J. R.-S.
Why have you been interested in animism recently?
Bach
I noticed that a lot of other cultures are animist, and then I noticed that our culture is in some sense the only one that is not animist. I was basically growing up in a world which sees animism as superstition, and also it’s possible that it’s a superstition where it is basically a belief in ghosts. The difference between living and dead nature is that somehow there is this vis vitalis animating power that exists in part of nature, and somehow seems to violate physics, and physics and biology don’t have a need for such theories. We try to explain reality through mechanisms. And in this mechanist world there doesn’t seem to be a place for spirits that float around between the organisms and so on. So very weird set of beliefs.
On the other hand, there is this nagging suspicion that these cultures which are animist, the members of these cultures don’t seem to be intrinsically all dumber than ours. Right? So for many thousand years since there’s recorded history in these different cultures, they all seem to converge on a set of beliefs that is somehow different from mine. And when other smart people have ideas that are different from mine, it’s easy to think that they all have broken epistemology, that they don’t know how to make sense of reality. But it’s also possible that you notice that maybe you are the one who is to do that. So maybe you try to get to work how they see reality. So basically, another person has ideas that you don’t share, you should go in and try to understand what is it exactly that they mean.
And so I asked some people, “What do you think does animism mean?” So, for instance, if somebody in Japan says that they are an animist, what does this Japanese person mean by this? And I got the response: Japanese animists believe that everything in the universe is alive and conscious. And I said, “Wait a moment, this cannot be the case.” Because I believe that Japanese people know that the person under anesthesia is not conscious, or a person that has died is not alive. So they probably won’t mean to say that everything in the universe is alive and conscious, except of an unconscious person or a dead person. You just mistranslated the term. The word means something else, you just translate it into our cultural context, in which the exact words that the animists are using to describe reality mean something else. And so you basically lost something in translation there, that might be the crucial thing that they are pointing at.
And I filed this back in the head, in my back of my mind, until it made sense maybe at some point later in the future. And in my work as a computer scientist who tries to use the concept of AI to answer philosophical questions about the nature of mind and reality, I noticed that software is something that philosophers don’t look at very much. And software is kind of magic, if you think about it. It’s a causal pattern written in a language that is not the same language that we use to describe what happens in the transistors of your computer. Software does not break physics, but it’s able to control physics. Once you establish software, it’s a causal pattern that is determining what happens in reality to some point, and you can try to describe it via electricity and the transistors, but it’s not quite the point. And if you use different charges, use different electrons, different transistors, the software is still going to work—as long as the transistors have the property that they can sustain the software, right? So there needs to be a substrate. It has somewhat general generic properties, and the software is the specific stuff that runs on it.
And so there is a similarity to the way in which my brain is working. There are these activation patterns in my brain, and it doesn’t really matter which particular neurons are active in my brain, right? If a neuron dies, my mind can recruit a new neuron, and then train it with the necessary patterns and causal structure to run my mind on it. And my mind very much has the properties of what people call a spirit. It’s agentic, it’s capable of experience, right? And it’s a software that runs on the brain. My brain is probably not agentic and capable of experience. It’s my mind—that is, it’s me. It’s this representational pattern existing on it.
And then I stumbled on a different thing. (Sorry for talking so long; you can interrupt me any time.) So that is, it’s not just neurons that can process information, but every cell can send conditional messages to neighboring cells. And so when I, as a computer scientist, look at a multicellular organism, what I see is a Turing machine: a system that can process arbitrary information in arbitrary ways, as long as it finds some kind of agreement on how to do this via message passing. And so, under evolutionary conditions, a multicellular organism should evolve an architecture that allows it to generate self-organizing software. Right? So of course the software that exists on my brain and the software that exists in the cells of my body that talk to each other, they need to be self-organizing. And to be self-organizing, they need to find a way to work like a government—that is, basically training the individuals that participate in the organization of a society to speak the same language, to use a shared reward architecture, to pull in the same direction, to implement the same agent. And agency here means the ability to control future states: to perform control tasks that go beyond simple chemical reactions that happen in the present.
And so I got to this position that everything in nature seems to be running software, if it’s alive. And the thing that distinguishes living from dead stuff is self-organizing software that is agentic is running on the living stuff. And so I call this position “cyber animism,” but it’s tongue in cheek, because it’s not actually different from animism. It’s simply a reformulation of animism within the constraints of the terminology of our own culture, in which we basically discover—all the computer scientists have discovered what it actually means to be a spirit. Not in a metaphorical way, but very real. There is spirits running on the computers that are not agentic and capable of experience. But in order to run on biological brains they need to have self-organizing properties which make them agentic, and probably also require that they measure their relationship to the environment in such a way that they can become capable of experience.
J. R.-S.
When you make this analogy between spirits, and then software, and then think about how our brain is in a way not exactly identical to the mind—because, as you say, I mean, imagine just after you were to die, then a lot of your brain might still be alive in a sense. But there’s like a critical point where it’s like not connected. It’s not self-organized, you might say, in a way—
Bach
Because it loses the order. The software they’re running on it crashes. Right? And so the individual cells are still there, but they are no longer organized into a brain. And when your entire body dies, right, and you’re not just brain dead, what happens is that the organization of your body dies. And so it’s the individual cells, they partially survive. They try to head off to the hills, and some of them even make it. Some of them can survive the death of your body. But the organism itself is gone. And it’s important to realize the organism is not a physical thing. It’s not an object like an electron or something. It’s something that is a function that describes the organization between the cells. What you’re actually looking at when you look at an organism is a few trillion cells that behave as if they were one thing. And this behavior as if they’re one thing, this coherent behavior, is the result of a causal communication pattern between these cells—and this is a software. And so it’s not an analogy, it’s literally the same thing.
J. R.-S.
It makes me think about the start of life and abiogenesis, and then where does this… at what point is it self-organizing enough such that we can say that there is a spiritual software agent that’s now self-organizing it, or maybe it was always the case from the—I’m just trying to think about, like, when we think about patterns that self-organize matter, and that is what the organism is, not the matter in biology exactly, I’m trying to think about, like, is that pattern a thing in and of itself? Like, does it exist before it and then come down to embody the material? I’m just trying to think through it.
Bach
Yeah. Take money as an example. And money is also a software—money is not actually a bunch of coins. It uses these coins as a substrate, but it does not rely on the particular coins. It can use different materials, they can use computer accounts instead of pieces of paper, or pieces of metal, or whatever you use as your means of expressing currency, and so on. It’s also not actually currency, it’s also the regulation of currency. So, in some sense, money is a software that, once it’s invented, becomes real. And it starts as a human invention, as some kind of convention that exists in the beliefs of people. But once enough people start believing it, it becomes real, right? You can no longer explain the world without money once money is invented. And it also doesn’t require people to believe in it anymore. You can run it with computers. The computers can play a stock market. And regardless of what people believe about this, it’s still working. Right? And so money is a pattern in reality that has causal power.
And you could ask yourself, “Okay, does this mean that money is somehow ghostly?” But it’s not more ghostly than a sound wave, right? A sound wave is also a pattern in molecules in the air. And is a sound wave less real than, say, a photon? Maybe it is, but it could also be that the photon can basically be understood as some pattern in the ether, as something that is underlying the structure of the photon. But for the sound wave, you can basically look below the sound wave, we can resolve molecules below the sound, and we can zoom in so much that sound disappears as a concept. In the same way as we can zoom into the world until money disappears at a concept. Right? You can only see this when you coarse grain the world; when you go out and project it onto some coarser surface where you can see this causal structure of this pattern. But when you zoom in, we cannot zoom in basically below the particles that form us; the elementary particles are the stuff that composes us and our measuring instruments. So we have difficulty to go below it. And we can make theories below it but we cannot directly observe it. But it still makes sense to assume that the photon is not somehow the most elementary thing that exists. Below the photon matter, in our current theories, there are quarks and then there are probably patterns—for instance, in the quantum gravity you have informational patterns that could use photons. And so, in some sense, everything in physics can be understood as dynamic, persistent (or more or less persistent) causal patterns. And so everything in this way is seen as information that is happening and undergoing transformations in regular ways, and physics is the way to talk about transformation in information that we can measure across multiple dimensions.
J. R.-S.
Like the—
Bach
Yeah. So it’s all software, in a way. But the further you zoom out, the more complicated and ephemeral the patterns become. And so society is also a software, you could say. And this question that you asked—at which point does it start?—is a very interesting and important question. At which point does the spirit of an organism start? You can ask yourself: at which point does the spirit of a society start? And a society is ultimately composed of human individuals that talk to each other, and have beliefs about each other, and establish norms about how to interact in the future, and expectations, and so on. And it’s a gradual process, but there are some principles that turn a group of people into a society. For instance governance, where you basically establish institutional structures that make sure that the society gets a degree of stability and regulation. And this government is itself software, right? It’s a set of rules that society gives itself.
And I think it’s a very good analogy to think about how society starts with a few people who decide that they want to be the seed of a new government, that they want to have a new structure that is growing into a larger thing that scales and tells people how to behave at scale, and when new people enter it can entrain them with this pattern of interaction, and make the society real. In the same way, if a tree is growing, it needs to start with the seed, and the seed starts out from a single cell. And the single cell is an idea on how to subdue the chaos in its environment, and then assimilate nutrients until it can divide itself, and then has multiple cells that obey the same rules. And these cells talk to each other and specialize until they become a tree.
J. R.-S.
Okay. Yeah, I like this. That the software patterns that help bring together and shape organisms, but then why treat a photon as a single entity? What level should we look at that? Or bring it to money and societies, civilizations and, and—but there’s also interestingly a difference between, as you say, it’s a bit vaguer when you think about the patterns in society. And you can still think about it maybe in the same way. It has an origin state where some people have the idea, the seed, in their mind to sort of instantiate and embody the patterns into the society. So you need both. But I wonder, tell me: do you have any thoughts about biological organisms and how they embody such software patterns, versus high level group structures and different coherency, and sort of… yeah, how embodied and real are they? Are they equally real, these patterns in different levels of organization?
Bach
You talk about yourself here, right? You are an organism, or you could say you are a reflection of what it would be like if you were one. And what actually exists according to biology is a bunch of cells that talk to each other. And you are a pattern in this communication between the cells. And the reason why you say, “Oh, there is a tree”—what is the unit of tree-ness that makes you draw a boundary about around the tree? I think there are two types of reasons for this. One is: it allows us to make sense of the world. So it’s a good data compression. By separating the tree from its environment, you get a good model that allows you to predict that you can remove the tree from its environment without the environment changing very much beyond the removal of the tree. Right? And maybe you are able to plant down the tree somewhere else, and it takes root and continuous growing somewhere else. And then you divide the tree in the middle and say it doesn’t work quite the same way. Right? So there is some kind of natural boundary that depends on how you can interact with the world and how you can perceive the world. So we try to make sense of the world in terms of what we can perceive, and in terms of how it allows us to control reality. This is one side. This is the side of us, the observer, that makes models of reality, and so we draw these boundaries around what makes sense to us, what helps us to make sense of it and control it.
But the other one is: the tree itself has this particular shape because it’s the result of a certain causal pattern that starts in a single cell, that becomes a seedling, and then becomes this tree structure. And so it’s the software of the tree itself that makes the tree a tree. The tree itself is the result of a control process. And when you look at the world, you discover the traces of these control processes. So the boundary of the tree is also given by the way in which the tree works itself. There is something that makes the tree specific and repeatable that makes one tree of the same type similar to another tree of the same kind, of the same species, and so on. And that is, basically, they run similar programs. They run similar software that is related.
J. R.-S.
Whereas then comparing that to something like an economy or high-level organization, it makes me think that there is perhaps no easy way to cut a line through it and destroy the self-organizing patterns. Whereas in a tree you see it as a whole organism, and you can know, okay, this can be destroyed, and there’s a wholeness to it. Do you think that’s right? That then something that [???] is different in that way?
Bach
Yeah. You have to, of course, realize that the concepts that you’re talking about are concepts in some kind of representational language in our minds. So they’re not the physics outside of us, because we cannot experience the world outside of us directly, or get a grasp on it directly. All we get is patterns that your perceptual system is extracting from reality through the learning in your mind, and the generalization in your mind. And some of these patterns are about language that other people speak, and this allows us to share these concepts that we have, and to converge on shared conceptualizations of reality. But this makes us a collective mind, to some degree, where humanity is a collective modeling system of reality. But the objects that we’re talking about are not objectively there, they are there from the perspective of this modeling observer. And when you find better concepts to model reality, then your perception of reality changes.
And so animism is an excellent example to see this difference. In every culture there are some people who can see ghosts and other people who cannot see ghosts. But in some cultures the people who see ghosts are embarrassed, in other cultures the people who cannot see ghosts are embarrassed. This is the difference, right? Do ghosts exist or not? That’s a weird question. So there is something that people can conceptualize as a ghost, and model as a ghost with some degree of reliability. And so there is some ghost that is running on you. Right? You have a particular kind of spirit that makes you recognizable to others. And when the spirit crashes, the software crashes, then the organization of your body crashes, and your body is going to turn into mush. Right.
It’s of course the question: is something retained outside of your body after you die? And this is something—it’s an empirical question. You would have to measure this. You would have to measure: is it possible for the organization of your mind to persist when the organization of your cells is disappearing? Right? Similar question, maybe: what happens to the social structure of a society when the society crashes? Right? So when the Roman civilization died, what happened to the Roman civilization? Well, it didn’t completely disappear. There is a ghost of the Roman civilization in the environment that is informing things that we do today. Many of the words that the Romans invented made it into the future, and turned into concepts in the English language, and too in the Anglo-Saxon culture, for instance. And when the U.S. was designed, it was designed largely based on influences that people had when they read books that describe Roman society. And so you could say that the ghost of Roman society is still alive and still around. But of course it doesn’t run on people anymore, in the sense that it can control the region of physical reality. So it’s only here in this disembodied way. Right?
J. R.-S.
I wonder if, then, the ghost of Rome, or of anyone, if there was—at least before the spirit crashed in one’s organism—a spirit, then can that be re-embodied? Does that spirit still exist? I just wonder almost in terms of metaphysics, does the spirit exist, then, outside separately, and then so…? Or—
Bach
Yes, some spirits can. So for instance, if you take a program on your computer that you don’t like anymore—like, maybe you don’t like VS Code very much and you delete it on your computer. And at some point in the future, you decide you want to have it again. Right? Then you can basically get a recipe to re-embody VS Code on your computer again, download it, and there again it is. But this works precisely because VS Code is designed in such a way that the ingredients that are necessary for reviving it can be stored in such a way. And this is also true for some organisms, right? You can take the blueprint of an organism, it’s stored in the spore, and the organism dies, and at some point in the future you revive this spore, and you get an organism that has the equivalent structure. And then, if you want to revive yourself on somebody else, of course, you cannot get the stuff that depends on your particular substrate, on your particular biography—in the same way as you cannot get the same plant or the same mushroom by activating one of its spores. What you get is something that is in the same way as the original one, also a function of the quirks, the things that led to its genesis. So for instance, if you don’t have the same soil, the same biography, the same random ingredients that went into it, you’re going to get something that is slightly different.
And so if you manage to—say like the Dalai Lama—to reincarnate on a different human being after you die, you’re not going to rescue the personality traits of the human being that the previous Dalai Lama was running on. What you can get on the new being is the belief that you are an institution of governance that is a direct continuation of the previous ideas, of the previous Dalai Lama, to the degree that you can recover it from the teachings of the previous Dalai Lama that made it in his diaries and in books and to his advisors, and that you can then learn in the next life. Right? So you need to dis-identify from your mortal parts, from the parts that depend on the vehicle that you cannot retain from the substrate that you cannot reconstruct, and you can only identify yourself with those things that you can reconstruct. And that’s easy for VS Code and it’s much harder for people.
J. R.-S.
Tell me more about the analogy between a computer—like your laptop and you’re deleting VS Code—versus the software on, say, an organism. How perfect or imperfect do you think the metaphor is?
Bach
I think it’s a perfect metaphor. It’s not a metaphor—it’s the same thing. But they are two different types of software. The software that we have on computers is designed to be easily designable by people. So it’s written in such a way that people can understand it sequentially, and to make that possible that we can reason about software with our feeble brains, those computers that are completely deterministic—or so deterministic that we can ignore the remaining indeterminism. And our computers, of course, physical objects, so there is noise in them. And to some degree there’s a probability that one of the transistors fails. Maybe not even permanently, but in the sense that a bit in your computer randomly flips. And we build some error correction into the computers that make them robust against this to some degree. And it doesn’t need to be perfect. If your computer only crashes once in two years due to one of those bit flips that you don’t error correct for, maybe that’s good enough.
Also, you can tune your hardware to such a point that it becomes very, very reliable. And that allows you to build software that itself is not error correcting, that does not deal with the fact that its substrate is random. But the more random your subset becomes—for instance, when your internet connection becomes more noisy—you need to make more error correction in the system so it doesn’t crash. And your mind is running on a subset that is extremely noisy. Because neurons are noisy, right? Sometimes they don’t want to do what you want them to do. And sometimes they get too hot or too exhausted or too hungry to do your bidding, or they simply die, or they get diseased and… right? All this is something that our mind has evolved to adapt with. And so it’s multiple layers of very powerful error correction that our mind emerges on.
And it’s also a second-order design. It’s basically not a direct implementation of the structure of your mind, but something that wants to grow into it. The same way as you want to have a tree, you cannot just start with a small tree, you need to have a seedling, or that before was a seed, and the seed is not a seedling, the seedling is not a tree. It’s something that wants to become the next stage. Same way as you want to build a new society: you don’t start with the big society, you start with something that is a revolutionary movement. That is very different from a society. Or when you start a new religion, you need to get a small group of apostles together that are trying to spread your aesthetics. And these group of apostles is very unlike a society. Right? It’s something that slowly transforms society, and there are going to be multiple stages until the whole thing is implemented.
And so the kind of genesis of the software in biological systems is very different than this software that you write in your computer, where you start in the beginning and end at the end, and write it down like a story. Often you have to edit your drafts and go back and so on. But still, our current human software development process is something that is the result of how we use tools and how our minds were by using tools. And the software that exists in nature is not written like this. It’s more grown. Right? So it starts out with the seed for the software that is then adding complexity to itself.
J. R.-S.
When we think in our modern day about then creating artificial intelligence, do we want it to be like an organism which has a seed which then changes a lot, and it’s not the thing it is going to be at the end? Like, so they’re both software agents in a sense if we want to create AI, but we could use computers, we could use biology, and those examples, or are there different examples? Like, what should we—could artificial intelligence look like… what should we want it to look like?
Bach
So there are different reasons why we build AI. So, for instance, the large AI companies built AI because they are economically useful. And to make them economically useful they should be predictable and controllable, yet on the other hand powerful enough to build things that we cannot predict yet. That creates interesting tensions, right? How can you master your software and how can you master the tools that you are building? And how easy is it to make these tools? How efficient are the tools running? And it’s an open question. If we can get fully developmental AI to work, and if this is a desirable thing, and if the fully developable AI is going to be more efficient than the transformer architectures or liquid foundation models or state space models that we’re currently building. Right?
And so for me personally, as a philosophically interested researcher, I want to understand how this works in nature. So I want to make computer simulations that lead to self-organizing software, that lead to systems that are self-improving and so on. And I’m not necessarily interested to build products from this, because maybe this is dangerous. Right? So maybe I want to make these experiments only on very small scale to see how it works. Or maybe it’s not efficient. Maybe it turns out this is actually much slower than running transformers on nVidia chips.
And so I’m not interested in this particular question. My own question—that’s why I started a nonprofit research initiative, the California Institute for Machine Consciousness—is: I want to understand how it actually works in organisms, how it works in us. How does our consciousness emerge? And to do this I need to figure out these principles by which organisms grow minds. And I want to recreate them, because I’m really, really interested in figuring out who we are and how we work.
J. R.-S.
Tell me more about that consciousness element. Does the spirit software patterns that we’ve been talking about matter? And why is consciousness a problem for you or for anyone? I hear it often how we need to solve the problem of consciousness. Tell me about it.
Bach
So I think if we had no consciousness, we had no problems. An unconscious person has no problems. And that’s because consciousness is related to the ability to care, to be someone, to be an entity that becomes a person that cares about things, that is somewhere in the world and develops moral agency and so on, instead of just being some kind of mechanism. And so it’s, I think, quite crucial to understand what we actually mean by consciousness. And I think when we look inside of us, we notice that we perceive to be perceiving. It’s not just that there is content there, but we also perceive that we are perceiving that content. And we create our self over this notion of being an observer. It’s not just that something is being observed, but observation itself is happening, and something is the agent of observation, the observer. And this is how we get to our own self model.
So it’s basically a model of a third stage. The first degree of modeling would be simply: there is content present, you model reality, and you find a pattern in it which looks like a tree, or which looks like a human shape, and it allows you to predict what’s going to happen. And at a deeper level, you notice that something is noticing this and making these models, and you observe how these models are being made. And then, in order to get better at this, you need to figure out: oh, I am the thing that is making these models. And if I do X, and this model is looking differently in this particular way, and this makes me better at modeling or worse at modeling. And so by fiddling all these knobs inside of myself, I become better at making sense of reality. And in this way our cognitive development starts, and we become human beings, become interesting minds, and become sentient agents that make a model of who they are and how they relate to the rest of reality.
J. R.-S.
Tell me more about how and why consciousness would develop, as you say, in a high-level reciprocal structure, so that we model what we’re modeling, and is it possible that there’d be a complex organism, but without that? Tell me a bit more about it. It’s just very intriguing.
Bach
That’s a very interesting question. Some philosophers think that consciousness is so difficult to explain and also it’s unclear what it is good for. For instance, you can wake up at prison at night and the sleepwalker and the sleepwalker does basically random stuff, but does that? Like sleepwalker might be able to get out of bed and open the fridge and make dinner. But when you ask the sleepwalker, why are you doing this? They don’t make sense. There’s nobody home. So my hypothesis is that consciousness is not necessary for executing behaviors. It’s necessary for making decisions, for creating coherence, or becoming a coherent agent. In the same way, to have trains, you don’t need to have a government, but to build trains, you may need to have one. And so what we observe in people is that when they don’t both become conscious at the beginning, before birth even, and they never wake up, they don’t learn anything. They remain vegetative. They don’t learn to move, they don’t learn to become people. And nobody of us seems to have something that’s simpler than consciousness. So a natural hypothesis is that consciousness might be related to a biological learning algorithm, an important aspect of it. And it’s the prerequisite for who we become. It’s the simplest way to train a self-organizing system, maybe. Maybe it’s only the simplest way to train a nervous system, in which case only beings with a nervous system could be conscious. Right? Or only beings that have something that is sufficiently similar to a nervous system, or has a similar functional role. But maybe it’s for every self-organizing system, in that case, trees would be conscious too. Right? And ultimately, that’s very interesting questions. Are trees conscious or not? If there are conscious, they’re probably conscious at a very different timescale than us, because if they don’t have a nervous system, their information transfer is going to be much slower and maybe more noisy. And maybe it’s too noisy to maintain the state of utter coherence. Maybe a tree is more like a Roomba that has only some kind of heart-wired thing. But ultimately, the question is, if I was an evolutionary process and I produced dynamic complexity in the world that minimizes free energy, is the simple solution to get something that looks like a tree to have an agentic software that knows what it’s doing, and has a model of its own perceiving reality in some form of real time, even if the real time is much slower than ours, and uses different code than ours, and a different access to reality and different models of reality than ours. It might be almost inconvincible to us. Right? Or is it simpler to heart-wire this into some kind of Roomba or self-driving car like architecture that simply exploits determinism of the substrate. And nature often thinks are so brittle that you cannot heart-wire complicated structure into it. For instance, the architecture of our brain is so complicated, there’s so many connections that need to work right to make you behave in the right way, but you cannot encode all of this in your genome. And so your genome is not a blueprint of your brain. It’s more like a set of hints on how to evolve your brain. The individual cells solve problems together, they coordinate with each other, and they need to solve lots and lots of local problems. And the genome is directing this evolution by biasing in certain ways, by getting humans to differentiate in particular ways before they try others. And this leads to the organization of your brain. But it also makes it very robust. So when you have a few mutations, you typically still get to a valid brain, or a valid nose, or a valid set of eyes. Right. And does consciousness play a role in that? If there are mistakes, you don’t need to hard-code everything, you can resolve problems. Because when you talk about why would you need consciousness, it makes me think about how the literature on consciousness as helping us learn new and novel things, but then often we can push that back into the unconscious and not need to keep consciously thinking about something. And so there’s this tension, like this time if you want to be more conscious, this time is when it’s not as necessary. I think that we can distinguish degrees of awareness. I tend to use consciousness and awareness almost synonymously, but I think that awareness comes in degrees, but consciousness is almost binary. You’re either conscious or you’re unconscious. It’s, for instance, at night, you either dream or you are in deep sleep, and you don’t dream. And a person is either awake or anesthetized, and the region between is very slim and uninteresting. It’s mostly just some kind of phase transition. But you can have different degrees of awareness. You might be noticing that you’re just sitting there and staring at something, or you might be observing that, oh, I’m actually observing how I’m staring at that thing, and what becomes apparent or doesn’t become apparent. And I’m actually a person that looks at this stuff, and you might also go a step higher and be like, oh, I’m not actually a person. There’s just a story that my brain tells itself, and then what is consciousness that maintains models of what it would be like a person that stares at the thing. And so you have different degrees of awareness that you can have. And you can also develop these degrees of awareness depending on your environment, the other consciousnesses that you’re vibing with, and maybe you find yourself sometimes in the presence of another consciousness, and you go, wow, my God, I feel much more aware now. I wonder what emotions and intuitions are, or elements of our awareness that don’t necessarily feel like they come from us, such that they’re in our awareness, and it must be internally from me, but you know, you can have emotions and you don’t even want to have that emotion, like, oh, this is annoying or upsetting. Yeah, you don’t make your emotions. The emotions are upstream from yourself. The outer mind makes your emotions. So the way in which I interpret this is, I think very similar to the story that is being told in Genesis one in the Bible. I think that Genesis one is not a story about a supernatural being creating a physical universe and the story was written the physical universe was not invented yet. I think it’s a story about the world that exists inside of the dream of reality that each of us experiences, and it starts out with your first consciousness this infant consciousness, finding an uninitialized substrate. It’s a world that is too, without form and void. And then it begins to make contrast in this world and then it assigns intensity of the contrast to brightness to the color of the day and the flatness of the contrast to dark color of the night, and using this gradient from light to dark or intense to flat, warm to cold, hot to cool, right, it is now able to represent dimensions and by combining dimensions, we can build objects. And one of the objects it makes is the space. Simple space is the plane. It’s this two dimensional space and associates with the ground that you’re standing on. And then it makes space above the ground and it fills it with organic shapes and solid shapes and liquids and so on and makes animals and plants and gives them their names. And this is the creation of the world in the mind during cognitive development. Every one of us does this very much as the eye models are doing it right now. And at some point it realizes that the purpose of this exercise is to negotiate the interaction between an agent and its environment. And so it makes the model of this agent of you as a person, puts this model of this person would be like the model of what the future of the trillions else would be if there were one single thing into this world, and then consciousness binds to this perspective of this being and it looks at the simulated world and your mind from the perspective of that being. But your emotions, your motivation feelings, there are in information by the underlying structure that tells this character this puppet that you are how you should feel about yourself how you relate to the world what is the success of the organism in the world. What is your score in this game that you’re playing. And you cannot control the score generation generation because otherwise you would cheat. Right, it would be an amazing cheat mode if you can control your emotions and some people try to learn how to control the emotions precisely to cheat. Wouldn’t it be much nicer if I only had good emotions ever all the time. Probably not right you want to have appropriate emotions sometimes your emotions need to be unpleasant to perform that because something unpleasant has happened and you should make sure that it doesn’t happen again or you should avoid it. Right so being afraid that your child dies maybe is a useful emotion, unless you already understand everything about this and you will do exactly the right thing this out having that strong emotion and you notice that when you are capable of doing the right thing and all the other thing is that you’re emotional is far less strong, right it’s no longer crippling you, you can just do what you need to do and so your emotion is more something like a reflex that is given to you from the outside, and you respond to it in voluntarily you are changed by it. And that’s the purpose of emotion to set up yourself to do things to relate to things in a way that you don’t choose, but something bigger than you choose this, this outer mind of yours. When you, and then of course, yeah, Genesis is associating this outer mind with God. And this might be a particular Hebrew thing with the Abrahamic religion that basically you say the outer mind the thing that determines your emotion is not simply how your organism feels about you. But so how does the organism feel about the self but it’s how the tribe feels about you or how the larger spirit of things feels about you. Right so it’s you basically replaced this outer mind of the individual organism is an outer mind of an entire tribe. This is what monotheism is doing. It basically finds your emotion generation and motive generation to a collective agent. There’s so many things here but one thing that resonates with me is in the Gilchrist work on the right and left hemisphere and the way in which there’s this bimodal way in which one the right hemisphere can see holes and it’s connected to everything more intrinsically it sounds more like that emotional states that you can’t cheat on. But but that’s mediated and you need also the left hemisphere that is being able to be more goal oriented and tell stories about itself. Does that does that feel resonant. I suspect that is strict hemispheric separation is a bit of superstition. I don’t think that the brain is necessarily that strictly naturalized, and they also don’t think that the lateralization is the same for people. I would expect if the different hemispheres of the brain are performing radically different tasks that they would also be morphologically more different, because the architecture of our brain is optimized for being as fast as possible. Your brain, every hop between two neurons is expensive so you try to minimize the number of processing steps. And so, if there would be radically different functions on the left versus the right hemisphere, I think you should expect that the brain is falling in slightly different ways or that you see different numbers of layers. And instead, what we see is more differentiation between front to back, and this front to back differentiation is much stronger than the left to right differentiation. So we have basically two halves of the visual cortex in the back of the brain and the structure of the visual cortex looks slightly different defined structure than for say the prefrontal cortex. And I think that’s because the data structures that are being modeled here looks slightly different. What is the prefrontal cortex doing? Like I’ve heard for example Andrew Huberman talk about how actually a lot of the front cortical parts of the brain are actually sort of stopping us from doing what we might automatically more intuitively do. And there’s a way in which it’s doing a lot of inhibiting. Like I’m just interested in, yeah, what that is that of all consciousness does. It seems to be related to directing attention and executive control in general. And some indication is that when you have prefrontal lesions, most people lose executive control. And in the same way, this is very famous when he has gauge story of this guy who worked at construction of, I think, train lines, and that an exploding what shot through his forehead and recovered from this but lost a portion of his prefrontal brain and as a result, he lost the ability to behave well and public and he became compulsive gambler and basically exhibited extremely poor judgment became very impulsive. And this is maybe a good example of what’s happening but what we also notice is that people that have brain lesions very early that have a hydrocephalus that kills the part of the neurons in their brain and so on, or that have accidents very early in their mental development. They still develop functioning architecture. And so, as long as you have enough plasticity to compensate it seems to be that our brain is relative general architecture that can compensate for many of the missing abilities, especially in the neocortex that is after all a relatively generic structure. You can also see that if you take mice embryos and you excise a region of the neocortex and plunk it in somewhere else, then it adapts to the new environment and performs the function of the new environment. So, it could be what matters is not so much the particular topology of the organ, but the topology of the problem. But in the same way as when you look at an AI architecture, and you have many layers of a neural network and you train the same problem into different architectures and different neural networks, you still find that it is an equivalent structure across all these different AI models. And the structure is given by the mathematics of the problem itself that the model is trying to solve. And the same way, there is a problem that our organism needs to solve. And so, the structure of the information processing in our brain is probably more a function of the problem that the brain is to solve than of the brain itself. We’ve talked about the problems and the necessity of consciousness, why we have it, the problems it solves. And we’ve also talked about AI and we want it to be a bit more controllable, contained. We may not want it on biological, more noisy substrates. I then wonder, would we want conscious AI, would that be something valuable? This is an unknown question. And it’s a question that I’m not wise enough to decide. So, when we say we assume by we you mean, you as society under Trump, or do you mean a group of sociologists or a carter of effective artists or who is we right and so there is no no such thing as a homogenous V because society is not super coherent, especially right now. And an interesting more interesting question is, what are your thoughts on this right what do you understand about the consequences of this. What is the direction in which you think the world should be going or not the question is, what do you think the direction of the world is going to be going into. Because ultimately, you and me individually are not very important almost everyone of us is quite replaceable. It’s very rare that we actually control a bifurcation of society where we individually are instrumental in making it go down this path or that path. And so often it’s a much better perspective to think about what is going to happen in the world. And what can we do about it, because most of the things that are going to happen are going to happen anyway. And it’s you as an individual can largely not control the larger conditions. We are more like molecules in air balloon, and when there are more molecules in the air balloon at some point it might burst, but it’s not an individual molecule that makes it burst. And so one interesting question is more, is it going to burst or what’s going to happen to it. And what can we do while this is happening. How can we, how is society going to adapt to the things that are going to come. And so my perspective is, it’s honest, almost inevitable that we will have conscious systems. It’s that we have conscious AI. And the question is, under which conditions do we build it. Are we able to understand what we’re getting into. Can we start a research initiative right now that is not happening in some kind of random uncontrolled way or directed on breaking the stock market or enabling a big crypto project or whatever. But maybe let’s take this perspective of saying, let’s do a philosophical project that is deliberately done in a way that is as safe as we can make it and as ethical as we can make it and study it under these circumstances and learn something about this. And then we can publish about this we can talk about this, and people who build products and make decisions or people who build regulation can make decisions. When we think about how society at large has ongoing structures and projects that we are only a tiny part of. Can we think of those as well as spirits and software patterns that are controlling groups and huge conglomerations of parts and and or all the multiple spirits. Like how should we think about these things that we’re a part of like we can be part of multiple different moving organizing groups and societies. Yes, I think this is the correct perspective. Most of the agents that you observe in nature are collective agents which means just made out of smaller agents that decide to serve a larger agent. And to the degree that they individual agents explicitly or implicitly serve other agents and the degree to which they become coherent in their interaction this is the degree to which the transcendent agent the agent on the next layer becomes real to which the degree to which it comes implemented becomes the causal pattern that’s actually structure reality. So for instance, if you have a bunch of people in your family or in your group house, who decide that they want to act in a coherent way. Right then, suddenly your family or your group house becomes an agent. It becomes something that makes decisions collectively that decides what’s good for the whole of us. And the whole of us is more than the sum of its parts it’s something else. Right, it’s an entity that exists at the next level of organization, and individually individual can become exchangeable against something else that performs the same role. And so it makes sense to see societies families villages nation states movements and so on, also from the perspective of seeing society as this society is basically a spirit Japanese society is an organism, and the religion in Japan is not animism animism is not a religion it’s a method physical framework. It’s basically just a way to assign names to objects. It’s a different than Western society does which calls organisms mechanisms and tries to explain things mechanically rather than via this self organizing, agentic patterns. And I think this notion of a self organizing agentic pattern, a soul or psyche or spirit or ghost is very powerful and useful and in Japanese society, religion is the society itself. It’s slightly different from Christianity, which is one that is directed on some idealized version of reality. At the reality that we as Christian society sees that it’s very imperfect. For instance, in Christian society, one of the core values is innocence, the protection of innocence. And was Mary and Jesus are representing aspects of this right. Jesus represents the sacrifice and God’s love, and Mary represents purity. And this innocence is what Christian society uses to justify violence. And it’s not simply the survival the switch with Jesus is that Jesus doesn’t try to survive. Jesus tries to be as innocent as he can. And if you value this, then you need to build a society that facilitates this innocence that allows individuals to get a wave as being innocent. And that’s different from the Roman society, which has no problem with an innocent person dies in the colosseum. It needs to be lawful. You cannot break the law when you kill a person. But that innocent people died is just how nature works. The innocence of a mouse doesn’t protect it from the cat. And it’s just the reality of the society. So the Christian society has a very different aesthetics. It has a different spirit than the Roman society. And also different spirit than the traditional Japanese society. And in the Christian society, you basically have an idea of innocence that can only be imperfectly represented. Then you have competing organisms that might want to have each other’s lunch. There is war and brutal things in an evolutionary environment. And so the world that the Christians want is one that is transcending this one. It’s a different organization in which we are much more tightly aligned with each other and basically become a single mind. This is, I think the idea of heaven is that the emerging of all the minds into one that basically have shared interests. It’s radical non duality. And in the Japanese society, that’s immanentist. It’s basically what’s currently a society is itself the organism that you need to serve. And so when Japanese people bow to each other, they acknowledge the spirits and status that they have with respect to each other in the society. It’s similar to the muscle cells bowing to the neurons that tell them what to do. But it’s really interesting. But then I have a question nagging me and I thought about it before is let’s say we take a multistagular organism or whole society. There’s a way in which single cells can kill themselves and for the sake of the whole organism or we have martyrs in societies, which is a weird thing. And how does this work? How would BSP possible? How would everyone agree to do this? Is it because the high level spirit is sort of shaping things and so you can identify with that high level spirit and so the cell will willingly die for the organism, a person for the society? Imagine that you want to take over another organism and you are an organism that has the capacity to take over other organisms, like for instance, quality cells, fungus, that is able to take over the brain of an ant. And to do this, it needs to send cells into the ant that are going to try to reprogram the ant. And the ant has an immune system, so not all of these cells are going to be successful. Right, so many of them will have to sacrifice themselves for the good of the fungus. And from the perspective of the fungus, these cells that sacrifice themselves while attacking the ant, they are martyrs. They basically try to subvert the spirit of the ant into the spirit of the fungus to turn the ant into a servant of a different organism. And a similar thing is happening when one religious organization and a religion is an institution that maintains a spiritual organization in society, is trying to impose its own aesthetics on another society. And so if you were to start at a religion that is trying to take over another culture that has a different spirit, that has different aesthetics, that has a different idea of innocence and of gender relations and of reproduction and of progress and of governance and how to relate to nature and so on, that has a different sense of innocence. Right, then you will have to have notars, you will have to have individuals that are going to try to infect the spirit of the other society and subvert it. And from the perspective of the other society, this will look like corruption. Right, and if you have corruption, you need to have an immune response that tries to eradicate the corruption because otherwise you will die. So what you see in the warfare between different religions is actually very similar to the warfare between different organisms. I wonder how you’re thinking about the metaphysics of the world, because earlier you talked about Genesis and it actually being a story to explain the development of consciousness within, say, a child, an organism. But I wonder why it can’t be both the organism developing in its own life, but then also the creation of the world itself. Like, maybe you could have some other metaphysics like an idealism like from Bernardo Kastrop and then that is also the creation of the world itself and the person. And there is a certain field in philosophy that is concerned about how do you know something. And that’s called epistemology. And epistemology is not the result of you teaching it, it can be discovered independently and has been discovered again and again in many cultures. And so, a thing with epistemological claims is, this is always you have me to be able to justify how you know something and why the alternative is not true. And so when Bernardo Kastrop says the universe is a big brain, my problem is that I feel compared to asking how do you know that simply this is a kind of beautiful idea is not good enough. Is it possible that the universe is actually not a brain, but it’s a liver. Maybe it is not a liver and it’s not a brain maybe it’s something else entirely. A more interesting question is the universe the result of intelligent design or not. What does it mean that it has intelligent design. Does it mean that there needs to be an entity that is congruent with the mythology of my favorite religion that is somehow involved in the creation of flat space time. Or do I get, if that is the case, can I show this somehow does this follow from some mathematics that are in covertly true, right and that I can actually walk through, and that I understand or is this simply the result of somebody else thinking it’s useful if you’re stupid enough to believe that thing that you listen to my guidance because you’re not able to tell your left foot from your right foot if you believe stories like this. So you better have guidance that better comes from smart source your past like myself. And a lot of religion is like this that you’re basically, if you’re a person who is stupid enough to actually believe in virgin birth, then maybe you need guidance, maybe you need strict rules, maybe you don’t need to learn whether Jesus comes from the vet market or doesn’t have a zoonotic origin. Because you are not able to make sense of the world. And if you are able to make sense of the world then the disclaims make very little sense because how would you know that you would need to make a mathematical model that shows you how the universe could originate, or you would be able to necessarily to have a communication to such an entity that explains to you how it was made, and then also understand whether this entity is some random mountain spirit or whether it’s actually a creator deity. How would you know that it’s actually a D me or do you think ultimately it’s an unknowable question then the origins of our world. No, I think that at some level it is unknowable because you could be always living in a simulation. And if you are in a simulation you don’t know what the source code of the simulation is. If you live in Minecraft you cannot know what kind of computer you’re running on because these simulation insulates you from the ground truth in some way and so regardless of what kind of universe you’re in if it has enough compute. Maybe you just exist in the memory is of an agi right now. And so in this sense you cannot actually know how you exist. But if you go from the other way around from first principles, which universes could randomly form, or by themselves form that have the necessary and sufficient conditions to contain an entity like you. Maybe there is a relatively small space of such solutions. Maybe there are not that many ways and physics could work. And maybe you figure out such ways from first principles and then you discover. Oh my god the physical universe that we observe looked exactly like that. That would be quite compelling. Right. And this is in some sense what physicists are trying to do. So they see the observing that either you don’t need a lot of look ahead and planning to construct a universe like ours, or the universe is the result of some designer that needed to adjust many many knobs and screws to make it look at this very interesting non random way and this is, in some sense, part of the question that physicists are trying to solve how much fine tuning is there, and is everything is nature. Right. The definition everything that is is nature. But so the notion of supernatural doesn’t make a lot of sense except by hobbling your own mind. But the question is what kind of process is it. Human beings are also nature computers also nature everything that we’re building computer games are also nature. The latest base to make computer games and spirits are also natural entities that patterns in physics. I think it’s the easiest explanation of what they are. This has been great. This is this is really gone deep enough so I appreciate your time and any anyway you want to point people to your work anything else you want to say if you finish this. Yeah, have a look at CIMC dot AI that’s the website of the California Institute for machine consciousness. And also have the pleasure to work as an advisor for liquid AI, which is an AI company that is building models of information processing that are more inspired by biology, than the original transformer models are and in this way we make AI models more efficient and more easy to run. And we hope more useful to the world.