Donald Hoffman: Reality is an Illusion – How Evolution Hid the Truth
生物与进化哲学与宗教太空与探索音乐与艺术AI 与机器学习
📋 章节目录
暂无章节信息
🔑 关键词
spaceconsciousnesstheoryconsciousfundamentalrealitydonsayinginterfacedeeperinterestingstructuredatagoingexperienceswholeevolutionprobabilitydynamicssaid
💬 精彩语录
暂无语录
🎙️ 完整对话（4234 条）
Lex Fridman (00:00.000)
Whatever reality is, it's not what you see.
无论现实如何，它都不是你所看到的。
Lex Fridman (00:04.880)
What you see is just an adaptive fiction.
你所看到的只是一个改编小说。
Lex Fridman (00:12.560)
The following is a conversation with Donald Hoffman,
以下是与唐纳德·霍夫曼的对话，
Lex Fridman (00:14.920)
professor of cognitive sciences at UC Irvine,
加州大学欧文分校认知科学教授，
Lex Fridman (00:17.920)
focusing his research on evolutionary psychology,
他的研究重点是进化心理学，
Donald Hoffman (00:21.200)
visual perception, and consciousness.
视觉感知和意识。
Lex Fridman (00:23.920)
He's the author of over 120 scientific papers
他是 120 多篇科学论文的作者
Donald Hoffman (00:27.760)
on these topics and his most recent book
关于这些主题和他最近的书
Lex Fridman (00:30.320)
titled The Case Against Reality,
标题为《反现实的案例》，
Lex Fridman (00:33.280)
Why Evolution Hid the Truth from Our Eyes.
为什么进化论向我们隐藏了真相。
Lex Fridman (00:36.700)
I think some of the most interesting ideas in this world,
我认为这个世界上一些最有趣的想法，
Donald Hoffman (00:39.600)
like those of Donald Hoffman's,
就像唐纳德·霍夫曼的作品一样，
Lex Fridman (00:41.400)
attempt to shake the foundation
试图动摇根基
Donald Hoffman (00:43.640)
of our understanding of reality,
我们对现实的理解，
Lex Fridman (00:45.920)
and thus they take a long time to internalize deeply.
因此他们需要很长时间才能深入内化。
Lex Fridman (00:50.320)
So proceed with caution.
因此请谨慎行事。
Lex Fridman (00:52.240)
Questioning the fabric of reality
质疑现实的结构
Donald Hoffman (00:54.440)
can lead you to either madness or to truth.
可以引导你走向疯狂或真理。
Lex Fridman (00:58.520)
And the funny thing is, you won't know which is which.
有趣的是，你不会知道哪个是哪个。
Donald Hoffman (01:02.440)
This is the Lex Friedman podcast.
这是莱克斯·弗里德曼的播客。
Lex Fridman (01:04.260)
To support it, please check out our sponsors
Donald Hoffman (01:06.420)
in the description.
Lex Fridman (01:07.640)
And now, dear friends, here's Donald Hoffman.
Donald Hoffman (01:12.040)
In your book, The Case Against Reality,
Lex Fridman (01:14.300)
Why Evolution Hid the Truth from Our Eyes,
Donald Hoffman (01:17.200)
you make the bold claim that the world we see
Lex Fridman (01:20.000)
with our eyes is not real.
Donald Hoffman (01:21.960)
It's not even an abstraction of objective reality.
Lex Fridman (01:24.560)
It is completely detached from objective reality.
Lex Fridman (01:29.680)
Can you explain this idea?
Lex Fridman (01:30.880)
Right, so this is a theorem
Donald Hoffman (01:32.400)
from evolution by natural selection.
Lex Fridman (01:34.160)
So the technical question that I and my team asked was,
Lex Fridman (01:38.540)
what is the probability that natural selection
Lex Fridman (01:41.380)
would shape sensory systems
Lex Fridman (01:43.000)
to see true properties of objective reality?
Lex Fridman (01:46.240)
And to our surprise,
Donald Hoffman (01:47.440)
we found that the answer is precisely zero,
Lex Fridman (01:49.320)
except for one kind of structure
Donald Hoffman (01:51.400)
that we can go into if you want to.
Lex Fridman (01:52.480)
But for any generic structure
Donald Hoffman (01:54.640)
that you might think the world might have,
Lex Fridman (01:56.240)
a total order, a topology, metric,
Donald Hoffman (02:00.280)
the probability is precisely zero
Lex Fridman (02:02.300)
that natural selection would shape any sensory system
Donald Hoffman (02:05.300)
of any organism to see any aspect of objective reality.
Lex Fridman (02:08.560)
So in that sense, what we're seeing
Donald Hoffman (02:11.680)
is what we need to see
Lex Fridman (02:16.600)
to stay alive long enough to reproduce.
Lex Fridman (02:18.960)
So in other words, we're seeing what we need
Lex Fridman (02:20.640)
to guide adaptive behavior, full stop.
Lex Fridman (02:23.480)
So the evolutionary process,
Lex Fridman (02:26.080)
the process that took us from the origin of life on Earth
Donald Hoffman (02:30.100)
to the humans that we are today,
Lex Fridman (02:33.260)
that process does not maximize for truth,
Donald Hoffman (02:37.320)
it maximizes for fitness, as you say, fitness beats truth.
Lex Fridman (02:41.280)
And fitness does not have to be connected to truth,
Donald Hoffman (02:45.000)
is the claim.
Lex Fridman (02:46.680)
And that's where you have an approach
Donald Hoffman (02:49.000)
towards zero of probability
Lex Fridman (02:51.320)
that we have evolved human cognition,
Donald Hoffman (02:55.940)
human consciousness, whatever it is,
Lex Fridman (02:58.340)
the magic that makes our mind work,
Donald Hoffman (03:00.640)
evolved not for its ability to see the truth of reality,
Lex Fridman (03:06.240)
but its ability to survive in the environment.
Donald Hoffman (03:09.160)
That's exactly right.
Lex Fridman (03:10.240)
So most of us intuitively think that surely
Donald Hoffman (03:14.200)
the way that evolution will make our senses more fit
Lex Fridman (03:18.600)
is to make them tell us more truths,
Donald Hoffman (03:21.120)
or at least the truths we need to know
Lex Fridman (03:22.960)
about objective reality, the truths we need in our niche.
Donald Hoffman (03:26.100)
That's the standard view, and it was the view I took.
Lex Fridman (03:27.960)
I mean, that's sort of what we're taught
Donald Hoffman (03:30.520)
or just even assume.
Lex Fridman (03:31.840)
It was just sort of like the intelligent assumption
Donald Hoffman (03:33.460)
that we would all make.
Lex Fridman (03:34.720)
But we don't have to just wave our hands.
Donald Hoffman (03:37.860)
Evolution of a natural selection
Lex Fridman (03:38.940)
is a mathematically precise theory.
Donald Hoffman (03:41.220)
John Maynard Smith in the 70s
Lex Fridman (03:44.200)
created evolutionary game theory.
Lex Fridman (03:45.600)
And we have evolutionary graph theory
Lex Fridman (03:48.140)
and even genetic algorithms that we can use to study this.
Lex Fridman (03:50.520)
And so we don't have to wave our hands.
Lex Fridman (03:52.360)
It's a matter of theorem and proof and or simulation
Donald Hoffman (03:55.480)
before you get the theorems and proofs.
Lex Fridman (03:56.740)
And a couple of graduate students of mine,
Donald Hoffman (03:59.440)
Chester Mark and Brian Marion,
Lex Fridman (04:01.180)
did some wonderful simulations that tipped me off
Donald Hoffman (04:03.960)
that there was something going on here.
Lex Fridman (04:06.480)
And then I went to a mathematician, Chetan Prakash,
Lex Fridman (04:08.960)
and Manish Singh, and some other friends of mine,
Lex Fridman (04:13.040)
Chris Fields.
Lex Fridman (04:14.440)
But Chetan was the real mathematician behind all this.
Lex Fridman (04:17.000)
And he's proved several theorems
Donald Hoffman (04:18.560)
that uniformly indicate that with one exception,
Lex Fridman (04:21.800)
which has to do with probability measures,
Donald Hoffman (04:25.160)
there's no, the probability is zero.
Lex Fridman (04:28.240)
The reason there's an exception for probability measures,
Lex Fridman (04:30.920)
so called sigma algebras or sigma additive classes,
Lex Fridman (04:36.680)
is that for any scientific theory,
Donald Hoffman (04:40.680)
there is the assumption that needs to be made
Lex Fridman (04:43.400)
that whatever structure,
Donald Hoffman (04:48.640)
whatever probabilistic structure the world may have
Lex Fridman (04:51.760)
is not unrelated to the probabilistic structure
Donald Hoffman (04:55.520)
of our perceptions.
Lex Fridman (04:56.360)
If they were completely unrelated,
Donald Hoffman (04:57.360)
then no science would be possible.
Lex Fridman (04:59.560)
So this is technically the map from reality to our senses
Donald Hoffman (05:05.080)
has to be a so called measurable map,
Lex Fridman (05:07.000)
has to preserve sigma algebras.
Lex Fridman (05:08.640)
But that means it could be infinite to one,
Lex Fridman (05:10.360)
and it could collapse all sorts of event information.
Lex Fridman (05:14.320)
But other than that, there's no requirement
Lex Fridman (05:17.040)
in standard evolutionary theory
Donald Hoffman (05:18.960)
for fitness payoff functions, for example,
Lex Fridman (05:22.460)
to preserve any specific structures of objective reality.
Lex Fridman (05:25.560)
So you can ask the technical question.
Lex Fridman (05:27.120)
This is one of the avenues we took.
Donald Hoffman (05:30.920)
If you look at all the fitness payoffs
Lex Fridman (05:32.920)
from whatever world structure you might want to imagine.
Lex Fridman (05:37.960)
So a world with say a total order on it.
Lex Fridman (05:41.360)
So it's got end states and they're totally ordered.
Lex Fridman (05:44.100)
And then you can have a set of maps from that world
Lex Fridman (05:48.160)
into a set of payoffs, say from zero to a thousand
Donald Hoffman (05:50.560)
or whatever you want your payoffs to be.
Lex Fridman (05:52.560)
And you can just literally count all the payoff functions
Lex Fridman (05:56.240)
and just do the combinatorics and count them.
Lex Fridman (05:58.120)
And then you can ask the precise question,
Lex Fridman (05:59.840)
how many of those payoff functions preserve the total order?
Lex Fridman (06:04.000)
If that's what you're looking for,
Lex Fridman (06:04.840)
or how many preserve the topology?
Lex Fridman (06:07.320)
And you just count them and divide.
Lex Fridman (06:08.800)
So the number that are homomorphisms
Lex Fridman (06:11.880)
versus the total number, and then take the limit
Donald Hoffman (06:14.160)
as the number of states in the world
Lex Fridman (06:16.660)
and the number of payoff values goes very large.
Lex Fridman (06:19.960)
And when you do that, you get zero every time.
Lex Fridman (06:21.600)
Okay, there's a million things to ask here.
Lex Fridman (06:24.400)
But first of all, just in case people
Lex Fridman (06:28.240)
are not familiar with your work,
Donald Hoffman (06:30.880)
let's sort of linger on the big bold statement here,
Lex Fridman (06:35.880)
which is the thing we see with our eyes
Donald Hoffman (06:41.040)
is not some kind of limited window into reality.
Lex Fridman (06:45.040)
It is completely detached from reality,
Donald Hoffman (06:47.860)
likely completely detached from reality.
Lex Fridman (06:49.600)
You're saying 100% likely.
Donald Hoffman (06:52.580)
Okay, so none of this is real in the way we think is real.
Lex Fridman (06:57.780)
In the way we have this intuition,
Donald Hoffman (07:00.000)
there's like this table is some kind of abstraction,
Lex Fridman (07:05.000)
but underneath it all, there's atoms.
Lex Fridman (07:07.880)
And there's an entire century of physics
Lex Fridman (07:09.880)
that describes the functioning of those atoms
Lex Fridman (07:12.040)
and the quirks that make them up.
Lex Fridman (07:13.720)
There's many Nobel Prizes about particles and fields
Lex Fridman (07:19.840)
and all that kind of stuff that slowly builds up
Lex Fridman (07:23.320)
to something that's perceivable to us,
Donald Hoffman (07:25.200)
both with our eyes, with our different senses as this table.
Lex Fridman (07:29.960)
Then there's also ideas of chemistry
Donald Hoffman (07:33.560)
that over layers of abstraction, from DNA to embryos,
Lex Fridman (07:38.600)
the cells that make the human body.
Lex Fridman (07:42.960)
So all of that is not real.
Lex Fridman (07:46.660)
It's a real experience,
Lex Fridman (07:48.320)
and it's a real adaptive set of perceptions.
Lex Fridman (07:52.560)
So it's an adaptive set of perceptions, full stop.
Donald Hoffman (07:56.200)
We want to think that the perceptions are real.
Lex Fridman (07:58.640)
So their perceptions are real as perceptions, right?
Donald Hoffman (08:01.720)
We are having our perceptions,
Lex Fridman (08:03.640)
but we've assumed that there's a pretty tight relationship
Donald Hoffman (08:06.920)
between our perceptions and reality.
Lex Fridman (08:09.000)
If I look up and see the moon,
Donald Hoffman (08:11.720)
then there is something that exists in space and time
Lex Fridman (08:15.160)
that matches what I perceive.
Lex Fridman (08:18.960)
And all I'm saying is that if you take evolution
Lex Fridman (08:24.600)
by natural selection seriously, then that is precluded.
Donald Hoffman (08:29.600)
That our perceptions are there.
Lex Fridman (08:31.880)
They're there to guide adaptive behavior, full stop.
Donald Hoffman (08:35.160)
They're not there to show you the truth.
Lex Fridman (08:36.720)
In fact, the way I think about it is
Donald Hoffman (08:38.760)
they're there to hide the truth
Lex Fridman (08:40.660)
because the truth is too complicated.
Donald Hoffman (08:42.520)
It's just like if you're trying to use your laptop
Lex Fridman (08:45.720)
to write an email, right?
Lex Fridman (08:47.480)
What you're doing is toggling voltages in the computer,
Lex Fridman (08:50.280)
but good luck trying to do it that way.
Donald Hoffman (08:52.600)
The reason why we have a user interface
Lex Fridman (08:54.280)
is because we don't want to know that quote unquote truth,
Donald Hoffman (08:56.600)
the diodes and resistors and all that terrible hardware.
Lex Fridman (08:59.600)
If you had to know all that truth,
Donald Hoffman (09:02.040)
your friends wouldn't hear from you.
Lex Fridman (09:04.000)
So what evolution gave us was perceptions
Donald Hoffman (09:08.360)
that guide adaptive behavior.
Lex Fridman (09:10.200)
And part of that process, it turns out,
Donald Hoffman (09:12.000)
means hiding the truth and giving you eye candy.
Lex Fridman (09:16.680)
So what's the difference between hiding the truth
Lex Fridman (09:20.700)
and forming abstractions,
Lex Fridman (09:22.840)
layers upon layers of abstractions
Donald Hoffman (09:26.560)
over low level voltages and transistors
Lex Fridman (09:30.320)
and chips and programming languages
Donald Hoffman (09:35.480)
from assembly to Python that then leads you
Lex Fridman (09:38.560)
to be able to have an interface like Chrome
Donald Hoffman (09:41.040)
where you open up another set of JavaScript and HTML
Lex Fridman (09:45.780)
programming languages that lead you
Donald Hoffman (09:47.360)
to have a graphical user interface
Lex Fridman (09:49.360)
and which you can then send your friends an email.
Donald Hoffman (09:53.040)
Is that completely detached from the zeros and ones
Lex Fridman (09:58.480)
that are firing away inside the computer?
Donald Hoffman (10:01.560)
It's not.
Lex Fridman (10:02.880)
Of course, when I talk about the user interface
Donald Hoffman (10:04.800)
on your desktop, there's this whole sophisticated
Lex Fridman (10:10.120)
backstory to it, right?
Donald Hoffman (10:11.540)
That the hardware and the software
Lex Fridman (10:13.120)
that's allowing that to happen.
Lex Fridman (10:15.040)
Evolution doesn't tell us the backstory, right?
Lex Fridman (10:17.200)
So the theory of evolution is not going to be adequate
Donald Hoffman (10:20.400)
to tell you what is that backstory.
Lex Fridman (10:23.040)
It's gonna say that whatever reality is,
Lex Fridman (10:27.160)
and that's the interesting thing,
Lex Fridman (10:28.000)
it says whatever reality is, you don't see it.
Donald Hoffman (10:31.260)
You see a user interface,
Lex Fridman (10:32.600)
but it doesn't tell you what that user interface is,
Lex Fridman (10:36.900)
how it's built, right?
Lex Fridman (10:38.840)
Now, we can try to look at certain aspects
Donald Hoffman (10:42.600)
of the interface, but already we're gonna look at that
Lex Fridman (10:45.400)
and go, okay, before I would look at neurons
Lex Fridman (10:47.960)
and I was assuming that I was seeing something
Lex Fridman (10:49.640)
that was at least partially true.
Lex Fridman (10:52.840)
And now I'm realizing that it could be like looking
Lex Fridman (10:54.880)
at the pixels on my desktop or icons on my desktop
Lex Fridman (10:59.240)
and good luck going from that to the data structures
Lex Fridman (11:02.600)
and then the voltages and I mean, good luck.
Donald Hoffman (11:04.840)
There's just no way.
Lex Fridman (11:06.960)
So what's interesting about this is that
Donald Hoffman (11:08.600)
our scientific theories are precise enough
Lex Fridman (11:13.000)
and rigorous enough to tell us certain limits,
Donald Hoffman (11:17.160)
but, and even limits of the theories themselves,
Lex Fridman (11:20.040)
but they're not going to tell us what the next move is
Lex Fridman (11:23.240)
and that's where scientific creativity comes in.
Lex Fridman (11:25.880)
So the stuff that I'm saying here, for example,
Donald Hoffman (11:28.960)
is not alien to physicists.
Lex Fridman (11:31.040)
The physicists are saying precisely the same thing
Donald Hoffman (11:33.680)
that space time is doomed.
Lex Fridman (11:35.240)
We've assumed that space time is fundamental.
Donald Hoffman (11:37.240)
We've assumed that for several centuries
Lex Fridman (11:39.360)
and it's been very useful.
Lex Fridman (11:40.760)
So all the things that you were mentioning,
Lex Fridman (11:41.980)
the particles and all the work that's been done,
Donald Hoffman (11:43.920)
that's all been done in space time,
Lex Fridman (11:45.120)
but now physicists are saying space time is doomed.
Donald Hoffman (11:47.520)
There's no such thing as space time fundamentally
Lex Fridman (11:51.720)
in the laws of physics.
Lex Fridman (11:54.080)
And that comes actually out of gravity
Lex Fridman (11:58.580)
together with quantum field theory,
Donald Hoffman (11:59.920)
which just comes right out of it.
Lex Fridman (12:01.080)
It's a theorem of those two theories put together,
Lex Fridman (12:05.640)
but it doesn't tell you what's behind it.
Lex Fridman (12:08.000)
So the physicists know that their best theories,
Donald Hoffman (12:11.880)
Einstein's gravity and quantum field theory put together
Lex Fridman (12:15.160)
entail that space time cannot be fundamental
Lex Fridman (12:17.440)
and therefore particles in space time cannot be fundamental.
Lex Fridman (12:20.880)
They're just irreducible representations
Donald Hoffman (12:22.520)
of the symmetries of space time.
Lex Fridman (12:23.760)
That's what they are.
Lex Fridman (12:24.760)
So we have, so space time, so we put the two together.
Lex Fridman (12:27.600)
We put together what the physicists are discovering
Lex Fridman (12:29.700)
and we can talk about how they do that.
Lex Fridman (12:32.200)
And then we, the new discoveries
Donald Hoffman (12:33.780)
from evolution of a natural selection.
Lex Fridman (12:35.360)
Both of these discoveries are really in the last 20 years.
Lex Fridman (12:38.740)
And what both are saying is space time
Lex Fridman (12:41.520)
has had a good ride.
Donald Hoffman (12:43.520)
It's been very useful.
Lex Fridman (12:44.360)
Reductionism has been useful, but it's over.
Lex Fridman (12:46.520)
And it's time for us to go beyond.
Lex Fridman (12:48.400)
When you say space time is doomed,
Donald Hoffman (12:50.520)
is it the space, is it the time,
Lex Fridman (12:53.200)
is it the very hard coded specification of four dimensions?
Donald Hoffman (12:59.800)
Or are you specifically referring
Lex Fridman (13:01.560)
to the kind of perceptual domain
Lex Fridman (13:05.160)
that humans operate in, which is space time?
Lex Fridman (13:07.360)
You think like there's a 3D, like our world
Donald Hoffman (13:12.080)
is three dimensional and time progresses forward.
Lex Fridman (13:15.440)
Therefore, three dimensions plus one, 4D.
Lex Fridman (13:18.120)
What exactly do you mean by space time?
Lex Fridman (13:20.560)
And what do you mean by space time is doomed?
Donald Hoffman (13:24.080)
Great, great.
Lex Fridman (13:24.920)
So this is, by the way, not my quote.
Donald Hoffman (13:26.600)
This is from, for example, Nima Arkanihaim Ed
Lex Fridman (13:29.880)
at the Institute for Advanced Study at Princeton.
Donald Hoffman (13:31.760)
Ed Witten, also there.
Lex Fridman (13:34.340)
David Gross, Nobel Prize winner.
Lex Fridman (13:36.560)
So this is not just something the cognitive scientists,
Lex Fridman (13:39.360)
this is what the physicists are saying.
Donald Hoffman (13:40.680)
Yeah, the physicists, they're space time skeptics.
Lex Fridman (13:45.120)
Well, yeah, they're saying that,
Lex Fridman (13:46.680)
and I can say exactly why they think it's doomed.
Lex Fridman (13:49.600)
But what they're saying is that,
Donald Hoffman (13:51.640)
because your question was what aspect of space time,
Lex Fridman (13:53.880)
what are we talking about here?
Donald Hoffman (13:55.000)
It's both space and time.
Lex Fridman (13:56.920)
They're union into space time as an Einstein's theory.
Donald Hoffman (13:59.760)
That's doomed.
Lex Fridman (14:01.280)
And they're basically saying that even quantum theory,
Donald Hoffman (14:06.960)
this is with Nima Arkanihaim Ed, especially.
Lex Fridman (14:09.240)
So Hilbert spaces will not be fundamental either.
Lex Fridman (14:12.920)
So that the notion of Hilbert space,
Lex Fridman (14:15.560)
which is really critical to quantum field theory,
Donald Hoffman (14:19.000)
quantum information theory,
Lex Fridman (14:21.000)
that's not going to figure
Donald Hoffman (14:22.000)
in the fundamental new laws of physics.
Lex Fridman (14:25.280)
So what they're looking for
Donald Hoffman (14:26.840)
is some new mathematical structures beyond space time,
Lex Fridman (14:31.960)
beyond Einstein's four dimensional space time
Donald Hoffman (14:35.240)
or super symmetric version,
Lex Fridman (14:38.040)
geometric algebra signature two comma four kind of.
Donald Hoffman (14:41.160)
There are different ways that you can represent it,
Lex Fridman (14:43.280)
but they're finding new structures.
Lex Fridman (14:45.560)
And then by the way, they're succeeding now.
Lex Fridman (14:47.320)
They're finding, they found something
Donald Hoffman (14:48.440)
called the amplituhedron.
Lex Fridman (14:49.680)
This is Nima and his colleagues,
Donald Hoffman (14:51.520)
the cosmological polytope.
Lex Fridman (14:53.600)
So there are these like polytopes,
Donald Hoffman (14:57.400)
these polyhedra in multi dimensions,
Lex Fridman (15:00.320)
generalizations of simplices that are coding for,
Donald Hoffman (15:05.720)
for example, the scattering amplitudes of processes
Lex Fridman (15:08.240)
in the Large Hadron Collider and other colliders.
Lex Fridman (15:10.880)
So they're finding that if they let go of space time,
Lex Fridman (15:14.040)
completely, they're finding new ways
Donald Hoffman (15:16.960)
of computing these scattering amplitudes
Lex Fridman (15:18.760)
that turn literally billions of terms into one term.
Donald Hoffman (15:23.760)
When you do it in space and time,
Lex Fridman (15:25.040)
because it's the wrong framework,
Donald Hoffman (15:26.800)
it's just a user interface from,
Lex Fridman (15:29.560)
that's not from the evolutionary point of view,
Donald Hoffman (15:30.920)
it's just user interface.
Lex Fridman (15:32.160)
It's not a deep insight into the nature of reality.
Lex Fridman (15:34.600)
So it's missing deep symmetry
Lex Fridman (15:36.920)
is something called a dual conformal symmetry,
Donald Hoffman (15:39.040)
which turns out to be true of the scattering data,
Lex Fridman (15:40.960)
but you can't see it in space time.
Lex Fridman (15:42.720)
And it's making the computations way too complicated
Lex Fridman (15:46.280)
because you're trying to compute all the loops
Donald Hoffman (15:47.920)
in the Feynman diagrams and all the Feynman integrals.
Lex Fridman (15:50.320)
So see the Feynman approach to the scattering amplitudes
Donald Hoffman (15:53.000)
is trying to enforce two critical properties of space time,
Lex Fridman (15:56.520)
locality and unitarity.
Lex Fridman (15:58.600)
And so by, when you enforce those,
Lex Fridman (16:00.440)
you get all these loops and multiple,
Donald Hoffman (16:03.400)
different levels of loops.
Lex Fridman (16:04.600)
And for each of those,
Donald Hoffman (16:05.440)
you have to add new terms to your computation.
Lex Fridman (16:07.600)
But when you do it outside of space time,
Donald Hoffman (16:11.080)
you don't have the notion of unitarity.
Lex Fridman (16:13.800)
You don't have the notion of locality.
Donald Hoffman (16:15.760)
You have something deeper
Lex Fridman (16:17.320)
and it's capturing some symmetries
Donald Hoffman (16:18.840)
that are actually true of the data.
Lex Fridman (16:20.800)
And, but then when you look at the geometry
Donald Hoffman (16:23.000)
of the facets of these polytopes,
Lex Fridman (16:25.360)
then certain of them will code for unitarity and locality.
Lex Fridman (16:30.800)
So it actually comes out of the structure
Lex Fridman (16:32.400)
of these deep polytopes.
Lex Fridman (16:33.520)
So what we're finding is there's this whole new world.
Lex Fridman (16:36.520)
Now beyond space time that is making explicit symmetries
Donald Hoffman (16:42.080)
that are true of the data
Lex Fridman (16:43.000)
that cannot be seen in space time.
Lex Fridman (16:45.040)
And that is turning the computations
Lex Fridman (16:46.720)
from billions of terms to one or two or a handful of terms.
Lex Fridman (16:50.400)
So we're getting insights into symmetries
Lex Fridman (16:53.280)
and all of a sudden the math is becoming simple
Donald Hoffman (16:55.400)
because we're not doing something silly.
Lex Fridman (16:56.880)
We're not adding up all these loops in space time.
Donald Hoffman (16:59.040)
We're doing something far deeper.
Lex Fridman (17:00.840)
But they don't know what this world is about.
Donald Hoffman (17:02.840)
Also, they're in an interesting position
Lex Fridman (17:07.080)
where we know that space time is doomed.
Lex Fridman (17:09.080)
And I should probably tell you why it's doomed,
Lex Fridman (17:11.360)
what they're saying about why it's doomed.
Lex Fridman (17:12.840)
But they need a flashlight to look beyond space time.
Lex Fridman (17:15.600)
What flashlight are we gonna use
Lex Fridman (17:17.400)
to look into the dark beyond space time?
Lex Fridman (17:19.720)
Because Einstein's theory and quantum theory
Donald Hoffman (17:22.280)
can't tell us what's beyond them.
Lex Fridman (17:23.920)
All they can do is tell us that when you put us together,
Donald Hoffman (17:26.240)
space time is doomed at 10 to the minus 33 centimeters,
Lex Fridman (17:30.080)
10 to the minus 43 seconds.
Donald Hoffman (17:31.600)
Beyond that, space time doesn't even make sense.
Lex Fridman (17:34.000)
It just has no operational definition.
Donald Hoffman (17:37.000)
So, but it doesn't tell you what's beyond.
Lex Fridman (17:39.000)
And so they're just looking for deep structures
Donald Hoffman (17:41.560)
like guessing is really fun.
Lex Fridman (17:43.680)
So these really brilliant guys, generic brilliant men
Lex Fridman (17:47.160)
and women who are doing this work, physicists,
Lex Fridman (17:49.760)
are making guesses about these structures,
Donald Hoffman (17:52.280)
informed guesses, because they're trying to ask,
Lex Fridman (17:54.120)
well, okay, what deeper structure could give us
Donald Hoffman (17:56.560)
the stuff that we're seeing in space time,
Lex Fridman (17:58.360)
but without certain commitments
Donald Hoffman (17:59.960)
that we have to make in space time, like locality.
Lex Fridman (18:02.840)
So they make these brilliant guesses.
Lex Fridman (18:04.680)
And of course, most of the time you're gonna be wrong,
Lex Fridman (18:06.640)
but once you get one or two that start to pay off
Lex Fridman (18:09.560)
and then you get some lucky breaks.
Lex Fridman (18:11.320)
So they got a lucky break back in 1986.
Donald Hoffman (18:15.520)
Couple of mathematicians named Park and Taylor
Lex Fridman (18:18.680)
took the scattering amplitude for two gluons coming in
Donald Hoffman (18:22.640)
at high energy and four gluons going out at low energy.
Lex Fridman (18:25.560)
So that kind of scattering thing.
Lex Fridman (18:27.200)
So it's like apparently for people who are into this,
Lex Fridman (18:30.280)
that's sort of something that happens so often
Donald Hoffman (18:32.040)
you need to be able to find it and get rid of those
Lex Fridman (18:34.320)
cause you already know about that and you need to.
Lex Fridman (18:36.160)
So you needed to compute them.
Lex Fridman (18:37.280)
It was billions of terms and they couldn't do it
Donald Hoffman (18:39.880)
even though for the supercomputers couldn't do that
Lex Fridman (18:41.960)
for the many billions or millions of times per second
Donald Hoffman (18:44.560)
they needed to do it.
Lex Fridman (18:45.400)
So the experimentals begged the theorists,
Donald Hoffman (18:49.040)
please, you got it.
Lex Fridman (18:51.280)
And so Park and Taylor took the billions of terms,
Donald Hoffman (18:53.160)
hundreds of pages and miraculously turned it into nine.
Lex Fridman (18:58.120)
And then a little bit later,
Donald Hoffman (18:59.360)
they guessed one term expression
Lex Fridman (19:01.240)
that turned out to be equivalent.
Lex Fridman (19:02.480)
So billions of terms reduced to one term,
Lex Fridman (19:07.240)
that so called famous Park Taylor formula, 1986.
Lex Fridman (19:10.720)
And that was like, okay, where did that come from?
Lex Fridman (19:14.240)
This is a pointer into a deep realm, beyond space and time,
Lex Fridman (19:18.840)
but no one, I mean, what can you do with it?
Lex Fridman (19:21.480)
And they thought maybe it was a one off,
Lex Fridman (19:23.120)
but then other formulas started coming up.
Lex Fridman (19:25.840)
And then eventually Neymar, Connie, Hamid and his team
Donald Hoffman (19:28.440)
found this thing called the amplituhedron,
Lex Fridman (19:30.200)
which really sort of captures the whole,
Donald Hoffman (19:32.640)
a big part of the whole ball of wax.
Lex Fridman (19:34.600)
I'm sure they would say, no, there's plenty more to do.
Lex Fridman (19:37.800)
So I won't say they did it all by any means.
Lex Fridman (19:40.440)
They're looking at the cosmological polytope as well.
Lex Fridman (19:42.640)
So what's remarkable to me is that two pillars
Lex Fridman (19:48.040)
of modern science, quantum field theory with gravity
Donald Hoffman (19:51.640)
on the one hand and evolution by natural selection
Lex Fridman (19:54.240)
on the other, just in the last 20 years
Donald Hoffman (19:56.840)
have very clearly said space time has had a good run.
Lex Fridman (1:00:06.080)
If they came up, they were gone in one generation,
Donald Hoffman (1:00:07.840)
they just weren't.
Lex Fridman (1:00:09.440)
So they came and went even just in one generation.
Donald Hoffman (1:00:14.720)
They just are not good enough.
Lex Fridman (1:00:16.340)
The ones that were just tracking,
Donald Hoffman (1:00:18.020)
their senses just were tracking the fitness payoffs
Lex Fridman (1:00:20.900)
were far more fit than the truth seekers.
Lex Fridman (1:00:25.900)
So an answer at one level,
Lex Fridman (1:00:29.260)
I want to give an answer at a deeper level,
Lex Fridman (1:00:30.500)
but just with evolutionary game theory,
Lex Fridman (1:00:32.860)
because my attitude as a scientist is,
Donald Hoffman (1:00:36.140)
I don't believe any of our theories.
Lex Fridman (1:00:38.900)
I take them very, very seriously.
Donald Hoffman (1:00:40.220)
I study them, I look at their implications,
Lex Fridman (1:00:42.020)
but none of them are the gospel.
Donald Hoffman (1:00:43.580)
They're just the latest ideas that we have.
Lex Fridman (1:00:46.020)
And so the reason I study evolutionary game theory
Donald Hoffman (1:00:49.340)
is because that's the best tool we have right now
Lex Fridman (1:00:52.460)
in this area.
Donald Hoffman (1:00:53.780)
There is nothing else that competes.
Lex Fridman (1:00:56.300)
And so as a scientist, it's my responsibility
Donald Hoffman (1:00:58.500)
to take the best tools and see what they mean.
Lex Fridman (1:01:01.020)
And the same thing the physicists are doing.
Donald Hoffman (1:01:02.620)
They're taking the best tools
Lex Fridman (1:01:03.820)
and looking at what they entail.
Lex Fridman (1:01:06.220)
But I think that science now has enough experience
Lex Fridman (1:01:10.660)
to realize that we should not believe our theories
Donald Hoffman (1:01:14.180)
in the sense that we've now arrived.
Lex Fridman (1:01:17.020)
In 1890, a lot of physicists thought we'd arrived.
Donald Hoffman (1:01:21.500)
They were discouraging bright young students
Lex Fridman (1:01:25.780)
from going into physics, because it was all done.
Lex Fridman (1:01:27.900)
And that's precisely the wrong attitude forever.
Lex Fridman (1:01:31.540)
It's the wrong attitude forever.
Donald Hoffman (1:01:33.700)
The attitude we should have is a century from now,
Lex Fridman (1:01:37.740)
they'll be looking at us and laughing
Donald Hoffman (1:01:39.700)
at what we didn't know.
Lex Fridman (1:01:40.820)
And we just have to assume that that's going to be the case.
Donald Hoffman (1:01:43.300)
Just know that everything that we think
Lex Fridman (1:01:45.300)
is so brilliant right now, our final theory.
Donald Hoffman (1:01:48.460)
A century from now, they'll look at us
Lex Fridman (1:01:50.060)
like we look at the physicists of 1890 and go,
Lex Fridman (1:01:52.780)
how could they have been so dumb?
Lex Fridman (1:01:54.620)
So I don't want to make that mistake.
Lex Fridman (1:01:56.900)
So I'm not doctrinaire about any
Lex Fridman (1:02:00.220)
of our current scientific theories.
Donald Hoffman (1:02:02.100)
I am doctrinaire about this.
Lex Fridman (1:02:05.500)
We should use the best tools we have right now.
Donald Hoffman (1:02:08.260)
That's what we've got.
Lex Fridman (1:02:09.100)
And with humility.
Donald Hoffman (1:02:10.340)
Well, so let me ask you about game theory.
Lex Fridman (1:02:13.740)
I love game theory, evolutionary game theory.
Lex Fridman (1:02:18.100)
But I'm always suspicious of it, like economics.
Lex Fridman (1:02:23.500)
When you construct models,
Donald Hoffman (1:02:25.860)
it's too easy to construct things that oversimplify
Lex Fridman (1:02:31.380)
just because we, our human brains,
Donald Hoffman (1:02:34.420)
enjoy the simplification of constructing a few variables
Lex Fridman (1:02:39.020)
that somehow represent organisms or represent people
Lex Fridman (1:02:43.100)
and running a simulation that then allows you
Lex Fridman (1:02:45.580)
to build up intuition and then it feels really good
Donald Hoffman (1:02:48.460)
because you can get some really deep
Lex Fridman (1:02:50.420)
and surprising intuitions.
Lex Fridman (1:02:51.980)
But how do you know your models aren't,
Lex Fridman (1:02:55.260)
the assumptions underlying your models
Lex Fridman (1:02:57.200)
aren't some fundamentally flawed?
Lex Fridman (1:02:58.900)
And because of that,
Donald Hoffman (1:03:00.700)
your conclusions are fundamentally flawed.
Lex Fridman (1:03:03.020)
So I guess my question is what are the limits
Donald Hoffman (1:03:06.220)
in your use of game theory, evolution game theory,
Lex Fridman (1:03:08.860)
your experience with it?
Lex Fridman (1:03:10.220)
What are the limits of game theory?
Lex Fridman (1:03:12.460)
So I've gotten some pushback from professional colleagues
Lex Fridman (1:03:15.780)
and friends who have tried to rerun simulations
Lex Fridman (1:03:19.340)
and try to, the idea that we don't see the truth
Donald Hoffman (1:03:21.840)
is not comfortable and so many of my colleagues
Lex Fridman (1:03:24.220)
are very interested in trying to show that we're wrong.
Lex Fridman (1:03:26.420)
And so the idea would be to say that somehow
Lex Fridman (1:03:28.620)
we did something, as you're suggesting,
Donald Hoffman (1:03:30.580)
maybe something special that wasn't completely general.
Lex Fridman (1:03:33.740)
We got some little special part of the whole search space
Donald Hoffman (1:03:36.980)
in evolutionary game theory in which this happens to be true
Lex Fridman (1:03:39.580)
but more generally organisms would evolve
Donald Hoffman (1:03:42.100)
to see the truth.
Lex Fridman (1:03:42.940)
So the best pushback we've gotten is from a team at Yale.
Lex Fridman (1:03:48.100)
And they suggested that if you use
Lex Fridman (1:03:52.180)
thousands of payoff functions,
Lex Fridman (1:03:53.980)
so we in our simulations, we just use a couple,
Lex Fridman (1:03:57.140)
one or two, because it was our first simulations, right?
Lex Fridman (1:04:00.300)
So that would be a limit.
Lex Fridman (1:04:01.220)
We had one or two payoff functions,
Donald Hoffman (1:04:02.380)
we showed the result of those,
Lex Fridman (1:04:05.100)
at least for the genetic algorithms.
Lex Fridman (1:04:07.060)
And they said, if you have 20,000 of them,
Lex Fridman (1:04:10.460)
then we can find these conditions in which
Donald Hoffman (1:04:14.260)
truth seeing organisms would be the ones
Lex Fridman (1:04:17.260)
that evolved and survived.
Lex Fridman (1:04:19.340)
And so we looked at their simulations
Lex Fridman (1:04:21.140)
and it certainly is the case that you can find
Donald Hoffman (1:04:25.180)
special cases in which truth can evolve.
Lex Fridman (1:04:27.580)
So when I say it's probability zero,
Donald Hoffman (1:04:29.140)
it doesn't mean it can't happen.
Lex Fridman (1:04:30.140)
It can happen, in fact, it could happen infinitely often.
Donald Hoffman (1:04:32.980)
It's just probability zero.
Lex Fridman (1:04:34.380)
So probability zero things can happen infinitely often.
Donald Hoffman (1:04:38.260)
When you say probability is zero, you mean probability
Lex Fridman (1:04:40.380)
close to zero.
Donald Hoffman (1:04:42.020)
To be very, very precise.
Lex Fridman (1:04:43.140)
So for example, if I have a unit square on the plane
Lex Fridman (1:04:48.780)
and I use a measure on a probability measure
Lex Fridman (1:04:53.180)
in which the area of a region is this probability.
Donald Hoffman (1:04:58.340)
Then if I draw a curve in that unit square,
Lex Fridman (1:05:02.420)
it has measure precisely zero,
Donald Hoffman (1:05:05.500)
precisely not approximately, precisely zero.
Lex Fridman (1:05:07.820)
And yet it has infinitely many points.
Lex Fridman (1:05:10.100)
So there's an object that for that probability measure
Lex Fridman (1:05:12.180)
has probability zero, and yet there's
Donald Hoffman (1:05:14.020)
infinitely many points in it.
Lex Fridman (1:05:16.260)
So that's what I mean when I say that things
Donald Hoffman (1:05:19.060)
that are probability zero can happen
Lex Fridman (1:05:20.300)
infinitely often in principle.
Donald Hoffman (1:05:21.780)
Yeah, but infinity, as far as, and I look outside often,
Lex Fridman (1:05:26.500)
I walk around and I look at people.
Donald Hoffman (1:05:29.180)
I have never seen infinity in real life.
Lex Fridman (1:05:32.860)
That's an interesting issue.
Donald Hoffman (1:05:35.980)
I've been looking, I've been looking.
Lex Fridman (1:05:37.540)
I don't notice it, infinitely small or the infinitely big.
Lex Fridman (1:05:41.220)
And so the tools of mathematics,
Lex Fridman (1:05:43.360)
you could sort of apply the same kind of criticism
Donald Hoffman (1:05:45.700)
that it is a very convenient interface into our reality.
Lex Fridman (1:05:49.220)
That's a big debate in mathematics,
Donald Hoffman (1:05:50.500)
the intuitionists versus the ones who take,
Lex Fridman (1:05:52.260)
for example, the real numbers as real.
Lex Fridman (1:05:55.140)
And that's a fun discussion.
Lex Fridman (1:05:57.080)
Nicholas Giesen, a physicist,
Donald Hoffman (1:05:59.060)
has really interesting work recently
Lex Fridman (1:06:00.900)
on how if you go with intuitionist mathematics,
Donald Hoffman (1:06:04.160)
you could effectively quantize Newton,
Lex Fridman (1:06:10.020)
and you find that the Newtonian theory
Lex Fridman (1:06:12.460)
and quantum theory aren't that different
Lex Fridman (1:06:14.460)
once you go with it.
Donald Hoffman (1:06:16.540)
It's funny.
Lex Fridman (1:06:17.380)
It's really quite interesting.
Lex Fridman (1:06:18.220)
So the issue he raises is a very, very deep one,
Lex Fridman (1:06:21.040)
and one that I think we should take quite seriously,
Donald Hoffman (1:06:23.780)
which is how should we think about the reality
Lex Fridman (1:06:27.660)
of the contours hierarchy?
Donald Hoffman (1:06:30.420)
Aleph one, aleph two, and all these different infinities
Lex Fridman (1:06:35.860)
versus just a more algorithmic approach, right?
Lex Fridman (1:06:41.660)
So where everything's computable,
Lex Fridman (1:06:44.700)
in some sense, everything's finite,
Donald Hoffman (1:06:46.420)
as big as you want, but nevertheless finite.
Lex Fridman (1:06:50.620)
So yeah, that ultimately boils down to
Donald Hoffman (1:06:52.560)
whether the world is discrete or continuous
Lex Fridman (1:06:56.820)
in some general sense.
Lex Fridman (1:06:59.320)
And again, we can't really know,
Lex Fridman (1:07:01.220)
but there's just a mind breaking thought,
Donald Hoffman (1:07:05.500)
just common sense reasoning,
Lex Fridman (1:07:07.340)
that something can happen,
Lex Fridman (1:07:09.880)
and as yet, probability of it happening is 0%.
Lex Fridman (1:07:13.820)
That doesn't compute for common sense computer.
Donald Hoffman (1:07:18.140)
Right.
Lex Fridman (1:07:18.980)
This is where you have to be a sharp mathematician
Donald Hoffman (1:07:21.780)
to really, and I'm not.
Lex Fridman (1:07:23.460)
Sharp is one word.
Lex Fridman (1:07:24.980)
What I'm saying is common sense computer is,
Lex Fridman (1:07:27.420)
I mean that in a very kind of,
Donald Hoffman (1:07:33.340)
in a positive sense,
Lex Fridman (1:07:35.120)
because we've been talking about perception systems
Lex Fridman (1:07:37.300)
and interfaces, if we are to reason about the world,
Lex Fridman (1:07:42.140)
we have to use the best interfaces we got.
Lex Fridman (1:07:45.060)
And I'm not exactly sure that game theory
Lex Fridman (1:07:50.340)
is the best interface we got for this.
Donald Hoffman (1:07:52.780)
Oh, right.
Lex Fridman (1:07:53.620)
In application of mathematics, tricks and tools
Donald Hoffman (1:07:57.740)
in mathematics, the game theory is the best we got
Lex Fridman (1:08:00.620)
when we are thinking about the nature of reality
Lex Fridman (1:08:03.980)
and fitness functions and evolution, period.
Lex Fridman (1:08:07.020)
Right.
Donald Hoffman (1:08:07.860)
Well, that's a fair rejoinder,
Lex Fridman (1:08:10.080)
and I think that that was the tool that we used.
Lex Fridman (1:08:14.100)
And if someone says, here's a better mathematical tool
Lex Fridman (1:08:17.360)
and here's why, this mathematical tool
Donald Hoffman (1:08:20.260)
better captures the essence of Darwin's idea,
Lex Fridman (1:08:23.300)
John Maynard Smith didn't quite get it
Donald Hoffman (1:08:24.980)
with evolutionary game theory.
Lex Fridman (1:08:26.300)
There's this thing.
Donald Hoffman (1:08:27.540)
Now there are tools like evolutionary graph theory,
Lex Fridman (1:08:30.660)
which generalize evolutionary game theory,
Lex Fridman (1:08:32.900)
and then there's quantum game theory.
Lex Fridman (1:08:35.580)
So you can use quantum tools like entanglement,
Donald Hoffman (1:08:41.800)
for example, as a resource in games
Lex Fridman (1:08:44.460)
that change the very nature of the solutions,
Donald Hoffman (1:08:48.660)
the optimal solutions of the game theory.
Lex Fridman (1:08:50.700)
Well, the work from Yale is really interesting.
Donald Hoffman (1:08:54.420)
It's a really interesting challenge of these ideas
Lex Fridman (1:08:58.400)
where, okay, if you have a very large number
Donald Hoffman (1:09:00.940)
of fitness functions, or let's say you have
Lex Fridman (1:09:04.380)
a nearly infinite number of fitness functions
Donald Hoffman (1:09:07.500)
or a growing number of fitness functions,
Lex Fridman (1:09:09.220)
what kind of interesting things start to emerging
Lex Fridman (1:09:13.740)
if you are to be an organism?
Lex Fridman (1:09:15.540)
If to be an organism that adapts means
Donald Hoffman (1:09:18.860)
having to deal with an ensemble of fitness functions.
Lex Fridman (1:09:23.340)
Right, and so we've actually redone some of our own work
Donald Hoffman (1:09:28.500)
based on theirs, and this is the back and forth
Lex Fridman (1:09:30.300)
that we expect in science, right?
Lex Fridman (1:09:32.300)
And what we found was that in their simulations,
Lex Fridman (1:09:36.380)
they were assuming that you couldn't carve the world
Donald Hoffman (1:09:39.140)
up into objects, and so we said,
Lex Fridman (1:09:42.040)
well, let's relax that assumption.
Donald Hoffman (1:09:43.940)
Allow organisms to create data structures
Lex Fridman (1:09:45.980)
that we might call objects,
Lex Fridman (1:09:47.720)
and an object would be you take,
Lex Fridman (1:09:49.420)
you would do hierarchical clustering
Donald Hoffman (1:09:51.580)
of your fitness payoff functions,
Lex Fridman (1:09:53.180)
the ones that have similar shapes.
Donald Hoffman (1:09:54.860)
If you have 20,000 of them, maybe these 50
Lex Fridman (1:09:58.300)
are all very, very similar,
Lex Fridman (1:09:59.540)
so I can take all the perception, action, fitness stuff
Lex Fridman (1:10:03.620)
and make that into a data structure,
Lex Fridman (1:10:05.700)
and we'll call that a unit or an object.
Lex Fridman (1:10:08.300)
And as soon as we did that,
Donald Hoffman (1:10:09.640)
then all of their results went away.
Lex Fridman (1:10:11.600)
It turned out they were the special case
Lex Fridman (1:10:13.400)
and that the organisms that were allowed
Lex Fridman (1:10:16.660)
to only see, that were shaped to see only fitness payoffs
Donald Hoffman (1:10:21.460)
were the ones that were.
Lex Fridman (1:10:22.940)
So the idea is that objects then,
Lex Fridman (1:10:25.340)
what are objects from an evolutionary point of view?
Lex Fridman (1:10:27.140)
This bottle, we thought that when I saw a bottle,
Donald Hoffman (1:10:30.180)
it was because I was seeing a true object
Lex Fridman (1:10:31.780)
that existed whether or not it was perceived.
Donald Hoffman (1:10:34.400)
Evolutionary theories suggest a different interpretation.
Lex Fridman (1:10:37.980)
I'm seeing a data structure that is encoding
Donald Hoffman (1:10:42.300)
a convenient way of looking at various fitness payoffs.
Lex Fridman (1:10:45.420)
I can use this for drinking.
Donald Hoffman (1:10:48.300)
I could use it as a weapon, not a very good one.
Lex Fridman (1:10:50.740)
I could be somewhere with head with it.
Donald Hoffman (1:10:52.840)
If my goal is mating, this is pointless.
Lex Fridman (1:10:56.580)
So I'm seeing for, what I'm coding here
Donald Hoffman (1:10:59.640)
is all sorts of actions and the payoffs that I could get.
Lex Fridman (1:11:04.180)
When I pick up an apple,
Donald Hoffman (1:11:05.460)
now I'm getting a different set of actions and payoffs.
Lex Fridman (1:11:08.680)
When I pick up a rock, I'm getting, so for every object,
Lex Fridman (1:11:11.700)
what I'm getting is a different set of payoff functions
Lex Fridman (1:11:16.100)
and with various actions.
Lex Fridman (1:11:18.180)
And so once you allow that,
Lex Fridman (1:11:20.660)
then what you find is once again that truth goes extinct
Lex Fridman (1:11:25.620)
and the organisms that just get an interface
Lex Fridman (1:11:28.020)
are the ones that win.
Lex Fridman (1:11:29.540)
But the question, just sneaking up on, this is fascinating.
Lex Fridman (1:11:34.960)
From where do fitness functions originate?
Lex Fridman (1:11:38.180)
What gives birth to the fitness functions?
Lex Fridman (1:11:40.100)
So if there's a giant black box
Donald Hoffman (1:11:43.340)
that just keeps giving you fitness functions,
Lex Fridman (1:11:45.180)
what are we trying to optimize?
Donald Hoffman (1:11:46.220)
You said that water has different uses than an apple.
Lex Fridman (1:11:55.320)
So there's these objects.
Lex Fridman (1:11:57.000)
What are we trying to optimize?
Lex Fridman (1:11:58.860)
And why is not reality a really good generator
Lex Fridman (1:12:02.740)
of fitness functions?
Lex Fridman (1:12:05.380)
So each theory makes its own assumptions and says,
Donald Hoffman (1:12:07.760)
grant me this, then I'll explain that.
Lex Fridman (1:12:09.660)
So evolutionary game theory says,
Lex Fridman (1:12:11.020)
grant me fitness payoffs, right?
Lex Fridman (1:12:13.420)
And grant me strategies with payoffs.
Lex Fridman (1:12:16.340)
And I can write down the matrix
Lex Fridman (1:12:18.540)
for this strategy interacts with that strategy.
Donald Hoffman (1:12:20.340)
These are the payoffs that come up.
Lex Fridman (1:12:21.660)
If you grant me that,
Donald Hoffman (1:12:22.500)
then I can start to explain a lot of things.
Lex Fridman (1:12:24.560)
Now you can ask for a deeper question like,
Donald Hoffman (1:12:26.460)
okay, how does physics evolve biology
Lex Fridman (1:12:32.100)
and where do these fitness payoffs come from, right?
Donald Hoffman (1:12:36.100)
Now that's a completely different enterprise.
Lex Fridman (1:12:41.380)
And of course, evolutionary game theory then
Donald Hoffman (1:12:43.160)
would be not the right tool for that.
Lex Fridman (1:12:45.360)
It would have to be a deeper tool
Donald Hoffman (1:12:46.500)
that shows where evolutionary game theory comes from.
Lex Fridman (1:12:50.680)
My own take is that there's gonna be a problem
Donald Hoffman (1:12:55.220)
in doing that because space time isn't fundamental.
Lex Fridman (1:13:01.820)
It's just a user interface.
Lex Fridman (1:13:03.360)
And that the distinction that we make
Lex Fridman (1:13:06.220)
between living and nonliving
Donald Hoffman (1:13:08.500)
is not a fundamental distinction.
Lex Fridman (1:13:10.620)
It's an artifact of the limits of our interface, right?
Lex Fridman (1:13:15.140)
So this is a new wrinkle and this is an important wrinkle.
Lex Fridman (1:13:19.380)
It's so nice to take space and time as fundamental
Donald Hoffman (1:13:22.260)
because if something looks like it's inanimate,
Lex Fridman (1:13:24.280)
it's inanimate and we can just say it's not living.
Donald Hoffman (1:13:27.180)
Now it's much more complicated.
Lex Fridman (1:13:30.740)
Certain things are obviously living.
Donald Hoffman (1:13:32.200)
I'm talking with you, I'm obviously interacting
Lex Fridman (1:13:35.560)
with something that's alive and conscious.
Donald Hoffman (1:13:38.660)
I think we've let go of the word obviously
Lex Fridman (1:13:40.700)
in this conversation.
Donald Hoffman (1:13:42.020)
I think nothing is obvious.
Lex Fridman (1:13:43.500)
Nothing's obvious, that's right.
Lex Fridman (1:13:45.280)
But when we get down to like an ant,
Lex Fridman (1:13:48.700)
it's obviously living, but I'll say it appears to be living.
Lex Fridman (1:13:52.580)
But when we get down to a virus, now people wonder
Lex Fridman (1:13:55.600)
and when we get down to protons,
Donald Hoffman (1:13:57.260)
people say it's not living.
Lex Fridman (1:13:58.840)
And my attitude is look, I have a user interface.
Donald Hoffman (1:14:02.780)
Interface is there to hide certain aspects of reality
Lex Fridman (1:14:05.780)
and others to, it's an uneven representation,
Donald Hoffman (1:14:11.100)
put it that way.
Lex Fridman (1:14:11.980)
Certain things just get completely hidden.
Donald Hoffman (1:14:14.920)
Dark matter and dark energy are most of the energy
Lex Fridman (1:14:18.580)
and matter that's out there.
Donald Hoffman (1:14:19.780)
Our interface just plain flat out hides them.
Lex Fridman (1:14:23.380)
The only way we get some hint is because
Donald Hoffman (1:14:25.600)
gravitational things are going wrong within our,
Lex Fridman (1:14:28.740)
so most things are outside of our interface.
Donald Hoffman (1:14:31.980)
The distinction between living and nonliving
Lex Fridman (1:14:35.140)
is not fundamental.
Donald Hoffman (1:14:35.980)
It's an artifact of our interface.
Lex Fridman (1:14:37.200)
So if we really, really want to understand
Donald Hoffman (1:14:41.900)
where evolution comes from,
Lex Fridman (1:14:44.740)
to answer the question, the deep question you asked,
Donald Hoffman (1:14:46.740)
I think the right way we're gonna have to do that
Lex Fridman (1:14:48.660)
is to come up with a deeper theory than space time
Donald Hoffman (1:14:52.140)
in which there may not be the notion of time
Lex Fridman (1:14:54.300)
and show that whatever this dynamics of that deeper theory
Donald Hoffman (1:15:00.420)
is, by the way, I'll talk about how you could have dynamics
Lex Fridman (1:15:03.900)
without time, but the dynamics of this deeper theory,
Donald Hoffman (1:15:07.140)
when we project it into, in certain ways,
Lex Fridman (1:15:11.020)
then we do get space time and we get what appears
Donald Hoffman (1:15:13.060)
to be evolution by natural selection.
Lex Fridman (1:15:15.060)
So I would love to see evolution by natural selection,
Donald Hoffman (1:15:17.760)
nature, red and tooth and claw, people fighting,
Lex Fridman (1:15:20.220)
animals fighting for resources and the whole bit,
Donald Hoffman (1:15:22.220)
come out of a deeper theory in which perhaps
Lex Fridman (1:15:24.260)
it's all cooperation, there's no limited resources
Lex Fridman (1:15:27.260)
and so forth, but as a result of projection,
Lex Fridman (1:15:30.660)
you get space and time, and as a result of projection,
Donald Hoffman (1:15:33.580)
you get nature, red and tooth and claw,
Lex Fridman (1:15:35.420)
the appearance of it, but it's all an artifact
Donald Hoffman (1:15:38.420)
of the interface.
Lex Fridman (1:15:39.260)
I like this idea that the line between living
Lex Fridman (1:15:43.220)
and nonliving is very important
Lex Fridman (1:15:46.620)
because that's the thing that would emerge
Donald Hoffman (1:15:48.660)
before you have evolution, the idea of death.
Lex Fridman (1:15:55.220)
So that seems to be an important component
Donald Hoffman (1:15:58.900)
of natural selection, and if that emerged,
Lex Fridman (1:16:01.060)
because that's also asking the question,
Lex Fridman (1:16:05.540)
I guess, that I ask, where do fitness functions come from?
Lex Fridman (1:16:09.100)
That's like asking the old meaning of life question, right?
Lex Fridman (1:16:12.980)
It's the why, why, why?
Lex Fridman (1:16:17.500)
And one of the big underlying whys,
Donald Hoffman (1:16:20.300)
okay, you can start with evolution on Earth,
Lex Fridman (1:16:22.660)
but without living, without life and death,
Donald Hoffman (1:16:26.100)
without the line between the living and the dead,
Lex Fridman (1:16:28.700)
you don't have evolution.
Lex Fridman (1:16:30.500)
So what if underneath it, there's no such thing
Lex Fridman (1:16:32.540)
as the living and the dead?
Donald Hoffman (1:16:35.020)
There's no, like this concept of an organism, period.
Lex Fridman (1:16:39.500)
There's a living organism that's defined
Donald Hoffman (1:16:42.700)
by a volume in space time that somehow interacts,
Lex Fridman (1:16:48.340)
that over time maintains its integrity somehow.
Donald Hoffman (1:16:52.820)
It has some kind of history, it has a wall of some kind.
Lex Fridman (1:16:56.540)
The outside world, the environment,
Lex Fridman (1:16:58.340)
and then inside, there's an organism.
Lex Fridman (1:17:00.740)
So you're defining an organism,
Lex Fridman (1:17:02.900)
and also you define that organism
Lex Fridman (1:17:04.860)
by the fact that it can move, and it can become alive,
Donald Hoffman (1:17:10.220)
which you kind of think of as moving,
Lex Fridman (1:17:12.540)
combined with the fact that it's keeping itself
Donald Hoffman (1:17:14.700)
separate from the environment,
Lex Fridman (1:17:15.980)
so you can point out that thing is living,
Lex Fridman (1:17:17.900)
and then it can also die.
Lex Fridman (1:17:21.060)
That seems to be all very powerful components of space time
Donald Hoffman (1:17:26.340)
that enable you to have something
Lex Fridman (1:17:28.380)
like natural selection and evolution.
Donald Hoffman (1:17:31.660)
Well, and there's a lot of interesting work,
Lex Fridman (1:17:33.140)
some of it by collaborators of Carl Friston and others,
Donald Hoffman (1:17:36.180)
where they have Bayes net kind of stuff
Lex Fridman (1:17:40.940)
that they build on the notion of a Markov blanket.
Lex Fridman (1:17:43.780)
So you have some states within this network
Lex Fridman (1:17:47.180)
that are inside the blanket, then you have the blanket,
Lex Fridman (1:17:49.100)
and then the states outside the blanket.
Lex Fridman (1:17:50.780)
And the states inside this Markov blanket
Donald Hoffman (1:17:52.940)
are conditionally independent of the states
Lex Fridman (1:17:54.260)
outside the blanket conditioned on the blanket.
Lex Fridman (1:17:57.380)
And what they're looking at is that the dynamics inside
Lex Fridman (1:18:02.020)
of the states inside the Markov blanket
Donald Hoffman (1:18:04.460)
seem to be trying to estimate properties of the outside
Lex Fridman (1:18:07.140)
and react to them in a way.
Lex Fridman (1:18:08.980)
So it seems like you're doing probabilistic inferences
Lex Fridman (1:18:11.260)
in ways that might be able to keep you alive.
Lex Fridman (1:18:14.220)
So there's interesting work going on in that direction.
Lex Fridman (1:18:17.540)
But what I'm saying is something slightly different,
Lex Fridman (1:18:21.540)
and that is, like, when I look at you,
Lex Fridman (1:18:24.780)
all I see is skin, hair, and eyes, right?
Donald Hoffman (1:18:26.380)
That's all I see.
Lex Fridman (1:18:27.420)
But I know that there's a deeper reality.
Donald Hoffman (1:18:31.140)
I believe that there's a much deeper reality.
Lex Fridman (1:18:32.580)
There's the whole world of your experiences,
Donald Hoffman (1:18:34.220)
your thoughts, your hopes, your dreams.
Lex Fridman (1:18:35.780)
In some sense, the face that I see
Lex Fridman (1:18:39.820)
is just a symbol that I create, right?
Lex Fridman (1:18:42.020)
And as soon as I look away, I delete that symbol.
Lex Fridman (1:18:44.740)
But I don't delete you.
Lex Fridman (1:18:46.220)
I don't delete the conscious experience,
Donald Hoffman (1:18:48.420)
the whole world of your...
Lex Fridman (1:18:50.180)
So I'm only deleting an interface symbol.
Lex Fridman (1:18:53.420)
But that interface symbol is a portal, so to speak.
Lex Fridman (1:19:00.740)
Not a perfect portal, but a genuine portal
Donald Hoffman (1:19:04.300)
into your beliefs, into your conscious experiences.
Lex Fridman (1:19:07.260)
That's why we can have a conversation.
Donald Hoffman (1:19:09.940)
Your consciousness is genuinely affecting mine,
Lex Fridman (1:19:12.140)
and mine is genuinely affecting yours,
Donald Hoffman (1:19:13.820)
through these icons, which I create on the fly.
Lex Fridman (1:19:17.580)
I mean, I create your face.
Donald Hoffman (1:19:18.700)
When I look, I delete it.
Lex Fridman (1:19:20.060)
I don't create you, your consciousness.
Donald Hoffman (1:19:22.020)
That's there all the time, but I do...
Lex Fridman (1:19:24.940)
So now, when I look at a cat,
Donald Hoffman (1:19:27.380)
I'm creating something that I still call living,
Lex Fridman (1:19:29.860)
and I still think is conscious.
Donald Hoffman (1:19:31.860)
When I look at an ant, I create something
Lex Fridman (1:19:34.660)
that I still would call living, but maybe not conscious.
Donald Hoffman (1:19:38.100)
When I look at something I call a virus,
Lex Fridman (1:19:40.660)
now I'm not even sure I would call it living.
Lex Fridman (1:19:42.860)
And when I look at a proton, I would say,
Lex Fridman (1:19:45.380)
I don't even think it's not alive at all.
Donald Hoffman (1:19:48.860)
It could be that I'm nevertheless interacting
Lex Fridman (1:19:53.580)
with something that's just as conscious as you.
Donald Hoffman (1:19:55.660)
I'm not saying the proton is conscious.
Lex Fridman (1:19:57.420)
The face that I'm creating when I look at you,
Donald Hoffman (1:19:59.380)
that face is not conscious.
Lex Fridman (1:20:00.540)
That face is a data structure in me.
Donald Hoffman (1:20:03.660)
That face is an experience.
Lex Fridman (1:20:06.180)
It's not an experiencer.
Donald Hoffman (1:20:08.660)
Similarly, a proton is something that I create
Lex Fridman (1:20:12.660)
when I look or do a collision
Donald Hoffman (1:20:15.020)
in the Large Hadron Collider or something like that.
Lex Fridman (1:20:18.180)
But what is behind the entity in space time?
Lex Fridman (1:20:21.380)
So I've got this space time interface,
Lex Fridman (1:20:23.140)
and I've just got this entity that I call a proton.
Lex Fridman (1:20:25.540)
What is the reality behind it?
Lex Fridman (1:20:27.460)
Well, the physicists are finding these big, big structures.
Donald Hoffman (1:20:30.500)
The amplitude hadron, the sociahedron,
Lex Fridman (1:20:33.100)
cause what's behind those?
Donald Hoffman (1:20:36.020)
Could be consciousness, what I'm playing with.
Lex Fridman (1:20:38.740)
In which case, when I'm interacting with a proton,
Donald Hoffman (1:20:42.140)
I could be interacting with consciousness.
Lex Fridman (1:20:43.740)
Again, to be very, very clear,
Donald Hoffman (1:20:45.260)
because it's easy to misunderstand,
Lex Fridman (1:20:46.980)
I'm not saying a proton is conscious.
Donald Hoffman (1:20:49.300)
Just like I'm not saying your face is conscious.
Lex Fridman (1:20:51.340)
Your face is a symbol I create and then delete as I look.
Lex Fridman (1:20:56.060)
So your face is not conscious,
Lex Fridman (1:20:57.380)
but I know that that face in my interface,
Donald Hoffman (1:21:00.060)
the Lex Friedman face that I create,
Lex Fridman (1:21:01.860)
is an interface symbol that's a genuine portal
Donald Hoffman (1:21:04.260)
into your consciousness.
Lex Fridman (1:21:06.060)
The portal is less clear for a cat,
Donald Hoffman (1:21:09.980)
even less clear for an ant.
Lex Fridman (1:21:11.860)
And by the time we get down to a proton,
Donald Hoffman (1:21:13.380)
the portal is not clear at all.
Lex Fridman (1:21:15.900)
But that doesn't mean I'm not interacting
Donald Hoffman (1:21:17.140)
with consciousness, it just means my interface gave up.
Lex Fridman (1:21:20.140)
And there's some deeper reality that we have to go after.
Lex Fridman (1:21:23.020)
So your question really forces out a big part
Lex Fridman (1:21:26.980)
of this whole approach that I'm talking about.
Lex Fridman (1:21:29.100)
So it's this portal and consciousness.
Lex Fridman (1:21:30.580)
I wonder why you can't,
Donald Hoffman (1:21:33.300)
your portal is not as good to a cat,
Lex Fridman (1:21:36.820)
to a cat's consciousness than it is to a human.
Donald Hoffman (1:21:40.420)
Does it have to do with the fact that you're human
Lex Fridman (1:21:45.100)
and just similar organisms, organisms of similar complexity
Lex Fridman (1:21:49.820)
are able to create portals better to each other?
Lex Fridman (1:21:53.420)
Or is it just as you get more and more complex,
Lex Fridman (1:21:55.620)
you get better and better portals?
Lex Fridman (1:21:57.860)
Well, let me answer one aspect of it
Donald Hoffman (1:22:00.220)
that I'm more confident about,
Lex Fridman (1:22:01.180)
then I'll speculate on that.
Lex Fridman (1:22:03.540)
Why is it that the portal is so bad with protons?
Lex Fridman (1:22:07.180)
Well, and elementary particles more generally.
Lex Fridman (1:22:09.500)
So quarks, leptons and gluons and so forth.
Lex Fridman (1:22:12.260)
Well, the reason for that is because those are just
Donald Hoffman (1:22:16.100)
symmetries of space time.
Lex Fridman (1:22:19.220)
More technically, they're irreducible representations
Donald Hoffman (1:22:21.060)
of the Poincare group of space time.
Lex Fridman (1:22:22.940)
So they're just literally representations
Donald Hoffman (1:22:26.940)
of the data structure of space time that we're using.
Lex Fridman (1:22:30.940)
So that's why they're not very much insightful.
Donald Hoffman (1:22:33.260)
They're just almost entirely tied
Lex Fridman (1:22:35.700)
to the data structure itself.
Donald Hoffman (1:22:37.020)
There's not much,
Lex Fridman (1:22:38.940)
they're telling you only something about the data structure,
Donald Hoffman (1:22:40.980)
not behind the data structure.
Lex Fridman (1:22:42.420)
It's only when we get to higher levels
Donald Hoffman (1:22:44.180)
that we're starting to, in some sense,
Lex Fridman (1:22:46.060)
build portals to what's behind space time.
Donald Hoffman (1:22:49.820)
Sure.
Lex Fridman (1:22:50.660)
Yeah, so there's more and more complexity built
Donald Hoffman (1:22:55.940)
on top of the interface of space time with the cat.
Lex Fridman (1:22:59.940)
So you can actually build a portal, right?
Donald Hoffman (1:23:01.340)
Yeah.
Lex Fridman (1:23:02.180)
Yeah, right.
Donald Hoffman (1:23:06.700)
Yeah, this interface of face and hair and so on, skin.
Lex Fridman (1:23:14.820)
There's some syncing going on between humans though,
Donald Hoffman (1:23:18.100)
where we synced, like you're getting
Lex Fridman (1:23:21.100)
a pretty good representation of the ideas in my head
Lex Fridman (1:23:24.500)
and starting to get a foggy view of my memories in my head.
Lex Fridman (1:23:30.700)
Even though this is the first time we're talking,
Donald Hoffman (1:23:34.540)
you start to project your own memories.
Lex Fridman (1:23:36.860)
You start to solve like a giant hierarchy of puzzles
Donald Hoffman (1:23:40.820)
about a human, because we're all,
Lex Fridman (1:23:43.700)
there's a lot of similarities, a lot of it rhymes.
Lex Fridman (1:23:46.660)
So you start to make a lot of inferences
Lex Fridman (1:23:48.340)
and you build up this model of a person.
Donald Hoffman (1:23:50.900)
You have a pretty sophisticated model
Lex Fridman (1:23:52.820)
what's going on underneath.
Donald Hoffman (1:23:55.540)
Again, I just, I wonder if it's possible
Lex Fridman (1:23:59.060)
to construct these models about each other
Lex Fridman (1:24:00.980)
and nevertheless be very distant from an underlying reality.
Lex Fridman (1:24:06.940)
There's a lot of work on this.
Lex Fridman (1:24:08.260)
So there's some interesting work called signaling games
Lex Fridman (1:24:10.460)
where they look at how people can coordinate
Lex Fridman (1:24:13.860)
and come to communicate.
Lex Fridman (1:24:17.340)
There's some interesting work that was done
Donald Hoffman (1:24:19.540)
by some colleagues and friends of mine,
Lex Fridman (1:24:21.420)
Louis Narens, Natalia Komarova, and Kimberly Jamieson,
Donald Hoffman (1:24:26.740)
where they were looking at evolving color words.
Lex Fridman (1:24:32.060)
So you have a circle of colors, the color circle,
Lex Fridman (1:24:36.020)
and they wanted to see if they could get people to cooperate
Lex Fridman (1:24:39.900)
and how they carved the color circle up into a circle.
Donald Hoffman (1:24:43.300)
Two units of words.
Lex Fridman (1:24:45.940)
And so they had a game theoretic kind of thing
Donald Hoffman (1:24:49.660)
that they'd had people do.
Lex Fridman (1:24:50.700)
And what they found was that when they included,
Lex Fridman (1:24:52.940)
so most people are trichromats,
Lex Fridman (1:24:54.940)
you have three kinds of cone photoreceptors,
Lex Fridman (1:24:57.940)
but there are some, a lot of men,
Lex Fridman (1:24:59.500)
7% of men are dichromats.
Donald Hoffman (1:25:01.460)
They might be missing the red cone photoreceptor.
Lex Fridman (1:25:04.420)
They found that the dichromats had an outsized influence
Donald Hoffman (1:25:09.020)
on the final ways that the whole space of colors
Lex Fridman (1:25:12.580)
was carved up and labels attached.
Donald Hoffman (1:25:14.900)
You needed to be able to include the dichromats
Lex Fridman (1:25:17.900)
in the conversation.
Lex Fridman (1:25:18.860)
And so they had a bigger influence
Lex Fridman (1:25:20.140)
on how you made the boundaries of the language.
Lex Fridman (1:25:23.020)
And I thought that was a really interesting kind of insight
Lex Fridman (1:25:25.700)
that there's going to be, again, a game,
Donald Hoffman (1:25:27.900)
perhaps a game where evolutionary or genetic algorithm
Lex Fridman (1:25:31.460)
kind of thing that goes on in terms of learning
Donald Hoffman (1:25:34.420)
to communicate in ways that are useful.
Lex Fridman (1:25:37.900)
And so, yeah, you can use game theory to actually explore
Donald Hoffman (1:25:41.020)
that are signaling games.
Lex Fridman (1:25:42.580)
There's a lot of brilliant work on that.
Donald Hoffman (1:25:44.740)
I'm not doing it, but there's work out there.
Lex Fridman (1:25:47.620)
So if it's okay, let us tackle once more
Lex Fridman (1:25:50.820)
and perhaps several more times
Lex Fridman (1:25:52.620)
after the big topic of consciousness.
Donald Hoffman (1:25:55.580)
Okay, this very beautiful, powerful things
Lex Fridman (1:25:59.540)
that perhaps is the thing that makes us human, what is it?
Donald Hoffman (1:26:03.100)
What's the role of consciousness in,
Lex Fridman (1:26:06.140)
let's say even just the thing we've been talking about,
Donald Hoffman (1:26:08.140)
which is the formation of this interface, any kind of ways
Lex Fridman (1:26:13.820)
you want to kind of start talking about it.
Donald Hoffman (1:26:18.460)
Well, let me say first what most of my colleagues say.
Lex Fridman (1:26:22.940)
99% are, again, assuming that space time is fundamental,
Donald Hoffman (1:26:27.620)
particles and space time, matter is fundamental,
Lex Fridman (1:26:30.900)
and most are reductionist.
Lex Fridman (1:26:33.740)
And so the standard approach to consciousness
Lex Fridman (1:26:37.100)
is to figure out what complicated systems of matter
Donald Hoffman (1:26:43.900)
with the right functional properties
Lex Fridman (1:26:45.940)
could possibly lead to the emergence of consciousness.
Lex Fridman (1:26:48.540)
That's the general idea, right?
Lex Fridman (1:26:51.420)
So maybe you have to have neurons,
Donald Hoffman (1:26:53.900)
maybe only if you have neurons, but that might not be enough.
Lex Fridman (1:26:58.620)
They have to certain kinds of complexity
Donald Hoffman (1:27:00.380)
in their organization and their dynamics,
Lex Fridman (1:27:02.780)
certain kind of network abilities, for example.
Lex Fridman (1:27:05.140)
So there are those who say, for example,
Lex Fridman (1:27:10.820)
that consciousness arises from orchestrated collapse
Donald Hoffman (1:27:14.700)
of quantum states of microtubules and neurons, certainly.
Lex Fridman (1:27:18.340)
So this is Hamroff and Penrose, that's kind of.
Lex Fridman (1:27:22.460)
So you start with something physical,
Lex Fridman (1:27:25.460)
a property of quantum states of neurons,
Donald Hoffman (1:27:30.420)
of microtubules and neurons,
Lex Fridman (1:27:32.180)
and you say that somehow an orchestrated collapse
Donald Hoffman (1:27:34.860)
of those is consciousness or conscious experiences.
Lex Fridman (1:27:38.740)
Or integrated information theory.
Donald Hoffman (1:27:40.500)
Again, you start with something physical,
Lex Fridman (1:27:42.620)
and if it has the right kind of functional properties,
Donald Hoffman (1:27:44.940)
it's something they call phi,
Lex Fridman (1:27:46.300)
with the right kind of integrated information,
Donald Hoffman (1:27:48.060)
then you have consciousness.
Lex Fridman (1:27:50.620)
Or you can be a panpsychist, Philip Goff, for example,
Donald Hoffman (1:27:54.940)
where you might say, well,
Lex Fridman (1:27:57.340)
in addition to the particles and space and time,
Donald Hoffman (1:28:01.340)
those particles are not just matter,
Lex Fridman (1:28:03.420)
they also could have, say, a unit of consciousness.
Lex Fridman (1:28:06.540)
And so, but once again, you're taking space and time
Lex Fridman (1:28:09.340)
and particles as fundamental,
Lex Fridman (1:28:11.220)
and you're adding a new property to them,
Lex Fridman (1:28:14.100)
say, consciousness, and then you have to talk about how
Donald Hoffman (1:28:16.780)
when a proton and an electron get together
Lex Fridman (1:28:21.060)
to form hydrogen, then how those consciousnesses
Donald Hoffman (1:28:24.340)
merge to or interact to create the consciousness
Lex Fridman (1:28:27.340)
of hydrogen and so forth.
Donald Hoffman (1:28:30.340)
There's attention schema theory,
Lex Fridman (1:28:31.540)
which again, this is how neural network processes
Donald Hoffman (1:28:35.780)
representing to the network itself,
Lex Fridman (1:28:38.340)
its attentional processes, that could be consciousness.
Donald Hoffman (1:28:42.820)
There's global workspace theory,
Lex Fridman (1:28:45.500)
and neuronal global workspace theory.
Lex Fridman (1:28:48.420)
So there's many, many theories of this type.
Lex Fridman (1:28:50.420)
What's common to all of them is they assume
Donald Hoffman (1:28:53.900)
that space time is fundamental.
Lex Fridman (1:28:56.220)
They assume that physical processes
Lex Fridman (1:28:57.940)
and space time is fundamental.
Lex Fridman (1:28:59.620)
Panpsychism adds consciousness as an additional thing,
Donald Hoffman (1:29:02.540)
it's almost dualist in that regard.
Lex Fridman (1:29:05.940)
And my attitude is our best science is telling us
Donald Hoffman (1:29:11.220)
that space time is not fundamental.
Lex Fridman (1:29:13.300)
So why is that important here?
Donald Hoffman (1:29:17.100)
Well, for centuries, deep thinkers thought of earth, air,
Lex Fridman (1:29:23.100)
fire, and water as the fundamental elements.
Donald Hoffman (1:29:26.260)
It was a reductionist kind of idea.
Lex Fridman (1:29:28.580)
Nothing was more elemental than those,
Lex Fridman (1:29:30.100)
and you could sort of build everything up from those.
Lex Fridman (1:29:33.460)
When we got the periodic table of elements,
Donald Hoffman (1:29:37.180)
we realized that, of course,
Lex Fridman (1:29:40.020)
we want to study earth, air, fire, and water.
Donald Hoffman (1:29:42.020)
There's combustion science for fire.
Lex Fridman (1:29:44.060)
There's sciences for all these other things,
Donald Hoffman (1:29:49.020)
water and so forth.
Lex Fridman (1:29:50.300)
So we're gonna do science with these things,
Lex Fridman (1:29:51.900)
but fundamental, no, no.
Lex Fridman (1:29:54.180)
If you're looking for something fundamental,
Donald Hoffman (1:29:56.180)
those are the wrong building blocks.
Lex Fridman (1:29:58.500)
Earth has many, many different kinds of elements
Donald Hoffman (1:30:02.180)
that project into the one thing that we call earth.
Lex Fridman (1:30:04.660)
If you don't understand that there's silicon,
Donald Hoffman (1:30:06.820)
that there's iron,
Lex Fridman (1:30:07.660)
that there's all these different kinds of things
Donald Hoffman (1:30:09.020)
that project into what we call earth,
Lex Fridman (1:30:11.260)
you're hopelessly lost.
Donald Hoffman (1:30:14.700)
You're not fundamental, you're not gonna get there.
Lex Fridman (1:30:17.020)
And then after the periodic table,
Donald Hoffman (1:30:19.820)
then we came up with quarks, leptons, and gluons,
Lex Fridman (1:30:22.260)
the particles of the standard model of physics.
Lex Fridman (1:30:26.500)
And so we actually now know
Lex Fridman (1:30:29.060)
that if you really want to get fundamental,
Donald Hoffman (1:30:33.380)
the periodic table isn't it.
Lex Fridman (1:30:34.660)
It's good for chemistry,
Lex Fridman (1:30:35.660)
and it's just wonderful for chemistry,
Lex Fridman (1:30:37.220)
but if you're trying to go deep fundamental,
Lex Fridman (1:30:39.860)
what is the fundamental science?
Lex Fridman (1:30:41.500)
That's not it.
Donald Hoffman (1:30:42.340)
You're gonna have to go to quarks, leptons,
Lex Fridman (1:30:44.580)
and gluons and so forth.
Donald Hoffman (1:30:46.100)
Well, now we've discovered space time itself is doomed.
Lex Fridman (1:30:51.660)
Quarks, leptons, and gluons
Donald Hoffman (1:30:53.020)
are just irreducible representations
Lex Fridman (1:30:54.980)
of the symmetries of space time.
Lex Fridman (1:30:57.540)
So the whole framework
Lex Fridman (1:31:00.300)
on which consciousness research is being based right now
Donald Hoffman (1:31:03.860)
is doomed.
Lex Fridman (1:31:05.460)
And for me, these are my friends and colleagues
Donald Hoffman (1:31:09.220)
that are doing this, they're brilliant.
Lex Fridman (1:31:11.020)
They're absolutely, they're brilliant.
Donald Hoffman (1:31:13.060)
I, my feeling is I'm so sad
Lex Fridman (1:31:19.260)
that they're stuck with this old framework
Donald Hoffman (1:31:21.780)
because if they weren't stuck with earth, air, fire,
Lex Fridman (1:31:25.100)
and water, you could actually make progress.
Lex Fridman (1:31:27.180)
So it doesn't matter how smart you are.
Lex Fridman (1:31:28.740)
If you start with earth, air, fire, and water,
Lex Fridman (1:31:30.260)
you're not gonna get anywhere, right?
Lex Fridman (1:31:32.020)
Can I actually just,
Donald Hoffman (1:31:33.860)
because the word doomed is so interesting,
Lex Fridman (1:31:36.060)
let me give you some options, multiple choice quiz.
Donald Hoffman (1:31:40.460)
Is space time, we could say is reality
Lex Fridman (1:31:43.220)
the way we perceive it doomed,
Lex Fridman (1:31:46.500)
wrong or fake?
Lex Fridman (1:31:54.180)
Because doomed just means it could still be right
Lex Fridman (1:31:59.300)
and we're now ready to go deeper.
Lex Fridman (1:32:02.700)
It would be that.
Lex Fridman (1:32:03.860)
So it's not wrong, it's not a complete deviation
Lex Fridman (1:32:08.340)
from a journey toward the truth.
Donald Hoffman (1:32:10.620)
Right, it's like earth, air, fire, and water is not wrong.
Lex Fridman (1:32:13.860)
There is earth, air, fire, and water.
Donald Hoffman (1:32:15.700)
That's a useful framework, but it's not fundamental.
Lex Fridman (1:32:19.060)
Right, well, there's also wrong,
Donald Hoffman (1:32:20.820)
which is they used to believe, as I recently learned,
Lex Fridman (1:32:24.460)
that George Washington was the president,
Donald Hoffman (1:32:27.300)
the first president of the United States,
Lex Fridman (1:32:28.740)
was bled to death for something
Donald Hoffman (1:32:31.460)
that could have been easily treated
Lex Fridman (1:32:34.020)
because it was believed that you can get,
Donald Hoffman (1:32:36.380)
actually, I need to look into this further,
Lex Fridman (1:32:38.120)
but I guess you get toxins out or demons out.
Donald Hoffman (1:32:40.820)
I don't know what you're getting out
Lex Fridman (1:32:41.860)
with the bleeding of a person.
Lex Fridman (1:32:43.540)
So that ended up being wrong,
Lex Fridman (1:32:47.140)
but widely believed as a medical tool.
Lex Fridman (1:32:50.740)
So it's also possible that our assumption of space time
Lex Fridman (1:32:55.900)
is not just doomed, but is wrong.
Donald Hoffman (1:32:58.940)
Well, if we believe that it's fundamental, that's wrong.
Lex Fridman (1:33:02.200)
But if we believe it's a useful tool, that's right.
Lex Fridman (1:33:05.320)
But bleeding somebody to death
Lex Fridman (1:33:08.140)
was believed to be a useful tool.
Lex Fridman (1:33:10.100)
And that was wrong.
Lex Fridman (1:33:11.140)
It wasn't just not fundamental.
Donald Hoffman (1:33:13.820)
It was very, I'm sure there's cases
Lex Fridman (1:33:17.260)
in which bleeding somebody would work,
Lex Fridman (1:33:19.020)
but it would be a very tiny, tiny, tiny percentage of cases.
Lex Fridman (1:33:23.540)
So it could be that it's wrong,
Donald Hoffman (1:33:25.820)
like it's a side road that's ultimately leading
Lex Fridman (1:33:29.140)
to a dead end as opposed to a truck stop or something
Donald Hoffman (1:33:32.600)
that you can get off of.
Lex Fridman (1:33:34.700)
My feeling is not the dead end kind of thing.
Donald Hoffman (1:33:37.340)
I think that what the physicists are finding
Lex Fridman (1:33:39.460)
is that there are these structures beyond space time,
Lex Fridman (1:33:41.700)
but they project back into space time.
Lex Fridman (1:33:44.260)
And so space time, when they say space time is doomed,
Donald Hoffman (1:33:48.220)
they're explicit.
Lex Fridman (1:33:49.100)
They're saying it's doomed in the sense
Donald Hoffman (1:33:50.420)
that we thought it was fundamental.
Lex Fridman (1:33:51.780)
It's not fundamental.
Donald Hoffman (1:33:53.220)
It's a useful, absolutely useful and brilliant data structure,
Lex Fridman (1:33:57.460)
but there are deeper data structures
Donald Hoffman (1:33:59.500)
like cosmological polytope and space time is not fundamental.
Lex Fridman (1:34:03.980)
What is doomed in the sense that it's wrong
Donald Hoffman (1:34:07.620)
is reductionism.
Lex Fridman (1:34:10.580)
Which is saying space time is fundamental, essentially.
Donald Hoffman (1:34:14.020)
Right, right.
Lex Fridman (1:34:14.940)
The idea that somehow being smaller in space and time
Donald Hoffman (1:34:20.380)
or space time is a fundamental nature of reality,
Lex Fridman (1:34:23.900)
that's just wrong.
Donald Hoffman (1:34:26.160)
It turned out to be a useful heuristic
Lex Fridman (1:34:28.220)
for thermodynamics and so forth.
Lex Fridman (1:34:29.900)
And in several other places,
Lex Fridman (1:34:31.300)
reductionism has been very useful,
Lex Fridman (1:34:33.200)
but that's in some sense an artifact
Lex Fridman (1:34:36.220)
of how we use our interface.
Donald Hoffman (1:34:39.420)
Yeah, so you're saying size doesn't matter.
Lex Fridman (1:34:41.620)
Okay, this is very important for me to write down.
Donald Hoffman (1:34:44.580)
Ultimately. Ultimately, right.
Lex Fridman (1:34:46.300)
It's useful for theories like thermodynamics
Lex Fridman (1:34:49.740)
and also for understanding brain networks
Lex Fridman (1:34:51.460)
in terms of individual neurons and neurons
Donald Hoffman (1:34:54.420)
in terms of chemical systems inside cells.
Lex Fridman (1:34:58.500)
That's all very, very useful,
Lex Fridman (1:35:00.340)
but the idea that we're getting
Lex Fridman (1:35:02.380)
to the more fundamental nature of reality, no.
Donald Hoffman (1:35:05.900)
When you get all the way down in that direction,
Lex Fridman (1:35:08.260)
you get down to the quarks and gluons,
Lex Fridman (1:35:09.980)
what you realize is what you've gotten down to
Lex Fridman (1:35:11.860)
is not fundamental reality,
Donald Hoffman (1:35:13.300)
just the irreducible representations of a data structure.
Lex Fridman (1:35:16.420)
That's all you've gotten down to.
Lex Fridman (1:35:17.780)
So you're always stuck inside the data structure.
Lex Fridman (1:35:21.660)
So you seem to be getting closer and closer.
Donald Hoffman (1:35:23.620)
I went from neural networks to neurons,
Lex Fridman (1:35:25.660)
neurons to chemistry, chemistry to particles,
Donald Hoffman (1:35:27.680)
particles to quarks and gluons.
Lex Fridman (1:35:29.980)
I'm getting closer and closer to the real.
Donald Hoffman (1:35:31.500)
No, I'm getting closer and closer to the actual structure
Lex Fridman (1:35:34.300)
of the data structure of space and time,
Donald Hoffman (1:35:36.660)
the irreducible representations.
Lex Fridman (1:35:38.220)
That's what you're getting closer to,
Donald Hoffman (1:35:39.680)
not to a deeper understanding of what's beyond space time.
Lex Fridman (1:35:43.180)
We'll also refer, we'll return again
Donald Hoffman (1:35:46.300)
to this question of dynamics
Lex Fridman (1:35:48.060)
because you keep saying that space time is doomed,
Lex Fridman (1:35:51.860)
but mostly focusing on the space part of that.
Lex Fridman (1:35:54.420)
It's very interesting to see why time gets the bad cred too
Donald Hoffman (1:35:59.020)
because how do you have dynamics without time
Lex Fridman (1:36:01.060)
is the thing I'd love to talk to you a little bit about.
Lex Fridman (1:36:02.940)
But let us return your brilliant whirlwind overview
Lex Fridman (1:36:09.380)
of the different theories of consciousness
Donald Hoffman (1:36:11.440)
that are out there.
Lex Fridman (1:36:14.420)
What is consciousness if outside of space time?
Donald Hoffman (1:36:18.760)
If we think that we want to have a model of consciousness,
Lex Fridman (1:36:20.860)
we as scientists then have to say,
Lex Fridman (1:36:23.940)
what do we want to write down?
Lex Fridman (1:36:25.140)
What kind of mathematical modeling
Lex Fridman (1:36:26.940)
are we gonna write down, right?
Lex Fridman (1:36:28.700)
And if you think about it, there's lots of things
Donald Hoffman (1:36:30.180)
that you might want to write down about consciousness.
Lex Fridman (1:36:32.580)
For all the complicated subject.
Lex Fridman (1:36:35.620)
So most of my colleagues are saying,
Lex Fridman (1:36:36.880)
let's start with matter or neurons
Lex Fridman (1:36:38.380)
and see what properties of matter
Lex Fridman (1:36:40.940)
could create consciousness.
Lex Fridman (1:36:42.500)
But I'm saying that that whole thing is out.
Lex Fridman (1:36:45.200)
Space time is doomed, that whole thing is out.
Donald Hoffman (1:36:47.740)
We need to look at consciousness qua consciousness.
Lex Fridman (1:36:51.420)
In other words, not as something that arises
Donald Hoffman (1:36:53.300)
in space and time, but perhaps as something
Lex Fridman (1:36:55.220)
that creates space and time as a data structure.
Lex Fridman (1:36:58.260)
So what do we want?
Lex Fridman (1:36:59.440)
And here again, there's no hard and fast rule,
Lex Fridman (1:37:02.020)
but what you as a scientist have to do
Lex Fridman (1:37:03.820)
is to pick what you think are the minimal assumptions
Donald Hoffman (1:37:09.740)
that are gonna allow you to boot up a comprehensive theory.
Lex Fridman (1:37:13.740)
That is the trick.
Lex Fridman (1:37:16.420)
So what do I want?
Lex Fridman (1:37:17.360)
So what I chose to do was to have three things.
Donald Hoffman (1:37:23.060)
I said that there are conscious experiences.
Lex Fridman (1:37:26.260)
Feeling of headache, the smell of garlic,
Donald Hoffman (1:37:28.960)
experiencing the color red.
Lex Fridman (1:37:32.240)
There are, those are conscious,
Lex Fridman (1:37:33.640)
so that's the primitive of the theory.
Lex Fridman (1:37:34.880)
And the reason I want few primitives, why?
Lex Fridman (1:37:36.820)
Because those are the miracles of the theory, right?
Lex Fridman (1:37:38.800)
The primitives, the assumptions of the theory
Donald Hoffman (1:37:40.720)
are the things you're not going to explain.
Lex Fridman (1:37:42.400)
Those are the things you assume.
Lex Fridman (1:37:43.980)
And those experiences, you particularly mean
Lex Fridman (1:37:46.320)
there's a subjectiveness to them.
Donald Hoffman (1:37:49.600)
That's the thing when people refer
Lex Fridman (1:37:51.680)
to the hard problem of consciousness,
Donald Hoffman (1:37:54.040)
is it feels like something to look at the color red, okay.
Lex Fridman (1:37:58.160)
Exactly right, it feels like something to have a headache
Donald Hoffman (1:38:00.400)
or to feel upset to your stomach.
Lex Fridman (1:38:02.760)
It feels like something.
Lex Fridman (1:38:04.520)
And so I'm going to grant that in this theory,
Lex Fridman (1:38:09.560)
there are experiences and they're fundamental in some sense.
Lex Fridman (1:38:12.480)
So conscious experience.
Lex Fridman (1:38:13.840)
So they're not derived from physics.
Donald Hoffman (1:38:15.720)
They're not functional properties of particles.
Lex Fridman (1:38:18.780)
They are sui generis, they exist.
Donald Hoffman (1:38:21.800)
Just like we assume space time exists.
Lex Fridman (1:38:23.760)
I'm now saying space time is just a data structure.
Donald Hoffman (1:38:26.320)
It doesn't exist independent of conscious experiences.
Lex Fridman (1:38:29.520)
Sorry to interrupt once again,
Lex Fridman (1:38:30.800)
but should we be focusing in your thinking on humans alone?
Lex Fridman (1:38:35.760)
Or is there something about in relation
Donald Hoffman (1:38:40.640)
to other kinds of organisms that have
Lex Fridman (1:38:42.560)
a sufficiently high level of complexity?
Donald Hoffman (1:38:44.640)
Or even, or is there some kind of generalization
Lex Fridman (1:38:50.320)
of the panpsychist idea that all consciousness permeates,
Lex Fridman (1:38:54.380)
all matter?
Lex Fridman (1:38:55.760)
Outside of the usual definition
Donald Hoffman (1:38:58.720)
of what matter is inside space time.
Lex Fridman (1:39:01.220)
So it's beyond human consciousness.
Donald Hoffman (1:39:04.320)
Human consciousness, from my point of view,
Lex Fridman (1:39:06.380)
would be one of a countless variety of consciousnesses.
Lex Fridman (1:39:10.520)
And even within human consciousness,
Lex Fridman (1:39:12.400)
there's countless variety of consciousnesses within us.
Donald Hoffman (1:39:15.840)
I mean, you have your left and right hemisphere.
Lex Fridman (1:39:18.460)
And apparently if you split the corpus callosum,
Donald Hoffman (1:39:20.760)
the personality of the left hemisphere
Lex Fridman (1:39:22.800)
and the religious beliefs of the left hemisphere
Donald Hoffman (1:39:24.400)
can be very different from the right hemisphere.
Lex Fridman (1:39:26.400)
And their conscious experiences can be disjoint.
Donald Hoffman (1:39:30.680)
One could have one conscious experience.
Lex Fridman (1:39:32.520)
They can play 20 questions.
Donald Hoffman (1:39:33.880)
The left hemisphere can have an idea in its mind
Lex Fridman (1:39:35.880)
and the right hemisphere has to guess.
Lex Fridman (1:39:37.400)
And it might not get it.
Lex Fridman (1:39:38.920)
So even within you,
Donald Hoffman (1:39:40.700)
there is more than just one consciousness.
Lex Fridman (1:39:43.080)
It's lots of consciousnesses.
Lex Fridman (1:39:45.240)
So the general theory of consciousness that I'm after
Lex Fridman (1:39:48.720)
is not just human consciousness.
Donald Hoffman (1:39:50.380)
It's going to be just consciousness.
Lex Fridman (1:39:51.940)
And I presume human consciousness is a tiny drop
Donald Hoffman (1:39:56.720)
in the bucket of the infinite variety of consciousnesses.
Lex Fridman (1:39:59.320)
That said, I should clarify that the black hole
Donald Hoffman (1:40:02.680)
of consciousness is the home cat.
Lex Fridman (1:40:07.400)
I'm pretty sure cats lack, is the embodiment of evil
Lex Fridman (1:40:11.840)
and lack all capacity for consciousness or compassion.
Lex Fridman (1:40:16.000)
So I just want to lay that on the table.
Donald Hoffman (1:40:17.120)
That's the theory I'm working on.
Lex Fridman (1:40:17.960)
I don't have any good evidence, but it's just an intuition.
Donald Hoffman (1:40:20.880)
It's just a shout out.
Lex Fridman (1:40:23.480)
Sorry to distract.
Lex Fridman (1:40:24.320)
So that's the first assumption.
Lex Fridman (1:40:25.720)
The first assumption, that's right.
Donald Hoffman (1:40:27.200)
The second assumption is that
Lex Fridman (1:40:29.200)
these experiences have consequences.
Lex Fridman (1:40:31.680)
So I'm going to say that conscious experiences
Lex Fridman (1:40:35.040)
can trigger other conscious experiences somehow.
Lex Fridman (1:40:38.360)
So really in some sense, there's two basic assumptions.
Lex Fridman (1:40:43.760)
There's some kind of causality.
Lex Fridman (1:40:46.200)
Is there a chain of causality?
Lex Fridman (1:40:47.800)
Does this relate to dynamics?
Donald Hoffman (1:40:50.320)
I'll say there's a probabilistic relationship.
Lex Fridman (1:40:55.040)
So I'm trying to be as nonspecific to begin with
Lex Fridman (1:40:58.840)
and see where it leads me.
Lex Fridman (1:41:01.080)
So what I can write down are probability spaces.
Lex Fridman (1:41:04.000)
So a probability space, which contains
Lex Fridman (1:41:06.640)
the conscious experiences that this consciousness can have.
Lex Fridman (1:41:09.560)
So I call this a conscious agent, this technical thing.
Lex Fridman (1:41:16.960)
Annika Harris and I've talked about this
Lex Fridman (1:41:19.320)
and she rightly cautions me that people will think
Lex Fridman (1:41:22.880)
that I'm bringing in a notion of a self or agency
Lex Fridman (1:41:25.120)
and so forth when I say conscious agent.
Lex Fridman (1:41:27.440)
So I just want to say that I use the term conscious agent
Donald Hoffman (1:41:30.000)
merely as a technical term.
Lex Fridman (1:41:32.280)
There is no notion of self in my fundamental definition
Donald Hoffman (1:41:35.640)
of a conscious agent.
Lex Fridman (1:41:36.640)
There are only experiences and probabilistic relationships
Donald Hoffman (1:41:41.080)
of how they trigger other experiences.
Lex Fridman (1:41:43.000)
So the agent is the generator of the conscious experience?
Donald Hoffman (1:41:46.280)
The agent is a mathematical structure
Lex Fridman (1:41:49.560)
that includes a probability measure,
Donald Hoffman (1:41:51.920)
the probability space of a possible conscious experiences
Lex Fridman (1:41:56.080)
and a Markovian kernel, which describes how
Donald Hoffman (1:42:00.080)
if this agent has certain conscious experiences,
Lex Fridman (1:42:02.800)
how that will affect the experiences
Donald Hoffman (1:42:04.280)
of other conscious agents, including itself.
Lex Fridman (1:42:07.520)
But you don't think of that as a self?
Donald Hoffman (1:42:09.920)
No, there is no notion of a self here.
Lex Fridman (1:42:13.720)
There's no notion of really of an agent.
Lex Fridman (1:42:17.520)
But is there a locality?
Lex Fridman (1:42:20.320)
Is there an organism?
Donald Hoffman (1:42:21.160)
There's no space.
Lex Fridman (1:42:22.000)
There's no.
Lex Fridman (1:42:22.880)
So this is, these are conscious units, conscious entities.
Lex Fridman (1:42:28.280)
But they're distinct in some way
Donald Hoffman (1:42:30.440)
because they have to interact.
Lex Fridman (1:42:32.280)
Well, so here's the interesting thing.
Donald Hoffman (1:42:33.640)
When we write down the mathematics,
Lex Fridman (1:42:36.080)
when you have two of these conscious agents interacting,
Donald Hoffman (1:42:39.280)
the pair satisfy a definition of a conscious agent.
Lex Fridman (1:42:43.720)
So they are a single conscious agent.
Lex Fridman (1:42:46.160)
So there is one conscious agent.
Lex Fridman (1:42:48.520)
But it has a nice analytic decomposition
Donald Hoffman (1:42:52.360)
into as many conscious agents as you wish.
Lex Fridman (1:42:53.200)
So that's a nice interface.
Donald Hoffman (1:42:55.520)
It's a very useful scientific interface.
Lex Fridman (1:42:58.720)
It's a scale free or if you like a fractal like approach
Donald Hoffman (1:43:03.200)
to it in which we can use the same unit of analysis
Lex Fridman (1:43:06.320)
at all scales in studying consciousness.
Lex Fridman (1:43:09.760)
But if I want to talk about,
Lex Fridman (1:43:12.240)
so there's no notion of learning, memory, problem solving,
Donald Hoffman (1:43:17.160)
intelligence, self, agency.
Lex Fridman (1:43:20.800)
So none of that is fundamental.
Donald Hoffman (1:43:24.160)
So, and the reason I did that was
Lex Fridman (1:43:26.360)
because I want to assume as little as possible.
Donald Hoffman (1:43:29.880)
Everything I assume is a miracle in the theory.
Lex Fridman (1:43:32.160)
It's not something you explain, it's something you assume.
Lex Fridman (1:43:34.520)
So I have to build networks of conscious agents.
Lex Fridman (1:43:38.760)
If I want to have a notion of a self,
Donald Hoffman (1:43:40.000)
I have to build a self.
Lex Fridman (1:43:41.520)
I have to build learning, memory, problem solving,
Donald Hoffman (1:43:43.920)
intelligence and planning, all these different things.
Lex Fridman (1:43:46.720)
I have to build networks of conscious agents to do that.
Donald Hoffman (1:43:49.480)
It's a trivial theorem that networks of conscious agents
Lex Fridman (1:43:52.400)
are computationally universal, that's trivial.
Lex Fridman (1:43:54.680)
So anything that we can do with neural networks
Lex Fridman (1:43:56.800)
or automata, you can do with networks of conscious agents.
Donald Hoffman (1:44:00.480)
That's trivial.
Lex Fridman (1:44:01.320)
But you can also do more.
Donald Hoffman (1:44:04.320)
The events in the probability space need not be computable.
Lex Fridman (1:44:08.200)
So the Markovian dynamics is not restricted
Donald Hoffman (1:44:11.800)
to computable functions
Lex Fridman (1:44:14.440)
because the very events themselves need not be computable.
Lex Fridman (1:44:17.280)
So this can capture any computable theory.
Lex Fridman (1:44:20.640)
Anything we can do with neural networks,
Donald Hoffman (1:44:22.000)
we can do with conscious agent networks.
Lex Fridman (1:44:24.640)
But it leaves open the door for the possibility
Donald Hoffman (1:44:27.520)
of noncomputable interactions between conscious agents.
Lex Fridman (1:44:31.960)
So if we want a theory of memory, we have to build it.
Lex Fridman (1:44:37.520)
And there's lots of different ways you could build.
Lex Fridman (1:44:39.080)
We've actually got a paper,
Donald Hoffman (1:44:40.000)
Chris Fields took the lead on this.
Lex Fridman (1:44:41.720)
And we have a paper called Conscious Agent Networks
Donald Hoffman (1:44:44.640)
where Chris takes the lead and shows how to use
Lex Fridman (1:44:47.160)
these networks of conscious agents to build memory
Lex Fridman (1:44:49.080)
and to build primitive kinds of learning.
Lex Fridman (1:44:53.280)
But can you provide some intuition
Donald Hoffman (1:44:56.160)
of what conscious networks,
Lex Fridman (1:44:58.880)
networks of conscious agents helps you?
Donald Hoffman (1:45:04.080)
First of all, what that looks like.
Lex Fridman (1:45:07.200)
And I don't just mean mathematically.
Donald Hoffman (1:45:08.920)
Of course, maybe that might help build up intuition.
Lex Fridman (1:45:11.600)
But how that helps us potentially solve
Donald Hoffman (1:45:14.000)
the hard problem of consciousness.
Lex Fridman (1:45:17.680)
Or is that baked in, that that exists?
Lex Fridman (1:45:21.640)
Can you solve the hard problem of consciousness,
Lex Fridman (1:45:27.120)
why it tastes delicious when you eat a delicious ice cream
Lex Fridman (1:45:31.640)
with networks of conscious agents?
Lex Fridman (1:45:33.840)
Or is that taken as an assumption?
Lex Fridman (1:45:36.040)
So the standard way the hard problem is thought of
Lex Fridman (1:45:40.080)
is we're assuming space and time and particles
Donald Hoffman (1:45:44.600)
or neurons, for example.
Lex Fridman (1:45:47.120)
These are just physical things that have no consciousness.
Lex Fridman (1:45:50.000)
And we have to explain how the conscious experience
Lex Fridman (1:45:51.760)
of the taste of chocolate could emerge from those.
Lex Fridman (1:45:54.920)
So the typical hard problem of consciousness
Lex Fridman (1:45:57.080)
is that problem, right?
Lex Fridman (1:45:58.720)
How do you boot up the taste of chocolate,
Lex Fridman (1:46:02.040)
the experience of the taste of chocolate from neurons, say,
Lex Fridman (1:46:06.280)
or the right kind of artificial intelligence circuitry?
Lex Fridman (1:46:10.560)
How do you boot that up?
Donald Hoffman (1:46:11.640)
That's typically what the hard problem of consciousness
Lex Fridman (1:46:14.320)
means to researchers.
Donald Hoffman (1:46:15.800)
Notice that I'm changing the problem.
Lex Fridman (1:46:18.360)
I'm not trying to boot up conscious experiences
Donald Hoffman (1:46:21.360)
from the dynamics of neurons or silicon
Lex Fridman (1:46:23.800)
or something like that.
Donald Hoffman (1:46:25.000)
I'm saying that that's the wrong problem.
Lex Fridman (1:46:27.840)
My hard problem would go in the other direction.
Donald Hoffman (1:46:29.920)
If I start with conscious experiences,
Lex Fridman (1:46:33.480)
how do I build up space and time?
Lex Fridman (1:46:35.640)
How do I build up what I call the physical world?
Lex Fridman (1:46:37.560)
How do I build up what we call brains?
Donald Hoffman (1:46:40.560)
Because I'm saying consciousness
Lex Fridman (1:46:43.080)
is not something that brains do.
Donald Hoffman (1:46:45.640)
Brains are something that consciousness makes up.
Lex Fridman (1:46:49.440)
It's among the experience,
Donald Hoffman (1:46:50.840)
it's an ephemeral experience in consciousness.
Lex Fridman (1:46:54.560)
I look inside, so to be very, very clear,
Donald Hoffman (1:46:57.040)
right now, I have no neurons.
Lex Fridman (1:46:59.920)
If you looked, you would see neurons.
Donald Hoffman (1:47:03.200)
That's a data structure that you would create on the fly,
Lex Fridman (1:47:05.320)
and it's a very useful one.
Donald Hoffman (1:47:06.440)
As soon as you look away,
Lex Fridman (1:47:08.560)
you garbage collect that data structure,
Donald Hoffman (1:47:10.000)
just like that Necker cube that I was talking about
Lex Fridman (1:47:11.800)
on the piece of paper.
Donald Hoffman (1:47:12.960)
When you look, you see a 3D cube you created on the fly.
Lex Fridman (1:47:17.360)
As soon as you look away, that's gone.
Donald Hoffman (1:47:19.600)
When you say you, you mean a human being scientist.
Lex Fridman (1:47:22.920)
Right now, that's right.
Donald Hoffman (1:47:24.600)
More generally, it'll be conscious agents,
Lex Fridman (1:47:26.960)
because as you pointed out,
Donald Hoffman (1:47:28.400)
am I asking for a theory of consciousness
Lex Fridman (1:47:30.920)
only about humans?
Donald Hoffman (1:47:31.760)
No, it's consciousness,
Lex Fridman (1:47:33.920)
which human consciousness is just a tiny sliver.
Lex Fridman (1:47:38.240)
But you are saying that there is,
Lex Fridman (1:47:40.160)
that's a useful data structure.
Lex Fridman (1:47:41.840)
How many other data structures are there?
Lex Fridman (1:47:43.760)
That's why I said you human.
Donald Hoffman (1:47:45.480)
If there's another Earth,
Lex Fridman (1:47:47.480)
if there's another alien civilization
Lex Fridman (1:47:49.640)
and doing these kinds of investigations,
Lex Fridman (1:47:51.640)
would they come up with similar data structures?
Donald Hoffman (1:47:54.280)
Probably not.
Lex Fridman (1:47:55.120)
What is the space of data structures,
Donald Hoffman (1:47:56.560)
I guess is what I'm asking.
Lex Fridman (1:48:00.200)
My guess is that if consciousness is fundamental,
Donald Hoffman (1:48:04.000)
consciousness is all there is,
Lex Fridman (1:48:07.200)
then the only thing that mathematical structure
Donald Hoffman (1:48:10.440)
can be about is possibilities of consciousness.
Lex Fridman (1:48:15.600)
And that suggests to me
Donald Hoffman (1:48:17.880)
that there could be an infinite variety of consciousnesses,
Lex Fridman (1:48:21.280)
and a vanishingly small fraction of them
Donald Hoffman (1:48:25.120)
use space time data structures
Lex Fridman (1:48:27.800)
and the kinds of structures that we use.
Donald Hoffman (1:48:29.520)
There's an infinite variety of data structures.
Lex Fridman (1:48:32.040)
Now, this is very similar
Donald Hoffman (1:48:33.120)
to something that Max Tegmark has said,
Lex Fridman (1:48:35.000)
but I want to distinguish it.
Donald Hoffman (1:48:36.040)
He has this level four multiverse idea.
Lex Fridman (1:48:40.440)
He thinks that mathematics is fundamental.
Lex Fridman (1:48:43.040)
And so that's the fundamental reality.
Lex Fridman (1:48:45.000)
And since there's an infinite variety of,
Donald Hoffman (1:48:47.360)
endless variety of mathematical structures,
Lex Fridman (1:48:49.000)
there's an infinite variety of multiverses in his view.
Donald Hoffman (1:48:52.560)
I'm saying something similar in spirit,
Lex Fridman (1:48:55.280)
but importantly different.
Donald Hoffman (1:48:56.960)
There's an infinite variety
Lex Fridman (1:48:57.800)
of mathematical structures, absolutely.
Lex Fridman (1:49:00.760)
But mathematics isn't the fundamental reality
Lex Fridman (1:49:03.160)
in this framework.
Donald Hoffman (1:49:04.960)
Consciousness is,
Lex Fridman (1:49:06.680)
and mathematics is to consciousness
Donald Hoffman (1:49:09.360)
like bones are to an organism.
Lex Fridman (1:49:12.000)
You need the bones.
Lex Fridman (1:49:12.880)
So mathematics is not divorced from consciousness,
Lex Fridman (1:49:16.480)
but it's not the entirety of consciousness by any means.
Lex Fridman (1:49:20.280)
And so there's an infinite variety of consciousnesses
Lex Fridman (1:49:24.440)
and signaling games that consciousnesses could interact via.
Lex Fridman (1:49:30.120)
And therefore worlds, your common worlds,
Lex Fridman (1:49:32.560)
data structures that they can use to communicate.
Lex Fridman (1:49:37.200)
So space and time is just one of an infinite variety.
Lex Fridman (1:49:40.200)
And so I think that what we'll find is that
Donald Hoffman (1:49:43.600)
as we go outside of our little space time bubble,
Lex Fridman (1:49:48.880)
we will encounter utterly alien forms
Donald Hoffman (1:49:51.760)
of conscious experience that we may not be able
Lex Fridman (1:49:54.920)
to really comprehend in the following sense.
Donald Hoffman (1:49:59.920)
If I ask you to imagine a color
Lex Fridman (1:50:03.120)
that you've never seen before,
Lex Fridman (1:50:04.480)
does anything happen?
Lex Fridman (1:50:06.480)
Nothing happens.
Donald Hoffman (1:50:09.080)
Nothing happens.
Lex Fridman (1:50:11.120)
And that's just one color.
Donald Hoffman (1:50:12.960)
I'm asking for just a color.
Lex Fridman (1:50:14.640)
We actually know, by the way,
Donald Hoffman (1:50:16.320)
that apparently there are women called tetraphams
Lex Fridman (1:50:21.520)
who have four color receptors, not just three.
Lex Fridman (1:50:25.600)
And Kimberly Jameson and others who've studied these women
Lex Fridman (1:50:28.840)
have good evidence that they apparently have
Donald Hoffman (1:50:31.080)
a new dimension of color experience
Lex Fridman (1:50:34.280)
that the rest of us don't have.
Lex Fridman (1:50:35.840)
So these women are apparently living in a world of color
Lex Fridman (1:50:40.240)
that you and I can't even concretely imagine.
Donald Hoffman (1:50:42.280)
No man can imagine them.
Lex Fridman (1:50:43.760)
Yeah.
Lex Fridman (1:50:44.600)
And yet they're real color experiences.
Lex Fridman (1:50:46.840)
And so in that sense, I'm saying,
Donald Hoffman (1:50:48.840)
now take that little baby step,
Lex Fridman (1:50:50.960)
oh, there are women who have color experiences
Donald Hoffman (1:50:52.960)
that I could never have.
Lex Fridman (1:50:53.800)
Well, that's shocking.
Donald Hoffman (1:50:55.000)
Now take that infinite.
Lex Fridman (1:50:57.520)
There are consciousnesses where every aspect
Donald Hoffman (1:51:00.600)
of their experiences is like that new color.
Lex Fridman (1:51:03.440)
It's something utterly alien to you.
Donald Hoffman (1:51:05.760)
You have nothing like that.
Lex Fridman (1:51:07.840)
And yet these are all possible varieties
Donald Hoffman (1:51:10.040)
of conscious experience.
Lex Fridman (1:51:11.320)
And when you say there's a lot of consciousnesses,
Donald Hoffman (1:51:13.360)
as a singular consciousness,
Lex Fridman (1:51:16.400)
basically the set of possible experiences you can have
Donald Hoffman (1:51:19.960)
in that subjective way,
Lex Fridman (1:51:22.360)
as opposed to the underlying mechanism.
Donald Hoffman (1:51:25.040)
Because you say that, you know,
Lex Fridman (1:51:28.000)
having a extra color receptor,
Donald Hoffman (1:51:32.440)
ability to have new experiences
Lex Fridman (1:51:34.080)
that somehow a different consciousness,
Donald Hoffman (1:51:36.640)
is there a way to see that as all the same consciousness,
Lex Fridman (1:51:39.680)
the subjectivity itself?
Donald Hoffman (1:51:41.480)
Right.
Lex Fridman (1:51:42.320)
Because when we have two of these conscious agents
Donald Hoffman (1:51:45.200)
interacting, the mathematics,
Lex Fridman (1:51:46.720)
they actually satisfy the definition of a conscious agent.
Lex Fridman (1:51:49.560)
So in fact, they are a single conscious agent.
Lex Fridman (1:51:52.240)
So in fact, one way to think about what I'm saying,
Donald Hoffman (1:51:55.480)
I'm postulating with my colleagues,
Lex Fridman (1:51:57.360)
Chaiton and Chris and others,
Donald Hoffman (1:51:58.800)
Robert Pretner and so forth.
Lex Fridman (1:52:01.920)
There is one big conscious agent, infinitely complicated.
Lex Fridman (1:52:05.320)
But fortunately, we can, for analytic purposes,
Lex Fridman (1:52:08.320)
break it down all the way to,
Donald Hoffman (1:52:10.160)
in some sense, the simplest conscious agent,
Lex Fridman (1:52:11.760)
which has one conscious experience, one.
Donald Hoffman (1:52:15.440)
This one agent can experience red 35, and that's it.
Lex Fridman (1:52:18.360)
That's what it experiences.
Donald Hoffman (1:52:20.280)
You can get all the way down to that.
Lex Fridman (1:52:22.840)
So you think it's possible that consciousness,
Donald Hoffman (1:52:27.720)
whatever that is,
Lex Fridman (1:52:30.320)
is much more, is fundamental,
Donald Hoffman (1:52:34.040)
or at least much more in the direction of the fundamental
Lex Fridman (1:52:37.520)
than is space time as we perceive it?
Donald Hoffman (1:52:40.520)
That's the proposal.
Lex Fridman (1:52:42.520)
And therefore, what I have to do,
Donald Hoffman (1:52:45.040)
in terms of the hard problem of consciousness,
Lex Fridman (1:52:47.560)
is to show how dynamical systems of conscious agents
Donald Hoffman (1:52:51.680)
could lead to what we call space and time
Lex Fridman (1:52:54.000)
and neurons and brain activity.
Donald Hoffman (1:52:56.720)
In other words, we have to show how you get space time
Lex Fridman (1:53:00.400)
and physical objects entirely from a theory
Donald Hoffman (1:53:05.440)
of conscious agents outside of space time,
Lex Fridman (1:53:07.720)
with the dynamics outside of space time.
Lex Fridman (1:53:10.560)
So that's, and I can tell you how we plan to do that,
Lex Fridman (1:53:13.960)
but that's the idea.
Donald Hoffman (1:53:15.800)
Okay, the magic of it, that chocolate is delicious.
Lex Fridman (1:53:19.800)
So there's a mathematical kind of thing
Donald Hoffman (1:53:22.760)
that we could say here, how it can emerge
Lex Fridman (1:53:24.480)
within this system of networks of conscious agents,
Lex Fridman (1:53:27.800)
but is there going to be at the end of the proof
Lex Fridman (1:53:34.120)
why chocolate is so delicious?
Lex Fridman (1:53:36.760)
Or no?
Lex Fridman (1:53:38.280)
I guess I'm going to ask different kinds of dumb questions
Donald Hoffman (1:53:41.960)
to try to sneak up.
Lex Fridman (1:53:43.200)
Oh, well, that's the right question, and when I say
Donald Hoffman (1:53:45.760)
that I took conscious experiences as fundamental,
Lex Fridman (1:53:48.080)
what that means is, in the current version of my theory,
Donald Hoffman (1:53:51.560)
I'm not explaining conscious experiences
Lex Fridman (1:53:54.120)
where they came from.
Donald Hoffman (1:53:55.720)
That's the miracle, that's one of the miracles.
Lex Fridman (1:53:58.160)
So I have two miracles in my theory.
Donald Hoffman (1:53:59.600)
There are conscious experiences, like the taste of chocolate,
Lex Fridman (1:54:02.320)
and that there's a probabilistic relationship.
Donald Hoffman (1:54:06.200)
When certain conscious experiences occur,
Lex Fridman (1:54:08.680)
others are more likely to occur.
Donald Hoffman (1:54:10.560)
Those are the two miracles that are possible.
Lex Fridman (1:54:12.680)
Is it possible to get beyond that
Lex Fridman (1:54:17.600)
and somehow start to chip away
Lex Fridman (1:54:19.600)
at the miracleness of that miracle,
Lex Fridman (1:54:22.440)
that chocolate is delicious?
Lex Fridman (1:54:24.720)
I hope so.
Donald Hoffman (1:54:25.560)
I've got my hands full with what I'm doing right now,
Lex Fridman (1:54:27.680)
but I can just say at top level how I would think about that.
Donald Hoffman (1:54:32.240)
That would get at this
Lex Fridman (1:54:38.000)
consciousness without form.
Donald Hoffman (1:54:40.160)
This is really tough, because it's consciousness without form
Lex Fridman (1:54:46.800)
versus the various forms that consciousness takes
Donald Hoffman (1:54:50.200)
for the experiences that it has.
Lex Fridman (1:54:53.440)
Right, right.
Lex Fridman (1:54:55.080)
So when I write down a probability space
Lex Fridman (1:55:01.320)
for these conscious experiences, I say,
Donald Hoffman (1:55:03.320)
here's a probability space
Lex Fridman (1:55:04.280)
for the possible conscious experiences, right?
Donald Hoffman (1:55:07.080)
It's just like when I write down a probability space
Lex Fridman (1:55:08.760)
for an experiment.
Lex Fridman (1:55:09.640)
Like I'm gonna flip a coin twice, right?
Lex Fridman (1:55:12.520)
And I want to look at the probabilities of various outcomes.
Lex Fridman (1:55:15.240)
So I have to write down a probability space.
Lex Fridman (1:55:16.400)
There could be heads, heads, heads, tails,
Donald Hoffman (1:55:18.640)
tails, heads, tails, tails.
Lex Fridman (1:55:20.440)
So any class of probability you're told,
Donald Hoffman (1:55:24.080)
write down your probability space.
Lex Fridman (1:55:25.400)
If you don't write down your probability space,
Donald Hoffman (1:55:26.760)
you can't get started.
Lex Fridman (1:55:28.200)
So here's my probability space for consciousness.
Lex Fridman (1:55:30.680)
How do I want to interpret that structure?
Lex Fridman (1:55:33.280)
The structure is just sitting there.
Lex Fridman (1:55:34.600)
There's gonna be a dynamics that happens on it, right?
Lex Fridman (1:55:37.560)
Experiences appear and then they disappear,
Donald Hoffman (1:55:39.640)
just like heads appears and disappears.
Lex Fridman (1:55:42.000)
So one way to think about that fundamental
Donald Hoffman (1:55:46.380)
probability space is that corresponds
Lex Fridman (1:55:49.480)
to consciousness without any content.
Donald Hoffman (1:55:53.080)
The infinite consciousness that transcends
Lex Fridman (1:55:57.120)
any particular content.
Donald Hoffman (1:55:58.800)
Well, do you think of that as a mechanism,
Lex Fridman (1:56:00.600)
as a thing, like the rules that govern the dynamics
Lex Fridman (1:56:05.520)
of the thing outside of space time?
Lex Fridman (1:56:08.560)
Isn't that, if you think consciousness is fundamental,
Donald Hoffman (1:56:10.720)
isn't that essentially getting like,
Lex Fridman (1:56:12.760)
it is solving the hard problem,
Donald Hoffman (1:56:14.920)
which is like from where does this thing pop up,
Lex Fridman (1:56:21.000)
which is the mechanism of the thing popping up,
Donald Hoffman (1:56:24.320)
whatever the consciousness is,
Lex Fridman (1:56:25.800)
the different kinds and so on, that mechanism.
Lex Fridman (1:56:29.680)
And also, the question I want to ask is how tricky
Lex Fridman (1:56:34.200)
do you think it is to solve that problem?
Donald Hoffman (1:56:38.280)
You've solved a lot of difficult problems
Lex Fridman (1:56:40.280)
throughout the history of humanity.
Donald Hoffman (1:56:42.960)
There's probably more problems to solve left
Lex Fridman (1:56:47.600)
than we've solved by like an infinity.
Lex Fridman (1:56:52.900)
But along that long journey of intelligent species,
Lex Fridman (1:56:58.640)
when will we solve this consciousness one?
Donald Hoffman (1:57:01.080)
Which is one way to measure the difficulty of the problem.
Lex Fridman (1:57:04.020)
So I'll give two answers.
Donald Hoffman (1:57:05.560)
There's one problem I think we can solve,
Lex Fridman (1:57:08.680)
but we haven't solved yet.
Lex Fridman (1:57:09.880)
And that is the reverse
Lex Fridman (1:57:11.160)
of what my colleagues call the hard problem.
Donald Hoffman (1:57:14.900)
The problem of how do you start with conscious experiences
Lex Fridman (1:57:17.180)
in the way that I've just described them and the dynamics
Lex Fridman (1:57:19.680)
and build up space and time and brains,
Lex Fridman (1:57:22.920)
that I think is a tough technical problem,
Lex Fridman (1:57:25.120)
but it's in principle solvable.
Lex Fridman (1:57:26.240)
So I think we can solve that.
Lex Fridman (1:57:27.760)
So we would solve the hard problem,
Lex Fridman (1:57:29.400)
not by showing how brains create consciousness,
Lex Fridman (1:57:31.280)
but how networks of conscious agents
Lex Fridman (1:57:33.080)
create what we call the symbols that we call brains.
Lex Fridman (1:57:38.520)
So that I think.
Lex Fridman (1:57:40.040)
But does that allow you to, so that's interesting.
Donald Hoffman (1:57:42.160)
That's an interesting idea.
Lex Fridman (1:57:43.480)
Consciousness creates the brain,
Donald Hoffman (1:57:44.720)
not the brain creates consciousness.
Lex Fridman (1:57:46.420)
But does that allow you to build the thing?
Donald Hoffman (1:57:49.440)
My guess is that it will enable unbelievable technologies.
Lex Fridman (1:57:53.800)
Once, and I'll tell you why.
Donald Hoffman (1:57:55.880)
I think it plugs into the work
Lex Fridman (1:57:57.760)
that the physicists are doing.
Lex Fridman (1:57:58.880)
So this theory of consciousness will be even deeper
Lex Fridman (1:58:01.920)
than the structures that the physicists are finding,
Donald Hoffman (1:58:03.960)
like the amplituhedron.
Lex Fridman (1:58:06.200)
But the other answer to your question is less positive.
Donald Hoffman (1:58:10.960)
As I said earlier, I think that there is no such thing
Lex Fridman (1:58:13.080)
as a theory of everything.
Lex Fridman (1:58:15.800)
So that I think that the theory that my team is working on,
Lex Fridman (1:58:21.720)
this conscious agent theory, is just a 1.0 theory.
Donald Hoffman (1:58:26.600)
We're using probability spaces and Markovian curls.
Lex Fridman (1:58:29.920)
I can easily see people now saying,
Donald Hoffman (1:58:31.800)
well, we can do better if we go to category theory.
Lex Fridman (1:58:35.040)
And we can get a deeper, perhaps more interesting.
Lex Fridman (1:58:38.840)
And then someone will say,
Lex Fridman (1:58:39.780)
well, now I'll go to topoi theory.
Lex Fridman (1:58:42.440)
So I imagine that there'll be conscious agents,
Lex Fridman (1:58:45.600)
five, 10, 3 trillion, 0.0, but I think it will never end.
Donald Hoffman (1:58:51.760)
I think ultimately this question
Lex Fridman (1:58:54.360)
that we sort of put our fingers on of,
Lex Fridman (1:58:56.680)
how does the formless give birth to form,
Lex Fridman (1:59:03.080)
to the wonderful taste of chocolate?
Donald Hoffman (1:59:06.800)
I think that we will always go deeper and deeper,
Lex Fridman (1:59:10.200)
but we will never solve that.
Donald Hoffman (1:59:12.680)
That in some sense, that will be a primitive.
Lex Fridman (1:59:15.120)
I hope I'm wrong.
Donald Hoffman (1:59:16.760)
Maybe it's just the limits of my current imagination.
Lex Fridman (1:59:23.080)
So I'll just say my imagination right now
Donald Hoffman (1:59:26.000)
doesn't peer that deep.
Lex Fridman (1:59:30.280)
By the way, I'm saying this,
Donald Hoffman (1:59:31.640)
I don't want to discourage some brilliant 20 year old
Lex Fridman (1:59:35.120)
who then later on proves me dead wrong.
Donald Hoffman (1:59:37.920)
I hope to be proven dead wrong.
Lex Fridman (1:59:39.360)
Just like you said, essentially from now,
Donald Hoffman (1:59:41.040)
everything we're saying now, everything you're saying,
Lex Fridman (1:59:43.320)
all your theories will be laughing stock.
Donald Hoffman (1:59:45.600)
They will respect the puzzle solving abilities
Lex Fridman (1:59:51.120)
and how much we were able to do with so little.
Lex Fridman (1:59:54.040)
But outside of that, you will all be just,
Lex Fridman (1:59:58.240)
the silliness will be entertainment for a teenager.
Donald Hoffman (20:01.080)
Reductionism has been a fantastic methodology.
Lex Fridman (20:03.800)
So we had a great ontology of space time,
Donald Hoffman (20:05.680)
a great methodology of reductionism.
Lex Fridman (20:07.920)
Now it's time for a new trick.
Lex Fridman (20:10.720)
But now you need to go deeper and show,
Lex Fridman (20:13.000)
but by the way, this doesn't mean we throw away
Donald Hoffman (20:14.920)
everything we've done, not by a long shot.
Lex Fridman (20:17.160)
Every new idea that we come up with beyond space time
Donald Hoffman (20:20.720)
must project precisely into space time.
Lex Fridman (20:23.280)
And it better give us back everything that we know
Lex Fridman (20:25.320)
and love in space time or generalizations,
Lex Fridman (20:28.680)
or it's not gonna be taken seriously and it shouldn't be.
Lex Fridman (20:30.800)
So we have a strong constraint on whatever we're going to do
Lex Fridman (20:34.640)
beyond space time, it needs to project into space time.
Lex Fridman (20:37.600)
And whatever this deeper theory is,
Lex Fridman (20:39.360)
it may not itself have evolution by natural selection.
Donald Hoffman (20:42.800)
This may not be part of this deeper realm.
Lex Fridman (20:44.440)
But when we take whatever that thing is beyond space time
Lex Fridman (20:47.480)
and project it into space time,
Lex Fridman (20:49.160)
it has to look like evolution by natural selection
Donald Hoffman (20:51.720)
or it's wrong.
Lex Fridman (20:52.920)
So that's a strong constraint on this work.
Lex Fridman (20:57.400)
So even the evolution by natural selection
Lex Fridman (21:00.880)
and quantum field theory could be interfaces
Donald Hoffman (21:06.440)
into something that doesn't look anything like,
Lex Fridman (21:11.760)
like you mentioned.
Donald Hoffman (21:12.600)
I mean, it's interesting to think that evolution
Lex Fridman (21:14.520)
might be a very crappy interface
Donald Hoffman (21:16.880)
into something much deeper.
Lex Fridman (21:18.360)
That's right.
Donald Hoffman (21:19.200)
They're both telling us that the framework that you've had
Lex Fridman (21:21.880)
can only go so far and it has to stop.
Lex Fridman (21:24.160)
And there's something beyond.
Lex Fridman (21:25.600)
And the very framework that is space and time itself.
Donald Hoffman (21:29.160)
Now, of course, evolution by natural selection
Lex Fridman (21:32.360)
is not telling us about like Einstein's relativistic
Donald Hoffman (21:35.960)
space time.
Lex Fridman (21:36.800)
So that was another question you asked a little bit earlier.
Donald Hoffman (21:38.280)
It's telling us more about our perceptual space and time,
Lex Fridman (21:42.360)
which we have used as the basis for creating
Donald Hoffman (21:46.160)
first Newtonian space versus time
Lex Fridman (21:49.680)
as a mathematical extension of our perceptions.
Lex Fridman (21:53.280)
And then Einstein then took that and extended it even further.
Lex Fridman (21:56.680)
So the relationship between what evolution is telling us
Lex Fridman (21:59.120)
and what the physicists are telling us is that
Lex Fridman (22:01.040)
in some sense, the Newton and Einstein space time
Donald Hoffman (22:07.160)
are formulated as sort of rigorous extensions
Lex Fridman (22:11.320)
of our perceptual space,
Donald Hoffman (22:14.040)
making it mathematically rigorous
Lex Fridman (22:15.520)
and laying out the symmetries that they find there.
Lex Fridman (22:19.080)
So that's sort of the relationship between them.
Lex Fridman (22:20.760)
So it's the perceptual space time
Donald Hoffman (22:22.440)
that evolution is telling us
Lex Fridman (22:24.040)
is just a user interface effectively.
Lex Fridman (22:27.760)
And then the physicists are finding
Lex Fridman (22:28.960)
that even the mathematical extension of that
Donald Hoffman (22:31.440)
into the Einsteinian formulation has to be as well,
Lex Fridman (22:36.160)
not the final story, there's something deeper.
Lex Fridman (22:38.120)
So let me ask you about reductionism and interfaces
Lex Fridman (22:43.200)
as we march forward from Newtonian physics
Donald Hoffman (22:47.960)
to quantum mechanics.
Lex Fridman (22:49.920)
These are all, in your view, interfaces.
Lex Fridman (22:56.280)
Are we getting closer to objective reality?
Lex Fridman (22:59.240)
How do we know if these interfaces in the process of science,
Donald Hoffman (23:04.840)
the reason we like those interfaces
Lex Fridman (23:06.880)
is because they're predictive of some aspects,
Donald Hoffman (23:09.720)
strongly predictive about some aspects of our reality.
Lex Fridman (23:14.120)
Is that completely deviating
Donald Hoffman (23:16.040)
from our understanding of that reality
Lex Fridman (23:19.560)
or is it helping us get closer and closer and closer?
Donald Hoffman (23:22.760)
Well, of course, one critical constraint
Lex Fridman (23:24.560)
on all of our theories
Donald Hoffman (23:25.400)
is that they are empirically tested
Lex Fridman (23:27.240)
and pass the experiments that we have for them.
Lex Fridman (23:30.800)
So no one's arguing against experiments being important
Lex Fridman (23:34.440)
and wanting to test all of our current theories
Lex Fridman (23:38.440)
and any new theories on that.
Lex Fridman (23:40.600)
So that's all there.
Lex Fridman (23:44.280)
But we have good reason to believe
Lex Fridman (23:48.040)
that science will never get a theory of everything.
Donald Hoffman (23:51.520)
Everything, everything.
Lex Fridman (23:52.760)
Everything, everything, right.
Donald Hoffman (23:53.720)
A final theory of everything, right.
Lex Fridman (23:55.640)
I think that my own take is, for what it's worth,
Donald Hoffman (23:58.440)
is that Gödel's incompleteness theorem
Lex Fridman (24:00.920)
sort of points us in that direction,
Donald Hoffman (24:02.480)
that even with mathematics,
Lex Fridman (24:05.240)
any finite axiomatization that's sophisticated enough
Donald Hoffman (24:08.760)
to be able to do arithmetic,
Lex Fridman (24:10.360)
it's easy to show that there'll be statements that are true,
Donald Hoffman (24:13.560)
that can't be proven,
Lex Fridman (24:16.280)
can't be deduced from within that framework.
Lex Fridman (24:19.480)
And if you add the new statements to your axioms,
Lex Fridman (24:21.920)
then there'll be always new statements that are true,
Lex Fridman (24:24.280)
but can't be proven with a new axiom system.
Lex Fridman (24:26.920)
And the best scientific theories in physics, for example,
Lex Fridman (24:32.640)
and also now evolution, are mathematical.
Lex Fridman (24:35.080)
So our theories are gonna be,
Donald Hoffman (24:36.320)
they're gonna have their own assumptions
Lex Fridman (24:38.400)
and they'll be mathematically precise.
Lex Fridman (24:41.720)
And there'll be theories, perhaps,
Lex Fridman (24:42.800)
of everything except those assumptions,
Donald Hoffman (24:44.440)
because the assumptions are,
Lex Fridman (24:46.280)
we say, please grant me these assumptions.
Donald Hoffman (24:48.280)
If you grant me these assumptions,
Lex Fridman (24:49.440)
then I can explain this other stuff.
Lex Fridman (24:52.000)
So you have the assumptions that are like miracles,
Lex Fridman (24:57.600)
as far as the theory is concerned.
Donald Hoffman (24:58.640)
They're not explained.
Lex Fridman (24:59.480)
They're the starting points for explanation.
Lex Fridman (25:01.520)
And then you have the mathematical structure
Lex Fridman (25:03.160)
of the theory itself, which will have the Gödel limits.
Lex Fridman (25:07.520)
And so my take is that reality,
Lex Fridman (25:12.520)
reality, whatever it is, is always going to transcend
Donald Hoffman (25:18.240)
any conceptual theory that we didn't come up with.
Lex Fridman (25:22.440)
There's always gonna be mystery at the edges.
Donald Hoffman (25:24.800)
Right.
Lex Fridman (25:27.640)
Contradictions and all that kind of stuff.
Donald Hoffman (25:29.400)
Okay.
Lex Fridman (25:31.560)
And truths.
Lex Fridman (25:32.880)
So there's this idea that is brought up
Lex Fridman (25:34.840)
in the financial space of settlement of transactions.
Donald Hoffman (25:39.440)
It's often talked about in cryptocurrency, especially.
Lex Fridman (25:42.560)
So you could do, you know, money, cash,
Donald Hoffman (25:44.600)
is not connected to anything.
Lex Fridman (25:48.640)
It used to be connected to gold, to physical reality,
Lex Fridman (25:52.200)
but then you can use money to exchange,
Lex Fridman (25:54.640)
to exchange value, to transact.
Lex Fridman (25:57.280)
So when it was on the gold standard,
Lex Fridman (25:59.760)
the money would represent some stable component of reality.
Donald Hoffman (26:04.760)
Isn't it more effective to avoid things like hyperinflation
Lex Fridman (26:12.280)
if we generalize that idea?
Donald Hoffman (26:14.160)
Isn't it better to connect your,
Lex Fridman (26:19.120)
whatever we humans are doing
Donald Hoffman (26:20.520)
in the social interaction space with each other,
Lex Fridman (26:23.040)
isn't it better from an evolutionary perspective
Donald Hoffman (26:26.000)
to connect it to some degree to reality
Lex Fridman (26:28.040)
so that the transactions are settled
Donald Hoffman (26:31.620)
with something that's universal,
Lex Fridman (26:33.740)
as opposed to us constantly operating
Lex Fridman (26:35.880)
in something that's a complete illusion?
Lex Fridman (26:38.080)
Isn't it easy to hyperinflate that?
Donald Hoffman (26:41.400)
Like where you really deviate very, very far away
Lex Fridman (26:49.720)
from the underlying reality,
Lex Fridman (26:51.000)
or do you not never get in trouble for this?
Lex Fridman (26:53.720)
Can you just completely drift far, far away
Lex Fridman (26:58.200)
from the underlying reality and never get in trouble?
Lex Fridman (27:01.560)
That's a great question, on the financial side,
Donald Hoffman (27:04.440)
there's two levels at least
Lex Fridman (27:05.560)
that we could take your question.
Donald Hoffman (27:06.880)
One is strictly like evolutionary psychology
Lex Fridman (27:09.800)
of financial systems, and that's pretty interesting.
Lex Fridman (27:13.480)
And there the decentralized idea,
Lex Fridman (27:15.100)
the DeFi kind of idea in cryptocurrencies
Donald Hoffman (27:18.520)
may make good sense
Lex Fridman (27:19.920)
from just an evolutionary psychology point of view.
Donald Hoffman (27:22.400)
Having human nature being what it is,
Lex Fridman (27:25.720)
putting a lot of faith in a few central controllers
Donald Hoffman (27:30.600)
depends a lot on the veracity of those
Lex Fridman (27:34.280)
and trustworthiness of those few central controllers.
Lex Fridman (27:37.080)
And we have ample evidence time and again
Lex Fridman (27:39.440)
that that's often betrayed.
Lex Fridman (27:41.880)
So it makes good evolutionary sense, I would say,
Lex Fridman (27:44.920)
to have a decentralized,
Lex Fridman (27:46.680)
I mean, democracy is a step in that direction, right?
Lex Fridman (27:49.600)
We don't have a monarch now telling us what to do,
Lex Fridman (27:52.240)
we decentralize things, right?
Lex Fridman (27:54.560)
Because if the monarch,
Donald Hoffman (27:55.880)
if you have Marcus Aurelius as your emperor, you're great.
Lex Fridman (27:58.600)
If you have Nero, it's not so great.
Lex Fridman (28:01.160)
And so we don't want that.
Lex Fridman (28:02.320)
So democracy is a step in that direction,
Lex Fridman (28:04.280)
but I think the DeFi thing is an even bigger step
Lex Fridman (28:08.800)
and is going to even make the democratization even greater.
Lex Fridman (28:13.120)
So that's one level of it.
Lex Fridman (28:14.840)
Also, the fact that power corrupts
Lex Fridman (28:16.480)
and absolute power corrupts absolutely
Lex Fridman (28:18.120)
is also a consequence of evolution.
Lex Fridman (28:24.200)
That's also a feature, I think, right?
Lex Fridman (28:26.960)
You can argue from the long span of living organisms,
Donald Hoffman (28:30.880)
it's nice for power to corrupt for you to,
Lex Fridman (28:33.840)
so mad men and women throughout history
Donald Hoffman (28:38.800)
might be useful to teach us a lesson about ourselves.
Lex Fridman (28:43.040)
We can learn from our negative example, right?
Donald Hoffman (28:44.800)
Exactly.
Lex Fridman (28:45.640)
Right, right, right.
Donald Hoffman (28:48.040)
Power does corrupt and I think that you can think about that
Lex Fridman (28:51.120)
again from an evolutionary point of view.
Lex Fridman (28:53.600)
But I think that your question was a little deeper
Lex Fridman (28:55.800)
when that was, does the evolutionary interface idea
Donald Hoffman (29:01.480)
sort of unhinge science from some kind of important test
Lex Fridman (29:07.920)
for the theories, right?
Donald Hoffman (29:08.840)
We don't want, it doesn't mean that anything goes
Lex Fridman (29:12.320)
in scientific theory, but there's no,
Donald Hoffman (29:14.560)
if we don't see the truth,
Lex Fridman (29:15.880)
is there no way to tether our theories and test them?
Lex Fridman (29:18.640)
And I think there's no problem there.
Lex Fridman (29:23.640)
We can only test things in terms of what we can measure
Donald Hoffman (29:27.800)
with our senses in space and time.
Lex Fridman (29:29.520)
So we're going to have to continue to do experiments
Donald Hoffman (29:33.400)
and, but we're going to re,
Lex Fridman (29:35.040)
we're going to understand a little bit differently
Lex Fridman (29:36.840)
what those experiments are.
Lex Fridman (29:38.440)
We had thought that when we see a pointer
Donald Hoffman (29:41.840)
on some machine in an experiment,
Lex Fridman (29:45.720)
that the machine exists, the pointer exists
Lex Fridman (29:48.240)
and the values exist even when no one is looking at them
Lex Fridman (29:51.280)
and that they're an objective truth.
Lex Fridman (29:52.760)
And our best theories are telling us no,
Lex Fridman (29:55.480)
the pointers are just pointers
Lex Fridman (29:58.480)
and that's what you have to rely on
Lex Fridman (2:00:01.760)
Especially the silliness when we thought
Donald Hoffman (2:00:03.120)
that we were so smart and we knew it all.
Lex Fridman (2:00:06.000)
So it would be interesting to explore your ideas
Donald Hoffman (2:00:08.320)
by contrasting, you mentioned Annika, Annika Harris,
Lex Fridman (2:00:12.560)
you mentioned Philip Goff.
Lex Fridman (2:00:15.720)
So outside of, if you're not allowed to say
Lex Fridman (2:00:19.480)
the fundamental disagreement is the fact
Donald Hoffman (2:00:21.640)
that space time is fundamental.
Lex Fridman (2:00:24.400)
What are interesting distinctions
Donald Hoffman (2:00:26.600)
between ideas of consciousness
Lex Fridman (2:00:28.440)
between you and Annika, for example?
Donald Hoffman (2:00:30.080)
You guys have, you've been on a podcast together,
Lex Fridman (2:00:33.360)
I'm sure in private you guys
Donald Hoffman (2:00:36.600)
have some incredible conversations.
Lex Fridman (2:00:38.080)
So where are some interesting sticking points,
Donald Hoffman (2:00:41.560)
some interesting disagreements,
Lex Fridman (2:00:44.040)
let's say with Annika first.
Donald Hoffman (2:00:45.680)
Maybe there'll be a few other people.
Lex Fridman (2:00:47.640)
Well, Annika and I just had a conversation this morning
Donald Hoffman (2:00:49.680)
where we were talking about our ideas
Lex Fridman (2:00:51.560)
and what we discovered really in our conversation
Donald Hoffman (2:00:53.800)
was that we're pretty much on the same page.
Lex Fridman (2:00:57.040)
It was really just about consciousness.
Donald Hoffman (2:01:00.920)
Our ideas about consciousness
Lex Fridman (2:01:02.040)
are pretty much on the same page.
Donald Hoffman (2:01:04.040)
She rightly has cautioned me to,
Lex Fridman (2:01:07.840)
when I talk about conscious agents,
Donald Hoffman (2:01:10.080)
to point out that the notion of agency
Lex Fridman (2:01:12.000)
is not fundamental in my theory.
Donald Hoffman (2:01:15.800)
The notion of self is not fundamental
Lex Fridman (2:01:17.240)
and that's absolutely true.
Donald Hoffman (2:01:18.880)
I can use this network of conscious agents,
Lex Fridman (2:01:22.520)
I now use as a technical term,
Donald Hoffman (2:01:25.000)
conscious agent is a technical term
Lex Fridman (2:01:26.400)
for that probability space with the Markovian dynamics.
Donald Hoffman (2:01:29.320)
I can use that to build models of a self
Lex Fridman (2:01:31.840)
and to build models of agency,
Lex Fridman (2:01:33.240)
but they're not fundamental.
Lex Fridman (2:01:35.200)
So she has really been very helpful
Donald Hoffman (2:01:40.320)
in helping me to be a little bit clear about these ideas
Lex Fridman (2:01:43.600)
and not say things that are misleading.
Donald Hoffman (2:01:45.880)
This is the interesting thing about language, actually,
Lex Fridman (2:01:50.000)
is that language, quite obviously,
Donald Hoffman (2:01:52.320)
is an interface to truth.
Lex Fridman (2:01:56.640)
It's so fascinating that individual words
Donald Hoffman (2:02:01.400)
can have so much ambiguity
Lex Fridman (2:02:05.680)
and the specific choices of a word
Donald Hoffman (2:02:10.760)
within a particular sentence,
Lex Fridman (2:02:12.120)
within the context of a sentence,
Donald Hoffman (2:02:13.600)
can have such a difference in meaning.
Lex Fridman (2:02:17.240)
It's quite fascinating,
Donald Hoffman (2:02:18.600)
especially when you're talking about topics
Lex Fridman (2:02:20.360)
like consciousness, because it's a very loaded term.
Donald Hoffman (2:02:23.960)
It means a lot of things to a lot of people
Lex Fridman (2:02:26.160)
and the entire concept is shrouded in mystery.
Lex Fridman (2:02:29.120)
So a combination of the fact that it's a loaded term
Lex Fridman (2:02:32.440)
and that there's a lot of mystery,
Donald Hoffman (2:02:34.320)
people can just interpret it in all kinds of ways.
Lex Fridman (2:02:36.920)
And so you have to be both precise
Lex Fridman (2:02:39.160)
and help them avoid getting stuck
Lex Fridman (2:02:43.640)
on some kind of side road of miscommunication,
Donald Hoffman (2:02:48.120)
lost in translation because you used the wrong word.
Lex Fridman (2:02:50.840)
That's interesting.
Donald Hoffman (2:02:51.680)
I mean, because for a lot of people,
Lex Fridman (2:02:54.380)
consciousness is ultimately connected to a self.
Donald Hoffman (2:03:01.080)
I mean, that's our experience of consciousness
Lex Fridman (2:03:04.920)
is very, it's connected to this ego.
Lex Fridman (2:03:08.840)
I mean, I just, I mean, what else could it possibly be?
Lex Fridman (2:03:12.840)
I can't even, how do you begin to comprehend,
Donald Hoffman (2:03:15.760)
to visualize, to conceptualize a consciousness
Lex Fridman (2:03:19.000)
that's not connected to like this particular organism?
Donald Hoffman (2:03:23.680)
I have a way of thinking about this whole problem now
Lex Fridman (2:03:26.520)
that comes out of this framework that's different.
Lex Fridman (2:03:30.900)
So we can imagine a dynamics of consciousness,
Lex Fridman (2:03:35.280)
not in space and time, just abstractly.
Donald Hoffman (2:03:37.660)
It could be cooperative for all we know.
Lex Fridman (2:03:40.560)
It could be very friendly, I don't know.
Lex Fridman (2:03:43.960)
And you can set up a dynamics, a Markovian dynamics
Lex Fridman (2:03:46.720)
that is so called stationary.
Lex Fridman (2:03:48.820)
And that's a technical term,
Lex Fridman (2:03:50.160)
which means that the entropy effectively is not increasing.
Donald Hoffman (2:03:53.400)
There is some entropy, but it's constant.
Lex Fridman (2:03:55.160)
So there's no increasing entropy.
Lex Fridman (2:03:56.960)
And in that sense, the dynamics is timeless.
Lex Fridman (2:04:01.540)
There is no entropic time, but it's a trivial theorem,
Donald Hoffman (2:04:05.620)
three line proof that if you have a stationary
Lex Fridman (2:04:10.620)
Markovian dynamics, any projection that you make
Donald Hoffman (2:04:13.860)
of that dynamics by conditional probability.
Lex Fridman (2:04:16.240)
And if you want, I can state a little bit more,
Donald Hoffman (2:04:17.980)
even more mathematically precisely
Lex Fridman (2:04:19.340)
for some readers or listeners.
Lex Fridman (2:04:21.920)
But if any projection you take by conditional probability,
Lex Fridman (2:04:25.120)
the induced image of that Markov chain
Donald Hoffman (2:04:28.980)
will have increasing entropy.
Lex Fridman (2:04:32.940)
You will have entropic time.
Lex Fridman (2:04:34.100)
So I'll be very, very precise.
Lex Fridman (2:04:36.140)
I have a Markov chain X1, X2 through Xn
Lex Fridman (2:04:40.740)
where Xn, n goes to infinity, right?
Lex Fridman (2:04:43.800)
The entropy H, capital H of Xn is equal to the entropy H
Donald Hoffman (2:04:49.680)
of Xn minus one for all n.
Lex Fridman (2:04:52.440)
So the entropy is the same.
Lex Fridman (2:04:55.660)
But it's a theorem that H of Xn,
Lex Fridman (2:05:00.660)
say given X sub one is greater than or equal to
Donald Hoffman (2:05:06.100)
H of Xn minus one given X1.
Lex Fridman (2:05:10.900)
Sure, where does the greater come from?
Donald Hoffman (2:05:13.580)
Because with the theorem, the three line proof,
Lex Fridman (2:05:17.220)
H of Xn given X1 is greater than or equal to H of Xn
Donald Hoffman (2:05:23.940)
given X1 and X2 because conditioning reduces.
Lex Fridman (2:05:27.500)
But then H of Xn minus one given X1, X2
Donald Hoffman (2:05:35.340)
is equal to H of Xn given X2,
Lex Fridman (2:05:39.620)
Xn minus one given X2 by the Markov property.
Lex Fridman (2:05:44.020)
And then because it's stationary, it's equal to H of X.
Lex Fridman (2:05:51.840)
I have to write it down.
Donald Hoffman (2:05:53.180)
Anyway, there's a three line proof.
Lex Fridman (2:05:56.820)
Sure, but the assumption of stationarity,
Donald Hoffman (2:06:02.100)
we're using a lot of terms that people won't understand,
Lex Fridman (2:06:04.540)
doesn't matter.
Lex Fridman (2:06:07.020)
So there's some kind of, some Markovian dynamics
Lex Fridman (2:06:10.220)
is basically trying to model some kind of system
Donald Hoffman (2:06:13.540)
with some probabilities and there's agents
Lex Fridman (2:06:15.700)
and they interact in some kind of way
Lex Fridman (2:06:17.260)
and you can say something about that system
Lex Fridman (2:06:19.980)
as it evolves stationarity.
Lex Fridman (2:06:22.900)
So a stationary system is one that has certain properties
Lex Fridman (2:06:28.600)
in terms of entropy, very well.
Lex Fridman (2:06:30.820)
But we don't know if it's stationary or not.
Lex Fridman (2:06:33.020)
We don't know what the properties.
Donald Hoffman (2:06:35.520)
Right.
Lex Fridman (2:06:36.500)
So you have to kind of take assumptions
Lex Fridman (2:06:38.220)
and see, okay, well, what does the system behave like
Lex Fridman (2:06:42.040)
under these different properties?
Donald Hoffman (2:06:43.580)
The more constraints, the more assumptions you take,
Lex Fridman (2:06:46.340)
the more interesting, powerful things you can say,
Lex Fridman (2:06:49.900)
but sometimes they're limiting.
Lex Fridman (2:06:52.060)
That said, we're talking about consciousness here.
Donald Hoffman (2:06:54.460)
Right.
Lex Fridman (2:06:55.980)
How does that, you said cooperative, okay, competitive.
Donald Hoffman (2:07:02.420)
It just, I like chocolate.
Lex Fridman (2:07:04.140)
I'm sitting here, I have a brain, I'm wearing a suit.
Donald Hoffman (2:07:08.500)
It sure as hell feels like I'm myself.
Lex Fridman (2:07:11.220)
Right.
Lex Fridman (2:07:12.060)
Now, what, am I tuning in?
Lex Fridman (2:07:14.660)
Am I plugging into something?
Donald Hoffman (2:07:16.820)
Am I a projection, a simple, trivial projection
Lex Fridman (2:07:20.700)
into space time from some much larger organism
Lex Fridman (2:07:23.980)
that I can't possibly comprehend?
Lex Fridman (2:07:25.980)
How the hell, you're saying some,
Donald Hoffman (2:07:28.020)
you're building up mathematical intuitions, fine, great.
Lex Fridman (2:07:31.500)
But I'm just, I'm having an existential crisis here
Lex Fridman (2:07:35.100)
and I'm gonna die soon.
Lex Fridman (2:07:36.700)
Well, I'll die pretty quickly.
Lex Fridman (2:07:37.940)
So I wanna figure out why chocolate's so delicious.
Lex Fridman (2:07:43.300)
So help me out here.
Lex Fridman (2:07:44.200)
So let's just keep sneaking up to this.
Lex Fridman (2:07:47.100)
Right, so the whole technical thing was to say this.
Donald Hoffman (2:07:52.500)
Even if the dynamics of consciousness is stationary
Lex Fridman (2:07:56.220)
so that there is no entropic time,
Donald Hoffman (2:07:58.720)
any projection of it, any view of it
Lex Fridman (2:08:03.940)
will have the artifact of entropic time.
Donald Hoffman (2:08:08.620)
That's a limited resource.
Lex Fridman (2:08:10.860)
Limited resources, so that the fundamental dynamics
Donald Hoffman (2:08:14.220)
may have no limits, limited resources whatsoever.
Lex Fridman (2:08:17.500)
Any projection will have certainly time
Donald Hoffman (2:08:19.980)
as a limited resource
Lex Fridman (2:08:21.180)
and probably lots of other limited resources.
Donald Hoffman (2:08:24.140)
Hence, we could get competition and evolution
Lex Fridman (2:08:28.140)
and nature, red and tooth and claw
Donald Hoffman (2:08:30.400)
as an artifact of a deeper system
Lex Fridman (2:08:32.700)
in which those aren't fundamental.
Lex Fridman (2:08:34.600)
And in fact, I take it as something
Lex Fridman (2:08:37.820)
that this theory must do at some point
Donald Hoffman (2:08:41.020)
is to show how networks of conscious agents,
Lex Fridman (2:08:42.740)
even if they're not resource limited,
Donald Hoffman (2:08:45.660)
give rise to evolution by natural selection
Lex Fridman (2:08:47.620)
via a projection.
Donald Hoffman (2:08:49.340)
Yeah, but you're saying,
Lex Fridman (2:08:51.100)
I'm trying to understand how the limited resources
Donald Hoffman (2:08:53.260)
that give rise to,
Lex Fridman (2:08:55.540)
so first the thing gives rise to time,
Donald Hoffman (2:08:57.940)
that gives rise to limited resource,
Lex Fridman (2:08:59.860)
that gives rise to evolution by natural selection,
Lex Fridman (2:09:03.140)
how that has to do with the fact that chocolate's delicious?
Lex Fridman (2:09:05.940)
Well, it's not gonna do that directly.
Donald Hoffman (2:09:08.060)
It's gonna get to this notion of self.
Lex Fridman (2:09:10.700)
Oh, it's gonna give you?
Donald Hoffman (2:09:12.140)
The notion of self.
Lex Fridman (2:09:12.980)
Oh, the evolution gives you the notion of self.
Lex Fridman (2:09:14.700)
And also of a self separate from other selves.
Lex Fridman (2:09:18.420)
So the idea would be that.
Donald Hoffman (2:09:20.140)
It's competition, it has life and death,
Lex Fridman (2:09:22.140)
all those kinds of things.
Donald Hoffman (2:09:23.300)
That's right.
Lex Fridman (2:09:24.140)
So it won't, I don't think,
Donald Hoffman (2:09:26.740)
as I said, I don't think that I can tell you
Lex Fridman (2:09:28.540)
how the formless gives rise
Donald Hoffman (2:09:30.180)
to the experience of chocolate.
Lex Fridman (2:09:32.380)
Right now, my current theory says
Donald Hoffman (2:09:33.660)
that's one of the miracles I'm assuming.
Lex Fridman (2:09:35.700)
Yeah.
Lex Fridman (2:09:36.660)
So my theory can't do it.
Lex Fridman (2:09:38.180)
And the reason my theory can't do it
Donald Hoffman (2:09:39.260)
is because Hoffman's brain can't do it right now.
Lex Fridman (2:09:41.460)
But the notion of self, yes.
Donald Hoffman (2:09:45.980)
The notion of self can be an artifact
Lex Fridman (2:09:49.060)
of the projection of it.
Lex Fridman (2:09:51.700)
So there's one conscious agent.
Lex Fridman (2:09:55.340)
Because anytime conscious agents interact,
Donald Hoffman (2:09:56.780)
they form a new conscious agent.
Lex Fridman (2:09:57.860)
So there's one conscious agent.
Donald Hoffman (2:10:00.300)
Any projection of that one conscious agent
Lex Fridman (2:10:02.620)
gives rise to time,
Donald Hoffman (2:10:05.060)
even if there wasn't any time in that one conscious agent.
Lex Fridman (2:10:07.660)
And it gives rise, I want to,
Donald Hoffman (2:10:10.060)
now I haven't proven this.
Lex Fridman (2:10:10.900)
So this is, so now this is me guessing
Donald Hoffman (2:10:13.220)
where the theory is going to go.
Lex Fridman (2:10:14.820)
I haven't done this.
Donald Hoffman (2:10:15.660)
There's no paper on this yet.
Lex Fridman (2:10:16.860)
So now I'm speculating.
Donald Hoffman (2:10:18.380)
My guess is I'll be able to show,
Lex Fridman (2:10:20.180)
or my brighter colleagues working with me
Donald Hoffman (2:10:22.460)
will be able to show
Lex Fridman (2:10:23.660)
that we will get evolution of a natural selection,
Donald Hoffman (2:10:26.660)
the notion of individual selves,
Lex Fridman (2:10:28.260)
individual physical objects and so forth
Donald Hoffman (2:10:29.940)
coming out as a projection of this thing.
Lex Fridman (2:10:31.620)
And that the self, this then will be really interesting
Donald Hoffman (2:10:35.900)
in terms of how it starts to interact
Lex Fridman (2:10:37.420)
with certain spiritual traditions, right?
Donald Hoffman (2:10:40.980)
Where they will say that there is a notion of self
Lex Fridman (2:10:44.100)
that needs to be let go,
Donald Hoffman (2:10:45.540)
which is this finite self
Lex Fridman (2:10:46.860)
that's competing with other selves
Donald Hoffman (2:10:48.300)
to get more money and prestige and so forth.
Lex Fridman (2:10:52.020)
That self in some sense has to die.
Lex Fridman (2:10:54.380)
But there's a deeper self,
Lex Fridman (2:10:56.460)
which is the timeless being
Donald Hoffman (2:11:01.660)
that precludes, not precludes,
Lex Fridman (2:11:04.340)
but precedes any particular conscious experiences,
Donald Hoffman (2:11:08.220)
the ground of all experience.
Lex Fridman (2:11:10.380)
That there's that notion of a deep capital self.
Lex Fridman (2:11:13.660)
But our little capital, lowercase s selves
Lex Fridman (2:11:17.980)
could be artifacts of projection.
Lex Fridman (2:11:20.820)
And it may be that what consciousness is doing
Lex Fridman (2:11:25.620)
in this framework is, right?
Donald Hoffman (2:11:26.780)
It's projected itself down into a self
Lex Fridman (2:11:30.300)
that calls itself dawn
Lex Fridman (2:11:31.420)
and a self that calls itself lax.
Lex Fridman (2:11:33.940)
And through conversations like this,
Donald Hoffman (2:11:36.020)
it's trying to find out about itself
Lex Fridman (2:11:37.900)
and eventually transcend the limits
Donald Hoffman (2:11:41.180)
of the dawn and lax little icons that it's using
Lex Fridman (2:11:45.860)
and that little projection of itself.
Donald Hoffman (2:11:49.020)
Through this conversation,
Lex Fridman (2:11:50.580)
somehow it's learning about itself.
Lex Fridman (2:11:53.620)
So that thing dressed me up today
Lex Fridman (2:11:57.700)
in order to understand itself.
Lex Fridman (2:11:59.940)
And in some sense, you and I are not separate
Lex Fridman (2:12:02.220)
from that thing and we're not separate from each other.
Donald Hoffman (2:12:03.940)
Yeah, well, I have to question the fashion choices
Lex Fridman (2:12:06.980)
on my end then.
Donald Hoffman (2:12:08.500)
All right, so you mentioned you agree
Lex Fridman (2:12:10.900)
in terms of consciousness on a lot of things with Anika.
Donald Hoffman (2:12:15.500)
Is there somebody, friend or friendly foe
Lex Fridman (2:12:20.260)
that you disagree with in some nuanced, interesting way
Donald Hoffman (2:12:25.540)
or some major way about consciousness,
Lex Fridman (2:12:28.580)
about these topics of reality that you return to?
Donald Hoffman (2:12:34.340)
Often, it's like Christopher Hitchens
Lex Fridman (2:12:38.660)
with Rabbi David Wolpe have had interesting conversations
Donald Hoffman (2:12:43.460)
through years that added to the complexity
Lex Fridman (2:12:46.260)
and the beauty of their friendship.
Donald Hoffman (2:12:47.660)
Is there somebody like that that over the years
Lex Fridman (2:12:51.940)
has been a source of disagreement with you
Lex Fridman (2:12:54.540)
that's strengthened your ideas?
Lex Fridman (2:12:56.180)
Hmm, my ideas have been really shaped by several things.
Donald Hoffman (2:13:02.600)
One is the physicalist framework
Lex Fridman (2:13:06.600)
that my scientific colleagues, almost to a person,
Donald Hoffman (2:13:10.460)
have adopted and that I adopted too.
Lex Fridman (2:13:12.800)
And the reason I walked away from it was
Donald Hoffman (2:13:15.120)
because it became clear that we couldn't start
Lex Fridman (2:13:20.420)
with unconscious ingredients and boot up consciousness.
Lex Fridman (2:13:22.880)
Can you define physicalist in contrast to reductionist?
Lex Fridman (2:13:28.720)
So a physicalist, I would say as someone
Donald Hoffman (2:13:33.120)
who takes space time and the objects within space time
Lex Fridman (2:13:35.920)
as ontologically fundamental.
Donald Hoffman (2:13:38.880)
Right, and then reductionist is saying
Lex Fridman (2:13:42.080)
the smaller, the more fundamental.
Donald Hoffman (2:13:43.760)
That's a methodological thing.
Lex Fridman (2:13:45.840)
That's saying within space time,
Donald Hoffman (2:13:47.960)
as you go to smaller and smaller scales in space,
Lex Fridman (2:13:50.520)
you get deeper and deeper laws, more and more fundamental
Donald Hoffman (2:13:53.080)
laws and the reduction of temperature
Lex Fridman (2:13:57.360)
to particle movement was an example of that.
Lex Fridman (2:14:01.280)
But I think that the reason that worked
Lex Fridman (2:14:03.960)
was almost an artifact of the nature of our interface.
Donald Hoffman (2:14:07.960)
That was for a long time and your colleagues,
Lex Fridman (2:14:10.320)
including yourself, were physicalists
Lex Fridman (2:14:12.060)
and now you broke away.
Lex Fridman (2:14:13.480)
Broke away because I think you can't start
Donald Hoffman (2:14:15.640)
with unconscious ingredients and boot up consciousness.
Lex Fridman (2:14:18.400)
And so even with Roger Penrose
Donald Hoffman (2:14:21.200)
where there's like a gray area.
Lex Fridman (2:14:23.840)
Right, and here's the challenge I would put
Donald Hoffman (2:14:26.840)
to all of my friends and colleagues
Lex Fridman (2:14:30.000)
who give one specific conscious experience
Lex Fridman (2:14:36.040)
that you can boot up, right?
Lex Fridman (2:14:37.640)
So if you think that it's integrated information
Lex Fridman (2:14:40.680)
and I've asked this of Giulio Tononi a couple times
Lex Fridman (2:14:44.120)
back in the 90s and then just a couple years ago.
Donald Hoffman (2:14:46.600)
I asked Giulio, okay, so great, integrated information.
Lex Fridman (2:14:49.040)
So we're all interested in explaining
Donald Hoffman (2:14:51.560)
some specific conscious experiences.
Lex Fridman (2:14:53.200)
So what is, you know, pick one, the taste of chocolate.
Lex Fridman (2:14:56.880)
What is the integrated information precise structure
Lex Fridman (2:15:00.860)
that we need for chocolate and why does that structure
Donald Hoffman (2:15:03.800)
have to be for chocolate and why is it
Lex Fridman (2:15:07.360)
that it could not possibly be vanilla?
Donald Hoffman (2:15:09.280)
Is there any, I asked him, is there any one specific
Lex Fridman (2:15:11.600)
conscious experience that you can account for?
Donald Hoffman (2:15:13.520)
Because notice, they've set themselves the task
Lex Fridman (2:15:18.360)
of booting up conscious experiences from physical systems.
Donald Hoffman (2:15:21.440)
That's the task they've set themselves.
Lex Fridman (2:15:22.880)
But that doesn't mean they're,
Donald Hoffman (2:15:25.440)
I understand your intuition,
Lex Fridman (2:15:26.840)
but that doesn't mean they're wrong
Donald Hoffman (2:15:28.280)
just because they can't find a way to boot it up yet.
Lex Fridman (2:15:31.320)
That's right.
Donald Hoffman (2:15:32.160)
No, that doesn't mean that they're wrong.
Lex Fridman (2:15:33.000)
It just means that they haven't done it.
Donald Hoffman (2:15:37.360)
I think it's principled.
Lex Fridman (2:15:38.920)
The reason is principled,
Lex Fridman (2:15:40.000)
but I'm happy that they're exploring it.
Lex Fridman (2:15:43.060)
But the fact is, the remarkable fact is
Donald Hoffman (2:15:45.080)
there's not one theory.
Lex Fridman (2:15:46.280)
So integrated information theory,
Donald Hoffman (2:15:49.360)
orchestrated collapse of microtubules,
Lex Fridman (2:15:52.640)
global workspace theory.
Donald Hoffman (2:15:54.800)
These are all theories of consciousness.
Lex Fridman (2:15:56.280)
These are all theories of consciousness.
Donald Hoffman (2:15:57.720)
There's not a single theory that can give you
Lex Fridman (2:16:01.040)
a specific conscious experience that they say,
Donald Hoffman (2:16:03.200)
here is the physical dynamics or the physical structure
Lex Fridman (2:16:06.680)
that must be the taste of chocolate
Donald Hoffman (2:16:08.020)
or whatever one they want.
Lex Fridman (2:16:09.120)
So you're saying it's impossible.
Donald Hoffman (2:16:11.180)
They're saying it's just hard.
Lex Fridman (2:16:13.040)
Yeah.
Donald Hoffman (2:16:14.320)
My attitude is, okay, no one said
Lex Fridman (2:16:18.200)
you had to start with neurons or physical systems
Lex Fridman (2:16:20.560)
and boot up consciousness.
Lex Fridman (2:16:21.400)
You guys are just taking that.
Donald Hoffman (2:16:22.220)
You chose that problem.
Lex Fridman (2:16:23.520)
So since you chose that problem,
Lex Fridman (2:16:26.080)
how much progress have you made?
Lex Fridman (2:16:27.320)
Well, when you've not been able to come up
Donald Hoffman (2:16:30.040)
with a single specific conscious experience
Lex Fridman (2:16:32.980)
and you've had these brilliant people
Donald Hoffman (2:16:34.520)
working on it for decades now,
Lex Fridman (2:16:36.680)
that's not really good progress.
Donald Hoffman (2:16:38.360)
Let me ask you to play devil's advocate.
Lex Fridman (2:16:41.520)
Can you try to steel man, steel man meaning
Lex Fridman (2:16:46.120)
argue the best possible case for reality?
Lex Fridman (2:16:49.840)
The opposite of your book title.
Donald Hoffman (2:16:51.200)
So, or maybe just stick into consciousness.
Lex Fridman (2:16:54.360)
Can you take the physicalist view?
Lex Fridman (2:16:56.980)
Can you steel man the physicalist view
Lex Fridman (2:16:58.780)
for a brief moment playing devil's advocate too?
Lex Fridman (2:17:01.760)
Or steel man the person you used to be?
Lex Fridman (2:17:05.700)
Right, right. She's a physicalist.
Donald Hoffman (2:17:07.200)
What's a good, like saying that you might be wrong
Lex Fridman (2:17:10.560)
right now, what would be a convincing argument for that?
Donald Hoffman (2:17:17.360)
Well, I think the argument I would give
Lex Fridman (2:17:21.440)
that I believed was, look,
Donald Hoffman (2:17:23.880)
when you have very simple physical systems,
Lex Fridman (2:17:25.920)
like a piece of dirt,
Donald Hoffman (2:17:28.400)
there's not much evidence of life or consciousness.
Lex Fridman (2:17:30.600)
It's only when you get really complicated physical systems
Donald Hoffman (2:17:32.800)
like that have brains and really,
Lex Fridman (2:17:35.120)
the more complicated the brains,
Donald Hoffman (2:17:36.400)
the more it looks like there's consciousness
Lex Fridman (2:17:39.240)
and the more complicated that consciousness is.
Donald Hoffman (2:17:41.200)
Surely that means that simple physical systems
Lex Fridman (2:17:45.220)
don't create much consciousness or if maybe not any,
Donald Hoffman (2:17:49.000)
or maybe panpsychists,
Lex Fridman (2:17:50.980)
they create the most elementary kinds
Donald Hoffman (2:17:52.320)
of simple conscious experiences,
Lex Fridman (2:17:54.120)
but you need more complicated physical systems to boot up,
Donald Hoffman (2:17:59.400)
to create more complicated consciousnesses.
Lex Fridman (2:18:02.020)
I think that's the intuition
Donald Hoffman (2:18:03.080)
that drives most of my colleagues.
Lex Fridman (2:18:04.920)
And you're saying that this concept of complexity
Donald Hoffman (2:18:09.000)
is ill defined when you ground it to space time.
Lex Fridman (2:18:13.760)
Oh, I think it's well defined
Lex Fridman (2:18:15.700)
within the framework of space time, right?
Lex Fridman (2:18:17.440)
No, it's ill defined relative to what you need
Donald Hoffman (2:18:21.100)
to actually understand consciousness
Lex Fridman (2:18:23.240)
because you're grounding complexity in just in space time.
Donald Hoffman (2:18:26.280)
Oh, got you, right, right.
Lex Fridman (2:18:27.640)
Yeah, what I'm saying is if it were true
Donald Hoffman (2:18:33.640)
that space time was fundamental,
Lex Fridman (2:18:37.000)
then I would have to agree
Donald Hoffman (2:18:38.040)
that if there is such a thing as consciousness,
Lex Fridman (2:18:40.240)
given the data that we've got,
Donald Hoffman (2:18:41.400)
that complex brains have consciousness and dirt doesn't,
Lex Fridman (2:18:45.880)
that somehow it's the complexity of the dynamics
Donald Hoffman (2:18:48.620)
or organization, the function of the physical system
Lex Fridman (2:18:52.120)
that somehow is creating the consciousness.
Lex Fridman (2:18:55.080)
So under those assumptions, yes,
Lex Fridman (2:18:58.240)
but when the physicists themselves are telling us
Donald Hoffman (2:19:00.040)
that space time is not fundamental, then I can understand.
Lex Fridman (2:19:03.680)
See, then the whole picture starts to come into focus.
Lex Fridman (2:19:07.000)
Why, my colleagues are brilliant, right?
Lex Fridman (2:19:10.960)
These are really smart people.
Donald Hoffman (2:19:12.000)
I mean, Francis Crick worked on this
Lex Fridman (2:19:14.080)
for the last 20 years of his life.
Donald Hoffman (2:19:16.780)
These are not stupid people.
Lex Fridman (2:19:17.800)
These are brilliant, brilliant people.
Donald Hoffman (2:19:19.680)
The fact that we've come up
Lex Fridman (2:19:20.720)
with not a single specific conscious experience
Donald Hoffman (2:19:22.980)
that we can explain and no hope.
Lex Fridman (2:19:25.920)
There's no one that says, oh, I'm really close.
Donald Hoffman (2:19:27.480)
I'll have it for you in a year.
Lex Fridman (2:19:29.200)
No, there's just like, there's this fundamental gap.
Lex Fridman (2:19:33.280)
So much so that Steve Pinker in one of his writings says,
Lex Fridman (2:19:36.360)
look, he likes the global workspace theory,
Lex Fridman (2:19:39.860)
but he says the last dollop of the theory
Lex Fridman (2:19:41.520)
in which there's something it's like to,
Donald Hoffman (2:19:44.000)
he says, we may have to just stipulate that as a brute fact.
Lex Fridman (2:19:50.640)
Pinker is brilliant, right?
Donald Hoffman (2:19:52.000)
He understands the state of play
Lex Fridman (2:19:54.960)
on this problem of the hard problem of consciousness,
Donald Hoffman (2:19:57.360)
starting with physicalist assumptions
Lex Fridman (2:20:00.880)
and then trying to put up consciousness.
Lex Fridman (2:20:02.640)
So you've set yourself the problem.
Lex Fridman (2:20:04.480)
I'm starting with physical stuff
Donald Hoffman (2:20:05.760)
that's not conscious.
Lex Fridman (2:20:07.880)
I'm trying to get the taste of chocolate out
Donald Hoffman (2:20:11.380)
as maybe some kind of function of the dynamics of that.
Lex Fridman (2:20:14.880)
We've not been able to do that.
Lex Fridman (2:20:16.060)
And so Pinker is saying, we may have to punt.
Lex Fridman (2:20:18.920)
We may have to just stipulate that last bit.
Donald Hoffman (2:20:21.920)
He calls it the last dollop.
Lex Fridman (2:20:23.840)
And just stipulate it as a bare fact of nature
Donald Hoffman (2:20:27.120)
that there is something it's like.
Lex Fridman (2:20:28.440)
Well, from my point of view as the physical,
Donald Hoffman (2:20:30.560)
the whole point, the whole promise of the physicalist
Lex Fridman (2:20:33.040)
was we wouldn't have to stipulate.
Donald Hoffman (2:20:34.520)
I was gonna start with the physical stuff
Lex Fridman (2:20:36.440)
and explain where the consciousness came from.
Donald Hoffman (2:20:38.800)
If I'm going to stipulate consciousness,
Lex Fridman (2:20:40.680)
why don't I just stipulate consciousness
Lex Fridman (2:20:42.600)
and not stipulate all the physical stuff too?
Lex Fridman (2:20:45.040)
So I'm stipulating less.
Donald Hoffman (2:20:46.840)
I'm saying, okay, I agree.
Lex Fridman (2:20:47.680)
Which is the panpsychist perspective.
Donald Hoffman (2:20:49.520)
Well, it's actually what I call
Lex Fridman (2:20:51.160)
the conscious realist perspective.
Donald Hoffman (2:20:52.640)
Consciousness.
Lex Fridman (2:20:53.760)
Panpsychists are effectively dualists, right?
Donald Hoffman (2:20:55.880)
They're saying there's physical stuff
Lex Fridman (2:20:57.200)
that really is fundamental and then consciousness stuff.
Lex Fridman (2:21:00.040)
So I would go with Pinker and say, look,
Lex Fridman (2:21:01.820)
let's just stipulate the consciousness stuff,
Lex Fridman (2:21:04.080)
but I'm not gonna stipulate the physical stuff.
Lex Fridman (2:21:05.900)
I'm gonna actually now show how to boot up
Donald Hoffman (2:21:08.080)
the physical stuff from just the consciousness stuff.
Lex Fridman (2:21:11.120)
So I'll stipulate less.
Donald Hoffman (2:21:12.360)
Is it possible, so if you stipulate less,
Lex Fridman (2:21:15.180)
is it possible for our limited brains to visualize reality
Lex Fridman (2:21:22.080)
as we delve deeper and deeper and deeper?
Lex Fridman (2:21:25.640)
Is it possible to visualize somehow?
Donald Hoffman (2:21:27.920)
With the tools of math, with the tools of computers,
Lex Fridman (2:21:31.000)
with the tools of our mind, are we hopelessly lost?
Donald Hoffman (2:21:34.600)
You said there's ways to intuit what's true
Lex Fridman (2:21:40.480)
using mathematics and probability
Lex Fridman (2:21:44.680)
and sort of a Markovian dynamics, all that kind of stuff,
Lex Fridman (2:21:50.500)
but that's not visualizing.
Donald Hoffman (2:21:51.920)
That's a kind of building intuition.
Lex Fridman (2:21:55.880)
But is it possible to visualize
Donald Hoffman (2:21:57.320)
in the way we visualize so nicely in space time
Lex Fridman (2:22:00.900)
in four dimensions, in three dimensions, sorry.
Donald Hoffman (2:22:04.240)
Well, we really are looking through a two dimensional screen
Lex Fridman (2:22:07.760)
until it's what we intuit to be a three dimensional world
Lex Fridman (2:22:12.720)
and also inferring dynamic stuff, making it 4D.
Lex Fridman (2:22:17.320)
Anyway, is it possible to visualize some pretty pictures
Lex Fridman (2:22:20.520)
that give us a deeper sense of the truth of reality?
Lex Fridman (2:22:25.840)
I think that we will incrementally be able to do that.
Donald Hoffman (2:22:29.640)
I think that, for example, the picture that we have
Lex Fridman (2:22:33.680)
of electrons and photons interacting and scattering,
Donald Hoffman (2:22:41.680)
it may have not been possible
Lex Fridman (2:22:43.360)
until Faraday did all of his experiments
Lex Fridman (2:22:45.320)
and then Maxwell wrote down his equations.
Lex Fridman (2:22:47.280)
And we were then sort of forced by his equations
Donald Hoffman (2:22:50.160)
to think in a new way.
Lex Fridman (2:22:52.920)
And then when Planck in 1900,
Donald Hoffman (2:22:56.760)
desperate to try to solve the problem
Lex Fridman (2:23:00.920)
of black body radiation,
Lex Fridman (2:23:02.240)
what they call the ultraviolet catastrophe
Lex Fridman (2:23:03.840)
where Newton was predicting infinite energies
Donald Hoffman (2:23:08.320)
where there weren't infinite energies
Lex Fridman (2:23:09.800)
in black body radiation.
Lex Fridman (2:23:11.480)
And he in desperation proposed packets of energy.
Lex Fridman (2:23:20.600)
Then once you've done that,
Lex Fridman (2:23:23.080)
and then you have an Einstein come along five years later
Lex Fridman (2:23:25.280)
and show how that explains the photoelectric effect.
Lex Fridman (2:23:29.360)
And then eventually in 1926, you get quantum theory.
Lex Fridman (2:23:33.560)
And then you get this whole new way of thinking
Donald Hoffman (2:23:35.320)
that was, from the Newtonian point of view,
Lex Fridman (2:23:38.380)
completely contradictory and counterintuitive, certainly.
Lex Fridman (2:23:45.400)
And maybe if Giesen is right, not contradictory.
Lex Fridman (2:23:47.680)
Maybe if you use intuitionist math, they're not contradictory,
Lex Fridman (2:23:50.600)
but still, certainly you wouldn't have gone there.
Lex Fridman (2:23:54.440)
And so here's a case where the experiments
Lex Fridman (2:23:57.740)
and then a desperate mathematical move,
Lex Fridman (2:24:01.760)
sort of we use those as a flashlight into the deep fog.
Lex Fridman (2:24:07.160)
And so that science may be the flashlight into the deep fog.
Lex Fridman (2:24:13.800)
I wonder if it's still possible to visualize
Donald Hoffman (2:24:16.080)
in the, like we talk about consciousness
Lex Fridman (2:24:20.080)
from a self perspective experience it.
Donald Hoffman (2:24:22.680)
Hold that idea in our mind,
Lex Fridman (2:24:25.700)
the way you can experience things directly.
Donald Hoffman (2:24:27.820)
We've evolved to experience things in this 3D world.
Lex Fridman (2:24:33.100)
And that's a very rich experience.
Donald Hoffman (2:24:35.960)
When you're thinking mathematically,
Lex Fridman (2:24:41.100)
you still in the end of the day have to project it down
Donald Hoffman (2:24:44.940)
to a low dimensional space to make conclusions.
Lex Fridman (2:24:49.700)
Your conclusions will be a number or a line
Donald Hoffman (2:24:53.220)
or a plot or a visual.
Lex Fridman (2:24:56.100)
So I wonder like how we can really touch some deep truth
Donald Hoffman (2:25:00.660)
in a subjective way, like experience it,
Lex Fridman (2:25:03.780)
really feel the beauty of it, you know,
Donald Hoffman (2:25:05.940)
in the way that humans feel beauty.
Lex Fridman (2:25:08.580)
Right, are we screwed?
Donald Hoffman (2:25:10.980)
I don't think we're screwed.
Lex Fridman (2:25:11.820)
I think that we get little hints of it
Donald Hoffman (2:25:14.460)
from psychedelic drugs and so forth.
Lex Fridman (2:25:17.580)
We get hints that there are certain interventions
Donald Hoffman (2:25:19.700)
that we can take on our interface.
Lex Fridman (2:25:21.460)
I apply this chemical,
Donald Hoffman (2:25:22.740)
which is just some element of my interface
Lex Fridman (2:25:25.980)
to this other, to a brain I ingested.
Lex Fridman (2:25:29.780)
And all of a sudden I seem like I've opened new portals
Lex Fridman (2:25:33.500)
into conscious experiences.
Donald Hoffman (2:25:36.380)
Well, that's very, very suggestive.
Lex Fridman (2:25:38.360)
That's like the black body radiation doing something
Lex Fridman (2:25:41.540)
that we didn't expect, right?
Lex Fridman (2:25:42.740)
It doesn't go to infinity
Donald Hoffman (2:25:44.100)
when we thought it was gonna go to infinity
Lex Fridman (2:25:45.460)
and we're forced to propose these quanta.
Lex Fridman (2:25:49.260)
So once we have a theory of conscious agents
Lex Fridman (2:25:53.580)
and this projection to space,
Donald Hoffman (2:25:55.260)
I should say, I should sketch what I think
Lex Fridman (2:25:57.180)
that projection is.
Lex Fridman (2:25:59.380)
But then I think we can then start
Lex Fridman (2:26:01.820)
to ask specific questions.
Donald Hoffman (2:26:03.540)
When you're taking DMT or you're taking LSD
Lex Fridman (2:26:08.780)
or something like that,
Donald Hoffman (2:26:10.780)
now that we have this deep model
Lex Fridman (2:26:12.060)
that we've reverse engineered space and time
Lex Fridman (2:26:14.620)
and physical particles,
Lex Fridman (2:26:16.200)
we've pulled them back to this theory of conscious agents.
Donald Hoffman (2:26:18.620)
Now we can ask ourselves in this idealized future,
Lex Fridman (2:26:23.380)
what are we doing to conscious agents
Lex Fridman (2:26:25.160)
when we apply five MEO DMT?
Lex Fridman (2:26:28.280)
What are we doing?
Lex Fridman (2:26:29.900)
Are we opening a new portal, right?
Lex Fridman (2:26:32.300)
So when I say that, I mean,
Donald Hoffman (2:26:33.780)
I have a portal into consciousness
Lex Fridman (2:26:35.980)
that I call my body of Lex Friedman that I'm creating.
Lex Fridman (2:26:39.260)
And it's a genuine portal, not perfect,
Lex Fridman (2:26:41.900)
but it's a genuine portal.
Donald Hoffman (2:26:43.020)
I'm definitely communicating with your consciousness.
Lex Fridman (2:26:45.940)
And we know that we have one technology
Donald Hoffman (2:26:49.340)
for building new portals.
Lex Fridman (2:26:51.440)
We know one technology and that is having kids.
Donald Hoffman (2:26:54.780)
Having kids is how we build new portals into consciousness.
Lex Fridman (2:26:59.220)
It takes a long time.
Lex Fridman (2:27:00.100)
Can you elaborate that?
Lex Fridman (2:27:01.380)
Oh, oh, oh, you mean like?
Donald Hoffman (2:27:04.780)
Your son and your daughter didn't exist.
Lex Fridman (2:27:07.220)
That was a portal.
Donald Hoffman (2:27:08.300)
You're having contact with consciousness
Lex Fridman (2:27:10.460)
that you never would have had before,
Lex Fridman (2:27:12.380)
but now you've got a son or a daughter.
Lex Fridman (2:27:14.100)
You went through this physical process,
Donald Hoffman (2:27:16.940)
they were born, then there was all the training.
Lex Fridman (2:27:19.980)
But is that portal yours?
Lex Fridman (2:27:22.780)
So when you have kids, are you creating new portals
Lex Fridman (2:27:25.300)
that are completely distinct from the portals
Lex Fridman (2:27:27.300)
that you've created with other consciousness?
Lex Fridman (2:27:29.300)
Like can you elaborate on that?
Donald Hoffman (2:27:31.300)
To which degree are the consciousness of your kids
Lex Fridman (2:27:35.380)
a part of you?
Donald Hoffman (2:27:37.340)
Well, so every person that I see,
Lex Fridman (2:27:39.720)
that symbol that I see, the body that I see,
Donald Hoffman (2:27:43.080)
is a portal potentially for me to interact
Lex Fridman (2:27:46.680)
with a consciousness.
Lex Fridman (2:27:50.120)
And each consciousness has a unique character.
Lex Fridman (2:27:53.600)
We call it a personality and so forth.
Lex Fridman (2:27:56.240)
So with each new kid that's born,
Lex Fridman (2:27:59.880)
we come in contact with a personality
Donald Hoffman (2:28:01.960)
that we've never seen before.
Lex Fridman (2:28:03.700)
And a version of consciousness
Donald Hoffman (2:28:05.700)
that we've never seen before.
Lex Fridman (2:28:06.960)
At a deeper level, as I said,
Donald Hoffman (2:28:08.640)
the theory says there's one agent.
Lex Fridman (2:28:10.480)
So this is a different projection of that one agent.
Lex Fridman (2:28:15.240)
But so that's what I mean by a portal
Lex Fridman (2:28:17.960)
is within my own interface, my own projection,
Lex Fridman (2:28:22.960)
can I see other projections of that one consciousness?
Lex Fridman (2:28:29.760)
So can I get portals in that sense?
Lex Fridman (2:28:31.960)
And I think we will get a theory of that,
Lex Fridman (2:28:36.480)
that we will get a theory of portals
Lex Fridman (2:28:38.240)
and then we can ask how the psychedelics are acting.
Lex Fridman (2:28:41.480)
Are they actually creating new portals or not?
Donald Hoffman (2:28:44.140)
If they're not, we should nevertheless then understand
Lex Fridman (2:28:47.960)
how we could create a new portal, right?
Donald Hoffman (2:28:50.220)
Maybe we have to just study what happens
Lex Fridman (2:28:51.880)
when we have kids.
Donald Hoffman (2:28:53.680)
We know that that technology creates new portals.
Lex Fridman (2:28:57.360)
So we have to reverse engineer that and then say,
Lex Fridman (2:28:59.440)
okay, could we somehow create new portals de novo?
Lex Fridman (2:29:04.440)
With something like brain computer interfaces, for example.
Donald Hoffman (2:29:09.440)
Yeah, well, maybe just a chemical or something.
Lex Fridman (2:29:10.700)
It's probably more complicated than a chemical.
Donald Hoffman (2:29:12.540)
That's why I think that the psychedelics may,
Lex Fridman (2:29:15.540)
because they might be affecting this portal
Donald Hoffman (2:29:17.460)
in certain ways that it turns it around and opens up.
Lex Fridman (2:29:20.660)
In other words, maybe once we understand
Lex Fridman (2:29:22.500)
what this thing is a portal, your body is a portal,
Lex Fridman (2:29:25.260)
and understand all of those complexities,
Donald Hoffman (2:29:26.580)
maybe we'll realize that that portal can be shifted
Lex Fridman (2:29:29.180)
to different parts of the deeper consciousness
Lex Fridman (2:29:32.220)
and can be used to create new portals.
Lex Fridman (2:29:34.420)
And give new windows on it.
Lex Fridman (2:29:36.420)
And so in that way, maybe yes,
Lex Fridman (2:29:38.860)
psychedelics could open up new portals
Donald Hoffman (2:29:41.180)
in the sense that they're taking something
Lex Fridman (2:29:42.300)
that's already a complex portal
Lex Fridman (2:29:43.740)
and just tweaking it a bit.
Lex Fridman (2:29:45.980)
Well, but creating is a very powerful difference
Donald Hoffman (2:29:48.600)
between morphing.
Lex Fridman (2:29:50.540)
Right, right, tweaking versus creating, I agree.
Lex Fridman (2:29:53.900)
But maybe it gives you intuition
Lex Fridman (2:29:55.800)
to at least the full space of the kinds of things
Donald Hoffman (2:29:58.840)
that this particular system is capable of.
Lex Fridman (2:30:01.060)
I mean, the idea of the consciousness creates brains.
Donald Hoffman (2:30:05.100)
I mean, that breaks my brain because,
Lex Fridman (2:30:08.540)
I guess I'm still a physicalist in that sense
Donald Hoffman (2:30:12.820)
because it's just much easier to intuit the world.
Lex Fridman (2:30:19.100)
It's practical to think there's a neural network
Lex Fridman (2:30:22.860)
and what are the different ways
Lex Fridman (2:30:25.660)
fascinating capabilities can emerge
Donald Hoffman (2:30:30.660)
from this neural network.
Lex Fridman (2:30:33.420)
I agree, it's easier.
Lex Fridman (2:30:34.420)
And so you start to,
Lex Fridman (2:30:36.460)
and then present to yourself the problem of,
Lex Fridman (2:30:38.580)
okay, well, how does consciousness arise?
Lex Fridman (2:30:40.860)
How does intelligence arise?
Lex Fridman (2:30:42.940)
How does emotion arise?
Lex Fridman (2:30:46.580)
How does memory arise?
Lex Fridman (2:30:49.820)
How do we filter within the system
Lex Fridman (2:30:52.580)
all the incoming sensory information
Donald Hoffman (2:30:54.980)
we're able to allocate attention
Lex Fridman (2:30:57.660)
in different interesting ways?
Lex Fridman (2:30:58.760)
How do all those mechanisms arise?
Lex Fridman (2:31:02.060)
To say that there's other fundamental things
Donald Hoffman (2:31:04.140)
we don't understand outside of space time
Lex Fridman (2:31:06.580)
that are actually core to how this whole thing works
Donald Hoffman (2:31:10.060)
is a bit paralyzing because it's like,
Lex Fridman (2:31:14.140)
oh, we're not 10% done, we're like 0.001% done.
Donald Hoffman (2:31:20.740)
It's the immediate feeling.
Lex Fridman (2:31:23.180)
Certainly understand that.
Donald Hoffman (2:31:24.420)
My attitude about it is,
Lex Fridman (2:31:26.200)
if you look at the young physicists
Donald Hoffman (2:31:29.760)
who are searching for these structures beyond space time,
Lex Fridman (2:31:32.600)
like amplitude and so forth,
Donald Hoffman (2:31:36.320)
they're having a ball.
Lex Fridman (2:31:38.360)
Space time, that's what the old folks did.
Donald Hoffman (2:31:41.340)
That's what the older generation did.
Lex Fridman (2:31:44.280)
We're doing something that really is fun and new
Lex Fridman (2:31:48.520)
and they're having a blast
Lex Fridman (2:31:51.880)
and they're finding all these new structures.
Lex Fridman (2:31:53.840)
So I think that we're going to
Lex Fridman (2:31:59.040)
succeed in getting a new deeper theory.
Donald Hoffman (2:32:03.520)
I can just say what I'm hoping with the theory
Lex Fridman (2:32:05.220)
that I'm working on, I'm hoping to show
Donald Hoffman (2:32:07.880)
that I could have this timeless dynamics of consciousness,
Lex Fridman (2:32:11.000)
no entropic time.
Donald Hoffman (2:32:12.720)
I take a projection and I show how this timeless dynamics
Lex Fridman (2:32:16.240)
looks like the Big Bang
Lex Fridman (2:32:19.000)
and the entire evolution of space time.
Lex Fridman (2:32:21.840)
In other words, I see how my whole space time interface.
Lex Fridman (2:32:25.120)
So not just the projection
Lex Fridman (2:32:28.000)
doesn't just look like space time,
Donald Hoffman (2:32:29.760)
you can explain the whole from the origin of the universe.
Lex Fridman (2:32:34.440)
That's what we have to do
Lex Fridman (2:32:35.720)
and that's what the physicists understand.
Lex Fridman (2:32:37.060)
When they go beyond space time to the amplitude heat
Lex Fridman (2:32:39.200)
and the cosmological polytope,
Lex Fridman (2:32:41.000)
they ultimately know that they have to get back
Donald Hoffman (2:32:43.200)
the Big Bang story and the whole evolution,
Lex Fridman (2:32:46.600)
that whole story where there were no living things.
Donald Hoffman (2:32:49.380)
There was just a point and then the explosion
Lex Fridman (2:32:53.680)
and then just particles at high energy
Lex Fridman (2:32:55.600)
and then eventually the cooling down
Lex Fridman (2:32:57.040)
and the differentiation and finally matter condenses
Lex Fridman (2:33:01.800)
and then life and then consciousness.
Lex Fridman (2:33:03.520)
That whole story has to come out of something
Donald Hoffman (2:33:05.800)
that's deeper and without time.
Lex Fridman (2:33:07.840)
And that's what we're up to.
Lex Fridman (2:33:12.040)
So the whole story that we've been telling ourselves
Lex Fridman (2:33:14.480)
about Big Bang and how brains evolved in consciousness
Donald Hoffman (2:33:17.140)
will come out of a much deeper theory.
Lex Fridman (2:33:18.720)
And yeah, for someone like me, it's a lot.
Lex Fridman (2:33:24.680)
But for the younger generation, this is like, oh wow,
Lex Fridman (2:33:29.480)
all the low cherries aren't picked.
Donald Hoffman (2:33:30.760)
This is really good stuff.
Lex Fridman (2:33:31.900)
This is really new fundamental stuff that we can do.
Lex Fridman (2:33:35.400)
So I can't wait to read the papers of the younger generation
Lex Fridman (2:33:40.120)
and I wanna see them.
Donald Hoffman (2:33:41.940)
Kids these days with their non space time assumptions.
Lex Fridman (2:33:48.980)
It's just interesting looking at the philosophical tradition
Donald Hoffman (2:33:51.660)
of this difficult ideas you struggle with.
Lex Fridman (2:33:53.780)
If you look like somebody like Emmanuel Kant,
Lex Fridman (2:33:57.680)
what are some interesting agreements and disagreements
Lex Fridman (2:34:00.300)
you have with a guy about the nature of reality?
Lex Fridman (2:34:04.900)
So there's a lot in agreement, right?
Lex Fridman (2:34:06.880)
So Kant was an idealist, transcendental idealist
Lex Fridman (2:34:10.960)
and he basically had the idea
Lex Fridman (2:34:15.180)
that we don't see nature as it is.
Donald Hoffman (2:34:19.340)
We impose a structure on nature.
Lex Fridman (2:34:26.020)
And so in some sense, I'm saying something similar.
Donald Hoffman (2:34:29.060)
I'm saying that, by the way,
Lex Fridman (2:34:30.540)
I don't call myself an idealist.
Donald Hoffman (2:34:31.660)
I call myself a conscious realist
Lex Fridman (2:34:33.500)
because idealism has a long history.
Donald Hoffman (2:34:35.620)
A lot of different ideas come under idealism
Lex Fridman (2:34:38.420)
and there's a lot of debates and so forth.
Donald Hoffman (2:34:40.620)
It tends to be identified with, in many cases,
Lex Fridman (2:34:44.540)
anti science and anti realism.
Lex Fridman (2:34:46.660)
And I don't want either connection with my ideas
Lex Fridman (2:34:49.380)
and so I just called mine conscious realism
Donald Hoffman (2:34:51.540)
with an emphasis on realism and not anti realism.
Lex Fridman (2:34:56.440)
But one place where I would, of course,
Donald Hoffman (2:34:58.900)
disagree with Kant was that he thought
Lex Fridman (2:35:00.300)
that Euclidean space time was a priori, right?
Donald Hoffman (2:35:05.560)
We just know that that's false.
Lex Fridman (2:35:07.220)
So he went too far on that.
Lex Fridman (2:35:11.780)
But in general, the idea that we don't start with space time,
Lex Fridman (2:35:16.220)
that space and time is in some sense
Donald Hoffman (2:35:17.740)
the forms of our perceptions.
Lex Fridman (2:35:19.240)
Yes, absolutely.
Lex Fridman (2:35:21.540)
And I would say that there's a lot in common
Lex Fridman (2:35:25.660)
with Berkeley in that regard.
Donald Hoffman (2:35:28.820)
There's a lot of ingenious arguments in Berkeley.
Lex Fridman (2:35:31.740)
Leibniz in his monodology understood very clearly
Donald Hoffman (2:35:36.740)
that the hard problem was not solvable.
Lex Fridman (2:35:38.860)
He posed the hard problem and basically dismissed it.
Donald Hoffman (2:35:41.900)
He just said, you can't do this.
Lex Fridman (2:35:43.740)
And so if he came here and saw where we are,
Donald Hoffman (2:35:47.100)
he said, look, guys, I told you this 300 years ago.
Lex Fridman (2:35:50.100)
And he had his monodology.
Donald Hoffman (2:35:51.300)
He was trying to do something like,
Lex Fridman (2:35:53.900)
it's different from what I'm doing,
Lex Fridman (2:35:56.980)
but he had these things that were not in space and time,
Lex Fridman (2:35:59.100)
these monads.
Donald Hoffman (2:36:00.420)
He was trying to build something.
Lex Fridman (2:36:03.300)
I'm trying to build a theory of conscious agents.
Donald Hoffman (2:36:05.300)
My guess is that if he came here,
Lex Fridman (2:36:08.460)
I could just, if he saw what I was doing,
Donald Hoffman (2:36:10.500)
he would say, he would understand it
Lex Fridman (2:36:13.100)
and immediately take off with it
Lex Fridman (2:36:15.300)
and go places that I couldn't.
Lex Fridman (2:36:17.500)
He would have no problem with this.
Donald Hoffman (2:36:19.780)
Right, there would be overlap of the spirit
Lex Fridman (2:36:22.860)
of the ideas would be totally overlapping.
Lex Fridman (2:36:25.460)
But his genius would then just run with it
Lex Fridman (2:36:26.860)
far faster than I could.
Donald Hoffman (2:36:28.060)
I love the humility here.
Lex Fridman (2:36:29.820)
So let me ask you about sort of practical implications
Donald Hoffman (2:36:32.620)
of your ideas to our world, our complicated world.
Lex Fridman (2:36:36.460)
When you look at the big questions of humanity,
Lex Fridman (2:36:38.780)
of hate, war, what else is there?
Lex Fridman (2:36:46.820)
Evil, maybe there's the positive aspects of that,
Donald Hoffman (2:36:51.500)
of meaning, of love.
Lex Fridman (2:36:54.700)
What is the fact that reality is an illusion perceived?
Lex Fridman (2:36:59.700)
What is the conscious realism when applied to daily life?
Lex Fridman (2:37:07.500)
What kind of impact does it have?
Donald Hoffman (2:37:09.300)
A lot, and it's sort of scary.
Lex Fridman (2:37:15.460)
We all know that life is ephemeral
Lex Fridman (2:37:18.300)
and spiritual traditions have said wake up to the fact
Lex Fridman (2:37:21.060)
that anything that you do here is going to disappear.
Lex Fridman (2:37:24.500)
But it's even more ephemeral than perhaps we've thought.
Lex Fridman (2:37:27.960)
I see this bottle because I create it right now.
Donald Hoffman (2:37:30.900)
As soon as I look away,
Lex Fridman (2:37:33.660)
that data structure has been garbage collected.
Donald Hoffman (2:37:36.060)
That bottle, I have to recreate it every time I look.
Lex Fridman (2:37:38.860)
So I spend all my money and I buy this fancy car.
Donald Hoffman (2:37:42.220)
That car, I have to keep recreating it
Lex Fridman (2:37:44.380)
every time I look at it.
Donald Hoffman (2:37:45.200)
It's that ephemeral.
Lex Fridman (2:37:46.540)
So all the things that we invest ourselves in,
Donald Hoffman (2:37:50.200)
we fight over, we kill each other over,
Lex Fridman (2:37:52.300)
we have wars over, these are all,
Lex Fridman (2:37:55.820)
it's like people in a virtual reality simulation, right?
Lex Fridman (2:37:59.420)
And there's this Porsche and we all see the Porsche.
Donald Hoffman (2:38:03.800)
Well, that Porsche exists when I look at it.
Lex Fridman (2:38:07.100)
I turn my headset and I look at it.
Lex Fridman (2:38:09.080)
And then if Joe turns his headset the right way,
Lex Fridman (2:38:12.220)
he'll see his Porsche.
Donald Hoffman (2:38:13.260)
It's not even the same Porsche that I see.
Lex Fridman (2:38:15.260)
He's creating his own Porsche.
Lex Fridman (2:38:17.780)
So these things are exceedingly ephemeral.
Lex Fridman (2:38:20.460)
And now just imagine saying that that's my Porsche.
Donald Hoffman (2:38:25.460)
Well, you can agree to say that it's your Porsche,
Lex Fridman (2:38:29.380)
but really the Porsche only exists as long as you look.
Lex Fridman (2:38:32.900)
So this all of a sudden,
Lex Fridman (2:38:34.640)
what the spiritual traditions have been saying
Donald Hoffman (2:38:36.600)
for a long, long time,
Lex Fridman (2:38:38.220)
this gets cashed out in mathematically precise science.
Donald Hoffman (2:38:41.380)
It's saying ephemeral, yes.
Lex Fridman (2:38:43.380)
In fact, it lasts for a few milliseconds,
Donald Hoffman (2:38:45.900)
a few hundred milliseconds while you look at it.
Lex Fridman (2:38:47.460)
And then it's gone.
Lex Fridman (2:38:48.740)
So the whole idea, why are we fighting?
Lex Fridman (2:38:53.740)
Why do we hate?
Donald Hoffman (2:38:57.740)
We fight over possessions
Lex Fridman (2:39:01.660)
because we think that we're small little objects
Donald Hoffman (2:39:05.200)
inside this preexisting space time.
Lex Fridman (2:39:07.820)
We assume that that mansion and that car
Donald Hoffman (2:39:11.220)
exists independent of us.
Lex Fridman (2:39:12.980)
And that somehow we, these little things
Donald Hoffman (2:39:16.060)
can have our sense of self and importance
Lex Fridman (2:39:19.540)
enhanced by having that special car
Donald Hoffman (2:39:21.060)
or that special house or that special person.
Lex Fridman (2:39:23.720)
When in fact, it's just the opposite.
Donald Hoffman (2:39:26.980)
You create that mansion every time you look.
Lex Fridman (2:39:29.540)
That's, you're something far deeper than that mansion.
Donald Hoffman (2:39:32.980)
You're the entity which can create that mansion on the fly.
Lex Fridman (2:39:37.140)
And there's nothing to the mansion
Donald Hoffman (2:39:39.860)
except what you create in this moment.
Lex Fridman (2:39:41.660)
So all of a sudden, when you take this point of view,
Donald Hoffman (2:39:46.780)
it has all sorts of implications
Lex Fridman (2:39:49.540)
for how we interact with each other,
Lex Fridman (2:39:51.620)
how we treat each other.
Lex Fridman (2:39:57.260)
And again, a lot of things
Donald Hoffman (2:39:58.740)
that spiritual traditions have said, it's a mixed bag.
Lex Fridman (2:40:02.500)
Spiritual traditions are a mixed bag.
Lex Fridman (2:40:03.740)
So let me just be right up front about that.
Lex Fridman (2:40:05.100)
I'm not promoting any particular,
Lex Fridman (2:40:06.780)
but they do have some insights.
Lex Fridman (2:40:08.380)
Yeah, they have wisdom.
Donald Hoffman (2:40:09.540)
They have certain wisdom.
Lex Fridman (2:40:10.460)
They have, I can point to nonsense.
Donald Hoffman (2:40:12.220)
I won't go into it,
Lex Fridman (2:40:13.060)
but I can also point to lots of nonsense.
Lex Fridman (2:40:14.500)
So the issue is to then to look for the key insights.
Lex Fridman (2:40:19.340)
And I think they have a lot of insights
Donald Hoffman (2:40:21.620)
about the ephemeral nature of objects in space and time
Lex Fridman (2:40:25.580)
and not being attached to them, including our own bodies.
Lex Fridman (2:40:28.820)
And reversing that I'm not this little thing,
Lex Fridman (2:40:31.680)
a little consciousness trapped in the body.
Lex Fridman (2:40:33.700)
And the consciousness itself is only a product of the body.
Lex Fridman (2:40:36.080)
So when the body dies, the consciousness disappears.
Donald Hoffman (2:40:38.900)
It turns completely around.
Lex Fridman (2:40:40.700)
The consciousness is fundamental.
Donald Hoffman (2:40:42.360)
The body, my hand exists right now
Lex Fridman (2:40:46.420)
because I'm looking at it.
Donald Hoffman (2:40:47.940)
My hand is gone.
Lex Fridman (2:40:49.780)
I have no hand.
Donald Hoffman (2:40:50.660)
I have no brain.
Lex Fridman (2:40:52.140)
I have no heart.
Donald Hoffman (2:40:53.220)
If you looked, you'll see a heart.
Lex Fridman (2:40:55.400)
Whatever I am is this really complicated thing
Donald Hoffman (2:41:00.100)
in consciousness.
Lex Fridman (2:41:01.240)
That's what I am.
Donald Hoffman (2:41:03.280)
All the stuff that I thought I was
Lex Fridman (2:41:05.660)
is something that I create on the fly and delete.
Lex Fridman (2:41:07.580)
So this is completely radical restructuring
Lex Fridman (2:41:11.940)
of how we think about possessions, about identity,
Donald Hoffman (2:41:16.940)
about survival of death and so forth.
Lex Fridman (2:41:20.420)
This is completely transformative.
Lex Fridman (2:41:22.740)
But the nice thing is that this whole approach
Lex Fridman (2:41:24.860)
of conscious agents, unlike the spiritual traditions,
Donald Hoffman (2:41:27.660)
which have said in some cases similar things,
Lex Fridman (2:41:30.660)
they've said it imprecisely.
Donald Hoffman (2:41:33.180)
This is mathematics.
Lex Fridman (2:41:34.900)
We can actually now begin to state precisely,
Donald Hoffman (2:41:38.260)
here's the mathematical model of consciousness,
Lex Fridman (2:41:40.140)
conscious agents, here's how it maps onto space time,
Donald Hoffman (2:41:42.400)
which I should sketch really briefly.
Lex Fridman (2:41:44.620)
And here's why things are ephemeral
Lex Fridman (2:41:50.380)
and here's why you shouldn't be worried
Lex Fridman (2:41:52.700)
about the ephemeral nature of things
Donald Hoffman (2:41:54.060)
because you're not a little tiny entity
Lex Fridman (2:41:57.340)
inside space and time, quite the opposite.
Donald Hoffman (2:41:59.940)
You're the author of space and time.
Lex Fridman (2:42:02.280)
The I and the am and the I am
Donald Hoffman (2:42:04.700)
is all kind of emerging through this whole process
Lex Fridman (2:42:07.300)
of evolution and so on that's just surface waves
Lex Fridman (2:42:12.400)
and there's a much deeper ocean
Lex Fridman (2:42:13.740)
that we're trying to figure out here.
Lex Fridman (2:42:15.300)
So how does, you said some of the stuff
Lex Fridman (2:42:18.220)
you're thinking about maps to space time,
Lex Fridman (2:42:19.860)
how does it map to space time?
Lex Fridman (2:42:21.500)
So just a very, very high level and I'll keep it brief.
Donald Hoffman (2:42:25.240)
The structures that the physicists are finding,
Lex Fridman (2:42:28.180)
like the amplituhedron, it turns out
Donald Hoffman (2:42:31.100)
they're just static structure, they're polytopes.
Lex Fridman (2:42:34.340)
But they, remarkably, most of the information in them
Donald Hoffman (2:42:37.940)
is contained in permutation matrices.
Lex Fridman (2:42:40.480)
So it's a matrix, like an end by end matrix
Donald Hoffman (2:42:45.720)
that just has zeros and ones.
Lex Fridman (2:42:49.520)
That contains almost all of the information
Lex Fridman (2:42:51.920)
and you can, they have these plebic graphs
Lex Fridman (2:42:54.160)
and so forth that they use to boot up the scattering.
Donald Hoffman (2:42:56.800)
You can compute those scattering amplitudes
Lex Fridman (2:42:59.000)
almost entirely from these permutation matrices.
Lex Fridman (2:43:03.560)
So that's just, now from my point of view,
Lex Fridman (2:43:07.200)
I have this conscious agent dynamics.
Donald Hoffman (2:43:09.360)
It turns out that the stationary dynamics
Lex Fridman (2:43:12.020)
that I was talking about,
Donald Hoffman (2:43:13.720)
where the entropy isn't increasing,
Lex Fridman (2:43:15.880)
all the stationary dynamics are sketched out
Donald Hoffman (2:43:19.720)
by permutation matrices.
Lex Fridman (2:43:24.200)
So there's so called Burkhoff polytope.
Donald Hoffman (2:43:27.680)
All the vertices of this polytope,
Lex Fridman (2:43:29.800)
all the points are permutation matrices.
Donald Hoffman (2:43:33.280)
All the internal points are Markovian kernels
Lex Fridman (2:43:37.440)
that have the uniform measure as a stationary measure.
Donald Hoffman (2:43:42.600)
Now I need to intuit a little better
Lex Fridman (2:43:44.320)
what the heck you're talking about.
Lex Fridman (2:43:46.000)
So basically, there's some complicated thing
Lex Fridman (2:43:50.680)
going on with the network of conscious agents
Lex Fridman (2:43:54.560)
and that's mappable to this,
Lex Fridman (2:43:56.560)
you're saying a two dimensional matrix
Lex Fridman (2:43:58.560)
that scattering has to do with what?
Lex Fridman (2:44:02.040)
With our perception, like that's like photon stuff?
Donald Hoffman (2:44:05.320)
I mean, I don't know if it's useful
Lex Fridman (2:44:06.520)
to sort of dig into detail.
Donald Hoffman (2:44:09.640)
I'll do just the high level thing.
Lex Fridman (2:44:11.000)
Yes.
Lex Fridman (2:44:11.880)
So the high level is the long term behavior
Lex Fridman (2:44:15.520)
of the conscious agent dynamics.
Lex Fridman (2:44:17.000)
So that's the projection of just looking
Lex Fridman (2:44:18.380)
at the long term behavior.
Donald Hoffman (2:44:20.680)
I'm hoping we'll give rise to the amplituhedron.
Lex Fridman (2:44:23.820)
The amplituhedron then gives rise to space time.
Lex Fridman (2:44:27.720)
So then I can just use their link
Lex Fridman (2:44:29.860)
to go all the way from consciousness
Donald Hoffman (2:44:31.140)
through its asymptotics to,
Lex Fridman (2:44:33.440)
through the amplituhedron into space time
Lex Fridman (2:44:35.280)
and get the map all the way into our interface.
Lex Fridman (2:44:37.880)
And that's why you mentioned the permutation matrix
Donald Hoffman (2:44:39.640)
because it gives you a nice thing to try to generate.
Lex Fridman (2:44:42.440)
That's right, it's the connection with the amplituhedron.
Donald Hoffman (2:44:44.480)
The permutation matrices are the core of the amplituhedron
Lex Fridman (2:44:47.440)
and it turns out they're the core
Donald Hoffman (2:44:49.120)
of the asymptotic description of the conscious agents.
Lex Fridman (2:44:52.280)
So not to sort of bring up the idea of a creator,
Lex Fridman (2:44:54.820)
but I like, first of all, I like video games
Lex Fridman (2:44:57.840)
and you mentioned this kind of simulation idea.
Donald Hoffman (2:45:01.140)
First of all, do you think of it as an interesting idea,
Lex Fridman (2:45:03.120)
this thought experiment that will live in a simulation?
Lex Fridman (2:45:06.440)
And in general, do you think we'll live in a simulation?
Lex Fridman (2:45:10.400)
So the Nick Bostrom's idea about the simulation
Donald Hoffman (2:45:14.320)
is typically couched in a physicalist framework.
Lex Fridman (2:45:17.920)
Yes.
Lex Fridman (2:45:18.840)
So there is the bottom level,
Lex Fridman (2:45:21.000)
there's some programmer in a physical space time
Lex Fridman (2:45:24.280)
and they have a computer that they've programmed
Lex Fridman (2:45:25.760)
really cleverly where they've created conscious entities.
Lex Fridman (2:45:30.600)
So you have the hard problem of consciousness, right?
Lex Fridman (2:45:32.720)
The standard hard problem.
Lex Fridman (2:45:33.560)
How could a computer simulation create a conscious,
Lex Fridman (2:45:36.680)
which isn't explained by that simulation theory.
Lex Fridman (2:45:39.120)
But then the idea is that the next level,
Lex Fridman (2:45:41.120)
the entities that are created in the first level simulation
Donald Hoffman (2:45:46.680)
then can write their own simulations
Lex Fridman (2:45:48.120)
and you get this nesting.
Lex Fridman (2:45:50.440)
So the idea that this is a simulation is fine,
Lex Fridman (2:45:55.780)
but the idea that it starts with a physicalist base,
Donald Hoffman (2:45:58.600)
I think, isn't fine.
Lex Fridman (2:46:00.120)
Well, there's different properties here.
Donald Hoffman (2:46:01.880)
The partial rendering, and to me that's the interesting idea
Lex Fridman (2:46:07.200)
is not whether the entirety of the universe is simulated,
Lex Fridman (2:46:11.360)
but how efficiently can you create interfaces
Lex Fridman (2:46:17.920)
that are convincing to all other entities
Lex Fridman (2:46:21.840)
that can appreciate such interfaces?
Lex Fridman (2:46:24.080)
How little does it take?
Donald Hoffman (2:46:25.920)
Because you said like partial rendering
Lex Fridman (2:46:27.920)
or like temporal, ephemeral rendering of stuff.
Donald Hoffman (2:46:31.800)
Only render the tree falling in the forest
Lex Fridman (2:46:33.920)
when there's somebody there to see it.
Donald Hoffman (2:46:36.340)
It's interesting to think,
Lex Fridman (2:46:38.240)
how can you do that super efficiently
Lex Fridman (2:46:39.760)
without having to render everything?
Lex Fridman (2:46:41.360)
And that to me is one perspective on the simulation,
Donald Hoffman (2:46:44.440)
just like it is with video games,
Lex Fridman (2:46:46.720)
where a video game doesn't have to render
Donald Hoffman (2:46:48.240)
every single thing.
Lex Fridman (2:46:49.480)
It's just the thing that the observer is looking at.
Donald Hoffman (2:46:52.300)
Right, there is actually, that's a very nice question.
Lex Fridman (2:46:55.600)
And there's whole groups of researchers
Donald Hoffman (2:46:58.080)
that are actually studying in virtual reality,
Lex Fridman (2:47:00.660)
what is the sort of minimal requirements on the system?
Lex Fridman (2:47:06.360)
How does it have to operate
Lex Fridman (2:47:07.760)
to give you an immersion experience,
Donald Hoffman (2:47:09.840)
to give you the feeling that you have a body,
Lex Fridman (2:47:12.600)
to get you to take it real?
Lex Fridman (2:47:14.240)
And there's actually a lot of really good work
Lex Fridman (2:47:15.960)
on that right now.
Lex Fridman (2:47:16.800)
And it turns out it doesn't take that much.
Lex Fridman (2:47:18.160)
You do need to get the perception action loop tight
Lex Fridman (2:47:21.600)
and you have to give them the perceptions
Lex Fridman (2:47:25.080)
that they're expecting if you want them to.
Lex Fridman (2:47:26.960)
But if you can lead them along,
Lex Fridman (2:47:30.080)
if you give them perceptions
Donald Hoffman (2:47:31.160)
that are close to what they're expecting,
Lex Fridman (2:47:32.320)
you can then maybe move their reality around a bit.
Donald Hoffman (2:47:35.480)
Yeah, it's a tricky engineering problem,
Lex Fridman (2:47:36.960)
especially when you're trying to create a product
Donald Hoffman (2:47:39.340)
that costs little, but that's,
Lex Fridman (2:47:41.240)
it feels like an engineering problem,
Donald Hoffman (2:47:43.000)
not a deeply scientific problem.
Lex Fridman (2:47:46.000)
Or meaning, obviously it's a scientific problem,
Lex Fridman (2:47:47.660)
but as a scientific problem,
Lex Fridman (2:47:49.040)
it's not that difficult to trick us descendants of apes.
Lex Fridman (2:47:53.320)
But here's a case for just us, you know, our own,
Lex Fridman (2:47:56.440)
if this is a virtual reality
Donald Hoffman (2:47:57.520)
that we're experiencing right now.
Lex Fridman (2:47:58.520)
So here's something you can try for yourself.
Donald Hoffman (2:48:01.680)
If you just close your eyes
Lex Fridman (2:48:04.600)
and look at your experience in front of you,
Donald Hoffman (2:48:08.720)
be aware of your experience in front of you,
Lex Fridman (2:48:09.880)
what you experience is just like a modeled dark gray,
Donald Hoffman (2:48:14.000)
where there's all sort of, there's some dynamics to it,
Lex Fridman (2:48:15.920)
but it's just dark gray.
Lex Fridman (2:48:17.620)
But now I ask you, instead of having your attention forward,
Lex Fridman (2:48:22.620)
put your attention backward.
Lex Fridman (2:48:24.320)
What is it like behind you with your eyes closed?
Lex Fridman (2:48:29.840)
And there, it's like nothing.
Donald Hoffman (2:48:34.000)
It's real.
Lex Fridman (2:48:35.420)
So what is going on here?
Lex Fridman (2:48:37.920)
What am I experiencing back there?
Lex Fridman (2:48:44.120)
Right?
Donald Hoffman (2:48:44.960)
Well, it's, I don't know if it's nothing.
Lex Fridman (2:48:47.160)
It's like, I guess it's the absence of,
Donald Hoffman (2:48:49.800)
it's not even like darkness or something.
Lex Fridman (2:48:51.800)
It's not even darkness.
Donald Hoffman (2:48:53.560)
There's no qualia to it.
Lex Fridman (2:48:58.680)
And yet there is a sense of being.
Lex Fridman (2:49:01.160)
And that's the interesting thing.
Lex Fridman (2:49:02.080)
There's a sense of being back.
Lex Fridman (2:49:03.580)
So I close my, when I put my attention forward,
Lex Fridman (2:49:06.760)
I have the qualia of a gray model thing.
Lex Fridman (2:49:08.900)
But when I put my attention backward,
Lex Fridman (2:49:10.380)
there's no qualia at all, but there is a sense of being.
Donald Hoffman (2:49:13.440)
Yeah.
Lex Fridman (2:49:14.400)
I personally, now you haven't been to that side of the room.
Donald Hoffman (2:49:18.560)
I have been to that side of the room.
Lex Fridman (2:49:20.040)
So for me, memories, I start playing the engine
Donald Hoffman (2:49:25.560)
of memory replay, which is like,
Lex Fridman (2:49:29.680)
I take myself back in time and think about that place
Donald Hoffman (2:49:32.880)
where I was hanging out in that part.
Lex Fridman (2:49:34.560)
That's where I see what I'm behind.
Lex Fridman (2:49:35.800)
So that's an interesting quirk of humans too,
Lex Fridman (2:49:38.720)
we're able to, we're collecting these experiences
Lex Fridman (2:49:41.200)
and we can replay them in interesting ways
Lex Fridman (2:49:43.080)
whenever we feel like it.
Lex Fridman (2:49:44.320)
And it's almost like being there,
Lex Fridman (2:49:46.640)
but not really, but almost.
Donald Hoffman (2:49:49.160)
That's right.
Lex Fridman (2:49:50.480)
And yet we can go our entire lives in this.
Donald Hoffman (2:49:53.160)
You're talking about the minimal thing for VR.
Lex Fridman (2:49:54.720)
We can go our entire lives and not realize
Donald Hoffman (2:49:56.720)
that all of my life, it's been like nothing behind me.
Lex Fridman (2:50:01.720)
Yeah, right.
Donald Hoffman (2:50:03.600)
We're not even aware that all of our lives,
Lex Fridman (2:50:06.280)
if you just pay attention to us behind me,
Donald Hoffman (2:50:10.640)
we're like, oh, holy smoke, it's totally scary.
Lex Fridman (2:50:13.280)
I mean, it's like nothing.
Donald Hoffman (2:50:14.720)
There's no qualia there at all.
Lex Fridman (2:50:16.120)
How did I not notice that my entire life?
Donald Hoffman (2:50:18.360)
We're so immersed in the simulation, we buy it so much.
Lex Fridman (2:50:21.280)
Yeah, I mean, you could see this with children, right?
Donald Hoffman (2:50:24.840)
Though with persistence, you could do the peekaboo game.
Lex Fridman (2:50:28.400)
You can hide from them and appear and they're fully tricked.
Lex Fridman (2:50:32.040)
And in the same way, we're fully tricked.
Lex Fridman (2:50:34.760)
There's nothing behind us and we assume there is.
Lex Fridman (2:50:37.800)
And that's really interesting.
Lex Fridman (2:50:39.320)
These theories are pretty heavy.
Donald Hoffman (2:50:42.320)
You as a human being, as a mortal human being,
Lex Fridman (2:50:46.640)
how has these theories been to you personally?
Donald Hoffman (2:50:49.920)
Like, are there good days and bad days
Lex Fridman (2:50:51.720)
when you wake up and look in the mirror
Lex Fridman (2:50:54.120)
and the fact that you can't see anything behind you?
Lex Fridman (2:50:57.760)
The fact that it's rendered,
Lex Fridman (2:50:58.800)
like, is there interesting quirks?
Lex Fridman (2:51:02.400)
Nietzsche with his, if you gaze long into the abyss,
Donald Hoffman (2:51:05.240)
the abyss gazes into you.
Lex Fridman (2:51:08.120)
How has these theories, these ideas,
Lex Fridman (2:51:10.240)
changed you as a person?
Lex Fridman (2:51:13.360)
It's been very, very difficult.
Lex Fridman (2:51:15.440)
And this stuff is not just abstract theory building
Lex Fridman (2:51:19.760)
because it's about us.
Donald Hoffman (2:51:21.760)
Sometimes I realize that there's this big division in me.
Lex Fridman (2:51:24.200)
My mind is doing all this science
Lex Fridman (2:51:26.600)
and coming up with these conclusions
Lex Fridman (2:51:28.520)
and the rest of me is not integrating.
Donald Hoffman (2:51:30.000)
I'm just like, I don't believe it.
Lex Fridman (2:51:31.600)
I just don't believe this.
Donald Hoffman (2:51:32.440)
I mean, it seems, so as I start to take it seriously,
Lex Fridman (2:51:35.920)
I get scared myself.
Donald Hoffman (2:51:37.000)
It's like, but it's very much,
Lex Fridman (2:51:41.040)
then I read these spiritual traditions
Lex Fridman (2:51:43.000)
and realize they're saying very, very similar things.
Lex Fridman (2:51:45.100)
Like, there's a lot of convergence.
Lex Fridman (2:51:48.120)
So for me, I have,
Lex Fridman (2:51:52.800)
the first time I thought it might be possible
Donald Hoffman (2:51:55.680)
that we're not seeing the truth was in 1986.
Lex Fridman (2:51:59.840)
It was from some mathematics we were doing.
Lex Fridman (2:52:02.580)
And when that hit me, it hit me like a ton of bricks.
Lex Fridman (2:52:05.920)
I had to sit down.
Donald Hoffman (2:52:06.760)
It was, it really, it was scary.
Lex Fridman (2:52:11.200)
It was really a shock to the system.
Lex Fridman (2:52:14.000)
And then to realize that everything
Lex Fridman (2:52:16.560)
that has been important to me,
Donald Hoffman (2:52:18.080)
like, you know, getting a house,
Lex Fridman (2:52:22.400)
getting a car, getting a reputation and so forth.
Donald Hoffman (2:52:26.500)
Well, that car is just like the car I see
Lex Fridman (2:52:28.760)
in the virtual reality.
Donald Hoffman (2:52:29.640)
It's just there when you perceive it and it's not there.
Lex Fridman (2:52:32.600)
So the whole question of, you know,
Lex Fridman (2:52:35.260)
what am I doing and why?
Lex Fridman (2:52:36.560)
What's worthwhile doing in life?
Donald Hoffman (2:52:39.440)
Clearly, getting a big house and getting a big car.
Lex Fridman (2:52:46.080)
I mean, we all knew that we were gonna die.
Lex Fridman (2:52:48.160)
So we tend not to know that.
Lex Fridman (2:52:50.500)
We tend to hide it, especially when we're young.
Donald Hoffman (2:52:52.720)
Before age 30, we don't believe we're gonna die.
Lex Fridman (2:52:54.680)
But we factually maybe know that you kind of
Donald Hoffman (2:52:58.080)
are supposed to, yeah.
Lex Fridman (2:52:59.440)
But they'll figure something out and we'll be the generation
Donald Hoffman (2:53:02.180)
that is the first one that doesn't have to die.
Lex Fridman (2:53:04.240)
That's the kind of thing.
Lex Fridman (2:53:05.180)
But when you really face the fact that you're going to die,
Lex Fridman (2:53:11.440)
and then when I start to look at it from this point of view
Donald Hoffman (2:53:13.360)
that, well, this thing was an interface to begin with.
Lex Fridman (2:53:16.460)
So what I'm really, is what I'm really gonna be doing,
Donald Hoffman (2:53:20.180)
just taking off a headset.
Lex Fridman (2:53:21.600)
So I've been playing in a virtual reality game all day
Lex Fridman (2:53:24.160)
and I got lost in the game and I was fighting over a Porsche.
Lex Fridman (2:53:27.700)
And I shot some guys up and I punctured their tires
Lex Fridman (2:53:31.440)
and I got the Porsche.
Lex Fridman (2:53:33.000)
Now I take the headset off and what was that for?
Donald Hoffman (2:53:35.120)
Nothing, it was just, it was a data structure
Lex Fridman (2:53:37.680)
and the data structure is gone.
Lex Fridman (2:53:39.080)
So all of the wars, the fighting and the reputations
Lex Fridman (2:53:42.960)
and all this stuff, it's just a headset.
Lex Fridman (2:53:49.680)
So my theory says that intellectually,
Lex Fridman (2:53:52.520)
my mind, my emotions rebel all over the place.
Donald Hoffman (2:53:57.520)
It's like I, you know, and so I have to meditate.
Lex Fridman (2:54:02.200)
I meditate a lot.
Lex Fridman (2:54:03.480)
What percent of the day would you say you spend
Lex Fridman (2:54:06.320)
as a physicalist sort of living life,
Lex Fridman (2:54:11.040)
pretending your car matters, your reputation matters?
Lex Fridman (2:54:16.840)
Like how much, what's that Tom Waits song,
Donald Hoffman (2:54:19.580)
I like my town with a little drop of poison.
Lex Fridman (2:54:22.480)
How much poison do you allow yourself to have?
Donald Hoffman (2:54:25.780)
I think my default mode is physicalist.
Lex Fridman (2:54:27.840)
I think that that's just the default.
Donald Hoffman (2:54:30.800)
When I'm not being conscious, consciously attentive.
Lex Fridman (2:54:37.280)
Then intellectually consciously attentive,
Donald Hoffman (2:54:39.000)
because if you're just, you're still,
Lex Fridman (2:54:40.820)
if you're tasting coffee and not thinking
Donald Hoffman (2:54:43.160)
or drinking or just taking in the sunset,
Lex Fridman (2:54:45.560)
you're not being intellectual,
Lex Fridman (2:54:47.440)
but you're still experiencing it.
Lex Fridman (2:54:49.480)
So it's when you turn on the introspective machine,
Donald Hoffman (2:54:53.660)
that's when you can start.
Lex Fridman (2:54:54.960)
And turn off the thinker,
Donald Hoffman (2:54:56.840)
when I actually just start looking without thinking.
Lex Fridman (2:55:00.960)
So that's when I feel like I,
Donald Hoffman (2:55:03.800)
all of a sudden I'm starting to see through.
Lex Fridman (2:55:06.840)
Sort of like, okay, part of the addiction to the interface
Donald Hoffman (2:55:13.280)
is all the stories I'm telling about it.
Lex Fridman (2:55:14.800)
It's really important for me to get that,
Donald Hoffman (2:55:15.960)
really important to do that.
Lex Fridman (2:55:18.040)
So I'm telling all these stories and so I'm all wrapped up.
Donald Hoffman (2:55:21.760)
Almost all of the mind stuff that's going on in my head
Lex Fridman (2:55:24.400)
is about attachment to the interface.
Lex Fridman (2:55:28.720)
And so what I found is that the,
Lex Fridman (2:55:34.520)
essentially the only way to really detach
Donald Hoffman (2:55:37.400)
from the interface is to literally let go
Lex Fridman (2:55:42.320)
of thoughts altogether.
Lex Fridman (2:55:44.360)
And then all of a sudden, even my identity,
Lex Fridman (2:55:49.200)
my whole history, my name, my education,
Donald Hoffman (2:55:52.040)
all this stuff is almost irrelevant
Lex Fridman (2:55:54.760)
because it's just now here is the present moment.
Lex Fridman (2:56:00.600)
And this is the reality right now.
Lex Fridman (2:56:03.840)
And all of that other stuff is an interface story.
Lex Fridman (2:56:07.160)
But this conscious experience right now,
Lex Fridman (2:56:09.520)
this is the only reality as far as I can tell.
Donald Hoffman (2:56:14.260)
The rest of it's a story.
Lex Fridman (2:56:17.160)
But that is, again, not my default.
Donald Hoffman (2:56:20.620)
That is, I have to make a really conscious choice
Lex Fridman (2:56:25.320)
to say, okay, I know intellectually
Donald Hoffman (2:56:28.400)
this is all an interface.
Lex Fridman (2:56:30.440)
I'm gonna take the headset off and so forth.
Lex Fridman (2:56:33.560)
And then immediately sink back into the game
Lex Fridman (2:56:36.540)
and just be out there playing the game and get lost in it.
Lex Fridman (2:56:39.760)
So I'm always lost in the game
Lex Fridman (2:56:41.520)
unless I literally consciously choose to stop thinking.
Donald Hoffman (2:56:46.520)
Isn't it terrifying to acknowledge
Lex Fridman (2:56:50.840)
that, to look beyond the game?
Lex Fridman (2:56:56.240)
Isn't it?
Lex Fridman (2:56:57.680)
It scares the hell out of me.
Donald Hoffman (2:56:59.800)
It really is scary because I'm so attached.
Lex Fridman (2:57:03.400)
I'm attached to this body.
Donald Hoffman (2:57:04.480)
I'm attached to the interface.
Lex Fridman (2:57:05.560)
Are you ever worried about breaking your brain a bit?
Donald Hoffman (2:57:09.840)
Meaning like, it's, I mean, some of these ideas,
Lex Fridman (2:57:14.840)
some of these ideas, when you think about reality,
Donald Hoffman (2:57:17.360)
even with like Einstein, just realizing,
Lex Fridman (2:57:21.280)
you said interface, just realizing that light,
Donald Hoffman (2:57:26.960)
that there's a speed of light
Lex Fridman (2:57:28.160)
and you can't go faster than the speed of light
Lex Fridman (2:57:29.880)
and what kind of things black holes can do with light,
Lex Fridman (2:57:34.680)
even that can mess with your head.
Donald Hoffman (2:57:37.080)
Yes.
Lex Fridman (2:57:38.160)
But that's still space time.
Donald Hoffman (2:57:41.040)
That's a big mess, but it's still just space time.
Lex Fridman (2:57:42.760)
It's still a property of our interface.
Donald Hoffman (2:57:44.760)
That's right.
Lex Fridman (2:57:45.600)
But it's still like, even Einstein realized
Donald Hoffman (2:57:49.880)
that this particular thing,
Lex Fridman (2:57:51.480)
some of the stories we tell ourselves
Donald Hoffman (2:57:53.000)
is constructing interfaces
Lex Fridman (2:57:56.040)
that are oversimplifying the way things work
Donald Hoffman (2:58:00.400)
because it's nice.
Lex Fridman (2:58:01.640)
The stories are nice.
Donald Hoffman (2:58:03.640)
Stories are nice.
Lex Fridman (2:58:04.920)
I mean, just like video games, they're nice.
Lex Fridman (2:58:07.880)
Right, but Einstein was a realist, right?
Lex Fridman (2:58:10.200)
He was a famous realist in the sense
Donald Hoffman (2:58:12.160)
that he was very explicit in a 1935 paper
Lex Fridman (2:58:15.160)
with Podolsky and Rosen, the EPR paper, right?
Donald Hoffman (2:58:19.680)
They said, if without in any way disturbing a system,
Lex Fridman (2:58:26.160)
I can predict with probability one,
Donald Hoffman (2:58:28.640)
the outcome of a measurement,
Lex Fridman (2:58:30.960)
then there exists in reality that element, right?
Donald Hoffman (2:58:36.320)
That value that, and we now know from quantum theory
Lex Fridman (2:58:39.360)
that that's false.
Donald Hoffman (2:58:41.000)
Einstein's idea of local realism is strictly speaking false.
Lex Fridman (2:58:46.600)
Yeah.
Lex Fridman (2:58:47.440)
And so we can predict, we can set up,
Lex Fridman (2:58:50.400)
in quantum theory, you can set up,
Lex Fridman (2:58:52.320)
and there's a paper by Chris Fuchs, quantum Bayesianism,
Lex Fridman (2:58:55.960)
where he scouts this out.
Donald Hoffman (2:58:58.040)
It was done by other people,
Lex Fridman (2:58:58.880)
but he gives a good presentation of this,
Donald Hoffman (2:59:00.840)
where they have a sequence of something
Lex Fridman (2:59:02.360)
like nine different quantum measurements that you can make.
Lex Fridman (2:59:05.640)
And you can predict with probability one
Lex Fridman (2:59:08.160)
what a particular outcome will be,
Lex Fridman (2:59:10.000)
but you can actually prove that it's impossible
Lex Fridman (2:59:14.880)
that the value existed before you made the measurement.
Lex Fridman (2:59:18.200)
So you know with probability one what you're gonna get,
Lex Fridman (2:59:20.240)
but you also know with certainty
Donald Hoffman (2:59:22.480)
that that value was not there
Lex Fridman (2:59:23.560)
until you made the measurement.
Lex Fridman (2:59:25.320)
So we know from quantum theory
Lex Fridman (2:59:27.480)
that the act of observation is an act of fact creation.
Lex Fridman (2:59:32.680)
And that is built into what I'm saying
Lex Fridman (2:59:35.160)
with this theory of consciousness.
Donald Hoffman (2:59:36.720)
If consciousness is fundamental,
Lex Fridman (2:59:38.960)
space time itself is an act of fact creation.
Donald Hoffman (2:59:42.240)
It's an interface that we create, consciousness creates,
Lex Fridman (2:59:44.840)
plus all the objects in it.
Lex Fridman (2:59:46.040)
So local realism is not true.
Lex Fridman (2:59:50.280)
Quantum theory has established that.
Donald Hoffman (2:59:51.840)
Also noncontextual realism is not true.
Lex Fridman (2:59:54.600)
And that fits in perfectly with this idea
Donald Hoffman (2:59:57.200)
that consciousness is fundamental.
Lex Fridman (2:59:59.600)
These things are, these exist as data structures
Donald Hoffman (30:00.120)
for making your judgments.
Lex Fridman (30:02.880)
But even the pointers themselves
Donald Hoffman (30:07.640)
are not the objective reality.
Lex Fridman (30:10.480)
So, and I think Gödel is telling us that,
Donald Hoffman (30:13.600)
not that anything goes, but as you develop
Lex Fridman (30:17.760)
new axiom systems, you will find out what goes
Donald Hoffman (30:19.760)
within that axiom system
Lex Fridman (30:21.640)
and what testable predictions you can make.
Lex Fridman (30:23.720)
So I don't think we're untethered.
Lex Fridman (30:25.800)
We continue to do experiments.
Lex Fridman (30:28.000)
What I think we won't have that we want
Lex Fridman (30:31.160)
is a conceptual understanding
Donald Hoffman (30:34.040)
that gives us a theory of everything
Lex Fridman (30:35.560)
that's final and complete.
Donald Hoffman (30:37.560)
I think that this is, to put it another way,
Lex Fridman (30:40.280)
this is job security for scientists.
Donald Hoffman (30:44.000)
Our job will never be done.
Lex Fridman (30:45.240)
It's job security for neuroscience.
Donald Hoffman (30:47.760)
Because before we thought that when we looked in the brain,
Lex Fridman (30:50.760)
we saw neurons and neural networks
Lex Fridman (30:52.440)
and action potentials and synapses and so forth.
Lex Fridman (30:57.560)
And that was it, that was the reality.
Donald Hoffman (31:00.320)
Now we have to reverse engineer that.
Lex Fridman (31:01.760)
We have to say, what is beyond space time?
Lex Fridman (31:04.760)
What is going on?
Lex Fridman (31:05.760)
What is a dynamical system beyond space time?
Donald Hoffman (31:08.400)
That when we project it into Einstein's space time,
Lex Fridman (31:10.520)
gives us things that look like neurons
Lex Fridman (31:12.360)
and neural networks and synapses.
Lex Fridman (31:15.240)
So we have to reverse engineer it.
Lex Fridman (31:16.520)
So there's gonna be lots more work for neuroscience.
Lex Fridman (31:19.000)
It's gonna be far more complicated
Lex Fridman (31:20.840)
and difficult and challenging.
Lex Fridman (31:23.840)
But that's wonderful, that's what we need to do.
Donald Hoffman (31:26.000)
We thought neurons exist when they are perceived
Lex Fridman (31:28.400)
and they don't.
Donald Hoffman (31:29.320)
In the same way that if I show you,
Lex Fridman (31:31.040)
when I say they don't exist,
Donald Hoffman (31:32.000)
I should be very, very concrete.
Lex Fridman (31:34.440)
If I draw on a piece of paper,
Donald Hoffman (31:36.240)
a little sketch of something that is called the Necker cube,
Lex Fridman (31:40.480)
it's just a little line drawing of a cube, right?
Donald Hoffman (31:42.840)
It's not a flat piece of paper.
Lex Fridman (31:44.000)
If I execute it well, and I show it to you,
Donald Hoffman (31:46.160)
you'll see a 3D cube and you'll see it flip.
Lex Fridman (31:48.360)
Sometimes you'll see one face in front,
Donald Hoffman (31:49.760)
sometimes you'll see the other face in front.
Lex Fridman (31:51.920)
But if I ask you, which face is in front
Lex Fridman (31:54.360)
when you don't look?
Lex Fridman (31:57.200)
The answer is, well, neither face is in front
Donald Hoffman (31:59.600)
because there's no cube.
Lex Fridman (32:01.320)
There's just a flat piece of paper.
Lex Fridman (32:03.200)
So when you look at the piece of paper,
Lex Fridman (32:05.120)
you perceptually create the cube.
Lex Fridman (32:08.440)
And when you look at it,
Lex Fridman (32:09.960)
then you fix one face to be in front and one face to be.
Lex Fridman (32:13.120)
So that's what I mean when I say it doesn't exist.
Lex Fridman (32:16.000)
Space time itself is like the cube.
Donald Hoffman (32:18.120)
It's a data structure that your sensory systems construct,
Lex Fridman (32:21.960)
whatever your sensory systems mean now,
Donald Hoffman (32:23.720)
because we now have to not even take that for granted.
Lex Fridman (32:27.320)
But there are perceptions that you construct on the fly
Lex Fridman (32:31.400)
and they're data structures in a computer science sense,
Lex Fridman (32:34.000)
and you garbage collect them when you don't need them.
Lex Fridman (32:35.680)
So you create them and garbage collect them.
Lex Fridman (32:37.480)
But is it possible that it's mapped well
Lex Fridman (32:40.880)
in some concrete, predictable way to objective reality?
Lex Fridman (32:45.640)
The sheet of paper, this two dimensional space,
Donald Hoffman (32:48.640)
or we can talk about space time,
Lex Fridman (32:51.240)
maps in some way that we maybe don't yet understand,
Lex Fridman (32:55.840)
but we'll one day understand what that mapping is,
Lex Fridman (32:59.620)
but it maps reliably.
Donald Hoffman (33:00.920)
It is tethered in that way.
Lex Fridman (33:02.940)
Well, yes.
Lex Fridman (33:03.780)
And so the new theories that the physicists are finding
Lex Fridman (33:06.240)
beyond space time have that kind of tethering.
Lex Fridman (33:08.120)
So they show precisely how you start with an epileptic hedron
Lex Fridman (33:11.880)
and how you project this high dimensional structure
Donald Hoffman (33:15.260)
into the four dimensions of space time.
Lex Fridman (33:18.280)
So there's a precise procedure that relates the two.
Lex Fridman (33:22.320)
And they're doing the same thing
Lex Fridman (33:23.820)
with the cosmological polytopes.
Lex Fridman (33:25.120)
So they're the ones that are making the most concrete
Lex Fridman (33:29.760)
and fun advances going beyond space time.
Lex Fridman (33:32.840)
And they're tethering it, right?
Lex Fridman (33:35.360)
They say this is precisely the mathematical projection
Donald Hoffman (33:38.440)
from this deeper structure into space time.
Lex Fridman (33:41.600)
One thing I'll say about, as a non physicist,
Lex Fridman (33:44.640)
what I find interesting is that they're finding just geometry,
Lex Fridman (33:48.840)
but there's no notion of dynamics.
Donald Hoffman (33:51.120)
Right now, they're just finding
Lex Fridman (33:52.960)
these static geometric structures, which is impressive.
Lex Fridman (33:57.400)
So I'm not putting them down.
Lex Fridman (33:58.400)
This is what they're doing is unbelievably complicated
Lex Fridman (34:01.160)
and brilliant and adventurous, it's all those things.
Lex Fridman (34:08.280)
And beautiful from a human aesthetic perspective
Donald Hoffman (34:11.720)
because geometry is beautiful.
Lex Fridman (34:12.920)
It's absolutely.
Lex Fridman (34:14.160)
And they're finding symmetries that are true of the data
Lex Fridman (34:16.400)
that can't be seen in space time.
Lex Fridman (34:18.660)
But I'm looking for a theory beyond space time
Lex Fridman (34:22.940)
that's a dynamical theory.
Donald Hoffman (34:25.320)
I would love to find, and we can talk about that
Lex Fridman (34:27.520)
at some point, a theory of consciousness
Donald Hoffman (34:29.560)
in which the dynamics of consciousness itself
Lex Fridman (34:33.100)
will give rise to the geometry
Donald Hoffman (34:35.500)
that the physicists are finding beyond space time.
Lex Fridman (34:37.960)
If we can do that,
Donald Hoffman (34:38.800)
then we'd have a completely different way
Lex Fridman (34:40.440)
of looking at how consciousness is related
Donald Hoffman (34:42.620)
to what we call the brain or the physical world
Lex Fridman (34:45.280)
more generally, right?
Donald Hoffman (34:46.280)
Right now, all of my brilliant colleagues,
Lex Fridman (34:49.280)
well, 99% of them are trying to,
Donald Hoffman (34:53.880)
they're assuming space time is fundamental.
Lex Fridman (34:56.720)
They're assuming that particles are fundamental,
Donald Hoffman (34:59.160)
quarks, gluons, leptons, and so forth.
Lex Fridman (35:02.000)
Elements, atoms, and so forth are fundamental
Lex Fridman (35:04.040)
and that therefore neurons and brains
Lex Fridman (35:06.400)
are part of objective reality.
Lex Fridman (35:08.800)
And that somehow when you get matter
Lex Fridman (35:10.880)
that's complicated enough,
Donald Hoffman (35:12.640)
it will somehow generate conscious experiences
Lex Fridman (35:16.240)
by its functional properties.
Donald Hoffman (35:17.840)
Or if you're panpsychist, maybe you,
Lex Fridman (35:20.520)
in addition to the physical properties of particles,
Donald Hoffman (35:22.680)
you add your consciousness property as well.
Lex Fridman (35:27.260)
And then you combine these physical and conscious properties
Donald Hoffman (35:30.880)
to get more complicated ones.
Lex Fridman (35:32.220)
But they're all doing it within space time.
Donald Hoffman (35:36.360)
All of the work that's being done on consciousness
Lex Fridman (35:38.960)
and its relationship to the brain
Donald Hoffman (35:41.880)
is all assumed something that our best theories
Lex Fridman (35:45.000)
are telling us is doomed, space time.
Lex Fridman (35:46.800)
Why does that particular assumption bother you the most?
Lex Fridman (35:50.440)
So you bring up space time.
Donald Hoffman (35:53.660)
I mean, that's just one useful interface
Lex Fridman (35:56.960)
we've used for a long time.
Donald Hoffman (35:59.680)
Surely there's other interfaces.
Lex Fridman (36:01.720)
Is space time just one of the big ones
Donald Hoffman (36:04.700)
that you, to build up people's intuition
Lex Fridman (36:06.800)
about the fact that they do assume a lot of things strongly?
Donald Hoffman (36:10.360)
Or is it in fact the fundamental flaw
Lex Fridman (36:15.040)
in the way we see the world?
Donald Hoffman (36:17.480)
Well, everything else that we think we know
Lex Fridman (36:20.640)
are things in space time.
Donald Hoffman (36:23.380)
Sure.
Lex Fridman (36:24.220)
And so when you say space time is doomed,
Donald Hoffman (36:27.720)
this is a shot to the heart of the whole framework,
Lex Fridman (36:32.880)
the whole conceptual framework that we've had in science.
Donald Hoffman (36:35.900)
Not to the scientific method,
Lex Fridman (36:37.720)
but to the fundamental ontology
Lex Fridman (36:40.800)
and also the fundamental methodology,
Lex Fridman (36:42.400)
the ontology of space time and its contents,
Lex Fridman (36:45.760)
and the methodology of reductionism,
Lex Fridman (36:47.360)
which is that as we go to smaller scales in space time,
Donald Hoffman (36:51.940)
we will find more and more fundamental laws.
Lex Fridman (36:55.160)
And that's been very useful for space and time for centuries,
Donald Hoffman (36:59.440)
reductionism for centuries.
Lex Fridman (37:01.440)
But now we realize that that's over.
Donald Hoffman (37:04.680)
Reductionism is in fact dead, as is space time.
Lex Fridman (37:08.720)
What exactly is reductionism?
Lex Fridman (37:10.600)
What is the process of reductionism
Lex Fridman (37:13.120)
that is different than some of the physicists
Donald Hoffman (37:17.640)
that you mentioned that are trying to think,
Lex Fridman (37:19.440)
trying to let go of the assumption of space time?
Donald Hoffman (37:22.080)
Looking beyond, isn't that still trying to come up
Lex Fridman (37:24.560)
with a simple model that explains this whole thing?
Lex Fridman (37:27.360)
Isn't it still reducing?
Lex Fridman (37:29.400)
It's a wonderful question,
Donald Hoffman (37:30.240)
because it really helps to clarify two different notions,
Lex Fridman (37:33.120)
which is scientific explanation on the one hand,
Lex Fridman (37:36.520)
and a particular kind of scientific explanation on the other,
Lex Fridman (37:39.840)
which is the reductionist.
Lex Fridman (37:40.860)
So the reductionist explanation is saying,
Lex Fridman (37:43.240)
I will start with things that are smaller in space time
Lex Fridman (37:47.520)
and therefore more fundamental,
Lex Fridman (37:49.320)
where the laws are more fundamental.
Lex Fridman (37:51.080)
So we go to just smaller and smaller scales.
Lex Fridman (37:54.880)
Whereas in science more generally,
Donald Hoffman (37:58.000)
we just say like when Einstein
Lex Fridman (37:59.520)
did the special theory of relativity,
Donald Hoffman (38:01.480)
he's saying, let me have a couple of postulates.
Lex Fridman (38:03.600)
I will assume that the speed of light is universal
Donald Hoffman (38:06.160)
for all observers in uniform motion,
Lex Fridman (38:12.000)
and that the laws of physics,
Lex Fridman (38:13.360)
so if you're for uniform motion are,
Lex Fridman (38:16.780)
that's not a reductionist.
Donald Hoffman (38:18.080)
Those are saying, grant me these assumptions.
Lex Fridman (38:20.020)
I can build this entire concept of space time out of it.
Donald Hoffman (38:23.400)
It's not a reductionist thing.
Lex Fridman (38:24.560)
You're not going to smaller and smaller scales of space.
Donald Hoffman (38:27.780)
You're coming up with these deep, deep principles.
Lex Fridman (38:30.600)
Same thing with his theory of gravity, right?
Lex Fridman (38:33.080)
It's the falling elevator idea, right?
Lex Fridman (38:35.600)
So this is not a reductionist kind of thing.
Donald Hoffman (38:37.760)
It's something different.
Lex Fridman (38:39.800)
So simplification is a bigger thing than just reductionism.
Donald Hoffman (38:45.520)
Reductionism has been a particularly useful
Lex Fridman (38:47.720)
kind of scientific explanation,
Lex Fridman (38:49.520)
for example, in thermodynamics, right?
Lex Fridman (38:51.720)
Where the notion that we have of heat,
Donald Hoffman (38:53.360)
some macroscopic thing like temperature and heat,
Lex Fridman (38:56.640)
it turns out that Neil Boltzmann and others discovered,
Donald Hoffman (38:59.600)
well, hey, if we go to smaller and smaller scales,
Lex Fridman (39:02.000)
we find these things called molecules or atoms.
Lex Fridman (39:04.480)
And if we think of them as bouncing around
Lex Fridman (39:06.440)
and having some kind of energy,
Donald Hoffman (39:08.680)
then what we call heat really can be reduced to that.
Lex Fridman (39:14.720)
And so that's a particularly useful kind of reduction,
Donald Hoffman (39:19.100)
is a useful kind of scientific explanation
Lex Fridman (39:21.400)
that works within a range of scales within space time.
Lex Fridman (39:25.480)
But we know now precisely where that has to stop.
Lex Fridman (39:28.480)
At 10 to the minus 33 centimeters
Lex Fridman (39:30.240)
and 10 to the minus 43 seconds.
Lex Fridman (39:32.720)
And I would be impressed
Donald Hoffman (39:34.360)
if it was 10 to the minus 33 trillion centimeters.
Lex Fridman (39:37.520)
I'm not terribly impressed at 10 to the minus 33 centimeters.
Donald Hoffman (39:43.560)
I don't even know how to comprehend
Lex Fridman (39:44.880)
either of those numbers, frankly.
Donald Hoffman (39:47.360)
Just a small aside,
Lex Fridman (39:49.080)
because I am a computer science person,
Donald Hoffman (39:51.520)
I also find cellular automata beautiful.
Lex Fridman (39:54.120)
And so you have somebody like Stephen Wolfram,
Donald Hoffman (39:57.840)
who recently has been very excitedly exploring
Lex Fridman (3:00:01.840)
when we create them.
Donald Hoffman (3:00:03.760)
As Chris Fuchs says, the act of observation
Lex Fridman (3:00:06.720)
is an act of fact creation.
Lex Fridman (3:00:08.760)
But I must say on a personal level,
Lex Fridman (3:00:12.480)
I'm having to spend,
Donald Hoffman (3:00:16.360)
I spend a couple hours a day
Lex Fridman (3:00:19.920)
just sitting in meditation on this
Lex Fridman (3:00:22.480)
and facing the rebellion in me
Lex Fridman (3:00:27.480)
that goes to the core,
Donald Hoffman (3:00:28.720)
it feels like it goes to the core of my being,
Lex Fridman (3:00:30.360)
rebellion against these ideas.
Lex Fridman (3:00:31.960)
So here it's very, very interesting
Lex Fridman (3:00:33.560)
for me to look at this because,
Lex Fridman (3:00:34.880)
so here I'm a scientist and I'm a person.
Lex Fridman (3:00:37.320)
The science is really clear.
Donald Hoffman (3:00:39.160)
Local realism is false.
Lex Fridman (3:00:40.280)
Noncontextual realism is false.
Donald Hoffman (3:00:42.220)
Space time is doomed.
Lex Fridman (3:00:43.280)
It's very, very clear.
Donald Hoffman (3:00:44.560)
It couldn't be clearer.
Lex Fridman (3:00:47.600)
And my emotions rebel left and right.
Donald Hoffman (3:00:50.240)
When I sit there and say, okay,
Lex Fridman (3:00:52.040)
I am not something in space and time.
Donald Hoffman (3:00:55.080)
Something inside of me says, you're crazy.
Lex Fridman (3:00:57.720)
Of course you are.
Lex Fridman (3:00:58.560)
And I'm completely attached to it.
Lex Fridman (3:01:00.160)
I'm completely attached to all this stuff.
Donald Hoffman (3:01:02.000)
I'm attached to my body.
Lex Fridman (3:01:02.960)
I'm attached to the headset.
Donald Hoffman (3:01:04.640)
I'm attached to my car.
Lex Fridman (3:01:06.920)
I'm attached to people.
Donald Hoffman (3:01:07.760)
I'm attached to all of it.
Lex Fridman (3:01:09.960)
And yet I know as an absolute fact,
Donald Hoffman (3:01:12.800)
I'm gonna walk away from all of it.
Lex Fridman (3:01:14.680)
I'm gonna die.
Donald Hoffman (3:01:19.920)
In fact, I almost died last year.
Lex Fridman (3:01:21.560)
COVID almost killed me.
Donald Hoffman (3:01:24.520)
I sent a goodbye text to my wife.
Lex Fridman (3:01:26.560)
So I thought I was done.
Donald Hoffman (3:01:28.320)
You really did.
Lex Fridman (3:01:29.240)
I sent her a goodbye.
Donald Hoffman (3:01:30.480)
I was in the emergency room and it had attacked my heart
Lex Fridman (3:01:35.480)
and it had been at 190 beats per minute for 36 hours.
Donald Hoffman (3:01:40.400)
I couldn't last much longer.
Lex Fridman (3:01:41.360)
I knew I couldn't, they couldn't stop it.
Lex Fridman (3:01:43.720)
So that was it.
Lex Fridman (3:01:46.440)
So that was it.
Lex Fridman (3:01:47.280)
So I texted her goodbye from the emergency room.
Lex Fridman (3:01:50.440)
I love you, goodbye kind of thing.
Donald Hoffman (3:01:52.240)
Yeah, right.
Lex Fridman (3:01:53.560)
Yeah, that was it.
Donald Hoffman (3:01:54.800)
So, so.
Lex Fridman (3:01:55.640)
Were you afraid?
Lex Fridman (3:01:57.360)
God, it scares the hell out of you, right?
Lex Fridman (3:01:59.120)
But there was, you're just feeling so bad anyway
Donald Hoffman (3:02:02.600)
that sort of you're scared, but you're just feeling so bad
Lex Fridman (3:02:06.800)
that in some sense you just want it to stop anyway.
Lex Fridman (3:02:10.840)
So I've been there and faced it just a year ago.
Lex Fridman (3:02:16.840)
How did that change you, by the way?
Donald Hoffman (3:02:18.840)
Having this intellectual reality that's so challenging
Lex Fridman (3:02:22.600)
that you meditate on, it's just an interface.
Lex Fridman (3:02:25.480)
And one of the hardest things to come to terms with
Lex Fridman (3:02:28.840)
is that that means that it's gonna end.
Lex Fridman (3:02:35.480)
How did that change you having come so close
Lex Fridman (3:02:37.480)
to the reality of it?
Donald Hoffman (3:02:38.320)
It's not just an intellectual reality,
Lex Fridman (3:02:39.920)
it's a reality of death.
Donald Hoffman (3:02:43.760)
It's forced, I've meditated for 20 years now.
Lex Fridman (3:02:47.720)
And I would say averaging three or four hours a day.
Lex Fridman (3:02:52.920)
But it's put a new urgency,
Lex Fridman (3:02:57.440)
but urgency is not the right word
Donald Hoffman (3:02:59.280)
because it's riveted my attention, I'll put it that way.
Lex Fridman (3:03:05.200)
It's really riveted my attention and I've really paid,
Donald Hoffman (3:03:10.520)
I spent a lot more time looking up
Lex Fridman (3:03:12.040)
what spiritual traditions say.
Donald Hoffman (3:03:15.280)
I don't, by the way, again, not taking it with the,
Lex Fridman (3:03:19.800)
take it all with a grain of salt.
Lex Fridman (3:03:21.440)
But on the other hand, I think it's stupid for me
Lex Fridman (3:03:23.400)
to ignore it.
Lex Fridman (3:03:24.320)
So I try to listen to the best ideas
Lex Fridman (3:03:28.880)
and to sort out nonsense from,
Lex Fridman (3:03:32.240)
and we all have to do it for ourselves, right?
Lex Fridman (3:03:34.480)
It's not easy.
Lex Fridman (3:03:35.520)
So what makes sense?
Lex Fridman (3:03:37.360)
And I have the advantage of some science
Lex Fridman (3:03:39.280)
so I can look at what science says
Lex Fridman (3:03:40.800)
and try to compare with spiritual tradition.
Donald Hoffman (3:03:43.080)
I try to sort it out for myself.
Lex Fridman (3:03:46.560)
But then I also look and realize
Donald Hoffman (3:03:48.360)
that there's another aspect to me,
Lex Fridman (3:03:49.600)
which is this whole emotional aspect.
Donald Hoffman (3:03:51.440)
The, I seem to be wired up
Lex Fridman (3:03:56.040)
as evolutionary psychology says I'm wired up, right?
Donald Hoffman (3:04:00.680)
All these defensive mechanisms, you know,
Lex Fridman (3:04:03.400)
I'm inclined to lie if I need to.
Donald Hoffman (3:04:06.040)
I'm inclined to be angry, to protect myself,
Lex Fridman (3:04:10.080)
to have an in group and an out group,
Donald Hoffman (3:04:12.280)
to try to make my reputation as big as possible,
Lex Fridman (3:04:16.800)
to try to demean the out group.
Donald Hoffman (3:04:18.640)
There's all these things
Lex Fridman (3:04:19.480)
that evolutionary psychology is spot on.
Donald Hoffman (3:04:22.320)
It's really bright about the human condition.
Lex Fridman (3:04:25.400)
And yet I think evolution, as I said, evolutionary theory
Donald Hoffman (3:04:29.640)
is a projection of a deeper theory
Lex Fridman (3:04:31.160)
where there may be no competition.
Lex Fridman (3:04:33.920)
So how, so I'm in this very interesting position
Lex Fridman (3:04:37.800)
where I feel like, okay,
Donald Hoffman (3:04:40.320)
according to my own theory, I'm consciousness.
Lex Fridman (3:04:42.200)
And maybe this is what it means
Donald Hoffman (3:04:43.720)
for consciousness to wake up.
Lex Fridman (3:04:46.480)
It's not easy.
Donald Hoffman (3:04:48.400)
It's almost like I have,
Lex Fridman (3:04:52.240)
I feel like I have real skin in the game.
Donald Hoffman (3:04:54.680)
It really is scary.
Lex Fridman (3:04:55.760)
I really was scared when I was about to die.
Donald Hoffman (3:04:58.720)
It really was hard to say goodbye to my wife.
Lex Fridman (3:05:02.360)
It really, it really pained.
Lex Fridman (3:05:04.640)
And to then look at that and then look at the fact
Lex Fridman (3:05:09.360)
that I'm gonna walk away from this anyway
Lex Fridman (3:05:11.560)
and it's just an interface.
Lex Fridman (3:05:12.800)
How do I, so it's trying to put all this stuff together
Lex Fridman (3:05:16.240)
and really grok it, so to speak,
Lex Fridman (3:05:19.320)
not just intellectually, but grok it at an emotional level.
Donald Hoffman (3:05:22.680)
Yeah, what are you afraid of,
Lex Fridman (3:05:23.920)
you silly evolved organism
Lex Fridman (3:05:26.680)
that's gotten way too attached to the interface?
Lex Fridman (3:05:30.000)
What are you really afraid of?
Donald Hoffman (3:05:32.000)
That's right.
Lex Fridman (3:05:33.880)
Is there a...
Donald Hoffman (3:05:34.880)
Very personal, you know, it's very, very personal.
Lex Fridman (3:05:36.640)
Yeah.
Donald Hoffman (3:05:37.480)
Yeah.
Lex Fridman (3:05:38.920)
I mean, speaking of the text,
Lex Fridman (3:05:40.960)
what do you think is this whole love thing?
Lex Fridman (3:05:43.920)
What's the role of love in our human condition?
Donald Hoffman (3:05:49.200)
This interface thing we have,
Lex Fridman (3:05:51.040)
is this somehow interweaved,
Lex Fridman (3:05:53.040)
interconnected with consciousness?
Lex Fridman (3:05:54.640)
This attachment we have to other humans
Lex Fridman (3:05:56.600)
and this deep, like some quality to it
Lex Fridman (3:06:02.920)
that seems very interesting, peculiar.
Donald Hoffman (3:06:07.440)
Well, there are two levels I would think about that.
Lex Fridman (3:06:11.200)
There's love in the sexual sense
Lex Fridman (3:06:12.880)
and there's love in a deeper sense.
Lex Fridman (3:06:15.400)
And in the sexual sense,
Donald Hoffman (3:06:16.960)
we can give an evolutionary account of that and so forth.
Lex Fridman (3:06:20.040)
And I think that's pretty clear to people.
Lex Fridman (3:06:24.480)
In this deeper sense, right?
Lex Fridman (3:06:27.640)
So of course, I love my wife in a sexual sense,
Lex Fridman (3:06:32.320)
but there is a deeper sense as well.
Lex Fridman (3:06:34.040)
When I was saying goodbye to her,
Donald Hoffman (3:06:35.080)
there was a much deeper love that was really at play there.
Lex Fridman (3:06:38.440)
That's one place where I think
Donald Hoffman (3:06:40.560)
that the mixed bag from spiritual traditions
Lex Fridman (3:06:43.200)
has something right.
Donald Hoffman (3:06:44.080)
When they say, love your neighbor as yourself,
Lex Fridman (3:06:46.240)
that in some sense, love is fundamental.
Donald Hoffman (3:06:49.440)
I think that they're onto something,
Lex Fridman (3:06:51.880)
something very, very deep and profound.
Lex Fridman (3:06:54.800)
And every once in a while,
Lex Fridman (3:06:57.480)
I can get a personal glimpse of that,
Donald Hoffman (3:06:58.960)
especially when I'm in the space with no thought.
Lex Fridman (3:07:03.800)
When I can really let go of thoughts,
Donald Hoffman (3:07:06.000)
I get little glimpses of a love
Lex Fridman (3:07:10.200)
in the sense that I'm not separate.
Donald Hoffman (3:07:11.960)
It's a love in the sense that I'm not different from that.
Lex Fridman (3:07:18.440)
If you and I are separate, then I can fight you.
Lex Fridman (3:07:21.240)
But if you and I are the same, if there's a union there.
Lex Fridman (3:07:25.120)
The togetherness of it, yeah.
Lex Fridman (3:07:26.720)
What, who's God?
Lex Fridman (3:07:29.520)
All those gods, the stories that have been told
Donald Hoffman (3:07:32.840)
throughout history, you said through the spiritual traditions.
Lex Fridman (3:07:36.680)
What do you think that is?
Lex Fridman (3:07:37.840)
Is that us trying to find that common thing at the core?
Lex Fridman (3:07:44.320)
Well, in many traditions, not all.
Donald Hoffman (3:07:50.760)
The one I was raised in, so my dad was a Protestant minister.
Lex Fridman (3:07:54.760)
We tend to think of God as a being.
Lex Fridman (3:08:00.000)
But I think that that's not right.
Lex Fridman (3:08:02.760)
I think the closest way to think about God is being, period.
Donald Hoffman (3:08:06.840)
Not a being, but being, the very ground of being itself is God.
Lex Fridman (3:08:12.280)
I think that's the deep, and from my point of view,
Donald Hoffman (3:08:16.280)
that's the ground of consciousness.
Lex Fridman (3:08:17.560)
So the ground of conscious being is what we might call God.
Lex Fridman (3:08:22.400)
But the word God has always been,
Lex Fridman (3:08:25.160)
for example, you don't believe the same God as my God,
Lex Fridman (3:08:27.560)
so I'm gonna fight you, or we'll have wars over,
Lex Fridman (3:08:30.400)
because the being, the specific being that you call God
Donald Hoffman (3:08:34.400)
is different from the being that I call God,
Lex Fridman (3:08:35.880)
and so we fight.
Donald Hoffman (3:08:36.840)
Whereas if it's not a being, but just being,
Lex Fridman (3:08:40.920)
and you and I share a being,
Donald Hoffman (3:08:42.880)
then you and I are not separate,
Lex Fridman (3:08:45.000)
and there's no reason to fight.
Donald Hoffman (3:08:46.840)
We're both part of that one being,
Lex Fridman (3:08:48.720)
and loving you is loving myself,
Donald Hoffman (3:08:51.960)
because we're all part of that one being.
Lex Fridman (3:08:54.000)
The spiritual traditions that point to that,
Donald Hoffman (3:08:57.240)
I think are pointing in a very interesting direction,
Lex Fridman (3:09:01.280)
and that does seem to match with the mathematics
Donald Hoffman (3:09:04.720)
of the conscious agent stuff
Lex Fridman (3:09:05.760)
that I've been working on as well,
Donald Hoffman (3:09:07.360)
that it really fits with that, although that wasn't my goal.
Lex Fridman (3:09:11.520)
Is there, you mentioned,
Donald Hoffman (3:09:15.320)
you mentioned that the young physicists that you talk to,
Lex Fridman (3:09:19.480)
or whose work you follow, have quite a lot of fun
Donald Hoffman (3:09:23.720)
breaking with the traditions of the past,
Lex Fridman (3:09:26.160)
the assumptions of the past.
Lex Fridman (3:09:28.640)
What advice would you give to young people today,
Lex Fridman (3:09:31.160)
in high school, in college, not just physicists,
Lex Fridman (3:09:34.080)
but in general, how to have a career they can be proud of,
Lex Fridman (3:09:38.760)
how they can have a life they can be proud of,
Lex Fridman (3:09:41.440)
how to make their way in the world,
Lex Fridman (3:09:43.000)
from the lessons, from the wins and the losses
Lex Fridman (3:09:45.960)
in your own life, what little insights could you pull out?
Lex Fridman (3:09:50.200)
I would say the universe is a lot more interesting
Donald Hoffman (3:09:53.920)
than you might expect, and you are a lot more special
Lex Fridman (3:09:58.520)
and interesting than you might expect.
Donald Hoffman (3:09:59.960)
You might think that you're just a little, tiny,
Lex Fridman (3:10:03.720)
irrelevant, 100 pound, 200 pound person
Donald Hoffman (3:10:09.400)
in a vast billions of light years across space,
Lex Fridman (3:10:14.240)
and that's not the case.
Donald Hoffman (3:10:15.920)
You are, in some sense, the being that's creating
Lex Fridman (3:10:18.920)
that space all the time, every time you look.
Donald Hoffman (3:10:21.520)
So, waking up to who you really are,
Lex Fridman (3:10:25.280)
outside of space and time,
Donald Hoffman (3:10:27.960)
as the author of space and time,
Lex Fridman (3:10:29.360)
as the author of everything that you see.
Donald Hoffman (3:10:32.040)
The author of space and time, sorry.
Lex Fridman (3:10:36.320)
You're the author of space and time,
Lex Fridman (3:10:38.240)
and I'm the author of space and time,
Lex Fridman (3:10:40.000)
and space and time is just one little data structure.
Donald Hoffman (3:10:42.640)
Many other consciousnesses are creating
Lex Fridman (3:10:44.120)
other data structures, they're authors
Donald Hoffman (3:10:46.000)
of various other things.
Lex Fridman (3:10:48.240)
So, realizing, and then realizing that,
Donald Hoffman (3:10:52.040)
I had this feeling growing up, going to college,
Lex Fridman (3:10:54.920)
reading all these textbooks, oh man, it's all been done.
Donald Hoffman (3:10:59.080)
If I'd just been there 50 years ago,
Lex Fridman (3:11:00.560)
I could have discovered this stuff,
Lex Fridman (3:11:01.520)
but it's all in the textbooks now.
Lex Fridman (3:11:03.800)
Well, believe me, the textbooks are gonna look silly
Donald Hoffman (3:11:07.240)
in 50 years, and it's your chance
Lex Fridman (3:11:10.160)
to write the new textbook.
Donald Hoffman (3:11:11.160)
So, of course, study the current textbooks.
Lex Fridman (3:11:14.600)
You have to understand them.
Donald Hoffman (3:11:15.920)
There's no way to progress until you understand
Lex Fridman (3:11:19.200)
what's been done, but then,
Donald Hoffman (3:11:23.440)
the only limit is your imagination, frankly.
Lex Fridman (3:11:26.000)
That's the only limit.
Donald Hoffman (3:11:26.840)
The greatest books, the greatest textbooks
Lex Fridman (3:11:29.160)
ever written on Earth are yet to be written.
Donald Hoffman (3:11:31.880)
Exactly.
Lex Fridman (3:11:35.080)
What do you think is the meaning of this whole thing?
Lex Fridman (3:11:36.880)
What's the meaning of life from your limited interface?
Lex Fridman (3:11:40.160)
Can you figure it all out, like why?
Donald Hoffman (3:11:43.400)
So, you said the universe is kind of trying to figure
Lex Fridman (3:11:46.040)
itself out through us.
Lex Fridman (3:11:49.000)
Why?
Lex Fridman (3:11:50.320)
Why?
Donald Hoffman (3:11:54.040)
Yeah, that's the closest I've come.
Lex Fridman (3:11:55.400)
So, I'll give you, so I will say that I don't know,
Lex Fridman (3:12:00.400)
but here's my guess, right?
Lex Fridman (3:12:02.680)
That's a good first sentence.
Donald Hoffman (3:12:03.920)
That's a good starting point.
Lex Fridman (3:12:05.200)
And maybe that's gonna be a profound part
Donald Hoffman (3:12:08.600)
of the final answer is to start with the I don't know.
Lex Fridman (3:12:10.800)
It's quite possible that that's really important
Donald Hoffman (3:12:13.760)
to start with the I don't know.
Lex Fridman (3:12:15.240)
My guess is that if consciousness is fundamental
Lex Fridman (3:12:18.600)
and if Gödel's incompleteness theorem holds here,
Lex Fridman (3:12:22.960)
and there's infinite variety of structures
Donald Hoffman (3:12:27.600)
for consciousness to some sense explore,
Lex Fridman (3:12:34.240)
that maybe that's what it's about.
Donald Hoffman (3:12:37.080)
This is something that Annika and I talked about a little bit
Lex Fridman (3:12:39.400)
and she doesn't like this way of talking about it.
Lex Fridman (3:12:40.840)
And so I'm gonna have to talk with her some more
Lex Fridman (3:12:42.680)
about this way of talking.
Lex Fridman (3:12:43.920)
But right now I'll just put it this way
Lex Fridman (3:12:45.440)
and I'll have to talk with her more
Lex Fridman (3:12:46.440)
and see if I can say it more clearly.
Lex Fridman (3:12:48.920)
But the way I'm talking about it now is that
Donald Hoffman (3:12:55.680)
there's a sense in which there's being
Lex Fridman (3:13:01.640)
and then there's the experiences or forms
Donald Hoffman (3:13:03.840)
that come out of being.
Lex Fridman (3:13:05.680)
That's one deep, deep mystery.
Lex Fridman (3:13:09.800)
And the question that you asked, what is it all about?
Lex Fridman (3:13:13.680)
Somehow it's related to that.
Lex Fridman (3:13:15.240)
Why does being, why doesn't it just stay without any forms?
Lex Fridman (3:13:19.600)
Why do we have experiences?
Lex Fridman (3:13:22.560)
Why not just have, when you close your eyes
Lex Fridman (3:13:26.680)
and pay attention to what's behind you, there's nothing.
Lex Fridman (3:13:30.040)
But there's being.
Lex Fridman (3:13:30.960)
So why don't we just stop there?
Lex Fridman (3:13:35.960)
Why didn't we just stop there?
Lex Fridman (3:13:37.000)
Why did we create all tables and chairs
Lex Fridman (3:13:39.400)
and the sun and moon and people?
Lex Fridman (3:13:41.720)
All this really complicated stuff, why?
Lex Fridman (3:13:44.720)
And all I can guess right now,
Lex Fridman (3:13:49.440)
and I'll probably kick myself in a couple of years
Lex Fridman (3:13:51.840)
and say that was dumb, but all I can guess right now
Lex Fridman (3:13:53.960)
is that somehow consciousness wakes up to itself
Donald Hoffman (3:13:57.880)
by knowing what it's not.
Lex Fridman (3:13:59.080)
So here I am, I'm not this body.
Lex Fridman (3:14:02.280)
And I sort of saw that, it was sort of in my face
Lex Fridman (3:14:05.640)
when I sent a text goodbye.
Lex Fridman (3:14:08.320)
But then as soon as I'm better, it's sort of like,
Lex Fridman (3:14:10.520)
okay, I sort of don't wanna go there, right?
Donald Hoffman (3:14:13.320)
I, okay, so I just, so I am my body.
Lex Fridman (3:14:17.560)
I go back to the standard thing, I am my body
Lex Fridman (3:14:19.640)
and then I want to get that car.
Lex Fridman (3:14:21.560)
And even though I was just about to die a year ago,
Lex Fridman (3:14:24.080)
so that comes rushing back.
Lex Fridman (3:14:26.000)
So consciousness immerses itself fully
Donald Hoffman (3:14:31.760)
into a particular headset.
Lex Fridman (3:14:36.440)
Gets lost in it and then slowly wakes up.
Donald Hoffman (3:14:39.760)
Just so it can escape and that is the waking up,
Lex Fridman (3:14:41.920)
but it needs to have a negative.
Donald Hoffman (3:14:43.240)
It needs to know what it's not.
Lex Fridman (3:14:44.960)
It needs to know what you are.
Donald Hoffman (3:14:47.720)
You have to say, oh, I'm not that, I'm not that.
Lex Fridman (3:14:49.600)
That wasn't important, that wasn't important.
Donald Hoffman (3:14:52.600)
That's really powerful.
Lex Fridman (3:14:53.720)
Don, let me just say that because I've been
Donald Hoffman (3:14:57.760)
a long term fan of yours and we're supposed
Lex Fridman (3:15:01.120)
to have a conversation during this very difficult moment
Donald Hoffman (3:15:03.840)
in your life, let me just say you're a truly special person
Lex Fridman (3:15:06.840)
and I for one, I know there's a lot of others that agree.
Donald Hoffman (3:15:10.800)
I'm glad that you're still here with us on this earth
Lex Fridman (3:15:13.720)
if for a short time.
Lex Fridman (3:15:17.360)
So whatever, whatever the universe,
Lex Fridman (3:15:21.760)
whatever plan it has for you that brought you close
Donald Hoffman (3:15:25.480)
to death to maybe enlighten you some kind of way,
Lex Fridman (3:15:30.000)
I think it has an interesting plan for you.
Donald Hoffman (3:15:34.040)
You're one of the truly special humans
Lex Fridman (3:15:35.960)
and it's a huge honor that you would sit
Lex Fridman (3:15:37.480)
and talk with me today.
Lex Fridman (3:15:38.800)
Thank you so much.
Donald Hoffman (3:15:39.640)
Thank you very much, Lex.
Lex Fridman (3:15:40.520)
I really appreciate that, thank you.
Donald Hoffman (3:15:42.760)
Thanks for listening to this conversation
Lex Fridman (3:15:44.200)
with Donald Hoffman.
Donald Hoffman (3:15:45.480)
To support this podcast, please check out our sponsors
Lex Fridman (3:15:48.200)
in the description and now, let me leave you
Donald Hoffman (3:15:50.880)
with some words from Albert Einstein,
Lex Fridman (3:15:53.120)
relevant to the ideas discussed in this conversation.
Donald Hoffman (3:15:56.960)
Time and space are modes by which we think
Lex Fridman (3:16:01.240)
and not conditions in which we live.
Donald Hoffman (3:16:03.680)
Thank you for listening and hope to see you next time.
Lex Fridman (40:02.360)
a proposal for a data structure
Donald Hoffman (40:04.320)
that could be the numbers that would make you
Lex Fridman (40:07.600)
a little bit happier in terms of scale,
Donald Hoffman (40:09.240)
because they're very, very, very, very tiny.
Lex Fridman (40:12.680)
So do you like this space of exploration
Donald Hoffman (40:15.480)
of really thinking, letting go of space time,
Lex Fridman (40:18.600)
letting go of everything and trying to think
Lex Fridman (40:20.280)
what kind of data structures
Lex Fridman (40:21.660)
could be underneath this whole mess?
Donald Hoffman (40:23.780)
That's right.
Lex Fridman (40:24.620)
So if they're thinking about these as outside of space time,
Donald Hoffman (40:27.800)
then that's what we have to do.
Lex Fridman (40:29.120)
That's what our best theories are telling us.
Donald Hoffman (40:30.560)
You now have to think outside of space time.
Lex Fridman (40:32.600)
Now, of course, I should back up and say,
Lex Fridman (40:36.520)
we know that Einstein surpassed Newton, right?
Lex Fridman (40:40.360)
But that doesn't mean that there's not good work
Donald Hoffman (40:41.960)
to do on Newton.
Lex Fridman (40:42.940)
There's all sorts of Newtonian physics
Donald Hoffman (40:44.300)
that takes us to the moon and so forth,
Lex Fridman (40:46.220)
and there's lots of good problems
Donald Hoffman (40:47.240)
that we want to solve with Newtonian physics.
Lex Fridman (40:49.960)
The same thing will be true of space time.
Donald Hoffman (40:52.200)
It's not like we're gonna stop using space time.
Lex Fridman (40:53.960)
We'll continue to do all sorts of good work there.
Lex Fridman (40:56.420)
But for those scientists who are really looking
Lex Fridman (40:59.720)
to go deeper, to actually find the next,
Donald Hoffman (41:04.240)
just like what Einstein did to Newton,
Lex Fridman (41:06.160)
what are we gonna do to Einstein?
Lex Fridman (41:07.440)
How do we get beyond Einstein and quantum theory
Lex Fridman (41:09.800)
to something deeper?
Donald Hoffman (41:10.900)
Then we have to actually let go.
Lex Fridman (41:13.280)
And if we're gonna do like this automata kind of approach,
Donald Hoffman (41:18.800)
it's critical that it's not automata in space time,
Lex Fridman (41:21.180)
it's automata prior to space time,
Donald Hoffman (41:23.580)
from which we're going to show how space time emerges.
Lex Fridman (41:25.880)
If you're doing automata within space time,
Donald Hoffman (41:28.240)
well, that might be a fun model,
Lex Fridman (41:29.640)
but it's not the radical new step that we need.
Donald Hoffman (41:33.520)
Yeah, so the space time emerges from that whatever system.
Lex Fridman (41:36.760)
Like you're saying, it's a dynamical system.
Donald Hoffman (41:39.600)
Do we even have an understanding what dynamical means
Lex Fridman (41:42.520)
when we go beyond?
Donald Hoffman (41:45.480)
When you start to think about dynamics,
Lex Fridman (41:48.080)
it could mean a lot of things.
Donald Hoffman (41:50.360)
Even causality could mean a lot of things
Lex Fridman (41:53.000)
if we realize that everything's an interface.
Donald Hoffman (41:58.320)
Like how much do we really know is an interesting question.
Lex Fridman (42:01.400)
Because you brought up neurons,
Donald Hoffman (42:02.520)
I gotta ask you yet another tangent.
Lex Fridman (42:05.440)
There's a paper I remember a while ago looking at
Lex Fridman (42:07.900)
called Could a Neuroscientist Understand a Microprocessor?
Lex Fridman (42:11.480)
And I just enjoyed that thought experiment
Donald Hoffman (42:14.160)
that they provided, which is they basically,
Lex Fridman (42:16.740)
it's a couple of neuroscientists,
Donald Hoffman (42:18.880)
Eric Jonas and Conrad Cording,
Lex Fridman (42:22.320)
who use the tools of neuroscience
Donald Hoffman (42:24.840)
to analyze a microprocessor, so a computer chip.
Lex Fridman (42:30.400)
Now, if we lesion it here, what happens and so forth,
Lex Fridman (42:32.200)
and if you go and lesion in a computer,
Lex Fridman (42:35.400)
it's very, very clear that lesion experiments on computers
Donald Hoffman (42:38.280)
are not gonna give you a lot of insight into how it works.
Lex Fridman (42:40.440)
And also the measurement devices and the kind of sort of,
Donald Hoffman (42:42.660)
just using the basic approaches of neuroscience,
Lex Fridman (42:44.560)
collecting the data, trying to intuit
Donald Hoffman (42:47.160)
about the underlying function of it.
Lex Fridman (42:49.420)
And that helps you understand that
Donald Hoffman (42:52.540)
our scientific exploration of concepts,
Lex Fridman (42:57.720)
depending on the field,
Donald Hoffman (43:00.280)
are maybe in the very, very early stages.
Lex Fridman (43:05.000)
I wouldn't say it leaves us astray,
Donald Hoffman (43:08.480)
perhaps it does sometimes,
Lex Fridman (43:09.560)
but it's not anywhere close to some fundamental mechanism
Donald Hoffman (43:14.680)
that actually makes a thing work.
Lex Fridman (43:16.440)
I don't know if you can sort of comment on that
Donald Hoffman (43:18.960)
in terms of using neuroscience
Lex Fridman (43:20.440)
to understand the human mind and neurons.
Donald Hoffman (43:24.200)
Are we really far away potentially
Lex Fridman (43:26.320)
from understanding in the way we understand
Lex Fridman (43:30.480)
the transistors enough to be able to build a computer?
Lex Fridman (43:33.720)
So one thing about understanding
Donald Hoffman (43:37.960)
is you can understand for fun.
Lex Fridman (43:40.600)
The other one is to understand so you could build things.
Lex Fridman (43:45.600)
And that's when you really have to understand.
Lex Fridman (43:49.120)
Exactly.
Donald Hoffman (43:49.960)
In fact, what got me into the field at MIT
Lex Fridman (43:53.880)
was work by David Marr on this very topic.
Lex Fridman (43:57.620)
So David Marr was a professor at MIT,
Lex Fridman (43:59.820)
but he'd done his PhD in neuroscience,
Donald Hoffman (44:02.040)
studying just the architectures of the brain.
Lex Fridman (44:05.340)
But he realized that his work, it was on the cerebellum.
Donald Hoffman (44:10.040)
He realized that his work, as rigorous as it was,
Lex Fridman (44:15.040)
left him unsatisfied
Donald Hoffman (44:16.600)
because he didn't know what the cerebellum was for
Lex Fridman (44:19.440)
and why it had that architecture.
Lex Fridman (44:21.680)
And so he went to MIT and he was in the AI lab there.
Lex Fridman (44:25.480)
And he said, he had this three level approach
Donald Hoffman (44:29.600)
that really grabbed my attention.
Lex Fridman (44:30.680)
So when I was an undergrad at UCLA,
Donald Hoffman (44:32.640)
I read one of his papers in a class and said,
Lex Fridman (44:34.760)
who is this guy?
Donald Hoffman (44:35.600)
Because he said, you have to have a computational theory.
Lex Fridman (44:37.440)
What is being computed and why?
Lex Fridman (44:40.320)
An algorithm, how is it being computed?
Lex Fridman (44:42.360)
What are the precise algorithms?
Lex Fridman (44:44.720)
And then the hardware,
Lex Fridman (44:45.920)
how does it get instantiated in the hardware?
Lex Fridman (44:47.800)
And so to really do neuroscience, he argued,
Lex Fridman (44:50.400)
we needed to have understanding at all those levels.
Lex Fridman (44:52.760)
And that really got me.
Lex Fridman (44:54.400)
I loved the neuroscience, but I realized this guy was saying,
Donald Hoffman (44:57.080)
if you can't build it, you don't understand it effectively.
Lex Fridman (45:00.080)
And so that's why I went to MIT.
Lex Fridman (45:02.000)
And I had the pleasure of working with David
Lex Fridman (45:04.480)
until he died just a year and a half later.
Lex Fridman (45:09.180)
So there's been that idea that with neuroscience,
Lex Fridman (45:12.760)
we have to have, in some sense, a top down model
Donald Hoffman (45:15.880)
of what's being computed and why
Lex Fridman (45:18.940)
that we would then go after.
Lex Fridman (45:20.000)
And the same thing with the, you know,
Lex Fridman (45:21.480)
trying to reverse engineer a computing system
Donald Hoffman (45:24.240)
like your laptop.
Lex Fridman (45:25.520)
We really need to understand
Lex Fridman (45:27.640)
what the user interface is about
Lex Fridman (45:29.120)
and what are keys on the keyboard for and so forth.
Donald Hoffman (45:34.240)
You need to know why to really understand
Lex Fridman (45:37.440)
all the circuitry and what it's for.
Donald Hoffman (45:40.320)
Now, we don't, evolution of a natural selection
Lex Fridman (45:46.560)
does not tell us the deeper question that we're asking,
Lex Fridman (45:51.760)
the answer to the deeper question, which is why?
Lex Fridman (45:53.680)
What's this deeper reality and what's it up to and why?
Donald Hoffman (45:59.600)
All it tells us is that whatever reality is,
Lex Fridman (46:04.120)
it's not what you see.
Lex Fridman (46:05.500)
What you see is just an adaptive fiction.
Lex Fridman (46:12.180)
So just to linger on this fascinating, bold question
Donald Hoffman (46:15.420)
that shakes you out of your dream state.
Lex Fridman (46:18.980)
Does this fiction still help you in building intuitions
Lex Fridman (46:23.420)
as literary fiction does about reality?
Lex Fridman (46:27.780)
The reason we read literary fiction
Donald Hoffman (46:30.260)
is it helps us build intuitions and understanding
Lex Fridman (46:36.260)
in indirect ways sneak up to the difficult questions
Donald Hoffman (46:39.140)
of human nature, great fiction.
Lex Fridman (46:41.900)
Same with this observed reality.
Donald Hoffman (46:46.100)
Does this interface that we get, this fictional interface,
Lex Fridman (46:49.180)
help us build intuition about deeper truths
Lex Fridman (46:52.820)
of how this whole mess works?
Lex Fridman (46:55.020)
Well, I think that each theory that we propose
Lex Fridman (46:58.900)
will give its own answer to that question, right?
Lex Fridman (47:01.100)
So when the physicists are proposing these structures
Donald Hoffman (47:05.300)
like the amplituhedron and cosmological polytope,
Lex Fridman (47:08.260)
associahedron and so forth beyond space time,
Donald Hoffman (47:11.100)
we can then ask your question for those specific structures
Lex Fridman (47:14.200)
and say, how much information, for example,
Donald Hoffman (47:17.620)
does evolution by natural selection
Lex Fridman (47:19.240)
and the kinds of sensory systems that we have right now
Lex Fridman (47:24.540)
give us about this deeper reality?
Lex Fridman (47:26.940)
And why did we evolve this way?
Donald Hoffman (47:30.020)
We can try to answer that question from within the deep.
Lex Fridman (47:33.020)
So there's not gonna be a general answer.
Donald Hoffman (47:34.940)
I think what we'll have to do is posit
Lex Fridman (47:37.980)
these new deeper theories
Lex Fridman (47:39.300)
and then try to answer your question
Lex Fridman (47:41.480)
within the framework of those deeper theories,
Donald Hoffman (47:43.580)
knowing full well that there'll be an even deeper theory.
Lex Fridman (47:47.140)
So is this paralyzing though?
Donald Hoffman (47:49.860)
Because how do we know we're not completely adrift
Lex Fridman (47:53.380)
out to sea, lost forever from,
Lex Fridman (47:57.700)
so like that our theories are completely lost.
Lex Fridman (48:00.080)
So if it's all,
Donald Hoffman (48:04.240)
if we can never truly deeply introspect to the bottom,
Lex Fridman (48:09.220)
if it's always just turtles on top of turtles infinitely,
Lex Fridman (48:14.380)
isn't that paralyzing for the scientific mind?
Lex Fridman (48:18.360)
Well, it's interesting that you say introspect
Donald Hoffman (48:20.780)
to the bottom.
Lex Fridman (48:21.620)
Because there is one,
Donald Hoffman (48:26.940)
again, this isn't the same spirit of what I said before,
Lex Fridman (48:28.900)
which is it depends on what answer you give
Donald Hoffman (48:31.000)
to what's beyond space time,
Lex Fridman (48:32.800)
what answer we would give to your question, right?
Donald Hoffman (48:35.180)
So, but one answer that is interesting to explore
Lex Fridman (48:39.660)
is something that spiritual traditions have said
Donald Hoffman (48:41.260)
for thousands of years, but haven't said precisely.
Lex Fridman (48:43.640)
So we can't take it seriously in science
Donald Hoffman (48:45.860)
until it's made precise,
Lex Fridman (48:46.900)
but we might be able to make it precise.
Lex Fridman (48:49.120)
And that is that they've also said something like
Lex Fridman (48:53.140)
space and time aren't fundamental,
Donald Hoffman (48:54.780)
they're Maya, they're illusion.
Lex Fridman (48:56.820)
And, but that if you look inside, if you introspect
Lex Fridman (49:03.240)
and let go of all of your particular perceptions,
Lex Fridman (49:07.260)
you will come to something that's beyond conceptual thought.
Lex Fridman (49:11.260)
And that is, they claim,
Lex Fridman (49:15.340)
being in contact with the deep ground of being
Donald Hoffman (49:17.580)
that transcends any particular conceptual understanding.
Lex Fridman (49:21.220)
If that is correct, and I'm not saying it's correct,
Donald Hoffman (49:24.340)
but, and I'm not saying it's not correct,
Lex Fridman (49:26.020)
I'm just saying, if that's correct,
Donald Hoffman (49:28.340)
then it would be the case that as scientists,
Lex Fridman (49:30.740)
because we also are in touch with this ground of being,
Donald Hoffman (49:34.100)
we would then not be able
Lex Fridman (49:36.980)
to conceptually understand ourselves all the way,
Lex Fridman (49:40.140)
but we could know ourselves just by being ourselves.
Lex Fridman (49:43.540)
And so we would, there would be a sense
Donald Hoffman (49:46.100)
in which there is a fundamental grounding
Lex Fridman (49:48.900)
to the whole enterprise,
Donald Hoffman (49:50.840)
because we're not separate from the enterprise.
Lex Fridman (49:53.320)
This is the opposite of the impersonal third person science.
Donald Hoffman (49:57.380)
This would make science go personal all the way down.
Lex Fridman (50:01.620)
And, but nevertheless, scientific,
Donald Hoffman (50:04.020)
because the scientific method would still be
Lex Fridman (50:07.020)
what we would use all the way down
Donald Hoffman (50:09.320)
for the conceptual understanding.
Lex Fridman (50:10.420)
Unfortunately, I still don't know
Donald Hoffman (50:11.460)
if you went all the way down.
Lex Fridman (50:12.780)
It's possible that this kind of whatever consciousness is
Lex Fridman (50:15.940)
and we'll talk about it,
Lex Fridman (50:17.300)
is getting the cliche statement of be yourself.
Donald Hoffman (50:24.860)
It is somehow digging at a deeper truth of reality,
Lex Fridman (50:28.520)
but you still don't know when you get to the bottom.
Donald Hoffman (50:31.940)
A lot of people, they'll take psychedelic drugs
Lex Fridman (50:34.500)
and they'll say, well, that takes my mind to certain places
Donald Hoffman (50:37.980)
where it feels like that is revealing
Lex Fridman (50:41.140)
some deeper truth of reality,
Lex Fridman (50:43.120)
but you still, it could be interfaces on top of interfaces.
Lex Fridman (50:46.860)
That's, in your view of this, you really don't know.
Donald Hoffman (50:52.620)
I mean, it's Gato's incompleteness
Lex Fridman (50:54.100)
is that you really don't know.
Donald Hoffman (50:55.620)
My own view on it, for what it's worth,
Lex Fridman (50:59.740)
because I don't know the right answer,
Lex Fridman (51:00.700)
but my own view on it right now is that it's never ending.
Lex Fridman (51:05.780)
I think that there will never,
Donald Hoffman (51:07.220)
that this is great, as I said before,
Lex Fridman (51:09.020)
great job security for science.
Lex Fridman (51:12.020)
And that we, if this is true,
Lex Fridman (51:14.860)
and if consciousness is somehow important
Donald Hoffman (51:17.580)
or fundamental in the universe,
Lex Fridman (51:19.020)
this may be an important fundamental fact
Donald Hoffman (51:20.620)
about consciousness itself,
Lex Fridman (51:21.980)
that it's a never ending exploration
Donald Hoffman (51:25.080)
that's going on in some sense.
Lex Fridman (51:27.520)
Well, that's interesting.
Donald Hoffman (51:30.100)
Push back on the job security.
Lex Fridman (51:31.900)
Okay.
Lex Fridman (51:34.440)
So maybe as we understand this kind of idea
Lex Fridman (51:37.500)
deeper and deeper,
Donald Hoffman (51:39.140)
we understand that the pursuit is not a fruitful one.
Lex Fridman (51:42.940)
Then maybe we need to,
Donald Hoffman (51:45.180)
maybe that's why we don't see aliens everywhere,
Lex Fridman (51:48.380)
is you get smarter and smarter and smarter,
Donald Hoffman (51:51.260)
you realize that exploration is,
Lex Fridman (51:55.580)
there's other fun ways to spend your time than exploring.
Donald Hoffman (51:59.260)
You could be sort of living maximally
Lex Fridman (52:03.960)
in some way that's not exploration.
Donald Hoffman (52:05.860)
There's all kinds of video games you can construct
Lex Fridman (52:10.020)
and put yourself inside of them
Donald Hoffman (52:11.820)
that don't involve you going outside of the game world.
Lex Fridman (52:15.220)
It's a feeling, from my human perspective,
Lex Fridman (52:18.740)
what seems to be fun is challenging yourself
Lex Fridman (52:21.020)
and overcoming those challenges.
Lex Fridman (52:22.660)
So you can constantly artificially generate challenges
Lex Fridman (52:25.220)
for yourself, like Sisyphus and his boulder,
Donald Hoffman (52:28.260)
just, and that's it.
Lex Fridman (52:30.660)
So the scientific method
Donald Hoffman (52:32.300)
that's always reaching out to the stars,
Lex Fridman (52:34.060)
that's always trying to figure out
Donald Hoffman (52:35.200)
the puzzle on the bottom puzzle,
Lex Fridman (52:37.380)
we're always trying to get to the bottom turtle.
Donald Hoffman (52:40.540)
Maybe if we can build more and more the intuition
Lex Fridman (52:43.980)
that that's infinite pursuit,
Donald Hoffman (52:48.880)
we agree to start deviating from that pursuit
Lex Fridman (52:51.100)
and start enjoying the here and now
Donald Hoffman (52:53.180)
versus the looking out into the unknown always.
Lex Fridman (52:56.580)
Maybe that's looking out into the unknown
Donald Hoffman (52:58.960)
as a early activity for a species that's evolved.
Lex Fridman (53:07.580)
I'm just sort of saying, pushing back,
Donald Hoffman (53:09.820)
as you probably got a lot of scientists excited
Lex Fridman (53:12.180)
in terms of job security,
Donald Hoffman (53:13.620)
I could envision where it's not job security,
Lex Fridman (53:17.780)
where scientists become more and more useless.
Donald Hoffman (53:22.020)
Maybe they're like the holders of the ancient wisdom
Lex Fridman (53:25.500)
that allows us to study our own history,
Lex Fridman (53:29.660)
but not much more than that, just to push back.
Lex Fridman (53:34.500)
That's good pushback.
Donald Hoffman (53:36.500)
I'll put one in there for the scientists again,
Lex Fridman (53:39.340)
but sure, but then I'll take the other side too.
Lex Fridman (53:41.460)
So when Faraday did all of his experiments
Lex Fridman (53:46.700)
with magnets and electricity and so forth,
Donald Hoffman (53:49.340)
he came up with all this wonderful empirical data
Lex Fridman (53:52.060)
and James Clerk Maxwell looked at it
Lex Fridman (53:54.140)
and wrote down a few equations,
Lex Fridman (53:56.140)
which we can now write down in a single equation,
Donald Hoffman (53:58.220)
the Maxwell equation if we use geometric algebra,
Lex Fridman (54:00.180)
just one equation.
Donald Hoffman (54:03.900)
That opened up unbelievable technologies.
Lex Fridman (54:07.420)
People are zooming and talking to each other
Donald Hoffman (54:09.500)
around the world, the whole electronics industry.
Lex Fridman (54:13.660)
There was something that transformed our lives
Donald Hoffman (54:17.140)
in a very positive way.
Lex Fridman (54:19.380)
With the theories beyond space time,
Donald Hoffman (54:21.800)
here's one potential, right now,
Lex Fridman (54:25.440)
most of the galaxies that we see, we can see them,
Lex Fridman (54:29.440)
but we know that we could never get to them
Lex Fridman (54:31.040)
no matter how fast we traveled.
Donald Hoffman (54:32.640)
They're going away from us at the speed of light or beyond.
Lex Fridman (54:36.040)
So we can't ever get to them.
Lex Fridman (54:37.720)
So there's all this beautiful real estate
Lex Fridman (54:39.160)
that's just smiling and waving at us
Lex Fridman (54:41.080)
and we can never get to it.
Lex Fridman (54:42.560)
Yeah.
Lex Fridman (54:43.400)
But that's if we go through space time.
Lex Fridman (54:45.560)
But if we recognize that space time
Donald Hoffman (54:47.240)
is just a data structure, it's not fundamental.
Lex Fridman (54:50.240)
We're not little things inside space time.
Donald Hoffman (54:53.920)
Space time was a little data structure in our perceptions.
Lex Fridman (54:58.120)
It's just the other way around.
Donald Hoffman (54:59.960)
Once we understand that,
Lex Fridman (55:02.200)
and we get equations for the stuff that's beyond space time,
Donald Hoffman (55:07.080)
maybe we won't have to go through space time.
Lex Fridman (55:08.320)
Maybe we can go around it.
Donald Hoffman (55:09.640)
Maybe I can go to Proxima Centauri
Lex Fridman (55:11.080)
and not go through space.
Donald Hoffman (55:11.920)
I can just go right there directly.
Lex Fridman (55:14.680)
It's a data structure.
Donald Hoffman (55:15.520)
We can start to play with it.
Lex Fridman (55:17.200)
So I think that for what it's worth,
Donald Hoffman (55:21.680)
my take would be that the endless sequence of theories
Lex Fridman (55:27.760)
that we could contemplate building
Donald Hoffman (55:30.760)
will lead to an endless sequence of new remarkable insights
Lex Fridman (55:36.600)
into the potentialities, the possibilities
Donald Hoffman (55:39.400)
that would seem miraculous to us.
Lex Fridman (55:41.920)
And that we will be motivated to continue the exploration
Donald Hoffman (55:45.200)
partly just for the technological innovations
Lex Fridman (55:49.040)
that come out.
Lex Fridman (55:50.440)
But the other thing that you mentioned though,
Lex Fridman (55:53.960)
what about just being?
Lex Fridman (55:55.560)
What if we decide instead of all this doing and exploring,
Lex Fridman (55:58.960)
what about being?
Donald Hoffman (56:00.120)
My guess is that the best scientists will do both
Lex Fridman (56:04.400)
and that the act of being will be a place
Donald Hoffman (56:10.200)
where they get many of their ideas
Lex Fridman (56:12.640)
and that they then pull into the conceptual realm.
Lex Fridman (56:16.560)
And I think many of the best scientists,
Lex Fridman (56:18.480)
like Einstein comes to mind, right?
Donald Hoffman (56:19.880)
Where these guys say, look,
Lex Fridman (56:21.120)
I didn't come up with these ideas by a conceptual analysis.
Donald Hoffman (56:25.400)
I was thinking in vague images
Lex Fridman (56:28.080)
and it was just something nonconceptual.
Lex Fridman (56:31.760)
And then it took me a long, long time
Lex Fridman (56:33.760)
to pull it out into concepts
Lex Fridman (56:35.880)
and then longer to put it into math.
Lex Fridman (56:38.400)
But the real insights didn't come from data.
Donald Hoffman (56:41.040)
The real insights didn't come from just slavishly
Lex Fridman (56:44.520)
playing with equations.
Donald Hoffman (56:45.760)
They came from a deeper place.
Lex Fridman (56:48.360)
And so there may be this going back and forth
Donald Hoffman (56:51.640)
between the complete nonconceptual
Lex Fridman (56:54.720)
where there's essentially no end to the wisdom
Lex Fridman (56:57.680)
and then conceptual systems
Lex Fridman (56:58.840)
where there's the girdle limits that we have to that.
Lex Fridman (57:02.240)
And that may be, if consciousness is important
Lex Fridman (57:05.520)
and fundamental, that may be what consciousness,
Donald Hoffman (57:07.640)
at least part of what consciousness is about
Lex Fridman (57:09.400)
is this discovering itself, discovering its possibilities,
Lex Fridman (57:13.440)
so to speak, and we can talk about what that might mean,
Lex Fridman (57:17.240)
by going from the nonconceptual to the conceptual
Lex Fridman (57:20.920)
and back and forth.
Lex Fridman (57:23.400)
So you get better and better and better at being.
Donald Hoffman (57:26.440)
Right.
Lex Fridman (57:27.480)
Let me ask you just to linger on the evolutionary,
Donald Hoffman (57:31.080)
because you mentioned evolutionary game theory
Lex Fridman (57:33.160)
and that's really where you,
Donald Hoffman (57:35.280)
the perspective from which you come
Lex Fridman (57:37.120)
to form the case against reality.
Donald Hoffman (57:42.080)
At which point in our evolutionary history
Lex Fridman (57:45.360)
do we start to deviate the most from reality?
Lex Fridman (57:49.520)
Is it way before life even originated on Earth?
Lex Fridman (57:55.800)
Is it in the early development from bacteria and so on?
Donald Hoffman (58:02.280)
Or is it when some inklings of what we think of
Lex Fridman (58:05.880)
as intelligence or maybe even complex consciousness
Lex Fridman (58:11.680)
started to emerge?
Lex Fridman (58:12.920)
So where did this deviation,
Donald Hoffman (58:15.980)
just like with the interfaces in a computer,
Lex Fridman (58:19.520)
you start with transistors and then you have assembly
Lex Fridman (58:23.960)
and then you have C, C++, then you have Python,
Lex Fridman (58:28.160)
then you have GUIs, all that kind,
Donald Hoffman (58:30.320)
you have layers upon layers.
Lex Fridman (58:31.520)
When did we start to deviate?
Donald Hoffman (58:33.360)
Well, David Marr, again, my advisor at MIT,
Lex Fridman (58:37.520)
in his book, Vision,
Donald Hoffman (58:38.960)
suggested that the more primitive sensory systems
Lex Fridman (58:42.380)
were less realistic, less theoretical,
Lex Fridman (58:45.840)
but that by the time you got to something
Lex Fridman (58:47.040)
as complicated as the humans,
Donald Hoffman (58:48.280)
we were actually estimating the true shapes
Lex Fridman (58:51.640)
and distances to objects and so forth.
Lex Fridman (58:53.400)
So his point of view, and I think it was probably,
Lex Fridman (58:57.120)
it's not an uncommon view among my colleagues
Donald Hoffman (59:01.680)
that, yeah, the sensory systems of lower creatures
Lex Fridman (59:06.200)
may just not be complicated enough
Donald Hoffman (59:07.580)
to give them much, much truth.
Lex Fridman (59:10.040)
But as you get to 86 million neurons,
Donald Hoffman (59:12.880)
you can now compute the truth,
Lex Fridman (59:14.080)
or at least the parts of the truth that we need.
Donald Hoffman (59:17.120)
When I look at evolutionary game theory,
Lex Fridman (59:21.680)
one of my graduate students, Justin Mark,
Donald Hoffman (59:24.120)
did some simulations using genetic algorithms.
Lex Fridman (59:27.700)
So there he was just exploring,
Donald Hoffman (59:30.640)
we start off with random organisms,
Lex Fridman (59:32.160)
random sensory genetics and random actions.
Lex Fridman (59:36.360)
And the first generation was unbelievably,
Lex Fridman (59:38.560)
it was a foraging situation.
Donald Hoffman (59:39.720)
They were foraging for resources.
Lex Fridman (59:41.360)
Most of them stayed in one place,
Donald Hoffman (59:44.180)
didn't do anything important.
Lex Fridman (59:47.160)
But we could then just look at how the genes evolved.
Lex Fridman (59:51.180)
And what we found was,
Lex Fridman (59:55.080)
what he found was that basically you never even saw
Donald Hoffman (59:59.960)
the truth organisms even come on the stage.
🔗 相关节目
Alex Filippenko: Supernovae, Dark Energy, Aliens & the Expanding Universe
Alex Filippenko
Alex Garland: Ex Machina, Devs, Annihilation, and the Poetry of Science
Alex Garland
Alien Debate: Sara Walker and Lee Cronin
Alien Debate
Ariel Ekblaw: Space Colonization and Self-Assembling Space Megastructures
Ariel Ekblaw