Watch Episode Here
Listen to Episode Here
Show Notes
This is a ~55 minute conversation with Michael Johnson (https://t.co/YxAOZif0V2) covering topics of spectrum of consciousness down to the cell (and below) level, implicit memory and cellular pixels of experience, Platonic space, symmetry breaking, and intrinsic vs. extrinsic motivation in systems.
CHAPTERS:
(00:01) Framing consciousness science
(03:25) Are cells conscious
(09:30) Cells as qualia pixels
(14:41) Platonic space and chemistry
(24:05) Bioelectric symmetry breaking
(30:16) Structured water microstates
(39:09) Intrinsic motivation and play
(47:20) Symmetry theory in biology
PRODUCED BY:
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.
[00:01] Michael Johnson: Today I was interested in broaching a new topic. And thinking a little bit more about consciousness. We've discussed this a little bit. We talked about vessel computation last time. And I guess what I want to say today is maybe a potential meet-the-middle approach or merging what I would say Michael Levin thought with Michael Johnson thought and see where that can go. Sure.
[00:42] Michael Levin: That's great. Yeah, let's do it.
[00:44] Michael Johnson: Awesome. Just a few. I just have some notes here. I'll read from them. Just a few notes on my approach to consciousness: dealing a lot with formalism, structuralism, symmetry, valence, physics, and what I'm calling strong monism. And then a vasocomputation as the neural system and the vessel muscular system coordinating on patterns. A big theme is, how do we get to a proper science of consciousness? Maybe there are certain levels of organization that have somewhat unique affordances for understanding consciousness. To put some words in your mouth here, I read your work, and there's this beautiful multi-scale approach to everything. Collective intelligence and diverse systems having agents and goals. I've heard you mention polycomputing and everything is doing processing. Stressors, surprising competencies, diverse systems having predictive world models, and this focus on emergence. I want to pause here. Anything else that you would add?
[02:37] Michael Levin: I can also talk about some of the latest things that I've been talking about as far as the role of Platonic space. But the majority of my work is not about consciousness per se. I've been talking about it more recently. This morning a new talk has gone up, which was a talk I gave at a consciousness conference last week. I've said a few things about it, but I haven't made any strong claims about it. I don't yet have my own theory of consciousness to put out there, but I do think about it. I'm happy to play off of whatever you want to say about it.
[03:25] Michael Johnson: I want to talk about whether cells are conscious. You've dug pretty deeply into the biochemistry and electrical profile of other cells. One claim that I'd make here is that there are a lot of different theories of consciousness out there. That approach: do systems have a world model? Do systems have integrated information, Markov blankets, quantum coherence? Are they EM pockets? What is their shape in branchial space? Cells are this interesting system where any theory of consciousness you'd come to the topic with checks the box. I think cells are conscious, and I think you think cells are conscious, and it would be interesting to explore the biochemistry of that.
[04:41] Michael Levin: I guess the first question we should talk about is, do you think that's a binary question that we're asking in terms of things? Things either are or are not conscious. Is that how you're thinking about it? Or more of a continuum view.
[05:01] Michael Johnson: That's a good question. I think of consciousness as linked to space-time and as having boundaries in space-time. And this may be a difference in how we approach this. My meet-in-the-middle approach is that our themes of unconsciousness would identify the cell as a plausible conscious system. I think there are going to be a lot of edge cases. And then we can pretty confidently say, okay, healthy neurons are generally pretty conscious.
[06:18] Michael Levin: I think that it is not a binary thing; the question is what kind and how much. I suspect that. What we don't know yet is to what extent consciousness tracks intelligence. They're not the same thing for sure. Anil Seth has this diagram where they're pretty much orthogonal. He's got two perpendicular axes for it. I don't know if they're completely orthogonal or if they tend to track each other. I suspect they do. I think any system that has goal-directedness and is putting forth effort to try to reach particular states as opposed to other states is going to have an inner perspective that matters. The way I like to think about it is that in certain kinds of systems, the criticality of looking at the world from its perspective is different. If you have a bumpy landscape and there's a bowling ball on this landscape, your view as an outside observer basically tells you everything you need to know. You know exactly what's going to happen as a third-person observer. But if you have a mouse on that same landscape, your view of the landscape is irrelevant. What matters is the mouse's view of that landscape, because it could be completely different. He might have been rewarded and punished at certain areas. He might have different attention or preference. So the degree to which you have to adopt the perspective of the agent in order to know what's going on is relevant to how much of a first-person perspective they will have. Being able to recognize that is a two-way IQ test. If we don't know how to take that first-person perspective, we're really bad at it. People argue with me all the time: your liver can't be conscious. I'm conscious. Nobody actually has a story to tell why the electrical networks of the liver are somehow barred from the things that they think the electrical networks of the brain are doing. There is no story like that, but everybody assumes there is. They take their native certainty about these things—the priors that we got from our evolutionary history—and people often mistake that for some kind of good argument. To the extent that cells navigate spaces with valence and reward functions, and they have all the same mechanisms and the same evolutionary history and the same kinds of behavioral repertoires that we see complex organisms doing, at least to a smaller extent, a lot of the same stuff shows up. I see absolutely no reason why you wouldn't think that they have a degree of consciousness. Personally, I think it goes far below that. I don't think you need to be alive or anything like a cell to get onto that spectrum. The important thing is I do think it's a spectrum.
[09:30] Michael Johnson: That makes a lot of sense. And I think that the perspective of the agent-like process matters. I think that's very right. I think there might be an opportunity to figure out what is the typology of a cells world model, a cell-state world model, what sorts of things cells might sense and build. The overall metaphor that I'm going to use is cells as quality pixels in our canvas of experience. As much as we are a conglomeration of cells, our experience is also a conglomeration of many cellular microstates. Then you can dig into what kind of values these pixels can take. And what's happening when a cell depolarizes? Maybe the intensity of the pixel is the beam of the cell. From an evolutionary perspective, cells are faced with many informational imperatives where they had to understand: is my environment dangerous? Is it acidic? Some cells in an organism specialized in detecting hydrogen ions in the environment. They turned into sour taste buds. Likewise, other cells specialized in detecting umami in the environment: amino acids with hydrophilic side chains that could be useful, nutritious. There's an interesting typology of cell microstates, which could very cleanly map to microsensations. If you categorized all the different cell types, you would get a list of different possible types of quality values. Does that make sense?
[12:20] Michael Levin: And these equilia values are equilia of the cell, or when you say they're pixels, you think they somehow add up to the equilia of the animal that they're part of.
[12:35] Michael Johnson: I think the cell actually is conscious of that. That is what the cell feels like. And then we are a superset of these cells?
[12:46] Michael Levin: I think it's reasonable. Certainly the first part is reasonable, trying to figure out what the world of a cell looks like based on the things it cares about in physiological space, in the transcriptional space. There's a long history of this business of the Umwelt, trying to get inside a creature's head by asking yourself what matters to it. So while I am a panpsychist in that sense, I don't think that we are trying to solve the combination problem here. That is, I don't think that our consciousness is some sort of aggregate or amalgam of our components' consciousnesses. I think the cells inside us have some degree of consciousness, the tissues and the organs do as well, and so do we. But at every point, I don't think it's created by summing up the parts. I think the larger scale allows a better interface for an aggregation of a more complicated consciousness that comes from this Platonic space. I think our physical bodies, including embryos, biobots, robots, are haunted by these patterns in the same way that triangular objects are haunted by the truths of mathematics that pertain to triangles and to prime numbers. So I'm not trying to do any kind of a summation of consciousness of the parts, but I do think that the cells have it and quite probably the components within the cells as well, from what we can see.
[14:41] Michael Johnson: So I want to talk a little bit about the platonic realm. To close this loop, I think that my expectation is that the body, if you look at it in four dimensions, three dimensions plus time, consciousness is, probably, dominantly affected by the EM field. And there is probably one biggest chunk of consciousness. And then we call that our consciousness. But there may be smaller chunks in four dimensions. For example, the liver may have its own pocket of consciousness, which we don't have direct access to. And so we can interface with, but not control. So I was thinking about the platonic mind hypothesis that we are tapped into this larger and more beautiful space of dynamics of possibilities of these platonic forms of shapes. And you've written about this. And I'm wondering, to what degree could they be considered symmetry groups?
[16:19] Michael Levin: My current model is that Platonic space has levels or domains, parts of which are occupied by things that we recognize from math. So this is where the low-agency version of the truths of number theory lives. And then there are similar regions for more abstract things, like symmetry groups, possibly the kinds of things that Plato and others talked about, beauty, that may be related. There are regions that are also occupied by more complex dynamic forms that we would typically recognize as behavioral propensities or kinds of minds. Are they the same as symmetry groups? I probably wouldn't think so. How do you see it?
[17:20] Michael Johnson: I think that past a certain point, you get to this; a lot of things work out to be equivalent. The Platonic shapes are equivalent to some mathematical classes and so on. I don't know whether to anchor this to symmetry groups or to a more general Platonic frame. To take this back to the sensation stuff, I'm in awe of chemistry. There's this mesoscale structure, where chemistry is not necessarily inherently in the laws of physics. It's emergent from the laws of physics. One of your colleagues, Eric Pihole, has this wonderful "Causal Emergence 2.0" paper talking about how real various things are and whether the whole can be more real than the sum of its parts. I would observe that chemistry is surprisingly real — maybe what really exists is electrons or fields or strings or strands; there are many approaches in physics to what really exists. But chemistry is surprisingly real. It's the way of coarse-graining reality that is sturdy, stable, predictive, descriptive. I'm always looking for what analogous structures could look like in consciousness. How can we coarse-grain sensations in a similar way? I've thought about whether there is a periodic table of qualia to be found. But the periodic table is based on this harmonic structure in valence shells. The move that I would want to make is something like the atomic sensations that we have as humans: sourness, bitterness, sweetness, smell of citrus. I think David Ginty has written about 15 to 18 different types of touch receptors and done some great work there. Can we understand each of these as a different symmetry group or symmetry-breaking event? And then, if we could do that, we could slowly build up this basic alphabet of human sensation based on what's happening in the cells themselves.
[20:46] Michael Levin: Have any aspects of chemistry or chemical reactions been analyzed from the perspective of causal information theory? Because we just did something like that that's coming out in a couple of weeks. Have you seen anything like that?
[21:07] Michael Johnson: No, I haven't.
[21:09] Michael Levin: What we did was, and this is Federico Pagosi's work in my group. I'll take a step back. When you have a rat and let's say you do some associative conditioning, so the rat presses the lever, gets a reward. We know that no individual cell has both experiences. The foot of the rat touches the lever, the gut gets the delicious sugar, but in order to have that associative memory, you have to be a collective intelligence. You have to have an integration that allows the rat to know things that none of the individual cells know.
[21:52] Michael Johnson: Yeah.
[21:53] Michael Levin: Clear enough. What I wanted to know was, does it work in the opposite direction? That is, if you train something, does it become more of an integrated agent by virtue of being trained? In other words, does forming new memories raise your causal emergence? And so we looked at it in the context of models of gene regulatory networks. So this is just chemistry. There's no cell, there's nothing. All there is is a set of differential equations that control how certain chemicals turn other chemicals on or off. That's it. We have a few papers showing that when you have a system like that, it can learn. It can do about six different kinds of learning. It can do habituation, sensitization, associated with Pavlovian conditioning. So what Federico did was he looked at a measure of phi D of causal emergence as we train these things. He found that these networks divide into several different categories. We don't have a good name for it yet. In some of these categories, the more you train them, the higher the causal emergence goes. It reifies the process of learning new things as a collective, reifies the agent as a collective intelligence. You can quantitatively watch it happen. I wrote a blog post about it, and at the end, I have a diagram of Pinocchio, and he was told, if you want to be a real boy, you got to go to school. There are some other interesting aspects to it. Chemistry apparently already has these features. We have some other stuff that isn't public yet that takes it one step further and looks at the origin of these things. I don't know what we could do below that, if there's anything at the particle level that could be analyzed this way. But the chemistry is already doing it.
[24:05] Michael Johnson: Right. Nice. It does seem you're putting some optimization pressure on the integration term, and what comes to mind is Zurich has this quantum Darwinism brain that physics is the product of some natural selection for patterns that can persist and copy themselves into the environment.
[24:32] Michael Levin: When he says physics, does he mean specific physical phenomena or does he mean the laws of physics? Is he talking about a small, multiple-universe thing that's Darwinian, or does he mean within our universe the physical instances are trying to persist?
[24:51] Michael Johnson: I believe his work deals with the patterns in our universe.
[24:55] Michael Levin: Okay.
[24:56] Michael Johnson: The formal term would be motifs in the Hamiltonian, although I wouldn't want to define that. One thing that comes to mind here: there are always questions of multi-scale. The next question is what are the sweet spots to coarse-grain the system on to say this is a really interesting phenomenon that doesn't necessarily happen in the same way at other scales, but it does happen at this scale. My attention on the cell is this ping-pong between hyperpolarization and depolarization: neurons, muscle cells, and some immune system cells oscillate between a polarized and depolarized state. This is a very useful computational thing. There are cool things to do with how muscles move and how neurons fire. Movement and communication arise from this dance between the charged state and the state which happens when that charge gets released and collapses. The question is what happens to the cell's internal structure when it depolarizes. To say a few words, I'm looking at depolarization as a symmetry-breaking event: you pump energy into it and create some symmetries, and that, for most cell types, is the neutral state. Then you break that symmetry and it, maybe physically but more so electrically, collapses into a higher-entropy, more directional state. I'm wondering what your intuitions are in terms of if there's some origami — the cell is origami — and if you pump up its energy, it unfolds. When you release the energy, it collapses. What's that look like?
[28:01] Michael Levin: That's an interesting way of thinking about it. What we see from our work in non-neural cells is that the voltage change is slow and gradual. All of these things are relative because it has a unit associated with it, so they seem slow or not to us, but it's all relative. But slower than what you see in neuroscience. The symmetry breaking that we see is spatial at the level of a multicellular collective. You have an initial homogeneous pattern of cells, and you can set up local amplification–long-range inhibition loops that break symmetry: a certain cell will depolarize and become an organizer, and it will automatically tell everybody else, you don't do it, I'm doing it. It will suppress everybody else. Much like with Turing patterns, you can have symmetry breaking and spontaneous pattern formation in electrical networks with no underlying hardware differences, purely at the level of the physiology. We see that as a multi-scale kind of thing. One thing that I've always wanted to do, and I have a student that's going to try, is do some of the voltage mapping. We've already mapped within individual cells that the voltage is not homogeneous. We already know there are patterns within single cells, but most single cells are featureless in the plane. What I want to do is examine some very highly patterned cells, some ciliates — paramecium, Lacrymaria — this kind of thing that has very complex patterns. What does the voltage look like within a single cell? Are there regions? I'm almost certain. We did a little bit of Stentor in an old Danny Adams paper from my group a while back. But there needs to be a lot more of this.
[30:16] Michael Johnson: That seems really interesting. I think one question that comes up in thinking about this a lot is how do you proxy the internal structure of the cell and what do we even mean by internal structure? Just in terms of cell membrane polarity, Nick Lane has some great pieces. I think he gave a talk about "what is a feeling" in biophysical terms and talked about different places on the membrane corresponding to how the cell might feel. I thought that was a really clever approach. Another cluster of ideas, I've been speaking with Ben Anderson and his team and a friend, Nick Ford, about this a lot. It's this idea: is the water within the cell structured? This gets into Gilbert Ling's work, Albert St. Georgie. This could be an interesting proxy for what else is happening in the cell, but it also could be causal. It should be a very sensitive thing. Have you spent much time thinking about what could be happening with the water and the hydration shells around proteins?
[32:08] Michael Levin: No, I haven't. It's certainly an interesting thing. Jerry Pollack has written about this stuff a lot. I'm sure there's something to it. We have not studied it much.
[32:20] Michael Johnson: Okay.
[32:21] Michael Levin: It's just beyond. I've got my hands full at this point with all the stuff we do, and I don't have any expertise in that. There are a number of people looking at it, and I'm sure there's something there.
[32:37] Michael Johnson: Martin Picard has also written about cristae alignment. I might be pronouncing it wrong, but basically how mitochondria in the cell align or can get disordered as well. My optimistic hope here is that a lot of these metrics might overlap, that if you can measure cristae alignment, you're also proxying water structure, and you're also proxying EM fields, and you're also proxying anything that matters. But that's very weakly held.
[33:21] Michael Levin: I tend to think that pretty much all the materials inside a cell are being hacked by all the stuff around them, they're being used as a memory medium, they're being manipulated, and conversely have their own some degree of an agenda of what they're going to do in terms of the various ends, the goal states they're trying to achieve. I would think that water was probably part of that.
[33:50] Michael Johnson: To try to say something real about sensation and cells and whatnot. I would be hopeful that there could be low to mid hundreds of possible ways that cells can depolarize or collapse into a lower charge state.
[34:30] Michael Levin: When you say low to mid hundreds of ways, do you mean the channels that are causing it, or do you mean the specific physiological states that they can then occupy?
[34:41] Michael Johnson: The specific physiological states will definitely be correlated with the channels.
[34:47] Michael Levin: If we pretend the whole membrane has one value, then you're talking about a scale that goes from roughly 0 to roughly minus 80. As far as we can tell, the cells are only sensitive to plus or minus 5 millivolts; any given cell is probably not going to read any finer than that. So that tells you that you've got a small number of tens of distinct states. However, the cell membrane is not a single value. When we've looked at it, the domains that can be different voltages are about two to five microns in size. Potentially, a cell could be a soccer ball of different polygons on it. So that's a lot more. That would be probably in the thousands.
[35:45] Michael Johnson: Yeah. Interesting.
[35:46] Michael Levin: Because my guess is we don't know how finely cells react to that. How finely do they read that whole manifold? But my suspicion is that it can matter that there is a code there that it can interpret.
[36:03] Michael Johnson: The follow up question there would be, if there are dangers to cells, if there's some acid or there's a predator, there's some bad condition somewhere or there's some good conditions close by. What components of cells would the cell want to be very protective of? Just to tell this very simple story with water structure, Ling talks about how the water in a cell is structured around proteins and these proteins get unfurled and then water, being a dipole molecule, attaches to the charge sites and then other water attaches there, and they hold the dipole, such that it's a little bit more polarized. In theory, you can get chains of water molecules hydrating proteins. Ling thought of this as the living state, and it's a delicate balance. If you have hydrogen ions bumping into this, it would disorder this system. It would lead to symmetry breaking of this water matrix in a specific taste or flavor. Likewise, you'd have amino acids with hydrophobic side chains, so things that taste bitter. If this bumped into this water matrix, it would also disorder, it would also lead to symmetry breaking, but in a different motif with different flavor. I'm looking at cell microstates as corresponding to various symmetry breaks of this water matrix. Now, this is very loosely held. I think a big question is what is the cell trying to preserve? What is the cell trying to protect? This is one candidate. It's not the only candidate. But sensory states will revolve around the core things that the cell wants to maintain, we can say.
[39:09] Michael Levin: There's another issue here to think about, which is induced versus intrinsic motivation. In our work on sorting algorithms, these are short deterministic algorithms to sort numbers. What we found is that there's the thing that we make it do via the algorithm, which is to sort numbers. Also, there are these weird side quests that it takes that are nowhere in the algorithm. They're not prohibited by the algorithm, but neither are they instantiated by it. I've been playing with this notion of the reward function that we force on it, but then there's the intrinsic motivation. So I think you can see that in biology too. Evolution would be the forcing function that tries to say, you have to do this and this. But in the meantime, there's some others that, as long as you don't interfere too badly with it, you're also free to do some other stuff. We're very used to looking at biological functions from that evolutionary lens and saying, "Okay, why is it doing that? That's got to be good for reproduction," or it's a side effect of something else that's good for reproduction. But I have a feeling that there's also a bunch of other stuff that things are doing; even very simple things are probably doing other things that I don't think are coming directly from any of their experiences in the physical world. Whether we can flesh out our theory of Platonic space for it yet, I don't know. Some of the stuff we see the Xenobots and the anthrobots doing: they do things that were clearly evolutionarily important for their primary lifestyle, but when you take them out of that lifestyle, you get to find out what it would be doing if the other cells didn't force it to be a two-dimensional skin layer on the outside of the embryo. And normally all that is suppressed. Forcing a kid to sit in class and do math, you don't get to find out what else you'd be doing if you weren't doing that. But if you let up to some extent, then you get to find out what the intrinsic motivation is. And then possibly you work with that. There are educational philosophies that target that as opposed to trying to do a strict reward function. So I wonder, when we look at these cells, how much of that relates to some other conversations I've had with other people about whether problem solving—when we study intelligence, I define it as problem solving, goal-directed problem solving. But that's just for convenience. There are other aspects of being cognitive that have nothing to do with that. There's play, exploration. We all know what it looks like when birds and mammals play. You can see crows sliding down roofs on little flat things they've found somewhere; they're just having fun. It's not anything useful that they're doing. So the question is, what does it look like when cells do this? There is the evolutionary extrinsic motivation: you have to keep your pH at this level. If you don't do that, you're going to die. But alongside that, what does play look like on the cellular scale? What else are you doing? Some people say cells are too simple to do that. If bubble sort can do it, I'm pretty sure cells can do it. I think we're just bad at noticing it. We need to develop tools to be able to recognize play and exploration in unconventional embodiments.
[43:35] Michael Johnson: That's great. I'm such a fan of your work on Xenobots. Good stuff. I was thinking about what is now if we take the perspective of cells as qualia pixels, although I keep wanting to use the word quaxels as qualia pixels, I'm getting some pushback on that. Then taking a look at what does a lone Xenobot look like as a dynamic qualia pixel? How does its value change in different environments? Is it this unitary pixel — there's the Xenobot, there's the cell, and it has a value. Or is it heterogeneous in that we could think of its mitochondria as its pixels. I don't have a clear answer there.
[44:45] Michael Levin: I don't know. We'll have to see to what extent we end up needing to solve some kind of a summation function or not, or whether it's just completely different types of consciousness that show up when you make a particular interface. I'm not sure how constraining the parts are.
[45:19] Michael Johnson: Yeah.
[45:21] Michael Levin: For what you get the causal architecture is clearly important in some way.
[45:28] Michael Johnson: This gets into questions of: Does consciousness require definite extension and location in space and time? Or can it be more of a logical, computational thing?
[45:48] Michael Levin: I tend to think that a particular embodiment of consciousness will have location in space and time. That location will be fuzzy to some extent because there is no unified, there is no indivisible intelligence anywhere. We're all made of parts, we're all collective intelligence. Once you get to electrons or something, I don't know what the deal is in physics, but it's going to be a little bit fuzzy. The other question this brings up is to what extent there are lateral interactions within the platonic space among things that are not currently coming through any interface. If they are not static, which I strongly suspect is the case, then there will be some sort of chemistry of patterns in that space that are doing things regardless of their connection in the physical world. Then how much spatiality there is — I can imagine that it's not spatial the way we're used to; it doesn't have a location the way we're used to, but it's almost like a content-addressable memory instead of a location-addressable system. Instead of saying this is where this information is, it's like, what is this information about? Then it must be somewhere near this other thing, which is about the same thing.
[47:20] Michael Johnson: That makes sense. One hope that I have for this analysis is that it's always a question for me. Nine years ago I came out with the symmetry theory of balance. Similar to what you've said, geometric frustration is real frustration. If we had a mathematical representation of an experience, the symmetry of this representation would correspond to the pleasantness of the experience. I wrote a book on this.
[48:06] Michael Levin: Yeah.
[48:08] Michael Johnson: And so it's speaking about a formalism of an experience and not necessarily making a big claim in terms of how to create the formalism, but if we had a formalism, how to interpret it.
[48:22] Michael Levin: Yeah.
[48:23] Michael Johnson: I'm always eager to try to apply it to biological systems.
[48:30] Michael Levin: Yeah.
[48:31] Michael Johnson: There have been a lot of questions about how to apply it to a brain or to a nervous system. I'm optimistic that it can be applied to a cell's symmetry group, although there's a big question of how to coarse-grain a cell's symmetry.
[48:52] Michael Levin: I think it'd be interesting to try to apply some of these things to data in, for example, transcriptional space, so omics data. What does symmetry and beauty look like in that space? We're already trying to think about what it looks like to have barriers. What does a mirror test look like in transcriptional space? It's very hard to think about these things because we're so obsessed with the three-dimensional world. But I feel like all this can be defined, and it would be interesting to see what symmetry breaking looks like in these other spaces.
[49:35] Michael Johnson: One thing that comes to mind is I do think that symmetry breaking is directional, which is a very useful property. You have the symmetries of a system. A starfish is a pretty simple example where its nervous system is a ring of neurons. And then symmetry is the success condition: homeostatic success. If a fish comes and starts nibbling on a leg, the symmetry gets broken in a way that the different parts of an organism can tell where the problem is. There's this lensing effect. The starfish can move or adapt to that. Once it's safe again, the symmetry gets restored. In terms of looking at transcription networks and so on, it does seem there's some sort of directional, high-information perspective that the system can take, such that if you have a symmetry and it gets broken, then every part of the system knows a little bit about where the problem is.
[51:06] Michael Levin: That's very interesting. So what I'm hearing is symmetry breaking as a cognitive glue. So non-locality, I can see how that would be connected to geometric frustration. And you'd have a propagation speed of light thing for getting it around. I think that's very interesting. I think that's worth more development: symmetry breaking as a binding set of policies that create the collective intelligence. I think that's a cool idea.
[51:58] Michael Johnson: I'm mindful of your time. Any other cool things to talk about?
[52:15] Michael Levin: I was looking at my notes. We covered most of what I wanted. Let's go off and think about this, the symmetry breaking business. We could look for it if we knew how to recognize it in the abstract, in its general form.
[52:35] Michael Johnson: Yeah.
[52:38] Michael Levin: Yeah.
[52:42] Michael Johnson: Just a few words on that. There are different numbers of symmetry in different dimensions. I think in 2D there are 17 wallpaper symmetry groups of different ways things can get flipped or rotated. Frank Wilczek has this nice definition of symmetry as "change without change": anytime you can apply an operation to a system but leave it the same. Then there's 300-plus symmetries in 3D and I think almost 4,040, although you might want to check my numbers on this. Go ahead.
[53:30] Michael Levin: So I was just going to say that by itself is one of these. People often ask, what do you mean by facts that don't have a physics? That right there, the number of these groups under various circumstances, that's just what it is. There's no fact of physics, there's no history, there's nothing that's going to underlie that as a more reductive explanation. It just is what it is.
[53:57] Michael Johnson: Yeah, totally, totally.
[53:59] Michael Levin: I think that's really interesting.
[54:00] Michael Johnson: I also see that the symmetries of a self-organizing system are kaleidoscopic grooves that the system can follow to get back into a state of order. Water has tetrahedral symmetry. That's exactly 24 symmetries. I think of them as grooves in a kaleidoscope that you can follow back to order. It would be interesting to do this analysis for your gene regulatory networks.
[54:52] Michael Levin: I think it definitely would. We need to think about what kind of data you need, what goes in, and how to do the calculation. That would be quite interesting.
[55:05] Michael Johnson: Yeah.
[55:07] Michael Levin: We have genomic data, there's transcriptomic data, we have electrophysiological data, we have lots of simulations, so we could look at it in silico animats.
