This is a transcript of a conversation between Atai Barkai and QualiaNerd that took place on X. You can find the original conversation here. It revolves around Atai's paper "On the Psycho-Physical Parallelism."

(The transcript was generated via LLM, so it may contain a few minor errors.)

Kinematics and dynamics

QualiaNerd: I've been reading this paper you've written on the psychophysical parallelism. You've proposed a model of kinematics of reality. Maybe you could sum it up if you're going to use this as educational content. Perhaps some people will listen to this in the future, and then maybe I'll have some questions.

Atai Barkai: Sounds good, feel free to keep going and to guide this conversation.

QualiaNerd: So to maybe correct some of my potential misunderstanding - you propose a model of kinematics of reality. Perhaps we should introduce this concept of kinematics first. Kinematics, as opposed to dynamics, is the study of the fundamental constituents of reality or fundamental constituents of a particular system. It's about what the system is actually made of. It's not the interactions between the parts of the system but the irreducible states or irreducible components from which that system is made.

For example, you could have as these components of a system something like quantum fields in quantum field theory, or in Newtonian physics, you could have something like stable, rigid three-dimensional bodies. The dynamics are the rules of inference or the rules of transformation or the laws of physics that govern the state transitions and evolution in time of this particular system. And of course, you could capture both kinematics and dynamics in terms of a state space where you would add time as another dimension. That's the basic summary as I understand it of kinematics.

Atai Barkai: Yeah, that's honestly exactly right. This division between kinematics and dynamics is a little bit artificial. In the end, you specify a physics model that is supposed to represent the world. The way that all physics models work is, you first say what can exist under this model. To name a really silly example, an electron is not something that you can specify in any Newtonian universe. There's nothing, no states that you can write down that correspond to an electron because the electron has qualities that don't exist in a Newtonian universe.

So you specify what can exist - that's the kinematics of a theory. And then you're specifying, like you said, the regularities and ideally the irreducible regularities that relate the things that can exist under your model to one another. To some extent, you can say how they relate over time, but really time appears both in the kinematics and the dynamics of a physics theory. It's easiest to think of time in the concept of this three-plus-one dimensional universe.

You could just as easily - and to some extent that's what special and general relativity do - treat time almost like another dimension of space. There's this strange regularity across this time dimension where things are very continuous. If you have space, like a field of strawberries, there's a regularity - they all look the same one next to another, but then it can just end. The strawberry field can end and now we have a potato field.

You can look at time like a regularity that says these coordinates have this continuity, and discontinuity is honestly exactly the laws of physics - it's how things change over time. But I digress a little. The big point is that there is what can exist under your conception of reality, and then there is how these things that exist relate to one another. What are the regularities?

We all know that at a basic level, if you drop a ball on the floor, you expect it to go down. That's a regularity. And of course, you can make that more abstract going from "it falls to the floor on earth" to "actually it goes towards the center of earth even if you're in space" and "actually it's not just earth, it goes to the center of any mass." So you realize there's one simple regularity that accounts for all this, like Newton's law of gravitation.

Fundamental information integration and the binding problem

QualiaNerd: All right. Well, I mean, you are the physicist here, but as far as I know, it checks out. To continue with my understanding of the paper, you propose a model of what can exist - the kinematics of reality - that has two features: fundamental information integration and classical indescribability.

The fundamental information integration, if I'm not mistaken, has to do with what has been called the binding problem or the boundary problem in consciousness research.

Atai Barkai: That's right.

QualiaNerd: To put this in extremely simple terms - because people sometimes overcomplicate these things - for example, you see multiple colors at the same time. That already is an example of fundamental information integration. It's not just that you are paying attention to one particular color and not to others. Some people might get confused by that, thinking "what binding is there?" because they confuse attention with the broader field of which this point of attention is part.

Even if you're focusing on just one particular small part of your visual field, you are still seeing multiple colors at the same time. Unless you're colorblind, of course, but we could give other examples. You see multiple colors at the same time or you see colors and hear sounds at the same time. It's this property of an ontological monad that is not a point-like or particle-like thing.

Atai Barkai: Right. There's some complex structure embedded in there.

QualiaNerd: Yeah, right. It's a non-infinitesimal ontological entity.

Atai Barkai: Right. Exactly. And I really like your communication style of literally using color emojis to instantiate one of these clusters in the brain of the other person you're talking to. Instead of talking about it, you just instantiate it. I think that makes things clear.

QualiaNerd: Yeah, exactly! This was something I've been successful with to some extent, although with some people who have a very symbolic or analytical thinking style - nothing wrong with that, they can be great mathematicians - even that approach might come up short. For example, I recently had an interesting conversation with Pete Mandik on Twitter who's a philosopher. I tried this with the colored emoji, with a red circle. And his response was that it was not a well-formed sentence in English.

Atai Barkai: That's true.

QualiaNerd: Exactly, that's the point.

Atai Barkai: That's the problem.

QualiaNerd: One problem with the colored emoji is something that Andrés observed with Thomas Metzinger, who might say, "But the color can't really be dissociated from the shape. What's the redness of red that's separate from the circularity of the red circle?" And that is a fair point, at least in normal states of consciousness. It really is true that colors generally come in shapes.

Perhaps this might be why something like a smell or scent might be a better example of a non-symbolic structure - something that is not a thought or abstract model but completely unlike that.

Atai Barkai: Yeah.

QualiaNerd: It might be a better example than a color because color has shape, and to grasp the concept of color in itself, this non-symbolic qualia, you have to mentally dissociate the color from the shape. Some people might react poorly to that, creating a cascade of symbolic thinking where they immediately start thinking about frequencies of light and how the brain represents frequencies of light and so on.

Atai Barkai: I think I agree to some extent. The human brain is maybe literally the most complex structure we know to exist in the universe at this point. When you have human qualia clusters, there's a lot of complexity there, and I think it can get really tangled with many interesting open questions. But that can be almost a fog that obfuscates the simple core question, which I think you understand very well. That's what distracts or confuses people who get lost in the mire. I agree that to some extent, smell is simpler than color.

QualiaNerd: I've been thinking about the pedagogy of this - how do you approach talking about this topic with somebody who might be very intelligent, a very good philosopher or mathematician, but who doesn't have good meta-cognitive access to non-symbolic qualia? To respond with something like, "Well, that's not a well-formed English sentence. I don't know what you mean by that."

Atai Barkai: I would love in this conversation to get past that and get to the actual interesting stuff, because there are so few people that you can actually get past that point with and get to the interesting stuff.

QualiaNerd: Absolutely, no problem.

Atai Barkai: Because I agree that's a huge communication problem. I don't have a solution to it. It's a depressingly difficult problem.

Classical indescribability of qualia

QualiaNerd: Right. And the second part of this puzzle was the second property that you proposed of the kinematics of the correct model of reality: the classical indescribability of these ontological monads or irreducible components that form reality. I must admit that I don't quite understand what was meant by that. Please tell me if this understanding is correct.

If you think about how mathematics might have developed - this is obviously very simplified - perhaps people at first needed to count rocks or sticks or sheep, so they invented integers. Then maybe one day zero sheep returned, so they invented zero. Then they started borrowing stuff and invented negative numbers, and so on through irrational numbers and complex numbers.

But the thing here is that it's always a very limited qualia variety. You're always representing certain qualia clusters with other qualia clusters. When 16 sheep come back home and you count them, that's a qualia cluster - the abstract thinking of numbers - representing another qualia cluster, which is what you see in your visual field of 16 sheep returning. Obviously, there's a lot of processing going on, but it's roughly this - you're representing some qualia clusters with other qualia clusters.

And it would be impossible to describe or articulate something like quantum field theory only with the mathematical apparatus available to people thousands of years ago who only knew about integers. You couldn't do it.

Atai Barkai: Right. That's actually a really interesting thing. You mentioned complex numbers, and I spoke to somebody just the other day who drew a parallel that I hadn't noticed between my proposal of these classically indescribable structures and complex numbers. Complex numbers - the square root of negative one - is a nonsensical concept from a naive integer perspective, except it turns out it's a very rich structure. There's this whole discussion of Platonism, whether mathematics is discovered or not, but the bottom line is it's a really coherent structure that emerges when you assume there's such a thing as the square root of negative one.

There's something similar about this classically indescribable information. Maybe it's a good point to step back a little, because I come at this from two essentially completely distinct perspectives that I claim converge to the same point. One side is just pure physics, pure quantum mechanics. The other side is the consciousness stuff.

The kinematical problem of consciousness

Maybe I'll start with the consciousness stuff. Like you said, I think fundamentally the hard problem, as I argued in the second small paper, "The Kinematical Problem of Consciousness," is essentially a problem of kinematics. The real big problem is that we have these structures that we think exist in the universe - these qualia clusters, consciousness, whatever you want to call it - and the problem is we cannot represent them under our model of the universe.

Our model of the universe that's supposed to allow us to represent everything that can exist - you can't find where to fit them there. This model can accommodate lots of things: cars, black holes, stars, electrons, quarks - all these things you can represent under our current model of the universe. And you can even represent people and brains. That's all fine, and things that behave as if they're associated with qualia clusters.

QualiaNerd: Perhaps one thing I would say here is that we're just scratching the surface in terms of representational systems. You can represent some very primitive or simple features of your state of consciousness with mathematics or words or some representational system that usually takes the form of abstract thinking. The question is, where do we go from there? What new qualia varieties would need to be invented to more accurately represent qualia with other qualia? Because that's what's going on when you do mathematics, when you do philosophy. Words, thinking, numbers - when you think about the number three, there's also some particular type of qualia there.

Atai Barkai: Right, that's exactly the simple thing at the core of all the complex stuff. Essentially there are two qualities, as I argue, that we implicitly associate with qualia clusters that you cannot put into our current model of the universe. The first one is the binding problem - the fundamental integration. I didn't use the word "binding" when I wrote the paper because I wasn't familiar with that term when I wrote it back in 2018 or so, but that was just my ignorance. That's a well-recognized part of consciousness research.

Essentially the problem is that in a physics model, you can define all sorts of complex structures, but these structures are always defined essentially by the behavior of the system. There are emergent regularities that are not perfect, but they capture a lot of what's going on by assuming some complex thing to exist.

To name a simple example, if you look at the sun, fundamentally from a physics perspective it's a collection of trillions and trillions of particles. But you can think of it as a big sphere in the sky and you'll get pretty far if you think that's what it is. You won't get perfect ability to predict what's going on, but you get really far. So it's this emergent structure, but it's really just a function of convenience for making sense of the universe. The universe does not know there's a sphere there.

And this is true for the sun, it's true for everything else. There's nothing you can find, no boundary you can paint, that the universe itself recognizes. In a classical universe, there's no system that you can draw a boundary around where the universe knows the parts of these systems are related to one another. It's only through emergent analysis. It simplifies analysis, but the universe doesn't care about analysis.

The biggest difference is that with qualia clusters, that's not how we identify that the parts are part of a single thing. It's not that we analyze it and say, "Oh, it's really convenient to treat them as one thing." There's something different going on. The way that you identify different parts of a single qualia cluster are part of a single qualia cluster is not through analysis of behavior that simplifies things.

Even here there's an asterisk because the contents of qualia clusters also function essentially as the cognitive brain state of this intelligent agent, and that's a very important point. The intelligent part does emerge, just like with GPT-4 - there are all these transistors under the hood doing one bit flip at a time, and the collective thing is nowhere to be found, but if you zoom out, it behaves as if it's reasoning over this complex unit of structure.

So you have this emergent unity, and that's fine. Consciousness behaves in the same way in the sense that when you observe a person acting in the world, there's this complex structure that emerges over the behavior. But qualia clusters are used to implement this behavior, but that's not how we identify them in the first place.

QualiaNerd: Well, I mean, you identify them by being them. That's the thing - the best model of something is that thing, and anything else is just a lossy representation.

About GPT-4, this is actually a very interesting question - to what extent is there a match between the functional boundary around a particular system and the ontological boundary? In the limit, the two would probably have to be the same. A digital computer and a human brain might be functionally equivalent from a certain cherry-picked point of view, but they are not functionally equivalent if you are a hungry Hannibal Lecter.

In the limit, I would say the functional boundary will be the same as the ontological boundary. But from our cherry-picked point of view, the functional boundary between the computers implementing GPT-4 and the ontological structure - I would say there is quite a big mismatch there.

Michael Johnson has written about this - digital computers are these qualia clusters put together in ways that are computationally useful from our cherry-picked perspective, but there is a big mismatch between the functional boundary as we interpret it and the ontological boundary of what's actually happening there.

If you have something like a digital computer, there is no single ontological monad, no single qualia cluster, no single "topological pocket" to use Andrés's model, that instantiates the entire algorithm running on this digital computer. What you have are these relatively simple entities that don't contain anything even remotely similar to the information-rich real-time world simulation instantiated by a human brain. You can put these small things together in a clever way to get some computationally useful behavior, but in the limit, it will not work. You will not be able to simulate everything that a human brain can do.

The meta-problem of consciousness

Atai Barkai: There's a thread here at the risk of popping new threads, but I'll mention briefly - I think there are a few things that still haven't propagated as much as they should in the consciousness research community. One of these things is the meta-problem of consciousness.

QualiaNerd: Yeah, or the meta-problem constraint on theories of consciousness, because like you said, there's a really interesting fact about consciousness or qualia clusters, which is that we can talk about them.

Atai Barkai: Right, yeah. This means that they are causally significant. I mean, if you look at panpsychism, it's not compatible with that. Philip Goff talks about how consciousness is a problem of philosophy, not of science. But I think that's not a coherent point of view because in a panpsychist universe, as he imagines, everything is implemented via qualia clusters, but by definition, that fact cannot leak to the dynamics, the emergent dynamics of these brains talking across a Twitter space. But that's a whole other thing, maybe it was a mistake to bring that thread up.

QualiaNerd: Oh no, that's very interesting. I'm not sure why that would be so. What exactly would we be unable to formulate in a panpsychist universe?

Atai Barkai: There are different flavors of panpsychism. I think some of them can be logical, but the most dominant one is Russellian panpsychism that says physics only tells us the relations between things. It doesn't tell us what these things are implemented on top of.

QualiaNerd: Right, like what is the fire in the equations?

Atai Barkai: Yeah, like what is this structure over. They correctly recognize, "Hey, there's something that we think exists in the universe, but we can't find anywhere in our physics equations." That's a correct observation.

QualiaNerd: I would bring up one point - people sometimes do this interesting thing where they expect a description to be non-circular and to somehow fully capture everything there is to know about the thing described. But that's completely impossible.

This happened to me when thinking about valence. I was wondering, what's valence really? What determines valence? Is it symmetry? I was reading Michael Johnson and Andrés, studying symmetry theory of valence, and considering various other possibilities.

[Technical difficulties occur and are resolved]

QualiaNerd: So what I was saying was that I was thinking about this - what determines valence really? Is it symmetry or the fulfillment of goals? Then I realized what I actually had been doing was expecting to find some kind of non-circular description that would somehow fully capture everything there is to know about the thing I was trying to describe. And that's not really possible.

You have to accept that you will never be able to find a perfect description, aside from circular descriptions. If you have a circular description, then it fully captures everything there is to know about the thing you are trying to describe, but it's pointless because you can't really use it for anything. Everything is what it is, and that's it - but that's not helpful.

I realized that, to paraphrase Sam Harris - and I'm not suggesting anybody should actually do this - put your hand on a hot stove if you want to know what valence is. That's valence. Something like symmetry or any physical theory we might develop to describe valence will only be a description. As David Pearce would say, it will be a mathematical physical description of valence or a mathematical physical signature of valence, but it will not be the valence in itself.

It's similar with consciousness in general. You will never find a description of colors that will cause the visual field to disappear to be replaced by the description. No matter how you describe colors, the description will always be something different than the thing being described. This is something you can't overcome.

You've written about wanting an ideal theory of physics that would fully capture all of the processes that go on when somebody formulates a claim such as "I'm conscious" or "I see redness" or "I have qualia."

The hard problem of valence

Atai Barkai: Yes, on the valence side, my feeling is this is a really deep problem. I suspect there will be an answer that makes sense. I saw that Andrés posted "The Hard Problem of Valence" video the other day. I haven't watched it yet, but from the title, I agree. I think there's a really hard problem there.

QualiaNerd: That's a good talk, I can recommend it.

Atai Barkai: I'm going to listen to that. I think Philip Goff takes this as evidence of God. He says, "How come the function is aligned with the truth?" Because it could have been flipped, let's say - when you have positive valence, the person behaves as if it got negative feedback. I don't know what the solution is there. I don't think it's a miracle. I think there's some coherent explanation, but honestly, to me right now the answer is almost unimaginable.

QualiaNerd: To give an example of what this perfect description might look like that would fully capture everything that goes on when a person formulates the claim "I am conscious" or "I have qualia" - it would be a kind of full-spectrum superintelligence that would instantiate all of the qualia clusters of this person. The qualia clusters of which the person consists are probably not just the brain or just this one world simulation run by a brain. There's probably more of them.

Some kind of full-spectrum superintelligence that would instantiate all of them at the same time - at a bare minimum, you would need that. That already would be a kind of perfect account of what's going on there. Then you could create some model of that that would be different from the thing described because the description is not the same thing as the thing described unless it's circular.

If you want to have a perfect model of something, then the perfect model is that something and nothing else. So aside from that, I don't really see this as a problem that needs to be overcome somehow.

Atai Barkai: I think the question to me goes back to kinematics. Even the fact that there is such a thing as valence - forget which is which - where do you find that in the laws? And I think it does connect to the classical describability thing.

So maybe we should take a step back and talk about this more broadly. There's the binding problem - how do you define complex structure that you can objectively identify by looking at the state space of the universe, or at least identify by some mechanism other than looking at emergent behavior that simplifies analysis, which is how we generally identify complex structures in every other part of the universe? That one is relatively well understood by the consciousness research community.

But I think there's a second problem that everyone implicitly knows but isn't explicitly addressed enough, which is what I call classical describability. You represent this very well with the colored emoji. You ask, where does this exist in the universe? And you put the red circle emoji.

If you look at qualia clusters, there's the old question, "Is my red the same as your red?" But I think there's an even deeper thing underneath, which is well understood when you look at how you describe red. This circle thing - you can't describe it. It's not a sentence in the English language, but you can instantiate this qualia thing that you cannot describe in somebody else's brain.

QualiaNerd: I would say this is always happening. Every description is like that, even a description of the circular shape. I don't think anything is classically... well, you can draw an equation that captures all of the structure of a circle.

Atai Barkai: No, the equation will not capture it. What do you mean by "capture the structure"? There still needs to be this step of interpretation. You see the circle and then you create some way to represent it using some other qualia variety with an abstract equation.

Let's step away from qualia for a second. I can describe circles. I can describe this structure called a circle: x squared plus y squared equals some radius squared.

QualiaNerd: But that's still qualia.

Atai Barkai: It's true, but if we put ourselves for a second in a qualia-less universe as a thought experiment, you can represent circles in a classical universe. It will be an emergent thing, a cognitive thing.

QualiaNerd: I can't... a circle is...

Atai Barkai: Put it this way - nobody's asking "Is my circleness the same as your circleness?"

QualiaNerd: Oh yes, this is true.

Atai Barkai: You can draw on a piece of paper a representation of what makes something a circle. My claim is that you can't do the same for color. What you do is say, "I cannot describe to you red, green, or blue, but what I can do is tell you that these five things in my qualia cluster right now - I cannot describe their color, but I can tell you they're all the same one."

If I look at a tree and see a bunch of leaves, I can't describe greenness. I can say, "This thing that I cannot describe is shared by all these leaves on this tree." And I can see this car that's also green and say it's the same one too. So you can divide the world in that way. That's the part you can capture.

You can get more advanced and say it's the same qualia experience when I look at this particular RGB value. And then you can send that over the wire - you can send a recipe to instantiate this division of the colors of the qualia clusters. Essentially you send a JPEG or something - some representation of which pixels light up where. But the individual thing, this structure, is only part of what's there.

QualiaNerd: Do you see this as a fundamental difference? Would you say that the shape of the circle is somehow fundamentally more describable than the color?

Atai Barkai: By "color," I don't mean all aspects of color. Some things about color can be classically represented, particularly the relations between different color qualia across a single qualia cluster. That's a really important point.

There's the question of how do you know my red is the same as your red. But how do you know your red is the same as your red from three seconds ago?

QualiaNerd: That's true.

Atai Barkai: For you to know that, the implicit assumption is that your brain can record that redness from three seconds ago and compare it to now. But how can it record? The brain, as far as we know, can only record what you can also record on a piece of paper. I know how to record the shape of something on a piece of paper.

QualiaNerd: I wouldn't formulate it that way. I have this idea deeply ingrained that a description is always different from the thing described. So yes, you can write down what the brain does on a piece of paper, but the piece of paper and the brain will still not be the same thing.

Atai Barkai: It's true.

QualiaNerd: So you are saying that some parts are more describable than others. I agree with that - some parts of qualia clusters are more describable in terms of the current mathematical apparatus or linguistic apparatus than other parts. But I'm not sure if this is fundamental.

Atai Barkai: You can imagine an almost maximally simple qualia cluster, which is just yellowness. Who knows if that can actually be physically instantiated because when we look at the brain, it's this really complicated cognitive system. But still, you can imagine this thought experiment of a really simple qualia cluster. And there's something about this that you cannot write down. It's like, "Describe yellow to me." It's such a simple problem.

QualiaNerd: But how would I go about describing something like a square?

Atai Barkai: You can literally draw it. You can say, "Put your hand on a pencil, move right five coordinates, then down five coordinates, then left five coordinates." You can describe it. We do this all the time. You can send information that allows somebody to recreate this.

QualiaNerd: With this I agree - only fairly shallow characteristics of the qualia clusters that are our world simulations are really well describable by the current mathematical or linguistic apparatus that we have available.

Atai Barkai: But color should be one of these simple things. There are complex things like emotions - how do you describe emotions? But color is a really simple thing. You can imagine a qualia cluster that just has yellowness. The complexity is not the problem here. The problem is there's no language for this. There's no language, and the deeper problem is that you can't even imagine how there possibly could be.

For emotions, I can imagine a language that would evolve over time to describe a really precise emotion. Some meditators are coming up with extra precise language to describe various states of mind. So those parts are describable in principle. But when I look at yellowness or redness, how could anyone ever describe this to somebody else?

The persistence and representability of qualia

And I would apply this also to yourself three seconds ago. How do you know that what you call yellowness right now is the same thing it was three seconds ago? We take for granted that it's the same particular thing that was instantiated in the universe when you looked at the same emoji three seconds ago.

QualiaNerd: Well, that's not logically entailed, but for Occamian reasons, I would say it's probably more parsimonious.

Atai Barkai: In some ways yes, in some ways no, but we all assume that. We don't say, "Oh, it might have been different." We say, "No, I remember what it was three seconds ago and it was the same thing." But what I try to point out is that when you say "I remember what it was three seconds ago," you don't actually have access to the qualia cluster of three seconds ago.

QualiaNerd: Of course.

Atai Barkai: It's just a memory. You're reinstantiating a new quale in the current consciousness cluster. You're saying these two things are the same - my instantiation of the memory and the instantiation of what I'm seeing right now. But they're fundamentally across the same qualia cluster. There's an important point here - we can compare these colors, these classically indescribable things.

QualiaNerd: But you would say that you can't describe them with the qualia variety of abstract symbolic thought. That's what you mean by classically indescribable?

Atai Barkai: What I mean is that it's not isomorphic to any sequence of classical bits.

QualiaNerd: Well, but even that is already a bit of an abstraction. A sequence of classical bits, ones and zeros, even that needs to be interpreted somehow. If you write down ones and zeros, it doesn't really mean anything - it depends on how you interpret it.

Atai Barkai: But everything else we know of, you can capture in that way. You can make some interesting case about continuous phenomena, but can you digitally represent it?

QualiaNerd: Maybe a more accurate way of putting it would be that the representations we use to describe relationships between colors, as opposed to the colors themselves, can be further represented as classical bits. But mathematics itself is not a sequence of bits. There can be fairly complex and interesting mathematical qualia when you think about complex numbers or equations that would probably be in the realm of abstract symbolic thought, but it's not shifting around ones and zeros in your mind.

Atai Barkai: But the question is, can it be isomorphic to another system that is only shifting zeros and ones, that is doing the same thing?

QualiaNerd: I dislike this formulation of qualia formalism by Andrés and Michael Johnson for precisely this reason - they say that every qualia cluster, or as they would say, every moment of experience, is isomorphic to some mathematical object. Well, with enough reinterpretation or interpretative freedom, anything can really be isomorphic to anything. I don't think this isomorphism is some kind of iron-clad law of nature. Maybe an alien species would have entirely different mathematics and wouldn't use ones and zeros.

Atai Barkai: I think that goes to the binding problem - where do you draw the boundary? Because anything can represent this complex thing. You can take a bunch of zeros and ones and say, "If I look at it like that, it represents a picture of a couch." But if we look at it differently, maybe it'll represent something else.

I'm trying to go to the simplest possible version of this - just yellow, the yellow emoji. And if you have two of these yellow emojis, there's a particular thing there. There's the cognitive layer of yellow and how everything else you know is yellow - bananas, the sun. But there's also the thing you're just looking at right now - that's the qualia part, the qualia cluster part. If you try to decouple that from all the cognitive parts.

It exists in the universe, evidently. When you look at these two yellow emojis, there's a thing that can exist, be instantiated in our universe, which for lack of a better term, we call qualia. And I think it's a bad term to call it "yellow qualia" because it implies that it's not only the same across people, it's the same across yourself three seconds ago, which you don't have evidence for and would necessitate your brain being able to record yellowness, which obviously it can.

QualiaNerd: You have this memory of the ghost of the parted percept, and you check against that.

Atai Barkai: I think maybe it's actually helpful to look at the second part. I'm arriving from two directions at the same time. There's the consciousness part, and maybe we can put suspension of disbelief around there for a second. So there's the binding problem, there's this classical describability problem.

QualiaNerd: Even with classical describability, I'm on board with the idea that certain properties of our world simulations are much less describable by our current representation systems than others. The color yellow in itself is less describable by mathematics, or perhaps indescribable by mathematics, compared to high-level equations. I'm on board there. It's just that I don't see that as a fundamental distinction.

Atai Barkai: Let's put a pin in that for a second, because I see what you're saying and honestly I think it is a fundamentally open question, but there are hints that you can follow from a few different directions. Maybe let me get to that part now.

So you have the consciousness side, right? And again, fundamentally the problem here—the "hard problem," quote unquote—the claim here at least is that it's a kinematic problem. There's nothing in the states of the universe as we know it—with some asterisks that I'm going to add here—that is non-emergently integrated, where there's some complex information where the complexity is identified by something other than the fact that it simplifies the analysis of the dynamics.

And to this, again putting a pin in that, but this idea that any state that you cannot write down... and again, everything is a representation of something, but still, you can write down a perfect representation—not the thing itself, but a perfect representation—of the shape of a circle, let's say.

QualiaNerd: Well, what's a perfect representation? I don't know how I would put this.

Atai Barkai: I think I understand your hesitation there, and it's legitimate.

QualiaNerd: Like, when the sheep come home and you say that there are 16 sheep, there's some kind of an isomorphism between some high-level property in your world simulation and the abstract thinking of mathematics, where you are thinking in terms of integers. But you can't do anything like that with yellowness or greenness.

Atai Barkai: Right. I mean, the point is literally just write down on a piece of paper something that allows you to recreate this thing through definition. Because then you could just write down a recipe for making yellow paint, no? We can write about the light absorption of colors, and that is certainly capturing some of the structure, some of the color structure of qualia clusters, right? That captures that relationship part that we talked about.

So I can't describe to you yellowness, but I can tell you that these five things bring a very similar qualia. I can't describe any of them individually, but I can tell you all five of them are very similar. And that's what you get with this recipe. I can make a recipe that, if I apply it to these five elements in my field of vision, is going to reliably encode the same qualia in a single qualia cluster.

QualiaNerd: Well, the yellowness in and of itself isn't describable in terms of any mathematical qualia in a way in which, let's say, the circular shape is.

Atai Barkai: Right. But again, putting a pin in that for a second...

QualiaNerd: I agree with that. It's just that I don't see that as a fundamental issue. I see it more as a matter of creating better and better representational systems. It would really be a weird situation if the current very limited qualia varieties of symbolic thinking used now in our mathematical and linguistic representational systems could describe the entirety of the state space of all possible qualia clusters equally well.

Atai Barkai: I think where we do diverge is that I agree with that sentence, but I think most of the lack is accounted for by complexity. There's a saying that the Eskimos have like 20 words for snow or something.

QualiaNerd: Yeah.

Atai Barkai: The more detail you get, the harder it is to discern. That's the complexity part.

QualiaNerd: Even for simple ones. I think that color is a good example.

The measurement problem

Atai Barkai: Right, and of course, you can extend this to smell and other senses. But then, completely separate from consciousness, let's totally put consciousness aside. Imagine we're just physicists in a physics department.

QualiaNerd: Yeah.

Atai Barkai: They don't care about consciousness. Quantum mechanics is famously problematic, right? We've had this theory for 100 years at this point, and there are still really big problems with it. Fundamentally there are two problems: one is the so-called measurement problem. What is quantum mechanics fundamentally? It's a model of the universe where states of the universe are given by a list of prefixes plus structure.

What I mean is that you have the state, let's say of a quantum coin, that can be up or down. It can also be like "is the cat dead or alive," right? It can be arbitrarily complex structure. Then you have a prefix which is interpreted as what you can derive probabilities from. The model, the Schrödinger equation, tells you how to evolve these list of prefixes plus structure over time. It's just the Schrödinger equation.

When you do that, you can predict what the state of the universe will be in 10 seconds. And you see that these prefixes correspond to the measured probability of getting the structure when you do some "measurement."

QualiaNerd: Well, that's the Born rule, right? The probabilities?

Atai Barkai: Exactly. That's the Schrödinger equation.

QualiaNerd: Yeah, exactly.

Atai Barkai: And the question is: where does this Born rule come from? There's literally 100 years of discussion on this stuff.

QualiaNerd: What's the problem with the self-locating uncertainty in the Many Worlds interpretation?

Atai Barkai: Again, this has been discussed extensively. Let me tell you my problem with it: if you have a weighted quantum coin, not an even coin, but something like 9/10 and 1/10 probability. You throw this coin a million times. The sequence of heads or tails that you get is given by the structure. So you said before there's a prefix and the structure, right? Each possible heads-tails sequence appears in your evolution equation.

From these sequences you derive the probability of the coin. If 9/10 of the results are heads, then you know this is a 9/10 coin. So you get this whole structure. Basically every single probability of the coin exists in your wave function.

QualiaNerd: Right, like the Many Worlds interpretation.

Atai Barkai: Yes, exactly. It's true that if you look at the prefixes of the structures that match the probability of the coin, these prefixes converge to one, and the prefixes of all the other ones converge to zero essentially. You can make this statement mathematically precise. The prefixes of the "unlikely" structures are exponentially suppressed.

QualiaNerd: Right, but they're still in the wave function.

Atai Barkai: Exactly. Literally the Many Worlds says all of these things exist. Some of them have big prefixes, some of them have small prefixes, but all of them are equally real. The fact they're equally real is the fundamental point of the Many Worlds interpretation - that you don't have to add anything more to the Schrödinger equation. Essentially, every probability that you can give to this coin exists in the universe and they're all equally real.

So where does the Born rule come from?

QualiaNerd: But if you're an observer who is trying to find out which branch you find yourself in...

Atai Barkai: But the problem is specifically with the Born rule. It's not just uncertainty, not just "I don't know where I'm going to be." It's not just qualitative uncertainty. It's a very specific type of thing - it's the Born rule. There are papers and papers on this stuff, which in itself is a data point. When things are understood enough over 100 years, they shouldn't need this much explanation.

There's definitely something interesting there. There's all the proof from Zurek and in general, the fact that the prefixes are consistent. There's something consistent about the Born rule in this meta-structure. Essentially you have the probability encoded in two different parts of the wave function. In the coin example, it's encoded in the structure parts, in the relative frequencies of measuring heads and tails.

QualiaNerd: So it's encoded in both the Schrödinger equation and the Born rule.

Atai Barkai: Sort of, yeah. It's encoded both in the structure. If you have heads, tails, heads, tails, you literally have two to the millionth components in your wave function if you do this a million times. Each of them have a prefix and you can calculate the probability. You can essentially calculate the coin probability from the structure. If out of this million, 9/10 are heads and 1/10 is tails, then you'd say, "In this structure, this is a coin with 9/10 to 1/10 probability."

So it's encoded in the structure, and it's also encoded in the prefix. The prefix of each possible world is calculated from the Schrödinger equation plus the coin that you started with. There's an interesting consistency there. I don't think Many Worlds tells us nothing - there's something really interesting there. But this problem is a deep problem. I think even the most staunch adherents of Many Worlds generally say something like, "We don't totally understand how it gives rise to this, but if you do enough complicated linear algebra, then it comes out."

QualiaNerd: I think that's what Sean Carroll would say.

Atai Barkai: Right, and they try to find how to do that. There are more and more attempts, but nobody says, "This is exactly how it happens and we know that." Essentially nobody claims that, I think.

Anyway, that's one of the problems of quantum mechanics. Then there are other things that are not inherently problems, they're just weird. Weird doesn't have to mean there's something we don't understand. It could just be that our intuitions are limited - we're just monkeys, why should our intuitions for what's weird matter?

Einstein, Bell, and quantum entanglement

That's a fair analysis, but to point out these weird things: the most interesting thing about quantum mechanics arguably is entanglement, which was first identified by Einstein, Podolsky, and Rosen in the famous EPR paper.

QualiaNerd: I've read about it in papers but haven't fully understood it.

Atai Barkai: Let me give a quick background. Einstein is kind of famous for supposedly being "too old" to get quantum mechanics, but that's a totally bad portrayal of what actually happened. I think of his generation, he understood quantum mechanics better than anybody in my estimate. It was only after he died, with John Bell, that quantum mechanics was first properly understood.

What Einstein pointed out – he gets flak for saying "God doesn't play dice" as if he thought the world isn't probabilistic – but what he really identified is that you can't have quantum mechanical non-determinism without also accepting non-locality. He was really bothered by the non-locality part, as he should have been.

In the EPR paradox paper, he said, "Quantum mechanics tells us you have systems that can be in a superposition of up and down. It's fine when you have this in one particle, but you can have entangled systems." I don't think the word "entanglement" existed then. But he said, "We can have these two systems with quantum correlations, like spin up and spin down particles where the total spin is zero."

Quantum mechanics says you don't know until you measure – even the universe in principle does not know whether this particle is up or down. That was Bohr's claim at the time. Einstein said, "Let's take these two particles, send one to one end of the galaxy, the other one five galaxies away. You're saying until I measure them, I don't know what they are, right?" He continued, "If that's true, then when I measure one particle at one end of the universe, now on the other end the other particle's state has changed – from being non-deterministic in principle to determined but not known."

QualiaNerd: Right, but that can't be used for communication, because then the parties detecting would have to come together.

Atai Barkai: Exactly. That's the start of the weirdness where Einstein said, "Philosophically, what you're saying is that when I measure one particle, something changes on the other side of the universe, but we can't use this for communication." It's really weird – it really does change, but in such a perfect way that we cannot use this for communication. That's the non-communication theorem. In quantum mechanics, you can show that nothing you do to this particle here, even though it does change the state of the other one, changes the statistics over the states. So you can't use that to send any signals.

If you could use that, problem solved – the whole universe is non-local and that's that. Obviously that's a huge problem with relativity, but we'd figure out a theory where that works. But that's not what happens – you cannot use this to communicate, but it still changes.

In many respects, Einstein's solution was the reasonable one: hidden variables.

QualiaNerd: Hidden variables.

Atai Barkai: He said, "Look, there are two things you can assume. Either there's extra structure inside this particle that we haven't measured yet, which can account for our ignorance of whether it's up or down. In principle you can put arbitrarily complex structure in there – you can put a computer. Who's to say it isn't?" And it could be something simple, just a little bit of extra state.

The second option is, "Oh, the universe is non-local and you have cats that are alive and dead and all this stuff." And he asked, "Which one of these is more reasonable?" I think he was completely correct to point out the first one is more reasonable – there's just extra information we're not aware of. We're really jumping the gun just because our current theory has all these weird features.

That was in Einstein's lifetime, and I think EPR had the right side of this argument in my opinion. But then came John Bell. David Bohm took de Broglie's theory – the de Broglie wavelength.

QualiaNerd: Yeah, pilot wave.

Atai Barkai: Exactly. From the very early days of quantum mechanics, de Broglie first came up with this idea of frequency corresponding to energy – really one of the pioneers of quantum mechanics. He started trying to put together some deterministic theory of quantum mechanics, but he got totally shot down and left it alone. David Bohm picked it up like 20 years later and completed it with the pilot wave interpretation of quantum mechanics.

But then there was von Neumann, who had his hands in basically every piece of science of his time – from computers to nuclear engineering to quantum mechanics. He had this theorem that no hidden variable theory could account for the measured results of quantum mechanics. So you could imagine that there's a hidden variable theory, but it would have to disagree with the predictions of quantum mechanics. That was seen as the nail in the coffin – we can't have hidden variables.

But this proof was totally wrong.

QualiaNerd: That was before Bell's theorem.

Atai Barkai: Yes, before Bell's theorem. John Bell gave some interviews where he said, "Von Neumann's proof was not only false, it was foolish. It was stupid." It had a bad assumption in there. But because it was von Neumann, it went essentially unnoticed for a couple of decades. Nobody really understood it, and people said, "Well, I don't really get it, but von Neumann says so, he must be right."

John Bell first accepted this and said, "Okay, looks like we can't have hidden variables." But then he came across David Bohm's theory, the pilot wave theory, and he said, "Wait a second. I just saw a proof that said it should be impossible, but here's a theory that does this – a deterministic theory that perfectly predicts and perfectly agrees with quantum mechanics predictions." Obviously there was something wrong with von Neumann's proof because here was a theory with hidden variables that agreed with quantum mechanics predictions.

But the interesting thing about the pilot wave was that it has this feature of non-locality – you interact with one system and it affects a system on the other side of the universe. Bell wondered, "Could you have a theory that agrees with quantum mechanics predictions that doesn't have this feature?" He came up with Bell's theorem that shows, "No, actually there is a true non-local theorem that you can formulate about quantum mechanics."

That's Bell's inequalities. Essentially, there's no theory that can reproduce quantum mechanics statistics where doing something to one entangled particle doesn't affect the measurements done to the other entangled counterpart. The bottom line is that these two things are actually affecting one another – what you do to one particle affects the results on the other particle. There's no way to wish that away like Einstein wanted to.

Hidden variables and superdeterminism

QualiaNerd: If I'm not mistaken, doesn't this have something to do with statistical independence, where you could have a hidden variable theory if you abandoned statistical independence?

Atai Barkai: Yes, there's this thing called superdeterminism, which is a whole other thing. It's kind of a can of worms.

QualiaNerd: That the result of a measurement depends on what you decide to measure?

Atai Barkai: Yes. By the way, I'm about 80% convinced that you can actually formulate a theorem that shows that superdeterminism is incoherent, but I've not been able to write it yet. If you can formulate such a theorem like John Bell did, what is it? And if you can't, why can't you? I think there are some interesting hints there.

But superdeterminism basically says the way that John Bell got around this philosophical conundrum was that there are other things we can do to these two particles besides measure if they're up or down. We can choose the axis along which we measure if they're up or down.

QualiaNerd: Yeah.

Atai Barkai: What superdeterminism points out is that if you knew in advance what everybody was going to choose, which axis people are going to choose, then you can bake in the information – you can bake in whether the thing is up or down.

QualiaNerd: Exactly, given the choice in advance. It seems very parsimonious.

Atai Barkai: I haven't seen a version of this that I find remotely satisfying, and I've been looking. I have an open mind – I think there could be versions that are potentially intelligible, but I've yet to see any of them, including the most famous one by Sabine Hossenfelder.

QualiaNerd: Hossenfelder.

Atai Barkai: Anyway, that's a whole other discussion. Every thought here you can double-click on and zoom in and zoom in on. But Bell's inequalities are a fundamental part – as far away as you can get from controversial in the world of quantum mechanics.

So it turns out you cannot think of these two entangled systems as distinct subsystems, which is in itself a very big deal.

QualiaNerd: Yeah, the two entangled particles are somehow fundamentally...

Atai Barkai: They're fundamentally a single system.

QualiaNerd: Yeah, exactly.

Atai Barkai: And there's no theory where you think of these as distinct systems that agrees with quantum mechanics. Again, with the asterisk being that it's just statistics that have to agree.

That opens the door for the superdeterminism stuff, but I think there's an 80% chance you could formulate a theorem showing why it's incoherent. But let's not get into that.

This circles back to consciousness, almost. We're almost there. The thing still missing is – so these things are a single system. Yet there are a bunch of interesting things about quantum mechanics.

First, a qubit naively seems like it can represent a lot more information than a bit, right? A bit is just 0 or 1, and a qubit is 0 or 1 plus a prefix that can be a real number, like 0.63. So you'd think you can capture a lot more bits per qubit. But actually that turns out to not be true. There's this thing called Holevo's theorem which shows if you want to decode your information out of your qubits, you can only encode and decode one bit per qubit.

QualiaNerd: Extract?

Atai Barkai: Yes, extract only one bit of information per qubit. If you want to store information in a qubit or in 100 qubits and extract it somewhere else from the same system, you can only store and extract up to one bit per qubit. That's interesting but just curious.

But then there's this thing called superdense coding. If you have just one qubit system, you can only store one piece of information in it. But if you have an entangled system of two particles – individually you can store one qubit per particle, but these two particles are entangled – it turns out you can do something to one particle, send just that one particle to its entangled counterpart, and you can recover two bits out of these two qubits.

So you can encode two bits in the one particle that you send over. You can have four states, like a two-bit system: 00, 11, 01, 10. You send one particle across that individually can only store one bit, and yet you recover two bits on the other side as long as you have its counterpart available.

It's another kind of qualitative weirdness of Bell's inequalities. You really do change the state of the universe at the other end, but you do it in such a way where you can't send information.

QualiaNerd: If I understood correctly, this applies to entangled particles when you send this particle... Do you have to do that faster than the speed of light?

Atai Barkai: No, you can't use the entanglement itself to transmit information, but if these two particles were sent entangled, then by sending just one particle across you can encode two bits in it, even though if it wasn't entangled, you can only encode one bit in it.

QualiaNerd: Right, like if you send particles then you can send two bits.

Atai Barkai: Exactly. The two-qubit system is a two-bit system, and you can interact with the full two-qubit system with only one of its ends, only one of its qubits. So essentially this qubit, the single qubit, is kind of this gate into a two-qubit system that you can manipulate. You can literally encode and extract two bits of information out of it.

So it's really weird. Again, there's nothing wrong here, unlike the measurement problem where the theory doesn't tell you how the universe behaves. With probabilities, you want to get probabilities out of this theory, but nobody really knows how to get that.

The second issue is just weird. You do something to one particle, it affects the other side of the universe, but it does it in such a weird way that you can't use that fact to transmit information. In quantum foundations, some people analyze this as a fine-tuning problem – it's weird that you can literally affect something without it being detectable. There's only a small space of possible theories that would allow for this.

QualiaNerd: That is interesting.

Atai Barkai: And then in addition, to transmit information, you can interact with a full two-qubit system from only one of its ends.

QualiaNerd: Right, but respecting the speed of light.

Atai Barkai: Exactly. You have to send the particle to extract that back out. So there's just weird stuff. There are a bunch of other things of that nature where if you try to form some qualitative understanding of what's going on, you keep coming up short.

Again, that's not inherently problematic – maybe we're just monkeys, why should our intuitions matter? But it still feels uneasy, and everybody who studies this in depth feels uneasy, and this unease does not go away. You feel there's some formulation that we're just missing that would make the puzzle pieces click.

And that's again just a pure physics problem. In quantum foundations conferences where nobody cares about consciousness, people talk about these things. And what I claim is that there's a unity here.

Connecting quantum mechanics and consciousness

So you can forget consciousness for a second, just think of physics again. A mathematical physicist could have discovered this from scratch in principle. It's just weird, so it's easier to come at it from two different angles at the same time. But let's just imagine a structure that can exist under some theory of physics that is only partially isomorphic to any sequence of classical bits. Colloquially, you could say it's partially classically indescribable. And it's a weird thing to presuppose. Which is why nobody has, as far as I know. But it's not weirder than the square root of -1.

QualiaNerd: Right.

Atai Barkai: It's kind of like the information theory parallel to complex numbers. Honestly, somebody gave me this parallel just the other day. I didn't think of it myself.

QualiaNerd: Was this the person Andrés put you in contact with?

Atai Barkai: No, it was somebody who doesn't have that much physics or mathematics experience, but is a smart guy.

QualiaNerd: Got it.

Atai Barkai: Anyway, you can imagine this thing and start to play with this idea. What if you had a system that was only partially isomorphic to classical information? The classically describable projection of that system—the classical describable substructure—would have to be operationally indeterministic even if your universe is fundamentally deterministic. Why? Because if you have two systems that according to all their classical describable parts are the same, they still actually have distinctions under the hood.

QualiaNerd: But this is actually interesting because this perhaps permits the possibility that if we developed better representational systems, maybe then quantum entanglement could be used for communication.

Atai Barkai: That's the second part here—trying to make sense of this weirdness where it affects the other side of the universe, but you can't use that to transmit information. And yet if you have access to both of these things, you can extract information out of it.

So again, putting this in the language of classically describable information, imagine a system where you can extract information from the relationship between two individually indescribable pieces. You can say they are the same or they're opposite, but you can't characterize them individually. Of course you can't use this to transmit information because literally the substructure that you're affecting is the classically indescribable substructure.

So of course you change the state to the other end of the universe—it's really laid out there exclusively—you're changing the classically indescribable substructure of the system. And of course, you can't use that to extract information because it's classically indescribable. The fine-tuning problem goes away. And then of course, if you have both of these ends together, if you can measure the relationship between them, then you can extract two bits out of the thing.

If you think of this in a qualitative way, imagine you have this structure in the universe—I'm not saying anything about consciousness—where it has two parts in this two-qubit system. The parts individually are indescribable. I cannot write down "this one is seven and this one is nine." I take that as a constraint. It's a new type of structure, just like the square root of negative one. But the relationship between these two things is classically describable.

QualiaNerd: They're the same or they're different, right?

Atai Barkai: Exactly, how they're different or the same. That's my claim, though it's not fully finished—there's more math to work out.

Classically indescribable information and entanglement

QualiaNerd: That's actually quite interesting. So with regards to this classical indescribability, would you say that the situation with the entangled particles is like that of colors?

Atai Barkai: Basically. My claim is that now that I have this universe—which you could have arrived at just from pure math in principle—that agrees with all the predictions of quantum mechanics, I no longer have an issue with the kinematical problem of consciousness. I was missing a thing that was fundamentally integrated—integrated by something other than simple fish fusion of analysis.

QualiaNerd: That's the entanglement.

Atai Barkai: That's the entanglement part. And it supports classically indescribable information. That's literally one of the only elements in my theory of physics that I came up with to explain quantum mechanical observations.

On both the consciousness side and on the physics side, there's this really interesting interaction between indeterminism and non-locality. On the quantum mechanics side, that's what Einstein first noticed. They really come together. If you had only one or the other, you'd be in big trouble. Somehow the indeterminism is what enables the non-locality.

And on the consciousness side, it's also like you have these things that individually are indescribable, but the relationship between them is describable. That's what allows you to encode information over a substrate that is made of partially indescribable elements.

QualiaNerd: I'm thinking about what the parallel would be when it comes to the entangled pair of particles itself. Would you say something like the world simulation or the qualia cluster that is a fundamentally integrated world simulation instantiated by a human brain would also be like a pair of entangled particles? A fundamentally integrated system?

Atai Barkai: In my view, qualia clusters are almost fully decoupled from cognitive ideas such as world simulations, except in the sense that they can be used to implement—they can be used as the substrates for implementing—cognitive systems including world simulations.

The information that's encoded in your qualia cluster right now is deeply isomorphic to whatever information your senses absorb and what you're thinking about as a cognitive agent, with memories and emotions and all these things. But in principle, it could contain any information. The mechanism isn't there to allow for that because it's all driven by evolution for a purpose.

QualiaNerd: So the idea is that the classically indescribable substructure of the qualia cluster could be different and yet they would be carrying out or implementing the same cognitive functions?

Atai Barkai: Yes. You can imagine two qualia clusters that have exactly identical classically describable substructure—it would be sort of like having two identical pictures, like two people looking at the same picture, or you looking at the same picture now and in five seconds.

The classically describable substructure is identical. If you ask this cognitive system how it behaves, it interprets it identically because all the useful information there is the same. But they actually are distinct.

QualiaNerd: The classically describable information is the same.

Atai Barkai: Yes, the classically describable is the same, but the classically indescribable is distinct. And the way that this would show up in a physics experiment is that some of this behavior is indeterministic.

Which is funny—if you go into the free will stuff, it's kind of against free will in the sense of being non-deterministic. This says it just appears to be indeterministic. It's not actually indeterministic.

QualiaNerd: Right, it appears to be. So regardless of whether or not the universe is actually indeterministic, it will appear to be indeterministic because—assuming that there's this classically indescribable substructure—you can have two systems where every information that you can write down about them is the same, and yet they are actually different.

Atai Barkai: Exactly. And so they would behave differently under experiment. The claim is that in a universe that has these classically indescribable elements, it would appear operationally indeterministic, whether or not it were fundamentally deterministic.

QualiaNerd: That's interesting. I'm still trying to think about this. When you have a pair of entangled particles—another thing I'm thinking about is indirect realism. Whatever you detect on your screen of perception is of course not something in the external environment, but something internal to the brain. Does that also play into this somehow? Maybe a lot of the confusion even in quantum mechanics comes from not realizing indirect realism enough.

Atai Barkai: Can you expand on this? I don't totally follow, but maybe if you keep talking, I will.

QualiaNerd: Well, I've talked about Bernardo Kastrup. And he says something about quantum mechanics, but I don't want to misinterpret him. It's been a while, I'd have to go back and read his articles where he described this. I don't have perfect memory unfortunately.

Atai Barkai: I have to read it again too. It's been a while since I listened to him the last time. He has some interpretation of quantum mechanics as well.

QualiaNerd: I mean, I think it's largely a different abstraction layer. Of course you have to take all that into account. If we are talking about transmitting information, what we are really focusing on is the structure of the world simulations.

Atai Barkai: Yes, that's exactly right. In quantum mechanics, there's also this big discussion of operationism versus what you could call ontologicalism, where you say a theory of physics only tells you how things behave.

But consciousness is kind of the catch-all there because eventually all your measurements are, as you point out, not actually measurements of instruments in a lab. You need your theory to account for the fact that qualia clusters have to exist in your model of universe. Otherwise, you could come up with some theory where we're all just deeply insane and confused—and obviously a lot of people have. But if you swallow that, there's really no end—you can say that about anything. You can say we know nothing, we're totally delirious.

But if you don't want to say that, then your theory has to contain this structure in it that cannot be abstracted away. It's the only thing that actually has to exist.

QualiaNerd: Like the self-evident existence of consciousness.

Atai Barkai: I'm not even going to say the "self-evident" part, I just think it's an internally coherent story. If you have a coherent story, I think a lot of these questions will go away.

My theory makes experimentally testable predictions, so it's not just words. You can actually go and test these things, which is why it's not just an interpretation. These predictions would be different from what the Many-Worlds or Copenhagen interpretations would make.

The most simple prediction is that the human brain facilitates complex entanglement, which is from a naive physics perspective extremely unexpected. That's the same point I think that David Pearce makes.

QualiaNerd: But then how do you solve the boundary problem? Everything is entangled with everything else, right? It's just a matter of degree.

Atai Barkai: Actually that's not exactly right. That's a misconception. There's actually something called the monogamy of entanglement. Essentially what it says is that if you have two qubits that are maximally entangled with one another, then they cannot be entangled with anything else.

In a real system, it won't necessarily be maximally entangled, but it's not a resource that you can just go around infinitely.

QualiaNerd: So there are hard boundaries there with this entanglement?

Atai Barkai: Essentially, yes. If you have two qubits that are maximally entangled with one another—which means the statistics are such that they're perfectly correlated or anti-correlated—then you know for a fact that there's no other part of the universe that has the same entanglement relationship with them at all. And that extends beyond the two-qubit system. That extends to a million-qubit system also.

QualiaNerd: And how short-lived would this be? I imagine that in the brain it would be extremely short-lived.

Atai Barkai: Well, that's the criticism against this idea, right? Max Tegmark, whom I deeply respect and who is one of my favorite physicists, has this calculation that a brain is too hot and wet for quantum computing or for quantum entanglement. I think that's really bad science that set the field back, similar to von Neumann's proof—a really bad proof from a really authoritative figure.

Nobody's claiming that it's just meat. Basically the calculation assumes it's just meat at a certain temperature, made of a certain material, and then calculates the quantum coherence duration you should expect. But if you do the same calculation, then no brain should be able to talk or walk or build iPhones. Obviously these are extremely intricate machines designed by evolution that carry out incredibly complicated operations at every level of the stack—from the macro to the cognition to the blood flow to the cells themselves.

QualiaNerd: I wasn't saying that as a criticism. I was just asking if it would be short-lived.

Atai Barkai: It's not necessarily short-lived. I think that's where David Pearce and I diverge, but that's a point I've wanted to clarify with him for a while.

The way I understand it, he says that the information we think of as encoded in a qualia cluster will be found in entanglement relations. I think he expects it to be found in a second-time-scale thing. What I don't get is if that were the case, then how could that be causally significant? Which I think we both agree it would have to be because of the meta-problem constraint—that's how we can talk about it. If it was just there but wasn't causally significant, that couldn't correspond to the truth.

Causal significance of consciousness

QualiaNerd: But if consciousness was the ontological substrate from which the fields of physics were made, then they would be causally significant, right?

Atai Barkai: In a sense yes, but you don't want them to just be that—that's the panpsychist view where they're technically causally significant because everything with cause and effect is embedded in the structure. But you want the causal significance to appear in the structure.

Whatever is embedded in these entanglement relations—the world of particles, like in brains and all that—that's the world where we speak about consciousness. It's not qualia that speaks about consciousness. It's brains and tongues and muscles that are speaking about consciousness.

QualiaNerd: But if the brains and tongues and muscles are all made from qualia clusters, what are you saying? That the same dynamics would be implemented, or even perhaps the same kinematics could be...

Atai Barkai: You can imagine that your underlying substrate has these qualities, but the emergent dynamics that describe the classical world we live in lose that part. Just like in a computer simulation—you can have an Nvidia GPU and put a simulation of a two-dimensional universe on it. The fact that it's implemented on top of some microchip architecture is totally lost to the simulation. It doesn't make it into that dynamics.

QualiaNerd: I'm worried about getting lost in abstraction here. What exactly do you mean by that—that it does not get translated into the dynamics of the simulation? It depends on what exactly you mean by the simulation.

Atai Barkai: I just mean the embedded regularities. You have the underlying regularities and the underlying kinematics. But then you have the imagined kinematics and imagined regularities.

QualiaNerd: That are a coarse graining, right? If this picture is true, then it's a coarse grain. It's not really corresponding to reality, but it's still a valid coarse graining.

Atai Barkai: What you need to happen is that if it were the case that the coarse graining did not contain evidence of the substrate's kinematics, then that would not be the universe we actually live in. That would describe a universe where consciousness exists, qualia clusters exist, but nobody can talk about them. There'd be no hard problem, no problem at all. These terms would never be coined.

QualiaNerd: You would expect that people wouldn't, for example, say that they are puzzled by colors.

Atai Barkai: Right. They might think that in some weird way—the confusion might be encoded, but that confusion would never leak out into the classically describable part, into the world of particles—like fingers typing on a keyboard in a philosophy journal. That causal structure is closed.

I think that's the most critical observation, to some extent much more critical than everything else in my thinking, because it transcends the particulars of any theory. It means there has to be some physics of consciousness because we can literally talk about it.

The world of physics—Newtonian mechanics and chemistry and all that—describes how right now I'm talking, how my tongue is moving. There's some link between that world and qualia clusters, as evidenced by the fact that I'm talking about qualia clusters.

If you could trace this causal structure, like debugging a computer program, somewhere in there you must find either that there's no such thing as qualia clusters, which would be deeply unsettling and confusing, or that they exist and participate in this causal structure. And then the question is: how could that possibly be? My theory is one possible way that could happen, and it has certain experimental signatures.

QualiaNerd: It's not just that they participate in it—there's actually another point. That they participate is already entailed by panpsychism or by David Pearce's theory where everything is caused only by consciousness.

Atai Barkai: Exactly. Participate and leak.

QualiaNerd: There's an extra constraint here, which is that the underlying kinematics somehow show up in the high-level description because we are making a high-level description referencing it. That's proof that it happens.

Atai Barkai: Exactly, which is what I claim rustic panpsychism cannot accommodate. But that's a slightly different thread, though I think it's actually the deeper point.

QualiaNerd: But this theory is compatible with panpsychism as well, right? It's just a more strained panpsychism.

Atai Barkai: It is compatible, but now it's a deeply scientific panpsychism in the sense that the question of whether it's true or not is turned into an experimentally testable question. You have this signature.

In principle it could be true, and honestly I think it's more likely than not. I find it more elegant. It's more parsimonious for the universe to contain only one type of thing, and there's one type of thing that we know has to exist, which is qualia clusters. So it'd be nice if everything's just made of that.

On the other hand, there's Shakespeare's "There are more things in heaven and earth, Horatio, than are dreamt of in your philosophy." I think that's also true. Consciousness itself is almost the perfect example of that because even though it's the most familiar thing we have, it's still deeply confusing. Who could have thought of classically indescribable information?

QualiaNerd: That's interesting. With regards to this classically indescribable information, to clarify a bit more—the claim is that certain very simple structures, like yellowness for example, cannot be described by our current mathematics or physics or the qualia varieties of our current representational systems in the way that the relationships between these classically indescribable qualia clusters can be.

Atai Barkai: That's right.

QualiaNerd: But that doesn't preclude the possibility that—similar to how you couldn't describe quantum field theory with only integers, and integers are also a particular type of qualia variety—the qualia of mathematicians 3,000 years ago were different from the qualia of mathematicians today who think about complex numbers.

Atai Barkai: Wait, I'm really sorry. I noticed my phone is at 2% battery right now, so I don't want it to cut off.

QualiaNerd: No problem. I'm available every day more or less at this time.

Atai Barkai: Maybe we can chat tomorrow. It's Memorial Day weekend in the United States now.

QualiaNerd: And if Andrés could join, that would be great.

Atai Barkai: That would be great.

QualiaNerd: Thank you. That was very informative.

Atai Barkai: Likewise, great discussion. I really enjoyed it.

QualiaNerd: Thank you.

Atai Barkai: All right, I see there's a bunch of folks listening. Thanks to everybody for tuning in.

QualiaNerd: Thanks as well. Bye.

Atai Barkai: Bye.