Into the Impossible With Brian Keating - Donald Hoffman On Free Will, Positive Geometries of Consciousness & Spacetime [Ep. 455]

Episode Date: August 25, 2024

Will physics ever be able to explain consciousness? What if our entire understanding of reality is just an illusion shaped by evolution? Could the universe be a sophisticated virtual reality created by our own minds for our own good? And does free will exist after all?  Joining me today to explore these profound questions is none other than Donald Hoffman! Donald Hoffman is a renowned philosopher, famous TED talk speaker, cognitive scientist, and deep thinker.  He is a professor in the Department of Cognitive Science at the University of California, Irvine, and the author of the provocative book The Case Against Reality. This is his third time on my show because we simply never run out of topics, and it’s always refreshing to hear his thoughts and ideas!  So, without further ado, let’s jump right in! — Key Takeaways:  00:00:00 Intro  00:01:11 Sam Harris, Robert Sapolsky, and free will  00:07:59 Positive geometries of consciousness  00:17:17 Will physics ever be able to explain consciousness? 00:26:05 Stepping outside of spacetime  00:36:51 Markovian dynamics  00:48:10 Theory of perception  00:52:37 Audience questions  01:05:44 The future of academia  01:13:24 Outro — Additional resources:  ➡️ Learn more about Don Hoffman:  📚 The Case Against Reality: https://a.co/d/dy3tfCc — ➡️ Follow me on your fav platforms: ✖️ Twitter: https://twitter.com/DrBrianKeating  🔔 YouTube: https://www.youtube.com/DrBrianKeating?sub_confirmation=1  📝 Join my mailing list: https://briankeating.com/list  ✍️ Check out my blog: https://briankeating.com/cosmic-musings/  🎙️ Follow my podcast: https://briankeating.com/podcast  — Into the Impossible with Brian Keating is a podcast dedicated to all those who want to explore the universe within and beyond the known. Make sure to follow/subscribe so you never miss an episode! Learn more about your ad choices. Visit megaphone.fm/adchoices

What if our entire understanding of reality is just an illusion shaped by evolution? Could the universe be a sophisticated virtual reality created by our own minds, or evolution itself for our own good? And does free will even exist after talking just this year alone to Sam Harris, Robert Sapolsky, and Daniel Dennett? Three prominent philosophers on the subject of free will, two of whom would deny the existence of free will, I'm thrilled to converse with Donald Hoffman, who offers a completely different, highly mathematical and grounded perspective. Hoffman is a renowned philosopher and a colleague of mine in the University of California. He's the author of the provocative breakout bestseller,

the case against reality. This is his third time on the show because we simply never run out of topics and it's always refreshing to hear his brilliant thoughts on these issues and more. So buckle up. This is going to be the deepest dive into the fabric of reality that you've ever been on. Let's go. is indistinguishable from magic. Open the pod bay doors, both Sam Harris and Robert Sapolsky have recently been guests on The Into the Impossible podcast, and their perspective on consciousness is quite different from yours.

How do you view their perspectives in contrast to your own, Don? One of the things that they focus on is that they don't think that there's free will. At least Sam doesn't think that there's free will. And the theory of conscious agents that I'm working on does seem to have what we call a decision kernel in it. So a Markovian kernel that sort of models the decisions that a conscious agent takes. Of course, the math is just the math. It doesn't tell you how you have to interpret it.

But it's sort of tempting to interpret it as a notion of free will, but a very interesting notion of free will, because agents aren't in a vacuum. They're in a milieu of other agents. It's a social network of interacting conscious agents. And as we'll discuss, they can combine. So agents can interact, and if they are compatible, they can combine. And in that process, what will you see is that there's this, it's not just a social network of agents,

but the agents themselves are combining or dividing. And so the decision kernel of each agent is not the only decision kernel that's going on. There's lots of these decision kernels interacting. but the fact that you can combine means that in some sense my decisions are part and parcel of perhaps a bigger agent

of which I'm a part and so if there is a notion of free will and the notion of conscious agents I would say that it's a notion that's unusual that in some sense my free choices

are entirely aligned with the higher agents of which I'm apart, and yet my actual, if you look at the details of my choices, they're unique to me. So I have my own unique, you know, Markovian colonel for my own choices, but is compatible with all the other choices. So that's one interesting direction. So our work on consciousness provides an interesting framework for thinking about free will and perhaps more less philosophical, more mathematical kind of way. But But then there's another aspect where we may differ on consciousness, although people's views evolve.

I take consciousness as fundamental. Most neuroscientists, and I think Sam, at least in some of the things I've heard of him, heard him say, wants to talk about the neural underpinnings of consciousness. Now, you know, maybe that's no longer the case. I won't pin that on Sam. I won't say that that's what he's saying. But suppose if he is, then, and that's sort of the standard view, is that in some sense, there is a physical substrate to consciousness.

So consciousness in that sense isn't fundamental. It's the neural circuits, for example, that are fundamental in their activity. Or if it's an AI, maybe the circuits of an AI or the software in an AI that's fundamental and it's a substrate that gives rise to consciousness. my attitude is that of course that's a natural thing to look at it fits in with the normal framework of science of you know a productionism and and so forth but it hasn't worked out yet there's not a single specific conscious experience that we've been able to reduce to a neural substrate for example so there's or or any of the current physicalist theories so we're we're batting zero on that

so so those are two areas where i won't claim that we're fundamentalists. mentally disagreeing, but where one could imagine there's disagreement. And when I think about those two individuals, they both, you're right, kind of deny the essence of the question of whether or not we have free will. In one case, it seems to be, you know, strictly determined in Sam Harris' case, I should say. And in Sopulski's case, you know, where I kind of diverge from him or at least, I give him intellectual honesty points because he says, I act as if I have free will, and I believe in my heart and to my shame, he said to me that people should be punished for their action. I found

very odd. If I go up to Stanford and occupy his office and so forth, or worse, I steal his manuscript in latex form from his desktop, not his virtual mental and evolutionary desktop that you have pioneered, but the real desktop on his computer, I, assume he'd be pursuing me in a court of law if I steal his next book, you know, really determined or, you know, super determined as he probably is one. So how do you come down on people that, that say that they don't really put their money where their mounts are, you know, and I'm glad that they don't because nobody acts as if they have no free will. I mean, even people that believe, like Sam Harris acts as if he has free will, even though he believes everything is effectively

determined. How do you, what do you make of such people? I mean, are they, you know, do you respect their honesty, but you kind of deny their legitimacy. Where do you come down on the case of, you know, for example, crime and punishment, you know, cause and effect? How would you adjudicate such issues? These are difficult issues and they have a long philosophical history and I'm delighted that different people are taking different points of view and we're having a debate. So when I read Sam's stuff, for example, it's challenging and it forces me to think out of the box and so forth. And when you see the neural evidence that we can predict from neural activity in certain parts of the motor cortex, what you're going to do, you know, several seconds before you know what you're going to do,

there's a prima facie case that can be made that the neural activity is, in fact, the cause of your behavior and not your, what you thought was your free choice. It's an obvious prima facie case that can be made. Now, I think that it doesn't go beyond prima facie. I think it, you know, when you push on it, I think it dissolves. But I certainly would not in any way charge them with, intellectual dishonesty or anything like that. Absolutely not. I think I'm delighted to have intellects of that stature taking points of view that I disagree with

so that it gives me something challenging to have to push back on and so forth. So I think it's all good. And, you know, reductionism has been a very powerful framework for a long, long time. I think it's no longer going to work going forward. But it, you know, there's good reason to try given the long history of some major successes.

You gave a wonderful talk in the science of consciousness, which is put on by our mutual friend and many-time guests, Stuart Hammeroff and others at the University of Arizona. It's been going on 30 plus years, I believe. I spoke at it about 10 years ago. I was never invited back. No. But I've had on all the wonderful luminaries like your fellow colleagues. but you gave just a spectacular talk and you've been gracious enough to agree

that you'll present, we'll present your slides here and we'll walk through it and maybe I'll ask some questions as an interested layperson. You know, I'm always frustrated when I talk to, not to you, but to others, you know, just seems to be so little progress that the progress in string theory seems to be like meteoric, you know,

compared to consciousness. And that's saying a lot, as you know, my feelings about string theory. But let's walk through your slides and then we'll show them on the screen as you go through it, and I'll try not to interrupt, but it's such a fascinating talk. I couldn't let the opportunity to share this with my audience pass me by.

So the title of my talk is positive geometries of consciousness. I first start off and talk about the fact that we have lots of conscious experiences. We can experience estimates are thousands of flavors or millions of colors. And I was surprised to find out. the field of smell now thinks that we could be experiencing trillions of smells. And then we have all sorts of experiences of like sounds and bodily feelings. And so all in all, we have trillions of specific conscious experiences that we enjoy or don't enjoy as human beings.

And so you would think that, you know, there's been a lot of work now in cognitive neuroscience and AI and so forth trying to get theories of conscious. consciousness, you know, to try to understand and explain these specific conscious experiences. And most of these theories assume, of course, that space and time are fundamental, and particles, elementary particles in space and time are sort of the fundamental building blocks. And when you get, you know, a large enough group of these particles interacting the right way, you get macroscopic objects like neurons and brains, and neurons and brains with the right properties give rise to consciousness or to the illusion of consciousness.

Some physicalists think that consciousness isn't real, it's just an illusion that's made, say, by our nervous system. And then there are panpsychists of various stripes, but the ones I'm interested in are panpsychists who also take space time as fundamental and certain physical objects in space time, like elementary particles as fundamental objects, but that in addition to their physical properties,

they have some kind of property of consciousness. some unit of consciousness, say an electron has, has physical properties, but also has some kind of qualia experiences associated with it. So there are a number of theories. There are, you know, higher order theories, there are integrated information theory, neuron or monocytubule, quantum states, collapse, orchestrated collapse, global workspace theory. There's a number of theories. And recently, you know, the, there's been some adversarial collaborations that have been funded where a couple of these theories can go head to head and make predictions say about the neural correlates of consciousness that they have and so forth. So it looks like you might think, well, okay, well, it's a good feel.

We have lots of theories out there and a good background. So, you know, where are we? And, you know, there's trillions of experiences out there that, you know, for these theories to explain. So So it must be, you know, you think it would be like, you know, shooting fish in a barrel. You know, just pick, there's trillions, so pick one, and you use your physicalist theory to explain it. Well, and so how are we doing? How are, you know, how many experiences have these theories explained? And the remarkable fact is, is that there's zero. There's no physicalist theory right now that can explain, for example, the taste of mint.

You know, what orchestrated collapse of quantum states of nonoralachotribles must be the taste of and couldn't be the smell of garlic or something like that. So we're batting zero, and it's quite remarkable. We look at all these theories, and some of the theories will tell you exactly what you need to do. For example, an integrated information theory, to have a substrate for a conscious experience, you need to have the right causal structure of the substrate, and that causal structure can be specified by a Markovian kernel. So the natural question is, okay, great.

So give me a Markovian kernel for the taste of mint. You know, how big a Markovian kernel does it have to be? You know, does it have, you know, a thousand rows and a thousand columns and therefore like a million entries? And if so, why that number of rows and columns and what are the specific numbers and the entries that it must be that it must have? Or the class of entries that it must have to be the taste of mint and couldn't be the taste of chocolate. And there's nothing. nothing's been done. So some of these theories tell you what they would have to do to have a theory,

but in fact, when it comes right down to it, where's the beef? Where is your matrix? Or where is your precise orchestrated collapse of quantum states of microtubas that must be the taste of vanilla? Or what's the global workspace architecture and contents of the shared memory that must be the taste of chocolate? So, and part of the problem is that as soon as you say, okay, well, here's the matrix that must be, you know, this is the causal matrix that must be the taste of mint, the obvious question is, give me a break. Why in the world must this thousand by thousand matrix and those particular entries be the taste of mint? Please, you know, give me an explanation. And you can say, well, you can see where there's going to be some problems here doing that.

Or what orchestrated collapse of quantum state? And this is not, by the way, a new problem. Leibniz saw this way back 300 years ago in his monodology. And he actually gave physicalist theory short shrift. He gave them one paragraph. He said, yeah, obviously can't be done. You can't boot up consciousness from unconscious, mechanical, physical ingredients. He spent a paragraph on it and figured, you know, intelligent readers can move on.

They could have a sight of him. You know, he'd have a good H index. It'd be good for his career by now. These are some of the conscious experiences that you might want them to try to explain, the taste of chocolate, smell of a rose and so forth. Or, you know, the experience of space. You know, if you have a theory of space, you know, the experience of space. Why must this, for example, pattern of integrated information be the experience of visual space

and why could it not be the experience of haptic space or auditory space? and similarly for time. So again, no answers to this kind of thing. For example, an integrated information theory and experience qualities identical to maximally irreducible intrinsic cause-effect structure of a mechanism in a state, and the cause-effect structure can be

written as a transition probability matrix. And so the question then is, so what is a probability matrix that must be, you know, the taste of mint? And why those specific entries in the Markovian kernel. And what do these matrices operate on? What's the vector analog here?

I think they're not so much viewing them as operators here as representing a causal structure. It could be viewed as maybe a flow of information through a particular, you know, circuit or something like that. So, yeah, so why should, you know, if you have a theory that you're claiming there's a theory of the experience of space, what is the matrix for visual versus auditory? versus haptic. And why must it be that matrix and so forth? And again, this is just asking them to do what their own theories say that they need to do for, right? So orchestrated collapse of no neural

microtubules, it would be, okay, so what is giving me the precise pattern of orchestrated collapse of non-nar microtubules that must be the taste of mint? And why must it be that? And so we're batting zero on that. This is a little striking because if I came to you and said, I've got a theory of particle interactions. And you said, oh, wonderful. So, you know, what particular particle interactions can you, can you model in your theory? I say, well, you know, you might say, well, can you do, like electron photon interactions like this one, or cork and gluon interactions? And if I said to you, well, oh, I can't do any specific particle interactions. I can't actually, you know, predict any specific ones. I just have a general theory of particle interactions. You might wonder what in the

world I'm talking about. What does it mean to have a theory of particle interactions that doesn't explain any particle interactions. And so the same thing is here. What does it mean to have a theory of conscious experiences or of consciousness that doesn't account for any specific conscious experiences? It's important to understand that what a scientific theory is. Every scientific theory makes certain assumptions. And it says if you grant me these assumptions, then I can explain some other wonderful stuff. As a result, because every theory has assumptions, there is no theory of everything. The theory has a scope and it has some limits. Newton's theory, Newtonian physics, has incredible scope. A lot of mathematics in the

theory to explore the power of that scope. But it has a hard limit. When we came to look at chemistry, we realized that basically Newton can't do chemistry. No way that Newtonian physics is going to do. We need to quantum theory. So every scientific theory has its own scope and limit. We sometimes talk about the hard problem of consciousness, but there are also certain theories that are just going to be the wrong problem for, you know, it'll be the wrong problem for them. So chemistry is the wrong problem for Newton. It's the right problem, perhaps for quantum mechanics.

Every good theory gives you mathematical tools to explore its scope. Great theory gives you the mathematical tools to actually find their limits. And that's an important aspect of science, is that we have certain concepts like Einstein, for example, had a concept of space time being fundamental and the speed of light being universal for all observers, at least inertial observers. So he turned his concepts into a mathematically precise framework, special and general relativity. And then later on, that combined with quantum theory and quantum field theory, led to discovery that space time is not fundamental.

The space time falls apart at the plank scale. So every theory has its own scope and limits, and a great theory. Actually, you can use the mathematics of the theory to show the fundamental limits of the basic concepts of that theory. For example, in Einstein's case, Space Time itself, which is what is modeling, turns out to have a limited range of applicability. It has no operational meaning beyond 10 to the minus 33 centimeters. So I want to propose, in part because space time is not fundamental, that physicalism cannot do consciousness. So it's not just that consciousness is a hard problem for physicalist approaches, neural reductionist approaches or AI reductionist approaches or whatever it might be, or integrated information identifying it with certain causal structures in a physical substrate.

That class of theories assumes that spacetime is fundamental and assumes the whole reductionist paradigm. And I'm saying that that whole paradigm is just the wrong paradigm. And consciousness is the wrong problem, not a hard problem, but an impossible problem for any physicalist reductionist approach. So that's sort of one of the main points that I want to make here. So, and it's not, of course, that's one thing for cognitive scientists to say, but it's the high energy theoretical physicists who are saying this. So, Neymar Connie Ahmed and others for, so Nathan Seiberg, for example, says, I'm almost certain that space and time are illusions. These are primitive notions that will be replaced by something more sophisticated.

Andrew Strominger at Harvard says the notion of space time is something, you've got clearly something we're going to have to give up. David Gross, who won the Nobel Prize for his work on the Strong Force, says, I believe that space for sure and presumably time as well will be emergent. And then I think Nima says the very notion of space time is not a fundamental one. Space time is doomed. There's no such thing as space time fundamentally in the laws of physics. David Gross then explains why.

He explains that there's no operational meaning to distances smaller than the Planck scale. If you're trying to measure smaller and smaller objects, you need to have more and more powerful microscope. And for that, effectively you're using light or some kind of radiation with smaller and smaller wavelengths, so you can resolve the details of the object that you're trying to look at. theory allows that. I mean, you can make a smaller wavelength as you wish. It requires more energy. You know, the E equals H-new, so as the frequency goes up, the wavelength is smaller than the energy

is going up. But when you combine that with Einstein's theory of space time, and especially, you know, the fact that energy and matter and matter-energy curve space-time, you get a problem at 10 to the minus 33 centimeters that with the wavelengths at 10 to the minus 33 centimeters, You have so much energy in such a small region of space that you actually create a black hole and you destroy the very thing that you're looking at. And so what this means is that space time is a fairly shallow data structure. It falls apart at 10 to the minus 33 centimeters, not 10 to the minus 33 trillion centimeters, just 10 to the minus 33 centimeters. That sounds impressive to us now because we don't have the technologies to probe it. But once we get the technologies to probe it, 10 to the minus 33 centimeters, then, you know, it's going to be a hard limit to what we can do.

inside space time. So space time is doomed. We thought it was fundamental, but it's not. I also talk about evolution by natural selection, and it also agrees with the physicists, the high-energy theoretical physicists, that space time is doomed. You can use Darwin's theory in mathematical form, in evolutionary game theory, to ask a technical question, which is, what is the probability that any organism has ever been shaped to see any aspect of objective reality, truly? as it is. So you can ask that as a technical question. Do we see the truth? What's the probability that our senses show us the truth about objective reality? We see what appears to us in our senses are this space and time and physical objects inside space and time. So the question I'm asking,

does evolution entail that those are going to be true, true perceptions of the objective nature of reality or not? And it turns out, so we can ask the question, does natural selection favor of radical perceptions? And the answer on the next slide, is no we vertical perceptions go extinct in fact the probability is zero that any sensory system of any organism has ever been shaped by natural selection to see any structural aspect of objective reality correctly so what we find is the two main pillars of modern science which is high-energy theoretical physics you know quantum field theory for example and also evolution of a natural

selection, both agreeing that space time isn't fundamental, and objects in space time aren't fundamental. And therefore, the whole physicalist reductionist paradigm will be called into question. So what is it that we are seeing? From an evolutionary point of view, we get a hint that evolution shapes our sensory systems to guide adaptive behavior. That's right. We are shaped by evolution to have sensory systems that guide adaptive behavior, period. We thought that seeing the truth was going to help you have more adaptive behavior and that's just a wrong intuition. When you actually look at evolution, what's really going on is in some sense, evolution isn't showing you a window on reality. Evolution gives you a user interface,

like the desktop interface, or gives you a VR headset, if you want it, to play the game of life. If you're playing a virtual reality game like Grand Theft Auto, what you're really doing in this analogy is you're toggling millions of voltages and a precise sequence in some computer, some powerful computer somewhere. But if you had to do that to try to win the game, good luck. It would be very, very difficult to toggle those voltages and win the game. Someone who has a little virtual reality headset and they see a steering wheel and a dashboard and gas pedal and so forth is going to beat you. They don't need to see the truth. They need to control the truth, but they don't need to know what the truth is even. You don't have to know about diodes and resistors

and voltages to win the game if you are given the right interface. And so that's sort of the idea of what Darwin's theory tells us. is that evolution shaped us with a headset that lets us play the game of life and it hides the truth. So that means that reductionism is doomed. If space time is doomed, then the idea of reductionism, which I think is the next slide, that fundamentally the laws of physics are given by some laws at the ultra-most microscopic scale. And we just have to go to the smaller scale to see what's ultimately going on. And physicists are telling us is that that reductionist paradigm is wrong.

As we go to smaller and smaller scales, at some point you just stop. You can't go any smaller, 10 to the minus 33. And actually, if you try to go smaller, you start actually going the other direction. You actually get black holes that get bigger and bigger, bigger. So you get more macroscopic. So the reductionist paradigm is a problem. And it's been the foundation for a lot of science. We used to think the foundation was like Earth, Air, Fire, and Water, four basic elements.

And then we went to the periodic table of elements. We thought that was fundamental. And now we have the standard model. of physics with the bosons, leptons, and quarks. And that's our current fundamental reductionist approach. But we know that space time is doomed so that even the bosons, leptons, and quarks aren't going to be the final answer in terms of what's the nature of the basic elements of objective reality. So the idea that has guided a lot of consciousness research is, again, a reductionist one, which is that we start with fundamental particles like the bosons, leptons, and cork.

works and when they have the right combinations, we get more larger objects like these pyramidal neurons and if you go in larger, you get objects like the brain. And then the brain with the right properties, right causal or other kinds of properties, somehow gives rise to conscious experiences. That's the theory in global workspace and illusionism and panpsychism and so forth. But there's a problem with all this, and Stephen Pinker points it out, quite nicely in his 2018 book Enlightenment now, what I'll call the stipulation problem of consciousness.

And what Pinker says is, but the last dollop in the theory, and he was talking about the global workspace theory, which is the theory that Pinker seems to like in that book. But the last dollop in the global workspace theory that it subjectively feels like something to be such circuitry, the global workspace circuitry,

may have to be stipulated as a fact about reality where explanation stops. What Steve is saying is that, you know, we have the global workspace theory and all of its architecture, and is trying to explain conscious experiences when you get right down to it, like it takes the smell of a rose or the taste of chocolate, it doesn't give it to you.

You are assuming the global workspace architecture, and then you're having to also stipulate the conscious experience of the rose or the chocolate of the cake. So you're stipulating the physical system. You're also stipulating the experience. There's no explanation here. It's all stipulation. It's topological, yeah.

Yeah, and it's not... It's not very satisfying science when the thing you're trying to explain has to be stipulated. At the very last step, you say, you know, you have to stipulate. So that's a big problem. These reductionist approaches are failing, and I think it's for principled reasons because reductionism is doomed. And that's what the high-energy theoretical physicists are telling us. Space-time is doomed. Therefore, reductionism is doomed.

And therefore, our field, the field of consciousness studies, needs to get a cue from the experts on space time, namely the high-energy theoretical physicists, that it's over for space-time and reductionism. And we want to start there with our theory of consciousness, you know, we're starting with the wrong foundation. And not only is it space-time not the right foundation, but also quantum mechanics. So again, these theorists are pointing out

that quantum theory is not fundamental either. As NEMA puts it, so there's some other structure that we're looking for in some way of thinking about interpreting the structure beyond space time that will let us see space time and quantum mechanics emerge simultaneously

and joined at the hip. One reason why they're saying that space time is doomed and quantum theory is doomed is that they've, you know, when they're trying to

understand the probabilities or the amplitudes for particles interacting like two glows smashing into each other and six gluons spraying out, when you do the computation inside space time,

it's a mess. It's hundreds of pages of algebra for a single event. Hundreds of them, thousands, millions of terms. But they've discovered recently these new structures outside of space time, yeah, like the amplitude hedron in cosmological polytopes and the sociahedron. So they found these new structure. This is all fairly new, last 10 or 11 years, that allow them to compute scattering amplitudes much more cleanly, much, much more simply. Millions of terms

turn into four or five terms that you can compute by hand. And you see new symmetries, something called the Infinite Yangian symmetry, for example, in certain interactions that you can't see inside space time. What we're finding, and then they're finding these structures called decorated permutations that help classify, allow them to classify these new geometric structures. And so there's this whole new field called the field of positive geometries that's coming out now. We used to think space time is fundamental, but it's not. Beyond space time, there are positive geometries like the amplitude hydrant, and then beyond them are these combinatorial objects like decorated permutations, which allow us to classify

these objects. So this is a brand new frontier. We've stepped outside of space time for the first time as a species in the last 10 or 11 years. We're outside of space time and we're finding stuff. Hey there, Students of the Impossible. I'm sure you're enjoying this conversation with my good friend Don Hoffman. And I want to make sure that you help me help you get even more great guests on the Into the Impossible podcast. It's been in a wild ride. This year alone, I've had Sam Harris, Dan Dennett, Robert Sapolsky, and soon you'll hear from Richard Dawkins two special episodes. But I want you to help me help you, which is to make sure you're subscribed and following the podcast on audio and video formats. It really helps out. And if you don't mind extra credit homework assignments from a professor,

please make sure you leave a comment or leave a review of the podcast on Spotify or on the Apple podcast app. You're able to do that. It will really help me out so much. So now back to the episode with Don. Yeah, and this is right up our alley as physicist, but it might not be abundantly clear to the audience. How does it come back to the brain? I mean, is it that, It's a no-go theorem because we currently don't understand and we need physics beyond the standard model to understand physical reality and the ultimate building blocks are particles, but we don't actually know if there are strings that go even below that or matrices or loop quantum gravity is correct. Is that the relevance of that? I mean, it's fascinating to me as a physicist, but what is the relevance to the brain and the mind and consciousness? Well, the relevance is if we're trying to find a fundamental framework from which to boot up,

consciousness the question is is space time itself a deep enough structure for us to try to boot up consciousness from it and my take on what's what's happening in high-energy theoretical physics is they're saying no space time and quantum theory themselves are merely projections of much deeper structures and much more interesting structures space time has four dimensions maybe 11 or 12 depending on what what other theory might have, the amplitude hedron could have billions of dimensions.

These are different. So these are not little things curled up inside space time. These are brand new structures. And so we're very much taken off the headset. We're looking outside of space time. And what we're finding are obelisks. We're finding these geometric objects outside of space time.

And I think about it as like the scene in 2001 of Space Odyssey, where there's the obelisk and all the apes are, you know, hooting and hollering and pounding. And they know it's really important and it's very significant, but they don't know what it means. And I think that's where we are, because we've only been outside of space time for the last 10 or 11 years. We're finding these positive geometries, but the question is, who ordered that and why? And what does

this mean? And space time is clearly not fundamental. What is fundamental? So we're at this big, you know, this big new vista has opened up in science that says reductionism inside space time is not where it's at. And there's something new that's really important. It's positive geometries, but we don't know what it means yet. For example, we have no dynamics. What is a dynamics outside of space time that gives rise to these positive geometries and these decorated permutations that they found? This is all brand new, but what it says to me is to my field of consciousness studies, if we're trying to build our theory of consciousness on space time physics, that's an old horse now. We don't want to hop on that old horse. We need to get on something that's more

fundamental. And by the way, just in February, there was a big workshop in Germany, bringing it together 100 PhDs in physics and high energy theoretical physics in what's called the Universe Plus project that's been funded by the European Research Council, the ERC. It's a 10 million Euro grant international study effort to understand these positive geometries and the combinatorial objects like the decorated permutations. So this is where the high-energy theoretical physics is headed right now and consciousness studies will ignore this to their own peril. We'll effectively be in the same place as someone trying to do chemistry and believing that Newton is the way to do it. So I'm going to stick with Newton.

Newton. I don't care what this new fangled quantum stuff is. I'm sticking with Newton for my theory of chemistry. Well, good luck. And that's where we are right now in our consciousness studies. We're going to stick with space time, even though that horse is not the horse that the physicist themselves are riding at the very high energy level. So what I want to propose then with, again, with a number of collaborators, including, you know, Cheaton-Trenton, Robert Prentner, Vinich Singh, and a number of other Schwapen, Chattapaget and others. A theory of what we call conscious agents, a network of interacting conscious agents as the fundamental nature of reality. And this is going to be conscious agents not emerging from space time, but prior to space time. And the idea will be that these conscious agents are, in fact, prior to the positive geometries that high energy physics is finding outside of space time. So, you know, I'll talk about that in a moment. So there's a whole laundry list of things that a theory of consciousness must explain. I've got learning, memory, problem-solving, attention, memory, the self, intelligence, semantics. There's lots of things that you would like a theory to do.

But, you know, following Akins-Razer, I want to keep my assumptions to a minimum. So we're going to only pick a couple. We're going to use qualia. So conscious experiences exist and actions. That there are probabilistic relationships among conscious experiences. That's all we're going to have in our theory. And then we put this in a math. mathematical formalism, which I'll only talk about briefly, we have these things for conscious agents, and an agent has a set of experiences.

Based on the experiences, it then decides what action is going to take. And once it does that decision, that it acts on the network, the big social network of other conscious agents, which then acts back on the conscious agent that we're sort of focusing on here. So we get this loop, a perception, decision, action loop. And we modeled this with Markovian kernels, so probability spaces for the, for the set of possible experiences and so forth, and then Markovian kernels for the arrows. It turns out that the set of conscious agents is computationally universal. So it's easy to prove that anything that you can do with a neural network, you can also use a conscious agent network to build. So there is a learning, memory, problem solving, and so forth can be.

And it's Markovian in that the current state, you know, provides input for the future state, but no further kind of memory, you know, hysteresis or affects linger, or is there some other implementation of Markovian processes? We're going to implement it as, you know, Markov, literal Markov kernels in the, where those arrows are. So we'll literally have Markovian kernels. But what you pointed out is correct. Markovian dynamics is one in which the current state,

is all the information that governs the transition to the next state. So it's a finite memory. I would point out that you can make that state as complex as you want. So what it really means is you can have an arbitrary but finite amount of memory in your decision. So it's very much like in Turing machines where we say you have a tape, it's a finite tape, but there's as much tape as you want. You'll never run out of tape when you're doing your computation. And that's sort of the way it is with the Markovia.

colonel states. It's going to be finite, but as big a state as you want, we can give you that, and we can turn it into Markovian. So it's not much of a limitation, actually. Then I went through some fairly technical stuff here. Turns out that we discovered an order. So we can actually model individual conscious agents with what we call a quality kernel. We can take the perception, decision, action kernels that I talked about in that loop and combine them together into a single kernel that we call the qualia kernel. So now we have a much simpler thing, and we can actually then look at a single Markovian

kernel, which describes if I'm a conscious agent, I have the experience of red now, what's the probability that I'll experience green next or blue next or whatever? So it gives us what we call a quali kernel. What we've shown is that there's actually a logic on the set of these Markovian kernels. So this is apparently a new contribution to the mathematics of Markov-Chang. And we'll be publishing that in the paper that we're writing right now. We discovered that there's what's called a partial order on the set of Markovian kernels, and this gives a logic.

You can take the and the and the and the and the and the end and the ore, the meat and the join and so complements in certain certain cases and so forth. We give it a non-bullying logic of Markovian kernels, which then turns into a non-bullying logic of conscious agents. So we can actually talk about agents, consciousness, is combining or disassociating, decombining. And we can talk about also a notion of observation. One conscious agent, if it's sort of what's called a trace chain of a bigger agent,

then it observes the bigger agent. Then there's a whole logic that we have on that. So then our big idea then is that we have this network, this social network, like the Twitterverse, of conscious agents. And it's beyond space time. And it's actually beyond the geometry. and permutations that the physicists have found outside of space-time. But our proposal is that we can project the dynamics of conscious agents onto the decorated permutations.

And from there, the physicists tell us how to project into the positive geometries and into space-time. So we've actually, in a paper last year, we actually got a projection onto decorated permutations from Markovian kernels. And when we did that, we first just went in the literature, we assumed it had already been done. just how do you represent Markovian kernels and you know classify them with with decorated permutations and it turned out it hadn't been done so last year we did it and we we actually contributed as far as I can tell something new to the mathematics of Markov-Markov chains we actually showed how they can be classified by decorated permutations so now what we're what we're doing is we're looking at how to now

project we've got the projection of the dynamics of conscious agents onto decorated permutations. Now we want to get all the way into space time, right? We want to make, you know, what does the mass, what are particles in space time? What do they correspond to in the Markovian dynamics and so forth? What does mass, momentum, spin, distance, and so forth. And I can just say at top level, in the paper we're writing right now, we're going to publish the idea that the mass of particles inside space time is a projection of the so-called entropy rate of communicating classes of these Markovian kernels of conscious agents.

So an entropy rate is an interesting thing. It's very important in communication theory. If you have a source, communication source, that has an entropy rate that's greater than the channel capacity than you're going to get distortion. You won't get the message through clearly. And also entropy rate is related to a minimum description

for a signal. minimum discretion link. So the entropy rate is an intro, and if people want, I can go into detail about the mathematical definition of entropy rate, but we make a proposal that the entropy rate is what projects to mass, and that makes a real clean prediction, because if the entropy rate is zero, then we're going to be dealing with massless particles, right? And therefore, we have to make a prediction that this is going to be a spin-one particle, plus or minus one, And we have to make the prediction that this thing is going at the speed of light. There's a maximum speed and this thing is going.

So in other words, we don't just get to say, oh, entropy rate is mass. There are consequences for saying that. And it turns out that they work in the more coden dynamics. When you actually have an entry rate of zero, you get a kernel, looks like the P kernel down here on the bottom right. It's all zeros and ones. Those are the entry rates are zero. And those, in fact, have determined it plus or minus one. So if we say the spin is a projection of the determinant, then we get that.

And then they also have the fastest commuting times. So commuting time between two states is the expected number of steps it takes you to get from that state to the second state and then back again. So it's the round-trip time. And it turns out that when you have the zero entropy rate, then you are, effectively, the commuting times are the fastest you can possibly have. And so there is a maximum speed limit. and it happens precisely when you're massless in this theory and when the determinant is plus or minus one.

So internally consistent. It's internally consistent. It doesn't mean we're right, but it means that the first little test is passing the first sniff test. We'll see if it goes all the way. So yeah, what's nice is that Einstein assumed that the speed of light is the maximum speed, and it's the same for all observers. But this theory, if this pans out, we'll actually explain why there is a maximum speed

because there is a maximum computing time and it happens to be for those systems that have zero entropy rate and therefore are massless and have spin one. Things that are assumptions in Einstein's theory will be things that we can explain from deeper assumptions in the theory of consciousness. So how do we test all this?

There's a number of ways to try to test it. You might say, well, why don't you go after a neural correlates consciousness. I mean, that's clearly the thing you want to test. And eventually we do, but I'm more interested in doing something simpler. Not that is simple, but, but I'm looking in this little example here, I want to look at the inside of a proton. And the many physicists have spent decades studying particle collisions that are inelastic collisions. They destroy the proton to let you look inside, you know, at the quarks and gluons and so forth inside. And so, This is just sort of a high-level picture of, you know, where quarks and gluons and protons fit in the whole scheme of matter.

I have to point out that nucleus and the atom is not charged neutral. There's something going on with that. It only has two protons. It should be helium, but it's helium five or something. Oh, okay. Oh, interesting. So when you look inside of a proton, what at the top level, what the decades of experiments have shown is that at the lowest resolution, what they call Bjork and X and Q squared, inside of a proton, you find it's dominated by three valence quarks, two up quarks and a down quark.

As you increase the resolution in space and time, you get what they call a quark C. You start to get quark-ant-quark pairs appearing, and you also get a bunch of gluons. Then when you get to the highest resolution, highest bureconnex and Q squared, then you get just a vast gluon C. So what you see inside the proton depends on essentially the scale, the temporal and spatial scales at which you're looking. And so we want to show that we can model all of this. We can get all of these momentum distributions at the different scales of resolution in space and time and actually get the exact momentum distribution. So that's going to be one of the acid tests we want to give our theory.

In fact, there's a team of researchers from Caltech and elsewhere that are, just working with us on this right now, they're going to be trying to build these simulations. So these are some of the momentum distributions that will try to match. And for those who are interested in some of the published details on this, we have a paper called Fusions of Consciousness, which is free online. You can read about the mapping to a Markov change to decorated permutations and so forth. Hey there. It's me again with a great offer for you, which is free, and you can cancel at any time, but I hope you won't, it's to join my Monday Magic mailing list.

And for one lucky listener each month, I select you to win a real fragment of a 4.3 billion-year-old segment of our solar system, a meteorite, which fell to Earth in Argentina some few thousand years ago. I'll send you information about meteorites and all sorts of cool things and how you can observe and maybe find some yourself. But I'll also send you every week the latest and greatest information from around the world of STEM, science, technology, engineering, and math, and many, many other thoughts that I,

develop and go through it. Every week, people really seem to enjoy it. Recommendations, tips, tricks, hacks, and all sorts of cool things from around the world of academia and beyond. So if you have a dot edu email address, the offer is even sweeter. You're guaranteed to win one of these beauties if you live in the United States. For those of you with a dot edu email, just go to briancating.com slash edu and send your details there. Now back to the episode with Don. That's excellent. No, that was really fun. So I've got a bunch of questions of my own, and then I've got a lot of questions from my audience as well. Can I ask, you know, since we talked about your first book,

the most recent book, The Case Against Reality, have there been updates to your sort of evolution in this realm with respect to the interface theory of perception? Have there been new developments since 2021 when we talked last on the podcast live about that book? Any new developments, either from you or criticism, Or, you know, kind of joint research programs? Yeah, there's been some criticisms that from philosophers that have been quite sharp and published. So it's not just been verbal, but they've actually published articles.

And the argument goes like this, that we use evolutionary game theory to show that physical objects like organisms and resources in space and time are not fundamental. that they're just a user interface illusion. Several philosophers have said, that is logically shooting yourself in the foot. You're a cognitive scientist, you don't know philosophy, you don't know about logic, well enough to know that you've actually committed some basic fundamental self-refutation going on here.

So the argument is this. Hoffman, you're using evolutionary game theory, which is a mathematician of Darwin's ideas. You're using evolutionary game theory to prove that physical objects like organisms and resources are not fundamental. They're not part of the fundamental furniture of reality. They're just user illusions, so to speak. Now, so their argument goes like, not because.

Now, either the mathematics of evolutionary game theory faithfully represents Darwin's ideas, or it doesn't. If it doesn't, you couldn't use it for the purpose that you've used for. And if evolutionary game theory does faithfully represent Darwin's ideas, then it couldn't possibly refute those ideas. They couldn't possibly end up saying those ideas cannot be fundamental. So either way, you're in what they've actually in print said, an unfortunate dialectical situation.

Okay, in other words, I shot myself on the foot. And so my reply is that argument completely misunderstands the very foundations of how science works. First, there are two critical things to know about science. first every scientific theory makes assumptions very very simple thing to note as a result there is no theory of everything because you don't explain your assumptions you assume your assumptions your assumptions of your theory are miracles for for that theory the flat-out miracles because you don't explain them and of course you can give me a deeper theory that explains those assumptions but your new theory

will have its own miracles its own assumptions and this goes on at infinitum so there is no theory So that means that every scientific theory has a scope, if it was a good theory. If it was a bad theory, we don't care about. But if it's a good theory, it has a scope, and it gives you the mathematical tools to explore the power of its scope. But it will have a limit. And if it's a great theory, the theory will give you the tools to discover the limits. And in case of Einstein's theory of space time, he gave us essential tools to discover the limits of his very central concepts of space and time, namely 10 to the minus 33 centimeters and 10 to the minus 43 seconds. It's over for space time. So it's not that we wouldn't want to say, oh, well, Einstein is, you know, Einstein's mathematics has refuted, poor Einstein, he got himself caught in a logical bind. He shot himself in the foot. He had these ideas about space and time, curving and so forth. And then, but then, you know, he came up with this mathematics and the mathematics came back and showed, well, that idea of space time falls apart at 10 to the minus 30. Oh, well, Einstein, you dark. You, you, you, you, you, you, you're,

dummy you look well no that's not our attitude it's like no it's a great theory because Einstein gave us the tools to find out the the not only the scope but the limits of these very fundamental concepts these were professional philosophers publishing in some cases in professional philosophical journals so so the the misunderstanding of science is profound in that so that's the fun interesting thing this happened in the last I've got a paper coming a couple of and reply to this episode is brought to you by Netflix most valuable promotions in Netflix are hosting a blockbuster triple headliner Saturday May 16th ronda rousey returns to face fellow woman's m-mae pioneer jina carano in the main event plus comains Nate Diaz versus mike Perry and the best have

you wait in the world francis angano versus Felipe lins watch ronda rousey versus jina carano live only on Netflix Saturday may 16th at 9 p.m eastern center time 6 p.m. Pacific time Please. Another question that a lot of my audience has had, and I've had it too, is this trope that the map is not the territory. You know, that you're explaining a modality to explain something, but now it's not clear where these matrices end and like real reality would begin. What do you say in those cases?

What do you make of that as a criticism, that describing something as a markoff process us is inherently the, which is inherently the byproduct of a human brain, someone called Markov, you know, originally, but, you know, where does it end? Is it important to truly have, you know, some base layer where you say this is where we start from? And these are axioms. And then even if you could do that, sorry to ask a double-barrow question, but if you could do that, you know, what would Gertil say, you know, that, how can it be complete given that? So, The map is not the territory. How do you answer that? And then maybe Gerdel's objection. I couldn't agree more with the first objection. And so when I say that there is no scientific theory of everything, that includes Hoffman's theory.

And that's the predicament of science, is that it, well, it's, you know, this is a predicament. If you wanted a theory of everything, then this is a bitter pill to swallow, right? But there is no theory of everything. If you want job security, this is great. So job security for science, this is really great. And my own view about scientific theories is that whatever reality is, a scientific theory is merely a description of a particular perspective at best. At best, it's a description of a particular perspective on reality.

And my guess is that there's an infinite number of perspectives that one could take, not just one or two, an infinite number of different perspectives that one can take. and in some sense the perspective that we're taking is what we what we call space time that's our little headset that's the little headset that we use to get our little you know projection of whatever this this reality is and again i when i say that of course i'm already using terms of my own theory and so at this point now i can all i say is those terms are used as pointers there is a reality and and we're just pointing to a description of a projection of that of that reality. But it turns out most of our theories are not even good at that. So most of the stuff that we come up with is not even good as a description of a projection of reality, right? But every once in a while we get something that is a good description of a projection of reality, and then we can build technology with it. So there is something useful to be done here. I don't think we're getting closer to the truth. If you ask me, what is the truth? My own guess is you are the

the truth. But to know that truth, you have to let go of any concept of who you are. You have to let go of all concepts and just be with yourself. Then you're knowing with that. It's a non-conceptual knowing of who you are as the truth. So now that gets to the Girdle question, right? So I take scientific theories as mathematically precise systems, right? And they have the, at least the power of arithmetic, which is something that Gertl was looking at. So, they will in general have the power of arithmetic. And so therefore there will be statements using girdle's approach that are true, but cannot be proven within whatever axiom system you've chosen for your scientific theory.

So I think that that girdle incompleteness thing is one, another reason why I'm saying pretty emphatically, there is no scientific theory of everything. And in fact, we don't even get close, right? that in some sense, the reality, whatever it is, infinitely transcends any scientific theory, including my own, infinitely transcends. But it's not all, hope is not all lost, because I'm suggesting that what you are is that truth, that reality, just looking at itself through a particular avatar. In this case, there's a, you know, a Brian and a Don avatar, but it's the one, there's this one

reality looking at itself through different avatars, and a particularly simple headset, and only a four-dimensional headset. So it was a fairly trivial, and I can imagine that there's an infinity of other headsets, for example, with maybe billions of dimensions, trillions of dimensions. Who knows? And we're probably, you know, the more dumb one. Okay, some questions from my audience. Abe Chung asked, Dr. Hoffman, I'd love to hear more about the interface for reality in the context of linguistics. particularly text linguistics. I am a translator by trade for 20 years,

and studying this field constantly reminds me of the lectures you have given. My studies are not enough to capture anything at the level that you do, but I have a gut feeling you've seen more down there. Can you say something about linguistics and applications of the interface for reality in that context? The language that we use

represents in some sense a worldview, in part. And the way we learn language, How do you actually learn the words of a language when you're teaching a kid, for example, at 18 months of age or something like that? You have Johnny there, and there's a rabbit on a rug in your living room. When the time is right, you're there with Johnny and you point your finger to the rabbit, and you say rabbit. And if Johnny's old enough and attentive, he gets it. it's actually that's what we call ostensive definition so in this case we almost every term that we learn in a language is by ostensive definition

and notice that there's an infinite ambiguity if I point and say rabbit what in the world am I referring to it could be the color of the rabbit it's white or it's fur the furry texture or it could be the left ear and the right front paw or it could be the tail and the carpet. So an object composed of the tail of the rabbit and the carpet that is sitting on. So what does the word rabbit refer to? And yet somehow, and also there's something called the level of which you're doing at the basic level. So when you point, you're going to say rabbit, you're not going to say, oh, biped or quadrupet.

You won't say, oh, Johnny, quadrupet. It is a quadruped, but you know that you'd be messing up your kid. The first thing you said was not rabbit, but quadruped. Or even if you tried to say the color first, you learn that you have to say the shape first, and then the color and higher things. So we have, what's really interesting is Johnny has to share this perceptual interface with mom and dad, and whoever it's, right, you have to have the same evolutionary interface. So this whole interface, so now I'm tying in the interface theory of perception with this linguistic thing.

And for Johnny to even pick up what you're talking about, he has to have the same kinds of entities in his interface. Otherwise, I mean, if you try to do that with, you know, a rabbit, if you take a rabbit and point and say rabbit, it's not going to get it. It's got a different interface. So, yeah, there's a really big connection. And in some sense, everything that you know, all the words that refer to objects and properties in a language, you only learn by ostensive definition. That means you assume a shared headset. So there's, of course, you can see a huge, the language depends hugely on an assumption of some shared properties of headsets. Hey, there's a good chance you might be a scientist or an engineer aspiring to be, maybe going to school, graduate school,

or after school, or maybe you're a professor like me. If you're wanting to learn the greatest tips and ways to become your best scientist, you might want to get my book into The Impossible. Think like a Nobel Prize winner, with a forward by my friend Nobel Laureate, Barry Barish. In it, we describe an incredible series of tips on how to collaborate better,

unlock your creative genius, and get over common pitfalls like the Imposter Syndrome. I hope you'll take a deep dive into it, and I know you'll enjoy it. You can read a free chapter at my website, website, Brian Keating.com, slash books, and you can buy it at Amazon.com, an e-book, audiobook, or in physical hard copy or paperback form. Thanks a lot. And maybe the last question I have from a listener is, his name is Los and Donio, which I considered for one of my children's

names. He says like this, suppose the argument for only conscious agents existing depends on an evolutionary argument. In that case, it seems like the biological entities and their systems involved in perception are somehow outside of or more fundamental than perceptions of a conscious agent. Yet they, our brains, eyes, et cetera, must also be merely products of evolution and predictive modeling and not real entities. Does this lead to an internal contradiction in your theory of conscious agents? If you approach it that way, you might get a contradiction, but I approach it in a different way. So the way I approach it is the following. we can write down to

a more coven dynamics of conscious agents that is called stationary. And that means that the entropy at each step is the same. So there's no entropic arrow of time in this dynamics. But it's a theorem,

very simple theorem to prove that if you take a projection of that stationary dynamics, say by conditional probability, you'll get a new dynamics, the Markovian dynamics, that's a projection, But the projected dynamics will have increasing entropy.

So it will have an entropic arrow of time. So my take on this is, I think that the conscious agent dynamics of consciousness is stationary. So when we get an arrow of time in our headset, that's entirely an artifact of the projection. So time is not fundamental. The entropic arrow of time is not fundamental. Now, this now ties into the evolution question. What is the most fundamental limited resource in evolution? It's time.

If I don't breathe in time, I die. If I don't eat in time, I die. If I don't meet in time, I don't reproduce. Time is the most fundamental limited resource. So what I'm saying is I love Darwin's theory. I've studied it. I think it's incredible. If you're going to talk inside of our space-time headset about biological evolution, use Darwin.

That's the right theory inside of our headset. having said all those nice things about Darwin, and I'll now say every aspect of that theory, organisms competing for resources, limited time, nature, red, and tooth and claw, every aspect of Darwin's theory is not an insight into a deeper reality. Every aspect is entirely an artifact of loss of information and a projection into this headset.

So the theory of conscious agents need have no limited resources, no competition, no nature, red, and tooth and claw. cooperation and love possibly right it could be all that but in the projection when you lose information it will look like there's an arrow of time nature red and tooth and claw predictive processing mark out blankets the whole all that whole thing of me versus the separation between entities will all appear there so you can see how i i'm reframing the entire question i'm not trying to boot i don't take evolution what i did with evolution was i said that's our current theory i'm I'm trying to find the scope and limits of our current theories, right?

So I used evolution to show that organisms in space and time can't be the fundamental reality. Right? That's what, it's the inside game that you play in science. You use your own theory to show its own limits. That's what we did. So given that, great. So now I have to make a leap. I make a leap entirely beyond space time.

Evolution doesn't tell me what leap to take. I'm just making a leap. I'm proposing there are these positive geometries. There's decorative permutations. and beyond them, a theory, a network of conscious agents. Now, of course, what I have to do is what I just sketched. If I start with the theory of conscious agents outside of space time, how do I get evolution inside space time?

Answer, as an artifact of loss of information in the projection. Then it looks like nature red and truth and claw, fighting for resources in limited time. But that's not true of the deeper picture. But the deeper picture shows how this limited headset picture arises as a loss of information in a special case. And that's the way science works. I want to ask one last question with my host prerogative, and that's just, I notice that

you are now listed as emeritus professor at UC Irvine just up the road as an anteater. I want to ask you just quick reflections on a career in academia, but with a particular eye towards the future of academia for, you know, slightly younger, though not much younger, but folks like me and even a new professor, will academia look the same in 20 years when I'm, you know, emeritus or when a new assistant professor today that we just hired and she's going to be emeritus in 40 years, what will academia look like? And what will in particular the field of artificial intelligence, obviously very closely related, a lot of your wonderful images came from AI generated art, which I love. How will AI change our profession or how will other forces protest and disruptions and encampments? What do you see as a future of academia with the benefit of this story, legendary career?

that you've had? My best guess is that AI is going to have a profound impact. In fact, when you're asking the question, that was the first thing that came to mind is that AI, right now, AI is my closest collaborator, right? Because on mathematical things, I always ask Chaiton, but sometimes my mathematician friend, Chaiton, I've worked with love for 35, 40 years, right? But, you know, he's a person, he's got a life. And he can't be at my beck and call every time I, Oh, Chaiton, what? You know, tell me about this math. He's not a grad student anymore.

That's right. So I just go and ask the AI to explain the stuff to me, and it does. It does a good job. So already, my, just in the last year, my whole style of research has really actually accelerated because I don't have to wait. And I can actually ask a series of questions and I can get a pretty, I mean, you have to tell when the AI, sometimes the AI gets it wrong. So you have to be careful.

Well, I think that students, Ultimately, I'm not sure we're going to need us to teach, frankly. It won't happen right away. But if you're looking 20 years down the line, I don't see too many people being able to compete with AI as 20 years down the line in terms of didactic skills and so forth. And right now is the worst AI is going to be, right? I mean, it's only going to get better.

And imagine my kids, a 10-year-old, and, you know, he or she has had AI, you know, reading over their emails and let's say it's benevolent, you know, who knows about Google. But, you know, everything will be on Google. You know, I wasn't on Google when I was 10. It didn't exist for them, you know, it took 20 more years, but almost before it existed. So imagine everything, every single thing, every thought, every email to collaborator. And again, it could be malevolent.

But let's just assume it's benevolent. Just the imagine, you know, it's truly a second, third, fourth, fifth brain and just the capability of that. But, you know, I wonder, I had some conversations last week with David Berlinski, who you probably know. He's a wonderful thinker and Raccontour. And he was talking about a conversation he had in 1965 with Noam Chomsky and B.F. Skinner and all these incredible minds that are heroes, right, to many people. And the question that came up of, you know, will, you know, computers ever really be able to reason? and he asked this of Chomsky, and Chomsky said, you know, the probability of, you know, in the stochastic system of the next word being selected is zero.

You know, basically there's an infinite number of words and then you add in context, grammar, syntax, lexicon, vocation, all these. So the probability, and yet I'm speaking to you right now. And I'm not even thinking about what I'm saying. It's probably obvious to you. But the point being that Chomsky, even the great Chomsky, thought that these LMs were impossible. you know, as recently as, you know, 30 years, 40 years ago even, maybe even more recent than that. So, and yet I push back on David, who's sort of even suggesting that AI will come up with these laws of physics. I mean, I don't have to remind you, but, you know, Einstein here, you know, he said his happiest thought was that if he was in free fall, he'd experienced no gravitational field.

And that led to the Einstein equivalence principle, which then overlaid onto it, you know, Riemannian geometry and everything. to what extent, you know, can chat GPT produce an amputahedron, you know, or an Einstein equivalent's principle? Is it just a matter of tokens? Because I'm very optimistic about the need for humans, although I agree our days, at least as we do it, are numbered. But I'm no Einstein. And the question of, you know, creativity at the level of, you know, I made a joke the other day

on X, I said, you know, is it, I assume, I stipulate that computers were. will always be better at humans than chess, but I don't believe a computer will ever invent a game like chess. So where is the magic, you know, where's the room for physicists, for cognitive scientists? Is it in creativity? Are there things that LLM certainly cannot replicate just based on their architecture? Earlier I said that I think that you are the truth. You are the reality and there's this one infinite reality that transcends any concepts

that's talking to itself through a Don and Brian. avatars and so forth. And that's relevant to this question, because if we look at an LLM or future AI, it's going to be some kind of computational system. And computational systems have their limits, right? We have the girdle limits at a minimum. I'm betting that you and I are avatars of an infinite intelligence. that's the true source of all creativity and that Einstein and any scientist taps into when they come up with their truly original ideas. The first step will be to say,

if it's just a computational device, then it doesn't transcend. It doesn't transcend and doesn't get to this true originality. But then there's a deeper level to trying to address your question. Because in some sense, okay, well, what is this thing we call Don and Brian? Well, these are just avatars in a particular headset, right? So why can't we create a new avatar in our headset that looks like a computer that is actually an avatar of the deep intelligence, right?

If we're going to only stick with a computational thing, then I think there's no problem. But somehow the Brian avatar is a portal into the infinite consciousness. And the Dawn avatar is a portal into it. And we know one way of creating new portals into this infinite conscience, and that is having kids. So that's one technology that allows us to do that. So could we, once we understand space time as a headset, and we understand the headset and at least some layers of its relationship to the infinite reality, that's a source of true insight and true originality and creativity.

Could we figure out new ways to open up genuine new portals? And might that look like AI? There I'll have to say the jury's out. So the first one, I would say, if it's just an LLM kind of, or just a computation device, no, then there's a level of creativity that is going to miss. But is it possible that we could actually open new portals into the consciousness behind? then it's a different game. Excellent.

Well, Don, this is a wonderful. I could talk to you forever. Maybe I'll come up to Irvine. Next time we'll do it in person. Oh, it would be a pleasure. Absolutely. Great pleasure, Brian. Yeah.

Don, you're such a fan favorite, my favorite. Love talking to you. And hopefully we'll do it again, like I said, in person sometime. I would love to do that, Brian. Great to talk. Okay. Stay well. Bye, Don.