"New Bodies, Unconventional Minds: a continuum of cognition (& consciousness)

Watch Episode Here

Listen to Episode Here

Show Notes

This is a ~39 minute talk titled "New Bodies, Unconventional Minds: a perspective on a continuum of cognition (& consciousness)" for a consciousness conference about our laboratory model systems relevant to the study of minds, my Platonic Space model, and a few opinions on consciousness (at the end, since I don't have any strong claims about consciousness per se).

CHAPTERS:

(00:00) Continuum of novel minds

(07:11) Molecular networks and learning

(11:42) Xenobots and anthrobots

(18:25) Paying for competencies

(28:21) The space of patterns

(33:28) Rethinking nature of consciousness

PRODUCED BY:

https://aipodcast.ing

SOCIAL LINKS:

Podcast Website: https://thoughtforms-life.aipodcast.ing

YouTube: https://www.youtube.com/channel/UC3pVafx6EZqXVI2V_Efu2uw

Apple Podcasts: https://podcasts.apple.com/us/podcast/thoughtforms-life/id1805908099

Spotify: https://open.spotify.com/show/7JCmtoeH53neYyZeOZ6ym5

Twitter: https://x.com/drmichaellevin

Blog: https://thoughtforms.life

The Levin Lab: https://drmichaellevin.org

Transcript

This transcript is automatically generated; we strive for accuracy, but errors in wording or speaker identification may occur. Please verify key details when needed.

Main Episode

[00:00] Alright. Yeah. Thank thank you for having me at, this amazing conference. I apologize. I can't be there in person. What I am going to, try to do today is to show you in the time that I have a few model systems that I think challenge the conventional philosophy of mind approaches. I'm only gonna show you, because of lack of time, one or two examples of things that I usually spend about an hour on each, including, I'll I'll I'll describe a little bit of a of a pre Darwinian phenomenon that I think will be of, of of of some interest in this specific section. Then I'm going to, give a proposal that's even weirder than, the idea that consciousness is a property of all matter. Then I will close with a few personal thoughts on consciousness that I can't really approve, but I'll just tell you what, what I what I think. So the reason that I'm interested in all this stuff, is that we have to, I think, get... Start to gain a science of where novel minds and their properties come from. We have stories of, selection, evolution. Engineers think that they can, create novel minds with specific properties. But, we really do need to understand, as we fill in the whole spectrum of, possible embodiments, I think the distinction between machines and humans and life, these terms are not gonna survive the next couple of decades because there's going to be a very rich, spectrum and continuum of cyborg and, chimeric cases that are not handled by the kinds of, categories that we have today. I've given talks to, psychiatrists and psychoanalysts trying to point out that because of regenerative medicine approaches and in general, freedom of embodiment, the, the the clientele list of of beings with with whom they're going to have to work to, improve their lives to figure out what their collective unconscious kinds of symbols mean and dreams and all of this for beings that have not been here with us at the same evolutionary stream. So we don't really have a science of this, and this is this is what what motivates my work. Now I primarily work in, third person science. We do experiments on intelligence, on, cognition and problem solving and all kinds of weird model systems, cells, molecular networks, tissues. I don't normally, say much about consciousness because I think it's very hard to do actual experiments on consciousness. So I... What I don't claim is to have any kind of a new, theory of consciousness today. I don't claim to have, data supporting one, theory of consciousness, but I will make three claims. First of all, that I think, the near universal focus on neurons is way too narrow. I think that patterns of anatomical form, three-dimensional space behavior, so conventional behavior, physiology, gene expression. All of these are a class that we need to understand as part of, cognition. These are navigational strategies of problem spaces, they go far beyond neurons. I think that, very briefly, I'll I'll say what I spent the whole time on last year, which is that for the same reasons you associate consciousness with complex brains, in other words, the way we solve the problem of other minds with other humans, for those exact same reasons, you need to consider very seriously the possibility of consciousness in many body structures. And finally, I'll spend most of the time talking about my my contention that, physicalism is false, and the implications of that in an important sense go all the way down. So, the first thing is that we know from developmental biology is that it is a smooth continuous process. We start out being the subjects of chemistry as an unfertilized oocyte. Eventually, we move through developmental physiology, behavior science, maybe psychoanalysis and psychiatry, maybe some detours into other disciplines. But fundamentally, while we have names for journals and departments and funding bodies, the substrate is continuous. There is no magic bright line that turns chemistry into mind. So what we're looking for are models of scaling. We're not looking for hard categories of things like is intelligent, is not intelligent, and none of that is going to do the trick. We need to understand what exactly is being scaled. And because, fundamentally, if we take seriously both evolutionary theory and, developmental biology, we need to understand that the null hypothesis is continuity. If someone wants to argue for emergent phase transitions, that is sharp categories, we're gonna need really strong evidence for it. I think the null hypothesis should be continuity not only with molecular components from which we, we all started life, but also with many other kinds of things that are are already now possible both on the biological end and on the technological end. It's continual everywhere. The other thing I wanna talk about is embodiment, to point out that we as humans have a lot of, what I call mind blindness. We are okay with noticing, intelligence in medium sized objects moving at medium speeds through three-dimensional space. But biology has been solving problems in many other spaces long before brains and muscles appeared. So, different components of your body have been navigating, transcriptional space, of gene expression states, of physiological states, of anatomical states, so anatomical morphospace, doing all the kinds of things that we see the the the cognitive loops doing in in conventional behavior. So embodiment doesn't just mean being able to run around in the three-dimensional world.

[05:07] Your body is full of agents that have been embodied in spaces that are hard for us to recognize. They operate in different scales of space and time and at different different kinds of spaces. So the first part that I'll go through very, very quickly is is this idea that, within each of our bodies, there are systems that meet all of the four basic criteria. There's a fifth one, which, we can have to do with, various, phi like metrics and causal emergence and so on. And and that... With that one, we're still investigating, but I think I think it's looking quite good. The bottom line is this. Most of us attribute consciousness to others even though we cannot directly verify it because of roughly four things. Behavior, so the ability to solve to solve problems, to communicate, to have goals, and and and and systems for which a first person perspective is really useful to understand what that system is going to do. You need to take their perspective to really know what's going on. So so that's one behavior. Two is evolutionary kinship, meaning they come from the same, stream that we do. So if we're conscious, maybe so are they. Conserve mechanisms if if their brain operates in the same way that our brain does, maybe maybe then we take that seriously. And, possible, measurements of causal emergence and integration. So so what I'll just say about that is that at the level of morphogenesis, all of the amazing things that our brain does in terms of fall... Create memories, be able to entertain counterfactual patterns, make decisions, integrate information, all all these kinds of things are basically sped up, evolutionarily sped up versions of dynamics that were going on back at the time of bacterial biofilms. So every cell in your body has ion channels. Most cells have electrical synapses. A lot of cells have neurotransmitters outside the brain. All of these systems form electrical networks that while your brain sped up to use those networks to make decisions about your movement in three-dimensional space, These networks were guiding your movement through anatomical space. They were taking you from being an egg to, being an organism in anatomical space, solving problems along the way, going around barriers, dealing with novelty, and I... We we could spend hours on all those examples. But I think, I think what, we've we've developed the first molecular tools to read and write the mind of the body, so non neural cells using all of the tools of neuroscience. And, these are just a couple of examples, where we can we can read out, in some cases, interpret, and most importantly, write back. And this is the backbone of our, regenerative medicine, program in the... Cells to make eyes and limbs and normalize cancer and these kinds of things by communicating with the cellular collective and rewriting their electrical pattern memories. All of the tools of neuroscience work outside of the brain. The concept as well. The tools and the concepts do not make the distinction that, that we make. And I think what evolution has been doing is basically pivoting some of the same strategies across different problem spaces. But today, what I want to dive in on a little bit is what's going on below the single cell level. So below the single cell level where you have molecular networks, what we found is that even, very simple molecular networks describable by small sets of ordinary differential equations are are are exhibiting a learning phenomena that are familiar to, any behavior scientist, meaning that there's about six forms of learning that molecular networks can do out of the box. You don't need neurons. You don't need a brain. You don't need a cell. Even it's just the molecular networks that can do this. They can do habituation, sensitization, Pavlovian conditioning. They can count to small numbers, probabilistic inference. That's Walter Fontana's work. And, we're we're taking advantage of that, through... In in biomedical applications and things like drug conditioning, trying to, trying to train the actual molecular pathways inside of cells. And, you can read about, all of that here. But here's the interesting thing, and this is the this is the prescientific, pre Darwinian, I mean, version of this, which is that what we've what we've noticed, and this is Federico Pagosi's work, from our group, is that for a large class of these networks, as you train them, their causal emergence goes up. So as they gain... As they learn and they gain experience and they they... Their causal emergence goes up. Sometimes it goes, consistently up. Other times, it it rises when you're when you're interacting with them. So when you stimulate them, they sort of wake up in between your communicating with them. They sort of disappear back into the into the void, so to speak, but then they wake up with with higher and higher causal emergence every time you every time you stimulate them. So here's what here's what we discovered. Increases in causal emergence make for better learning. They become better learners. Learning, in fact... Learning in in in turn raises causal emergence. So you get a positive feedback loop remarkably forcing them to forget, and we spend a lot of time figuring out how to make molecular networks forget things. Biomedically, that's very important. Making them forget does not erase the gains that they made in causal emergence. In other words, this loop is asymmetric. It points in one direction.

[10:09] It's unidirectionally pointing to higher levels of of intelligence and and agency. But the most amazing part is that as it turns out, random networks already do this. They are almost optimal for this ratchet when you look at random, not evolutionarily selected, not designed, random networks already have this amazing property of this asymmetric tendency to to learn and to become an integrated agent. So this is this is wild because it means that that property is a free gift from math. It does not require evolution. It does not rest on any aspect of physics. It it... This is anything that is... Any system that is describable by a a a small set of ordinary differential equations is, potentially, is potentially able to do this, and and, many networks used in biology can do this, but not because evolution optimized them. Of course, evolution can can, then to fine tune them for different purposes depending on, ecological need. But the material going into the process of of evolution already is optimized for this. Not the chemistry, but the math. It's a free gift for mathematics. It comes from the properties of of causal emergence and of network dynamics. Right? So so evolution gets this free gift from math, and there's a there's a preprint here that, that you can check out what this means in an actual model of origin of life. I think I think this is a this is a major input into the, into into what happens prior to, Darwinian selection taking off the ground because there are no replicators at the time that we're that we're doing this. The second thing that I wanna talk to you about is, some model systems, that, are kind of a stepping stone for us to try to understand. Their their model systems were asking where do, cognitive properties come from. So what we're doing is creating novel, novel living systems that are made of cells. The cells do have an evolutionary past. We take these cells from existing organisms. We don't make cells from scratch, of course. But what the cells do is composely, autonomously compose novel life forms with system level properties that were not engineered by us. They had never been selected for because they never existed nor were they trained. So one example is what we call the xenobots. Xenopus lavis is the name of the frog these cells come from. We rescue these epithelial cells from, their boring life of, being at a a sort of static covering on the outside of a of a frog from from this, embryo. We put them in the dish, and what they do is... You can see this here. Each one of these circles is a single cell, so they form these little clumps. They don't all look like a little horse. I just thought that was a cute video. They will eventually, combine together into, into this kind of thing. They don't die. They don't crawl away from each other. They don't form a cell culture layer. They form this. They form these little these little, creatures that have cilia, and they can have lots of different kinds of motion. They can go in circles. They can go back and forth. You can sculpt them into different, like this floating donut. They have collective behaviors, but there's no neurons. These are just epithelial cells. They have some cool spontaneous behaviors. If you put one in a maze like this, you'll find that it it goes for a while. It takes corners without having to bump into the opposite wall. At some point, spontaneously, it it turns around, goes back where where it came from. So it has internally derived spontaneous activities. It's impossible for for these guys to replicate in the normal froggy fashion because they're just epithelial cells. But if you provide them with loose, loose epithelial cells, they will, basically implement von Neumann's dream, which is a machine that goes out and makes copies of itself from things it finds in the environment. It it collects these these little cells. It it polishes them together into this sort of ball. The ball matures into the next generation of xenobots, guess what it does? The same thing. It makes more xenobots next generation next generation next generation. Then we call this kinematic cell for replication. There is... As far as we know, there's no other organism on earth that replicates this way. Certainly, none of our ancestors, none of their ancestors replicated this way. This is a novel thing. They have other novel skills. We look to see what new genes they might express. Hundreds of genes that frog embryos do not express. Some of them were a cluster related to, sensory perception of sound. So we stuck a speaker under the dish, and we found out that, yes, indeed, they respond behaviorally to acoustic stimuli. We found out more recently that they can learn. They can form two distinct memories of past experiences and hold on to them for at least twenty four hours, which we can then read out the memories. We we now figured out a way to sort of read their read their experiential history. And, if, if we inject a calcium, calcium reporter dye so that you can see, you can see all their sort of computations that they're doing, we can run the typical metrics on them as, people do in in neuro... Neurological context to, for example, find out if you're looking at a patient where there's, quote, unquote, someone home versus just a, you know, a pile of neurons. So, the integrated information here between Xenobots and their null model is, about the same, if not bigger, than the difference between human fMRI and its null model.

[15:18] I'm not saying xenobots are like brains. I'm saying both xenobots and brains are expressing dynamics that are that are not unique to, to neural tissue. You can give them some neural tissue If you wanna know what a nervous system would look like in an organism that's never existed on Earth, we we we took some, some neurons and randomly shoved them into a xenobot, and what you get is something like this. It's a very... It has very different, behaviors in terms of swimming around. Here's what the nervous system looks like. You can sorta you can sorta track it. It does have, does have specific layout. If you think this is some specific, thing about amphibians, this is this is what your cells would do if I liberated them from your body. So these are adult tracheal epithelial cells taken from, human patients who go to a biopsy. We then buy the leftover cells. They self assemble into this little creature we call an anthrobot. Again, has its own, dynamic motion, has lots of interesting behaviors. For example, anthrobots have over 9,000 differentially expressed genes compared to the cells, in the body. So half the genome is differentially expressed in a very specific way. They have four distinct behaviors, you can draw an ethogram of their transition probabilities. Not 11, not two, four specific behaviors. If you take a plate of human neurons and put a big scratch through it and throw in some anthrobots, they they find a spot. They assemble on that spot into what you could call a super bot cluster, and then what they start to do is knit together across the gap. They start to heal. If you lift them up, this is what you see. They start to they start to heal the the damage. So who would have thought that your tracheal epithelial cells, which sit there quietly in your airway passively for, decades dealing with, with smoke and and mucus and so on, given the opportunity to be to be, liberated from the rest of the body, they form a novel being with novel gene expression, novel physiology, novel functionality, novel behaviors that you would have absolutely zero clue about by looking at the human genome. We didn't engineer them. We didn't, put in any synthetic biology circuits. We didn't train them. We didn't give them any nanomaterials. No scaffolds. If you sequence them, all you will see is is 100% normal Homo sapiens. So so now now we understand, what we have to do. We have to try to we have to try to figure out where these very specific patterns come from. It's not enough to say that if you take cells, they'll do something interesting. We, as a society, who make lots of novel systems, social systems, financial systems, Internet of things, swarm robotics, we still have really no idea how to predict the properties, the goals, the the intelligence level, the motivations of novel minds. So these are simple model systems where you can start to ask the question, where do these things come from? So, for the last part of the talk, I want to, sort of give you give you my view of this, which is, really kind of, kind of unusual, and I think most most people are are, don't don't don't sign on to this, but, but this is what I think is happening. So the first thing we need to do is to say that there's a there's a, there's an invariance here. There's a symmetry here that behavior, anatomical shape, physiological, kinds of, policies, All of these are part of a class. These are patterns. These are ways of navigating specific problem spaces. These are behavioral policies in different different spaces. So anatomy... So your shape and your behavior are not radically different things. They're just different ways for a die... Different collective intelligences to navigate spaces. Once you've said that, then then we can ask, okay. Is there a common, source from where the specifics of these patterns come from, especially ones that cannot be told as a story of evolutionary selection from past histories, from physics and and and so on? Where where do they come from? And what... Is it possible that if some... If if these patterns are not just static order for free, which Stu Kaufman told us about decades ago. But what if what if there's more than that? What if it's not just order for free? What if some of these patterns are actually competencies recognizable to behavior scientists? They're scientists. They're behavioral policies that you would recognize as simple or maybe not so simple kinds of minds. So the way that we study them in our group is try... Is to try to quantify the following. We want to quantify the cost of specific competencies, both in biologicals and in, quote, unquote, machines. We have basically three ways of paying for for competencies. And, when when you try to make something, whether it be living or or nonliving, we generally assume you're going to have to pay for anything good that the system knows how to do. You will either engineer it directly, so rational design if you already know how to do it. If you don't know how to do it, you could evolve it. So, you know, make make trillions of variants, kill off everything that doesn't do the job, and keep the keep the the ones that do. Or you could train it. So give it history with the problem, and then eventually, maybe it'll learn to do the right thing. So either way, you have to pay. You either pay by your work in engineering, and and and creating the thing. You pay by generating test enormous number of variants, or you pay for it by training it. And so then you can compute a delta.

[20:20] You can look to see what you got in a given circumstance, and then you can see, is there anything left over? When we subtract the effort of... From what we put in, from from what we actually got, what is the what is the delta? And when there is a delta, I claim that, it's not it's not enough to just say emergence. I think emergence doesn't even begin to scratch the surface of of what's actually going on. Because emergence just basically says you got surprised, and it says that, there's some set of, amazing, sort of facts that they happen to hold in our world. I think that's that's too mysterious and too pessimistic. I think as scientists, we have to assume that there's a deeper order. There's a there's a space, a structured latent space from which specific things that emerge, whatever they may be, from which they are drawn. And we have to study that space. We have to understand the contents of that space. We can't just be sort of surprised whenever emergent things happen. I will I will also say for people that are that are interested is that, these things, can also be be be linked to, what we call inspiration. And I... There's a whole separate talk that I do about about that, what what this means for inspiration all the way from human creatives down to, quote, unquote, machines, which I'm not sure there are any machines. But the the bottom line is this, that, in a... For example, in a system like this... So so we know when we paid the computational cost of making a nice frog. So so for eons, this genome, bashed against various environments and, created, so it was selected for these specific, these these specific, developmental sequences and then eventually behavioral sequences. Here here... Here's an early xenobot. Here's an 83 old xenobot. It's turning into something. I have no idea what it's turning into. Then you've got these behaviors. There's never been any xenobots. There's never been any anthrobots. There's never been selection to be a good anthrobot or xenobot. When did we pay for all of this? And it's not enough to just say, well, at the time you were selected for you were selecting for frogs and humans, you also somehow selected for xenobots. That... The the whole point of of evolutionary theory is that it's supposed to paint a tight relationship between the properties of some creature now and the set of past environments which got you to those properties. There's something, additional going on here. This plasticity, this ability to make a new, being with the exact same hardware with novel, properties, is something that we actually... We we absolutely have to understand. Now just in case, you know, you're feeling smug about being a complex life form and and and saying, isn't, you know, isn't isn't life amazing? It it can do these things that are not, fully circumscribed by, by by the by the design, the... This the selection and the training that that were done. I will just point out that this this... It it doesn't take complexity, it doesn't, like, take life for this to be true. We've been studying very minimal computational systems. I'm talking extremely minimal, not trillion parameter language models, but little algorithms, six, seven lines of code, everything transparent, no new, mechanisms that you don't know about, no quantum anything. Classical... Purely classical, simple algorithms. And even there, and you can read about it here, even there, we see what you might call spontaneous side quests, meaning that, yeah, it does the thing you ask it to do, but at the same time, it also does interesting things that are not in the algorithm. They are nowhere in the algorithm. And the things that it's doing, if you know how to look, because these algorithms have been around for, eighty years, and nobody noticed this before, because we always assume they just... These are dumb machines that just do what you ask them to do. The things that they do are not come... Not just complexity, not just order, not just, unpredictability. They are competencies that are recognizable to any behavior scientist. There are things like delayed gratification, the various kinds of problem solving. So I I think what we're learning here is that, we we need a bit of humility in the following sense. I think I think most people would say that, okay. The formal models of chemistry do not capture the human mind. Right? It's... We we we are more than than than than within than the chemical machines that we're supposedly underlie our our our embodiment. But, that's okay because over here, there are these dumb machines, and they can never be like us. And they they are fully encompassed by the models we have of them. So Turing models, algorithms, things like that. My point is that, nothing really is fully encompassed by its formal model. That's just the basic philosophical point that I think we have to we have to, we have to accept. But, specifically, I think that as as limiting as machine metaphors are for life and for our minds, those metaphors are also not fully capturing what we think of as machines. These these these, these metaphors are are limiting in both cases, and I think this this thing that we're talking about, which I'll I'll I'll spell out momentarily, is something that goes all the way down. It is not just for humans. It is not just for biologicals. It is not just for extreme complexity or the product of evolution. It goes all the way down, and it starts very, very early on the complexity ladder. And I think what we need is not emergence, but a systematic investigation of the possibility space from which policies are drawn that, show up in our systems that we did not put in by design, by selection, or by training.

[25:21] Now we we should already know, from clinical cases that the hardware spec doesn't always tell the story. If you were to take a CT scan of this patient and see pretty much no brain, you would assume that you're looking at a profoundly, impaired function. And most of the time, that's true, except not all the time. And there's a set of cases that Karina Kaufman and I reviewed here, clinical cases of patients with extreme reductions in brain function, but normal or above normal IQ levels and and performance. So there there are many, many... It's not just about Zenobots. There are many scenarios where there's something something is is is going on where, it is not paid for by the conventional three, three sources of, of our typical paradigms. So what else is there besides besides physics and biology? Where could, where could order possibly come from? Because in the life sciences, we like our history, meaning meaning evolutionary history and biology, and we like physics. I mean, the physical environment imposes constraints. There's a lot of stories about how how constraint, shapes intelligence and so on. But there's something else, and and we've always known there's something else from from the time of Pythagoras, and that's the patterns of mathematics. So when you look at this thing, this is a Halley map of the simple function and complex numbers. The very specific features of this, okay, the the the the specifics of why this looks like... This pattern looks like that and not like something else, is something that you cannot explain via selection. There is no history of selection that got you here. It is not explained by any fact of physics. There are no physical properties that define this. In fact, there's nothing you can do in the physical universe to change this. If you were around at the Big Bang tweaking all of the fundamental constants of the universe, you could not make this be different. You couldn't make the value of e, the natural logarithm, be any any different. Where does this come from? So so we we already know for for thousands of years, we know that not all important facts are physics facts. There seem to be, there seem to be a set of patterns, of of of all different kinds that are drawn from a different option space than the patterns of of of physics. They're they're neither explainable nor controllable by the typical things we do in in in in physics. And yet many of these patterns have major implications for physics, for biology, for for for cognitive science. Often, if... No matter where you start, if you just keep asking why long enough, like a five year old, you just keep asking why, why, why, eventually, you end up in the in the math department. So, you know, if you wanna know why the cicadas come out at 13 and 17 years, the biologists will tell you it's because they wanna avoid predators from timing their cycle. And you say, okay. But why why, you know, thirteen and seventeen? And they'll say, ah, because those are prime. And you say, why are they prime? Go talk to the math department. So sooner or later, this is... You know, when you start asking these things, you eventually end up the turns out to be, well, it's some property of this mathematical object. That's that's why. So so that's interesting. That tells us that at least in some cases, there are sources of information that that do not come from selection. They do not come from physics. But why would we assume that the Platonist mathematicians have a monopoly on the, on these things? Why why only why only math? So here's the here's the... Here's my here's my proposal. What I think is happening here is that there is a space of patterns. Some of these patterns are very simple things that, simple low agency patterns that sit still long enough to be the subject of formal proofs and derivations and things like that. Those are the subjects of mathematics. And I would call math the behavioral science of just one layer of this space of of of patterns. But I think other, contents of that space... This is a hypothesis that other contents of that space are things that we would recognize in other sciences and other disciplines. They impact, the patterns of developmental biology. They impact the patterns of behavior in in cognitive science and so on. I'll give you a simple example of something that that is intermediate. Like I told you before, groups of ordinary differential equations. They are, on the one hand, certainly a mathematical object. On the other hand, they have Pavlovian conditioning competencies and and and probabilistic inference and so on. And so they sort of start to become the subject of of of behavior science. So I I just think we ought to loosen the assumption that this that this nonphysical space is just for, low level truths of of math and think about other high, high agency kinds of patterns that would be recognizable if they if they were were embodied. And actually, we're sort of even testing that because we're giving robotic bodies to this... To these kind of patterns as controllers to see what their behaviors would do because I don't want to assume that these things are dumb. I think even those assumptions are extremely, extremely dangerous. So what I think in the end is happening here is that, everything we make, bodies, cells, robots, organisms, embryos, all of it, are basically front end thin clients. The reason that our paradigms of design, training, and selection don't tell the whole story is because those are all theories of of the front end. They're they're all theories of this interface through which we're getting the patterns. And what we haven't done, except for the mathematicians in their cases, what we haven't done is study the option space of these patterns that interact with various kinds of embodiments. And we haven't yet... And this is what my group is doing.

[30:23] I have about, I don't know, five or six people working on on experiments on this, figuring out what kinds of embodiments are are are good hosts for what kind of patterns and what are the what are the properties of these. And so at this point, most of you are probably thinking, but this is this is old school. Interactionism is supposed to be dead. And, and my claim is simply this. I'm not saying too much past what Pythagoras already said other than to say that this isn't just for math. This is, this is for for all of the things that we're interested in in behavioral and cognitive science. I think physicalism was already not viable in the classic universe. I don't think you need, quantum, you know, quantum kinds of magic for it because already the classic universe was was haunted by patterns that make a huge difference to what happens, but are not themselves discoverable or changeable. They don't belong to that same physical world. We already knew there was a weird kind of interaction that is not the usual billiard ball causation. That is not an energy bearing interaction. So so we we are... We already knew that that was true, and I'm working with Lauren Ross, who's a philosopher of causation, to do a very careful sort of philosophical analysis of it. But but I think there's no there's no hiding from it. So you can you can read about you can read about all that here. This this this paper just came out. The... Fundamentally, what I'm proposing is that the mind body relationship is basically the same as that between the patterns of mathematics and physics, physics, just kind of on steroids. And I think it... It's extremely pervasive. I think once once once we understand that this is happening, we can find examples everywhere that we previously chalked up to emergent of... Emergent facts of selection or or something like that, we've... When we didn't have a good story. So what we're doing in the lab is, building, communication interfaces to some very unusual beings. So, specifically, you can you can find some of our, some of our papers on this. We're, using games as an interface to, to various kinds of systems. For example, we've got gene regulatory networks playing Atari games. We're trying to communicate the cells, the tissues to, to mathematical mathematical, objects. We want to be able to, ultimately, for medical purposes, you wanna be able to talk to your liver in in in plain English. You won't be talking about current events. You'll be talking about physiological states, but to talk to your various organs and to cells. And maybe this will have, some impact on on consciousness studies, but for sure, it will have impact on, cognitive science to broaden the, the set of cognitive kin to whom, to whom we can, we can meaningfully meaningfully communicate. So, in in summary, here's here's, what I think. I feel somewhat confident of the following claims. I I think nothing useful in this field is binary, not not consciousness, not intelligence. I think we need models of scaling and transformation, not propping up prescientific categories. I think none of this is about brains or about embodiment in three-dimensional space. I think there are much more generic and fundamental dynamics here that are that are really critical. I think our formal models don't tell the whole story either for life or for quote unquote machines. I'm not sure there are any machines in that in the in the sense that these these these dumb things that only do what you what you force them to do. I think that that that this that this notion of cognitive patterns goes all the way down. And I think it's not just complexity. It's not just unpredictability. It's competencies of navigating various problem spaces that we did not pay for in the conventional ways, and that require us now to have a research program to identify what is this what is this fourth thing that's contributing to all of So those things I I feel pretty confident about. In progress, our... We're we're working on tools for communicating with unconventional minds. We're trying to understand what aspects of the architecture matter for specific types of ingressions. For example, one interesting thing we might wanna know is whether fully fully synthetic embodiments, such as such as AIs, robotics, whatever, are whether they're pulling down some of the same patterns as biological media is... Pulls down or whether it's a completely separate set. So these are the kinds of things we're we're we're studying. We want to understand the mapping between the material pointers, so the the the construct of the body, and the specific set of the the ecosystem of patterns that end up that end up... That they end up hosting. And and we're trying to figure out what kinds of minds are buried within mathematical patterns. If we give them embodiments, so robotic bodies, what are their behaviors like? If you ever wanted to have a conversation with the natural logarithm or with a fractal or something like that, it it may end up possible. We're we're we're testing all that up. Very speculative kinds of things that I don't know the answer to is how well consciousness actually tracks intelligence. I don't know. You might think there would be some relationship, but I'm not sure what what that is. I I... I'm not sure how to gain a first person understanding of these exotic minds. I think we can study them in in third person. I don't know if we can ever know what it's like to be them. And my my strong suspicion is that we are not physical beings that occasionally get impinged upon by patterns of... From this space. I think we are the patterns. I think the distinction between thoughts and thinkers is, is is not real.

[35:29] I think, there's there's one continuum of of patterns, and some of them are larger than others, and large patterns can spawn smaller patterns and and so on. So at the very end, I wanna say a couple things about consciousness. I haven't said much about consciousness this whole time because I don't have any strong data, and so I have no strong claims about consciousness. But I figure I should say say what I think at least. I think that consciousness isn't made. I don't think it's a late product of evolution. I don't think it... It's a property of matter even. And I'm very skeptical of this notion of, unconscious processes. We we can... We we think we can imagine, and in fact, that seems to be the null hypothesis that everybody has that, that the the the most of the universe is sort of dead, and then we have to make arguments if we think something is conscious. I'm I'm I'm skeptical that that it's even coherent, to say that that there's such a thing as an unconscious process. We have no evidence for such a thing. Our n of one, example that we can actually check, which is ourselves, seems to be conscious. All of our third person data is calibrated on human verbal reports. When you have, the patient or or the the the subject say that, well, I had no experience... I had no conscious experience of learning this or I was anesthetized or something. Yeah. You've talked to the to the language user within the body, but, you have no idea what's going on with, with some of the other systems, and what their experience might have been. We simply have no no, negative control for for for this kind of thing. So, I I suspect that consciousness is a property of patterns. It isn't a property of of physical objects. It's... Consciousness is what we call the perspective of of patterns like us, ingressing through physical space. I I don't know whether they can have, experiences, in the absence of ingressing into physical space, but but I think that's what we call conventional consciousness. I think that theories of consciousness should output protocols. In other words, not predictions of behavior or physiology because then they're theories of behavior and physiology. I think a good theory of consciousness would output a protocol that puts you in first person in the same state... In the same cognitive, conscious state as some other systems, so then you really find out what what it's like. And that means at at least, at the very least, devices, for communication and some kind of mind sharing, with other unconventional systems. We have to exercise our, our imagination and the ability to... You know, when when we say, well, it's not... It's it's not anything that it's it's like to be, the, you know, sort of the the criteria for consciousness. It's something to be like, We have we have terribly limited imagination about what it's like to, to to be something else or or even someone else. And so I think we need devices. We need technologies to help us share minds with other beings. And I think a successful theory of consciousness should turn facts about a system into ethical imperatives. In other words, it should it should make the Humean bridge between the is, the description of a system and what's what's what's going on with it, and the ought, which is the set of ethical imperatives or how are we supposed to relate it, what do we owe that system, how... What do we treat it. That... That's what a good theory of consciousness should do. It should it should let us look at a system and what we know about it and and generate a set of a set of moral moral claims about how we're supposed to relate to it. And I think that's gonna be critical because, all of the standard of, Darwinian endless forms most beautiful are like a tiny corner of this possible space of of bodies and minds. And whether through regenerative medicine and then freedom of embodiment as we learn to communicate with ourselves and change our bodies and, you know, going into the future, humanity certainly not gonna keep the same hominid, form factor that it's had. But pretty much every combination of evolved material, engineered material, and software is some kind of viable agent. There are already cyborgs and and and early early stage chimeras of every kind. And and all of them are going to be hosts to various patterns of of goal states, of cognitive propensities, of preferences, and so on that we... That we simply have no ability to predict at this point. And what we want is to have a a nice ethical symbiosis with with all of these beings going forward. And for that, we really need to understand better what we are and, what the possibility space looks like. So I'll stop here. I'll thank the people, who did the biology that I... And the computer science that I showed you today. Lots of amazing collaborators. Some of these folks might be might be in the audience. And then I have to do a disclosure. There are four companies that have licensed some of the discoveries that have come from our work. So thank you so much for listening. I'll stop here.