Andrew Huberman: Neuroscience of Optimal Performance
Guests: Andrew Huberman
Full transcript · Read the full conversation in order, with its original context.
Contents19 chapters
§01Transcript · 00:00–08:37§02Transcript · 08:40–16:23§03Transcript · 16:23–25:05§04Transcript · 25:09–33:31§05Transcript · 33:33–41:18§06Transcript · 41:20–49:48§07Transcript · 49:50–58:00§08Transcript · 58:01–1:06:16§09Transcript · 1:06:18–1:13:40§10Transcript · 1:13:41–1:21:33§11Transcript · 1:21:35–1:28:53§12Transcript · 1:28:56–1:36:45§13Transcript · 1:36:49–1:45:24§14Transcript · 1:45:26–1:53:29§15Transcript · 1:53:33–2:02:46§16Transcript · 2:02:50–2:11:25§17Transcript · 2:11:26–2:20:18§18Transcript · 2:20:21–2:28:51§19Transcript · 2:28:54–2:32:08§01Transcript · 00:00–08:37
The following is a conversation with Andrew Huberman,
a neuroscientist at Stanford,
working to understand how the brain works,
how it can change through experience,
and how to repair brain circuits
damaged by injury or disease.
He has a great Instagram account @hubermanlab,
where he teaches the world
about the brain and the human mind.
Also, he's a friend and an inspiration
in that he shows that you can be humble,
giving, and still succeed in the science world.
Quick mention of each sponsor,
followed by some thoughts related to the episode.
8Sleep, a mattress that cools itself
and gives me yet another reason to enjoy sleep.
SEMrush, the most advanced SEO optimization tool
I've ever come across,
and Cash App, the app I use to send money to friends.
Please check out these sponsors in the description
to get a discount and to support this podcast.
As a side note, let me say that I heard from a lot of people
about the previous conversation I had with Yaron Brook
about objectivism.
Some people loved it, some people hated it.
I misspoken some parts,
was more critical on occasion than I meant to be,
didn't push on certain points that I should have,
was undereducated or completely unaware
about some major things that happened in the past
or major ideas out there.
I bring all that up to say
that if we are to have difficult conversations,
we have to give each other space to make mistakes,
to learn, to grow.
Taking one or two statements from a three-hour podcast
and suggesting that they encapsulate who I am,
I was, or ever will be,
is a standard that we can't hold each other to.
I don't think anyone could live up to that kind of standard.
At least I know I can't.
The conversation with Yaron is mild
relative to some conversations
that I will likely have in the coming year.
Please continue to challenge me,
but please try to do so with love and with patience.
I promise to work my ass off to improve.
Whether I'm successful at that or not, we shall see.
If you enjoy this thing, subscribe on YouTube,
review it with 5 Stars on Apple Podcasts,
follow on Spotify, support on Patreon,
or connect with me on Twitter @LexFriedman.
And now, here's my conversation with Andrew Huberman.
You've mentioned that in your lab at Stanford,
you induce stress by putting people into a virtual reality
and having them go through one of a set of experiences.
I think you mentioned this on Rogan or with Whitney,
that scare them.
So just on a practical, psychological level,
and maybe on a philosophical level,
what are people afraid of?
What are the fears?
What are these fear experiences
that you find to be effective?
- Yeah, so it depends on the person, obviously.
And we should probably define fear, right?
'Cause you can, without going too far down the rabbit hole
of defining these things,
you can't really have fear without stress,
but you could have stress without fear.
And you can't really have trauma without fear and stress,
but you could have fear and stress without trauma.
So, we can start playing the word game,
and that actually is one of the motivations
for even having a laboratory that studies these things,
is that we really need better physiological,
neuroscientific, and operational definitions
of what these things are.
I mean, the field of understanding emotions and states,
which is mainly what I'm interested in, is very complicated.
But we can do away with a lot of complicated debate
and say, in our laboratory,
what we're looking for to assign it a value of fear
is a big inflection in autonomic arousal,
so increases in heart rate, increases in breathing,
perspiration, pupil dilation,
all the hallmark signature features of the stress response.
And in some cases,
we have the benefit of getting neurosurgery patients
where we've got electrodes in their amygdala
and their insula and the orbital frontal cortex
down beneath the skull.
So these are chronically implanted electrodes,
we're getting multi-unit signals,
and we can start seeing some central features
of meaning within the brain.
And what's interesting is that,
as trivial as it might seem in listening to it,
almost everybody responds to heights
and falling from a high virtual place
with a very strong stress, if not fear response.
And that's because the visual vestibular apparati, right?
The optic flow and how it links
to the balanced semicircular canals of the inner ears,
all this technical stuff,
but really all of that pulls all your physiology,
the feeling that your stomach is dropping,
the feeling that suddenly you're sweating
even though you're not afraid
of falling off this virtual platform,
but you feel as if you're falling
because of the optic flow, that one is universal.
So we've got a dive with great white sharks experience
where you actually exit the cage.
We went out and did this in the real world
and brought back 360 video that's built out pretty-
- Oh, so this is actually 360 video.
- 360 video. - That's awesome.
- And this was important to us, right?
So when we decided to set up this platform,
a lot of the motivation was that a lot of the studies
of these things in laboratories,
I don't want to call them lame
because I want to be respectful
of the people that did this stuff before,
but they'd study fear by showing subjects
a picture of a bloody arm or a snake
or something like that.
And it just, unless you have a snake phobia,
it just wasn't creating a real enough experience.
So we need to do something where people
aren't going to get injured,
but where we can tap into the physiology
and that thing of presence of people momentarily,
not the whole time, but momentarily
for getting there in a laboratory.
And so heights will always do it.
And if people want to challenge me on this,
I like to point to that movie "Free Solo,"
which was wild because it's incredible movie,
but I think a lot of its popularity
can be explained by a puzzle,
which is you knew he was going to live
when you walked in the theater
or you watched it at home.
You knew before that he survived,
and yet it was still scary that people somehow
were able to put themselves into that experience
or into Alex's experience enough
that they were concerned or worried
or afraid at some level.
So heights always does it.
If we get people who have generalized anxiety,
these are people who wake up and move through life
at a generally higher state of autonomic arousal
and anxiety, then we can tip them
a little bit more easily with things
that don't necessarily get everyone afraid.
Things like claustrophobia, public speaking,
that's going to vary from person to person.
And then if you're afraid of sharks,
like my sister, for instance, is afraid of sharks,
she won't even come to my laboratory
because there's a thing about sharks in it.
That's how terrified some people are
of these specific stimuli.
But heights gets them every time.
- Yeah, I'm terrified of heights.
- We have you step off a platform, virtual platform,
and it's a flat floor in my lab, but you're up there.
- Well, you actually allow them the possibility
in the virtual world to actually take the leap of faith.
- Yeah, maybe I should describe
a little bit of the experiment.
So without giving away too much
in case someone wants to be a subject
in one of these experiments,
we have them playing a cognitive game.
It's a simple lights out kind of game
where you're pointing a cursor
and turning out lights on a grid,
but it gets increasingly complex and it speeds up on them.
And there's a failure point for everybody
where they just can't make the motor commands fast enough.
And then we surprise people essentially
by placing them virtually.
All of a sudden, they're on a narrow platform
between two buildings.
And then we encourage them or we cue them
by talking to them through a microphone
to continue across that platform to continue the game.
And some people, they actually will get down on the ground
and hold onto a virtual beam
that doesn't even exist on a flat floor.
And so what this really tells us is the power of the brain
to enter these virtual states as if they were real.
And we really think that anchoring the visual
and the vestibular, the balance components
of the nervous system are what bring people
into that presence so quickly.
§02Transcript · 08:40–16:23
There's also the potential, and we haven't done this yet,
to bring in 360 sound.
So the reason we did 360 video
is that when we started all this back in 2016,
a lot of the VR was pretty lame, frankly.
It was CGI.
It just wasn't real enough.
But with 360 video,
we knew that we could get people into this presence
where they think they're in a real experience more quickly.
And our friend, Michael Muller,
who I was introduced to because of the project,
I reached out to some friends.
Michael Muller's a very famous
portrait photographer in Hollywood,
but he dives with great white sharks and he leaves the cage.
And so we worked with him to build a 360 video apparatus
that we could swim underwater with,
went out to Guadalupe Island, Mexico,
and actually got the experience.
It was a lot of fun.
There were some interesting moments out there of danger,
but it came back with that video
and built that for the sharks.
And then we realized we need to do this for everything.
We need to do it for heights.
We need to do it for public speaking,
for claustrophobia.
And what's missing still is 360 sound
where 360 sound would be, for instance,
if I were to turn around
and there was a giant attack dog there,
the moment I would turn around and see it,
the dog would growl.
But if I turned back toward you, then it would be silent.
So, and that brings a very real element
to one's own behavior
where you don't know what's gonna happen
if you turn a corner.
Whereas if there's a dog growling behind me
and I turn around and then I turn back to you
and it's still growling,
that might seem like more of an impending threat,
unsustained threat,
but actually it's when you start linking
your own body movements to the experience.
So when it's closed loop,
where my movements and choices
are starting to influence things
and they're getting scarier and scarier,
that's when you can really drive people's nervous system
down these paths of high states of stress and fear.
Now, we don't wanna traumatize people, obviously,
but we also study a number of tools
that allow them to calm themselves in these environments.
So the short answer is heights.
- Heights.
- Yeah.
- Well, from a psychology
and from a neuroscience perspective,
this whole construction that you've developed
is fascinating.
We did this a little bit with autonomous vehicles.
So to try to understand the decision-making process
of a pedestrian when they cross the road
and trying to create an experience of a car
that can run you over,
so there's the danger there.
I was so surprised how real that whole world was.
And the graphics that we built
wasn't ultra realistic or anything,
but I was still afraid of being hit by a car.
Everybody we tested were really afraid
of being hit by that car.
- Even though it was all a simulation?
- It was all a simulation.
It was kind of boxy, actually.
I mean, it wasn't like ultra realistic simulation.
It was fascinating.
- Looms and heights.
So any kind of depth,
we're just programmed to not necessarily recoil,
but to be cautious about that edge and that depth.
And then looms, things coming at us that are getting larger.
There are looming sensing neurons, even in the retina,
at a very, very early stage of visual processing.
And incidentally, the way Muller
and folks learn how to not get eaten by great white sharks
when you're swimming outside the cage
is as they start lumbering in, you swim toward them.
And they get very confused when you loom on them,
because clearly you're smaller.
Clearly they could eat you if they wanted to,
but there's something about forward movement
toward any creature that that creature questions
whether or not it would be a good idea
to generate forward movement toward you.
And so that's actually the survival tool
of these cage exit white shark divers.
- Are you playing around with,
like one of the critical things
for the autonomous vehicle research
is you couldn't do 360 video because there's a game
theoretic, there's an interactive element
that's really necessary.
So maybe people realize this, maybe they don't,
but 360 video, you obviously,
well, it's actually not that obvious to people,
but you can't change the reality that you're watching.
- That's right.
- So, but you find that that's,
is there something fundamental about fear and stress
that the interactive element is essential for?
Or do you find you can arouse people with just the video?
- Great question.
It works best to use mixed reality.
So we have a snake stimulus.
I personally don't like snakes at all.
I don't mind spiders.
We also have a spider stimulus,
but like snakes, I just don't like them.
There's something about the slithering
and it just creates a visceral response for me.
Some people, not so much,
and they have lower levels of stress and fear in there.
But one way that we can get them to feel more of that
is to use mixed reality,
where we have an actual physical bat
and they have to stomp out the snake,
as opposed to just walk to a little safe corner,
which then makes the snake disappear.
That tends to be not as stressful
as if they have a physical weapon.
And so you've got people in there, you know,
banging on the floor against this thing.
And there's something about engaging
that makes it more of a threat.
Now, I should also mention,
we always get the subjective report
from the subject of what they experienced,
because we never want to project our own ideas
about what they were feeling.
But that's the beauty of working with humans
is you can ask them how they feel.
- Exactly.
- And humans aren't great at explaining how they feel,
but it's a lot easier to understand what they're saying
than a mouse or a macaque monkey is saying.
So it's the best we can do is language,
plus these physiological and neurophysiological signals.
- Is there something you've learned about yourself,
about your deepest fears?
Like you said snakes, is there something that,
like if I were to torture you, so I'm Russian,
so, you know, I always kind of think,
how can I murder this person that entered the room?
But also how could I torture you
to get some information out of you?
What would I go with?
- It's interesting you should say
that I never considered myself claustrophobic,
but, 'cause I don't mind small environments
provided they're well ventilated.
But I, before COVID, I started going to this Russian banya,
you know, and then we jumped here.
And I had never been to a banya.
So, you know, the whole experience
of really, really hot sauna.
And what do they call it?
The plaza, they're hitting you with the leaves.
And it gets really hot and humid in there.
And there were a couple of times where I thought,
okay, this thing is below ground.
It's in a city where there are a lot of earthquakes.
Like if this place crumbled and we were stuck in here,
and I'd start getting a little panicky.
And I realized, I'm like,
I don't like small confined spaces with poor ventilation.
So I realized, I think I have some claustrophobia.
And I wasn't aware of that before.
So I put myself into our own claustrophobia stimulus,
which involves getting into an elevator
and with a bunch of people, virtual people,
and the elevator gets stalled.
And at first you're fine, you feel fine.
But then as we start modulating the environment
and we actually can control levels of oxygen
in the environment if we want to,
it is really uncomfortable for me.
And I never would have thought, you know,
I fly, I'm comfortable in planes,
but it is really uncomfortable.
And so I think I've unhatched a bit of a claustrophobia.
- Yeah. - Yeah.
- Yeah, for me as well, probably.
That one, that one is pretty bad.
The heights I tried to overcome.
So I went to skydiving to try to overcome
the fear of heights, but that didn't help.
- Did you jump out?
§03Transcript · 16:23–25:05
- Yeah, I jumped out, but it was fundamentally
different experience than, I guess there could be
a lot of different flavors of fear of heights maybe.
But the one I have didn't seem to be connected
to jumping out of a plane.
It's a very different, 'cause like once you accept it,
you're going to jump, then it's a different thing.
I think what I'm afraid of is the moments before it
is the scariest part.
- Absolutely.
- And I don't think that's emphasized
in the skydiving experience as much.
And also just the acceptance of the fact
that it's going to happen.
So once you accept it is going to happen,
it's not as scary.
It's the fact that it's not supposed to happen
and it might, that's the scary part.
I guess I'm not being eloquent in this description,
but there's something about skydiving
that was actually philosophically liberating.
I was like, wow, it was the possibility
that you can walk on a surface.
And then at a certain point, there's no surface anymore
to walk on.
And it's all of a sudden the world becomes three dimensional
and there's this freedom of floating
that the concept of like, of earth disappears
for a brief few seconds.
I don't know, that was--
- Wild.
- That was wild, but I'm still terrified of heights.
So, I mean, one thing I want to ask just on fear,
'cause it's so fascinating is,
have you learned anything about
what it takes to overcome fears?
- Yes.
And that comes from two, from a research study standpoint,
two parallel tracks of research.
One was done actually in mice,
'cause we have a mouse lab also
where we can probe around different brain areas
and try and figure out what interesting brain areas
we might want to probe around in humans.
And a graduate student in my lab,
she's now at Caltech, Lindsay Sillay,
published a paper back in 2018,
showing that what at first might seem a little bit obvious,
but the mechanisms are not,
which is that there are really three responses to fear.
You can pause, you can freeze essentially.
You can retreat, you can back up, or you can go forward.
And there's a single hub of neurons in the midbrain,
it's actually not the midbrain,
but it's in the middle of the thalamus,
which is a four brain structure.
And depending on which neurons are active there,
there's a much higher probability that a mouse,
or it turns out, or a human will advance in the face of fear
or will pause or will retreat.
Now, that just assigns a neural structure
to a behavioral phenomenon.
But what's interesting is that it turns out
that the lowest level of stress or autonomic arousal
is actually associated with the pausing
and freezing response.
Then as the threat becomes more impending,
and we used visual looms in this case,
the retreat response has a slightly higher level
of autonomic arousal and stress.
So think about playing hide and go seek,
and you're trying to stay quiet
in a closet that you're hiding.
If you're very calm, it's easy to stay quiet and still.
As your level of stress goes up,
it's harder to maintain that level of quiet and stillness.
You see this also in animals that are stalking,
a cat will chatter its teeth.
That's actually sort of top-down inhibition
and trying to restrain behavior.
So the freeze response is actually an active response,
but it's fairly low stress.
And what was interesting to us is that
the highest level of autonomic arousal
was associated with the forward movement toward the threat.
So in your case, jumping out of the plane.
However, the forward movement in the face of threat
was linked to the activation of what we call collateral,
which means just a side connection,
literally a wire in the brain
that connects to the dopamine circuits for reward.
And so when one safely and adaptively,
meaning you survive,
moves through a threat or toward a threat,
it's rewarded as a positive experience.
And so the key,
it actually maps very well to cognitive behavioral therapy
and a lot of the existing treatments for trauma,
is that you have to confront
the thing that makes you afraid.
So otherwise you exist in this very low level
of reverberatory circuit activity
where the circuits for autonomic arousal are humming
and they're humming more and more and more.
And we have to remember that stress and fear and threat
were designed to agitate us so that we actually move.
So the reason I mentioned this is I think
a lot of times people think that
the maximum stress response or fear response
is to freeze and to lock up.
But that's actually not the maximum stress response.
The maximum stress response is to advance,
but it's associated with reward.
It has positive valence.
So there's this kind of,
everyone always thinks about the bell,
the sort of hump shape curve for,
at low levels of arousal performance is low
and as it increases performance goes higher
and then it drops off as you get really stressed.
But there's another bump further out
of the distribution where you perform very well
under very high levels of stress.
And so we've been spending a lot of time in humans
and in animals exploring what it takes
to get people comfortable to go to that place.
And also to let them experience
how there are heightened states of cognition there.
There's changes in time perception that allow you
to evaluate your environment
at a faster frame rate, essentially.
This is the matrix as a lot of people think of it.
But we tend to think about fear as all the low level stuff
where things aren't worked out.
But there are many,
there are a lot of different features to the fear response.
And so we think about it quantitatively
and we think about it from a circuit perspective
in terms of outcomes.
And we try and weigh that against the threat.
So we never want people to put themselves
in unnecessary risk, but that's where the VR is fun
because you can push people hard
without risk of physically injuring them.
- And that's, like you said, the little bump
that seems to be a very small fraction
of the human experience, right?
So it's kind of fascinating to study it
because most of us move through life
without ever experiencing that kind of focus.
- Well, everything's in a peak state there.
I really think that's where optimal performance lies.
- There's so many interesting words here,
but what's performance and what's optimal performance?
We're talking about mental ability to what?
To perceive the environment quickly,
to make actions quickly.
What's optimal performance?
- Yeah, well, it's very subjective
and it varies depending on task and environment.
So one way where we can make it a little bit more operational
and concrete is to say there is a sweet spot, if you will,
where the level of internal autonomic arousal,
aka stress or alertness, whatever you want to call it,
is ideally matched to the speed of whatever challenge
you have to be facing in the outside world.
So we all have perception of the outside world
as exteroception and the perception
of our internal real estate interoception.
And when those two things,
when interoception and exteroception are matched
along a couple of dimensions,
performance tends to increase
or tends to be in an optimal range.
So for instance, if you're, I don't play guitar,
but I know you play guitar.
So let's say you're trying to learn something new
on the guitar.
I'm not saying that being in these super high states
of activation are the best place for you
to be in order to learn.
It may be that your internal arousal needs to be at a level
where your analysis of space and time
has to be well-matched to the information coming in
and what you're trying to do in terms of performance,
in terms of playing chords and notes and so forth.
Now, in these cases of high threat
where things are coming in quickly
and animals and humans need to react very quickly,
the higher your state of autonomic arousal, the better,
because you're slicing time more finely,
just because of the way the autonomic system works.
The pupil dilation, for instance,
and movement of the lens essentially changes your optics.
And that's obvious, but with the change in optics
is a change in how you bin time and slice time,
which allows you to get more frames per second readout.
With the guitar learning, for instance,
it might actually be that you want to be almost sleepy,
almost in a kind of drowsy state to be able to,
§04Transcript · 25:09–33:31
and I don't play music, so I'm guessing here,
but sense some of the nuance in the chords
or the ways that you're to be relaxed enough
that your fingers can follow an external cue.
So matching the movement of your fingers
to something that's pure exteroception.
And so there is no perfect autonomic state for performance.
This is why I don't favor terms like flow,
because they're not well operationally defined enough,
but I do believe that optimal or peak performance
is going to arise when internal state is ideally matched
to the space-time features of the external demands.
- So there's some slicing of time that happens,
and then you're able to adjust,
so slice time more finely or less finely
in order to adjust to the dynamics of the stimulus.
What about the realm of ideas?
So like, I'm a big believer,
there's a guy named Cal Newport
who wrote a book about deep work.
- Oh yeah, I love that book.
- Yeah, he's great.
I mean, one of the nice things,
I've always practiced deep work,
but it's always nice to have words
put to the concepts that you've practiced.
It somehow makes them more concrete
and allows you to get better.
It turns it into a skill that you can get better at.
But I also value deep thinking,
where you think, it's almost meditative.
You think about a particular concept
for long periods of time.
The programming you have to do that kind of thing for.
You just have to hold this concept,
like you hold it and then you take steps with it,
you take further steps,
and you're holding relatively complicated things
in your mind as you're thinking about them.
And there's a lot of,
I mean, the hardest part is there's frustrating things,
like you take a step
and it turns out to be the wrong direction,
so you have to calmly turn around and take a step back.
And then it's, you're kind of like exploring
through the space of ideas.
Is there something about your study of optimal performance
that could be applied to the act of thinking
as opposed to action?
- Well, we haven't done too much work there,
but I think I can comment on it
from a neuroscience perspective.
- Yeah, what's your intuition?
- Which is really all I do is,
well, I mean, we do experiments in the lab,
but looking at things through the lens of neuroscience.
So what you're describing can be mapped fairly well
to working memory,
just keeping things online and updating them
as they change in information
and it's coming back into your brain.
Jack Feldman, who I'm a huge fan of
and fortunate to be friends with,
is a professor at UCLA,
works on respiration and breathing,
but he has a physics background.
And so he thinks about respiration and breathing
in terms of ground states
and how they modulate other states.
Very, very interesting and I think important work.
Jack has an answer to your question.
So I'm not going to get this exactly right
'cause this is lifted from a coffee conversation
that we had about a month ago.
Apologies in advance,
but I think I can get mostly right.
So we were talking about this,
about how the brain updates cognitive states
depending on demands and thinking in particular.
And he used an interesting example,
I'd be curious to know if you agree or disagree.
He said, "Most great mathematics is done by people
"in their late teens and 20s,
"and even you could say early 20s,
"sometimes into the late 20s,
"but not much further on."
Maybe I just insulted some mathematicians.
- No, that's true.
- And I think that it demands,
his argument was there's a tremendous demand
on working memory to work out theorems in math
and to keep a number of plates spinning,
so to speak, mentally,
and run back and forth between them, updating them.
In physics, Jack said,
and I think this makes sense to me too,
that there's a reliance on working memory,
but an increased reliance on some sort of deep memory
and deep memory stores,
probably stuff that's moved out of the hippocampus
and forebrain and into the cortex,
and is more some episodic and declarative stuff,
but really, so you're pulling from your library, basically.
It's not all RAM, it's not all working memory.
And then in biology,
and physicists tend to have very active careers
into their 30s and 40s and 50s and so forth,
sometimes later.
And then in biology,
you see careers that have a much longer arc,
kind of these protracted careers often,
people still in their 60s and 70s
doing really terrific work,
not always doing it with their own hands
'cause the people in the labs are doing them, of course,
and that work does tend to rely on insights gained
from having a very deep knowledge base,
where you can remember a paper,
or maybe a figure in a paper,
you could go look it up if you wanted to,
but it's very different
than the working memory of the mathematician.
And so when you're talking about coding
or being in that tunnel of thought
and trying to iterate and keeping a lot of plates spinning,
it speaks directly to working memory.
My lab hasn't done too much of that.
- With working memory?
- But we are pushing working memory
when we have people do things
like these simple lights out tasks,
while they're under,
we can increase the cognitive load
by increasing the level of autonomic arousal
to the point where they start doing less well.
And everyone has a cliff.
This is what's kind of fun.
We've had SEAL team operators come to the lab,
we've had people from other units in the military,
we've had a range of intellects and backgrounds
and all sorts of things,
and everyone has a cliff.
And those cliffs sometimes show up
as a function of the demands of speed of processing
or how many things you need to keep online.
I mean, we're all limited at some point
in the number of things we can keep online.
So what you're describing is very interesting
because I think it has to do with how narrow
or broad the information set is.
And I'm not an active programmer,
so this is a regime I don't really fully know,
so I don't want to comment about it in any way.
Doesn't suggest that,
but I think that what you're talking about
is top-down control.
So this is prefrontal cortex
keeping every bit of reflexive circuitry at bay.
The one that makes you want to get up and use the restroom,
the one that makes you want to check your phone,
all of that,
but also running these anterior thalamus
to prefrontal cortex loops,
which we know are very important for working memory.
- Yeah, let me try to think through this a little bit.
So,
reducing the process of thinking
to working memory access is tricky.
It's probably ultimately correct,
but if I were to say some of the most challenging things
that an engineer has to do,
and a scientific thinker,
I would say it's kind of depressing
to think that we do that best in our 20s,
but is this kind of first principles thinking step
of saying,
you're accessing the things that you know,
and then saying,
well, let me,
how do I do this differently than I've done it before?
This weird like,
stepping back, like,
is this right?
Let's try it this other way.
That's the most mentally taxing step,
is like,
you've gotten quite good at this particular pattern
of how you solve this particular problem.
So, there's a pattern recognition first.
You're like, okay,
I know how to build a thing
that solves this particular problem in programming, say.
And then,
the question is,
but can I do it much better?
And I don't know if that's,
I don't know what the hell that is.
I don't know if that's accessing working memory.
That's almost,
maybe it is accessing memory in a sense
§05Transcript · 33:33–41:18
that's trying to find similar patterns
in a totally different place
that it could be projected onto this.
But you're not querying facts,
you're querying like functional things, like.
- Yeah, it's patterns.
I mean, you're running out,
you're testing algorithms.
- Yeah.
- Right, you're testing algorithms.
So, I want to just,
because I know some of the people listening to this,
and you have basis in scientific training
and have scientific training.
So, I want to be clear.
I think we can be correct about some things
like the role of working memory
in these kinds of processes without being exhaustive.
We're not saying that the only thing,
we're not, you know,
we can be correct,
but not assume that that's the only thing involved, right?
And I mean, neuroscience, let's face it,
is still in its infancy.
I mean, we probably know 1% of what there is
to know about the brain.
You know, we've learned so much,
and yet there may be global states that underlie this
that make prefrontal circuitry work differently
than it would in a different regime,
or even time of day.
I mean, there's a lot of mysteries about this.
But, so I just want to make sure that we sort of are,
we're aiming for precision and accuracy,
but we're not going to be exhaustive.
So, there's a difference there.
And I think, you know,
sometimes in the vastness of the internet,
that gets forgotten.
So, the other is that,
you know, we think about,
you know, we think about these operations
at, you know, really focused,
keeping a lot of things online.
But what you were describing is actually,
it speaks to the very real possibility,
probably that with certainty,
there's another element to all this,
which is when you're trying out lots of things,
in particular, lots of different algorithms,
you don't want to be in a state
of very high autonomic arousal.
That's not what you want,
because the higher level of autonomic arousal
and stress in the system,
the more rigidly you're going to analyze space and time.
And what you're talking about
is playing with space-time dimensionality.
And I want to be very clear.
I mean, I'm the son of a physicist.
I am not a physicist.
When I talk about space and time,
I'm literally talking about visual space
and how long it takes for my finger
to move from this point to this point.
- You are facing a tiger and trying to figure out
how to avoid being eaten by the tiger.
- And that's primarily going to be determined
by the visual system in humans.
We don't walk through space, for instance,
like a scent hound would,
and look at three-dimensional scent plumes.
You know, when a scent hound goes out in the environment,
they have depth to the odor trails they're following.
And they don't think about them,
we don't think about odor trails.
You might say, oh, well, the smell's getting more intense.
Aha, but they actually have three-dimensional odor trails.
So they see a cone of odor.
See, of course, with their nose,
with their olfactory cortex.
We do that with our visual system
and we parse time often subconsciously,
mainly with our visual system,
also with our auditory system.
And this shows up for the musicians out there,
metronomes are a great way to play with this.
You know, bass drumming,
when the frequency of bass drumming changes,
your perception of time changes quite a lot.
So in any event, space and time are linked
in the through the sensory apparati,
through the eyes and ears and nose,
and probably through taste too,
and through touch for us, but mainly through vision.
So when you drop into some coding
or iterating through a creative process
or trying to solve something hard,
you can't really do that well
if you're in a rigid high level of autonomic arousal,
because you're plugging in algorithms
that are in this space regime, this time regime matches.
It's space time matched.
Whereas creativity, I always think the lava lamp
is actually a pretty good example,
even though it has these counterculture
new agey connotations,
because you actually don't know
which direction things are gonna change.
And so in drowsy states, sleeping and drowsy states,
space and time become dislodged from one another somewhat,
and they're very fluid.
And I think that's why a lot of solutions come to people
after sleep and naps.
And this could even take us into a discussion
if you like about psychedelics.
And what we now know, for instance,
that people thought that psychedelics work
by just creating a spontaneous bursting of neurons
and hallucinations, but the 5H, 2C and 2A receptors,
which are the main sites for things like LSD and psilocybin
and some of the other, the ones that create hallucinations,
the drugs that create hallucinations,
the most of those receptors are actually
in the collection of neurons that encase the thalamus,
which is where all the sensory information goes into,
a structure called the thalamic reticular nucleus.
And it's an inhibitory structure that makes sure
that when we're sitting here talking,
that I'm mainly focused on whatever I'm seeing visually,
that I'm essentially eliminating
a lot of sensory information.
Under conditions where people take psychedelics
and these particular serotonin receptors are activated,
that inhibitory shell, it's literally shaped like a shell,
starts losing its ability to inhibit
the passage of sensory information.
But mostly the effects of psychedelics
are because the lateral connectivity in layer five
of cortex across cortical areas is increased.
And what that does is that means
that the space-time relationship for vision,
like moving my finger from here to here,
very rigid space-time relationship, right?
If I slow it down, it's slower, obviously,
but there's a prediction that can be made
based on the neurons in the retina and the cortex.
On psychedelics, this could be very strange experience.
But the auditory system has one
that's slightly different space-time
and they're matched to one another
in deeper circuits in the brain.
The olfactory system has a different
space-time relationship to it.
So under conditions of these increased activation
of these serotonin receptors,
space and time across sensory areas starts being fluid.
So I'm no longer running the algorithm
for moving my finger from here to here
and making a prediction based on vision alone.
I'm now, this is where people talk about hearing sites,
right?
You start linking, this might actually make a sound
in a psychedelic state.
Now I'm not suggesting people run out and do psychedelics
because it's very disorganized,
but essentially what you're doing
is you're mixing the algorithms.
And so when you talk about being able
to access new solutions,
you don't need to rely on psychedelics.
If people choose to do that, that's their business.
But in drowsy states,
this lateral connectivity is increased as well.
The shell of the thalamus shuts down.
And what's, these are through these so-called
PONS, Chiniculate Occipital Waves.
And what's happening is you're getting
whole brain activation at a level
that you start mixing algorithms.
And so sometimes I think solutions come
not from being in that narrow tunnel of space-time
and strong activation of working memory
and trying to, well, iterate if this, then this,
very strong deductive and inductive thinking
and working from first principles,
but also from states where something
that was an algorithm that you never had in existence before
suddenly gets lumped with another algorithm.
And all of a sudden a new possibility comes to mind.
And so space and time need to be fluid
and space and time need to be rigid
in order to come up with something meaningful.
And I realize I'm riffing long on this,
but this is why I think, you know,
there was so much interest a few years ago
with Michael Pollan's book and other things happening
about psychedelics as a pathway to exploration
and all this kind of thing.
But the real question is what you export back
§06Transcript · 41:20–49:48
from those experiences.
'Cause dreams are amazing,
but if you can't bring anything back from them,
they're just amazing.
- I wonder how to experiment with the mind
without any medical assistance first.
I push my mind in all kinds of directions.
I definitely want to, I did shrooms a couple of times.
I definitely want to figure out how I can experiment
with psychedelics.
I'm talking to Rick Doblin, I think.
- Doblin. - Doblin.
Soon, I went back and forth.
So he does all these studies in psychedelics
and he keeps ignoring the parts of my email
that asks like, how do I participate in these studies?
- Yeah, well, there are some legality issues.
I mean, conversation, I won't be very clear.
I'm not saying that anyone should run out
and do psychedelics.
I think that drowsy states and sleep states
are super interesting for accessing
some of these more creative states of mind.
Hypnosis is something that my colleague, David Spiegel,
associate chair of psychiatry at Stanford, works on,
where also, again, it's a unique state
because you have narrow context.
So this is very kind of tunnel vision
and yet deeply relaxed, where new algorithms,
if you will, can start to surface.
Strong state for inducing neuroplasticity.
And I think, so if I had a, I'm part of a group
that it's called the Liminal Collective,
is a group of people that get together
and talk about just wild ideas, but they try and implement.
And it's a really interesting group.
Some people from military, from Logitech
and some other backgrounds, academic backgrounds.
And I was asked, what would be,
if you could create a tool,
if you just had a tool like your magic wand
and wish for the day, what would it be?
I thought it'd be really interesting
if someone could develop psychedelics
that have on-off switches.
So you could go into a psychedelic state
very deeply for 10 minutes,
but you could launch yourself out of that state
and place yourself into a linear real-world state
very quickly so that you could extract
whatever it was that happened in that experience
and then go back in if you wanted.
Because the problem with psychedelic states
and dream states is that, first of all,
a lot of the reason people do them is they're lying.
They say they want plasticity and they want all this stuff.
They want a peak experience
inside of an amplified experience.
So they're kind of seeking something unusual.
I think we should just be honest about that
because a lot of times
they're not trying to make their brain better.
They're just trying to experience something really amazing.
But the problem is space and time are so unlocked
in these states, just like they are in dreams,
that you can really end up with a whole lot of nothing.
You can have an amazing amplified experience
housed in an amplified experience
and come out of that thinking you had
a meaningful experience when you didn't bring anything back.
- You didn't bring anything back.
All you have is a fuzzy memory
of having a transformational experience,
but you don't actually have tools to bring back.
Or I'm just sorry, actually concrete ideas to bring back.
Yeah, it's interesting.
Yeah, I wonder if it's possible to do that with a mind
to be able to hop back and forth.
- Well, I think that's where the real power
of adjusting states is gonna be.
It probably will be with devices.
I mean, maybe it'll be done through pharmacology.
It's just that it's hard to do on/off switches
in human pharmacology that we have them for animals.
I mean, we have, you know, Cree flip recombinases
and we have, you know, channel opsins and halo rhodopsins
and all these kinds of things.
But to do that work in humans is tricky,
but I think you could do it with virtual reality,
augmented reality and other devices
that bring more of the somatic experience into it.
- You're of course a scientist who's studying humans
as a collective.
I tend to be just a one person scientist
of just looking at myself.
And, you know, I play, when these deep thinking,
deep work sessions, I'm very cognizant,
like in the morning, that there's times when my mind
is so like eloquent at being able to jump around
from ideas and hold them all together.
And I'm almost like I step back
from a third person perspective and enjoy that.
Whatever that mind is doing,
I do not waste those moments.
And I'm very conscious of this like little creature
that woke up that's only awake for,
if we're being honest, maybe a couple hours a day.
- Early part of the day for you.
- Early part of the day.
Not always, well, early part of the day for me
is a very fluid concept.
So. (laughs)
- You're one of those.
Yeah, you're one of those.
- Being single, one of the problems,
single and no meetings, I don't schedule any meetings.
I've been living at like a 28 hour day.
So I like, it drifts.
So it's all over the place.
But after a traditionally defined full night's sleep,
whatever the heck that means,
I find that like in those moments,
there's a clarity of mind that's just,
this everything is effortless
and it's the deepest dives intellectually that I make.
And I'm cognizant of it.
And I try to bring that to the other parts of the day
that don't have it and treasure them even more
in those moments 'cause they only last
like five or 10 minutes.
'Cause of course, in those moments,
you wanna do all kinds of stupid stuff
that are completely is worthless,
like check social media or something like that.
But those are the most precious things
in intellectual life is those mental moments of clarity.
And I wonder, I'm learning how to control them.
I think caffeine is somehow involved.
I'm not sure exactly.
- Sure.
Well, because if you learn how to titrate caffeine,
and everyone's slightly different with this,
what they need.
But if you learn to titrate caffeine with time of day
and the kind of work that you're trying to do,
you can bring that autonomic arousal state
into a close to perfect place.
And then you can tune it in with,
sometimes people want a little bit of background music,
sometimes they want less, these kinds of things.
The early part of the day is interesting
because the one thing that's not often discussed
is this transition out of sleep.
So there's a book,
I think it's called "Winston Churchill's Nap."
And it's about naps and the transition
between wake and sleep as a valuable period.
I've a long time ago,
someone who I respect a lot was mentoring me said,
"Be very careful about bringing in
someone else's sensory experience early in the day."
So when I wake up, I'm very drowsy.
I sleep well, but I don't emerge from that very quickly.
I need a lot of caffeine to wake up and whatnot.
But there's this concept of getting the download from sleep,
which is, in sleep,
you were essentially expunging the things
that you don't need,
the stuff that was meaningless from the previous day,
but you were also running variations on these algorithms
of whatever it is you're trying to work out in life
on short time scales like the previous day
and long time scales like your whole life.
And those lateral connections in layer five
of the neocortex are very robustly active
and across sensory areas.
And you're running an algorithm or a,
you know, it's a brain state that would be useless in waking
you wouldn't get anything done.
You'd be the person talking to yourself in the hallway
or something about something that no one else can see.
But in those states, you do,
the theory is that you arrive at certain solutions
and those solutions will reveal themselves
in the early part of the day,
unless you interfere with them by bringing in,
social media is a good example of you,
immediately enter somebody else's
space-time sensory relationship.
Someone is the conductor of your thoughts in that case.
And so many people have written about this.
What I'm saying isn't entirely new,
but allowing the download to occur
in the early part of the day and asking the question,
am I more in my head or external,
am I in more of an interoceptive or exteroceptive mode?
And depending on the kind of work you need to do,
if it's, it sounds like for you,
it's very interoceptive in the end,
very, you got a lot of thinking going on
§07Transcript · 49:50–58:00
and a lot of computing going on,
allowing yourself to transition out of that sleep state
and arrive with those solutions from sleep
and plug into the work really deeply.
And then, and only then allowing things like music,
news, social media,
doesn't mean you shouldn't talk to loved ones
and see faces and things like that.
But some people have taken this to the extreme.
When I was a graduate student at Berkeley,
there was a guy, there was a professor,
he's brilliant, odd, but brilliant,
who was so fixated on this concept
that he wouldn't look at faces in the early part of the day
because he just didn't want anything else to impact him.
Now, he didn't have the most rounded life, I suppose.
But if you're talking about cognitive performance,
this could actually be very beneficial.
You said so many brilliant things.
So one, if you read books
that describe the habits of brilliant people,
like writers, they do control that sensory experience
in the hours after wake.
Like many writers,
they have a particular habit of several hours
early in the morning of actual writing,
they do, not doing anything else for the rest of the day,
but they control,
they're very sensitive to noises and so on.
I think they make it very difficult to live with them.
I try to, I'm definitely like that.
I love to control the sensory,
how much information is coming in.
There's something about the peaceful,
just everything being peaceful.
At the same time, and we're talking,
meet your friend, Whitney Cummings,
who has a mansion, a castle on top of a cliff
in the middle of nowhere.
She actually purchased her own island.
So she wants silence.
She wants to control how much sound is coming in.
- She's very sensitive to sound and environment.
- Yeah. - Yeah.
Beautiful home and environment,
but like clearly puts a lot of attention into details.
Yeah.
And very creative.
- Yeah.
And that allows her creativity to flourish.
I'm also, I don't like, that feels like a slippery slope.
So I enjoy introducing
noises and signals and training my mind
to be able to tune them out.
'Cause I feel like you can't always control
the environment so perfectly
because your mind gets comfortable with that.
I think it's a skill that you want to learn
to be able to shut it off.
Like I often go to like, back before COVID,
to a coffee shop.
It really annoys me when there's sounds and voices
and so on, but I feel like I can train my mind
to block them out.
So it's a balance, I think.
- Yeah.
And I think, two things come to mind
as you're saying this.
First of all, yeah.
I mean, we're talking about what's best for work
is not always what's best for completeness of life.
I mean, autism is probably many things.
Like when you hear autism, just like feet,
there are probably 50 ways to get a fever.
There are probably 50 ways that the brain can create
what looks like autism or what people call autism.
But there's an interesting set of studies
that have come out of David Ginty's lab at Harvard Med
looking at, these are mouse mutants,
where these are models for autism
where nothing is disrupted in the brain proper
and in the central nervous system,
but the sensory neurons, the ones that innervate the skin
and the ears and everything are hypersensitive.
And this maps to a mutation
in certain forms of human autism.
So this means that the overload of sensory information
and sensory experience that a lot of autistics feel,
they're like, they can't tolerate things
and then they get the stereotype behaviors,
the rocking and the kind of the shouting.
It, you know, we always thought of that as a brain problem.
In some cases it might be, but in many cases,
it's because they just can't, they seem to have a,
it's like turning the volume up on every sense.
And so they're overwhelmed
and none of us want to become like that.
I think it's very hard for them
and it's hard for their parents and so forth.
So I like the coffee shop example
because the way I think about trying to build up resilience,
you know, physically or mentally or otherwise is one of,
I guess we could call it limb,
I like to call it limbic friction.
That's not a real scientific term and I acknowledge that.
I'm making it up now
because I think it captures the concept,
which is that, you know, we always hear about resilience.
It makes it sound like, oh, you know, under stress
where everything's coming at you, you're gonna stay calm.
But there's another, you know, so limbic,
the limbic system wants to pull you in some direction.
Typically in the direction of reflexive behavior.
And the prefrontal cortex through top-down mechanisms
has to suppress that and say, no,
we're not gonna respond to the banging
of the coffee cups behind me or I'm gonna keep focusing.
That's pure top-down control.
So limbic friction is high in that environment.
You've put yourself
into a high limbic friction environment.
It mean that the prefrontal cortex has to work really hard.
But there's another side to limbic friction too,
which is when you're very sleepy, there's nothing incoming.
You can be completely silent and it's hard to engage
and focus because you're drifting off
and you're getting sleepy.
So their limbic friction is high,
but for the opposite reason, autonomic arousal is too low.
So they're turning on Netflix in the background
or looping a song might boost your level of alertness
that will allow top-down control to be
in exactly the sweet spot you want it.
So this is why earlier I was saying,
it's all about how we feel inside
relative to what's going on on the outside.
We're constantly in this,
I guess one way you could envision it spatially,
especially if people are listening to this just on audio,
is I like to think about it kind of like a glass barbell
where one sphere of perception and attention
can be on what's going on with me.
And one sphere of attention can be on what's going on
with you or something else in the room or in my environment.
But this barbell isn't rigid.
It's not really glass.
Would plasma work here?
I don't know anything about plasma.
[laughing]
Sorry, I don't know.
Okay, but so imagine that this thing can contort.
The size of the globes at the end of this barbell
can get bigger or smaller.
So let's say I close my eyes and I bring all my experience
into what's going on through interoception internally.
Now it's as if I've got two orbs of perception
just on my internal state,
but I can also do the opposite
and bring both orbs of perception outside me.
I'm not thinking about my heart rate or my breathing.
I'm just thinking about something I see.
And what you'll start to realize
as you kind of use this spatial model is that two things.
One is that it's very dynamic
and that the more relaxed we are,
the more these two orbs of attention,
the two ends of the barbell can move around freely.
The more alert we are,
the more rigid they're going to be tethered in place.
And that was designed so that if I have a threat
in my environment, it's tethered to that threat.
I'm not going to, if something's coming to attack me,
I'm not going to be like,
"Oh, my breathing cadence is a little bit quick."
That's not how it works.
Why?
Because both orbs are linked to that threat.
And so my behavior is now actually being driven
by something external, even though I think it's internal.
And so I don't want to get too abstract here
because I'm a neuroscientist, I'm not a theorist.
But when you start thinking about models
of how the brain works, I mean, brain works, excuse me,
there are only really three things that neurons do.
They're either sensory neurons,
they're motor neurons, or they're modulating things.
And the models of attention and perception
that we have now, 2020, tell us that we've got interoception
and exteroception.
They're strongly modulated by levels of autonomic arousal.
And that if we want to form the optimal relationship
to some task or some pressure or something,
whether or not it's sleep, an impending threat, or coding,
we need to adjust our internal space-time relationship
with the external space-time relationship.
And I realize I'm repeating what I said earlier,
but we can actually assign circuitry to this stuff.
It mostly has to do with how much limbic friction there is,
how much you're being pulled to some source.
That source could be internal.
§08Transcript · 58:01–1:06:16
If I have pain, physical pain in my body,
I'm going to be much more interoceptive
than I am extroceptive.
You could be talking to me
and I'm just going to be thinking about that pain.
It's very hard.
And the other thing that we can link it to
is top-down control, meaning anything in our environment
that has a lot of salience will tend to bring us
into more exteroception than interoception.
And again, I don't want to litter the conversation
with just a bunch of terms,
but what I think it can be useful for people
is to do what essentially you've done, Lex,
is to start developing an awareness.
When I wake up, am I mostly in a mode of interoception
or exteroception?
When I work well, is that, what does working well look like
from the perspective of autonomic arousal?
How alert or calm am I?
What kind of balance between internal focus
and external focus is there?
And to sort of watch this process throughout the day.
- Can you linger just briefly on,
'cause you use this term a lot,
and it'd be nice to try to get a little more color to it,
which is interoception and exteroception.
What are we exactly talking about?
So like what's included in each category
and how much overlap is there?
- Interoception would be an awareness of anything
that's within the confines or on the surface of my skin
that I'm sensing.
- So literally physiological.
- Physiologically, like within the boundaries of my skin
and probably touch to the skin as well.
Exteroception would be perception of anything
that's beyond the reach of my skin.
So that bottle of water, a scent, a sound,
although, and this can change dramatically actually,
if you have headphones in,
you tend to hear things in your head,
as opposed to a speaker in the room.
This is actually the basis of ventriloquism.
So there are beautiful experiments done by Greg Reckensohn
up at UC Davis,
looking at how auditory and visual cues are matched
and we have an array of speakers,
and you can, this will become obvious as I say it,
but obviously the ventriloquist doesn't throw their voice.
What they do is they direct your vision
to a particular location,
and you think the sound is coming from that location.
And there are beautiful experiments
that Greg and his colleagues have done
where they suddenly introduce a auditory visual mismatch
and it freaks people out
because you can actually make it seem
from a perception standpoint,
as if the sound arrived from the corner of the room
and hit you, like physically, and people will recoil.
And so sounds aren't getting thrown across the room,
they're still coming from this defined location
and array of speakers,
but this is the way the brain creates
these internal representations.
And again, not to, I don't wanna go down a rabbit hole,
but I think as much as you're,
I'm sure the listeners appreciate this,
but everything in the brain is an abstraction, right?
I mean, the sensory apparati,
there are the eyes and ears and nose and skin
and taste and all that are taking information
and with interoception,
it's taking information from sensors inside the body,
the enteric nervous system for the gut,
I've got sensory neurons that innervate my liver, et cetera.
Taking all that and the brain is abstracting that
in the same way that if I took a picture of your face
and I handed it to you and I'd say, that's you,
you'd say, yeah, that's me.
But if I were an abstract artist,
I'd be doing a little bit more of what the brain does,
where if I took a pen, pad and paper,
maybe I could do this 'cause I'm a terrible artist,
and I could just mix it up and I'd,
let's say I would make your eyes like water bottles,
but I'd flip them upside down
and I'd start assigning fruits and objects
to the different features of your face.
And I showed you, I say, Lex, that's you.
Say, well, that's not me.
And I'd say, no, but that's my abstraction of you.
But that's what the brain does.
The space time relationship of the neurons that fire,
that encode your face, have no resemblance to your face.
- Right.
- And I think people don't really,
I don't know if people have fully internalized that,
but the day that I,
and I'm not sure I fully internalized that
because it's weird to think about,
but all neurons can do is fire in space and in time,
different neurons in different sequences,
perhaps with different intensities.
It's not clear the action potential is all or none,
although neuroscientists don't like to talk about that,
even though it's been published in "Nature" a couple times.
The action potential for a given neuron
doesn't always have the exact same waveform.
People, it's in all the textbooks,
but you can modify that waveform.
- Well, I mean, there's a lot of fascinating stuff
with neuroscience about the fuzziness
of all the, of the transfer of information
from neuron to neuron.
I mean, we certainly touch upon it
every time we at all try to think about the difference
between artificial neural networks
and biological neural networks,
but can we maybe linger a little bit on this,
on the circuitry that you're getting at?
So the brain is just a bunch of stuff firing
and it forms abstractions that are fascinating
and beautiful, like layers upon layers
upon layers of abstraction.
And I think it, just like when you're programming,
you know, I'm programming in Python,
it's awe-inspiring to think that underneath it all,
it ends up being zeros and ones.
And the computer doesn't know about,
no stupid Python or Windows or Linux.
It only knows about the zeros and ones.
In the same way with the brain,
is there something interesting to you
or fundamental to you about the circuitry of the brain
that allows for the magic that's in our mind to emerge?
How much do we understand?
I mean, maybe even focusing on the vision system.
Is there something specific about the structure
of the vision system, the circuitry of it
that allows for the complexity of the vision system
to emerge or is it all just a complete chaotic mess
that we don't understand?
- It's definitely not all a chaotic mess
that we don't understand if we're talking about vision.
And that's not just 'cause I'm a vision scientist.
- Let's stick to vision.
- Well, because in the beauty of the visual system,
the reason David Hubel and Torsten Wiesel
won the Nobel Prize was because they were brilliant
and forward thinking and adventurous
and all that good stuff.
But the reason that the visual system
is such a great model for addressing
these kinds of questions and other systems are hard
is we can control the stimuli.
We can adjust spatial frequency,
how fine the gratings are, thick gratings, thin gratings.
We can adjust temporal frequency,
how fast things are moving.
We can use cone isolating stimuli.
There's so many things that you can do in a controlled way.
Whereas if we were talking about cognitive encoding,
like encoding the space of concepts or something.
I, like you, if I may,
am drawn to the big questions in neuroscience.
But I confess in part because of some good advice
I got early in my career,
and in part because I'm not perhaps smart enough
to go after the really high level stuff,
I also like to address things that are tractable.
And I want, you know, we need to address
what we can stand to make some ground on at a given time.
- They construct brilliant controlled experiments
just to study, to really literally answer questions about.
- Yeah, I mean, I'm happy to have a talk
about consciousness, but it's a scary talk.
And I think most people don't wanna hear what I have to say,
which is, you know, which is,
we can save that for later perhaps.
- It's an interesting question of,
we talk about psychedelics,
we can talk about consciousness,
we can talk about cognition.
Can experiments in neuroscience be constructed
to shed any kind of light on these questions?
So I mean, it's cool that vision,
I mean, to me, vision is probably
one of the most beautiful things about human beings.
Also from the AI side,
computer vision has some of the most exciting applications
of neural networks is in computer vision.
But it feels like that's a neighbor
of cognition and consciousness.
It's just that we maybe haven't come up
with experiments to study those yet.
- Yeah, the visual system is amazing.
We're mostly visual animals to navigate, survive.
Humans mainly rely on vision, not smell or something else,
§09Transcript · 1:06:18–1:13:40
but it's a filter for cognition
and it's a strong driver of cognition.
Maybe just 'cause it came up
and then we're moving to higher level concepts.
Just the way the visual system works
can be summarized in a few relatively succinct statements,
unlike most of what I've said,
which has not been succinct at all.
- Let's go there.
- The retina. - What's involved.
- Yeah, so the retina is this three layers
of neuron structure at the back of your eyes,
about as thick as a credit card.
It is a piece of your brain.
And sometimes people think I'm kind of wriggling
out of a reality by saying that.
It's absolutely a piece of the brain.
It's a four brain structure that in the first trimester,
there's a genetic program that made sure
that that neural retina,
which is part of your central nervous system,
was squeezed out into what's called the embryonic eye cups.
And that the bone formed with a little hole
where the optic nerve is gonna connect it
to the rest of the brain.
And that window into the world
is the only window into the world for a mammal,
which has a thick skull.
Birds have a thin skull,
so their pineal gland sits and lizards too.
And snakes actually have a hole
so that light can make it down into the pineal directly
and in trained melatonin rhythms
for time of day and time of year.
Humans have to do all that through the eyes.
So three layers of neurons that are a piece of your brain,
their central nervous system.
And the optic nerve connects to the rest of the brain.
The neurons in the eye,
some just care about luminance,
just how bright or dim it is.
And they inform the brain about time of day.
And then the central circadian clock
informs every cell in your body about time of day
and make sure that all sorts of good stuff happens
if you're getting light in your eyes at the right times.
And all sorts of bad things happen
if you are getting light randomly
throughout the 24 hour cycle.
We could talk about all that,
but this is a good incentive
for keeping a relatively normal schedule,
a consistent schedule of light exposure.
Consistent schedule, try and keep a consistent schedule.
When you're young, it's easy to go off schedule and recover.
As you get older, it gets harder,
but you see everything from outcomes in cancer patients
to diabetes improves when people are getting light
at a particular time of day
and getting darkness at a particular phase
of the 24 hour cycle.
We were designed to get light and dark
at different times of the circadian cycle.
That's all being,
all that information is coming in
through specialized type of neuron in the retina
called the melanopsin,
intrinsically photosensitive ganglion cell
discovered by David Burson at Brown University.
That's not spatial information.
It's subconscious.
You don't think, oh, it's daytime.
Even if you're looking at the sun, it doesn't matter.
It's a photon counter.
It's literally counting photons.
And it's saying, oh, even though it's a cloudy day,
lots of photons coming in.
It's winter in Boston, it must be winter.
And your system is a little depressed.
It's spring, you feel alert.
That's not a coincidence.
That's these melanopsin cells signaling the circadian clock.
There are a bunch of other neurons in the eye
that signal to the brain.
And they mainly signal the presence of things
that are lighter than background or darker than background.
So a black object would be darker than background,
a light object, lighter than background.
And that all come, it's mainly, it's looking at pixels.
Mainly, they look at circles
and those neurons have receptive fields,
which not everyone will understand,
but those neurons respond best
to little circles of dark light
or little circles of bright light.
Little circles of red light versus little circles
of green light or blue light.
And so it sounds very basic.
It's like red, green, blue, and circles brighter or dimmer
than what's next to it.
But that's basically the only information
that's sent down the optic nerve.
And when we say information, we can be very precise.
I don't mean little bits of red
traveling down the optic nerve.
I mean spikes, neural action potentials in space and time,
which for you is like makes total sense.
But I think for a lot of people,
it's actually beautiful to think about
all that information in the outside world
is converted into a language that's very simple.
It's just like a few syllables, if you will.
And those syllables are being shouted down the optic nerve,
converted into a totally different language,
like Morse code, beep, beep, beep, beep, beep, beep.
Goes into the brain and then the thalamus
essentially responds in the same way that the retina does.
Except the thalamus is also waiting things.
It's saying, you know what?
That thing was moving faster than everything else,
or it's brighter than everything else.
So that signal I'm gonna get up,
I'm gonna allow up to cortex.
Or that signal is much redder than it is green.
So I'm gonna let that signal go through.
That signal is much, eh,
it's kind of more like the red next to it.
Throw that out.
The information just doesn't get up into your cortex.
And then in cortex, of course, is where perceptions happen.
And in V1, if you will, visual area one,
but also some neighboring areas,
you start getting representations
of things like oriented lines.
So there's a neuron that responds to this angle of my hand
versus vertical, right?
This is the defining work of Hubel and Wiesel's Nobel.
And it's a very systematic map of orientation,
line orientation, direction of movement, and so forth.
And that's pretty much, and color.
And that's how the visual system is organized
all the way up to the cortex.
So it's hierarchical.
You don't build, I wanna be clear,
it's hierarchical because you don't build up that line
by suddenly having a neuron that responds to lines
in some random way.
It responds to lines by taking all the dots
that are aligned in a vertical stack,
and they all converge on one neuron.
And then that neuron responds to vertical lines.
So it's not random.
There's no abstraction at that point, in fact.
In fact, if I showed you a black line,
I could be sure that if I were imaging V1,
that I would see a representation of that black line
as a vertical line somewhere in your cortex.
So at that point, it's absolutely concrete.
It's not abstract.
But then things get really mysterious.
Some of that information travels further up
into the cortex, so that,
and goes from one visual area to the next,
to the next, to the next,
so that by time you get into an area
that Nancy Kanwisher at MIT has studied
much of her career, the fusiform face area,
you start finding single neurons
that respond only to your father's face
or to Joe Rogan's face,
regardless of the orientation of his face.
I'm sure if you saw Joe, 'cause you know him well,
from across the room and you just saw his profile,
you'd be like, "Oh, that's Joe."
Walk over and say hello.
The orientation of his face isn't there.
You wouldn't even see his eyes necessarily,
but he's represented in some abstract way by a neuron
that actually would be called the Joe Rogan neuron
or black Joe neurons.
- He might have limits.
I might not recognize him if he was upside down
or something like that.
It'd be fascinating to see what the limits
of that Joe Rogan concept is.
- So Nancy's lab has done that
because early on she was challenged by people that said,
"There aren't face neurons.
"There are neurons that they only respond to space and time,
"shapes and things like that,
"moving in particular directions and orientations."
And it turns out Nancy was right.
They use these stimuli called Grebel stimuli,
which any computer programmer would appreciate,
which kind of morphs a face into something gradually
that eventually just looks like this alien thing
they call the Grebel.
And the neurons don't respond to Grebels.
In most cases, they only respond to faces
and familiar faces.
§10Transcript · 1:13:41–1:21:33
Anyway, I'm summarizing a lot of literature
and forgive me, Nancy, and for those of the Grebel people,
if there are, they're like,
"Don't come after me with pitchforks.
"Actually, you know what?
"Come after me with pitchforks.
"I think you know what I'm trying to do here."
So the point is that in the visual system,
it's very concrete up until about visual area four,
which has color pinwheels
and seems to respond to pinwheels of colors.
And so the stimuli become more and more elaborate,
but at some point you depart that concrete representation
and you start getting abstract representations
that can't be explained by simple point-to-point wiring.
And to take a leap out of the visual system
to the higher level concepts,
what we talked about in the visual system
maps to the auditory system where you're encoding,
what, frequency of tone, sweeps.
So this is gonna sound weird to do, but you know,
like a Doppler, like hearing something,
a car passing by, for instance.
But at some point you get into motifs of music
that can't be mapped to just a,
what they call a tonotopic map of frequency.
You start abstracting.
And if you start thinking about concepts of creativity
and love and memory, like what is the map of memory space?
Well, your memories are very different than mine,
but presumably there's enough structure
at the early stages of memory processing
or at the early stages of emotional processing
or at the earlier stages of creative processing
that you have the building blocks,
your zeros and ones, if you will,
but you depart from that eventually.
Now, the exception to this,
and I wanna be really clear
'cause I was just mainly talking about neocortex,
the six layered structure on the outside of the brain
that explains a lot of human abilities,
other animals have them too,
is that subcortical structures are a lot more like machines.
It's more plung and chug.
And what I'm talking about is the machinery
that controls heart rate and breathing and receptive fields,
neurons that respond to things like temperature
on the top of my left hand.
And one of the,
I came into neuroscience from a more of a perspective
initially of psychology,
but one of the reasons I forced upon myself
to learn some electrophysiology,
not a ton, but enough,
and some molecular biology and about circuitry
is that one of the most beautiful experiences
you can have in life, I'm convinced,
is to lower an electrode into the cortex
and to show a person or an animal,
you do this ethically, of course,
stimulus like an oriented line or a face.
And you can convert the recordings
coming off of that electrode into an audio signal
or an audio monitor,
and you can hear what they call hash.
It's not the hash you smoke,
it's the hash you hear.
And it sounds like,
it just sounds like noise.
And in the cortex, eventually you find a stimulus
that gets the neuron to spike and fire action potentials
that are converted into an auditory stimulus
that are very concrete, crack, crack, crack,
sounds like a bat cracking,
like home runs or outfield balls.
When you drop electrodes deeper into the thalamus
or into the hypothalamus
or into the brainstem areas that control breathing,
it's like a machine.
You never hear hash.
You drop the electrode down.
This could be like a grungy old tungsten electrode,
not high fidelity electrode,
as long as it's got a little bit of insulation on it.
You plug it into an audio monitor,
it's picking up electricity.
And if it's a visual neuron
and it's in the thalamus or the retina,
and you walk in front of that animal or person,
that neuron goes, [imitates electric noise]
and then you walk away and it stops.
And you put your hand in front of the eye again,
and it goes, [imitates electric noise]
and you could do that for two days.
And that neuron will just,
every time there's a stimulus, it fires.
So whereas before, it's a question of how much information
is getting up to cortex,
and then these abstractions happening
where you're creating these ideas.
When you go subcortical, everything is-
- There's no abstractions.
- It's two plus two equals four.
There's no abstractions.
And this is why I know we have some common friends
at Neuralink, and I love the demonstration
they did recently.
I'm a huge fan of what they're doing
and where they're headed.
And no, I don't get paid to say that.
And I have no business relationship to them.
I'm just a huge fan of the people and the mission.
But my question was to some of them,
when are you going to go subcortical?
'Cause if you want to control an animal,
you don't do it in the cortex.
The cortex is like the abstract painting
I made of your face.
Removing one piece or changing something
may or may not matter for the abstraction.
But when you are in the subcortical areas of the brain,
a stimulating electrode can evoke an entire behavior
or an entire state.
And so the brain, if we're gonna have a discussion
about the brain and how the brain works,
we need to really be clear which brain.
Because everyone loves neocortex.
It's like, oh, canonical circuits in cortex,
we can get the cortical connectome.
And sure, necessary, but not sufficient.
Not to be able to plug in patterns
of electrical stimulation and get behavior.
Eventually we'll get there.
But if you're talking subcortical circuits,
that's where the action is.
That's where you could potentially cure Parkinson's
by stimulating the subthalamic nucleus.
Because we know that it gates motor activation patterns
in very predictable ways.
So I think for those that are interested in neuroscience,
it pays to pay attention to like,
is this a circuit that abstracts the sensory information?
Or is it just one that builds up hierarchical models
in a very predictable way?
And there's a huge chasm in neuroscience right now,
because there's no conceptual leadership.
No one knows which way to go.
And this is why I think Neuralink
has captured an amazing opportunity, which was,
okay, well, while all you academic research labs
are figuring all this stuff out,
we're gonna pick a very specific goal
and make the goal the end point.
And some academic laboratories do that.
But I think that's a beautiful way
to attack this whole thing about the brain,
because it's very concrete.
Let's restore motion to the Parkinsonian patient.
Academic labs want to do that too, of course.
Let's restore speech to the stroke patient.
But there's nothing abstract about that.
That's about figuring out the solution
to a particular problem.
So anyway, those are my,
and I admit I've mixed in a lot of opinion there.
But having spent some time, like 25 years,
digging around in the brain and listening to neurons firing
and looking at them anatomically,
I think given it's 2020, we need to ask the right,
you know, the way to get better answers,
ask better questions.
And the really high level stuff is fun.
It makes for good conversation.
And it has brought enormous interest.
But I think the questions about consciousness
and dreaming and stuff, they're fascinating.
But I don't know that we're there yet.
- So you're saying there might be a chasm
in the two views of,
the power of the brain arising from the circuitry
that forms abstractions or the power of the brain
arising from the majority of the circuitry
that's just doing very brute force, dumb things
that are like, that don't have any fancy
kind of stuff going on.
That's really interesting to think about.
- And which one to go after first.
And here I'm poaching badly from someone I've never met,
but whose work I follow, which is,
and it was actually on your podcast.
I think Elon Musk said, you know, basically the brain is a,
well, you say a monkey brain with a supercomputer on top.
And I thought that's actually probably the best description
of the brain I've ever heard
because it captures a lot of important features
like limbic friction, right?
But we think of like, oh, you know,
when we're making plans, we're using the prefrontal cortex
and we're executive function and all this kind of stuff.
§11Transcript · 1:21:35–1:28:53
But think about the drug addict
who's driven to go pursue heroin or cocaine.
They make plans.
So clearly they use their frontal cortex.
It's just that it's been hijacked by the limbic system
and all the monkey brain, as you refer to.
It's really not fair to monkeys though, Elon,
because actually monkeys can make plans.
They just don't make plans as sophisticated as us.
I've spent a lot of time with monkeys,
but I've also spent a lot of time with humans.
Anyway, I'm-
- But you're putting, you're saying like,
there's a lot of value to focusing on the monkey brain
or whatever the heck you call it.
- I do, because let's say I had an ability
to place a chip anywhere I wanted in the brain today
and activate it or inhibit that area.
I'm not sure I would put that chip in neocortex,
except maybe to just kind of have some fun
and see what happens.
The reason is it's an abstraction machine.
And especially if I wanted to make a mass production tool,
a tool in mass production that I could give
to a lot of people, because it's quite possible
that your abstractions are different enough than mine
that I wouldn't know what patterns of firing to induce.
But if I want, let's say I want to increase my level
of focus and creativity, well, then I would love
to be able to, for instance,
control my level of limbic friction.
I would love to be able to wake up and go,
oh, you know what?
I have an eight o'clock appointment.
I wake up slowly.
So between seven, eight,
but I want to do a lot of linear thinking.
So you know what?
I'm going to just, I'm going to turn down the limbic friction
and, or ramp up prefrontal cortex's activation.
So there's a lot of stuff that can happen in the thalamus
with sensory gating.
For instance, you could shut down that shell
around the thalamus and allow more creative thinking
by allowing more lateral connections.
These would be some of the,
those would be the experiments I'd want to do.
So they're in the subcortical, quote unquote, monkey brain,
but you could then look at what sorts of abstract thoughts
and behaviors would arise from that, rather than,
and here I'm not pointing the finger at Neuralink at all,
but there's this obsession with neocortex.
But I, I'm going to, well, I might lose a few friends,
but I'll hopefully gain a few.
And also one of the reasons people spend so much time
in neocortex, I have a fact and an opinion.
One fact is that you can image there
and you can record there.
Right now, the two photon and one photon microscopy methods
that allow you to image deep into the brain
still don't allow you to image down really deep
unless you're jamming prisms in there and endoscopes.
And then in the endoscopes are very narrow.
So you're getting very, you know,
it's like looking at the bottom of the ocean
through a spotlight.
And so you much easier to look at the waves up on top.
Right?
So let's face it, folks,
a lot of the reasons why there's so many recordings
in layer two, three of cortex
with all this advanced microscopy
is 'cause it's very hard to image deeper.
Now the microscopes are getting better.
And thanks to amazing work,
mainly of engineers and chemists and physicists,
let's face it, they're the ones who brought
this revolution to neuroscience in the last 10 years or so.
You can image deeper.
But we don't really,
that's why you see so many reports on layer two, three.
The other thing, which is purely opinion,
and I'm not going after anybody here,
but is that as long as there's no clear right answer,
it becomes a little easier to do creative work
in a structure where no one really knows how it works.
So it's fun to probe around
because anything you see is novel.
If you're gonna work in the thalamus or the pulvinar
or the hypothalamus,
so these structures that have been known about
since the sixties and seventies
and really since the centuries ago,
you are dealing with,
you have to combat existing models.
And whereas in cortex, no one knows how the thing works.
Neocortex, six layer cortex.
And so there's a lot more room for discovery.
There's a lot more room for discovery.
And I'm not calling anyone out.
I love cortex.
We've published some papers on cortex.
It's super interesting.
But I think with the tools that are available nowadays
and where people are trying to head
of not just reading from the brain,
monitoring activity, but writing to the brain,
I think we really have to be careful
and we need to be thoughtful
about what are we trying to write?
What script are we trying to write?
Because there are many brain structures
for which we already know what scripts they write.
And I think there's tremendous value there.
I don't think it's boring.
The fact that they act like machines makes them predictable.
Those are your zeros and ones.
Let's start there.
But what's sort of happening in this field
of writing to the brain is there's this idea.
And again, I want to be clear,
I'm not pointing at Neuralink.
I'm mainly pointing at the neocortical jockeys out there
that you go and you observe patterns
and then you think replaying those patterns
is going to give rise to something interesting.
I should call out one experiment or two experiments
which were done by Susumu Tonogawa,
Nobel Prize winner from MIT.
Done important work in memory and immunology, of course,
is where he got his Nobel,
as well as Mark Mayford's lab at UC San Diego.
They did an experiment where they monitored
a bunch of neurons while an animal learned something.
Then they captured those neurons
through some molecular tricks
so they could replay the neurons.
So now there's like perfect case scenario.
It's like, okay, you monitor the neurons in your brain.
Then I say, okay, neurons one through 100
were played in the particular sequence.
So you know the space time,
you know the keys on the piano that were played
that gave rise to the song, which was the behavior.
And then you go back and you reactivate those neurons
except you reactivate them all at once,
like slamming on all the keys once on the piano,
and you get the exact same behavior.
So the space time code may be meaningless
for some structures.
Now that's freaky.
That's a scary thing because what that means
is that all the space time firing in cortex,
the space part may matter more than the time part.
So, you know, rate codes and space time codes,
we don't know.
And, you know, I'd rather have,
I'd rather deliver more answers
in this discussion questions,
but I think it's an important consideration.
- You're saying some of the magic is in the early stages
of what the closest to the raw information.
- I believe so.
I believe so.
You know the stimulus,
you know the neuron then codes that stimulus.
So you know the transformation.
When I say this for those of you
that don't think about sensory transformations,
it's like, I can show you a red circle.
And then I look at how many times the neuron fires
in response to that red circle.
And then I could show the red circle a bunch of times,
green circle, see if it changes.
And then essentially the number of times
that is the transformation.
You've converted red circle into like three action
potentials, you know, beep, beep, beep,
or whatever you want to call it, you know,
for those that think in sound space.
So that's what you've created.
You know the transformation and you march up the,
it's called the neuro axis as you go from the periphery
up into the cortex.
And we know that, and I know Lisa Feldman Barrett,
or is it Barrett Feldman?
- Barrett Feldman. - Barrett Feldman,
excuse me, Lisa, that talked a lot about this,
that, you know, birds can do sophisticated things
and whatnot as well.
But humans, there's a strong, what we call cephalization.
A lot of the processing has moved up into the cortex
and out of these subcortical areas,
but it happens nonetheless.
And so as long as you know the transformations,
you are in a perfect place to build machines
or add machines to the brain that exactly mimic
what the brain wants to do,
which is take events in the environment
and turn them into internal firing of neurons.
§12Transcript · 1:28:56–1:36:45
- So the mastery of the brain can happen
at their early level.
You know, another perspective of it is,
you saying this means that humans aren't that special.
If we look at the evolutionary time scale,
the leap to intelligence is not that special.
So like the extra layers of abstraction
isn't where most of the magic happens of intelligence,
which gives me hope that maybe if that's true,
that means the evolution of intelligence
is not that rare of an event.
- I certainly hope not.
I hope there are other forms of intelligence.
I mean, I think what humans are really good at,
and here I want to be clear that this is not a formal model,
but what humans are really good at
is taking that plasma barbell
that we were talking about earlier
and not just using it for analysis of space,
like the intermediate environment,
but also using historical information.
Like I can read a book today about the history of medicine.
I happen to be doing that lately
for some stuff I'm researching.
And I can take that information,
and if I want, I can inject it into my plans for the future.
Other animals don't seem to do that
over the same time scales that we do.
Now, it may be that the chipmunks
are all hiding little like notebooks everywhere
in the form of like little dirt castles
or something that we don't understand.
I mean, the waggle dance of the bee
is in the most famous example.
Bees come back to the hive,
they orient relative to the honeycomb and they waggle.
There's a guy down in Australia named Srinivasan
who studied this, and it's really interesting.
No one really understands it, except he understands it best.
The bee waggles in a couple of ways
relative to the orientation of the honeycomb,
and then all the other bees see that, it's visual,
and they go out and they know the exact coordinate system
to get to that source of whatever it was,
the food, and bring it back.
And he's done it where they isolate the bees,
he's changed the visual flight environment, all this stuff.
They are communicating,
and they're communicating something
about something they saw recently,
but it doesn't extend over very long periods of time.
The same way that you and I can both read a book
or you can recommend something to me,
and then we could converge on a set of ideas later.
And in fairness, 'cause she was the one that said it,
and I didn't, and I hadn't even thought of it.
When you talked to Lisa on your podcast,
she brought up something beautiful,
which is that it never really occurred to me,
and I was sort of embarrassed that it hadn't,
but it's really beautiful and brilliant,
which is that we don't just encode senses
in the form of like color and light
and sound waves and taste,
but ideas become a form of sensory mapping.
And that's where the really, really cool
and exciting stuff is,
but we just don't understand what the receptive fields are
for ideas.
What's an idea receptive field?
- And how they're communicated between humans,
because we seem to be able to encode those ideas
in some kind of way.
Yes, it's taking all the raw information
and the internal physical states,
that sensory information put into this concept blob
that we cut in the store,
and then we're able to communicate that.
- Yeah, your abstractions are different than mine.
I actually think the comment section,
on social media,
is a beautiful example of where the abstractions
are different for different people.
So much of the misunderstanding of the world
is because of these idea receptive fields.
They're not the same.
Whereas I can look at a photoreceptor neuron
or olfactory neuron or a V1 neuron,
and I am certain, I would bet my life
that yours look and respond exactly the same way
that Lisa's do and mine do.
But once you get beyond there, it gets tricky.
And so when you say something or I say something,
and somebody gets upset about it or even happy about it,
their concept of that might be quite a bit different.
They don't really know what you mean.
They only know what it means to them.
- Yeah, so from a neural link perspective,
it makes sense to optimize the control
and the augmentation of the more primitive circuitry.
So like the stuff that is closer
to the raw sensory information.
- Go deeper.
- I think we should go deeper into the brain.
And to be fair, so Matt McDougall,
who's a neurosurgeon and a neural link
and also a clinical nurse, a great guy, brilliant.
They have amazing people.
I have to give it to them.
They have been very cryptic in recent years.
Their website was just like nothing there.
They really know how to do things with style.
And they've upset a lot of people, but that's good too.
But Matt is there.
I know Matt, he actually came up through my lab at Stanford,
although he was a neurosurgery resident.
He spent time in our lab.
He actually came out on the shark dive
and did great white shark diving with my lab
to collect the VR that we use in our fear stuff.
I've talked to Matt and I think, you know,
he and other folks there are hungry
for the deeper brain structures.
The problem is that damn vasculature, all that blood supply.
It's not trivial to get through and down into the brain
without damaging the vasculature in the neocortex,
which is on the outer crust.
But once you start getting into the thalamus
and closer to some of the main arterial sources,
you really risk getting massive bleeds.
And so it's an issue that can be worked out.
It just is hard.
- Maybe it'd be nice to educate, I'm showing my ignorance.
So the smart stuff is on the surface.
So I didn't quite realize, 'cause you keep saying deep.
- Yeah, so--
- So like the early stages are deep?
- Yeah, so--
- In actually physically in the brain.
- Yeah, so the way that, you know,
of course you've got your deep brain structures
that are involved in breathing and heart rate
and kind of lizard brain stuff.
And then on top of that, this is the model of the brain
that no one really subscribes to anymore,
but anatomically it works.
And then on top in mammals, and then on top of that,
you have the limbic structures,
which gate sensory information and decide
whether or not you're gonna listen to something more,
that you're gonna look at it,
or you're gonna split your attention to both
kind of sensory allocation stuff.
And then the neocortex is on the outside.
And that is where you get a lot of this abstraction stuff.
And now not all cortical areas are doing abstraction.
Some like visual area one, auditory area one,
they're just doing concrete representations.
But as you get into the higher order stuff,
that when you start hearing names like infraparietal cortex,
and you know, when you start hearing multiple names
in the same, then you're talking about higher order areas.
But actually there's an important experiment
that drives a lot of what people wanna do
with brain machine interface.
And that's the work of Bill Newsome,
who is at Stanford and Tony Movshin,
who's at runs the Center for Neural Science at NYU.
This is a wild experiment.
And I think it might freak a few people out
if they really think about it too deeply,
but anyway, here it goes.
There's an area called MT in the cortex.
And if I showed you a bunch of dots all moving up,
and this is what Tony and Bill,
and some of the other people in that lab did way back when,
is they show a bunch of dots moving up.
Somewhere in MT, there's some neurons that respond,
they fire when the neurons move up.
And then what they did is they started varying
the coherence of that motion.
So they made it so only 50% of the dots moved up
and the rest move randomly.
And then neuron fires a little less.
And eventually it's random and that neuron stops firing
'cause it's just kind of dots moving everywhere.
It's awesome.
And there's a systematic map so that other neurons
are responding and things moving down,
and other things are responding left,
and other things are moving right.
Okay, so there's a map of direction space.
Okay, well, that's great.
You could lesion MT, animals lose the ability
to do these kind of coherence discrimination
or direction discrimination.
But the amazing experiment,
the one that just is kind of eerie,
is that they lowered a stimulating electrode into MT,
found a neuron that responds to when dots go up.
§13Transcript · 1:36:49–1:45:24
But then they silenced that neuron.
And sure enough, the animal doesn't recognize
the neurons are going up.
And then they move the dots down.
They stimulate the neuron that responds to things moving up.
And the animal responds, 'cause it can't speak,
it responds by doing a lever press,
which says the dots are moving up.
So in other words, the sensory,
the dots are moving down in reality on the computer screen.
They're stimulating the neuron
that responds to dots moving up.
And the perception of the animal
is that dots are moving up.
Which tells you that your perception of external reality
absolutely has to be a neuronal abstraction.
It is not tacked to the movement of the dots
in any absolute way.
Your perception of the outside world depends entirely
on the activation patterns of neurons in the brain.
And you can hear that and say, well, duh,
because if I stimulate the stretch reflex
and you kick or something or whatever,
the knee reflex and you kick,
of course there's a neuron that triggers that,
but it didn't have to be that way.
Because A, the animal had prior experience,
B, you're way up in this higher order cortical areas.
What this means is that,
and I generally try and avoid conversations
about this kind of thing,
but what this means is that we are constructing our reality
with this space time firing the zeros and ones.
And it doesn't have to have anything to do
with the actual reality.
And the animal or person can be absolutely convinced
that that's what's happening.
- Are you familiar with the work of Donald Hoffman?
So he's, so he makes an evolution argument
that's not important,
that we, our brains
are completely detached from reality
in the sense that he makes a radical case
that we have no idea what physical reality is.
And in fact, it's drastically different
than what we think it is.
- Oh my.
- So he goes, that's scary.
So he doesn't say like, there's just,
'cause you're kind of implying there's a gap.
There might be a gap.
We're constructing an illusion
and then maybe using communication
to maybe create a consistency
that's sufficient for our human collaboration or whatever,
or mammal, just maybe even just life forms
are constructing a consistent reality
that's maybe detached.
I mean, that's really cool
that neurons are constructing that,
like that you can prove that this is when you're a science
at its best, vision science.
But he says that like our brain is actually
just lost its shit on the path of evolution
to where we're normal.
We're just playing games with each other
in constructing realities that allow our survival.
But it's completely detached from physical reality.
- Like we're missing a lot.
- We're missing like most of it, if not all of it.
- Well, this was, it's fascinating
because I just saw the Oliver Sacks documentary.
There's a new documentary out about his life.
And there's this one part where he's like,
I've spent part of my life trying to imagine
what it would like to be a bat or something,
to see the world through the sensory apparati of a bat.
And he did this with these patients
that were locked into these horrible syndromes
that to pull out some of the beauty
of their experience as well,
not just communicate the suffering,
although the suffering too.
And as I was listening to him talk about this,
I started to realize, it's like,
well, what, you know, like there are these mantis shrimps
that can see 60 shades of pink or something.
And they see this stuff all the time
and animals that can see UV light.
Every time I learn about an animal
that can sense other things in the environment
that I can't like heat sensing,
well, I don't crave that experience
the same way Sacks talked about craving that experience,
but it does throw another penny in the jar
for what you're saying, which is that it could be that most,
if not all of what I perceive and believe
is just a neural fabrication.
And that for better, for worse,
we all agree on enough of the same neural fabrications
in the same time and place that we're able to function.
- Not only that, but we agree with the things
that are trying to eat us enough
to where they don't eat us.
Meaning like that it's not just us humans, you know.
- Oh, I see, because it's interactive.
- It's interactive.
So like, now I think it's a really nice thought experiment.
I think because Donald really frames it in a scientific,
like he makes a hard,
like as hard as our discussion has been now,
he makes a hard scientific case
that we don't know shit about reality.
I think that's a little bit hardcore,
but I think it's-
- It is hardcore.
It is hardcore.
- I think it's a good thought experiment
that kind of cleanses the palette
of the confidence we might have
about, 'cause we are operating in this abstraction space,
and the sensory spaces might be something very different.
And it's kind of interesting to think about
if you start to go into the realm of Neuralink
or start to talk about just everything
that you've been talking about with dream states
and psychedelics and stuff like that,
which part of the, which layer can we control
and play around with
and maybe look into a different slice of reality?
- You just gotta do the experiment.
The key is to just do the experiment
in the most ethical way possible.
I mean, that's the beauty of experiments.
This is why, there's wonderful theoretical neuroscience
happening now to make predictions,
but that's why experimental science is so wonderful.
You can go into the laboratory and poke around in there
and be a brain explorer and listen to and write to neurons.
And when you do that, you get answers.
You don't always get the answers you want,
but that's the beauty of it.
I think when you were saying this thing about reality
and the Donald Hoffman model,
I was thinking about children.
Like when I have an older sister, she's very sane.
But when she was a kid, she had an imaginary friend
and she would play with this imaginary friend.
And it had, there was this whole, there was a consistency.
This friend was like, it was Larry,
lived in a purple house.
Larry was a girl.
It was like all this stuff that a child,
a young child wouldn't have any issue with.
And then one day she announced that Larry had died.
And it wasn't traumatic or traumatic.
And that was it.
And she just stopped.
And I always wonder what that neurodevelopmental event was
that A, kept her out of a psychiatric ward
had she kept that imaginary friend.
But it's also, there was something kind of sad to it.
I think the way it was told to me,
'cause I'm the younger brother,
I wasn't around for that.
But my dad told me that there was a kind of a sadness
because it was this beautiful reality
that had been constructed.
And so we kind of wonder, as you're telling me this,
whether or not, as adults,
we try and create as much reality for children as we can
so that they can make predictions and feel safe.
Because the ability to make predictions
is a lot of what keeps our autonomic arousal in check.
I mean, we go to sleep every night
and we give up total control.
And that should frighten us deeply,
but unfortunately autonomic arousal yanks us down under
and we don't negotiate too much.
So you sleep sooner or later.
I don't know.
I was a little worried we'd get into discussions
about the nature of reality.
'Cause it's interesting in the laboratory,
I'm very much like, what's the experiment?
What's the analysis gonna look like?
What mutant mouse are we gonna use?
What experience are we gonna put someone through?
But I think it's wonderful that in 2020,
we can finally have discussions about this stuff
and look, kind of peek around the corner and say,
well, Neuralink and people,
others who are doing similar things
are gonna figure it out.
They're gonna, the answers will show up
and we just have to be open to interpretation.
- Do you think there could be an experiment
centered around consciousness?
I mean, you're plugged into the neuroscience community.
I think for the longest time,
§14Transcript · 1:45:26–1:53:29
the quote unquote C word was totally not,
was almost anti-scientific.
But now more and more people are talking about consciousness.
Elon is talking about consciousness.
AI folks is talking about consciousness.
It's still, nobody knows anything,
but it feels like a legitimate domain of inquiry
that's hungry for a real experiment.
- So I have fortunately three short answers to this.
The first one is a-
- Vlogs later.
- I'm not particularly succinct, I agree.
The joke I always tell is,
there are two things you never wanna say to a scientist.
One is, what do you do?
And the second one is, take as much time as you need.
And you definitely don't wanna say them
in the same sentence.
I have three short answers to it.
So there's a cynical answer,
kind of, and it's not one I enjoy giving,
which is that if you look into the '70s
back at the 1970s and 1980s,
and even into the early 2000s,
there were some very dynamic, very impressive speakers
who were very smart in the field of neuroscience
and related fields,
who thought hard about the consciousness problem
and fell in love with the problem,
but overlooked the fact that the technology wasn't there.
So I admire them for falling in love with the problem,
but they gleaned tremendous taxpayer resources,
essentially for nothing.
And these people know who they are.
Some of them are alive, some of them aren't.
I'm not referring to Francis Crick,
who was brilliant by the way,
and thought the Klaustrum was involved in consciousness,
which I think is a great idea.
It's this obscure structure that no one's really studied.
People are now starting to study it.
So I think Francis was brilliant and wonderful,
but there were books written about it.
It makes for great television stuff
and thought around the table
or after a couple of glasses of wine or whatever.
It's an important problem nonetheless.
And so I do think the consciousness,
the issue is it's not operationally defined, right?
That psychologists are much smarter
than a lot of hard scientists
in that for the following reason,
they put operational definitions.
They know that psychology,
if we're talking about motivation, for instance,
they know they need to put operational definitions on that
so that two laboratories can know
they're studying the same thing.
The problem with consciousness
is no one can agree on what that is.
And this was a problem for attention when I was coming up.
So in the early 2000s, people would argue,
what is attention?
Is it spatial attention, auditory attention?
Is it, and finally people were like, you know what?
We agree.
- Did they agree on that one?
- Sort of. - I remember.
- Sort of. - I remember sort of.
- Hearing people scream about attention.
- Right, they couldn't even agree on attention.
So I was coming up as a young graduate student.
I'm thinking like,
I'm definitely not gonna work on attention
and I'm definitely not gonna work on consciousness.
And I wanted something that I could solve or figure out.
I wanna be able to see the circuit or the neurons.
I wanna be able to hear it on the audio
and I wanna record from it.
And then I wanna do gain of function and loss of function.
Take it away, see something change, put it back,
see something change in a systematic way.
And that takes you down into the depths of some stuff
that's pretty plug and chug, you know?
But, you know, I'll borrow from something in the military
'cause I'm fortunate to do some work
with units from special operations
and they have beautiful language around things
'cause their world is not abstract.
And they talk about three meter targets,
10 meter targets and 100 meter targets.
And it's not an issue of picking the 100 meter target
'cause it's more beautiful
or because it's more interesting.
If you don't take down the three meter targets
and the 10 meter targets first, you're dead.
So that's, I think scientists could pay to, you know,
adopt a more kind of military thinking in that sense.
The other thing that is really important
is that just because somebody conceived of something
and can talk about it beautifully
and can glean a lot of resources for it,
doesn't mean that it's led anywhere.
So, but this isn't just true of the consciousness issue.
And I don't wanna sound cynical,
but I could pull up some names of molecules
that occupied hundreds of articles
in the very premier journals
that then were later discovered
to be totally moot for that process.
And biotech companies folded everyone and the lab pivots
and starts doing something different with that molecule.
And nobody talks about it
because as long as you're in the game,
we have this thing called anonymous peer review.
You can't afford to piss off anybody too much
unless you have some other funding stream.
And I've avoided battles most of my career,
but I pay attention to all of it.
And I've watched this and I don't think it's ego driven.
I think it's that people fall in love with an idea.
I don't think there's any,
there's not enough money in science
for people to sit back there
rubbing their hands together.
The beauty of what Neuralink and Elon and team,
'cause obviously he's very impressive,
but the team as a whole is really
what gives me great confidence in their mission,
is that he's already got enough money,
so it can't be about that.
He doesn't seem to need it at a level of,
I don't know him,
but he doesn't seem to need it
at a kind of an ego level or something.
I think it's driven by genuine curiosity.
And the team that he's assembled
include people that are very kind of abstract,
neuro neocortex, space-time coding people.
There are people like Matt, who's a neurosurgeon.
You can't, I mean, you know,
you can't BS neurosurgery.
Failures in neurosurgery are not tolerated.
So you have to be very good to exceptional
to even get through the gate, and he's exceptional.
And then they've got people like Dan Adams,
who was at UCSF for a long time,
is a good friend and known him for years,
who is very concrete, studied the vasculature in the eye
and how it maps to the vasculature in cortex.
When you get a team like that together,
you're gonna have dissenters,
you're gonna have people that are high-level thinkers,
people that are coders.
When you get a team like that,
it no longer looks like an academic laboratory
or even a field in science.
And so I think they're gonna solve
some really hard problems.
And again, I'm not here, they don't, you know,
I have nothing at stake with them.
But I think that's the solution.
You need a bunch of people
who don't need first author papers,
who don't need to complete their PhD,
who aren't relying on outside funding,
who have a clear mission,
and you have a bunch of people
who are basically will adapt to solve the problem.
- I like the analogy of the three-meter target
and the hundred-meter target.
So the folks at Neuralink are basically,
many of them are some of the best people in the world
at the three-meter target.
Like you mentioned Matt and neurosurgery,
like they're solving real problems.
There's no BS, philosophical,
smoke some weed and look back and look at the stars.
But,
so both on Elon and because I think like this,
I think it's really important to think about the hundred-meter
and the hundred-meter is not even a hundred-meter,
but like the stuff behind the hill that's too far away,
which is where I put consciousness.
Maybe, I tend to believe that consciousness
can be engineered.
I think part of the reason,
part of the business I wanna build leverages that idea.
That consciousness is a lot simpler
than we've been talking about.
- Well, if someone can simplify the problem.
- Right. - That will be wonderful.
I mean, the reason we can talk about something as abstract
as face representations, infusive form, face area,
is because Nancy Kanwisher had the brilliance
to tie it to the kind of lower level statistics
of visual scenes.
It wasn't 'cause she was like, "Oh, I bet it's there."
That wouldn't have been interesting.
So people like her understand how to bridge that gap
§15Transcript · 1:53:33–2:02:46
and they put a tractable definition.
So that's what I'm begging for in science,
is a tractable definition.
- This is what, but I want people to sit in the,
I want people who are really uncomfortable
with woo-woo like consciousness, like high-level stuff,
to sit in that topic and sit uncomfortably
because it forces them to then try to ground
and simplify it into something that's concrete.
Because too many people are just uncomfortable
to sit in the consciousness room
because there's no definitions.
It's like attention or intelligence
in the artificial intelligence community.
But the reality is it's easy to avoid that room altogether,
which is what, I mean, there's analogies
to everything you've said
with the artificial intelligence community,
with Minsky and even Alan Turing
that talked about intelligence a lot,
and then they drew a lot of funding and then it crashed
because they really didn't do anything with it.
And it was a lot of force of personality and so on,
but that doesn't mean the topic of the Turing test
and intelligence isn't something we should sit on
and think like, think like what is,
well, first of all, I mean,
Turing actually attempted this with the Turing test.
He tried to make concrete
this very question of intelligence.
It doesn't mean that we shouldn't linger on it
and we shouldn't forget that ultimately
that is what our efforts are all about.
In the artificial intelligence community
and in the people, whether it's neuroscience
or whatever bigger umbrella you wanna use
for understanding the mind,
the goal is not just about understanding
layer two or three of the vision.
It's to understand consciousness and intelligence
and maybe create it
or just all the possible biggest questions of our universe.
That's ultimately the dream.
- Absolutely, and I think what I really appreciate
about what you're saying is that everybody,
whether or not they're working on a kind of
a low level synapse, that's like a reflex in musculature
or something very high level abstract
can benefit from looking at those who prefer three,
everyone's going after three meter, 10 meter
and a hundred meter targets in some sense,
but to be able to tolerate the discomfort
of being in a conversation where there are real answers,
where the zeros and ones are known, zeros and ones,
are the equivalent of that in the nervous system.
And also, as you said, for the people that are very much
like, oh, I can only trust what I can see and touch,
those people need to put themselves
into the discomfort of the high level conversation
because what's missing is conversation
and conceptualization of things at multiple levels.
I think one of the, this is, I don't gripe about,
my lab's been fortunate.
We've been funded from the start and we've been happy
in that regard and lucky, and we're grateful for that.
But I think one of the challenges of research
being so expensive is that there isn't a lot of time,
especially nowadays, for people to just convene
around a topic because there's so much emphasis
on productivity.
And so there are actually, believe it or not,
there aren't that many concepts,
formal concepts in neuroscience right now.
The last 10 years has been this huge influx of tools.
And so people have been in neural circuits
and probing around and connectomes, it's been wonderful.
But 10, 20 years ago, when the consciousness stuff
was more prominent, the C word, as you said,
what was good about that time is that people
would go to meetings and actually discuss ideas and models.
Now it's sort of like demonstration day
at the school science fair where everyone's got their thing
and some stuff is cooler than others.
But I think we're gonna see a shift.
I'm grateful that we have so many computer scientists
and theoreticians or theorists,
I think they call themselves.
Somebody tell me what the difference is someday.
And psychology and even dare I say philosophy,
these things are starting to converge.
Neuroscience, the name neuroscience,
there wasn't even such a thing
when I started graduate school or as a postdoc,
it was neurophysiology or you were a neuroanatomist.
Now, it's sort of everybody's invited and that's beautiful.
That means that something's useful
is gonna come of all this.
And there's also tremendous work, of course,
happening on it for the treatment of disease.
And we shouldn't overlook that.
That's where eliminating, reducing suffering
is also a huge initiative of neuroscience.
So there's a lot of beauty in the field,
but the consciousness thing continues to be a,
it's like an exotic bird.
It's like no one really quite knows how to handle it
and it dies very easily.
- Well, yeah.
I think also from the AI perspective,
so I view the brain as less sacred.
I think from a neuroscience perspective,
you're a little bit more sensitive to BS,
like BS narratives about the brain or whatever.
I'm a little bit more comfortable
with just poetic BS about the brain,
as long as it helps engineer intelligent systems.
You know what I mean?
- Well, and I have to, I confess ignorance
when it comes to most things about coding
and I have some quantitative ability,
but I don't have strong quantitative leanings.
And so I know my limitations too.
And so I think the next generation coming up,
a lot of the students at Stanford are really interested
in quantitative models and theory and AI.
And I remember when I was coming up,
a lot of the people who were doing work ahead of me,
I kind of rolled my eyes at some of the stuff
they were doing, including some of their personalities,
although I have many great senior colleagues everywhere.
- Way of the world.
- So it's the way of the world.
So nobody knows what it's like to be a young graduate student
in 2020, except the young graduate students.
So I know there are a lot of things I don't know.
And in addition to why I do a lot of public education,
increased scientific literacy
and neuroscientific thinking, et cetera,
a big goal of mine is to try and at least pave the way
so that these really brilliant and forward thinking
younger scientists can make the biggest possible dent
and make what will eventually be all us old guys
and gals look stupid.
I mean, that's what we were all trying to do.
That's what we were trying to do.
So yeah.
- So from the highest possible topic of caution,
cautiousness to the lowest level topic of David Goggins.
Let's go.
- I don't know if it's low level, he's high performance.
- High performance, but like low, like there's no,
I don't think David has any time for philosophy.
Let's just put it this way.
- Well, I mean, I think we can tack it
to what we were just saying in a meaningful way,
which is whatever goes on in that abstraction
part of the brain, he's figured out how to dig down
in whatever the limbic friction,
he's figured out how to grab a hold of that,
scruff it and send it in the direction
that he's decided it needs to go.
And what's wild is that he's,
what we're talking about is him doing that to himself.
Right?
It's like he's scruffing himself and directing himself
in a particular direction
and sending himself down that trajectory.
And what's beautiful is that he acknowledges
that that process is not pretty, it doesn't feel good.
It's kind of horrible at every level,
but he's created this rewarding element to it.
And I think that's what's so, it's so admirable.
And it's what so many people crave,
which is regulation of the self at that level.
- And he practices, I mean, there's a ritual to it.
There's a, every single day, like no exceptions.
There's a practice aspect to the suffering
that he goes through.
- It's principled suffering.
- Principled suffering.
I mean, I just, I mean, I admire all aspects of it,
including him and his girlfriend/wife, I'm not sure.
She'll probably know this.
- I don't know.
- I'm not asking him.
- No, no, we've only, I've only communicated with her
by text about some stuff that I was asking David,
but yeah, they clearly have formed a powerful team.
- Yeah, good cop and bad cop.
- And it's a beautiful thing to see people working
in that kind of synergy.
- It's inspiring to me, same as with Elon,
that a guy like David Goggins can find love.
(laughing)
That you find a thing that works, which gives me hope
that like whatever flavor of crazy I am,
you can always find another thing that works with that.
But I've had the, so maybe let's trade Goggins stories,
you from a neuroscience perspective,
me from a self-inflicted pain perspective.
§16Transcript · 2:02:50–2:11:25
I somehow found myself in communication with David
about some challenges that I was undergoing.
One of which is we were communicating every single day,
email, phone, about the particular 30 day challenge
that I did that stretched for longer of pushups and pull-ups.
- You made a call out on social media.
- Yeah, social media.
- Actually, I think that was the point.
I knew of you before, but that's where I started tracking
some of what you were doing with these physical challenges.
- Yeah, what the hell's wrong with that guy?
- Well, no, I think I actually,
I don't often comment on people's stuff,
but I think I commented something like,
neuroplasticity loves a non-negotiable rule.
But no, I said a non-negotiable contract.
Because at the point where, yeah,
neuroplasticity really loves a non-negotiable contract
because, and I've said this before, so forgive me,
but the brain is doing analysis of duration, path,
and outcome, and that's a lot of work for the brain.
And the more that it can pass off duration, path,
and outcome to just reflex,
the more energy it can allocate to other things.
So if you decide there's no negotiation
about how many pushups, how far I'm gonna run,
how many days, how many pull-ups, et cetera,
you actually have more energy for pushups,
running, and pull-ups.
- And when you say neuroplasticity,
you mean like the brain, once the decision is made,
it'll start rewiring stuff to make sure
that we can actually make this happen.
- That's right, I mean, so much of what we do
is reflexive at the level of just core circuitry,
breathing, heart rate, all that boring stuff, digestion,
but then there's a lot of reflexive stuff
like how you drink out of a mug of coffee
that's reflexive too, but that you had to learn
at some point in your life earlier
when you were very little,
analyzing duration, path, and outcome,
and that involves a lot of top-down processing
with the prefrontal cortex.
But through plasticity mechanisms, you now do it.
So when you take on a challenge,
provided that you understand the core mechanics
of how to run pushups and pull-ups
and whatever else you decided to do,
once you set the number and the duration and all that,
then all you have to do is just go.
But people get caught in that tide pool of just,
well, do I really have to do it?
How do I not do that?
What if I get injured?
Can I sneak this?
And that's work.
And to some extent, look, not David Goggins, obviously,
nor do I claim to understand his process partially,
but maybe a little bit,
which is that it's clear that by making the decision,
there's more resources to devote
to the effort of the actual execution.
- Well, that's a really,
like what you're saying was not a lesson
that was obvious to me, and it's still not obvious.
It's something I really work at,
which is there is always an option to quit.
And I mean, that's something I really struggle with.
I mean, I've quit some things in my life.
It's like stupid stuff.
And one lesson I've learned is if you quit once,
it opens the door that,
like it's really valuable to trick your brain
into thinking that you're gonna have to die before you quit.
Like it's actually really convenient.
So actually what you're saying is very profound,
but you shouldn't intellectualize it.
Like it took me time to develop,
like psychologically in ways
that I think it would be another conversation
'cause I'm not sure how to put it into words,
but it's really tough on me
to do certain parts of that challenge.
- Well, it's a huge, you know, it's a huge output.
- The number that, see,
I thought it would be the number would be hard,
but it's not.
It's the entirety of it,
especially in the early days was just spending,
I'm kind of embarrassed to say how many hours this took.
So I didn't say publicly how many hours,
'cause I knew people would be like,
aren't you supposed to do other stuff?
Like the hell are you doing?
- Again, I don't wanna speculate too much,
but occasionally David has said this publicly
where people will be like, don't you sleep or something?
And his process used to just be
that he would just block, delete, you know, like gone.
But it's actually, it's a super interesting topic.
And because self-control and directing our actions
and the role of emotion and quitting,
these are vital to the human experience
and they're vital to performing well at anything.
And obviously at a super high level,
being able to understand this about the self is crucial.
So I have a friend who was also in the teams.
His name is Pat Dossett.
He did nine years in the SEAL teams.
And in a similar way,
there's a lore about him among team guys
because of a kind of funny challenge he gave himself,
which was, so he and I swim together,
although he swims further up front than I do
and he's very patient.
But, you know, he was on a,
he was assigned when he was in the teams
to a position that gave him a little more time
behind a desk than he wanted.
And there's not as much time out in deployments,
although he did deployments.
So he didn't know what to do at that time,
but he thought about it and he asked himself,
what does he hate the most?
And it turns out the thing that he hated doing the most
was bear crawls, you know, walking on your hands and knees.
So he decided to bear crawl for a mile for time.
So he was bear crawling a mile a day, right?
And I thought that was an interesting example that he gave
because, you know, like why pick the thing
you hate the most?
And I think it maps right back to limbic friction.
It's the thing that creates the most limbic friction.
And so if you can overcome that, then there's carry over.
And I think the notion of carry over
has been talked about psychologically
and then kind of in the self-help space,
like, oh, if you run a marathon,
it's going to help you in other areas of life.
But will it really?
Will it?
Well, I think it depends on whether or not
there's a lot of limbic friction.
'Cause if there is, what you're exercising
is not a circuit for bear crawls or a circuit for pull-ups.
What you're doing is you're exercising a circuit
for top-down control.
And that circuit was not designed to be for bear crawls
or pull-ups or coding or waking up in the middle of the
night to do something hard.
That circuit was designed to override limbic friction.
And so neural circuits were designed to generalize, right?
The stress response to an incoming threat
that's a physical threat was designed to feel the same way
and be the same response internally as the threat
to an impending exam or divorce or marriage
or whatever it is that's stressing somebody out.
And so neural circuits are not designed to be
for one particular action or purpose.
So if you can, as you did, if you can train up
top-down control under conditions
of the highest limbic friction,
that when the desire to quit is at its utmost,
either because of fatigue or hyper arousal,
being too stressed or too tired,
you're learning how to engage a circuit.
And that circuit is forever with you.
And if you don't engage it, it sits there,
but it's atrophied.
It's like a plant that doesn't get any water.
And a lot of this has been discussed in self-help
and growth mindset and all these kinds of ideas
that circle the internet and social media.
But when you start to think about how they map
to neural circuits, I think there's some utility
'cause what it means is that the limbic friction
that you'll experience in, I don't know,
maybe some future relationship to something or someone,
it's a category of neural processing
that should immediately click into place.
It's just like the limbic friction you experienced
trying to engage in the God knows how many pushups,
pull-ups and running, you know, runs you were doing.
- 25,000, who's commenting?
- So folks, if Lex does this again,
more comments, more likes.
(both laughing)
- No, well-
- This is the problem with you getting more followers
is you're gonna- - Get more, yeah.
- Actually, I should say that's the benefit.
I don't know, maybe it's not politically correct for me.
I asked, but there is this stereotype
about Russians being- - Politically correct.
- No, like being really durable.
And I started going to that Russian banya
way back before COVID,
and they could tolerate a lot of heat,
and they would sit very stoic.
§17Transcript · 2:11:26–2:20:18
No one was going, "Oh, it's hot in here."
They would just kind of ease into it.
So maybe there's something there, who knows?
- Might be something there,
but it could be also just personal.
I just have some, I found myself, everyone's different,
but I've found myself to be able to do something unpleasant
for very long periods of time.
Like I'm able to shut off the mind,
and I don't think that's been fully tested.
And I feel- - Monkey mind
or the supercomputer?
(both laughing)
- Well, it's interesting.
I mean, which mind tells you to quit exactly?
Limbic friction tells you-
- Well, limbic friction is the source of that,
but who are you talking with exactly?
- So there's a, we can put something very concrete to that.
So there's a paper published in Cell,
super top tier journal, two years ago,
looking at effort.
And this was in a visual environment
of trying to swim forward toward a target and a reward.
And it was a really cool experiment
'cause they manipulated virtually the visual environment.
So the same amount of effort was being expended every time,
but sometimes the perception
was you're making forward progress.
And sometimes the perception was you're making no progress
because stuff wasn't drifting by, meant no progress.
So you can be swimming and swimming
and not making progress.
And it turns out that with each bout of effort,
there's epinephrine and norepinephrine
is being released in the brainstem.
And glia, what traditionally were thought of
as support cells for the neurons,
but they do a lot of things actively too,
are measuring the amount of epinephrine
and norepinephrine in that circuit.
And when it exceeds a certain threshold,
the glia send inhibitory signals
that shut down top-down control.
They literally, it's the quit, you stop.
There's no more, you quit enduring.
It can be rescued, endurance can be rescued with dopamine.
So that's where the subjective part really comes into play.
So you quit because you've learned how to turn that off,
or you've learned how to,
some people will reward the pain process so much
that friction becomes the reward.
And when you talk about people like Goggins
and other people I know from special operations
and people have gone through cancer treatments three times,
you hear about, just when you hear about people,
the Viktor Frankl stories,
I mean, you hear about Nelson Mandela,
you hear about these stories,
I'm sure the same process is involved.
Again, this speaks to the generalizability
of these processes as opposed to a neural circuit
for a particular action or cognitive function.
So I think you have to learn to subjectively self-reward
in a way that replenishes you.
Goggins talks about eating souls.
It's a very dramatic example.
In his mind, apparently, that's a form of reward,
but it's not just a form of reward where you're,
it's like you're picking up a trophy or something.
It's actually, it gives you energy.
It's a reward that gives more neural energy,
and I'm defining that as more dopamine
to suppress the noradrenaline and adrenaline circuits
in the brainstem.
- So ultimately maps to that.
Yeah, he creates enemies.
He's always fighting enemies.
I never, I think I have enemies,
but they're usually just versions of me inside my head.
So I thought about, through that 30-day challenge,
I tried to come up with fake enemies.
It wasn't working.
The only enemy I came up with is David.
- Well, now you have,
you certainly have a formidable adversary in this one.
- I don't care.
David, I'm willing to die on this one.
So let's go there.
- Well, let's hope you both survive this one.
- My problem is the physical.
So everything we've been talking about in the mind,
there's a physical aspect that's just practically difficult,
which is like, I can't,
like when you injure yourself at a certain point,
like you just can't function.
- Or you're doing more damage.
- Yeah. - You're talking about
taking yourself out of running for, yeah.
For the rest of your life, potentially,
or like, you know, or for years.
So, you know, I'd love to avoid that, right?
There's just like stupid physical stuff
that you just want to avoid.
You want to keep it purely in the mental.
And if it's purely in the mental,
that's when the race is interesting.
But yeah, the problem with these physical challenges,
as David has experienced, I mean,
it has a toll on your body.
I tend to think of the mind as limitless,
and the body is kind of, unfortunately, quite limited.
- Well, I think the key is to dynamically control
your output, and that can be done by reducing effort,
which doesn't work for throughout,
but also by restoring through these
subjective reward processes.
And we don't want to go down the rabbit hole
of why this all works, but these are ancient pathways
that were designed to bring resources to an animal
or to a person through foraging,
for hunting or mates or water, all these things.
And they work so well because they're down in those
circuits where we know the zeros and ones.
And that's great because it can be subjective
at the level of, oh, I reached this one milestone,
this one horizon, this one three-meter target.
But if you don't reward it, it's just effort.
If you do self-reward it, it's effort minus one
in terms of the adrenaline output.
- I have to ask you about this.
You're one of the great communicators in science.
I'm really a big fan of yours,
and enjoying in terms of the educational stuff
you're putting on neuroscience.
- Thank you.
- What's the, do you have a philosophy behind it
or is it just an instinct?
(laughing)
- Oh, my.
- Unstoppable force?
Do you have, what's your thinking?
Because it's rare and it's exciting.
I'm excited that somebody from Stanford,
so I, okay, I'm in multiple places in the sense of
like where my interests lie.
And one, politically speaking,
academic institutions are under fire,
for many reasons we don't need to get into.
I get into it in a lot of other places,
but I believe in places like Stanford
and places like MIT as one of the most magical institutions
for inspiring people to dream, people to build the future.
I mean, it's, I believe that it is a really special,
these universities are really special places.
And so it's always exciting to me
when somebody as inspiring as you represents those places.
So it makes me proud that somebody from Stanford is,
like somebody like you is representing Stanford.
So maybe you could speak to what's,
how did you come to be who you are in being a communicator?
- Well, first of all, thanks for the kind words,
especially coming from you.
I think Stanford is an amazing place as is MIT.
And it's such a--
- MIT is better, by the way.
(laughing)
It's okay, I'll let it out, anything you say at this point.
- You know, I got many friends at MIT.
- Yeah.
- You know, hi, Ed Boyden.
- Smarter friends, yeah.
- Ed Boyden is best in class, you know,
among the best in class.
There's some people--
- Number one.
- Not me that can hold a candle to him,
but not many, maybe one or two.
I think the great benefit of being in a place like MIT
or Stanford is that when you look around,
you know, that the average is very high, right?
You have many best in class among the, you know,
one or two or three best in the world at what they do.
And it's a wonderful privilege to be there.
And one thing that I think also makes them
and other universities like them very special
is that there's an emphasis on what gets exported
out of the university.
What, you know, not keeping it ivory tower
and really trying to keep an eye on what's needed
in the world and trying to do something useful.
And I think the proximity to industry in Silicon Valley
and in the Boston area and Cambridge
also lends itself well to that.
And there are other institutions too, of course.
- So the reason I got involved in educating on social media
was actually because of Pat Dossett,
the Bear, Myle Bearcaw guy.
I was at the turn of 2018 to 2019.
We had formed a good friendship and we were,
§18Transcript · 2:20:21–2:28:51
he talked me into doing these early morning cold water swims.
I was learning a lot about pain and suffering,
but also the beauty of cold water swims.
And we were talking one morning and he said,
"So what are you going to do to serve the world in 2019?"
It's like, that's the way that like a Texan
former SEAL talks.
Like we're just literally,
what are you gonna do to serve the world in 2019?
Like, well, I run my lab.
It's like, no, no, what are you gonna do that's new?
And he wasn't forceful in it,
but I was like, that's interesting question.
I said, well, if I had my way,
I would just teach people, everyone about the brain
'cause I think it's amazing.
He goes, "We'll do it."
I go, "All right."
He goes, "Shake on it."
So we did it, you know?
And so I started putting out these posts
and it's grown into to include a variety of things,
but you asked about a governing philosophy.
So I want to increase interest in the brain
and in the nervous system and in biology generally.
That's one major goal.
I'd like to increase scientific literacy,
which can't be rammed down people's throats
of talking about how to look at a graph and statistics
and Z-scores and P-values and genetics.
It has to be done gradually in my opinion.
I want to put valuable tools into the world,
mainly tools that map to things that we're doing in our lab.
So these will be tools centered around
how to understand a direct one states of mind and body.
So reduce stress, raise one stress threshold.
So it's not always just about being calm.
Sometimes it's about learning how to tolerate
not being not calm.
Raise awareness for mental health.
There's a ton of micro missions in this,
but it all really maps back to, you know,
like the eight and 10 year old version of me,
which is I used to spend my weekends
when I was a kid reading about weird animals.
And I had this obsession with like medieval weapons
and stuff like catapults.
And then I used to come into school on Monday
and I would ask if I could talk about it
to the class and teach.
And I just, it's really, I promise,
and some people might not believe me,
but it's really, I don't really like
being the point of focus.
I just get so excited about these gems
that I find in the world in books and in experiments
and in discussions with colleagues
and discussions with people like you
and around the universe.
And I can't just compulsively,
I got to tell people about it.
So I try and package it into a form that people can access.
You know, I think if I've,
I think the reception has been really wonderful.
Stanford has been very supportive, thankfully.
I've given, done some podcasts even with them
and they've reposted some stuff on social media.
It's a precarious place to put yourself out there
as a research academic.
I think some of my colleagues,
both locally and elsewhere probably wonder
if I'm still serious about research, which I absolutely am.
And I also acknowledge that, you know,
their research and the research coming out of the field
needs to be talked about.
And not all scientists are good at translating that
into a language that people can access.
And I don't like the phrase, dumb it down.
What I like to do is take a concept
that I think people will find interesting and useful
and offer it sort of like a,
you would offer food to somebody visiting your home.
You're not going to cram foie gras in their face.
You're going to say, like, do you want a cracker?
Like, and they say, yeah.
And like, do you want something on that cracker?
Like, do you like cheese?
Like, yeah.
Like, do you want Swiss cheese
or you want that really like stinky, like French,
I don't like cheese much, but,
or do you want foie gras?
Like, what's that?
Like, so you're trying,
the best information prompts more questions of interest,
not questions of confusion, but questions of interest.
And so I feel like one door opens,
then another door opens, then another door opens.
And pretty soon the image in my mind
is you create a bunch of neuroscientists
who are thinking about themselves neuroscientifically.
And I don't begin to think
that I have all the answers at all.
I cast a neuroscience,
sometimes a little bit of a psychology lens
onto what I think are interesting topics.
And, you know, I, you know,
someday I'm going to go into the ground
or the ocean or wherever it is I end up.
And I'm very comfortable with the fact
that not everyone's going to be happy
with how I deliver the information,
but I would hope that people would feel
like some of it was useful and meaningful
and got them to think a little bit harder.
- Since you mentioned going into the ground
and Viktor Frankl, "Man's Search for Meaning,"
I reread that book quite often.
What, let me ask the big ridiculous question about life.
What do you think is the meaning of it all?
Like, and maybe why do you,
do you mention that book from a psychologist perspective,
which Viktor Frankl was,
or do you ever think about the bigger philosophical questions
that he raises about meaning?
What's the meaning of it all?
- One of the great challenges in assigning a good,
you know, giving a good answer to the question of like,
what's the meaning of life is I think illustrated best
by the Viktor Frankl example,
although there are other examples too,
which is that our sense of meaning is very elastic
in time and space.
And I'm, we talked a little bit about this earlier,
but it's amazing to me that somebody locked in a cell
or a concentration camp can bring the horizon
in close enough that they can then micro slice
their environment so that they can find rewards
and meaning and power and beauty,
even in a little square box or a horrible situation.
And I think this is really speaks
to one of the most important features of the human mind,
which is we could do, let's take two opposite extremes.
One would be, let's say the alarm went off right now
in this building and the building started shaking.
Our vision, our hearing, everything would be tuned
to this space time bubble for those moments.
And everything that we were processed,
all that would matter, the only meaning would be
get out of here safe, figure out what's going on,
contact loved ones, et cetera.
If we were to sit back, totally relaxed,
we could do the, you know, what is it?
I think it's called pale blue dot thing or whatever,
where we could imagine ourselves in this room.
And then they were in the United States
and this continent and the earth,
and then it's peering down on us.
And all of a sudden you get back,
it can seem so big that all of a sudden
it's meaningless, right?
If you see yourself as just one brief glimmer
in all of time and all of space, you go to, I don't matter.
And if you go to, oh, every little thing that happens
in this text thread or this, you know,
comment section on YouTube or Instagram,
your space time bubble is tiny.
Then everything seems inflated and the brain
will contract and dilate at space time,
vision and time, but also sense of meaning.
And that's beautiful.
And it's what allows us to be so dynamic
in different environments.
And we can pull from the past and the present and future.
It's why examples like Nelson Mandela
and Viktor Frankl had to include.
It makes sense that it wasn't just about grinding it out.
They had to find those dopamine rewards,
even in those little boxes they were forced into.
So I'm not trying to dodge an answer,
but for me personally, and I think about this a lot
because I have this complicated history in science
where my undergraduate graduate advisor
and postdoctoral advisor all died young.
So, you know, and they were wonderful people
and had immense importance in my life.
But what I realized is that we can get so fixated
on the thing that we're experiencing
how holding tremendous meaning,
but it only holds that meaning for as long
as we're in that space time regime.
And this is important because what really gives meaning
is the understanding that you can move
between these different space time dimensionalities.
And I'm not trying to sound like a theoretical physicist
or anyone that thinks about the cosmos in saying that.
It's really the fact that sometimes we say
and do and think things and it feels so important.
And then two days later, we're like, what happened?
Well, you had a different brain processing algorithm
§19Transcript · 2:28:54–2:32:08
entirely, you were in a completely different state.
And so what I want to do in this lifetime is I want to,
I want to engage in as many different levels
of contraction and dilation of meaning as possible.
I want to go to the micro.
I sometimes think about this.
I'm like, if I just pulled over the side of the road,
I bet you there's an ant hill there
and their whole world is fascinating.
You can't stay there.
And you also can't stay staring up at the clouds
and just think about how we're just these little beings
and it doesn't matter.
The key is the journey back and forth,
up and down that staircase, back and forth
and back and forth.
And my goal is to get as many trips up and down
that staircase as I can before the reaper comes for me.
- Oh, beautiful.
So the dance of dilation and contraction
between the different spaces, zoom in, zoom out
and get as many steps in on that staircase.
- That's my goal anyway.
And I've watched people die.
I watched my postdoc advisor die, wither away.
My graduate advisor, it was tragic,
but they found beauty in these closing moments
because their bubble was their kids in one case
or like one of them was a Giants fan
and like got to see a Giants game, you know,
in her last moments.
And like, and you just realize like it's a Giants game
but not in that moment because time is closing.
And so those time bins feel huge
because she's slicing things so differently.
So I think learning how to do that better
and more fluidly, recognizing where one is
and not getting too tacked to the idea
that there's one correct answer.
Like that's what brings meaning.
That's my goal anyway.
- I don't think there's a better way to end it, Andrew.
I really appreciate that you would come down
and contract your space time
and focus on this conversation for a few hours.
It is a huge honor.
I'm a huge fan of yours, as I told you.
I hope you keep growing and educating the world
about the human mind.
Thanks for talking today.
- Thank you.
I really appreciate the invitation to be here.
And people might think that I'm saying it
just 'cause I'm here, but I'm a huge fan of yours.
I send your podcasts to my colleagues and other people.
And I think what you're doing isn't just amazing,
it's important.
And so thank you.
- Thanks for listening to this conversation
with Andrew Huberman.
And thank you to our sponsors.
Asleep, a mattress that cools itself
and gives me yet another reason to enjoy sleep.
SEMrush, the most advanced SEO optimization tool
I've ever come across.
And Cash App, the app I use to send money to friends.
Please check out the sponsors in the description
to get a discount and to support this podcast.
If you enjoy this thing, subscribe on YouTube,
review it with five stars on Apple Podcasts,
follow on Spotify, support on Patreon,
or connect with me on Twitter @LexFriedman.
And now let me leave you with some words from Carl Jung.
I am not what happened to me.
I am what I choose to become.
Thank you for listening and hope to see you next time.
(upbeat music)
(upbeat music)