309 | Christof Koch on Consciousness and Integrated Information

0:00:00.5 Sean Carroll: Hello, everyone. Welcome to the Mindscape Podcast. I'm your host, Sean Carroll. Despite multiple episodes here on Mindscape devoted to the topic, we have still not figured exactly out what consciousness is or how it works. We've had philosophers on talking about what it means to have an experience. We've had neuroscientists talking about what happens in the brain, all to no avail. It's not that there's not been progress. I'm slightly tongue-in-cheek here, but this is a hard problem. What exactly is consciousness? Because it's one of those things we're very, very familiar with, but it's kind of unique in that we're most familiar with it from the inside, right? Rather than at an objective stance, measuring things with calipers or microscopes or whatever, we're experiencing it all the time. That's an opportunity, but it also creates certain unique problems.

0:00:49.6 SC: So today we're talking to someone who is a neuroscientist, but is also very, very willing to mix it up with the big philosophical questions. Christof Koch is currently at the Allen Institute for Brain Science, where he was originally the chief scientist and president of the institute. Before that, he was a professor at Caltech for quite a while. In fact, I met Christof when, believe it or not, we were both consulting on a movie [chuckle] and the idea of consciousness came up, and the idea of consciousness in a computer came up, and Christof pulls out his iPhone and points to it and says, how do you know this isn't conscious?

0:01:27.8 SC: At the time, I did not take that seriously, but now I understand the issues a lot better. And I know that he was gesturing toward the idea of panpsychism, which we've talked about before, of course, on the podcast. And Christof is not quite a panpsychist, but he's panpsychist adjacent as a proponent of the IIT, or Integrated Information Theory, approach to consciousness. It's a very down to earth theory in the sense that there are numbers and equations and you try to measure the connectivity and the integration of the information in some system. The system might be a human brain, or it might be an AI system, or it might be a pile of rocks on the beach. You can measure its integrated information, but the point is you always get some number.

0:02:11.6 SC: And Christof's point was, well, that number is not going to be zero for most systems. It might be very, very tiny, nothing like what we consider to be consciousness, but maybe you can talk about it as a wee tiny bit of consciousness without actually saying that electrons have feelings. So this is years later and a lot has happened. And we're going to talk about a number of things that Christof has done. He's been very involved in figuring out ever since his early work with Francis Crick what are called the neural correlates of consciousness. So forgetting about what consciousness is, what is it that happens in the brain at the level of neurons when you're having conscious experiences, doing very down to earth experiments in the brains of mice and seeing how maybe that can be extended to human experiences and then developing tools to understand, Is someone in a coma conscious? Is an artificial intelligence, large language model conscious? And so forth.

0:03:08.0 SC: His most recent book has the interesting title "Then I Am Myself the World: What Consciousness Is and How to Expand It." So he's gotten into the questions of, again, from the personal first person perspective, Are there things we can do through meditation or other forms of enhanced awareness to improve our own personal level of consciousness? So Christof is happy to be provocative about many things. We cover a lot of ground in a short conversation and I think you're going to find this one amusing. So let's go.

[music]

0:03:58.7 SC: Christof Koch, welcome to the Mindscape podcast.

0:04:00.5 Christof Koch: Thank you very much, Sean, for having me on Mindscape.

0:04:03.4 SC: You've been doing this for a while, going back to your collaboration with Francis Crick after he moved from molecular biology to thinking about consciousness. I wanted to start by just asking what is the major way, if there are any, that your views on what consciousness is and how it works have changed since you started thinking about this.

0:04:24.9 CK: So when Francis and I started, back when I was a young assistant professor at Caltech, partially overlapping with you, consciousness was still a complete no-no among serious working day scientists, particularly if, like me, you hadn't obtained the holy state of tenure yet. Right. And so it was something that you did after hours and where grad students who are always in the know, they routinely roll their eyes when, Oh yeah, that topic now. Right. And then what we instituted this purely empirical, pragmatic program. Okay, let's forget about the philosophy because we haven't made much progress in the last two millennia there. Let's focus on what we can do.

0:05:07.2 CK: So we talked about certain model systems in monkeys, in people, of course, in mice, how to go about it, which techniques to use, which visual illusions to use, and what are some of the basic question. And that really caught on this purely empirical program and led to what's now known, the search for the NCC or the neuronal or the neural correlates of consciousness. In other words, what are the footprints in the brain for any one conscious experience, like right now, I see you and I hear you in my head, this voice inside my head. Well, nothing in physics right now tells us why there should be voices or sights or smells or feeling of being angry, upset, in love, right? Nothing in quantum mechanics, nothing in general relativity, nothing in periodic table of chemistry, nothing in the endless ATGC chatter tells me that we have these things, these voices, these experiences.

0:06:00.9 CK: And so the search for the NCC is neutral with respect to your particular philosophical, metaphysical underpinning as long as you believe, and not even all philosophers believe, that consciousness is something that needs to be explained. Some philosophers, most famously Dan Dennett and the Churchlands, they claim it's all a big confusion. It doesn't really exist in any sort of way that you and I believe it exists. And it's just a big distraction. But among the vast majority of people, and even the majority of philosophers, it cries out for an explanation. Looking for the footprint is an empirical program.

0:06:38.2 CK: However, and this is also what has changed since I worked with Francis Crick for 14 years. Let's say in the fullness of time, we will have an explanation about the neuronal footprint. So anytime, Sean, that you feel bored, the feeling, the experience of being bored, it feels like something to be bored, right? Very different from being in love. We know, ultimately we will know it involves these neurons or maybe that mechanism, or maybe even the collapse of the wave function. Okay, whatever it is. But then of course, you want to know, well, why this mechanism and not that? Why these neurons here, not those neurons? If it's a collapse, why should the collapse feel like anything? Right?

0:07:20.8 CK: So this is of course, also known as a hard problem by David Chalmers. So ultimately, we need a theory of consciousness that precisely gets at the question you asked earlier on. You warned me, you're going to ask, is a fetus conscious? And if so, when? Is a machine ever conscious? Is a particle ever conscious? Is a parametrium ever conscious? So for all of that, you need a theory of consciousness. You need an empirical, verifiable, scientific, objective theory that tells us which bits and pieces of matter, which bits and pieces of organs like this, the most complex piece of matter in the known universe, have feelings.

0:07:58.0 CK: And Francis and I, we purposely didn't go there. But that, of course, was 30 years ago. And so in the meantime, we have theories, in particular, one of the two most popular theories, integrated information theory of consciousness, that Francis would have liked because it's causal and in fact he met the originator, Giulio Tononi. We had lunch. He came over twice to have lunch with Francis and myself at his house. Yeah. So that's the biggest difference.

0:08:31.5 SC: So we want...

0:08:32.4 CK: And perhaps the metaphysical assumptions, but we can leave that for later.

0:08:36.2 SC: We might get there. So in other words, you from the start were looking for the things that happen in the brain that correlate with the conscious experience. Now you're more taken with the idea that that's important but also we need a theory so that we can ask about things we can't immediately quiz, like the fetus or the AI.

0:08:58.7 CK: Yeah, exactly. Where we right now completely depend on our intuitions, our religious or philosophical biases. Right. For some people, of course, the fetus is conscious. Right. Way back to inception. Other people say, no, it's only born. Only when it's born to. Is a dog conscious? I know you're a cat lover. Are cat conscious? How far does consciousness go down the tree of life, all of that. Right now we have intuition. Everyone has a different intuition, but that's not good enough. Right? We all have the intuition that whales are fishes. Well, it turns out actually they're not fish, they're mammals.

0:09:37.1 SC: Well, you mentioned the idea that you're talking to me, you're seeing a little picture of me, you're hearing what I say. What is the distinction between consciousness, which seems so laden and philosophical with simple awareness of things? Are they the same thing?

0:09:53.8 CK: Yeah. So for us, so Francis and I wrote about it explicitly. For us, they're just different words. Some people, particularly early on. Now you can use consciousness even in nature and in PIL. You couldn't do that 30 years ago, unless maybe you're Roger Penrose. So people use this softer thing, awareness. Awareness is just a more generalized sort of, you're aware that you're alive, you're aware that you're awake, you're aware of that there's a president here. It's more background consciousness, but it's all about... Ultimately, you can also call it mental states, subjective states, phenomenal. Philosophers like to use the word phenomenal. It's all the same. It feels like something to be alive, to be awake, to be bored, to be hungry.

0:10:43.0 SC: And how much progress have we made on the literal neural correlates? Do we know which little neurons are firing when I'm hungry, when I'm feeling the feeling of being hungry?

0:10:54.0 CK: Well, even if we knew, of course, we know a lot about neurons and we know a lot about brain areas, but those are, of course, all correlation. Yeah, so we know. We know. I mean, there are hundreds of experiments. Right now I see you and you see me. So we have a pretty good understanding what are the neural pathways involved in vision. Vision, because it's the best studied modality of all, but of course that's different from... That's just correlation. Which of those bits and pieces are actually necessary?

0:11:21.5 CK: So for instance, the eye. Clearly photons are entering my eye and then are transformed ultimately into vision. Do I need my eyes? No. I can close my eyes and I can see you still ghostly. And of course tonight, if I dream about you, I dream in the dark and my eyes are closed, so I don't need the eyes. So the challenge has been to sort of disambiguate mere correlation from what is actually the causal agent. And there it's still very controversial.

0:11:52.0 CK: You may have heard I had this famous bet with the philosopher Dave Chalmers 27 years ago now, where after a late night pop call in Bremen at a conference on consciousness, he challenged me and says, Look, even in a quarter of a century, which, I was, we were all much younger, seemed forever, we won't find these. It's going to take us much longer. He didn't say it's impossible. He agrees the NCC is a purely empirical, operational defined project. With enough people, money, funding, research, we will figure it out.

0:12:26.6 CK: And so we had this meeting two years ago where it turned out as part of this adversarial collaboration. So the Templeton foundation started this cool project where they put, where they pitted the two dominant theories of consciousness, integrated information theory and global neuronal workspace theory, directly against each other, were a little bit like the famous experiments 1919, predicting the solar eclipse. During the solar eclipse, whether the shift of light follows Newtonian physics or Einstein general relativity, we directly pit the two theories against each other, trying to resolve empirically, well, what are the empirical predictions? Which one is actually true? And so at this meeting in New York, in May of 2023, the agreement was, well, none of the theories are totally correct.

0:13:24.2 CK: At least the empirical manifestation of these theories in the brain with respect to where is the neural correlate. So IIT says neural correlate is in the back, the sensory cortices, while global neuronal workspace make the argument, it's primarily in the prefrontal cortex, a part of the brain that's most expanded in humans compared to other mammals. And the field still hasn't settled down on that. That paper is finally coming out in Nature in a couple of months from now.

0:13:53.2 SC: Okay.

0:13:54.1 CK: Because there was a lot of controversy about it. It turns out, big surprise here, Sean. It turns out people who work for 20 years on a theory, then if you go into this adversarial collaboration and the theory or some of its prediction is disproven, they don't just say, ah, you got me, my theory is wrong. I admit, 25 years of my life was wasted. Sorry about that.

0:14:17.0 SC: I don't believe it.

0:14:18.3 CK: Yes, people actually don't do that. Surprise, surprise. Newton wasn't around in 1917, 1919. Otherwise he would have objected again. Right. So we still haven't converged, but we're getting closer. And what we do realize and what came out in this debate, that the prefrontal cortex, and this is relevant to your other question, are computers ever going to be conscious? Is much more closely associated with doing, with planning, with intelligence. This is the difference between consciousness and intelligence and also artificial consciousness and artificial intelligence.

0:15:00.5 CK: Consciousness is ultimately about being, being in love, being happy, being angry, seeing things, hearing things, et cetera. While intelligence is really about short-term, medium-term, long-term planning to act. Right. I have to save money now in order to collect it or retirement. That's an intelligent action. So ultimately that's about doing stuff. It's about functions, et cetera. And that's different from being. And so therefore it's not that surprising that the parts and bits of the brain that are involved in planning and reasoning, in thinking, in moral reasoning, in decision making, are different from the bits and pieces of the brain that seem to be the substrate for feeling, for having experiences.

0:15:47.7 SC: Well, I love the idea of the adversarial collaboration. I think it should be more widely adopted, although some adversaries would just not be able to collaborate, in my experience. So congratulations to you folks for being able to do that. But now I need to ask this question, even if it takes a long time to answer. What are these two theories? What is integrated information theory? Global workspace theory? I had never heard the idea that they focus or they propose different parts of the brain as being central for consciousness.

0:16:18.5 CK: Well, I mean, that was a good thing that came out of this, in this meeting, out of this meeting. Well, let's step back. Most people, in fact, I just attended just before a meeting at Google, I mean virtual, at DeepMind in London about what's called computational functionalism and consciousness. So most people who study consciousness, particularly everyone in AI and big tech, makes this philosophical, metaphysical assumption that was proposed formally by Putnam, the American philosopher called Turing functionalism or computational functionalism. Consciousness has one or more functions, let's call it planning, or summarizing the current state of the world in my mind and acting on it, long term memory, broadcasting information in the brain, what probably a list of functions. And any system that implements these functions will therefore ipso be conscious, including a Turing machine.

0:17:18.1 CK: So if we can get a Turing machine to mimic everything that people do that seems to involve consciousness, like talking about consciousness and reasoning based on it, then machines like large language models will also be conscious. And so now it's just a practical question we can discuss. Is ChatGPT 4.0? Is O3? Is DeepSeek? Is Gemini? Are they conscious or are they still missing? Well, maybe they're still missing a little bit of something. You know, maybe they have to take in more tokens or to do a little bit longer memory or whatever. But it's really just a question around pragmatics. So that's computational functionalism.

0:18:00.4 CK: So ultimately consciousness is a computation, it's a hack. If I program machine, it's substrate independent, it can run in the cloud, it can run on a quantum computer, whatever. It's a clever hack. And once you instantiate this hack, of course these machines are conscious. You can ask them and they talk about it at great length, how conscious they are, how they feel. Right. Versus...

0:18:22.3 SC: Sorry, that's according to that view, the Putnam view.

0:18:26.2 CK: The Putnam view, which is very, very prevalent. And of course it is a dominant one in the industry. And I get it's so dominant because if you interact with an LLM, they seem incredible, not only intelligent, but also, yeah, they talk about feelings.

0:18:42.6 SC: Aware. Yeah.

0:18:43.8 CK: Now of course they talk about feelings because they've been fed on every novel humans have ever written. And all novels are about the feelings of the protagonist. Right. Their love and hate and being slighted and being insulted and hurting and et cetera, et cetera. And they've ingested all of that like vampires. So of course they can reproduce it. It's all deep fake. That's what I believe.

0:19:05.2 CK: Okay, so the other theory on the other side, there's really only one dominant theory, which is integrated information theory, which says no, consciousness is not a clever hack, it's not a computation, it's not a process, it's not a function, although it may be associated with all those things. Ultimately consciousness is, as I said, a state of being. It's a structure in a very high-dimensional causal space. So ultimately consciousness is about a system, a substrate like this brain having causal power upon itself. And this is the particular way the system acts upon itself. In other words, can determine its own future and be determined by its past. I can unpack that. The more this system feels like something.

0:19:51.6 CK: So it's a little bit like saying complex systems that can act upon themselves, in other words, that are not just randomly determined where the next state is a function of the previous state, any such system, if you look at the mathematical formulation in terms of its transition probability matrix, has what's called causal power. The current state co-determines the next state. There may be some noise, so it may not be fully deterministic, but determines the next state, or the past determines the present and the present determines the future. The more internal causal power the system has to determine its own fate, the more it is conscious.

0:20:34.2 CK: And the theory describes two aspects of it. 8. [0:20:37.1] ____ It has a number called phi, the Greek number phi, which you can also think about the irreducibility of the system. The more the system is above and beyond the sum of its part, the higher the phi, the more it is conscious. So you can have a system that's phi zero, which means it's totally... It does not exist as a whole. Its irreducibility is zero. It's not really a system. It's more two independent system. That system wouldn't be conscious. The higher the phi, the more conscious the system is.

0:21:16.3 CK: And most importantly, and this is never stressed by other theories, you also have to have a theory that tells you, why does being in love feel the particular way it does? Why does time flow? We all experience time as flowing forward. It never moves backward. Why? Space. Whether it's visual space or seen space or heard space, feels extended. Well, what is it about the underlying substrate that gives rise to the feeling of extendedness or boldness [0:21:47.7] ____ or being bored or being angry? So the theory also explains by the structure of this unfolded causal space why one experience feels the way it is, very different from any other experience.

0:22:02.0 SC: And so is the global workspace theory the sort of descendant of the functionalism of Putnam?

0:22:08.3 CK: Yeah, okay, sorry, you asked me that. Yeah. So global workspace is a computer science metaphor from the '50s, right. The idea is you have a central process and that writes information on a blackboard. It's also called the blackboard architecture. And then every local processes can access this blackboard. So the idea here is my brain does many things, but unconsciously, like you and I, like all humans, we move our eyes four times a second. Okay. The regulation of eye movements is very sophisticated, highly controlled, but it totally bypasses our consciousness. Just like the way I adjust my gait, I move my hands, I sort of infer syntax, we speak incorrect sentences, but we don't constantly put those things together. It just happens. Those are all local processes.

0:23:03.0 CK: Every information that's conscious is globally accessible. What's meant, and I think it's a correct insight from psychologists, once I'm conscious of something, oh, there is something funny over there, then all the different processes in my brain are informed about it. I can put it into long-term memory. I of course have it into short-term memory. I can use that information I just became conscious of to do long-term planning. I can think about it, I can reason about it. So that's the idea that once information becomes global, it's accessible to all the information processing system in the brain. And that's what consciousness is.

0:23:43.5 CK: So the metaphor is really you broadcast information and [0:23:52.2] ____ and his collaborators have identified a set of neurons that are situated in the prefrontal cortex that project the information back to the rest of the brain. So that is this broadcast. So the claim is every time information percolates from, let's say, from the visual system that starts at the back of your brain, propagates forward, is still local, is still local, then reaches the front of the brain and then that information, if it's important enough and we attend to it becomes, it's broadcast to all the rest of cortex, to the neocortex, the outermost layer of the brain. That act of broadcasting, that function is what constitutes consciousness. So any system that instantiates that broadcast will be conscious.

0:24:42.4 SC: So if I understand it correctly, then the global workspace theory seems to focus on the spotlight of attention that consciousness involves, whereas integrated information theory, IIT is more about this sort of, well, the integrated information that we use to plan and control our own destinies.

0:25:01.5 CK: Yeah, but it's not just destiny at a high level. At a high level, it's just looking out in... I mean, causal power is just the ability of, let's say, some elements, whether neurons or transistors or whatever to influence... Electric charge has extrinsic causal power because a positive charge here has the ability to influence the negative charge over here. Gravity has causal power due to curvature in space time. So those are example of extrinsic causal power. Intrinsic causal power is just, you have a little cellular network of neurons or cells or transistors, whatever. And if the neural or these elements are in this state, then the next state will be this. And if they're in that state, then the next state will be something else.

0:25:48.4 CK: So causal power is something you can actually evaluate. If you have a complete mathematical description of a system, how its states evolve over time and how the individual elements influence each other, you can fully compute, explicitly compute all of its causal powers. And it's not an airy-fairy, it's an operational. So in that sense it's very operational. It doesn't appeal to anything mysterious. So it's not dualist. Some people say IIT is dualist. Well, it's not because it doesn't appeal to any substance. It's just the claim that fundamentally what experience feels like is this unfolded causal power. So any system that is integrated, that's more than the sum of the individual parts, maybe even a very simple system like a parametrium will, because of its vast complexity of the... In the parametrium, there are probably a billion different molecules of maybe a thousand or two thousand different types, proteins, et cetera. They all interact causally with each other. And this causal interaction feels itsy bits like something like being alive maybe.

0:26:56.2 SC: So this is why you are not running in fear from the idea of panpsychism. Even though you're not like a full panpsychist, but you kind of get why that would be a way of thinking about things. Consciousness is not something that pops up at a threshold. It's like there's little bits of it everywhere.

0:27:16.6 CK: That's correct. I mean that's entirely correct. It's also parameter free. Right? You don't have to say, well if your phi is 42, then you're conscious, but if below 42, sorry, you just didn't make it, because that seems totally arbitrary. I mean, why 42 and particularly if you're... So, you know, I'm also a biologist, so we study mouse brains that are a tiny fraction the size of human brain, in fact almost exactly a thousand times. Yet if you look at the hardware, it's almost impossible. If I focus in with a microscope on a tiny quinoa-size grain of mouse cortex, it's almost impossible to tell it apart from the human brain. And so there's no doubt you can go even down to smaller animals, that if you just observe animals, they have complex behavior, they have memory. That is quite likely that all animals, certainly animalia, may well feel like something. And this is of course this very ancient intuition of that, that's called panpsychism.

0:28:22.5 SC: I should phrase this as a question. Would you classify yourself as a physicalist? Do you think that it's just physical stuff? There's no spooky essences involved here.

0:28:34.5 CK: Well, so something that's always bothered me about physicalism from way, way back is the mental is clearly not physical. My state of seeing, of being or thinking or whatever is not physical. By definition, it's phenomenal. In fact, Schrödinger. So I was recently rereading Erwin Schrödinger. Right. And he makes this very explicit that even as a physicist, I have to look, I have to experience the trace of an, on an oscilloscope, what means I have to see. That's a conscious experience. I have to hear other people saying something when like Einstein, I'm imagining some, the elevator and free fall, et cetera. Well, I use my imagination. That's consciousness.

0:29:26.1 CK: So we cannot escape everything we have... The only thing, Sean, we have access to, the only thing is our experiences of the world. I don't have access to atoms. I never had. No one has. Yeah, you can show me little pictures, you can show me equations that are compatible with atomism and all of that, but you don't have direct access. All you have access to is the phenomena. All these beautiful shapes painted on the inside of your cave that you call your reality. That's your bespoke reality. Right. That's the only thing everyone knows of. That's the only thing, philosopher have the phrase you are directly acquainted with.

0:30:06.1 CK: So physicalism is an additional assumption, says, well, there's not only these phenomena, but then I infer there's something above and beyond. So now inverting that and saying, well, now I'm going to do away with this observer altogether. So there are no phenomena, everything is physical has always struck me as ludicrous. But until recently, you couldn't say that in polite academic society if you wanted to remain a scientist. Okay. You can't say that because then people say, oh, now we're talking about woo-woo. Right.

0:30:38.4 SC: Yeah, I would say that. But I'm glad you're liberated. I'm glad that you feel the courage to admit these inclinations.

0:30:44.3 CK: Yes, in fact, I have. After a particular type of experience I had, I became reacquainted. So I grew up, obviously, in a Germanic household. And so I grew up around not only Richard Wagner, but also Arthur Schopenhauer, who is probably one of the best known, at least modern, Western idealist. And now there's this resurgent in philosophy, not only in panpsychism, it's becoming more popular now but idealism, good old idealism. And so I'm thinking of this computer scientist. I think I would warmly recommend you interview him. Very smart guy, computer scientist and philosopher. He has two PhDs. Bernardo Kastrup, who has this, he calls it analytic idealism, which is really sort of Schopenhauer for the modern world, where essentially he says, well, the simplest explanation, we can all agree, following Occam's razor and all of that, the simplest explanation if we only postulate one thing.

0:31:43.7 CK: Well, the only thing I have directly acquainted with is phenomenal. It's phenomenal all the way to the horizon because that's the only thing I and you ever know. So then everything is phenomenal. And in fact the physical has to be explained within is ultimately arising from the phenomenal. That even the physical stuff like atoms and charges and brain and elementary particles and quarks and whatever you have ultimately has to be explained with respect to the phenomena. And then this philosophy now meets foundational physics where ever since Bell's inequality and the entangled particles and Einstein's spooky action at a distance, we now know there may be no truly observer-independent events. And then maybe the observer has to be right there, central part of physics, including maybe saying, well, that's really fundamentally what truly exists.

0:32:43.3 CK: Because here's one point, Sean, which is really central to IIT. IIT distinguish between absolute existence and relative existence. Absolute existence is existence for itself. You exist for yourself. Tonight, when you're going to go to sleep three hours earlier than I am, you will, particularly in the early phase of your night, of your sleep, you'll go into deep sleep. So if I awake you during your deep sleep and I ask you, Sean, Sean, wake up. Where did you just come from? You would say, typically you would say from nowhere because you were deep sleep. These delta waves are crisscrossing your brain. You were not there. You were not there to yourself. Your partner. Your bed partner can definitely see your body is there. So you are there for others, but not for yourself.

0:33:36.7 CK: Then later on, as you continue to, as your body continues to sleep, you'll wake up inside your sleeping body and suddenly you are something. You are flying. You're meeting long lost, dead friends, lovers, pets, whatever. You're dreaming. It's a conscious state. And so this cup only exists for others. This cup does not exist for itself. Only conscious creatures, only conscious system exists for themselves. So that's really the fundamental gap that runs straight through everything. Are you on this? Do you exist for yourself or do you only exist for others? You can see that is the most critical distinction there is. And so that really puts the onerous on the phenomenal, on experience.

0:34:22.8 SC: This is great. It's wild stuff. I'm very glad you're saying it. I could not possibly disagree more with everything that you're saying, but I love it.

0:34:30.8 CK: Wait, wait, wait, wait. Okay. Why? What do you disagree?

0:34:36.8 SC: I'm a physicalist through and through. I think that quantum mechanics doesn't say what you say it says. I think that as scientists we take in data from the world, from the phenomenal experiences that we have. That's absolutely true. Completely agree with everything you say that of course we, by construction we get what we get as our phenomenal selves by construction. But then we use that to make a hypothesis about how the world works. And the best hypothesis we have is the true existence of a physical world of which we are emergent parts.

0:35:08.9 CK: Okay, so I like, I have no trouble with the hypothesis. That's of course what I do every day in my daily work, studying brains. But I guess we can differ to what extent quantum mechanics now reintroduces the notion often of or is thinking of getting rid... There was this paper in Nature Physics, a year ago by Howard Wiseman and people from Australia, right, questioning the absolute, the absoluteness of events that every, every advent that facts, there are no observer-independent facts. Now I agree that hasn't been resolved. It's ongoing. But that's always been a challenge for quantum mechanics to see what is a fact and what facts do not depend on an observer.

0:36:03.8 SC: Well, I think that this is one of the difficulties in understanding quantum mechanics because it's the one theory that we have in physics, putting aside consciousness and psychology, et cetera. But in physics, quantum mechanics is the one theory that introduces the notion of an observer when it gives you the rules of the game. Right. I just wrote my own paper in Nature that appeared a week ago, sort of rehearsing the history of this. I'll send it to you.

0:36:30.1 CK: Oh, yes, please.

0:36:31.3 SC: But of course, there are since the 1950s, straightforwardly physicalist models that explain why observations in quantum mechanics seem to be special without making observers a part of the fundamental ontology. And then it's an empirical question if and which of these models are actually correct.

0:36:55.0 CK: I agree. Ultimately it's an empirical... Well, the metaphysics, of course, for better or worse, is immune from verification or falsification. Right. The metaphysical interpretation. Yes. Ultimately it's an empirical question. What is the best theory within? Is it a psionic theory, a Bayesian theory, or many worlds theory, et cetera, within quantum mechanics? Yes, I totally agree.

0:37:24.6 SC: Yeah, I mean, my inclination. And I'll try not to go on too long because this will. Because there's so much good stuff I want to talk about. So otherwise we should have a whole separate conversation. But my impression is that assuming physicalism and deriving mentality from it is enormously more productive scientifically than biting the bullet of idealism and trying to derive the physical world from that.

0:37:50.4 CK: I grant you that, but that's a purely pragmatic consideration. And I want to know before I die. I want to know what is the true state of the universe, not which is the most pragmatic state to hold. Right.

0:38:04.8 SC: Well, that's an interesting distinction right there that we could talk about. But good. Anyway, I completely am on board with the idea that idealism in its modern guise is an intellectually respectable way to go, whether it's epistemic approaches to quantum mechanics or idealism or panpsychism, et cetera. I'm just not going there. That's why I'm very glad to have such an articulate spokesperson for the idea. But without getting too distracted then, you can probably anticipate I'm going to ask a question about former Mindscape guest Scott Aaronson's famous objection to integrated information theory.

0:38:42.4 SC: I mean, the great thing about IIT, everyone agrees is that it sticks its neck out, right? It has a quantitative thing and says this. This is the thing we're going to compute, this number phi, it's going to help tell us where the consciousness is. And Scott's objection is that he can come up with examples of things that on face value don't look conscious at all, like a large rectangular array of logic gates, which have a huge value for this number you're computing, but don't seem to map on to our intuitive view of consciousness. So what is your response to that?

0:39:17.2 CK: Yeah, so he's correct. IIT, like any good theory, if you extrapolate, it makes some rather strange predictions. In fact, I would argue this part of the brain, the posterior part of the neocortex that empirically seems to be like the substrate of consciousness in mammals, including humans, is like a very large array of neurons. And you can approximate neurons as a particular type of logical array that have a topographic connectivity, so very, not unrelated to the type of arrays that he postulated. Yeah. So I think he is right. There's some very non-intuitive predictions. And so now we have to do the science. This has always been the case.

0:40:03.9 CK: People way back argued, at the time of the Greeks and of course in the Renaissance that the Earth cannot be a sphere because people would fall off. Of course they fall off on the other side. That's the most ridiculous things. You can't be on the backside of a sphere without falling off. Right. So we have to be careful with things where we think, well that's so ridiculous. Sometimes we have to take things seriously like black holes. Right. That were also laughed. People laughed about them. And then actually try to do an experiment to see is it true. So for instance, Sean, people can now start to build cerebral organoids. You've heard about mini brains in a dish, right?

0:40:46.4 SC: Yeah. Yep.

0:40:47.5 CK: So in principle, in 10, 20 or 30 years, we'll be able to build them like very large tissue. So if I were to take your mind neocortex, the outermost layer of the brain that's really most closely associated with vision and consciousness intelligence in us, it's really a pizza-like tissue, it's two to three millimeter thick like a pizza, and it's 12 to 14 inch across and it's highly folded and you have two of them, left and right brain. So in the fullness of time, in 20, 30 years, people will be able to grow these things in a dish, right? Yeah. I mean, it seems sort of ridiculous, but I think that's going to happen because you can reuse it to fix.

0:41:27.2 CK: Let's say if you have Alzheimer or some other pathology, you can fix it. Well, in principle, this piece of tissue may feel like something. Maybe it just feels a vast spatial extendedness like you get during certain meditative states when the meditators just talk about this vast luminosity. Right. So consciousness may end up being found in very strange spaces. And we have to. The only way we can verify any theory of consciousness is in us, in you and I and other people. And once we have a theory verified, then we have to go where it takes us.

0:42:02.8 SC: Okay. So your answer is to bite the bullet and say yes, these examples that don't seem to map on to our intuitive idea of consciousness should nevertheless count as conscious. That's your resolution.

0:42:14.6 CK: And conversely, things that look very conscious, like LLMs, same logic applies. We have to test, are they really conscious or are they just faking it?

0:42:24.0 SC: Has anyone computed phi for an LLM?

0:42:28.2 CK: Yeah. So there's a paper that's on bioRxiv now that we're trying to submit that shows formally within... So this is only assuming IIT is true because it's very formal and it's very specific. So what we can formally show, you have a simple model of a von Neumann [0:42:48.4] ____ machine. Okay? It's a very simple, with 116 gates or something like that. And you can show an actual... I mean it's all simulated. So in one case we have a simple automata that's heavily nonlinear. In the other case we map this automata onto a simple but perfectly general 4-bit computer. And you can show that the input output states of these two systems here a physical set of four nonlinear gate, here 116 gates as you would typically find them in a von Neumann [0:43:25.4] ____ architecture, functionally they're totally equivalent. They perform the same input, give rise to the same output. So you cannot distinguish them.

0:43:33.5 CK: So by Turing, they're exactly the same. But you can show phenomenal. This has some causal power. Well, this one doesn't exist at all at the level of the whole, at the whole system, 116 gates and only very small subset of them have a tiny bit of causal power. Why is the difference? Well, because machines on a typical CPU, one transistor will talk to two or three other transistors. But then of course, you have endless billions of these transistors arrayed to give you universal logic. While the way the brain is wired, you have one neuron gets input from 50,000, projects to 50,000 with heavily heavy overlap. And so this notion of causal power is radical different in one from the other.

0:44:23.9 CK: So you can show functional equivalent does not equate to phenomenal equivalence, including in the limit you have two systems that are functionally identical, this one is highly conscious, this one isn't conscious at all, which I think is the case for LLMs. This is not. You can't do such a proof for neuromorphic computers, nor for quantum computers. So it may well be that you can build. So there's no question, IIT says there isn't anything supernatural about the brain. So clearly you can build artifacts that are conscious, but just not the way we build them today. Not the computers running in the Cloud.

0:45:01.1 SC: Are you at all sympathetic to the idea, just to finish up the Scott Aaronson conversation, that IIT might be better thought of as a necessary but not sufficient condition for consciousness to happen? Like if you don't have this integrated information this high quantity phi, you're not conscious. But just because you do, doesn't mean you are. If you're not representing the world, if you have no semantics, et cetera, then maybe we shouldn't count it as conscious.

0:45:30.7 CK: Well, okay, so A, you can perfectly well, and in fact we've done this. There are a number of papers where you can evolve. In a computer, you can evolve simple creatures that have high phi that represent the world in a sense that there's a mapping between... These are creatures that you have 10,000 of them. You let them run through mazes, you collect the best one, you perturb their genome, you give rise to another brain, you send them through that. You do that thousands of times. You start with randomly connected brains, and you end up with brains that have mappings from the outside, the labyrinth or whatever they have to work, they have to deal with, onto their internal brain.

0:46:11.6 CK: So IIT doesn't preclude at all that there is this lawful mapping like you and I have, between the external world and the visual brain or the auditory world. I mean, who knows? I cannot think about the space of all possible theories. But within IIT... In fact, there's a thing called the central identity as part of IIT that says your conscious experience is identical to or can be fully accounted for, that's a better formulation, one to one, with nothing left over on either side by the totality of the intrinsic causal power. Every aspect of your experience, including every memory, every thought, every dream, can be fully accounted for. Now, this may be wrong, it's a theory, but that's what the theory states.

0:47:04.4 SC: And I do want to get on other things, but maybe a quick opportunity to talk about, like the nitty gritty of looking at brains. You've mentioned mouse brains already. Are we testing? Does it even make sense to ask the question? Are we... [laughter] Christof is showing us.

0:47:20.5 CK: That's a mouse brain.

0:47:21.3 SC: A little key ring with a mouse brain on it. Yes. Can you take the ideas from IIT and either test them in the mouse brain or learn about the mouse brain using these ideas? And if so, how do we do that?

0:47:36.4 CK: Yeah, I mean, the mouse brain. People think a mouse. Well, what's the mouse? It doesn't talk. I'll never have dinner conversation with it. All of that's true. But again, if I take a grain of mouse brain, dog brain, elephant brain, human brain, I mean, basic hardware is all the same. It's not identical. We evolved the last common answer between us and mice 60 million years ago or so, but the components are remarkable the same to the extent that we just published a whole series of Nature papers showing. So the brain consists of 5,000 [0:48:09.8] ____ 5,000 different types of brain cells, burkina [0:48:12.9] ____ cells, pyramidal cells, very different types of pyramidal cells, et cetera. The same 5000 cells can more or less be found in the human brain. It's just like genes. Our genome is on the order of 20,000 genes, the mouse genome is on the order of 20,000 genes. It's all very, very similar. But the difference is, of course, in the mouse it's much easier to do experiments and for ethical reasons, there are certain questions. You can't ask the mouse about language, of course, but you can do the basic.

0:48:42.0 CK: They're highly tactile, they're highly olfactory, they're less visual, they have vision, but they're less visual than us. So there are lots of experiments that you can do in a mouse or in similar creatures that are very similar or relate closely to experiments that people can do in humans where you of course limit it. Typically you don't access the human brain, you only see it indirectly. So you have to use EEG and fMRI. So we are all closely related. We are all nature children. We have some specialization, like a hyper-developed sense of importance. That's one big difference.

0:49:15.7 SC: I don't know. The mice might. I just don't know. But then, okay. I mean, so you're now still affiliated with the Allen Brain Institute, is that right?

0:49:24.3 CK: Yes. Yeah, correct.

0:49:25.7 SC: And is the Allen Brain Institute a hotbed of IITism or is it more ecumenical?

0:49:32.2 CK: No, no, no, it's purely so... We are now 1000 people. When I came, we are 100. When I was the chief scientist and the president, we went to 300. I stepped down and now we are 1000 people. So we focus on a few very large projects like cataloging all the cell types in the mouse brain or all the cell types in the human brain. Doing this. Also doing development. What happens in a very young immature brain as it develops? How do cell types change? What about the Atlas? We're doing these connectome projects. What we do at the finest level possible, where we don't image with photons but with electrons, cutting up tiny slivers, two or three nanometers thick of brain and cutting millions of these slices, putting them together. So we do a few very large projects that typically are not going to be done by industry and can't be done at a university. By and large, we don't think too much about consciousness and such things. We leave that to academics like me or like many other academics at various universities.

0:50:38.9 SC: Well, I mean, it's good. I'm very glad you said this because I think that it's fun to talk about the big picture philosophical questions. And they're not the same as but they absolutely do rely on this kind of empirical work where you're trying to figure out how many different kinds of neurons there are in the brain.

0:50:56.2 CK: Totally. Otherwise we still sit around like Plato and Aristotle at the Academy days about 2,400 years ago in Athens and shoot the breeze without ever making progress. But that's a great thing about science, including about psychedelics, including quantum mechanics. You can ask nature the right question in the form of the correct experiment, right, and you will get an answer. That's the best thing about science, better than any other human activity.

0:51:22.7 SC: I agree. I completely agree. And so, given what we've learned and what we're still hypothesizing about, there's a bunch of questions lying out there about consciousness that are pretty down to earth and operational as we've already alluded to. We want to know, are people conscious when they're dreaming? Are they conscious when they're in a coma? Are they conscious when they're six weeks after conception? Do you think we've made progress with these questions?

0:51:49.9 CK: Yeah, so in fact, I helped start. So one of them very much... On all of those questions, except maybe the final one, let's come to that. So A, I helped start a company called Intrinsic Powers that brings a device that used to be a pure research prototype developed by Giulio Tononi, the architect of Integrated Information Theory and a close medical colleague of his, Marcello Massimini, to test for the presence of consciousness in behavioral unresponsive patients.

0:52:24.5 CK: So if you or I have, after this conversation, a heart attack or bleeding or a car accident, we get into the ICU, we are unresponsive. We're clearly alive, we're on a ventilator, we're unresponsive. And they ask you, Can you hear me? Can you see me? Can you track a flashlight with your eyes? If I pinch you, will you moan or will you move a limb? You don't do any of that. So you're considered behaviorally unresponsive. This used to be called vegetative state. Okay. The tragic is four to five days out, if you remain in this day for four or five days, typically the team, the medical team will initiate discussion with your loved one. Well, what did he want? Is it time to pull the plug? Literally. They pull ventilator support. And up to 90%, up to nine out of 10 of these patients die because of withdrawal of life critical care. Okay.

0:53:16.9 CK: Now we do know that one quarter of these patients are conscious, but covertly conscious. So what that means, for instance, if you ask them, so you have this person who's lying there, you ask them, sir, imagine playing soccer for 30 seconds. And then imagine just being quiet. Again, you ask him a second time. Imagine playing soccer for 30 seconds, imagine lying quiet. Then you can see, for instance, the nice modulation in your motor cortex response. So it means this patient who's unable to communicate, where you have no idea whether there's anyone home still can lawfully voluntarily regulate their brain activity. Okay.

0:54:01.0 CK: And so, and together we developed this device. So essentially what it does, it knocks the brain by sending in a magnetic pulse called a transcranial magnetic stimulation. And then you measure the response using high-density EEG, and essentially you complete compute the complexity of that in IIT. And it turns out there's an absolute threshold, 0.32, it's normalized. So zero would mean it's totally flatlined. You knock the brain and there's no response. That's brain death. Or one would mean every point, every electrode is totally independent of the other. This doesn't happen in real brains.

0:54:38.3 CK: So typically, brain complexities between 0.6 and 0.8, typically in you and I, for instance, right now, or when we know when you dream, same thing. When you're in a deep sleep, it's very small complexity below threshold 0.31. When you're in a dream state, your complexity is high. When you are on psychedelics, your complexity is high. When you are under anesthesia, your complexity is low. When you're in an unambiguous coma where it's clear that you're not there, your complexity is low. So it's this very nice threshold and with very high specificity of 0.95, we can determine whether you are actually... So it's a primitive consciousness detector. It's the first primitive conscious detector.

0:55:22.0 CK: So again, this really shows, independent of all these idealism, physicalism, panpsychism, we can actually make progress on these questions. We can build consciousness detectors. Right now for us, in the fullness of time, we will certainly we can develop them for every, for any mammal, and maybe in the fullness of time, for any system, but right now for humans, so that's real progress.

0:55:48.0 SC: Does it help us with the prospect of brain-computer interfaces and being able to talk with people who are in a coma?

0:55:55.4 CK: No. So this is a purely device just to see is the complexity of this brain, is it sufficient to support consciousness. Now, separately, you can then specifically take these patients and further do rehabilitation or talk about if they remain in this state. So typically what happens? So most people, the natural course of events, people either die or they recover some functionality. So if you remain in this state, a small subset, five to 10%, they become chronic and then they're typically in a home or sometimes they get actually taken to their own home and they remain in this state where they have either no or only minimal. Sometimes people sort of recover some ability to move their eyes. In the worst case, they may not recover at all.

0:56:49.3 SC: Yeah. Okay, that's... [laughter] I'm sorry to hear that, but I'm glad you're doing... I mean I'm sorry to be reminded of this terrible fact about human life, but I'm glad that you're making progress on distinguishing between the cases that really seem to be essentially gone and those that are recoverable. Then what about the youthfulness question? What about the fetus and becoming conscious? Is that something that we can talk quantitatively about?

0:57:15.5 CK: Well, so all we can say, and in fact I was participating in amicus brief in that anti-abortion lawsuit. We know the cortex doesn't really fully develop till the second trimester. So your nervous system develops at eight to 10 week in the first trimester, but it starts developing but then really sort of doesn't develop the connectivity and the connectivity to the external world through the thalamus. That doesn't happen till week 22 or 24. EG is also pretty much flat until then. And then you get the first so-called burst suppression EG where you get activity and then quiet. Activity and quiet. And the typical pattern in you and I, that doesn't happen till very late end of third trimester where you get a more normal looking EG with all the waves developing. So it's very difficult to see right now but before the first trimester... In the first trimester it probably may not feel... I mean it's very difficult inference.

0:58:28.8 SC: Yeah.

0:58:29.5 CK: As far as we can tell, it probably doesn't feel a lot to be a first trimester. At the end of the second trimester you begin to get some simple reflexivity. So in premature babies when they need some operation, they do already limb withdrawal reflexes. But that doesn't tell...

0:58:47.8 SC: Yeah, that's just a reflex.

0:58:48.9 CK: A Drosophila embryo does the same thing. It curves away from a source of heat, like a match. So is that reflex? It's very difficult to be certain.

0:59:00.2 SC: Is it exciting or somewhat scary to be part of an amicus brief in front of the Supreme Court?

0:59:06.1 CK: Well, it was exciting and we were trying to do the right thing, but it didn't amount to much. It wasn't decided on the merits of the scientific case. It was just decided we need to send this back to the States.

0:59:20.8 SC: Okay, and then you already mentioned a little bit about LLMs and your opinions about that. But is there from IIT or from any competing theory, do we suggest some kind of roadmap for making artificial computer programs that truly are conscious, or is that a very far way off?

0:59:40.2 CK: Well, first, why would you want to do that?

0:59:43.7 SC: People want to do lots of things.

0:59:45.9 CK: Okay, that's true. Yeah. So according to... So you cannot simulate it.

0:59:51.8 SC: Okay, that's a big claim.

0:59:54.4 CK: Yeah. Because look, you can perfectly well simulate the black hole at the center of our galaxy, right? Sagittarius A*. But have you ever worried, Sean, that if you turn on the simulation, you're next to the computer program in the computer center that runs this, you're going to be sucked into the black hole?

1:00:12.4 SC: No, I have not worried about that.

1:00:14.0 CK: Why not? Where's the difference?

[laughter]

1:00:16.8 CK: No, I kid you not. Where's the difference?

1:00:19.9 SC: Well, the simulated test particles in the computer would be sucked into the simulated black hole.

1:00:26.1 CK: Yes, but that's all simulation. It's a difference between the real and the simulated. Right. So because this computation by itself does not have causal power to bend space time, just like it never gets... I mean, Sean, Seth made this argument decades ago. It doesn't get wet inside a rain simulation. Right. Same thing. It doesn't have the causal power. Yes. So you can perfectly well, even if you were to build a computer simulation of the human brain, the complete human brain, like Henry Markham, Blue Brain Project wanted to do that. Right. In the future we'll be able to do it. But still. And this computer simulation, of course, will wake up and speak because it can do that, but it won't feel like anything because you can't simulate it. You actually have to build it.

1:01:07.4 CK: So if you want true consciousness, you have to build it. You either have to do neuromorphic engineering where you build hardware, not simulate, you actually build hardware in the image of the brain. People are trying to do that at IBM and Intel, et cetera, or possibly quantum computers, possibly like, Willows at Quantum Compute lab, et cetera.

1:01:31.5 SC: So embodiment might be very important for this project.

1:01:34.6 CK: Yes, yes. It's not going to be substrate independent.

1:01:37.4 SC: Right. Okay, good.

1:01:38.9 CK: And so I don't believe, by the way, in this mind upload. Right.

1:01:43.3 SC: No, I agree.

1:01:45.9 CK: That I cut my brain, I get my brain's connectivity, I simulate it up in the cloud, and then this thing will be conscious. No. It may be able to mimic me, including my accent and everything perfectly, but it doesn't feel like anything. It's a deep fake.

1:02:00.5 SC: You're not worried about the Matrix?

1:02:01.4 CK: No. I'm worried about many things, but not the Matrix.

1:02:05.8 SC: Well, you've written... I'm sorry that it's taken this long to get there, but you've written a provocative book just last year. Do you want to give the title or should I read it?

1:02:15.5 CK: Then I Am Myself the World, which is taken from the second act of Richard Wagner's opera Tristan Isolde, where the eponymous lovers have this union where, I mean, they make love on stage, basically, but they want to overcome their own self. And so they have what's in modern language called a mystical experience. They become, they lose the boundaries between themselves, between Tristan Isolde and the rest of the universe. So they feel they are one with the universe, which is what really one of the defining features of what's called a mystical experience.

1:02:52.0 SC: Well, this is what we can talk about when you're at a certain stage of your successful academic career. You're talking about expanding consciousness now, which is certainly fraught territory. [chuckle]

1:03:04.1 CK: Well, yeah, I mean, yes, I've been fortunate enough to have such an extraordinary experience. And it leaves you totally unmoored. It leaves you completely flabbergasted. It leaves you... I had this when I was 65, and I thought my metaphysics, my sense of what exists has been fully established and at 65, isn't going to budge anymore. And then suddenly you have this ontological earthquake where all your search and literally all the plates shift and everything you lived and everything you did suddenly looks very, very different. So it's for sure it's a transformative experience. In a sense, if you undergo this, there's a distinct before and after. You'll never look at things the same way again which is another defining hallmark of a transformative experience.

1:03:57.3 SC: So we're talking about some kind of psychedelics or other pharmaceuticals or a purely meditative state?

1:04:04.9 CK: No, this was with help.

1:04:10.8 SC: Yeah. Okay, but you're a neuroscientist. You know that you had an experience that was caused by prodding your neurons with chemicals. So how does that help you get insight into the fundamental nature of reality?

1:04:28.7 CK: Okay, so you're totally right. My brain was a substrate of this experience. I don't doubt that for one second, helped or not. So William James talks about it in this wonderful book. I warmly recommend it to everyone. Read it. It hasn't aged, has aged very well that he first wrote in 1906, the father of American psychology. It's a variety of religious experience where he talks about mystical experience. And he says they bring one of the defining hallmarks of them, this what he calls noetic quality, from news, knowledge, Greek, you bring back this experience and it has powerful authority over you. You cannot help but now try to seek.

1:05:20.8 CK: I'm still a rational scientist in person, so what I try to do is try to see how does what I experience, how is that compatible with my view of everything else, including science, including brain science, including knowledge that my brain was a substrate of this particular experience. So I think you can reconcile both of those things. Look, it's a little bit like, Imagine, Sean, the following scenario. No one dreams, okay? People, you've never read about dream, but when people go to sleep, they just sleep. That's it. And you have one experience. One. How would that change your life? If you only had one experience, one dream experience, where you met your long lost parents or friends or et cetera, it would sort of profoundly affect you, right?

1:06:18.0 SC: Absolutely. Would not. I hope it would not lead me to change my ontological view of the world.

1:06:25.2 CK: Well, because you haven't had. Look, and I take Nova [1:06:28.3] ____ and James says this explicitly. You can't convince other people of your views, and I'm not trying to. This is just my experience.

1:06:36.3 SC: Oh yeah. Sure.

1:06:37.8 CK: I do know that these types of experience that I write about are common in literature, common across all cultures. Lots of people in lots of different settings have written about them. So they're not that rare, it's not unheard of. And so the challenge is if you accept them, what does this tell you about the world? And in particular, what does it tell you about the metaphysical, the metaphysics about the world? So you can continue to do. I can continue to do all my science, all my neuroscience, et cetera, but it really shook the ontological foundation of which is why I'm now much more sympathetic to idealism, which I hadn't thought about for 50 years.

1:07:16.5 SC: Okay, so my question was going to be. So what is the shift in your metaphysics? But is that it, that you're much more sympathetic to idealism?

1:07:26.6 CK: Yeah, because what I experienced there was effectively the world at large, that there was no self anymore. Self was completely gone, Christof was gone. There wasn't a body, there wasn't anything like that. And I instead I experienced the universe at large, literally the galaxy, I mean, the universe at large. Yeah, I know it sounds corny and all of that, but that's what you experience.

1:07:49.7 SC: That's okay.

1:07:49.8 CK: And you're in this time. There's no passage of time. It's not too slow or too long. This moment. And so, yeah, so you run around. This was at midnight on a beach in somewhere in Brazil and you're totally confused the next weeks and months. Right. Because you're trying to fit in. How does this accord with everything else I know? And then sort of I thought that Schopenhauer came upon this remarkable quote of Schopenhauer where he talks about 200 years ago in his book, well, his only book, he really wrote, The World as Will and Representation, where he talks about states like that. And as far as we know, he was not a medit. I mean, meditation didn't exist at the time in the West. Psychedelics really didn't. He never had any experience like that. That was of course before we encountered opium and other substances. So he just thought about these things. And then, yeah, I looked around at idealism and then I came across this philosopher computer scientist, Bernardo Kastrup, who espouses this modern form of idealism.

1:08:58.5 SC: And in your book you're going to suggest that people can understand consciousness better and expand it through these kind of intense experiences?

1:09:08.4 CK: I would certainly suggest, yes. So I grew up at a time when I did... I was heavily into alcohol.

1:09:20.2 SC: We were all young once. Yeah.

1:09:21.3 CK: In the West, but nothing else. But that's changed over the last 10 years. Yeah, they can reveal you... So A, they can reveal these states of self where your ego is dissolved, ego dissolution, where there's no self and you can experience a profound beauty. Like here I'm looking at the island, the forest, the green lush forest. You can just see the transcendent beauty of it whole. And there's no self to get in the way. I mean, we need the self. It's necessary to do long-term planning, all of that, but it also constantly get in the way. It's always about me, me, me, me and me. I mean, look at the most senior political personality in this country who shall remain nameless. Right. Or the two of them. Both of them. Right.

1:10:05.5 SC: Yes.

1:10:06.6 CK: And it's all about me, me, me, me, me. And so suddenly you have this experience of loss of self and you realize, well, that's perfectly fine, the universe is profoundly beautiful without me. So I think that's really an essential experience that most everyone should have and can have in a safe, in a, if you do it in a safe manner. Now, not everyone, there are certain dangers. Some of these substances are very powerful. So you have to be very cautious. You have to use all the proper safeguards, et cetera. But I think for most people, they can reveal aspects of consciousness that they haven't thought about or they have never experienced or they have never imagined.

1:10:47.1 SC: Good. Should we wind things up by talking about quantum mechanics?

1:10:51.6 CK: Sure. It seems like a good thing to end on.

1:10:55.0 SC: We've talked about it already. But you have a paper out not too long ago about the possibility that quantum mechanics does play a role in consciousness.

1:11:02.9 CK: Yeah. So we initiated this collaboration with Hartmut Neven, who is a, he is a physicist and also his background is computational neuroscience like me. In fact, we had our same advisor back in Germany. But since then he's a vice president now at Google and heads the Quantum Compute Google group in Santa Barbara, building, busy building these large-scale quantum computers and then together with a bunch of experimentalists, particularly Ken Kosik who's at University of California in Santa Barbara, and Luca Turin, who's at a university in UK. And the idea is, well, so there's theoretical work and then right now there's experimental work trying to test it that sort of inverting Penrose really...

1:11:53.5 CK: Penrose famously argued within his objective reduction that the collapse of the wave function, you have a particular system, it's in a superposition that collapse because of interactions with the environment. And that gives rise to a little ping, a little bit, a little conscious moment. That's his argument. And then he got together with the anesthesiologist to hammer off microtubule and all of that. But let's divorce the instantiation of that microtubule from that idea. It's controversial, but has remained irresistible to many, many people. All right.

1:12:31.5 CK: So we are inverting that because we think, well, if you have a system that consists out of two or three or more than one qubit that are all entangled, something Roger Penrose never considered, then if that's true, you could then use that to communicate faster than light information violating relativity theory that no one really wants to violate. Right. So you can invert that by saying, well, no, it is when the system goes into superposition that you create a conscious moment. And if you have a system with 20, let's say qubits, then they go into this superposition of 20 qubits. But you, of course, the defining thing of consciousness is you only ever experience a definite experience, right?

1:13:22.0 SC: Not a superposition.

1:13:23.4 CK: So this is where the many worlds interpretation come in. I only travel down one of these branches of the two to the N possibility. If I just have up down states, 2 to the N universes and that's the one I consciously experience. And what you experience depends on the way these interact. And here you could use for example, IIT to explain that. So it sort of inverts Roger's proposal. And if that's true, you can then sort of build quantum systems that could interact with your consciousness as long as you can get a system. If the latest iPhone has quantum computing and is entangled with my qubits in my brain, then that would be one conscious mind. So you could expand your consciousness if you have the right technology.

1:14:15.2 CK: Now it's one of many crazy sounding ideas in the interface between consciousness and quantum mechanics. But the great thing is we're trying to test it now. And this again where we come back to this thing. If you ask the right experiment, you can get an answer. So there have been since 30 years, Klaus Schulten's work, these hypothesis that biological system, in particular bird navigation, avian navigation, uses a quantum compass, right? In quantum, in the interaction in the eye there you have this, the interaction between photons and the molecule. And this depends weakly seems depend on magnetic field. And that songbirds can use that to navigate across vast distances. Ultimately that's based on a radical pair mechanism. That's one hypothesis. Or on nuclear spin interaction, it's another one.

1:15:17.1 CK: So we are trying to test that. There has been one claim that xenon... So you take the rare gas xenon, it's an anesthetic. This is known. It's actually very good anesthetic because it's rare gas. It doesn't interact much with anything, it's very safe, it doesn't burn. So it can be used in the clinic. It's just very expensive. So it's not really used in day-to-day practice. But it is an anesthetic. There's one claim, there's an isotope dependency, that xenon 128, 129, 130 and 131 have different anesthetic potencies. Now the difference in atomic mass is minute, is under 1%, so the hypothesis is unlikely to be due to the difference in mass. But the most critical difference is that 128 and 130 has spin zero, while 129 and 131 has spin respectively 1/2 and 3/2.

1:16:15.3 CK: So now Luca Turin is trying to test this very simple experiment in flies. In other words, if you take a bunch of flies, you put them in a tube. So now you see 30 drosophila fruit flies in a tube. You introduce gas, various types, different types of xenon, different isotopes dependent xenon gas, and you compress. And at some point you see when you reach certain pressure that the flies are all immobile, they're anesthetized and then you release them again. So you can do this experiment multiple times. So the very simple test is, is it true that this dependency how deep you have to plunge the plunger into that vial with xenon gas until you achieve immobility, that here stands in for anesthetic potency. Does that depend on the isotope of xenon?

1:17:08.6 SC: Good.

1:17:08.9 CK: And a different model system that we're trying to do with Ken Kosik, what about humanoid, these human cerebral organoids made out of human pluripotent stem cells. If you can get them, if you put xenon on, again, ultimately the tissue will... Usually these tissue, these mini brains fire action potential. If you put sufficient xenon in that, that ceases, that stops the firing. So does that the point, the precise point, when that firing stops, does that depend on the isotope? Is it going to be different for Xenon-129 versus Xenon-130 ? That's the question we're asking.

1:17:46.8 SC: It's certainly an experiment very well worth doing. I guess it's late in the podcast so I can ask one technical question. The hypothesis that a system being in superposition is relevant for consciousness somehow. That's what bugged me a little bit when I saw the paper because being in superposition is sort of not an objective fact. Right. That is relative to what basis you're using to describe your system. So is there more physical criterion that we can replace for that?

1:18:16.1 CK: No, not right now.

1:18:17.5 SC: Okay. That's why it's science. We got to do it.

1:18:20.8 CK: Yes, yes, yes.

1:18:24.0 SC: What is your thought? What if you had to place Bayesian priors here? Do you think that quantum mechanics is super important for consciousness?

1:18:30.5 CK: Probably not in the human brain.

1:18:33.0 SC: Okay. But it would be cool.

1:18:35.6 CK: On the other hand, look, evolution, it's been around since four and a half billion years, roughly. Right. And so evolution is very clever. It's done all sorts of things. If it's not outlawed by the laws of physics, somewhere, and if it's really beneficial for the organism, then it may well have been exploited. Would I bet my house on it? No.

1:19:00.5 SC: Well, not at any point.

1:19:01.3 CK: That's why we're doing the experiment.

1:19:02.3 SC: Yeah. That's why we do the experiment. And that's a very good philosophy to end with. So, Christof Koch, thanks so much for being on the Mindscape podcast.

1:19:09.2 CK: Thank you very much for having me, Sean. That was very enjoyable. That was very enjoyable.

1:19:13.3 SC: Good.

1:19:15.2 CK: What is your paper about? You said you just published a paper 10 days ago.

1:19:18.8 SC: Oh, it was just a review article for Nature. An essay about the 100th anniversary of Quantum mechanics and why it's still controversial, why it's still difficult to really figure out and get a consensus on what quantum mechanics actually says.

1:19:32.6 CK: That's right. 1925.

1:19:36.1 SC: It's the quantum year. It's the United Nations has said that this is the year of the quantum. Yeah, so and we still don't understand it.

1:19:42.4 CK: It's not true for general relativity. Right. There are none of these controversies swelling around GR.

1:19:47.0 SC: No other theory has these controversies. There's not like a whole subset of people doing philosophy of electromagnetism. Right. Quantum mechanics is the one that... And I think it's an embarrassing situation for the physics community because we have not addressed the problem. I think that's an issue. Yeah.

1:20:06.0 CK: But it's also an opportunity.

1:20:07.9 SC: Exactly. Well, I like to say I personally am very slow. So I'm perfectly happy with the rest of the community not racing me to do this. I'm trying to figure it out myself.

1:20:17.2 CK: Yeah. So please send it to me, the article.

1:20:19.9 SC: I will do that. Thanks.

1:20:21.4 CK: Thank you. That was fun.

[music]