think, given my current lifestyle, that I have food in the fridge-

Matt Walker 1:20:45 Mm-hmm.

Lex Fridman 1:20:45 ... and, uh, a bed to sleep on, I think what Mother Nature wants me to do is to be lazy. (laughs)

Matt Walker 1:20:51 (laughs)

Lex Fridman 1:20:51 And so, I think I'm actually resisting Mother Nature, uh, in, because so many of my needs are satisfied, and so I h- um, I have to resist some of the natural forces of the body and the mind when I do some of the things I do.

Matt Walker 1:21:07 Yeah.

Lex Fridman 1:21:07 So, th- there's that dance, um, you know, like, I've been thinking about doing a startup, and that's obviously going against everything that my body and mind are telling me to do because it's going to be basically suffering. But the only reason I want-

Matt Walker 1:21:22 Life as you know it will be over. (laughs)

Lex Fridman 1:21:26 (laughs) Yes. I, but nevertheless, there's some kind of inner drive that wants me to do it, and then you start to ask a question, "Well, how do you optimize the things you can optimize, like sleep, like diet, like the people that you surround yourself with, in order to maximize happiness and performance and all those kinds of things, without also over optimizing?" And that's such an interesting idea from, from a engineer. So, as you may know, you don't often get that, those kinds of ideas from engineers. Engineers usually just don't read books about sleeping. (laughs) They're usually, like, the, the, they're not, they, they're not the healthiest of people. Um, I think that's changing over time, especially with Silicon Valley, especially with the tech sector. People are starting to understand what's a healthy lifestyle. But usually, they, they're kind of in, on the insane side, especially programmers. But it's nice to hear, uh, somebody like that use sleep and use some of the things you talk about strategically.

Matt Walker 1:22:30 Yeah.

Lex Fridman 1:22:31 On purpose.

Matt Walker 1:22:33 You know, to that idea of not just trying to use what Mother Nature gave, but seeing if you can do something more or different, in a conservative mindset, I would then pose the question, at what cost?

Lex Fridman 1:22:51 Right.

Matt Walker 1:22:52 Because when you do something, perhaps, that deviates from the typical pre-programmed, you know, Mother Nature's program, um, I suspect it usually comes at the cost of something else. So, maybe he is, you know, able to direct and focus his sleeping cognition-

Lex Fridman 1:23:14 Mm-hmm.

Matt Walker 1:23:14 ... on those particular topics that will gain him better problematic resolution the next day when he wakes up. The question is, though, at what cost of the other things that didn't make it onto the menu of the finger buffet of sleep that night?

Lex Fridman 1:23:31 Mm-hmm.

Matt Walker 1:23:32 And is it that you don't process the emotional difficulties or events, and therefore, you are less emotionally resolved the next day, but you are more problem resolved the following day? And so, I always try to think, and I, I truly don't want to sound puritanical either about sleep, and I think I've come off that way, um, many a time, especially when I started out in the public. Um, um, the tone of the book, in some ways, you know, I look back and think, "Could I have been a little softer?" And the reason was I, I was that way back in, when I started writing the book, which was probably something like 2014 or '15, sleep was the neglected step-sister in the health conversation-

Lex Fridman 1:24:19 Mm-hmm.

Matt Walker 1:24:20 ... of the day. And I was just so sad to see the amount of suffering and disease and sickness that was caused by insufficient sleep. And for years before, I'd been, you know, doing public speaking, and I'd s- tell people about the great things that happen when you get sleep. People would say, "That's fascinating." And then they would go back and keep doing the same thing about not sleeping enough. And then I realized-... you can't really speak about the good things that happen. It's like the news, what bleeds, leads. And if you speak about the alarmingly bad things that happen, people tend to have a behavioral change.

Lex Fridman 1:24:52 Mm-hmm.

Matt Walker 1:24:53 And so, the book, as a consequence, I think probably came out a little bit on the strong side of, you know, trying to convince, you know, people.

Lex Fridman 1:25:03 Well, you were trying to help a lot of people and that's a powerful way to help a lot of people.

Matt Walker 1:25:07 I w- I was genuinely trying to help people, but, you know, f- certainly for some people who, for whom sleep is not, does not come easy, then it was probably, you know, a tricky book to read too. And I think I, I feel more sensitive to those people now and empa- empathetically connected to them. Um, so I think the ... Again, the point was simply that I don't mean to sound too puritanical in, in all of this.

Lex Fridman 1:25:32 Yeah.

Matt Walker 1:25:32 And the same way with, you know, caffeine and coffee.

Lex Fridman 1:25:36 Mm-hmm.

Matt Walker 1:25:37 I am just a scientist, and I am not here to tell anyone how to live their life. That is not my job at all. And life is to be lived, to a degree, and life is to be lived if you want to do a startup.

Lex Fridman 1:25:50 Mm-hmm.

Matt Walker 1:25:51 All I want to do is empower people with the understanding of the science of sleep, and then you can make an informed choice as to how you want to live your life, and I offer no judgment on how anyone wishes to live their life. I just want to try and see if the information that I have about sleep would alternatively change how you would think about your life decisions, and if it doesn't, no problem, and if it does, I, I hope it's been of use.

Lex Fridman 1:26:18 Well,

Lex Fridman 1:26:19 maybe this is me trying to justify my lifestyle to you, but, uh, Dr. Seuss said, "You know you're in love when you can't fall asleep because reality is finally better than your dreams."

Matt Walker 1:26:32 (laughs) Mm-hmm. I love that quote too.

Lex Fridman 1:26:34 (laughs) Okay. My sleeping, uh, schedule is, uh, complicated, and it has to do primarily with the fact that I love, uh, basically everything that I do, and that love takes a form that may not appear to be love from the external observer perspective, 'cause it's often includes struggle, it often includes something that looks like stress, even though it's not stress. It's like this excitement, it's this-

Matt Walker 1:27:05 (laughs)

Lex Fridman 1:27:06 ... turmoil and chaos of passion, of struggling with a problem, of being sad and down, uh, to the point even depressed of how difficult the problem is, the disappointment that the last few weeks and months have been a failure and, uh, self-doubt, all that mix, but I, I, I love it. And a part of that is sometimes staying up all night, working on a thing I'm really passionate about, and that means sleep schedules that are just like, you know, uh, sometimes sleeping during the day, sometimes very often sleeping very little, but taking naps that are like an hour or two hours or so on, that kind of weird chaos. And, and-

Matt Walker 1:27:49 Yeah.

Lex Fridman 1:27:49 Now, I'll also s- to try to give, give myself backup.

Matt Walker 1:27:54 (laughs)

Lex Fridman 1:27:54 I was trying to like, uh, research yesterday, is anybody else productive while like this? And there is, of course, a lot of anecdotal evidence, and some of it, um, could be just narratives that people have told to the public when in reality they sleep way more, but there's a bunch of people that, uh, you know, have not, are f- are famous for not sleeping much.

Matt Walker 1:28:16 Right.

Lex Fridman 1:28:16 So, uh, on the topic of naps, uh, this, I read this a long time ago and I, I checked this. Churchill was big on big naps.

Matt Walker 1:28:25 Mm-hmm.

Lex Fridman 1:28:26 And is actually just reading more about Winston Churchill's sleep schedule is very much like mine. So, I