Self-Replicating Robots with Michael Levin, Professor of Biology at Tufts University

(Intro Narrator at 00:00:03) Hello, my friends. Today Joel is talking to Michael, professor of biology at Tufts University, and they discuss how Michael has learned to create reproducing robots from frog cells, what we can make these robots do and what they're already capable of, and how Michael's studies of cellular cognition are being applied to regenerating limbs. All of this right here, right now, on the Modern CTO podcast.

(Michael Levin at 00:00:34) Here we go.

(Joel Beasley at 00:00:35) This is the Modern CTO podcast. When I was meeting with the team, they showed me this video, and apparently it was living robots made of frog cells that were reproducing and running around. They were eating some stuff up. What is that?

(Michael Levin at 00:01:01) So this is some work that we've been doing in collaboration with Josh Bongard's lab at the University of Vermont. The idea behind it — so these are xenobots. The idea is to use cells extracted from various tissues. In this case, this was frog skin, to use them as a robotics platform for basically two reasons. One, to make useful synthetic living machines that are going to do interesting and helpful things, and also to learn more about where biological goals come from. So why is it that these standard cells with no genetic editing can do new things, can make new types of bodies, new behaviors, and so on? And the amazing thing is they weren't eating those cells that you saw. They were kind of in a field of loose skin cells. They weren't eating them. They were corralling them and pushing them into little piles because those piles make new xenobots. Those piles become new xenobots. So this was basically kinematic self-replication. It's kind of like Von Neumann's dream of a machine that goes around, collects a bunch of parts, and out of those parts makes a copy of itself. It was actually a kind of replication.

(Joel Beasley at 00:02:09) That's insane. That's crazy. So when I'm looking at that video, the xenobots are the clusters, right?

(Michael Levin at 00:02:20) Yeah. So the big, kind of light brown piles — the big light brown structures that are moving around, those are the xenobots. And that stuff that looks like sand that they're moving around, those are little tiny cells. And so what they'll do is — and this is a very short clip, but what they'll do, both singularly and as a group, is they will corral, like little bulldozers, they will corral those cells into little piles. And overnight, those little piles will further self-assemble into the next generation of xenobots, which guess what? Will then do the exact same thing, repeating the cycle again.

(Joel Beasley at 00:02:58) So simply putting these things into piles creates a xenobot. They're not doing anything to the material?

(Michael Levin at 00:03:05) Right. So here's the interesting thing. And one of the interesting things about this whole technology is that it really stretches our definitions of robotics, of engineering, and so on. Traditional engineering worked with passive materials. So when you work with wood or metal or something like that, it's on you to make it do everything that you want the machine to do. So the engineer does everything, the material does very little. It's basically just structural. And then we sort of a little bit after that, we had some computational materials that you can count on to do computations and so on. This is the next level of that. This is an agential material. These cells have agendas on their own, meaning that if they are outside of the instructive influences of other cells in the body, so if they're left to their own devices and they're pushed together into a particular density in a particular environment, they are able to form the xenobot. The material knows how to do that. So this is engineering using two strategies. One is subtractive, meaning that we didn't add anything to these cells. We didn't give them new genes, new nanomaterials, we didn't do any of that. What we did was subtract constraints. What are the constraints? The constraints are within their normal context in an embryo. If you ask, well, what do skin cells want to do? The typical answer is, well, obviously they want to make this boring two-dimensional sort of cell layer on the outside of the organism that's going to keep out the pathogens. Well, that isn't what they normally want to do. That's what they're sort of bullied into doing by the other cells. Right? These are instructive interactions that keep cells doing this. In the absence of all of that, you see what they actually want to do. What they actually want to do is get together and make these little proto-organisms. So that's the — so the first principle is engineering by subtracting constraints, and the second principle is taking advantage of the competency of your material. Right? So all of this works. It works for us to make it and it works for the xenobots to make it because the cells will take over. Once you put them in these little piles, they take over and they do what they do best. And that actually has massive implications for regenerative medicine, and we can talk about that. We've used that kind of principle to regrow legs and form eyes and various other things by taking advantage of the competencies of the cells and not trying to micromanage it.

(Joel Beasley at 00:05:19) So you've actually done that? You've done the regeneration of limbs?

(Michael Levin at 00:05:24) Sure. In a frog model, we have. Yeah. Yeah. Yeah. So we have several papers showing that you can induce regeneration of legs in an adult animal that normally does not regenerate legs by interfering with the signaling of the cells after wounding for twenty-four hours. That leads to a year and a half of leg growth. So during those first twenty-four hours, that collection of cells at the wound is going to decide what they should do. And you have the ability to intervene in that decision and push them one way or the other. Not micro-specify how to make a leg, not tell them what to make, but just literally push them and say, not scarring and not nothing, but actually make whatever normal organ goes here. And that's it. And then you don't touch them again for a year and a half and they grow a leg. Yeah.

(Joel Beasley at 00:06:12) You're really expanding my mind right now. And I talk to technology people that do crazy things all day, but this is exceptionally interesting. How did — have you had thoughts about this and how it relates to consciousness in general?

(Michael Levin at 00:06:28) Very much so. Yeah. So my lab works at the intersection of developmental biology, computer science, and cognitive science. And the reason I weave those three disciplines together is because I think that all of this is fundamentally an example of collective intelligence and problem-solving. So cells getting together to build one thing versus another thing is the navigation of a cellular collective intelligence through a kind of problem space. It happens to be anatomical morphospace. So meaning the shape of the space of possible configurations of a hand, a head, an eye, you know, whatever it's going to be. So to me, this is all part of — when somebody asks what kind of science I do, I still think of myself as a computer scientist. I think that what we're studying is computation. It just happens to be in a living medium, but that doesn't matter. What matters is we really want to understand how these very competent subunits, including cells and molecular networks and so on, scale up towards very large goals. In other words, build a limb or regenerate a limb if it's damaged or keep maintaining a limb or those kinds of things. And there are some amazing examples of very intelligent navigation of that space, meaning it doesn't just — you know, we're sort of used to normal embryonic development. You start with a frog egg and you get a frog, you start with a fish egg and you get a fish, and you get this idea that, well, it's a hardwired process that just sort of rolls forward, each step forces the next step and so on. That's really not how it is. It's a very dynamic process that can make up for all sorts of interventions. You can make the cells bigger, larger. You can add cells, remove cells. Heck, in the human, you can cut the early embryo in half and you get two perfectly good monozygotic twins. Why is that? Because each half embryo rebuilds the other half. Right? So all of these things are — you know, this is incredibly plastic activity that meets William James' definition of intelligence, which was the ability to get to the same goal by different means. So this has — I mean, I don't say too much about consciousness per se, but I say a lot about cognition and the idea that what you're looking at is a collective intelligence solving problems in a space that you're not used to looking at. Right? We're used to looking at three-dimensional space and animals solving problems in 3D space by moving around. Right? You see, you know, monkeys, dogs, octopus, whatever is going to do these things. But actually, there are all kinds of intelligences in physiological space, in gene transcription space, in anatomical morphospace. So this problem is very much tied to cognition.

(Joel Beasley at 00:09:06) Have you been able to identify the method of communication they use?

(Michael Levin at 00:09:13) So, yeah. So they probably use many, and there are many more, I'm sure, that remain to be discovered, but the one that we really work on is bioelectricity. So all cells, not just your neurons, but all cells communicate electrically. I mean, if you think about — we are also collective intelligences. Right? So people often say, well, this termite mound, maybe that's a collective intelligence, right, or this flock of birds or something is a collective intelligence, but I am a centralized, you know, true intelligence. No. There's no sharp distinction there because you and I are also bags of neurons, and these individual cells still have to cooperate together in a very particular way to give rise to this singular, large-scale intelligence that's going to have memories, goals, preferences that don't belong to any of the individual pieces but belong to the whole. That scaling problem is just as important for us as it is for, you know, termites and ants and so on. And so you might ask, okay, how do brains solve this problem? Right? Brains solve it by bioelectricity. So neurons connect electrically, and they make these networks that process information in a way that gives rise to behavior and sentience and everything else. It turns out that — perhaps, you know, looking at it backwards, maybe not so surprising, is that brains didn't invent this trick. So how did brains learn to do this? Well, they exploited a system that was here long before even neurons appeared. So back around the time of bacterial biofilms, evolution discovered that electrical networks are a super convenient way to process information. They're very good for integrating into computations, for storing memories, for decision-making circuits, for homeostatic circuits, and so on. And evolution's been exploiting that ever since. So all of the decisions of developmental biology, remodeling, cancer suppression, maintaining and repairing organisms and regeneration, all of those things involve a lot of bioelectrical signaling. They also involve, of course, chemical signaling, biomechanics, you know, ultra-weak photons. There's a lot of physics that goes into it. But the bioelectrics is kind of special because just like in the brain, it's the medium of the cognitive activity. Right? If you want to know how do you read out the information in the brain, the commitment of neuroscience nowadays is that, well, you need to track the electrical activity and decode it. So the same is actually true for the rest of the body. If you want to know how cells know what to make, part of that answer is it's in the electrical network that they form with each other. It's those pattern memories in that electrical network.

(Joel Beasley at 00:11:42) Do you think they have some sort of encryption to the host in the sense that, like, my cells are going to know to work for what I am versus someone else sitting next to me?

(Michael Levin at 00:11:54) Yeah. There's two — there's several senses of encryption here. On the one hand, it's very likely, although we still don't really understand this very well, but it's very likely that there is a degree of encryption going on because such a powerful system is really a great target for parasites, for various evolutionary cheaters, for different ways for other animals to exploit you, bacteria and so on. And in fact, we've published that the number of heads that a planarian flatworm is going to grow is partially dependent on what the microbiome says. So the microbiome actually has input into that process. Right? So clearly, there's some sort of arms race whereby these various microbes are learning to hijack some of those native control principles, right, this bioelectrical network. So I'm sure the host probably is fighting back, making it possible, but not too easy to control itself using those tools. So I'm sure there's some kind of evolutionary arms race there. With respect to, you know, you and your neighbor, I mean, the reality is cells are incredibly interoperable. So you can make chimeras. You can make chimeras across multiple species. So in our lab, one of the things we make is frogolotls. So partly frog, partly axolotl. Those embryos work together just fine. The cells work together just fine. And in fact, people, you know, people instrumentize with inorganic electrodes and weird materials. Right? You know, all kinds of nanomaterials and all kinds of stuff. Cells will work with whatever they have. So, you know, that signaling may allow them to know whether they're sitting next to something that doesn't belong there, but they will certainly get along with it if they can.

(Joel Beasley at 00:13:36) This is some pretty intense stuff, man. This is — well, I — oh, wow. Yeah. And you've been studying this for how long?

(Michael Levin at 00:13:46) Well, that depends when you start counting. I mean, I've been thinking about these things since I was a kid. I was always interested in these kinds of questions of mind and embodiment and control and computation and so on. I did computer science first, and I was a programmer, and I got a degree in computer science and then biology, and I got a PhD in biology. And I started my own independent group in 2000. But I've been working on this stuff, you know, one way or another for many, many years.

(Joel Beasley at 00:14:17) That is so cool. I'm very glad that we've got smart people like you out there changing the world. The grey goo scenario from sci-fi, basically self-replicating bacteria turns us all into this giant grey goo thing. How do we make sure that doesn't happen?

(Michael Levin at 00:14:36) That's a great question. I mean, to be clear, the grey goo scenario is more for nanorobotics. It's not so much for the biological stuff that I deal with, so I'm in no way an expert on that. I can just, you know, sort of thinking about it generically, I'm not sure there's any way to make sure of anything. In other words, you know, how do we make sure we don't blow up the planet with nuclear weapons? I don't know how you make sure of that. I'm not sure there's a way to make absolutely sure of that. There are ways to, you know, sort of try to work towards a better future. Just even in the absence of gray goo, how do we make sure that we don't wipe ourselves out with some sort of crazy new virus or new bacteria that somebody creates somewhere? I'm not sure how — I don't know how you make sure of that.

(Michael Levin at 00:15:25) I think there are bad actors that will try to do it on purpose. I think it's actually not that hard. And so that's a very dangerous thing, much more dangerous than, let's say, conventional weapons or even nuclear weapons. I don't know how you make sure of that because all of these things are progressively easier and easier for people to play with. In other words, it's not just, okay, there's one central lab somewhere and those are the only people that can do this.

(Michael Levin at 00:15:55) All of this kind of stuff becomes easier and easier to access. So I don't know. I certainly don't have a solution to it, but I think that we have to be clear that making rules against it is okay, but it doesn't go terribly far because the people that are really interested in wrecking things are not interested in these rules. They're just not paying attention to these rules, right?

(Michael Levin at 00:16:16) So I'm not sure. I think I'll tell you what I think we absolutely need to have. The answer to this is more science, not less science. Because oftentimes people will say, "Well, this sounds scary, and I'd like to avoid these things happening, so let's not investigate them." And I think that's exactly the wrong approach because somebody's going to do this anyway.

(Michael Levin at 00:16:38) And even naturally, right, as we see, there are natural plagues and all kinds of natural disasters that are going to come along. If we do not have a scientific understanding of where these complex systems come from, how they evolve, what they're going to want to do, whether they be social structures or financial structures or Internet of Things or swarm robotics—you know, we really need a mature science of the scaling of basal cognition. How do these things scale into larger systems, and what do these larger systems want to do? We need a science of this.

(Michael Levin at 00:17:12) That's an existential-level need for humanity.

(Joel Beasley at 00:17:15) What's the name of that science?

(Michael Levin at 00:17:17) It's a good question. I mean, I don't know that it exists yet, really. People have called it complexity theory. People have called it emergence. I'm not sure it has a great name yet.

(Michael Levin at 00:17:30) I think it's basically this is a really emerging new field where people have known for a really long time that lots of simple little subunits performing simple local rules can give rise to something very complex. That's been known for a really long time. A really simple example of that that most people will recognize is a fractal, right?

(Michael Levin at 00:17:52) You have this beautiful—I mean, you've seen them—this incredibly beautiful image. That whole image is squeezed out of a little tiny formula that's usually just maybe ten, twelve characters long, right? All of that complexity in this tiny formula because the formula just tells you a simple rule.

(Michael Levin at 00:18:07) You iterate that rule and you get this incredibly complex thing. So people have known about emergent complexity for a long time. We all know that complex systems are difficult to predict. But there's actually way more than this going on, which is the actual scale of cognition, meaning that not only are these systems going to have behaviors that are hard to predict, but they're going to have goals, and they're going to have novel ways of reaching those goals. And that's a much more powerful level of activity, and we do not know where collective goals come from, right?

(Michael Levin at 00:18:37) Look at, just as a simple example, xenobots, right? There's never been any xenobots before, so you can't say that the reason xenobots have these features is because of evolution or selection to be a good xenobot. There's never been selection to be a good xenobot.

(Michael Levin at 00:18:52) So where does this come from, right? I mean, clearly they're exploiting some sort of laws of physics, computation, and mathematics to do this, but we are terrible at predicting these things. We don't know how to predict these goals, how to even recognize goal-directed systems. Oftentimes people will have this binary view where they'll say, "Well, these things have true goals and true cognitions"—and usually they mean humans, maybe some great apes, maybe some octopus, and they all sort of argue with each other about where it ends.

(Michael Levin at 00:19:24) And then there's all this stuff which is just physics. It doesn't have any of that. So I think that's a fundamental mistake, is trying to say that there's some kind of binary separation. I think that there is a massive continuum of very unconventional agents that have goals and solve problems in spaces that we're just not used to observing. And if we don't come to grips with that and develop this science, we're going to have very major problems.

(Joel Beasley at 00:19:49) I want to go back to the communication between the cells. What type of communication did you say that you could pick up on?

(Michael Levin at 00:19:56) So this is bioelectrical signaling. It's basically just like what's going on in the brain, but slower. And it's not specifically related to neurons.

(Joel Beasley at 00:20:05) Now, are you able to sort of isolate the xenobot and the material and track communication between the cells? Or is that too granular for what we can do today?

(Michael Levin at 00:20:20) No, we can do that, and we haven't specifically done that with xenobots yet. We've done it with all sorts of other—we've done it in embryonic development. We've done it in limb regeneration. We've done it in spinal cord regeneration, eye development, face development, cancer.

(Michael Levin at 00:20:35) We've done it in the context of tumorigenesis. Yeah, you can track—so we developed the first tools to do two things: track those bioelectrical conversations that the cells are having with each other, and rewrite those conversations, basically edit them on the fly to put in new information to change what the collective is going to do. It's a little bit like in neuroscience, they have these research programs for neural decoding. So can we read the electrical activity in your brain and see what you're thinking about, what your memories are, and so on?

(Michael Levin at 00:21:05) And then inception of false memories. So for example, there are researchers that can take mice—there's a group at MIT that did this—where you can take mice and you can use light, optogenetics, to write specific electrical states into their brain to give them memories of experiences they never actually had. I mean, that sounds kind of wild, but it should absolutely be possible if you believe that the electrical activity of your brain stores your memories. Then of course you ought to be able to write new ones that way.

(Michael Levin at 00:21:33) So we've done exactly that, but instead of the brain, we do it in the rest of the body to say, "Okay, you're a flatworm and your default electrical circuit says you need to have one head. But we can go ahead and manipulate that information and tell these groups of cells that now a proper worm should have two heads instead of one." And so when that worm is injured and it goes to regenerate, guess what it does? It makes two heads.

(Michael Levin at 00:22:00) So we've done that. We've made various other organs. Anyway, yeah.

(Joel Beasley at 00:22:06) So I want to share with you a quick story and you can tell me if it's true or not. Okay? So my dad is in the Air Force. They put the GPS system into the stealth bomber, so he had top-secret-type clearance. And he was always interested in that, being an electrical engineer and a software developer.

(Joel Beasley at 00:22:23) So he was always interested in the coolest things happening. And I remember when I was pretty young, I was probably, you know, early nineties, and I think we watched something on a show or a tape or something he had. And the experiment that they were running—I can't remember if it was fiction or nonfiction—but the experiment that they were running, they were isolating saliva cells, and they would take them out of the—you know, the person would spit. They'd run it through a centrifuge or something, and then they would have electrical monitoring on them.

(Joel Beasley at 00:22:59) And then they would tell the person to leave the room or go home, "We're done, we've taken your sample." And then someone would jump out and scare them, and they could pick up on the electrical signal actually happening in the vial of the cells that had already been removed from the person. True, or is that—was that something that you think is true, or is that too far from reality?

(Michael Levin at 00:23:17) You know, I've heard about those kind of studies. I've never done them myself. All I can say is that currently we don't have any known mechanism that would support that. I'm not saying it's impossible. I think saying that things are impossible is a really good way to turn out to be limited in imagination twenty years later.

(Michael Levin at 00:23:39) So I'm not going to say that it's impossible, but we don't have any way of knowing that—we don't have any theory right now that would predict that that would happen. Let's put it that way. So that's something that, you know, if that were true, it would be highly significant. Somebody should be researching it, publishing it, and so on. I haven't seen it, but I'm not going to say it's impossible.

(Joel Beasley at 00:23:59) So let's talk about wounds, right? You mentioned that you could track these conversations that they're having, sort of record them, and then manipulate them to change the instruction, which is great. And for background, I was a software engineer for seventeen years, and then this podcast got popular, so I haven't been writing code the past three years. But I spent a large part of my life doing it.

(Joel Beasley at 00:24:24) So I like the analogies that you're using. I like that you're calling these robots, machines, because it helps you take all the machine knowledge you have and apply it over. I really like that. But I kind of forgot what I was going to say.

(Joel Beasley at 00:24:41) Oh yeah, what's the distance with these wounds, right? So you have the cut and then you have all the cells that are right there. You're bleeding. There's cells all over you. What's the distance that they're affected? If you take one of those cells about three millimeters away, is it no longer connected to you? Or how does that work?

(Michael Levin at 00:25:02) Yeah. Distance, if you mean distance through the air, meaning physically disconnect them, then as far as we know, that's it. They're not connected. And again, there could be—there's data on ultra-weak photons and who knows what else, right? So I'm not going to say it's impossible. But with known bioelectrical communication, yeah, once you're physically disconnected, that's it.

(Michael Levin at 00:25:14) However, you can also talk about distance through the tissue. In other words, in your body, what do tissues know about each other across distance? So if you have a wound or a damage or an incipient tumor or something else going on, what other cells know about it?

(Michael Levin at 00:25:38) And so in our work, I don't know how this will play out in a human because we haven't done this on that scale. But in the frog model, I'll tell you two pieces of information. One is that if you inject an oncogene on one side of the animal, whether or not that thing makes a tumor has to do with the electrical conversations between those cells and their neighbors. You can affect that process by manipulating bioelectric states all the way on the other side of the animal. So it does not have to be local.

(Michael Levin at 00:26:07) You don't have to target the tumor cells themselves. The bioelectrics propagate across the whole animal. Now that's in the tadpole—that's not really that much distance. So I can't really say that it's going to go more than a couple centimeters, but it might be way bigger than that. We just don't know.

(Michael Levin at 00:26:22) In the frog legs, if you have a froglet and you amputate one of the legs, the other leg that you never touched has a bioelectric signal that kicks on within thirty seconds at the exact same place where the other leg got amputated. So the untouched leg knows where the other damage was. Not only does it know where, you can tell what kind of damage it was—basically a puncture wound or an amputation—based on tracking the bioelectrics in the other leg. So, you know, how far is that?

(Joel Beasley at 00:26:52) Stuff that we've already—

(Michael Levin at 00:26:52) That you could do today.

(Joel Beasley at 00:26:53) Without medically modifying the frog.

(Michael Levin at 00:26:56) Correct. And to be clear, it's not anything that we do. It's what the frog does, right? We haven't done anything. We've just discovered the fact that the tissue in the frog already knows what's going on, right?

(Michael Levin at 00:27:02) So that opens the door for some interesting things, like surrogate site diagnostics and who knows what you can tell about other tissues from looking at a particular location in the body. Again, it has to be done in humans to figure out what the actual length limit is. Maybe it is just a few centimeters.

(Michael Levin at 00:27:21) But what we see in our frog models is that information spreads very far. And again, it shouldn't be terribly shocking in the sense that that's what bioelectric networks are for, right? They're for integrating information across spatial distances. They're really good at that.

(Joel Beasley at 00:27:36) I want—dude, this is fun. It is really cool to get to talk to people like you. I really am interested now in somebody writing more about the science of the collection of things because I find some of the things interesting will be thoughts that I just randomly have. Walking around in life, thoughts that I've randomly had, and then somehow later feeling like they're sort of connected to areas that are already being researched. And I'm like, "Well, what prompted that?"

(Joel Beasley at 00:28:11) Was it simply just a likelihood of me consuming other information? Or, you know, is there some sort of field of consciousness? As you make these discoveries locally in a frog that it can do these things, and we know in a fractal nature, it's turtles all the way down, right? There's the ants, and then we're just a collection. It's like, what are the humans building? What's the goals of the humans and what are they doing and all of that?

(Joel Beasley at 00:28:35) Have you—do you think that—your discoveries, as you progress for your time here on Earth, do you think everything that's happened has led you to believe that the world's more magical or less magical in the sense that there's more undiscovered things? Or what are your thoughts there, if that made any sense?

(Michael Levin at 00:29:02) Yeah. Well, I hear two questions there. At the end, I think you're asking, you know, in terms of what percentage of the reality do I feel that modern science has got a good grasp on, right? How much is there to discover versus do we really pretty much just already have it nailed down?

(Michael Levin at 00:29:23) That I can answer, and I will say that my own feeling is that we know almost nothing. I think this idea that we have that, you know, we got these textbooks and things seem to be under pretty good control—I think we are just at the beginning. I think there are such profound areas of ignorance in many, many areas that are really important that we are going to be—you know, this is nowhere near the end of this journey.

(Michael Levin at 00:29:50) We're much closer to the beginning than we are to the end, I feel that. And that's clearly not—I mean, people write these books, you know, "The End of Science" and things like that, where they kind of feel like, "Hey, you know, we got the basic bones of everything, and yeah, there's some details to put on things, but we kind of understand how the world works." I don't think that's true at all. And so I can say that.

(Michael Levin at 00:30:11) With the first part of your question about the magic—well, it depends how you define magic. If you define magic as some sort of capricious breakdown of the world order, no, I don't see any evidence of that at all. I don't feel that at all. However, to me, where the magic lies—I see things that I think are magical every day, and they're not magical because they somehow break down logic or science or they can't be explained or any of that.

(Michael Levin at 00:30:37) I think all none of that stuff is useful. I think what is absolutely magical is the way that it is—it's many things. It's first of all, the fact that the causal structure of our mind in some way matches that of the universe so that we can make heads or tails. I mean, that to me is the most basic first piece of magic right there, right? That we can even begin to make heads and tails of what's going on. And the amazing—you know, there's something magical going on because when you find out certain things, they lead you to unearth other things that you didn't know about before. Right? So these are not invented. They're discovered.

(Michael Levin at 00:31:20) So whether they be theorems of mathematics or, you know, other things, you start off with something, and before you know it, it pulls you in towards finding something else. So it isn't that you just sort of came up with something that was expedient for you at the moment. There is some sort of a grand design behind it that is actually—once you start pulling on the threads, each thread leads to some other thread that you didn't know was there. And so the interconnectedness or consilience of these things, as some people call it, I think that's magical. I think the fact that we can make progress using these rules of logic, the fact that—what I also think is magical is the fact that to me, intelligence, not in a mystical sense, but in a, you know, sort of cybernetic problem-solving sense, seems to be baked into the universe all the way down, far below cells and things like that.

(Michael Levin at 00:32:15) And that to me is magical, the fact that we know what it's like to be a giant collection of cells. What's it like to be a swarm of cells? Well, you tell me because we are, you know, you and I, that's what we are. And so the fact that you can actually arrange matter in a particular way to make cells that solve problems, and then arrange those cells in a way that you have the centralized intelligence that has a first-person perspective that has preferences and goals and so on—the fact that that's even possible in the physical universe, that's magical to me.

(Michael Levin at 00:32:49) And it's not that, you know, not in the sense that it can't be explained or anything like that. I think every time we explain something and we get better insight into it, the magic goes up for me. It doesn't go down. You know, some people kind of feel that science squeezes the magic out of things. I think it's the exact opposite.

(Michael Levin at 00:33:04) I see more magic every day. The more I understand what's going on and what, you know, my colleagues and other people discover things, the magic rises. You know?

(Joel Beasley at 00:33:13) Oh, yes. I think we have similar views of the word "magic." Right? It's like the feeling that you get that's like awe. Yeah.

(Joel Beasley at 00:33:22) You mentioned something else. Like, so as you make these discoveries, like, more things seem possible and it gets exciting. And, you know, you said something—you said that a lot of people will think it's like sucking the fun out of it. Right? Well, another thing that I've come across a lot is I feel like a lot of people think that people that are in science, like, don't believe in God. Right? And I know I'm using God very loose term. It's highly subjective when I use that word, but I'm just gonna say the thing that is. Right?

(Joel Beasley at 00:33:56) Like, at the lowest level, like, I think it could be, like, a variety of things. Right? But we see this organization happening, and it seems to be more than just chemical reactions. And so I'm just curious. Like, when I see science making progress, I think it's almost like the act of discovering God.

(Joel Beasley at 00:34:14) And I'm using God as a placeholder for the thing that created this universe. Right? It's like our origins. And to me, that's really exciting because, uh, I think there's a lot of—well, obviously, there's a lot of mystery to life. And I just thought it was kind of funny how a lot of people will look at science as a reason to excuse God away, and I'm sitting there thinking, like, it's quite the opposite. They're trying to figure out creation.

(Joel Beasley at 00:34:41) And through that, understanding how we fit into this container and what this container is and how it was created.

(Michael Levin at 00:34:49) Yeah. Um, I think that it really depends on, obviously, your view of what you mean when you say "God." Right? So if people—some people mean by this an intelligence that's above the level of humans with whom you can have a personal conversation. You can ask for favors. You can ask forgiveness. You can whatever it's gonna be. Right? I don't see science providing any evidence for that level of activity. However, looked at a different way, all science begins with an act of faith.

(Michael Levin at 00:35:23) It has to. At the bedrock is a very basic piece of—is a very basic belief that you have to take on faith if you're gonna do science. That is the fact that the world is understandable and it's out there for you to rationally uncover. Because there isn't any actual reason for that to necessarily be the case. I mean, how would you know that that's the case? It's an unprovable assumption. You cannot—so if you don't believe that, you cannot do science. Science is only done by people who are who firmly believe that the world operates according to understandable principles. And by the way, this weird, sort of, you know, monkey-like ancestor happens to have a cognitive system that's going to be able to understand it. That's a lot of stuff to take on faith.

(Michael Levin at 00:36:09) Right? But let's be clear. You know, scientists often will say, "Yeah, I'm not into faith. I'm into experiment." That's great, but all of it rests on that one fundamental assumption. And I think it was—I'm trying to remember who specifically it was. It might have been Boltzmann or it might be older than that. But he basically pointed out that let's just imagine for a second that there were no laws in the universe whatsoever. Right? There were no laws of physics. It's just pure randomness, no rhyme or reason to anything. If that's the case, and it lasts an infinite amount of time, then guaranteed—mathematically guaranteed—there will be pockets of time during which things are happening that just happen to look like they're lawful. Right?

(Michael Levin at 00:36:51) The way that if you keep tossing a coin, eventually, you're gonna have, you know, a run of 20 heads. And if you just look at that run, you might say, "Well, I see the rule here. The rule is always heads. That's the rule." And what you don't realize is that there's just a little pocket of order in this incredibly random stream, right, where you tossed it, you know, billions of times. So, uh, whoever it was, probably Boltzmann, was arguing that we can't really detect—if you look around and you see things going according to the laws of physics for some amount of time, you actually don't know that that's what the universe is doing. It could just be a random pocket of order in an infinite series of completely random happenings. So in order to do science—and if that were true, you can't do science because like those 20 heads in a row, they could end at any point. There is no law.

(Michael Levin at 00:37:37) There's no law keeping it that way. So at any point, you might get to the end of the sequence and then whoop, that's it. It's gone. So in order to actually do science, you have to have an act of faith. You have to believe that we are not existing in a bubble of events that look like they're lawful, that the universe actually has laws.

(Michael Levin at 00:37:56) Because you're gonna put a lot of blood, sweat, and tears into trying to figure those laws out as a scientist. So all of that starts with an act of faith. And I think this is what—I'm no expert on this, but I think this is what people like Einstein and many other folks like that, when they said that they had some, you know, sort of deist sensibilities, I think this is what it is. It's this faith that everything else is based on—that the world is rationally understandable. And it's quite a leap, actually, because it doesn't have to be that way.

(Joel Beasley at 00:38:27) Yes. Yeah. And I just—I think it would be pretty cool if one day, like, we started to get some answers to questions like, you know, I won't have talked to a person a long time, and then randomly, they'll pop into my head, and later in the afternoon, they call me. It's like, what are—like, you know, things that happen like that, um, or, you know, people have dreams about things that'll happen to them the next day. Like, there's things that happen that I don't believe are completely out of the realm of possibility.

(Joel Beasley at 00:38:58) Um, but I think as we learn—I think one of the keys for a while, and I didn't know this before this interview, that you studied this communication between cells in this way. Because one of the things I've thought for a while is, like, if you could study this and understand how they're communicating, you would have a better understanding of what's going on in the universe around us. Right?

(Michael Levin at 00:39:22) Yeah. Look. I think we have to be humble about the fact that—I like I said, I do not believe that we have a great handle on some of the really fundamental things that are important about our reality. Having said all that, I will just say that the current understanding of how things work doesn't really give—you know, it doesn't really support or explain the things that you were talking about. So currently, I don't believe—right. Currently, we don't have a framework for that. But I think we really have to be humble about, um, you know, how much have we really figured out about what's going on. And I think there are major gaps, you know, major knowledge gaps to where we may be surprised about some of these things.

(Joel Beasley at 00:40:08) Yeah. I see it as opportunity. And I try to be really objective with things—as objective as you can be. Like, my intention is that, but, you know, we're humans, uh, because I'm interested in answers. I'm not necessarily—I'm kinda like rogue. I'm not on a specific team. I'm more interested in, like, the truth of it. Right? Like, let's figure out what's going on.

(Joel Beasley at 00:40:31) And to be completely honest with you, I'm kind of surprised that it's not like a bigger thing happening in life. Right? Like, it's—I mean, if you look at the money that we spend on science versus other things, it's relatively—you know, it's not the most important thing if you look at our budget as humanity.

(Michael Levin at 00:40:49) Yeah. Yeah. Yeah.

(Joel Beasley at 00:40:50) Well, rough topic a little bit. Um, I do have a good question, though. Um, because you can measure these signals and because all of these electro things are happening, um, sort of unconsciously, for lack of a better term, uh, within ourselves, um, have you come up with figuring out what percentage of the electro signaling is conscious versus unconscious in our bodies?

(Michael Levin at 00:41:22) Well, um, I mean, conscious is an interesting question. You could ask that of chemical signaling. So the vast majority of the chemical reactions that go on in your body are not accessible to your conscious waking self. Some of it may be recoverable under altered states, let's say hypnosis or something. Maybe you would be able to give answers about, you know, aspects of your body physiology that you don't normally feel.

(Michael Levin at 00:41:49) I don't know. But the vast majority of it is certainly not conscious. I mean, it's funny. If I were to say to you, "Yes, you know, with my conscious intention, I can change the voltage of 30% of my body cells," and you might say, "That's a crazy claim." Right? And then I would say, "Well, uh, every time you lift your arm, you are consciously—if you're laying there in bed and you say, 'I'm gonna lift my arm' and you lift up your arm, what is that? That's your conscious intent, whatever that is, and there's no good consensus on what that even is. But whatever that is, your conscious intent has changed the electrical potential in your muscle cells that allows them to do the work of lifting your arm.

(Michael Levin at 00:42:28) So all of us, all day long, as we go about our lives, are using our conscious will to alter the electrical state of your muscle cells. Can you do that to other types of cells that are not muscle cells? I'm not sure. Some people—I've actually been contacted from my Twitter feed where I talk about some of the science. So people who actually contact me—apparently, there are people that can make the hair on their arms stand on end consciously.

(Michael Levin at 00:42:52) Right? Uh, so some percentage of humans apparently can do that. I never heard of that before, but it's possible. So what else, you know, is sort of in your standard configuration or with biofeedback training with some other kind of practices that people might be able to do. You know, I know that people who do yoga and whatnot can learn to control all kinds of other things.

(Michael Levin at 00:43:12) So the vast majority of it is unconscious, but the jury's still out on what can be brought under conscious control if you work at it. You know? There are limits, right?

(Joel Beasley at 00:43:25) Yeah. Like, uh, I can move my ears up and down, which a lot of people can't do.

(Michael Levin at 00:43:29) That's nice.

(Joel Beasley at 00:43:30) And can you see that?

(Michael Levin at 00:43:32) Yeah. Yeah. Yeah. That's awesome.

(Joel Beasley at 00:43:34) Can you do that?

(Michael Levin at 00:43:35) I cannot do that. Nope.

(Joel Beasley at 00:43:37) See? And it's like, I didn't even try. I just—people notice when I smile, they're like, "Oh, when you smile, your ears always go up. That's how I can tell if you're fake smiling or real smiling." It's like, my wife would tell me that.

(Joel Beasley at 00:43:46) And I was like, "Oh, really?" And then, like, flaring nostrils. So there—I think, you know, some things, like, we just for whatever reason, when we're becoming ourselves, like, through the cellular creation. Right? Like, as it's happening, uh, we get access to certain things, and, um, we're all just a little bit different.

(Joel Beasley at 00:44:08) And there's enough of us that a lot of us have some things in common that we can do.

(Michael Levin at 00:44:12) Yeah. And the question is too—I mean, a lot of it is incredibly plastic and reprogrammable. Uh, there's an old experiment in rats where if you measure the temperature of a rat's ears and you're rewarded for the temperature differential, then rats can generate up to, like, a five-degree Celsius difference between the left and the right ear because it's learned that by doing that, it gets a reward. Right? So we can learn to do that.

(Michael Levin at 00:44:36) That's one more behavior that it can learn to do. And humans with biofeedback can do that kind of stuff too. I've seen all sorts of things where people can learn to do these kinds of things. So the question is plasticity. Have you ever, another example is, have you ever seen that rubber hand illusion?

(Joel Beasley at 00:44:51) Yeah.

(Michael Levin at 00:44:53) Right? So it's an amazing thing where, you know, you do this thing where they put this rubber hand in front of you and they sort of pat it while they're patting your other hand, and then they take out a hammer and they go to whack it. You kind of jump. Your body, the structure of your body having four limbs, has been nailed down by evolution for how many millions of years? And yet twenty minutes visual experience is enough to completely override that and make you believe that you've got five limbs. Right? It overrides that. Twenty minutes of just watching somebody touch that rubber arm overrides this default expectation that's been set in the tetrapod body plan that's been set for millions of years. So that's, you know, the plasticity is incredible. And so that's why there are all these prosthetics, and we're gonna have all kinds of cyborgs and everything else, because the plasticity is just massive.

(Joel Beasley at 00:45:44) Yes. I want to quickly touch on commercialization. Does your lab have a desire to get these xenobots out, or to use this technology of limb regeneration for first aid type things in humans ultimately? Or how are you gonna ultimately make money?

(Michael Levin at 00:46:05) Yeah. Well, right. Making money. I think for sure we have, it's a really important goal of mine to have this stuff out to relieve human suffering, medical issues. So birth defects. So the important thing is that all the stuff that we study, including xenobots and all the bioelectrics and everything else, cancer, birth defects, traumatic injury, aging, degenerative disease, all of these things have one thing in common, which is: how do collections of cells decide what they're going to build? If we had the answer to that question, we could basically solve all medical needs except for maybe infectious disease. Everything else rides on the simple question of how do you convince cells to build one thing, not another. And so I'm extremely motivated to push this stuff to the point where it can help actual humans and veterinary medicine and so on. We're doing that by rolling out a couple of different spin-off companies from the lab. So we have one that's called Fauna Systems, and this is, Josh Bongard and myself are the cofounders, and that's all about xenobots and the idea of learning, using the xenobots as a discovery platform to learn how to program collections of cells so that they can do things either out in the environment or in the body, or in fact, never mind the xenobots, but to use those same principles to get your body cells to regrow whatever it is that you need if you lose it or if you age or if you have a tumor or whatever. So there's one company, that's Fauna Systems. There's another company called Morphoceuticals Inc, which myself and David Kaplan at Tufts, also from Tufts, are the cofounders. And that one is all around bioelectric medicine and regeneration. So we are trying to achieve limb regeneration in mammals now, and ideally, you know, someday it'll be useful for human patients. And then we have another company that's kind of about to start in the area of bioelectric cancer reprogramming. So all of these things, you know, we do try to push these things out, hopefully. I mean, I have no experience in industry or business or anything like that, but I work with people who do, and so hopefully this will get out and help real people.

(Joel Beasley at 00:48:16) Yeah. I like it because in my business experience, I really like platform business models, which is kind of what you described, how you pushed it out, and people can learn from it. And if you have that platform, and then there's commercialization around it, then you can improve the platform continuously and just constantly make something better, and then people can just build on top of it. It just becomes like a giant app store.

(Michael Levin at 00:48:38) Exactly. Exactly. That's what I'm looking for. The goal is not to make one or a set of drugs. The goal is to make a discovery engine that fuses machine learning and biological intelligence to have this, like a cycle, a discovery cycle of new ways to control biological outcomes.

(Joel Beasley at 00:48:58) Yeah. You guys can do cancer. I'm gonna do fat cells. I'm gonna create, like, a cell that eats other fat cells, but like, only 80% or however much fat you need. Right? That way you don't get the gray goo scenario where all of a sudden all your fat's gone and you're dying.

(Joel Beasley at 00:49:14) So this man, this is great. One last question. How long, this is just a guess. This is like a fun, for context, everybody, this is fun. This is not investor updates or anything like that. How long do you think, if you were just a guesstimate in the marketplace, would there be a scenario where I get my arm cut off, the ambulance comes, and they have some sort of device, chemical, communication, something that would tell my arm to regrow instead of stop growing?

(Michael Levin at 00:49:48) Alright. I'm glad you started out by pointing out that it's fun. I will, people ask me that all the time, like, when? I mean, I get, I get unbelievable emails from people every day who have lost limbs. It's heartbreaking. It's absolutely heartbreaking. Well, I get two kinds of emails. I get emails from young healthy people saying this is scary, you should stop your work. And then I get emails from people whose kids are sick or they're sick or they have some kind of damage. And they say, what's taking you so long? Like, hurry up. And they want to know when. So I can tell you, I'd be lying if I could give you any kind of a realistic estimate because I have no idea. I have no idea what the market's gonna be like. I have no idea what the funding is gonna be like, how the science is going to go. I will tell you that, you know, in the spirit of sort of my own personal guess, I expect to see it in my lifetime. I'm 52 now. So I expect—

(Joel Beasley at 00:50:41) 52? What sort of drugs are you taking, dude? You got good genetics, buddy.

(Michael Levin at 00:50:46) Yeah. Yeah. Maybe. I mean, I'm not taking anything. In fact, I get emails all the time as well asking me what kind of drugs and then suggesting drugs that I should be taking. All kinds of nootropic stuff.

(Joel Beasley at 00:50:57) I typically get most of my medical information from random people emailing me too. That's how I do my diet and my health and my kids too.

(Michael Levin at 00:51:05) Yeah. Yeah. I figured that would be the case. So right. So no. I don't take anything pretty much. I expect, I hope to see it in my lifetime. I think that's not, I think that's not crazy. I think that there will be major advances soon, and there will be some engineering problems to solve to just scale it up for the large size of human arms and stuff like that, but I totally think it's doable. I think it's doable.

(Joel Beasley at 00:51:31) Does your dad look young too?

(Michael Levin at 00:51:34) I don't know if he looks young. He's super active. The guy is 76. He plays soccer twice a week. He walks three miles a day. He teaches computer science classes. Like, he's super active.

(Joel Beasley at 00:51:46) Okay. My dad looks really young, and I look super young. That's why I grew the beard. But yeah, it was always, like, looking far younger than I am, and I hated it, like, up to my mid-twenties. And right when I hit 30, I was like, I kind of like this. This isn't so bad.

(Michael Levin at 00:52:03) Yeah. Yeah. Yeah.

(Joel Beasley at 00:52:04) So, um, it's cool how life is like that. Anything else we want to get out there to the world? Any other cool things you're doing or topics we didn't touch on?

(Michael Levin at 00:52:14) There's a ton of stuff. I can just recommend, go to drmikelevin.org. That's the website. And all the papers are there, a lot of presentations, talks that I've given, a TED talk. I have a TED talk that's there that's linked to there. ICDO.org, which is the Institute for Computer Designed Organisms, which Josh Bongard and I are leading. And then, yeah, Morphoceuticals, you know, just go to the website, and you can see all this stuff. There's tons of stuff.

(Joel Beasley at 00:52:45) Amazing. And lastly, I just want to signal my support for, you mentioned earlier something along the lines of the answer is more science, and I fully agree with that. We should have more science. And then for the people who want to do it properly, we've got guidelines and sets of rules for how they can do it in the safest way known. Right? And, yeah, so I just, I feel pretty passionate about that because I'm definitely not the person to retract because of fear. And I think that that holds us up more often than not, you know?

(Michael Levin at 00:53:16) Yep. Yep.

(Joel Beasley at 00:53:17) I like that you're pushing for, well, basically, that's a big thank you, Michael. That's a big thank you for pushing this study forward and for doing this, and I really admire the work that you're doing.

(Michael Levin at 00:53:30) Thanks very much. I appreciate it. Thank you for the conversation. Thanks for having me on.

(Joel Beasley at 00:53:36) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.