Into the Impossible With Brian Keating - Christof Koch Exposes the Fatal Flaw in AI Consciousness [Ep. 484]

Episode Date: March 23, 2025

Pique is offering 20% off for life AND a free Starter Kit with your purchase—that's a rechargeable frother and glass beaker to make the perfect cup every time. Just go to http://piquelife.com/impossible Please join my mailing list here 👉 https://briankeating.com/list to win a meteorite 💥 What is consciousness? Will we ever be able to create digital twins of ourselves that are conscious? And can artificial intelligence ever truly experience consciousness? Here today to enlighten us on these elusive yet fascinating topics is Christof Koch! Christof is a renowned neuroscientist and one of the leading figures in the field of consciousness. He is best known for his work in exploring the neural bases of consciousness, particularly in relation to the brain's structure and function. A key proponent of Integrated Information Theory, Christof is also about to release his new book, Then I Am Myself the World, an essential read for anyone seeking to understand humankind and the future we are creating. Join Christof and me as we delve into the mystical and mysterious world of consciousness. — Key Takeaways:  00:00 Intro  00:43 Judging a book by its cover 03:03 Defining consciousness  05:55 Progress in understanding consciousness  07:43 Integrated information theory  12:42 Consciousness in non-human systems  15:58 Meditation, psychedelics and dancing  17:58 AI scientists and digital twins  26:56 The perception box  31:43 Quantum mechanics and consciousness 38:25 How to calculate phi  44:01 Quantum computers and consciousness 48:00 Audience questions  52:49 Outro   — Additional resources:  ➡️ Learn more about Christof Koch: 💻 Website: https://christofkoch.com/  📚 Then I Am Myself the World: https://a.co/d/agleHRB  ➡️ 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

This episode is brought to you by Netflix. Most valuable promotions in Netflix are hosting a blockbuster triple headliner Saturday, May 16th. Rhonda Rousey returns to face fellow woman's MMA pioneer Gina Carano in the main event. Plus co-main's Nate Diaz versus Mike Perry. And the best heavyweight in the world, Francis Ngano versus Felipe Lenz. Watch Rhonda Rousey versus Gina Carrano, live only on Netflix. Saturday, May 16th at 9 p.m. Eastern Center time, 6 p.m. Pacific time. They can do everything we can do or will be able to do everything we can do,

but they don't have any state of being. They don't exist for themselves. They're all just like my garbage collector. It doesn't feel like anything to be in LLM. They will never be what we are conscious. Any sufficiently advanced technology is indistinguishable from magic. Open the pod bay doors, Hal. Christoph, great to meet you at least virtually.

It's been 25 years since we were together at Caltech. How are you doing? I'm well. It's snowing here in Seattle, which is unusual. Christoph, so today we're here to talk about your really legendary career and the contributions that you've made, maybe more on the physics side to the theory of consciousness, et cetera, than you have talked about in the past. But I want to start, as I often do, with doing that thing you're never supposed to do, which is to judge a book by its cover. So they say not to do that, but what else do you have to go on if you've never read the book?

So if you'd show the book, there it is. Then I am myself the world. And this is kind of an exploration. It's almost like a travel log, at least in some ways, with a near-death experience. I wonder if you could tell us the origin of the title, the subtitle, and the artwork is, yeah, I could probably understand it, but tell me, besides the fact that it's the University of California colors, what does the title and subtitle represent? So the title comes actually not from a near-death experience,

but from a different experience, a mystical experience, and it's taken from, it's translated from the German, It's from Richard Wagner opera, the second act, Tristan Isolde, in which Tristan Isolde, the eponymous lovers, fall into love and they try to overcome the individuality. So Tristan wants to become Isolde, and Isolde wants to become Tristan, and then they have this moment of rapture

I've written to some of the most amazing music in the Western canon, where they become one with each other and with the universe. So it's one of the defining aspects of a mystical experience or spiritual experience when suddenly you feel the borders between you. I know I'm me and when I touch this cup, that's not me. Even if I shake your hand, where to shake your hand, I can recognize that's a different hand. That hand doesn't belong to my body. So under certain circumstances, these boundaries can dissolve. And then you can have this extraordinary experience where whatever remains of your consciousness, there isn't a self anymore in the conventional sense. There isn't Christoph anymore.

Instead, what there is, there is this identity with everything in the universe. You become one with the universe. And typically also space and time very often also disappears. In other words, the moment isn't too short or too long. It simply is. And the very perception of space can also disappear under these extraordinary circumstances. So that's what the title all alludes to. Now, when you talk about consciousness being the foundation of existence,

I get frustrated because I still don't have a good notion of what consciousness is. And I've talked to many of the leading luminaries, your colleagues, and many of the top luminaries. And it seems to be this tautological thing that either you can only define self-referentially or it is something so simplistic that we can say that everything participates in consciousness. So how do you, as a consciousness professional, what do you think of as consciousness? What is the base layer of reality when it comes to consciousness? Do you hear me?

Yeah, I hear you. Yeah. You hear me. Yes, I do. What is hearing? Hearing is a conscious experience. It's not only behavior. Yeah, when I hear you, you ask me a question, I can respond.

But it's just that conscious perception of hearing. And there's nothing in the loss of physics. that tells us that systems here. If you look at quantum mechanics, there isn't anything about hearing or seeing. And hearing is just one instance of a trillion different varieties of conscious experience. It's seeing, hearing, feeling my body,

being bored, being in love, hating, dreading, imagining, dreaming. Those are all different conscious states. You cannot imagine, in fact, many people, Schrodinger has written very evocative about that in Einstein as well. You cannot be a scientist without accessing the world by seeing it, by looking at a scope, by hearing other people talking about it, by doing

manipulations of mathematics in your head, all of those involve conscious experience. So conscious experience is the Omphalov, it's a center of existence. To me as a conscious being, when I go tonight to sleep, we all do that, particularly in the early phase of the night, you go into what's called deep sleep or delta sleep, which is characterized by these deep waves or at low frequency, 2 to 4 hertz, crisscrossing the brain. At that point, you do not exist for yourself. If I wake you up, if I wake up volunteers

while they're in deep sleep, a non-dream state, and ask, did anything go through your mind? Most people will say, no, nothing. I came from nowhere and suddenly I am. So at that point, you do not exist for yourself anymore. You exist for others. Your bad partner can still see or interact your sleeping body, but you don't exist for yourself.

So therefore, consciousness is really the only way it is so elementary. the only way we can define ourselves. When I'm anesthetized during surgery, I don't exist for myself. When I'm dead, presumably, I don't exist for myself. In that sense, prior to anything else comes consciousness, because that is the only way I exist for myself or things in general exist for themselves to the extent that it feels like something to be a dog or a person or fetus or a collection of So the collection of those experiences, a subjective experience, and yet we have the famous book, What Is it Like to Be a Bat, or the famous essay by Thomas Nagel. I want to write a book, What Does It Like to be Thomas Nagel by a bat?

The essence seems to be from a lay perspective, which is the only approach I can bring in it. We don't know and we can't know. We don't know what. You don't know what it's like to be a bat. In other words, there's a batness and essence of batteness that we can never exercise. That essay was written 52 years ago. And has there been progress? I mean, is there a sense of frustration that either we don't accept the definition or

and that there is an actual definition or we haven't made progress? Are we making progress to understanding at a deep level a fundamental accrete upon definition of consciousness? Or is that a fool's error? I don't think it's a question of making progress. We have the only definition of conscience that makes sense. Because it is private, because it is subjective by definition, that's a key character of conscience, you can't remove that.

So I don't think it's a question of, well, do we need to get different definition? What we don't have, what we have made some progress, but there's no agreed upon, universal agreed upon, is a theories of consciousness, a theory that would explain, is this conscious, this cup, and if not, why not? Well, what's wrong with it? Is it the constitution?

Is it somehow these carbon molecules in the ceramic, they don't have the right stuff and bilipid membranes do? Or is it the way they're wired? up, is it the functional interactions? Right? And now, we don't have that. Again, there's nothing in physics on chemistry in their biology that would tell us that certain systems, be the physical or chemical or biological, have this interior aspect.

So that's the brute fact of the brute fact that we're all confronted with that demands an explanation. Well, we have made progress on some individual theories of conscience, in particular integrated information theory. I'm biased, of course, because I've participated it. But that, let me step back. In terms of the landscape of theories of consciousness, most philosophers, including most philosophers like Dan Dennett, for example, and large number of people like him, are so-called functionalist computationalist computationalist. They say consciousness has one or more functions, whatever your function is, summarizing the environment, making plants, whatever. And once you instantiate those functions on a universal tooling machine or approximation they are all, of these systems will be conscious. And so by that definition, LLM's chat GPD either are conscious now already or very soon,

once you add the latest widget, take chat GPD 04 or 03 or deep seek or whatever. So either they're conscious or very soon will be conscious. Most theories, even in psychology and neuroscience, are of that ilk. So, for example, the global workspace theory, another very popular contender for theory of consciousness, says the function of consciousness is to take information and broadcast it widely throughout the brain. It derives from a computer architecture called Blackboard architecture, where different local processors write the output onto a central memory at Blackboard, and that information gets broadcasted through everyone. And the theory asserts that this act of broadcasting, that's what consciousness

is. And yes, if you build this into a computer or assimilated, then this thing will also be conscious. I.R.T. Integrated information theory is the minority view that says no, it's nothing to do with computation. It's not a question of a clever hack. There will never be a clever hack that can instantiate conscience. Conscience ultimately is about being. It has to do with the constitution of certain physical systems. You have to be built in a certain way. There's nothing supernatural about consciousness. But IAT is ultimately consciousness is not a process. It's not a function. It's not a computation, it's a structure. It's a causal structure. And so the way you, the way IAT starts, it starts with consciousness. Rather than taking the brain and try and squeeze it very hard and get

the juice of the wine of consciousness out of the water of the brain, out of the water of a mechanism, which is what most people try to do. They say, well, if I squeeze hard enough, it's 40 hertz oscillation. If I squeeze hard enough, it's global workspace. Or if I really squeeze hard enough, it's the collapse of the wave function a la Penrose, right? But in all three cases, why is it that the collapse of the wave function should go hand in hand with conscious? Why is it that broadcasting information to everyone should go hand in hand with consciousness? Why should 40 hertz go hand in hand with consciousness? Not 32 hertz or not 50 hertz? What's magical about all of these things? IAT starts differently. IT says, well, let's start with conscience. Conscience is central, and it's the

only thing that truly exists for itself. And consciousness has certain number of properties. It's one. It's specific. In other words, when I'm conscious right now of you, I have a particular experience of seeing you, of hearing you, of seeing, of feeling my body sitting on a chair while we're doing this interview. So it's very specific out of trillions of other experiences. It has boundaries. There are certain things that are inside consciousness and most things that are not inside it. It's also structured. It has parts and subparts. There's left and right and up and down and space and time and all of those things. So it starts with experience itself and then looks, okay, now I'm looking for a substrate, a causal substrate that reflect these properties of any one conscious experience. And then it ends up with a calculus. So if you give me a description of any system, a formal description of a system in terms of a transition probability matrix.

So you have a system that is so many elements, whether they're neurons or transistors or whatever, and they're in a particular state. And if you tell me, if I'm in this state, then it leads to that state, then it comes and then it goes to this state. If I have a complete description of a system, then IAT says, well, I can now, in principle, unfold the causal power of this system mathematically, and I can derive both its quantity of consciousness. This is called phi.

It's a number. It's a pure number. It can be zero. positive, the bigger the phi, the more the system exists for itself, the more the system is irreducible, the more the system is consciousness. But it also derives a mathematical structure, and that mathematical structure, that's a central claim of the theory. That mathematical structure is what conscious experience is. So the challenge is this mathematical structure has to explain why space feels extended, why time always flows, in fact, it always ever flows forward,

why I think like colors feel the way they are, why being bored feels different than being in love, etc. By that definition, could we say that in the panpsychist network approach to consciousness, or even say a mycelium network where you've got this massive largest structures on Earth, that seems to contain a lot of information, it seems to occupy space, it seems to propagate only forwards in time? I mean, would you say that it is conscious and if not, why not? So potentially even a single cell that has vast, molecular networks of interaction, right?

It contains maybe a billion molecules of 10,000 different proteins. They all interact in very complicated, nonlinear synergistic manner. Yes, it may well feel a little bit like something to be a paramecium. And once the membrane, the walls of the paramedium dissolves, it's dead. It doesn't feel like anything anymore. But it also says quite clearly not everything exists. So, for instance, this is not conscious. this isn't conscious and the combination of the two isn't conscious either.

By the way, the theory also explains the interesting fact that you and I, although we interact, are not conscious, there isn't a Brian Christof Uber consciousness. You're conscious. There's me, I'm conscious. We do interact causally. I ask you something. You respond to it, et cetera. But it's not.

Why not? Well, because the theory says what's conscious are always these local maxima of integrated information, local maximum of fire. So there's lots of integration within my brain, particularly in my cortex. There's lots of integrated information in your brain. But now if we look at the causal interaction among us, they're minute and they're totally dwarf compared to the interaction within my brain and within your brain. Therefore, there's a local maximum in you, that's you, Brian.

There's a local maximum in me. That's Christoph, but there isn't an Uber consciousness. Now, in principle, I can do an experiment where I can start connecting my brain to your brain with like Elon Musk-like neurolink technology. Right. You have some project like that too, right? So in principle, once we have enough connectivity, then what we could achieve, we could add so many wires between our brains, just like the wires that the 200 million, you probably

know that 200 million fibers called the corpus callosum that connect the left brain with the right hemisphere, right? And sometimes they can be cut during split brain surgery. So when I build this artificially, then it may well happen. that at a particular point once I add more and more wires between your brain and my brain, that Brian disappears, Christoph disappears, instead there will be this new amalgamation of Brian Christoph that has one consciousness but speaks through two mouths and has four legs and four arms, etc. Actually the inverse of what happens during split brain.

When I sever these connections between the left and the right hemisphere surgically, what I get are two conscious entity inside one body. Typically, only one speaks. Typically, it's the left hemisphere. But you can observe, if you observe these patients, particularly after they had surgery, you can observe intermanual conflicts, etc. As far as we can tell, there are two conscious entities, and they may as well be on other sides of the moon. This guy has no direct experience of what this hemisphere, they're both their own independent conscious minds. You mentioned in this book, and that's why I called it a travel log or an experimentalist guide in this sense. You talk about different experiences that you had from meditation to dancing to even psychedelics. How did those transform you? Did they give you a perspective that you wouldn't have had, say, if you didn't try psychedelics in this case? Which I think with mushrooms, you tried mushrooms in this case?

I started off being a dance and a rock climber. And what I experienced when I was younger are these wonderful states, which is both why dancing and rock climbing, particularly big walls. I love big walls in Yosemite, etc. can be very addicting because you lose yourself. You lose the sense of self. You go into

what people call the sort of a trans like the zone, the flow. The flow state, these are all different words where you're highly conscious, right, if you're dancing, you're incredible conscious of the music of course and where your partner is and where you are positioned in 3D space. When you climb, you're

highly conscious of the wall of any tiny indentation in the granite and you're always hyper aware of the ground, the big air below you, but you're not conscious of yourself. All your problems, you fight with your spouse and their homework and all of that is totally gone. It's the deep state of happiness, of contentment, which I think is why people love the states of these states of extended flow where you lose yourself. We need a sense of self, right?

We need a sense of self in order to get through PhD, through grad school, to do a difficult task. But they can also, the self can also be a cause to bear, right? I constantly worry about myself. I get anxious. What's going to happen with Trump and all my grants and all of that stuff. So it's good to occasionally experience the world without itself because you realize even with herself out of the way, the world is beautiful and transformative and transcendent. And so, of course, you can definitely get that in doing certain types of psychedelic states, right? When including up to complete ego dissolution where you're highly conscious, there is no more. voice in your head. There's no more crystal. Christopher's gone.

You mentioned in the book about digital twins and mind uploading, things like that. I am not a dumer the way my friend Max Tagmark is. And I have a very simple Turing test analog, which is that when an AI system can create a new law of physics, say, for example, the way that Einstein came up with the equivalence principle. It was by visualizing the visceral sensation of freefall. And he realized that an observer in free fall would not experience a gravitational force. field, I'm not very optimistic that a computer silicon version of Einstein, AI Einstein, can even visualize what happiness means. I mean, Einstein said it was the happiest thought of his life in 1907 when he had this realization. And then B, what does it even mean for a computer to visualize

the sensation of free fall? So my criterion has been, when can these systems come up with new laws of physics that can be verified through observational data, the kind of which my colleagues and I collect through our experiments, instruments, apparatus. And I think we're a long way away from that. And I think we are not likely to get there through GPUs and Hopper H-100s coupled to token-guessing large language models. Are you, first of all, are you more optimistic than I am, say, that an AI system can actually construct new, legitimate laws of nature rather than just assembling language that humans have already created based on its training data? Yes, I do think so. Do you know this paper?

Can't see who. I can see the title, AI scientist, yeah. What they did here, I think a group in Sakana, a company in Japan, they got an LLM to come up with its own hypothesis about the mathematics of diffusive learning. The algorithm, the AI, came up with its own computational way of testing this in simulated, this is not physics. This is in the domain of machine learning. To simulate that, to collect the results.

to generate graphs and to write a complete paper for new ribs. All at the, they mentioned this facetiously, all at the cost of $15. So essentially you get a graduate level, PhD level paper that goes to a high-impact journal for $15 by an LLM. I think anything that humans can do, machine can sooner or later, and I suspect very soon, not just 50 years from now, but maybe half a year from now, this will happen. Because people keep on saying over the last three years, we're now effectively in year three of the machine intelligence, right? We are now in year three of the age of intelligent machines, right? If you date year zero, the Big Bang to October 22, and every time people say, well, it can't

still do this, and you wait three months, and some LLM somewhere can learn to do it. Because why is it so surprising. They take everything, they're like vampires. They take everything humanity has produced, put it out on the web, including mathematical proofs, including text and computer code. So they're very good at take these symbols, predict the next symbol, whether that symbol is a token from human doing poetry or talking about their experiences or a symbol is a mathematical symbol or is a computer language, like a C code, right? It's all the same to them. And they just predict the next and the next. So they can do everything we can do or will be able to do everything we can do, but they don't have any state of being. They don't exist for themselves. They're all

just like my garbage collector. It doesn't feel like anything to be an LLM. And as long as they're built autonomy, it has to do with the hardware, of course. So as long as they use digital computers that have a radical different connectivity and from these guys, they have minimal phi, minimal integrated information so they will never be what we are conscious. There's no awareness that they have of their own. There is never any awareness, although of course they claim. Why? I mean, because they've read every novel and most novels about feeling, the interior lives of the protagonist. Well, so they can regurgitate that. They've been trained. They've sucked up every book written by Schumann. That's why I'm not as optimistic that they can construct new laws.

For example, the lack of being able to create a theory of everything is not predicated on it needs to know this Gladiator 2, the plot of Gladiator 2, and it hasn't gotten that yet because they haven't uploaded it to 2025 yet. They're very good. They have an IQ of 190 in every single subject that humans have, but since they're trained, as you said, on what humans do, they're bound to what humans can do. There's a phenomenon called the lock-in effect where allegedly the height of a rocket is determined by the width of a horse's butt in the Roman Empire, because that's how wide train tracks were. I don't know if it's true, but the I idea makes sense that these things are so successful, the architecture of LLMs, it's very different

than the way that you would do theoretical physics, though. For example, would this object come up with the wild curvature hypothesis of Sir Roger Penrose and say that consciousness depends on the value of the local gravitational vile curvature? I tend to think that probably wouldn't do, and I actually have a proof of it because we took a thousand years of Mercury's perihelian data, we put it into an LLM, and we said, given this, what's missing? What is causing Mercury's anomalous perihelian advance of the famous 70 arc seconds per century? We needed to tell it that you have to discretize space time. Did it come up with Vulcan? The Vulcan hypothesis? It didn't come up with Vulcan either. It just said we had to actually

force it to accept a Maxwell. We basically encoded gravitation as a new electromagnetic force and then it could understand and it could predict it for the next 20 centuries. But it couldn't come up with the fact that space time is curved to get back to my Einstein equivalence principle. In other words, there seems to be some ghost in the machine, a zeitgeist or whatever you might call it in German. But the point is, if it can do everything that a human can do, that means it's better than all the humans together, but it's not some new human. And I wonder, if you talk about these digital twins, what would it be like to upload a mind? I mean, would you want to turn off the computer? I mean, you said it doesn't have a sense of consciousness,

but what is the digital twin involved? What does that project actually do to instantiate a mind uploaded? Are you making the claim that AI's LLM at least will never be able to replicate the peak of human creativity? I think there's an essential qualia that theoretical physics represents. That would exclude most people on the planet, though. Of course, yeah. But AI is promised to have an IQ of 400 and just something completely different than we've ever had contact with. So all I'm saying is LLMs, as they are, token matching context, where they're able to predict very successfully the next token and even write latex code and make that paper that you showed me.

That's very different from coming up with this insight that observer in free fall would not feel gravitational force. I agree. I'm not saying the same. All I'm saying I've observed particularly over the last three years that people make this argument, it still can't do X. And then three months later, they can do X. And then they say, oh, no, but there's now why. It still can't do Y. right? And then again, so I'm skeptical about that claim. However, my central claim is that I'm not

skeptical about it because I can show mathematically within the, assuming that integrated information theory is the correct way of looking at consciousness, then you can show none of these LLMs running on machines in the cloud will ever be conscious. So my digital twin, in the future of time, if you could reconstruct my brain without killing me in the process by cutting up my brain in getting the so-called connectome, the complete wiring, and instantiating all my neurons, which are, of course, vastly more complex than simple point neurons, simulating them somewhere in the cloud, right? That's the idea of sort of digital immortality. Then this thing might well in the limit be able to behave like me and possibly even speak like me. I mean, I can do that already

today, right? I don't need a lot of training data to get an algorithm to speak like me with a Germanic accent, it won't feel like anything. Now, it can fool other people because it looks like me, right? You'll see an avatar that behaves like me. It sounds like me. So of course, you're going to say, well, that's Christop. But it's all a deep fake. It's all a deep fake. This thing, this digital stone will not feel like anything. And so we're going to live. If that future comes to pass, we would live in a world where we surround it by zombies, literally, all these things that don't exist, but that they pretend they exist. And we don't know the difference, including people, will fall love in with these digital, including, of course, people will ask, well, we should give

him rights. We should give them moral rights and legal rights, because it's clearly a conscious thing, but I think it's not. It's a simulacrum of something. It's a zombie, literally. It's a deep fake. But I think that's the future we are heading into. Similar concept. Peak my interest was this concept of the perception box. I've interpreted it loosely like Donald Hoffman has this desktop interface. Can you explain it to the audience? What do you mean by this? Well, okay, so the perception box is a metaphor coined by Elizabeth Koch. I'm not, although we share the last name, we're not related, is the idea that I think goes back in philosophy hundreds of years, in fact, particular to Emmanuel Kant, that your reality, Brian,

is constructed, is a construct ultimately of your brain. The way you experience the world, if you look at these colors, most people will think, well, there's objective matter of the fact about this. There's an objective blue and this is objective yellow. Yet we know there are lots of illusion. Do you know they're really a powerful illusion, The Dress? Hashtag the dress.

Okay, so this is this wedding dress that went viral in 2015 where roughly half of humanity thinks that, oh yeah, it's white and gold. And the other half sees it's blue and black. And it's not ambiguous. It's not that you have to look and say, is it one or the other? No.

Like I always see it's in white and yellow. completely unambiguous, where other people will see it as blue and black. What's going on there? People say, well, what's the real color? No, there is no color. What there is are photons from the sun that strike this very complicated fabric and then are reflected ultimately into my photoreceptors. And then my brain actively constructs this conscious experience that's attached to surfaces we call color. But it's in construction. And it turns out there are at least two ways you can construct it. And so half of us will see it one way and half of us will see it. One way. us will see other. But as Emmanuel Cantor already pointed out more than 200 years ago,

the same is true for everything, including space and time, the constructs of our mind. And we all know this because we can sit during a talk and we can experience this talk lasting forever, or we can be in a moment where we wanted to last forever, but it's painfully short. And so we know that the duration of a moment can be shorter or longer. On the extreme experience, like a near-death experience, you may totally lose the flow of time altogether, like you experience something that has no duration whatsoever anymore. So all of these things are ultimately construct of the mind. And depending on your genes, in other words, the predisposition of your brain, how you grow up, early childhood experiences, traumas, etc. You will see the world as good,

bad or evil or you're biased towards one group or you're biased against another group, depending on your upbringing, if I tell you October 7th or I tell you January 6th, you will interpret these radically different, right? We see this in politics all the time. This is polarization with the same, even reasonable people that otherwise perfectly reasonable, your friends or relatives and neighbors, suddenly when you come to a political event, they see it totally different. Well, why? Well, because their experience leads them to have a different. different construct. I'm always surprised at this. When you go on a freeway from UCSD, you can go at high speeds parallel in opposite direction next to each other and you wonder, well, am I not glad

that certain aspects of our reality actually shared? Because if it wouldn't be shared, if you deviate by five degrees, you're going to have two collisions at two times 50 miles per hour. So most things, the basic, our sensory perception of the basic of our environment is shared with some weird exception like colors, et cetera. But that's only true in the limit. And most things are ultimately, it's all a construct of your mind. And so that's what the perception box. You grow up in a particular perception box due to your experiences in your genes and your early childhood experience. But you can change it also. We know the therapy can change it. In fact, there are these classes of experience called transformative experience where single experience, good or bad,

can suddenly you reinterpret everything differently, an experience that can love. five minutes, ten minutes, twenty minutes can be spiritual experience, a mystical experience, and your death experience. It can be someone telling you something. Suddenly you say, oh my God, now I get it. And everything flips suddenly, everything, your entire narrative changes. So for example, when someone in you tell tells you, let's see, I've had an affair, then suddenly your life is turned upside down because now you interpret everything in suddenly in different light. So it means that this perception box we live in, we're not doomed to live in this perception work forever. We can actively do things. We're actors in this drama, in our own

narrative, and we can change it. And that's very important to realize, particularly today, where so many people are rather pessimistic about life and its progress. Well, speaking about perception is another word for it as observation. I can't help but think that there are various ways that physicists interpret quantum mechanics might come to bear on the amount of this metric IIT ascribes to consciousness-related structures. Can you talk about that? I mean, would a Copenhagen interpretation differ from an Everettian interpretation? How do the various interpretations influence the score of this metric for IT? Many papers published in IAT assume deterministic classical physics. So you have machines, whether you have gates, logical gates, whether they're neurons and transistors, but they're probabilistic, but they have a definite state. But there are at least two different groups that have now come up with quantum mechanical.

versions of IAT. The research is open as to which one is the best one, the one which most closely holds to what we observe. Most of us came of age during a time when physicalism rules supreme. That says things exist objectively. And the only true existence is this objective existence that you can quantify in terms of numbers, quantities, right? Mass, energy, space, time. Only they have existence, truly exist, everything else emerges, including liberal democracy, emerges out of matter and, of course, consciousness. Now, this is a promise that has never been fulfilled. Philosophy of mind has been utterly inadequate to explain why anything should have feelings, right? And we talked about at the start. Even knowing all the laws as currently conceived of physics,

chemistry, biology, we are at a loss to explain that certain systems like this have this interior aspect. It feels like something to be them. Okay. So that's one. Then be the challenge of to define actually what is a physical, right? And now we know with the violations of bells in a quality, an entangled electron, that there are these weird things that are certainly not my grandfather's materialism anymore. You can take two entangled photons, the half a universe away, you measure one instantaneous the state of the other one is determined. So it turns out, is it maybe true as Visman or other people now assert that there is no objective fact of the matter, that nothing has an existence independent of an observer, that everything always depends who and

how it is observed. Well, if that's true, then physicalism really goes out of the window, right? Because physicalism, this objective world simply doesn't exist. No, it always depends. It's sort of the participatory universe. It always depends on an observer. And maybe this is exactly what the function of consciousness. This is exactly what consciousness does. So maybe once we truly understand what's at the rock bottom of our theory of what truly exist, maybe we come and find suddenly we find consciousness, not at the top where supposedly it emerges after five billion years on a rocky planet, the last two million years we have these creatures that are conscious. No, it may be at the rock bottom of existence itself, where we define what

truly interacts with other things. You talk about extrinsic and intrinsic causal power. First of all, can you define what you mean exactly by intrinsic versus? extrinsic? And then second of all, how could they be experimentally differentiated? How could you differentiate what is intrinsic versus, because it certainly feels like everything I do is intrinsic, right? So what is intrinsic versus extrinsic? And then how could we experimentally differentiate? Okay. Intrinsic is simply within the system. So I have a particular system, right? So I have to define my system, right? What is the system? And it has boundaries because there are things within the system and there's everything outside the system, presumably the rest of the world. Okay.

So intrinsic causal power simply means the ability of a system, let's say, let's just talk about transistor for the sake of argument, right? So I have a chip, okay? I'm looking at the ability of this chip. If these transistors are opened and those are in an off state, what is the next state? How much ability to the system's current state have to determine its future? And likewise, how much is it determined by its immediate past? And you can quantify that.

So, for instance, if this system is incredible noisy, then its current state, If it's in the limit, random, then the current state will not determine anything about the next state, because it's totally subject to random, thermal fluctuation, let's say. Okay, then it's causal power zero. If it's very determined, then it has high causal power. Extrinsic causal power just means acting upon something else. So for instance, gravity curvature in space time has obviously power to make this cup move. Or charge has an influence on other charges, external to.

So that's a difference between intrinsic and extrinsic causal power. And IAT simply says causal power's ability of a system intrinsic to within itself to act upon its own future, to be determined by its paths and to act upon its own future. And IAT also precisely defines what is a system. So this is a question of meteorology that goes back all the way to Aristotle. How do you define a system? So for instance, how do you define me? Does this include me or not? This wedding band, is that included in me or not?

If I take out a filling, is that part of me or not? if I have a brain implant or kidney or whatever, is that part of me or not, right? So how do you define a system? This has always been in a challenge, in particular biology, and IAT has some very, it defines exactly what is the whole.

A whole is the part of the substrate that maximizes its intrinsic causal power. It's a very precise, for any particular system, I can precisely define what is a system. And this solves a fundamental problem that physicists have never even thought about, if there end people in a room, okay, so end people inside a room, people like you and me,

in a room, as we know from experience, there are probably n different conscious minds, right? We have no theory that tells us from first principle how many minds are in that room. Because all physics says, well, there's a bunch of molecules. There's lots of complicated structure within this room. Max Techmark, I mean, has made this point, right? I can, of course, empirically define, well, there's a difference in terms of, of the surface binding energy between my skin or my watch here and other things. But that's a very empirical ad hoc definition.

There's no rigorous definition to say, well, if I have in a room with N people, they're N minds. Well, there may actually end plus one mind because it turns out one of that person is pregnant. Or it turns out one of that person has a split brain. So there are actually two minds inside that one. So we need a theory among others that exactly can tell us that. and IAT can. I wonder if it would be helpful to the audience to go through a simple, maybe heuristic calculation

of how would you actually calculate five? For example, my computer mouse versus real mouse. Is it logarithmic? Is it exponential? Unfortunately, super because you have to take the subset. So if I have a system with N gates, okay, then I have to look at all single neurons, the effect of single gates onto other gates, all combination of. of two pairs, all three triplets.

So I have to look at all possible combination of all possible subsets, which is a very, very large number. So it scales as two to the two to the n, unfortunately. It scales badly. So that's a practical limitation right now. Of course, it's not a conceptual one. It's a practical one. Yeah, so if you have N bid, it's a gauge.

And in principle, you have to consider two to the end, so that's two to the 10, two, to the thousand possible states just for 10. So it quickly grows astronomical large. Photons in the universe, right? Yes, vastly. But think about it, Brian. Our consciousness, I mean, it has to reflect

the richness of consciousness. And again, there's another paradox here. Imagine, oh, you can just close your eyes and you're staring at an empty, well, if I close my eyes, all I see is black, right? And you could say, well, that's a simple. It's simple to describe. It's just black screen, right?

But that's simple to describe for me to you, okay, because we have these labels. This empty black, this empty space, so let's see I'm staring at a blackboard, black, blackboard, nothing on it. But still, it contains trillions of spatial relationship. There's left, there's right, there's up, there's down, there's distances, there's triangular relationship. So all of that encompasses my feeling of space. because what do people mean when they say, well, that's a vast extended space,

whether it's auditory space or felt space or visual space. It's extended. That's precisely what being extended means. And so you have to capture that. So any experience, even what seems like a simple experience of an empty black space, is incredible full with all these relationships that experience that to me as a conscious being immediate, obviously, I don't have to infer them. I don't have to compute them.

They're just there. When we think about doing experiments, it either seems very risky in that you have to experiment on yourself or fraught with ethical considerations if you're going to experiment on animals or children, whatever. And some say animals and children are very much the same types of entities. It seems like we could learn things from the types of studies that my colleague here, Ramashandran, used to do with synesthesia, things like that. Are there elements of the brain? if you were to do a mapping of phi throughout a single individual brain, are there more higher density regions of fineness in a human brain? Or how don't know, breakdown? And could you study brain damage people or enhance people using these two? Yeah. So if you had a qualioscope, so what you

would suspect on principle grounds, that areas that have dense topographic connectivity, where you have lots of very spatial scales, lots of connections locally, but then the density of connectivity falls off with space as compared to a random connectivity. So let's say in the front of cortex, I'm talking about neocortex, the outermost layer of cortex, that's most highly developed in us and in other primates, which is essentially, it's a two, well here, I can show you, it's a tissue close to, it's like this,

it's two to three millimeter, and it's the size of a big pizza, 12 inch, 12 to 14 inch, and you've got two of them, they're highly convolved, one here and one here, okay? And you can conceptually unfold, And in the back, towards the back of the brain, where we have the representation for visual space, for auditory space, for felt space, body images, etc. You have a lot of these topographic connections that fall off with distance, very dense. In the front of the brain that's more closely tied to intelligence, we have more random, seemingly

random connection and less much local. The phi, for mathematical reasons, phi is maximized by these local connections. connectivity and it's much smaller of using this random connectivity. And so there's a big prediction that if you truly care about patients that are conscious, not patients that are intelligent, but patient that are conscious, what really counts is the back part of cortex. And that's also what the clinical, surgical evidence of lesion shows. Well, the front of the brain is really more associated with things having to do with moral reasoning,

with what people conventionally say is intelligence. there's a lot of background noise suddenly. Adversarial collaboration that was started, oh, just in 2019, I think it's finally going to make its way into nature this year where a group of 14, a consortium of 14 labs tested the differences in prediction between integrated information theory in the one hand and global neuronal workspace on the other. Because the global neural workspace theory of conscience predicts is really the front of the brain,

that's critical for consciousness. While IAT predicts it's the back of the brain, the back of cortex, I'm talking about not the brain in general, but the front of cortex or the back of cortex. And it looks like it's going to come out in a couple of months. And then on the opposite side of the complexity scale in terms of brains or Boltzmann brains, how would they rank in terms of IIT scores? I think that's as interesting as talking about, well, maybe we're all a gigantic computer game being simulated in the next universe up.

Yeah, fine. I mean, you can endlessly speculate, and I know people love it. I mean, in the Middle Ages, they speculated how many pin, literally, how many angels can fit on a pin of a needle, right? Yeah, and you can speculate that and it's fun, but I don't think it pertains to anything in the real world that I'm concerned about. I detected a kind of not so positive opinion about computation, the functionalism in your remarks that you made earlier, that you're not so fond of the idea that computer could

effectively replicate, functionally replicate a brain. A digital computer. What about a quantum computer then? There's nothing supernatural about the brain, right? The brain's a piece of furniture like anything else in the universe, but it has these very special, very high-dense connectivity. It's by far the most complex piece of active matter in the known universe. And you cannot simulate consciousness.

We know this. You can write down Einstein's field equation for the black hole, right? Sagittarius A-star at the center of the galaxy. but you don't have to be afraid that when you turn on the simulation, you'll be sucked into the simulation, right? People say, well, that's ridiculous. It's a simulation. Just like it never gets wet inside a simulation of a rainstorm.

Well, same thing with consciousness. Consciousness is causal power, and you can simulate the behavior associated with consciousness, right? We talked about LLMs and all of that, but it doesn't mean the thing will be conscious. That's true for digital machine that have this very different connectivity where typically one transistor.

If you look actually where the rubber meets the road, right, where actually the computation happens. The shuttling of electric charge or electrons going from one gate of a transistor to the next one, you typically have one transistor that's connected to three or four other transistors. And then that's replicated trillion times, right, on an Nvidia chip, etc. Brains is radical different. You have one neuron that gets input from 10 or 50 or 100,000 other neurons and it projects to 10 or 50, 100,000 other neurons with massive overlap. Very different than digital machines.

Now, in quantum computers, if they're all entangled, no one has ever done the calculation so far, but sooner or later someone will, I suspect because of this massive, if it's truly, if it's really true that all these n bits are entangled, I think effectively that corresponds to something much closer to the brain. So I, so my personal feeling is that it may well be that quantum machines, if you build them big and complex enough, will at least in principle be able to feel like something. How could we ever resolve, just as again as we wrap up? I mean, how would we know who's right, the computational functionalists or the IAT? It's ultimately, we need to explain our own, because the only where we can all agree, even with animals, some people disagree, we can all agree that you and I are conscious, right? So ultimately, any theory and its prediction will have to be tested in humans and verified in humans, and one, we have a theory that accounts not for the behavior. but actually for conscious experience itself, not the fact that I can say, for example,

I can talk when I'm conscious, et cetera, because those are functions associated with conscience, and that's not really conscious itself. Then I can just like with Einstein, I can then go and inferred in very strange places, animals and fetuses, maybe even in cerebral organoids, right, or in machines. But ultimately, so this entire debate right now that's going on about machine consciousness really doesn't tell us a lot, because there's. that's always based on our intuition and our intuition is we grew up in a world. We're primarily talking apes, speaking apes, surrounded by other speaking apes.

So all of our intuition is based on language. But as I said before, that leads to deep fake to machines that pretend to be conscious but don't feel like anything. But are very good at aping us. Okay, I have a couple of questions as we finish up for my audience. Robinson 8491.91 asked, does time move one second per second? or is it our brain that makes it so? Will time move faster for a housefly and slower for a tree in their experience?

Very likely, because we also know that in our own experience, as we talk about, perception of time can be changed, right? When you're highly aroused, let's say when someone points a gun at you, people report this is a very well-known phenomenon, or when you, for the first time, jump out of an airplane, right, to do a parachute, time seems to slow down. It's incredible slow.

While if you're bored, it also changes. So we know even from our own experience that time, the perception of time is subjective like anything else. And so it's likely going to be also in other creatures, including animals and possibly in trees as well. Kadurim do ask, where does consciousness take place? What part of the brain is it? Is it the cells themselves or the structures, the cortexes, etc.?

It's the interaction of cells, this vast complex interaction of cells. So you have to ask at what time scale and what's the actual substrates? Because ultimately you can say, well, whatever, we know, the final substrates are elementary particles or brains or strings or whatever it is. But that's not really where conscience operates. Conscious operates at a particular scale that maximizes causal power. And right now it looks like that's on the order, probably neurons and assemblies of neurons, but probably not inside neurons. But that's ultimately an empirical question that we need to resolve empirically. But my feeling is it's large ensembles of millions of neurons.

Another question, I can't read their name. It's a string of unpronounceable syllables. What's the second most highest IIT score, a five score in the animal kingdom? Is it an octopus? Is it a dolphin? What would you say is ranks next to us in the natural world? It may be that certain dolphins or certain whales in fact have a conscious, I mean, why should we be at top? We like to be on top. We think we're the best, right? We're certainly the dogs. dominant species, but that's because of our intelligence and our hyperaggressiveness. It may well be, particularly if you look at cortex in animals like whales and dolphins, all of which are mammals, of course, they have cortex that can be twice as big as our cortex.

And it's not apparent why. And it may well be that some of these creatures or also elephants may experience the world in more conscious ways, yet are not as intelligent as we are, because intelligence and conscience is not one and the same. And while they co-vary, they may be low in sum while consciousness is higher in them. The way you experience the world is not necessarily the same as the way you can manipulate and plan and act in the world, which is what intelligence is. It's about doing while consciousness is about being.

Information loss in the context of black hole information theory. So would a black hole in some sense be a consciousness eraser? Could there be implications from the combined effect in information destruction for those that do believe that information can be destroyed in a black hole, as Hawking did at one point at least? What are your thoughts about the interplay between the structures, macroscopic structures in GR, and conscious entities? I mean, could a Boltzmann brain next to a black hole or really spelled trouble for consciousness five score? I don't know. Given the existence of the no hair theorem, once you fall into that blackthole, there isn't anything. even more substrate, right? It's all one mush. It's all one. So there's no more integration

because it's all one thing. It's all one state, essentially, at least if the no hair theorem is true. So I don't know. I don't think it's terrible, worthwhile to speculate about it. I'm still surprised that you didn't mention the Bernese Mountain Dog is the most impressive conscious phi score element entity besides human beings. Christoph, thank you so much for spending so much of your time with me. It's been a great and fascinating journey and I can't wait to. What are you working on next? Are you have another book in the works? It's been about a few years since this one came out. Yeah. Well, this one actually came out in May of last year. Yeah, I'm thinking about writing at this intersection of quantum mechanics, this crisis in physicalism, and then these mystical experience,

how can these be reconciled that many people talk about, speak about? Turns out a significant, including Schrodinger, a significant number of people have these experiences. and don't know what to make of them, so they just put them elsewhere, they park them there. But ultimately, if we want to arrive at a complete picture of reality, we have to integrate them. And then you take what's currently happening in foundational physics. It's really true that there are no objective states. Everything depends on being observed or how it's being observed. Well, then maybe we can try to reconcile those.

Very good. Well, that would be interesting. Can't wait until that comes out and I'll have to have you back on the show. Thank you, Christoph. Thank you so much for spending your valuable time with us today. Enjoy the snowstorm, the rare snowstorm. Thank you. Thank you. That was very enjoyable. It may be good.