Episode 105: Michael Levin’s Unseen World of Cell Cognition

• Links to this episode: Spotify / Apple Podcasts

Unofficial, reviewed transcript. It may still contain mistakes; check the podcast when wording matters.

Speakers are labeled by first name where identified. Unidentified-speaker labels may refer to different people. Bracketed question marks indicate uncertain wording.

Transcript

[00:00:00]  Peter: This week on The Theory of Anything podcast, Bruce speaks about the work of Michael Levin, who is a biologist known for his work on cell cognition, or the idea that electrical signals between cells influence the formation of biological systems. His work has potentially massive implications for cancer research and many other fields. He is also a prominent figure promoting third-way evolution that seeks to expand evolutionary theory beyond the alleged reductionism of a gene-centric or neo-Darwinian approach. Presumably these bioelectric effects could be considered a branch of epigenetics. He is, I should add, kind of an out-there guy even by the standards of some of the thinkers we discuss on this podcast. Of course, this is a podcast that deals with out-there theories.

[00:00:55]  Unidentified speaker: See Frank Tipler.

[00:00:57]  Peter: And Bruce does an excellent job exploring his work.

[00:01:00]  Unidentified speaker: I hope you all enjoy this.

[00:01:09]  Unidentified speaker: Welcome to The Theory of Anything podcast.

[00:01:11]  Bruce: Hey, Peter.

[00:01:11]  Peter: Hey, Bruce, how are you doing?

[00:01:13]  Unidentified speaker: Well, I'm doing okay.

[00:01:15]  Bruce: I've had the flu, so I can't exactly say I'm doing great.

[00:01:21]  Unidentified speaker: Okay.

[00:01:22]  Peter: Well, I'm sorry to hear that.

[00:01:24]  Bruce: Hopefully, your editing skills will remove my constant coughing. It's been a lot worse, and I'm drugged to within an inch of my life.

[00:01:33]  Unidentified speaker: Okay.

[00:01:35]  Peter: Well, you sound fine so far, so.

[00:01:38]  Unidentified speaker: The bad part's passed.

[00:01:39]  Bruce: It was really bad for a few days. But one of the worst sicknesses I've had in a long time.

[00:01:44]  Unidentified speaker: Oh, geez. I'm sorry.

[00:01:45]  Bruce: My wife points out that I just like never get sick.

[00:01:48]  Unidentified speaker: Yeah.

[00:01:48]  Bruce: And so to get hit so hard, and it was really weird. And like I had a hard time. Like when you have the flu, it's hard to like drag yourself into work as you're in.

[00:01:59]  Unidentified speaker: You have like body aches and like chills.

[00:02:03]  Bruce: And so yeah, it was everything.

[00:02:05]  Unidentified speaker: So I guess it could be COVID.

[00:02:07]  Bruce: I don't know for sure if it's the flu. But those have very similar symptoms.

[00:02:12]  Unidentified speaker: Yeah.

[00:02:13]  Unidentified speaker: Well, all right.

[00:02:15]  Bruce: Well, today we're going to talk about Michael Levin's agentic approach to evolution, or also known as do scientists misunderstand evolution? You had maybe wanted to make some comments about the episodes up to this point.

[00:02:31]  Peter: While I'm at the edge of my seat in terms of, I mean, this is a really this what I think what at least I tentatively titled the last episode or two episodes ago, the neo-Darwinism versus post-Darwinism debate. It's not something that in all my life until very recently, I'd ever even heard of, to be really honest. But, you know, I can't tell if—I truly can't tell if we're on to one of the most interesting debates that no one knows about or if we're talking about something completely pedantic, that 99 percent of the biologists would just scoff at.

[00:03:10]  Bruce: I am going to take a definitive stance on that question.

[00:03:14]  Unidentified speaker: Okay.

[00:03:14]  Unidentified speaker: Oh, good.

[00:03:16]  Unidentified speaker: And I haven't up to this point. I've been—

[00:03:19]  Unidentified speaker: I mean, this is typical me.

[00:03:20]  Bruce: I'm doing my best to represent each side.

[00:03:23]  Unidentified speaker: I'm trying very hard.

[00:03:24]  Bruce: I started with James Shapiro's viewpoint. I then went to Zach Hancock's.

[00:03:29]  Unidentified speaker: I found a lot to like with both.

[00:03:31]  Bruce: I, in general, would say I actually think Zach Hancock did a better job than James Shapiro.

[00:03:37]  Unidentified speaker: Okay.

[00:03:38]  Bruce: I don't really like Denis Noble's books very much.

[00:03:44]  Unidentified speaker: I find them problematic.

[00:03:46]  Peter: Yeah, you've been, what would you say, throwing some shade there?

[00:03:49]  Unidentified speaker: Yeah, definitely.

[00:03:53]  Bruce: Despite all that, we're going to use Michael Levin and his research program.

[00:03:58]  Unidentified speaker: Okay.

[00:03:59]  Bruce: As the, ultimately the deciding factor in this debate. I think it defines things, he defines things in such a way that we can come up with a definitive answer to the question.

[00:04:11]  Unidentified speaker: Okay.

[00:04:12]  Unidentified speaker: By asking the right question.

[00:04:14]  Unidentified speaker: He's going to explain what the right question is.

[00:04:16]  Unidentified speaker: Okay.

[00:04:16]  Peter: Well, a rare, unambiguous answer without qualifications.

[00:04:23]  Unidentified speaker: Yes.

[00:04:25]  Unidentified speaker: I know where you stand there.

[00:04:27]  Unidentified speaker: Okay.

[00:04:28]  Unidentified speaker: So, just a review.

[00:04:30]  Bruce: Two episodes ago, we went over James Shapiro's arguments that there was something wrong with quote neo-Darwinism or modern synthesis. He treats those two terms as interchangeable.

[00:04:42]  Unidentified speaker: Shapiro argued the following.

[00:04:44]  Unidentified speaker: He argued that gradualism is incorrect.

[00:04:46]  Bruce: Evolution often takes huge single-generation steps. For example, hybridization, new species formed within a single generation due to interbreeding. Or symbiogenesis, formation of the first eukaryotes would be the gold standard example of this.

[00:05:04]  Unidentified speaker: That is, cells with a nucleus.

[00:05:06]  Bruce: Eukaryotes would be cells with a nucleus. That that happened in a single generation by two organisms fusing together. That's apparently well accepted at this point. He argues against gene centricism.

[00:05:19]  Unidentified speaker: He says it's incorrect.

[00:05:20]  Bruce: He says evolution relies on natural genetic engineering or NGE. And that the genes know how to... The organism knows how to read and write DNA and uses it as a data store and then utilizes sequences by moving them around in the genome or even spreading them horizontally. Here's a quote from Denis Noble that's similar from his book The Music of Life, page 21. The logic of successful systems that win in the competition for survival lies in the system, not in the genes. It is the system, organisms, that live or die, not genes.

[00:05:57]  Unidentified speaker: So this is...

[00:05:58]  Bruce: this, I think, captures really what the third way Shapiro and Nobles are saying, what their real critique is. One question I asked, though, was even if Shapiro was right, would this really mean that neo-Darwinism and modern synthesis was incorrect? For example, would neo-Darwinists really see hybridization or symbiogenesis as somehow at odds with current theory? And our answer was no, they really wouldn't.

[00:06:28]  Peter: So I was trying to explain this to my Bumble [date?] last night, not gonna lie, and what I said, I explained what neo-Darwinism is. I think I did a pretty good job there. I mean, I don't think she'd ever heard of that, but which is fine. I mean, people have their own interests, which describing it is something that a gene-centric view on life, our bodies are basically tools for our genes to replicate themselves, is the best conception of life. And then I described—tried to describe the post-Darwinism view on life, meaning as genes were just sort of one piece of the puzzle. It's the whole body replicating itself through various mechanisms, some of which are more or less controversial.

[00:07:28]  Unidentified speaker: She'd heard of epigenetics for sure.

[00:07:31]  Peter: I'm not even fully aware of all the implications of what epigenetics is. Maybe you can define that, Bruce. In the context of this conversation, what is epigenetics?

[00:07:43]  Bruce: In the previous episode, I actually pulled the definitions from the actual researchers.

[00:07:48]  Unidentified speaker: Oh, you did?

[00:07:49]  Unidentified speaker: Okay.

[00:07:49]  Unidentified speaker: That's right.

[00:07:50]  Unidentified speaker: I haven't edited that one yet.

[00:07:52]  Bruce: So there's actually two definitions and they're not the same.

[00:07:57]  Unidentified speaker: Okay.

[00:07:57]  Bruce: And that's one of the things that Hancock pointed out, was that the word has had [meant?] different things over time.

[00:08:02]  Unidentified speaker: Fair enough.

[00:08:03]  Unidentified speaker: I kind of got that.

[00:08:04]  Unidentified speaker: Yeah, that's right.

[00:08:05]  Bruce: So one of them is just plasticity, okay? The idea that different genes manifest in different ways depending on the environment. So that could be considered epigenetics. That's not generally how we would think of epigenetics. Epigenetics would generally be thought of as some form of inheritance that wasn't through DNA.

[00:08:27]  Peter: So I was thinking like maybe the mom is stressed out while the baby's in the womb.

[00:08:31]  Unidentified speaker: That's right.

[00:08:31]  Peter: That's the kind of thing I had in mind.

[00:08:34]  Bruce: That's what I think it normally means today. That's not its original meaning. The first one is the original meaning, but over time it became this other thing. So it is non-DNA inheritance basically, is what epigenetics is.

[00:08:51]  Unidentified speaker: Okay.

[00:08:51]  Peter: So that's kind of really the whole core of the argument, then, is an epigenetics versus genetics thing, or maybe only genetics versus both working together.

[00:09:06]  Unidentified speaker: Yeah.

[00:09:07]  Bruce: So how would a neo-Darwinist look at epigenetics? Do they see it as at odds with neo-Darwinism?

[00:09:15]  Unidentified speaker: They don't.

[00:09:15]  Bruce: That's really what we came away with, with the Hancock episode.

[00:09:19]  Unidentified speaker: Okay.

[00:09:19]  Bruce: And I think this is the first problem we've bumped into, is that you've got this idea, I mean, you kind of pitched it as neo-Darwinism versus post-Darwinism, but you're taking the point of view of Noble and Shapiro, whether you realize it or not, right? Like that's the way they would put it. It isn't the way Hancock would put it.

[00:09:40]  Unidentified speaker: Okay.

[00:09:40]  Bruce: And they would say, no, neo-Darwinism entirely encompasses, it's about the Price equations.

[00:09:48]  Unidentified speaker: It's got nothing to do with genes.

[00:09:50]  Bruce: It's just inheritance of material of some sort. So we don't care if it's epigenetics or genetics.

[00:09:57]  Unidentified speaker: And it's still neo-Darwinism.

[00:09:58]  Unidentified speaker: And that's what they would say.

[00:09:59]  Bruce: So they would argue that there's no actual conflict, even with epigenetics. So now I pointed out that these counter examples that we've talked about, they do surprise me, right? Because they do break the way evolution was taught in schools to me, even at the college level in College Biology 101 or whatever, right? And so you've got this weird thing where I've got Hancock who would say, not just Hancock, Jerry Coyne says exactly the same thing. I watched the debate between Noble and Dawkins. Dawkins essentially says exactly the same thing, okay? Where to him, it's like, I don't even see how this is at odds with neo-Darwinism. And that's kind of their view, is that, yes, you're raising all these things, but none of this is actually at odds with neo-Darwinism.

[00:10:55]  Unidentified speaker: And that's the stance they took.

[00:10:56]  Unidentified speaker: Okay.

[00:10:57]  Peter: He didn't, Dawkins really didn't dispute, from what I recall at least, that many of the details of what Noble was talking about, right?

[00:11:04]  Unidentified speaker: Right.

[00:11:05]  Bruce: He just doesn't think it's that important.

[00:11:07]  Unidentified speaker: Fair enough.

[00:11:08]  Unidentified speaker: Yeah.

[00:11:08]  Unidentified speaker: Okay.

[00:11:09]  Bruce: So, I think what we've, what's going on here is that you've kind of got the actual theory of neo-Darwinism, which would be based on the gene-neutral Price equations, according to Hancock. And then you've got a whole body of philosophical commitments built around it that aren't actually part of the theory, but are so well accepted and so much of how we teach it and how we explain it, that it's very difficult to break away between what's actually part of the theory and what's actually not part of the theory. So one episode ago, I went over evolutionary biologist Zach Hancock's criticisms of Denis Noble and James Shapiro. From his YouTube video, Denis Noble is wrong about evolution. Zach's counter-argument could be summarized as, Noble and Shapiro don't really understand modern theory of biological evolution. The true modern theory is rooted in the non-gene-centric Price equations. Once you realize that, every supposed counter-example they raise actually fits nicely into the current modern understanding of neo-Darwinism. Now, I ended up agreeing with Hancock in our previous episode. In general, I think that the gene-neutral Price equations are a much better way to understand modern evolutionary theory than, say, the selfish gene theory, which Hancock downplayed quite a bit and really did not think that that's what neo-Darwinism was. Having said that, I had never, you know, heard of the Price equations prior to watching Hancock's video. Like, somehow, the central principles of evolution had entirely escaped my education, and I don't know what to make of that.

[00:12:56]  Bruce: I mean, it's kind of like there's an actual sort of issue here, and yet I kind of also agree with Hancock that really none of these examples we're raising are truly at odds with the strongest form of neo-Darwinism. And I also, I agree that we'll never actually see a return to Lamarckism, despite what Denis Noble says, unless we're defining Lamarckism in a way so squishy that it can be described by the Price equations, and thus is really just a traditional form of variation and selection. So, possibly not through genetic inheritance though, because the Price equations do not care if it's genetic inheritance or not. They will accept epigenetics just fine as part of inheritance. Now, Hancock did basically admit that epigenetics are real, that there is non-DNA inheritance, but he argued that it wasn't a very... except in plants, that it wasn't a very strong process and that it can't have big long-term effects. So, this might be something that Noble would try to argue with him. In fact, he does try to argue that with Dawkins in the debate that they had, right? That he thinks that the impact is going to be a lot more than what we think. Still, I mean, based on experiments done so far, epigenetics only last for about three generations and then they vanish. So, they probably don't have as strong, they're not as charmed as Dawkins would put this, compared to DNA. So, I can understand that even if we're accepting epigenetics as valid as part of neo-Darwinism, I can see why that it still ends up having the strong genetic centricism that kind of grows out of it.

[00:14:56]  Bruce: So, here is from Hancock's video at 1 hour, 50 minutes and 55 seconds. This is the fate of all third-way claims. They are specific details about specific systems, they are proximate mechanisms. But what evolutionary theory is all about are the ultimate mechanisms. What is the source of the origin of these systems, and what allowed them to spread and be manifest in a population? The answer to these questions lie in the basic modern synthetic paradigm, mutation, selection, and drift. So, that was all Hancock, okay? So, if we're talking about some form of mutation, some form of selection and some form of drift, it's neo-Darwinism, as far as he's concerned. And we don't need Lamarckism, we don't need anything else. This is all just part of The Theory, okay? So, in essence, then, you have the post-Darwins who are saying neo-Darwinism is defective, and you have the neo-Darwins who aren't arguing, they're not arguing, they're arguing that neo-Darwinism is more expansive than the post-Darwins think, and that it includes the examples that they're using.

[00:16:08]  Unidentified speaker: Does that make sense?

[00:16:10]  Unidentified speaker: It makes perfect sense, yes.

[00:16:12]  Unidentified speaker: Okay.

[00:16:13]  Bruce: But I'm not entirely convinced that Zach's overall view is correct, because if it were to be shown that there is a widespread misunderstanding of the term neo-Darwinian evolution, quote unquote, to, in most people's minds, means something closer to Richard Dawkins' reductive, selfish gene approach to evolution, then that would seem to indicate that Shapiro and Noble were correct to criticize the current popular understanding of neo-Darwinian evolution. So this would be true even if evolutionary biologists like Hancock understood the term in a more reasonable way than most other scientists. And this is the thing that I really didn't see Zach handle, and was probably where my strongest criticism of Hancock's video was. I then proceeded to give several examples of famous quote misunderstandings. So this included David Deutsch outright stating that Dawkins' view is the correct modern understanding of Darwinism, and going on to wire it deeply into his Constructor Theory of Knowledge in the form of the Two Sources Hypothesis, i.e. the idea that there are only two sources of knowledge, genes and memes. Note that another possibility here is that Hancock is incorrect that Dawkins isn't the best modern interpretation of Darwinian evolution, in which case Hancock's whole argument against Noble and Shapiro is actually wrong. And the reason why I tried to include David Deutsch into this is because it guaranteed that it created a triangle where something had to give, right?

[00:17:51]  Bruce: So, you either have to argue that David Deutsch misunderstands modern evolution—he's a world famous scientist, right?—or you have to argue that Hancock as an evolutionary biologist misunderstands modern evolutionary theory. Or you have to argue that Shapiro and Noble are correct that there tend to be people who misunderstand the theory, and that it's come to mean something different than what Hancock is saying. So that was why I liked that example there. I also gave examples of biologists getting angry with Barbara McClintock when she realized that genes can jump because that violated their understanding of how genes were central to evolution. Or the surgeon general misunderstanding how quickly bacteria can learn to resist antibiotics due to holding to the standard schoolroom understanding of evolution of genes rather than what actually happens in real life. But I still have to wonder how convincing these examples are. Sure, they're good examples, and I think they do show that misunderstandings abound, right? But it could be, for example, that Shapiro and me are just cherry picking. The problem is that Shapiro and Noble have no real alternative theory to offer. What we really need is a totally different paradigm that clearly is so different than the way biologists currently understand evolution, that it makes strange and weird unexpected predictions, that can then be verified, thus establishing that this new theory is not ad hoc.

[00:19:26]  Bruce: Yes, I'm intentionally using the word verified because that matters in this context, and crit rats that have a problem with the word verified do not understand this aspect of Popper's epistemology. You verify the experiments, the predictions, that's exactly what you verify.

[00:19:44]  Unidentified speaker: It's not the theory that you verify.

[00:19:49]  Unidentified speaker: This is really what we need here.

[00:19:50]  Bruce: Okay, so this is where the work of Michael Levin can be very useful. It might give us a more definitive resolution to the question of whether or not biologists have been missing important aspects of evolution due to actually holding too strongly to the gene-centric understanding of evolution, even though that isn't actually a part of modern neo-Darwinian evolutionary theory. So, now I'm going to cover a part of Zach Hancock's arguments that I left out from the previous episode because I feel like they have direct relevance to Michael Levin's work, so I saved it for this episode. Hancock has a section in his video called Agency Intentionality and Other Weasel Words.

[00:20:30]  Unidentified speaker: Let me give you a few sample quotes.

[00:20:32]  Bruce: Hancock quoting the Forbes interview with Noble at 1 hour 1 minute 56 seconds. He says, To Noble, this routine procedure offers clear evidence of the organism actively participating in its own evolution.

[00:20:48]  Unidentified speaker: It's doing natural selection.

[00:20:50]  Bruce: This is an alternative theory of evolution where cognition is fundamental to this theory. The smallest unit of life, cells, have some version of intelligence and intent that allows them to detect and respond to their environments. At 1 hour 2 minutes 40 seconds, Noble and friends contend that organisms, including single-celled organisms, are purposeful agents, capable of directing their own evolution, their goal-oriented, and they have the ability to make conscious or semi-conscious decisions about how to deal with challenges that the environment throws at them. These ideas are either remarkably profound or extremely trivial depending on how you define these terms. So let's take an example. This isn't the example that Hancock uses, but it's probably a more obvious example than the one he uses.

[00:21:42]  Unidentified speaker: So let's take sexual selection.

[00:21:43]  Unidentified speaker: Peacocks was the example from the previous episode.

[00:21:45]  Bruce: As our example, female peacocks prefer male peacocks with a beautiful but not exactly useful tail. In the last episode, I asked if this meant female peacocks are practicing, quote, Lamarckism. Now, of course, I'm trying to force you to think harder about what Lamarckism really is when I ask that. Because most people, even though it kind of fits the definition of Lamarckism, which is a very squishy idea, nobody would probably seriously consider it to be Lamarckism. But now let's ask a similar question. Does this mean peacocks are agents participating in their own evolutionary process? Well, sure, of course it means that.

[00:22:27]  Unidentified speaker: But so what?

[00:22:28]  Bruce: This doesn't seem like much of a revelation, does it? Hancock, using a similar example, then says, this is 1 hour 3 minutes 37 seconds, From such a definition, we see that organismal agency includes everything from foraging to taking a nap. At 1 hour 16 minutes 26 seconds, Noble is in effect suggesting that the woodpecker creating a cavity in a tree is an act of purposeful agency and directs the evolution of the nuthatch that might nest in it, therefore neo-Darwinism is dead. Now, I laughed pretty hard at that comment. In fact, Hancock points out that in evolutionary biology, they invented a new world, new word to deal with this called teleonomy as opposed to teleology. It's the idea that overall there is no purpose, but evolved purposes teleonomy does exist. So teleonomy, according to Hancock, is goal-directedness in biology, should be understood as emergent from underlying genetic mechanisms. At one hour, five minutes, and 17 seconds, Hancock says, not that the cells actually have some inherent purpose, since purpose implies a mind and cells don't have minds. One hour, 10 minutes, evolutionary biologists are hesitant about using words like agency because they have an anthropomorphic undertone and can be easily misunderstood by the general public. But it's not because we don't think organisms, including ourselves, make conscious decisions. That last statement needs a little bit of context. It wasn't Hancock claiming humans don't make conscious decisions. In context, he was responding to Noble quoting Jerry Coyne, who claimed that there was no free will.

[00:24:17]  Bruce: His actual point was that the question of humans having or not having free will played no bearing in a biologist's hesitancy to use terms like agency. Because of problems like this, Hancock takes issue with use of certain words that Noble and Shapiro use freely but other scientists are not likely to use. At one hour, 17 minutes, 19 seconds, words like intentionality, purpose and agency to me really have no place in evolutionary biology, says Hancock, especially when evolution is being explained to the general public, because they have a tendency to confuse and give false impressions. At one hour, 18 minutes, three seconds, my point is that it simply isn't necessary to use such loaded words as they indicate completely trivial concepts in the first place and only act to spread confusion about how evolution actually works. At one hour, 55 minutes, 11 seconds, Shapiro thinks terms like agency and purpose are themselves driving forces that are not merely emergent properties of chemicals.

[00:25:24]  Unidentified speaker: I'm going to say that again.

[00:25:25]  Bruce: Are not merely emergent properties of chemicals, but are something special. This kind of language walks dangerously close to vitalism, spiritualism and frankly creationism. Hancock quotes Peter Bowler as arguing that Lamarckism keeps coming back because it implies things that seemed more hopeful than regular evolution. Hancock summarizes the Bowler argument for why people like Lamarckism as, at 1 hour 53 minutes and 36 seconds, organisms choose their response to each environment challenge and thus direct evolution by their own efforts. Hancock goes on to opine, with or without any religious implications, that this is certainly a more hopeful vision than derived from Darwinism, says Hancock. Life becomes an active force in nature, not merely responding in a passive manner to environmental pressures. Hancock's argument is that the real reason that Noble and Shapiro keep bringing back Lamarckism and trying to make a big deal out of agency and cognition is because they, perhaps subconsciously, have a preference for a more hopeful and appealing version of evolution. Or put another way, Hancock, in response to Noble's claim that there were ideological reasons biologists struggled to drop false views of evolution, points out that Noble himself could be easily accused of the same and reversed.

[00:26:55]  Peter: You know, I'm going to go ahead and own that and say that, you know, I think this post-Darwinism, neo-Neo-Darwinism might be true, factually, but, you know, I'll also admit that I hope it is true. It does seem to be a more, like, optimistic view of reality, honestly, than just that the genes are just controlling everything.

[00:27:19]  Unidentified speaker: I don't know.

[00:27:19]  Peter: I can't even exactly say why, but the criticism does resonate.

[00:27:23]  Unidentified speaker: I'll say that.

[00:27:25]  Unidentified speaker: Okay.

[00:27:26]  Bruce: At one hour, 51 minutes, 48 seconds, Hancock says, Noble attributes our legacy of missteps to rigid assumptions put in place over a century ago. To stand in for lack of evidence, rigid assumptions, that's it. Evolutionary biology was hindered by, quote, rigid assumptions. This brazen disrespect, says Hancock, for the history of the field is insulting. As I pointed out in the historical section at the beginning of the video, evolutionary biologists came to the current paradigm, not by rigid assumptions, but by evidence. And then, quoting Joseph Fellenstein at 1 hour 57 minutes 27 seconds, we are left at the end, still waiting for a post-Neo-Darwinism Darwin, Darwinian theory that has not appeared almost 40 years later. And after many new iterations on the same old critiques, we're all still waiting for a better theory that has yet to materialize. So, let's summarize Hancock's claims here, which I have given to you without comment or criticism. He is saying we should avoid terms like agency and purpose in evolution, for they will be misunderstood by the public. But that evolutionary biologists are well aware of the fact that animals make choices as part of evolutionary theory, though this should be understood in terms of part of their genetic inheritance. This has come to be called teleonomy, not to be confused with teleology, to avoid confusing the public. But there is simply nothing new here, claims Hancock. Shapiro and Noble are raising issues that already fit into the theory and don't shed any new light on evolutionary theory.

[00:29:09]  Bruce: So at 1 hour, 56 minutes, 38 seconds, he says, Noble is feeding the social media algorithm. What the algorithm likes is controversy. No one likes to hear it's actually all good here. Now, this is all a good setup for what is so remarkable about Michael Levin's research. Listeners may recall that I used to be a creationist.

[00:29:35]  Unidentified speaker: This is when I was a kid.

[00:29:38]  Bruce: I'd ask evolutionists to explain via evolutionary theory how life even got started in the first place. The point I was making was that the theory had a large hole.

[00:29:46]  Unidentified speaker: In fact, this is correct.

[00:29:47]  Bruce: The theory does, in fact, have a large hole. But evolutionists arguing with me would never admit this. They would without fail refer me to the Miller-Urey experiment and act wrongly like that somehow solved the whole problem. They simply didn't want to admit to the existence of an unsolved problem in evolution lest the general public misunderstood and thought that this refuted evolutionary theory. I'm now 100% in the evolutionist camp. I'm 0% creationist, 100% evolutionist. But I still find that entire tactic abhorrent. If evolutionary theory has a hole about how life got started, that should have been admitted to. And the Miller-Urey experiment tells us basically nothing about that, and this should have been admitted to. So it shouldn't surprise anyone that I'm not going to be the most sympathetic person to arguments that we might quote mislead the general public. Having said that, I'm going to now walk that back a little bit. If saying cell cognition causes the public to think cells are conscious, and if that's actually happening, then it seems to me that he's making a fair point. Because obviously cells aren't actually conscious.

[00:31:05]  Unidentified speaker: I guess we don't know that for sure.

[00:31:06]  Unidentified speaker: I'm assuming that they're not conscious.

[00:31:08]  Unidentified speaker: That would be the general assumption.

[00:31:10]  Bruce: Since we don't have a theory of conscious, like who would know for sure, right? But generally speaking, there's an assumption that a cell is not conscious. And if saying cell cognition makes people think that they are, then yeah, I can see why Hancock's got a concern there. Now, I became interested in Levin's work back in episode 21, Evolution Outside the Genome.

[00:31:33]  Unidentified speaker: A little history here is needed.

[00:31:35]  Bruce: I started really thinking about artificial general intelligence after reading David Deutsch's books and his Aeon article on artificial intelligence. See episode 58, Deutsch's Creative Blocks a Decade Later. So not surprisingly, I initially held the Two Sources Hypothesis that the only two forms of knowledge creation were memes and genes. See episodes 75, Deutsch's Theory of Knowledge, The Walking Robot, and the episodes that follow 76 to 80 for my extended look and criticisms of the Two Sources Hypothesis. However, as I started to get interested in artificial intelligence, going back to work on a master's degree at one point to dig into it further, I found that most of the algorithms met all of Deutsch's criteria for being considered knowledge creating, including his own example of the robot that learns to walk using a genetic algorithm.

[00:32:26]  Unidentified speaker: This is what episode 75 was about.

[00:32:29]  Bruce: In a panel discussion, this would be episodes 12 to 14 of this show on AGI, I made a comment about how I suspected machine learning was a form of Popperian epistemology. Ella Hoeppner, who was part of the panel, told me to check out Donald Campbell's Theory of Evolutionary Epistemology. So I read Campbell's paper, more than one paper actually, but one in particular is the famous one, and I found it somewhat wanting, and I've actually been fairly critical of it. I talked about some of my concerns with it in episode 25, Universal Darwinism, Does Artificial Intelligence Create Knowledge? And particularly episode 26, Is Universal Darwinism the sole source of knowledge creation? Note that I used the term Universal Darwinism, which actually is a correct term for Campbell's theory, but it's actually the less common of the two terms. The correct term for his theory, the one that he coined for it, was evolutionary epistemology. Now, I really thought that was a term that pre-existed Campbell. He coined the term in this paper, and I did not realize that at the time that I made these episodes. So I called them Universal Darwinism, even though it's a worse term, because I was somehow trying to have a term for Campbell's theory, okay?

[00:33:46]  Unidentified speaker: I wouldn't make that mistake today.

[00:33:47]  Bruce: I am going to refer to his theory as evolutionary epistemology. Since I, at the time, hadn't realized that Campbell himself coined the term to refer to his theory. However, there was simply no denying that Campbell's theory, even though I have concerns with it, was strongly and unavoidably at odds with Deutsch's Two Sources hypothesis. So whereas Deutsch thinks that there are currently only two implementations of Popper's theory of knowledge in all of reality, genes and memes, Campbell thought Popper's theory of knowledge was ubiquitous, and that all quote inductive achievement or knowledge creation, Campbell uses both those terms. It's a little unclear if he means them as synonyms or not. I think he did, but I'll have to make an argument to that effect later. But he uses both terms, he refers to both, that they are common in nature. But that what's really going on when you have inductive achievement of any kind, so claims Campbell, is that really, really blind variation and selective retention is happening behind the scenes. Which is of course a form of Popper's epistemology, but a generalized form. So worse yet, Popper absolutely loved Campbell's theory. Popper had a few minor disagreements. He actually lists exactly what his disagreements are. So we know what his disagreements were.

[00:35:13]  Unidentified speaker: I was just going to ask you about that.

[00:35:15]  Peter: And which book is this that Popper talks about this?

[00:35:18]  Unidentified speaker: I'm curious.

[00:35:20]  Bruce: Well, the only book I found it in is called Evolutionary Epistemology, Rationality and the Sociology of Knowledge, with contributions by Sir Karl Popper, Donald T. Campbell, WW Bartley III, and a whole bunch of other names.

[00:35:34]  Unidentified speaker: So this is kind of obscure.

[00:35:35]  Peter: So he had just like a chapter in there or something.

[00:35:38]  Unidentified speaker: Yes.

[00:35:38]  Bruce: So you have one chapter by Campbell, you have another one with a follow-on paper he did, and then you have Popper's response to his original paper. So Popper in his response, it was glowing, like it was the most glowing I've ever seen Popper be towards another theory. He did have a few minor disagreements, but his enthusiastic endorsement, which I cover in episode 26, by the way, was so overwhelmingly positive that I started to realize that both Campbell and Popper would never endorse the two sources hypothesis. Moreover, despite some problems with Campbell's evolutionary epistemology, Popper and Campbell had some overwhelmingly good explanations for why they were, at least in general, right and Deutsch was, in general, wrong. Since Deutsch didn't write until after Popper and Campbell were dead, of course, what I mean is they anticipate his objections, and they do, and had already responded to his arguments in full. It was not long after this that I happened to see Michael Levin in a YouTube video presenting his research. Here was this overt example of exactly what Popper and Campbell were claiming, a miniature version of evolution that wasn't part of the genome. It was an outright falsification of Deutsch's Two Sources Hypothesis. This led me to trying to talk to other members of the CritRat community about some of the problems in Deutsch's Two Sources Hypothesis. Now, recall also that my concern was strictly with the Two Sources Hypothesis, not with Deutsch's Constructor Theory of Knowledge in general, which I think is quite compatible with Campbell's and Popper's views.

[00:37:16]  Bruce: The rest of the story you probably already know because I mentioned it several times in this podcast. I started asking questions and raising issues within the CritRat community, mostly around Donald Campbell's theories originally, but also with Levin's work. I found that many, though not all, CritRats would respond to potential falsifiers by ad hoc saving their theories. This surprised me because Popper is quite specific on what objectively counts as an ad hoc save and made it part of his critical methodology that you do not ad hoc save your theories. When I pointed that out to the CritRats, I'd get something like a blank stare. This is online, so it's not a literal blank stare. They'd have no idea what I was talking about. Or they'd even insist that Popper had no critical methodology. Or occasionally, they would explain to me that Deutsch showed that this part of Popper was incorrect and that there was no need for ad hocness anymore because now we have easy-to-variness, the easy-to-vary criteria, and it's replaced it. Now, I had been aware that it was common for people to misunderstand Popper, but I suddenly realized that even people that had read Popper in detail often misunderstood even basic aspects of his theory. This led me to wonder if it wasn't, quote, Popper's fault, something you've probably heard me bring up a few times, that his methodology hasn't caught on. Was he saying things that were causing confusion or misunderstanding?

[00:38:39]  Bruce: This is why much of this podcast is me exploring that idea in detail, and working out new ways to explain Popper's epistemology that avoid the misunderstandings, culminating in episodes like Episode 82, Popper's Ratchet, or Episode 93, Philosophical Theories vs. Bad Explanations, which was my attempt to generalize Popper's theory to include what today we would call Metaphysical Theories. Now, if I'm being honest, Levin's work was somewhat ancillary to all this. It wasn't even really Campbell's work that was the stone that started the avalanche for me. Because honestly, I didn't like Campbell's paper that much. It was really Popper's strongly worded endorsement of Campbell's Theory that forced me to start thinking much, much harder about Popper's epistemology. But Levin's theories did play a role in this, even if it was an important ancillary role. So I have a special warm place in my heart for his theories. Okay, so now you have my history of Levin's work and how it was a factor in adjusting my thinking about Popper's epistemology. And you have some strongly worded criticisms from Zach Hancock aimed at Noble. But as we'll see, the criticisms, particularly the ones I did at the beginning of this episode, apply to Michael Levin. And in this context, I want to go over Michael Levin's research, keeping Zach Hancock's criticisms of cell cognition in mind. It can serve as a way to tease out the truth from the Noble versus Hancock and everyone else argument. It is also a startling but real life corroboration of Campbell's and Popper's evolutionary epistemology.

[00:40:23]  Bruce: It is also a refutation of Deutsch's Two Sources Hypothesis. And it's even possibly a way forward in solving Ken Stanley's the problem of open-endedness and how life came out of non-life, although Stanley doubted that when I asked him that question. It is therefore possibly related to the problem of AGI, since AGI is related to the problem of open-endedness. For all these reasons, I feel Michael Levin's theories really demand more study and I feel they fit perfectly into A Four Strands worldview. And it also corrects one of the key errors in Deutsch's theories. Moreover, no one seems to actually disagree with Michael Levin. Go search for Denis Noble is wrong on YouTube or Google. You will get a lot of hits. Now go search for Michael Levin is wrong.

[00:41:15]  Unidentified speaker: You will get Zippo.

[00:41:17]  Bruce: At least at the time that I wrote this, the slides pack that I made for this podcast. I think there are two reasons for this. The first is that Michael Levin simply hasn't made wild, triggering claims like Lamarckism is back, or neo-Darwinian evolution is dead, like Denis Noble loves to make. Even though as we'll see, Levin's research program is everything Noble and Shapiro could have hoped for. The second is more important. The second reason for why nobody is arguing with Levin is important. Trying to fight Michael Levin is a little bit like picking a fight with Superman.

[00:41:55]  Unidentified speaker: You're going to lose.

[00:41:56]  Bruce: Michael Levin, unlike Noble and Shapiro, has developed testable theories and produced startling empirical evidence that makes his point for him. So he doesn't have to rely on armchair philosophizing, like Noble and Shapiro often rely on, or specific counter examples that as we can always find a way to tuck back into the Price equations. Okay, so let me just say also, there's quite a bit of speculation that Levin's work is going to win him the Nobel Prize. In fact, at the time I wrote the notes for this episode, I checked the odds, the betting odds for him winning the Nobel Prize, 44%. So, his work is kind of a big deal.

[00:42:38]  Unidentified speaker: And it's just groundbreaking.

[00:42:41]  Bruce: What is the relationship then, if any, between Levin and Noble? Levin has made it clear he is a big fan of Noble, and often quotes him approvingly. Consider this quote from YouTube video on Mendel's pod called Fourth Season Opener, Michael Levin on The New Biology. Yes, that's actually the title. At about 43 minutes and 56 seconds, Denis Noble is one of the best people that's been talking about this for decades about no privileged level of biological causation. That was Levin talking about Noble. Moreover, Levin actually interviewed Noble on a podcast that he is co-host on, saying many supportive things. The video is on a channel called Michael Levin's Academic Content, and it is called Aastha Jain Simes, and I Interview Denis Noble.

[00:43:35]  Unidentified speaker: I'm probably butchering her name.

[00:43:37]  Unidentified speaker: Sorry.

[00:43:39]  Bruce: I've never seen anyone claim that Levin is a third-wayer, but this seems to have more to do with Levin just being disinterested in positioning himself as a radical heretic, and less to do with him actually having any strong disagreements with Noble. Consider these statements from Levin to Noble in Levin's Interview of Denis Noble. Same podcast I just mentioned, 4135. In your Noble's work and in your writing, you've emphasized a couple of really fundamental changes of perspective that have so many trickle-down effects if they were more widely recognized. The incredibly important role of goal-directed activity in biology, the top-down causation, the work on no-privilege level of causation. I mean, I think these are incredibly fundamental ideas that once people take them on board, they make you see biology in a new way, and that then informs all the specific things you may do later. There's just no denying from this quote that Levin is fully on board with basically everything Noble's claiming. At least in terms of the high-level principles, if not the specifics, you'll never see, maybe you will, but I never found Levin claiming Lamarckism was back. But in terms of the basic ideas and principles, he seems to be just completely in agreement with Denis Noble.

[00:45:07]  Peter: But can I interject one thing? But to be fair, Michael Levin is a pretty out-there guy, right? I mean, I know he does a lot of talks and things with Bernardo Kastrup.

[00:45:24]  Unidentified speaker: Do you know who that is?

[00:45:26]  Unidentified speaker: I've kind of heard the name.

[00:45:27]  Unidentified speaker: But yeah.

[00:45:28]  Unidentified speaker: Okay.

[00:45:28]  Peter: Well, he's kind of one of these guys who's a critic of, I mean, I'm out of my depth here. I can't really fully explain what he does, but he's a critic of materialism and believes in a universal consciousness kind of a thing.

[00:45:46]  Unidentified speaker: I don't want to say-

[00:45:47]  Bruce: Do you have an actual quote from—from Levin actually saying he believes in those things?

[00:45:53]  Unidentified speaker: By the way, I think-

[00:45:53]  Unidentified speaker: No, no, I don't.

[00:45:54]  Unidentified speaker: I don't.

[00:45:55]  Unidentified speaker: I'm just talking.

[00:45:56]  Peter: I'm just, but I do know if you search for Michael Levin and Bernardo Kastrup, they do talks together and things all the time. So, I mean, I assume there's somewhat alignment on some of these things. I'm not even saying it's bad, but, you know, Bernardo Kastrup would probably be dismissed as almost like a Deepak Chopra kind of character by most mainstream scientists. I'm not rightly or wrongly, but-

[00:46:30]  Unidentified speaker: So, here's the thing.

[00:46:31]  Bruce: Just like Noble is willing to appear on shows with creationists, right?

[00:46:36]  Unidentified speaker: Fair enough.

[00:46:37]  Bruce: One of the things that Hancock brought up, even though he never actually is a creationist on these shows, Levin does much the same thing. Levin is very guarded in what he says. And so that's why I felt like I could, even though I've never seen these videos, I could with confidence say he's never actually said it, has he?

[00:46:55]  Unidentified speaker: Because he just doesn't, right?

[00:46:56]  Unidentified speaker: Okay.

[00:46:57]  Bruce: Now, he admits he has some beliefs that are a bit out there, but he won't say what they are. And he'll say, about 70% of my beliefs are entirely public.

[00:47:07]  Unidentified speaker: This is what you're going to find on my videos.

[00:47:09]  Bruce: I've got a 30% that informs it. There's no scientific basis for it. It's just part of my beliefs.

[00:47:16]  Unidentified speaker: I don't even bring it up.

[00:47:17]  Bruce: I don't even feel the need to raise it. So I think what you're going to find is that Levin at some level does have some woo-woo beliefs that he just never raises. And you will never find him actually bring them up. I think the worst case you will find is that he mentions he believes in a Platonic realm. Unfortunately, the term Platonic realm means so many different... It's a suitcase term that means so many different things in so many different circumstances. There may even be a reasonable way to understand that term, the way he's using it. Okay, so I think you're going to find that he really does a pretty good job of just not going there. Like, he's got beliefs that do inform his research program.

[00:48:03]  Unidentified speaker: But ultimately, it all...

[00:48:04]  Unidentified speaker: And I'll give you the quotes later.

[00:48:06]  Bruce: But ultimately, it comes down to, can you actually do experiments? And if you can't, then he's just not going to even bring it up.

[00:48:12]  Unidentified speaker: Okay.

[00:48:13]  Bruce: And to him, that's just a non-scientific belief. And so it doesn't even need to be raised.

[00:48:20]  Peter: You know, I'm definitely in favor of talking to all kinds of people. I'm not, like, trying to make a guilt by association kind of an argument.

[00:48:27]  Unidentified speaker: But I'm just putting that out there.

[00:48:30]  Unidentified speaker: Okay.

[00:48:33]  Bruce: So Levin is outright agreeing with Noble that people have missed massively important aspects of Darwinian evolution and biology due to failing to understand that there is no privileged level of causation, i.e. due to always insisting on the genes I view, that the genes create chemicals and that's what drives everything. So let's actually talk about Levin's actual experiments, because I think this is where the rubber meets the road.

[00:48:59]  Unidentified speaker: Okay.

[00:49:02]  Unidentified speaker: So this is one of my favorites.

[00:49:04]  Bruce: Levin put a flatworm, a planaria, into a solution of barium. Now, the barium blocks the ability of the cells to exchange potassium with the outside world. So the head of the planaria is full of neurons that exchange potassium. So being in this solution causes the head to explode. Now, recall that planaria can regenerate if you cut them in half. So the planaria will grow back a new head in a couple of weeks.

[00:49:31]  Unidentified speaker: Now, here's the stunner.

[00:49:33]  Bruce: The new head that grows back is now barium adapted and no longer has problems with barium.

[00:49:40]  Unidentified speaker: Now, Levin wanted to know how that could be.

[00:49:43]  Bruce: So, he looked at the genes of the barium adapted head and found that a small number of the genes were turned off and on to adapt to this novel stressor.

[00:49:52]  Unidentified speaker: Now, here's the amazing part.

[00:49:54]  Bruce: Planaria never see barium in the wild. There is no evolutionary history of knowing what to do when you're hit with barium. So, at seven minutes in the podcast called Can Cells Think? Michael Levin on the Well, he says, So, just imagine you're a bunch of cells and you're hit with this incredible physiological stressor you've never seen before. You've got tens of thousands of genes.

[00:50:21]  Unidentified speaker: What do you do?

[00:50:22]  Bruce: How do you know which gene to turn off and on? You don't have time to try combinations. You don't have time for exhaustive search. You don't have time for hill climbing. You don't have time to try random things because you'll probably kill the cell long before you solve the problem. And yet, you are able to navigate from where you are to where you need to be to escape this physiological stressor. And so, what I think evolution has done is pivoted some of the same tricks from very simple systems that only solve metabolic problems eventually to physiological and then to transcriptional problems. And when multicellularity comes on the scene, large scale anatomical problems. And so, it's never a question of is something physics and chemistry or is it cognitive? The question is what kind of cognition and how much? That was all from a quote from Levin. You probably just noticed that Michael Levin appears to be quite comfortable with the very kind of language that Zach Hancock and most biologists try to avoid, speaking of cell cognition and agency in particular, with particular ease. Consider this quote, 45 minutes, 45 seconds from the same podcast. And I think that kind of embodiment is critical, but it's not the physical space or the physical body that's critical.

[00:51:44]  Unidentified speaker: It's the loop.

[00:51:45]  Bruce: It's this having to be an agent, to say, what do I do next, and having a set of options and a set of pressures and constraints, and enough of a cognitive system which can be very minimal, that lets you do things where the future is going to be different from the past. That's what I mean by embodiment. Note how these are exactly the weasel words Hancock objected to in Shapiro and Noble, but this time we've got hard empirical experiments that demonstrate the previously unknown problem-solving ability of embodied cells. Now, of course, this idea that cells can solve novel problems isn't really in any way at odds with neo-Darwinian evolution. If by that term you mean Zach Hancock's version of it that's centered around the gene-neutral Price equations, is it really that surprising that Darwinian evolution gave this planaria a learning algorithm that it can solve a problem with a novel environmental stressor? I mean, isn't that exactly what our own immune system does? I admit this is a bit more interesting because planaria can reproduce both sexually and asexually. So, this barium-adapted head will get passed to a new planaria as part of its inheritance, even though the genes have technically not changed. So, if you can wrap your mind around the gene-neutral Price equation version of neo-Darwinism, this is still, quote, handled by the theory. But it's interesting how people have responded to this video when I sent it to them. And I feel like this really is important, even though we're talking about, you know, laymen.

[00:53:24]  Bruce: One person, a non-crit rat, I sent it to, immediately wanted to declare it a developmental artifact of how heads form in response to an environment. But then he acknowledged that he wasn't sure why the planaria had to turn off and on various genes. A crit rat I sent it to declared that there was probably a, quote, more robust head stored in the genes, that it only deploys when it's needed, and by chance it was barium-resistant. And then he said, well, or maybe it really did encounter barium or a barium-like substance in the past. And he said it would be impossible for you to ever prove otherwise. So this theory doesn't have any way that you can falsify, which he saw as a positive. So he argued that there was probably no actual learning or problem-solving except by regular genetic evolution. Now this answer is interesting because he's a crit rat, and this is clearly an ad hoc save. He's also leaving an explanation gap as to why the genes had to activate and deactivate to be able to create this barium-resistant head. Now, when I pointed out to him that he was ad hoc saving his gene-centric view of evolution, and that such saves were always possible, he didn't understand why he was making this point as a fellow critical rationalist. And so naturally, he was unimpressed with my argument. The argument didn't even seem slightly impressive to him. He was convinced that probably there was barium somewhere in the history, and you couldn't prove otherwise, so there was really no reason here to see this as a learning algorithm.

[00:54:52]  Bruce: Now, in both cases, there was this knee-jerk reaction to try to explain the experiment wholly from kind of the standard view of evolution that we all get taught in school, rather than acknowledge actual agentic learning and creation of new knowledge within the life of a single organism.

[00:55:08]  Unidentified speaker: Now, why?

[00:55:10]  Bruce: I'd suggest it's because Shapiro and Noble are correct that neo-Darwinism is understood almost entirely as equivalent to a single level of emergent, the genes I view, and that none of the others are generally understood, no other level of emergence is understood to be valid. Now, Levin's team, here's another experiment. They created Picasso tadpoles with scrambled facial features, like eyes on top of the head, but by disrupting normal development processes. Now, normally, tadpoles develop into frogs with specific facial features in specific places, and it was thought to be hardwired. I mean, of course, because we often think of genes as being blueprints. The Picasso tadpoles, despite their scrambled starting points, developed into normal frogs, showing that development isn't hardwired. Instead, the system is homeostatic error correction, meaning it corrects errors to reach its goal. The cell's actions are goal-directed, with the goal of making a correct frog face. The system remembers what a correct frog face should look like. Cells are pursuing goals in an anatomical space, or the space of all possible configurations of the body. The tadpole face develops without micromanagement of individual cells. The collective cells act with collective intelligence, rather than individual instructions. And he found that bioelectricity is the key. Cells communicate and integrate their intelligence using ion channels and electrical connections. This creates a network that stores a pattern of what a correct frog face should be.

[00:56:46]  Bruce: The idea of a collective intelligence communicated via bioelectricity, just like the brain uses, now mind you, is going to be a big factor in most of Levin's experiments. So this is kind of a big deal for understanding his point of view and what his theories are. In short, the Picasso Tadpole experiment shows that development uses a flexible error-correcting system, not a hardwired blueprint. Cells work together to achieve a goal using bioelectric patterns as a reference, showing a form of tissue-level intelligence. Notice that we're not even talking about cell intelligence anymore. We're talking about something even more basic than that. Tissue-level intelligence, okay?

[00:57:24]  Unidentified speaker: Maybe even smaller than cells.

[00:57:28]  Bruce: Now, at 14th season opener, Michael Levin on The New Biology, at 1358, he says, Keep in mind, this already is very much a taboo topic. Because in biology, we're not supposed to think about goals. We're supposed to think about emergence from chemical signals. You're not supposed to think that these things are in some way moving towards a goal. Now, when I say goal, too, a lot of people jump to a very grandiose definition of goal, which is a conscious, I know what my goal is, I know I can change my goal. You know, this is not what I'm claiming at all. I'm using a very cybernetic kind of definition.

[00:58:08]  Unidentified speaker: I'm lowered down on that spectrum.

[00:58:10]  Bruce: I'm not saying anything about the conscious experience of this.

[00:58:14]  Unidentified speaker: Notice how he's not a panpsychist here.

[00:58:17]  Bruce: All I'm saying is the system has, there's a set point. It tries very hard by expenditure of energy and various competencies to get to that set point. And it will do so even when you try to prevent it with some degree of intelligence.

[00:58:32]  Unidentified speaker: That's what I mean by goals.

[00:58:33]  Bruce: It's a cybernetic, it's a non-magical definition. Now another experiment he did was the limb regeneration. So adult frogs don't naturally regenerate limbs, but Levin's team can induce regeneration of the limbs. They do this by manipulating the bioelectric signals in the first 24 hours after amputation. Cells use electrical networks to communicate, which Levin calls a language. Levin does not micromanage individual cells, control gene expression or use 3D printing. Instead, he communicates with the cells at the level of the bioelectric network, and he sets a high level goal. The treatment involves telling cells to follow a limb growth path, not a scarring path. The cells are influenced by the bioelectric pattern or set point that is established. Levin's tools read and write the memories of the collective by observing and manipulating the electrical network states. He's working to crack the bioelectric code to induce specific outcomes. Genes are concerned only with building the hardware, but the bioelectric software directs cells. Levin can make cells do things outside their genetic code. Levin provided a communication interface to set a goal. The cells already know what a leg looks like and they build it. Cells follow a bioelectric pattern or set point that represents the goal. Changing this set point directs the cells. This shows the bioelectric language can be used to override genetic limitations and trigger complex regeneration processes.

[01:00:07]  Bruce: In essence, you cut the frog's leg off and then you send a signal that through experiment, you've come up with, you know means 'regrow this leg'. Even though frogs don't normally regrow legs, suddenly the frog will regrow its leg because that's knowledge that is contained within the bioelectric signals. It just simply is a different goal that's been given. That way, you don't have to try to figure out what genes to adjust and what chemicals to adjust. You basically just tell the body, regrow that limb, new goal, boom, it does it. In the podcast From Bioelectric Fields, a Paradigm Shift in Biology at 3741, he says, my claim about goal-directedness, and the key point is that by targeting these patterns through interventions, using techniques that modify memories, you can produce complex outcomes that were previously impossible. The micromanagement required would simply be too difficult. Now, he gives a really interesting example of this. So, from Bioelectric Fields, a Paradigm Shift in Biology, 3741, he says, a simple neuroscience example of this is, you have a rat, and you want the rat to do a circus trick, sit on a little bicycle or something, and somebody says, okay, I can see this thing is a deterministic molecular system.

[01:01:28]  Unidentified speaker: So, here's what we're going to do.

[01:01:29]  Bruce: We're going to find all the different muscle motions that need to happen to make the body do it. Then we're going to track back to all the nerves that need to be activated. Then we're going to track back all the way into the brain, and then we're going to go through, and I'm going to tell you exactly which pixels on the retina you need to activate to stimulate it to make it do the thing. Now, in theory, this is possible, says Levin. In reality, the sun will burn out before you figure out any of that. But someone who understands that the rat really is can say, no, no, no, you don't need to do any of that.

[01:02:04]  Unidentified speaker: You can train the rat.

[01:02:07]  Bruce: What do you mean you can train the rat?

[01:02:10]  Unidentified speaker: Top-down control.

[01:02:11]  Unidentified speaker: You collaborate with the rat.

[01:02:12]  Bruce: You incentivize the rat. What's magic about this is that the top-down control pushes all the complexity onto the system itself. You don't need to figure out which neurons need to fire. The rat's already very good at knowing what needs to be done, given certain top-down controls. It's a communication problem, not a micromanagement problem. The same thing is going on here. We know this is a goal-directed phenomenon because we change the goals.

[01:02:38]  Unidentified speaker: We don't touch the hardware.

[01:02:40]  Unidentified speaker: We don't change the DNA.

[01:02:42]  Bruce: We don't tell any of the stem cells what to do. All of that is taken care of by the system. All we do is reset the goals. That's the beauty of goal-directed systems.

[01:02:53]  Unidentified speaker: You can communicate with them.

[01:02:55]  Unidentified speaker: You can reset the goals.

[01:02:56]  Unidentified speaker: Maybe they learn.

[01:02:57]  Bruce: Some cells and tissues learn. Okay, but regrowing a limb is setting a goal that the genes already knew how to do. Can you actually set a goal that the genes didn't intend? Something entirely new, not part of the genome.

[01:03:13]  Unidentified speaker: So there was an experiment that Levin did to test that.

[01:03:17]  Bruce: It's the two-headed planaria experiment. So Levin's lab induced two-headed planaria flatworms by altering their bioelectric patterns without modifying the worm's DNA. So the standard biology suggests morphology is determined by genes. But these worms, when you cut them in two, continue to reproduce two-headed offspring, even without further bioelectric manipulations. This suggests, this indicates a heritable change in phenotype independent of genetics. So you have this two-headed planaria, you cut it, that's how it does asexual reproduction, and the new planaria that grows will also be two-headed, even though the genes haven't changed.

[01:04:02]  Unidentified speaker: Okay.

[01:04:04]  Bruce: The bioelectric circuit stores the information for the number of heads, effectively acting as a set point for the worm's morphology, which can be reprogrammed. The cells use this bioelectric pattern, not the DNA, to determine the number of heads to grow. The genome itself was not changed, showing that morphological information is not solely encoded in DNA.

[01:04:26]  Unidentified speaker: So now, why was this overlooked?

[01:04:29]  Unidentified speaker: So let's actually talk about this.

[01:04:31]  Bruce: So in, I think it's the same podcast episode, but now at 9.49, maybe I didn't write this one down right. He says, there are, you know, looking back, you can always tell a conventional story. And sometimes, especially with the outcomes of our work, you would have to put in some epicycle, so to speak. You would have to put in some extra bells and whistles to make it fit the conventional paradigm.

[01:04:55]  Unidentified speaker: But you always can.

[01:04:57]  Unidentified speaker: This is all Michael Levin.

[01:04:58]  Bruce: So when somebody says, look, you know, this is what you've shown, and somebody else says, oh, I'm sure there's a molecular explanation for it. Of course, there's a molecular explanation for it. There's always a reductive explanation for it afterwards. The real question is, what was the paradigm that got you to do the experiment in the first place? And why wasn't it done before? The first two-headed worms, I mean, there are many ways to make two-headed worms. The first two-headed worms were seen at the turn of the century, around 1903 and so. The question though, is since between then and 2009, when we did our experiment, why did nobody recut them and then find out that they inherited the two heads? To my knowledge, nobody ever recut them. The reason you wouldn't recut them is because if you have a standard belief that your genome is what determines your morphology and the genome hasn't been touched, well then why would you do it? Of course you had an expectation. Of course they will go back to being one head once you cut off the ectopic head. Our paradigm predicts the opposite, because we were thinking that the bioelectric circuits that store this information as a kind of memory. The most obvious thing about memory is that it keeps state, either temporarily or permanently, but it keeps state. So I think that's important, right?

[01:06:23]  Unidentified speaker: It's not just the phenotype.

[01:06:25]  Bruce: I mean, anybody can get a double head. There are many ways to get a double head. So just to answer your specific question, I think what we have here is a new paradigm. The new paradigm is this idea that everything exists on a spectrum of cognition and that you cannot say whether something is on that spectrum from previously held philosophical commitments.

[01:06:48]  Unidentified speaker: You have to do experiments.

[01:06:50]  Unidentified speaker: Now think about this for a moment.

[01:06:52]  Bruce: There had been experiments to create two-headed planaria, but it took over a century to realize that a two-headed planaria will reproduce and form new planaria with two heads.

[01:07:08]  Unidentified speaker: Why did it take a century?

[01:07:09]  Bruce: Because it just seemed obvious to biologists that since genes produced phenotypes, that if you did cut one, they'd go back to being one-headed. This was just so strongly built into the understanding of the theory that it didn't occur to them to try. Or to put it another way, neo-Darwinists couldn't think to do such an experiment due to rigid thinking rooted in a reductionist genes-eye view of neo-Darwinism.

[01:07:37]  Unidentified speaker: Now, that's not an offensive opinion from Noble.

[01:07:40]  Unidentified speaker: That's a factual statement about what actually happened in this case.

[01:07:44]  Unidentified speaker: Okay?

[01:07:46]  Bruce: Notice also that Levin is pointing out that the problem isn't the genes-eye view itself.

[01:07:50]  Unidentified speaker: And I think this is really important.

[01:07:52]  Bruce: Reductive theories like that can be entirely correct. The problem is reductionism or a belief, perhaps held subconsciously, that you have to explain neo-Darwinism and biology in terms of genes controlling chemicals and making taboos around emergent explanations that do a better job of explaining the same phenomenon. The genes-eye view isn't going away ever. No one nor does this reduce its importance. If anything, emergent explanations like this agentic view that we're talking about, this cell cognition view, enhance the reductive explanation of the same phenomena. The real issue is the taboo on the emergent explanation that explains it from a different level. Levin points out that if reductive explanations were really superior, then why not try to explain biology in terms of quantum foam? You pick the level of emergence that makes sense for the problem you're trying to solve. And for Levin, he decided to view cells and even tissue inside of cells as having a minimal form of intelligence or put another way, that they contain knowledge creating learning algorithms. And this turned out to be just the right paradigm to open up whole new ways of research, to research theory, to research evolutionary theory that everyone else had missed due to not wanting to quote mislead the public. Another one that honestly, this is very personal one. Cancer is caused by cells disconnecting from, according to Levin and according to his experiments, disconnecting electrically from their neighbors. This may not even be caused by genetic damage.

[01:09:28]  Bruce: So Levin's approach in the experiment was to reconnect cells electrically, restoring their participation in a mind meld of the tissue. This mind meld is a state of collective intelligence where cells adhere to the tissue's goals, preventing uncontrolled growth. Bioelectricity is the cognitive glue, according to Levin, that's his term, that binds cells together and enables them to act as a collective. Restoring electrical conditions redirects cancer cells to the goal of making normal tissue.

[01:09:57]  Peter: So I just kind of always thought that the cell went crazy, right? And just starts reproducing itself in some abhorrent way. But it sounds like what he's saying is that the cells in cancer cells just aren't connected to the other cells biochemically or something?

[01:10:14]  Unidentified speaker: Let me finish explaining the experiment.

[01:10:15]  Unidentified speaker: Okay, sorry.

[01:10:16]  Unidentified speaker: Sorry.

[01:10:17]  Bruce: This method, tried in frogs, prevents or removes tumors even with strong oncogene expression, that is to say abnormal cancer DNA, addressing the software without needing to fix the hardware. This is a top-down approach using communication rather than micromanagement of cells. Levin views cancer as a disorder of morphogenic intelligence where the boundary between cell and environment is disrupted. This is actually the answer to the first part of your question, which is a cancer cell is a cell that stops thinking of itself as part of a collective, and tries to become a single-celled animal that's trying to reproduce itself.

[01:10:55]  Peter: According to him, it's controversial, isn't it, or not?

[01:11:00]  Unidentified speaker: Well, let's keep going.

[01:11:01]  Unidentified speaker: Okay.

[01:11:02]  Bruce: So, yes, according to him, this approach uses insights from cognitive science, not just chemistry, to understand and influence cell behavior. Software signals alone can trigger cancer, suggesting that it is a problem of goals and not necessarily a genetic damage. So you can't fix the problem without fixing the genetic. You can fix the problem without fixing the genetic damage.

[01:11:26]  Unidentified speaker: You just have to communicate a different goal.

[01:11:28]  Unidentified speaker: Now, this is the experiment that he did.

[01:11:29]  Unidentified speaker: So let me actually quote him now.

[01:11:32]  Bruce: 14th Season Opener: Michael Levin on “The New Biology,” 31:52. Software is enough to trigger it, and the best part of it is that you can prevent and normalize tumors made from human [oncogene?] by forcibly reconnecting cells to their neighbors. The oncogene, mutated DNA causing cancer, is still there.

[01:11:51]  Unidentified speaker: We're not fixing it.

[01:11:52]  Unidentified speaker: There's no gene therapy.

[01:11:54]  Unidentified speaker: The oncoprotein is blazing.

[01:11:57]  Bruce: Because we can label it, you see [programming?]. It doesn't matter because it's the software that drives it, not the hardware that drives it in the end. In other words, you were saying, according to Levin, he did this as an experiment, right? He actually took cancer cell in frogs. And he sent a signal saying, actually, you now have a new goal as to be part of the collective. These cancer cells had the mutated cancer genes still. They had all the proteins that suggested they were still cancerous. And the tumor went away, even though all the bad DNA from cancer was still there. Because you don't have to fix the genes to make the cancer go away. You simply have to communicate to the cells that they're actually part of the overall collective, then the cancer goes away.

[01:12:53]  Unidentified speaker: Now, does that answer your question?

[01:12:56]  Unidentified speaker: Yeah.

[01:12:56]  Peter: I mean, I just would be curious if there are other perspectives on this research. I guess part of it is that there's just so much like weird ideas and pseudoscience out there about cancer that I just kind of like makes my makes me a little bit. The BS detector starts to get a little bit activated, even though for all I know, his research is great and it's all right.

[01:13:27]  Bruce: So I don't think anybody doubts that he did make cancer go away in a frog.

[01:13:30]  Unidentified speaker: Fair enough.

[01:13:31]  Bruce: Just by sending a bioelectrical signal. That really was a giant discovery.

[01:13:35]  Unidentified speaker: Okay.

[01:13:36]  Bruce: Now, we could ask questions like, would this work in humans?

[01:13:39]  Unidentified speaker: Well, we don't know.

[01:13:40]  Bruce: We have to actually experiment to figure that out. He is starting a company to try to cure cancer using bioelectric signals.

[01:13:46]  Unidentified speaker: Okay.

[01:13:47]  Bruce: So he's putting his money where his mouth is, right?

[01:13:50]  Unidentified speaker: Yeah.

[01:13:51]  Unidentified speaker: We're making this very personal now.

[01:13:52]  Bruce: Think about all the cancer research that's been done utilizing the genes I neo-Darwinian view, that were all steeped in and how unsuccessful it's been. No one even thought to try to tell the cells via bioelectric signals, and that they used to communicate, to just stop being a cancer cell and give them a new objective. It is so much easier to solve cancer at this level of emergence, at least in frogs. Of course, the reason no one thought to do this was because no one knew about bioelectric fields between cells, because they were blind to it because the entire field, literally everyone, had imbibed neo-Darwinism as requiring explanations about genes using chemicals to create proteins, and that's it. That just was what neo-Darwinism had come to mean. No one even conceived the possibility that tissue itself might be an intelligent learning algorithm that you can simply retrain. I suppose the proof is going to be in the pudding on this one. If we have cures for cancer in 50 years, thanks to Levin's research, then it really will be academic at that point, right? It may be that this is easier in frogs than it is in humans. So that often happens, of course. But this is kind of a big breakthrough in cancer research, that just nobody's even attempted this approach. Nobody even thought to even attempt this sort of approach, because it just didn't fit the paradigm as the way we normally think of evolution. So the next experiment are xenobots. This is probably the one he's most famous for. So xenobots are novel living machines. Xenobots are tiny living robots made from frog cells.

[01:15:32]  Bruce: They are biological constructs that exhibit novel behaviors. So they're constructed from repurposed cells. They're created by taking living cells from frog embryos and reassembling them in new forms. The cells are not genetically modified. So exploring latent biological potential, the xenobots are used to explore the range of what biological hardware can do, investigating forms and functions not normally seen in nature. So xenobots exhibit a range of surprising behaviors, such as self-propulsion, collective behavior, self-healing. They can make copies of themselves by gathering and assembling cells that they find in their environment.

[01:16:11]  Unidentified speaker: The cells are not genetically altered.

[01:16:13]  Bruce: Their behavior is guided by their original biological programming and their new configuration. Xenobots show that cells are capable of more than their usual behaviors, revealing possibilities for how biological materials can be organized. Xenobots represent a new kind of bioengineering that is based on understanding and manipulating the collective behavior of cells, not just micromanaging genes or proteins. The Xenobot experiment challenges traditional notions of what life is and how it behaves, moving beyond a focus solely on genetics to a focus on bioelectric networks, collective intelligence and the problem-solving capabilities of biological materials. Xenobot research represents a shift in focus for managing the hardware of cells, genes and molecules, to the software, [biological/bioelectric?] fields and the cell-to-cell communication. So this is now from Michael Levin, Why Intelligence Isn't Limited to Brains at 1957. What's happening is that biology, unlike our computer technology, biology is dealing with a very unreliable medium. So we work very hard so that every layer is isolated from the layer above. And when you're writing in high-level computer code, you're not worried that the metal is going to act weird and that suddenly your registers, the numbers are going to float off and be something else. You can trust that you have full trust in the layer underneath.

[01:17:34]  Unidentified speaker: This is regular computers, of course.

[01:17:36]  Bruce: Biology has the exact opposite approach because everything is unreliable. You have no idea exactly how many copies of anything you're going to have. The molecules come and go, everything degrades and things misfold. And you know, on an evolutionary time scale, things are going to change, guaranteed. Not only is the environment going to change, but your own parts are going to change because you're going to mutate.

[01:17:59]  Unidentified speaker: Everything is different each time.

[01:18:01]  Bruce: And for this reason, I think that evolution does not overtrain on the past. And this is why we could talk about it, but I have in my papers many, many examples of this. Biology is incredibly good at solving novel problems and making some kind of coherent being in all kinds of crazy configurations that either nature has done or that we do in the lab. And the reason is even the normal embryo was not a mechanical construction to begin with. It is solving that problem every single time. I like this idea of beginner's mind. It's like you're given some tools or you're given some hardware that's encoded by the genetics. You have no idea what it used to mean. You have no idea what kind of environment and how it was meant to be interpreted.

[01:18:57]  Unidentified speaker: And you have no allegiance to that.

[01:18:59]  Bruce: And you have no ability to know exactly what that is. All you know is you're going to use that information as best you can. You can right now do whatever the next thing is. And that creativity, that the requirement to confabulate really to tell a new story without any allegiance to the story used to be, is the intelligence ratchet that starts off the physiological and metabolic space and proceeds through anatomical morph space and eventually behavioral space and linguistic space and so on. So, from 14th season's opener, Michael Levin on The New Biology, 1314, the genetics did not give you a hardwired machine that does the same thing every single time. What it gave you is hardware that can actually adapt in a context-sensitive nature. So, hopefully with these experiments, I've at least convinced you that it's pretty common to have a strong reductionist bias when it comes to neo-Darwinian evolution. And that it is kind of a blinding thing. It has caused us to not understand certain aspects of how evolution actually work because we were so afraid of things like cell cognition and things like that. Didn't want to mislead the public, but instead we ended up misleading the entire field. Note that none of this changes or violates the Price equations. The reproduction of a two-headed planaria without a change in the genes, fits the gene-neutral Price equations just fine.

[01:20:33]  Bruce: And so while Hancock is technically correct that none of these new discoveries is truly quote devastating to neo-Darwinism per se, if you understand them as the gene-neutral Price equations, there's just no getting around the fact that Hancock is still ultimately pretty much entirely wrong. We all have been taught to have a rigid reductionist thinking when it comes to neo-Darwinian evolution. And even if the Price equations did not include that fact, it was still part of how nearly all of us were taught evolution. To drive this point home, here is Michael Levin's account of what evolution actually is. This is from Bioelectric Fields, A Paradigm Shift in Biology. Professor Michael Levin from Essentia Foundation. 23.05, what I think evolution is actually doing is developing problem-solving agents. And I don't just mean clever animals that do things in three-dimensional space. I mean molecular networks that can learn. Cells and tissues that solve problems in physiological space and anatomical space and so on. And so if you have this idea that what comes out of evolution are these problem-solving agents with competencies, with certain degrees of intelligence, that is up to us to discover, that it's up to us to discover, that opens up a whole new frontier of research programs around how do we detect unconventional intelligence, operating in different problem spaces. Do we communicate with it in ways that are effective to change their behavior? Or put another way, what evolution really evolves is learning algorithms, or in other words, evolutionary algorithms.

[01:22:07]  Bruce: Now if you have listened to this podcast since the beginning, you will be one of the very few people not surprised by this. We actually did a full podcast about this back on episode 38, Animal Learning and Popper's Epistemology. At that point in time, I had never heard Levin say any of this. But it isn't like I'm so smart that I made it up myself. I was quoting Karl Popper. And Levin is more or less a version of Donald Campbell's Theory of Evolutionary Epistemology. In other words, Karl Popper and Donald Campbell, but in particular this quote I'm about to give you from Karl Popper, wholly anticipated Michael Levin. And Popper did so back in an era, 1961, where words like software weren't even around. So he didn't know what words to use to refer to what today we would just call software. So he tried to use the analogy of a plane's autopilot, because that was the best analogy he could come up with to try to explain what he was referring to. He tried to explain that animals evolved such programs so they were able to utilize physical changes via a learning algorithm.

[01:23:12]  Unidentified speaker: So here's the actual quote.

[01:23:14]  Bruce: This is, I quoted it back in episode 38, but let me give you a kind of an extended version of the quote. This is from Objective Knowledge page 275 to 278. Now it seems clear, this is Karl Popper, that in a complicated organism, almost all accidental mutations will be disadvantageous. They will thus be eliminated by natural selection. This will hold special force for accidental mutations affecting more than one organ. He uses the example of the software of the autopilot system. Keep in mind that an autopilot system back then probably wasn't a computer software. It's probably a system that did some sort of checking and then trying to rebalance things.

[01:23:55]  Unidentified speaker: So you can think of it as soft.

[01:23:57]  Bruce: Today we would literally make it computer software, but back then it may not have actually literally been computer software. Okay, but this is why he had to use this example. So he uses the example of the software of an autopilot versus, say, the wings on the plane. Now he says, according to the monistic hypothesis, what does he mean by the monistic hypothesis? Well, this would be regular Darwinian, neo-Darwinian evolution. Okay, so according to the monistic hypothesis, regular Darwinian evolution, a favorable mutation of an organ, say an increase in the power of one of the engines, will always be used favorably, and that is all there is to it. Any favorable mutation is improbable, but its probability need not be vanishingly small. But according to the dualistic hypothesis, this is his hypothesis, the idea that animals have learning algorithms, and that this is fundamental to how evolution actually works, a favorable change to an organ would, in many cases, be potentially favorable, be only potentially favorable. To make any difference, the improvement would have to be used. And this new use might depend on complementary, accidental change in the central propensity. That's what Popper's term was for the software of the nervous system. Which would be, at the same time, both independent and complementary must indeed be vanishing.

[01:25:16]  Unidentified speaker: So he's raising this issue.

[01:25:18]  Unidentified speaker: How did evolution ever work?

[01:25:19]  Unidentified speaker: How did you ever have evolution work?

[01:25:21]  Bruce: Where you have this idea that an animal is born, it's got like, you know, better hands or, you know, something like that, and then it's able to use it, and it gives it an evolutionary advantage.

[01:25:32]  Unidentified speaker: But in real life, that would never happen.

[01:25:33]  Bruce: Like if you had a Boston Dynamic Robot that doesn't have learning, this is the example I used from episode 38, and you put a better engine inside of it, even though the engine is better, the software wasn't built with that engine.

[01:25:48]  Unidentified speaker: So the whole robot's going to fail.

[01:25:50]  Unidentified speaker: Okay, so he's raising this issue.

[01:25:52]  Bruce: How is it that you can even have physical changes evolve in the first place using the standard evolutionary theory, the monistic hypothesis? So he then offers his quote radical view. Okay, he explains why this isn't a problem under his theory, the dualistic hypothesis, because we can imagine skills and abilities in software simply evolving independently prior to physical changes that later then get co-opted. So Popper then explains his radical proposal. Quote, once a new aim or tendency or disposition, or a new skill, or a new way of behaving, so we're talking about the software, has evolved in the quote central propensity structure. That's obviously the software of the nervous system. He never calls it that because that wasn't a term back in 1961. This fact will include the effects of natural selection in such a way that previously unfavorable, though potentially favorable mutations become actually favorable if they support the new established [tenancy/tendency?]. But this means that the evolution of executive organs will become directed by this tendency or aim of the software and thus goal-directed.

[01:27:06]  Unidentified speaker: Page 278.

[01:27:08]  Unidentified speaker: Okay, that's Karl Popper.

[01:27:10]  Unidentified speaker: This is exactly what Michael Levin is talking about.

[01:27:12]  Bruce: Karl Popper back in 1961 actually anticipated the whole thing, okay? Now, once a mutation like this is established, another mutation, this is Karl Popper, another mutation which makes the skill structure more flexible may become more favorable. And by such mutations of the skill structure, the organism might acquire propensity to learn in the sense of improving its skills by trial and error. So, Popper is explaining that evolution really only makes sense if evolution had a tendency towards creating learning algorithms. Therefore, we should expect evolution, physical evolution to be rooted in learning algorithms from the ground up, from the beginning, okay? This is a really, Popper saw this as a radical proposal, and he says, I admit this is a radical proposal, I know I'm kind of out there, he says, in objective knowledge. Michael Levin is showing that he was right, that took decades later, how many decades is it now? Quite a few decades between 1961 and today. But this is actually what we're really starting to find, okay? It's an idea that I had only ever heard from Karl Popper, but it makes an insane amount of sense to me. If evolution had to adjust the software to account for each new hardware change as we would do with a Boston Dynamics robot, I actually think Boston Dynamics robots today do have learning algorithms. They didn't back when I made episode 38, they often didn't actually use learning algorithms.

[01:28:41]  Unidentified speaker: It would just never work.

[01:28:43]  Bruce: There was this intense, so there's this intense evolutionary pressure to evolve learning algorithms. This is exactly why it makes so much sense that you can explain life, cells, and tissue in terms of their learning algorithms, aka cell cognition like Michael Levin is doing. Okay, but this is just an analogy, right? I mean, clearly, we don't mean to imply that this is really a form of cognition, right? Surely, we at least agree with Hancock that referring to cells as having cognition is just going too far, right? Well, let's consider what Michael Levin has to say about this. Consider this quote from one of Levin's recent papers. In this paper, he takes a look at the fact that the processes we currently only associate with nervous systems actually evolved prior to the existence of the nervous system and are contained in the bioelectric systems that he has discovered. Quote from the abstract, However, the findings of evolutionary biology, this is from the abstract, developmental bioelectricity and synthetic bioengineering are revealing the ancient pre-neural roots of many mechanisms and algorithms occurring in brains, the implication of which is that minds may have preceded brains. Okay, let me see if I can explain why this is significant, okay?

[01:29:59]  Unidentified speaker: And it actually, it makes so much sense.

[01:30:01]  Bruce: It may be a little shocking, but it's more, this is by the way from brains and where else, mapping theories of consciousness to unconventional embodiments.

[01:30:10]  Unidentified speaker: Brains, where did nervous systems come from, right?

[01:30:15]  Bruce: They came from this bioelectric signals that the cells use that Levin has discovered, that affect the morphology, right?

[01:30:23]  Unidentified speaker: So, what are brains?

[01:30:25]  Unidentified speaker: What are nervous systems?

[01:30:26]  Unidentified speaker: They're bioelectric signals.

[01:30:29]  Bruce: There's nothing out of the ordinary with this idea that nervous systems use bioelectricity to coordinate the animal, the organism, you know, and that's what brains and nervous systems do.

[01:30:41]  Unidentified speaker: But that had to come from somewhere.

[01:30:43]  Bruce: There had to be something that came before it that was similar enough to it that it could evolve into nervous systems. Well, that is the bioelectricity of the cells. The bioelectricity between the cells, the collective intelligence of the cells, it's much slower than brains. Okay, he says it's more on the matter of minutes instead of, you know, milliseconds. But he is saying in this paper that everything that you would think of as being part of the nervous system or the brain, all of those same mechanisms exist in animals that don't have brains, because all cells have these and multicellular animals have these. So there is even possible implications for consciousness studies. The paper points out that many theories of consciousness are built on various assumptions about brain-specific activities. We now need to either eliminate those theories because it would make brain-free animals also conscious, or we need to accept that there isn't a need for nervous systems to be conscious. For what it's worth, I think the first is the correct answer. I don't believe animals without nervous systems are conscious.

[01:31:50]  Unidentified speaker: But maybe I'm wrong.

[01:31:51]  Unidentified speaker: Who knows?

[01:31:52]  Bruce: The key point here is everything that brains do today already existed in a simpler form prior to brains, very far back in evolutionary time. This is what Levin is discovering with his bioelectricity and cell collective intelligence. Another interesting aspect of Levin's philosophy is its ties to the Popperian concept of corroboration.

[01:32:14]  Unidentified speaker: Let me actually get some quotes here.

[01:32:16]  Bruce: This is from Bioelectric Fields, A Paradigm Shift in Biology. The easiest way to explain this is in terms of not having precommitments to armchair philosophical positions, but instead to formulate your theories as adjudicatable via empirical evidence. For example, now here's the quote from Levin at 52 minutes and 38 seconds. This is a question to Levin actually.

[01:32:40]  Unidentified speaker: Can we train the weather as an agent?

[01:32:43]  Bruce: Well, you know, your view then is, I guess you'd say even the weather has a mind. Levin answers this crazy sounding question, answer this crazy sounding question, makes explicit why making your theories testable always trumps philosophical commitments. He says, well, I don't say that it does.

[01:33:02]  Unidentified speaker: I say that it might.

[01:33:04]  Bruce: Now, the question is, could we do an experiment? So, for example, you have, I don't know, a tornado or some kind of weather system. We could, in theory, do an experiment if we had the tools to do it. We could, in theory, test whether they habituate. Habituate is a kind of learning algorithm that cells are known to do. We could, in theory, test the proposition, can you train a dynamical system? And you know, that might sound crazy. How would you train a hurricane or something? But all of us are just a kind of dynamical system right now. Now, we are certainly, we have some qualities that hurricanes and things like that don't have.

[01:33:45]  Unidentified speaker: But again, it's a spectrum.

[01:33:47]  Unidentified speaker: It isn't a categorical difference.

[01:33:49]  Bruce: And we are all temporary dynamical systems in metabolic space and things like that.

[01:33:54]  Unidentified speaker: And we can learn.

[01:33:56]  Bruce: It's entirely possible that if we had a way to simulate—stimulate weather patterns in a particular way, you might find that, wow, there really is some kind of habituation here. Some kind of sensitization here.

[01:34:11]  Unidentified speaker: Now, a couple of things to say.

[01:34:13]  Bruce: One is that we've done this with molecular pathways. This is why I was saying earlier that I don't think cells are the bottom level, because all it takes is half a dozen genes interacting with each other to show associative learning, meaning Pavlovian conditioning. So this is why he thinks that cells aren't the bottom level, that learning algorithms exist lower than a cell.

[01:34:38]  Unidentified speaker: Like that's it.

[01:34:39]  Unidentified speaker: You don't need the rest of the cell.

[01:34:41]  Unidentified speaker: You don't need anything else.

[01:34:44]  Bruce: Just from the math alone, for just a few interactive subunits, you can already give this to you.

[01:34:50]  Unidentified speaker: So could weather patterns do it?

[01:34:52]  Unidentified speaker: Could ecosystems do it?

[01:34:54]  Bruce: I mean, if I had a $10 bet right now, I would probably say they can, but we cannot assume that.

[01:35:00]  Unidentified speaker: We need to do experiments.

[01:35:04]  Bruce: So my point, and this is now Michael Levin, why intelligence isn't limited to brains at 35, 31.

[01:35:11]  Unidentified speaker: So my point is simply this.

[01:35:13]  Bruce: We have to be humble about the fact that we don't know how to just, you know, notice these things. And they are not, you can't have philosophical feelings about it. You have to do experiments and see what that does for you.

[01:35:26]  Unidentified speaker: Let's make it super practical.

[01:35:28]  Bruce: This is now from Biological Fields, The Paradigm Shift in Biology 2008-07.

[01:35:32]  Unidentified speaker: Let's make it super practical, okay?

[01:35:34]  Bruce: Because some of these philosophical things can be debated endlessly and that causes people to dismiss them. I want to make it, I want to make it extremely practical.

[01:35:44]  Unidentified speaker: Let's do this.

[01:35:45]  Bruce: Let's agree that for all of us, the acid test of any new theory is whether they facilitate new applications in medicine, new research and new discoveries. So I'm not interested in philosophical theories that paint extra stuff onto the science.

[01:36:00]  Unidentified speaker: Because it makes us feel good.

[01:36:02]  Bruce: My claim is, I'm only going to say things, and I'm only going to pursue things that I think are actively making us do better science, and which is empirically detectable. New capabilities and new medical applications. So that's our goal. So I want to be clear, that's my starting point. So boom, Michael Levin is talking about the Popperian concept of theories not being ad hoc, which to Popper meant they had to have independently testable consequences. This is Popper's theory of corroboration.

[01:36:35]  Unidentified speaker: It's actually not Popper's theory of corroboration.

[01:36:37]  Bruce: It's Popper's theory of falsification-ism, which is largely misunderstood. Falsification in Popper has often been understood by fans of Popper as being about justification-ism. And it is about justification-ism, but that's not the main thing it's about.

[01:36:52]  Unidentified speaker: Okay?

[01:36:54]  Bruce: What falsification really is about is about what makes a superior theory. What's the nature of a superior theory? I've often used the example of the theory of Oranges Stop Scurvy versus Vitamin C Stop Scurvy. Now, these two theories are in fact correct, so neither can truly ever be refuted. But Vitamin C Stop Scurvy is the better of the two theories, particularly after corroborating Oranges Stop Scurvy and then wanting to find out why that is. Precisely because it's the riskier theory, by which Popper means it's the theory that has more potential logical counter-examples. This idea that we prefer theories that are easier to be falsified, that is what falsificationism is in Popper's epistemology. It's all about our choice to formulate a riskier, bolder, or more empirically testable theory and thereby increase the content of your theory by increasing its reach.

[01:37:50]  Unidentified speaker: And this is what Levin is getting at.

[01:37:54]  Bruce: The goal is theories that say something about the world that might turn out to be incorrect.

[01:37:58]  Unidentified speaker: That is to say that they can be falsified.

[01:38:02]  Bruce: Let me actually put this in a slightly different way that I've been thinking about this. In Popper, we prefer a certain kind of theory. We don't prefer robust theories that are impossible to criticize or falsify. We prefer brittle theories that really shouldn't be standing up except they happen to be right.

[01:38:21]  Unidentified speaker: That's really what Popper is about.

[01:38:23]  Bruce: Okay, you want a really bold, risky theory that really should be easy to show is wrong if it were actually wrong. And that's what falsificationism actually is. This is why the whole debate over corroboration versus falsification never did make sense. Okay, because falsificationism isn't about justificationism. It's about the nature of the theory itself. Okay, the brittle, bold, risky theory that we prefer. This is now from 14 Season Opener, Michael Levin on The New Biology, 2211. If the answer from the actual experiment is actually it doesn't help to think that cells or tissue are being intelligent agents at all, then we give up and we say chemistry is the way to go and we continue status quo. But what we found actually is that that's not the case. We've made numerous discoveries that by using these tools that operate on principles of looking for memories, looking for goals, looking for intelligence, and so on, these kinds of tools are extremely, they pay off, they pay off wildly in developmental biology. And for that reason, and regenerative medicine. And so for that reason, here's my hypothesis.

[01:39:34]  Unidentified speaker: It was very simple.

[01:39:36]  Bruce: Morphogenesis, meaning the creation of the body, its repair during regeneration, resistance to cancer, aging, and so on. All of these things are the behavior of a collective intelligence in anatomical space. The behavior and other hypotheses—other hypotheses was that the intelligence in anatomical space could include the intelligence in the brain, could include intelligence in the organs. Yeah, I mean, so it's not controversial that we are collective intelligence of neurons operating in three-dimensional space, right?

[01:40:08]  Unidentified speaker: I mean, that's it.

[01:40:09]  Bruce: And so he's saying, look, it's just, I'm not really doing anything that's all that radical compared to what we already accepted. And we even know the glue, we even know the cognitive glue that binds the individual neurons into a single [unmodified?] mind, that would be bioelectric electricity like in the brain. My point is that the exact same thing happens in anatomical space. And it's, and again, the cognitive glue there is the biological network of the rest of the body. Now, at 2056, now, having started there, here's where we have to be fearless, says Levin. We cannot assume from philosophical arm, or from a philosophical armchair, what the optimal way to deal with biomedical issues is from the level of chemistry. That's a common assumption, but we have to be clear that's an assumption.

[01:40:56]  Unidentified speaker: We don't actually know that.

[01:40:58]  Bruce: If you're going to going to be actually scientific about this, you have to entertain the possibility that tools from other levels of organization, and the tools I like best are the tools that come from cognitive science, behavioral science, and neuroscience. Some of these tools are going to be useful for what we need to do.

[01:41:16]  Unidentified speaker: So my claim is very simple.

[01:41:17]  Bruce: I'm going to take, I'm going to appropriate from the neural and behavioral sciences tools having to do with memory, learning, and perceptual illusions, top-down control of executive goals over the chemistry, all the molecular tools, so all the ion channel channel kind of things, optogenetics, all the neurotransmitter drugs, you know, the active inference framework. I'm going to steal all of that stuff from behavioral science and I'm going to simply ask, what does it do for us to look at the control of the body, control of body shape and function? If the answer is that it doesn't help at all, then we give up and we say chemistry is the way to go and we continue with the status quo. So let's look back now at Zach Hancock's claim from his video at one hour, 17 minutes, 19 seconds. Words like intentionality, purpose, and agency, to me really have no place in evolutionary biology, especially when evolution is being explained to the general public, because they have a tendency to confuse and give false impressions. At 118.03, my point is that it simply isn't necessary to use such loaded words, as they indicate completely trivial concepts in the first place, and only act to spread confusion about how evolution actually works. At one hour, 55 minutes, 11 seconds, Shapiro thinks terms like agency and purpose are themselves driving forces that are not merely emergent properties of chemicals, but are something special. If you didn't know about Levin's work, Hancock's claims seem rather reasonable to me.

[01:42:53]  Bruce: But Levin's work is a counterexample to Hancock's claims that show that Hancock is mistaken. Using terms like purpose, goal, and agency is how Michael Levin came up with a new way of looking at evolution and biology and found the bioelectric field of cells of an organism and then used that to communicate and set goals and solve problems. It is how Levin showed that trying to force fit everything into evolution into the emergent properties of chemicals, to quote Hancock, was causing biologists to miss an entire level of emergence. So I want to go back to a statement Hancock made about Shapiro's theory of natural genetic engineering. At 1 hour, 18 minutes, 20 seconds, he says, The biochemist James Shapiro argues that organisms direct their own evolution, that is, they do so by directly modifying their own DNA. Shapiro calls this natural genetic engineering. If it is true, it would certainly be a non-trivial action to purpose. Indeed, if organisms were capable of modifying their own DNA in service of adaption, it would be a remarkable departure from the modern synthetic paradigm. Now, none of the experiments from Levin involved natural genetic evolution. None of the ones I just went over, none of them involve natural genetic evolution. So we might argue here that Hancock has been corroborated by Levin's experiments here. But at this point, Hancock's whole point seems entirely deflated.

[01:44:23]  Bruce: Okay, if we did find that the organism contained software to rearrange genes to adapt, not unlike the planaria that grew its head, a new head in barium, would it really be all that different from what we learned from Levin's experiments? It's possible that Shapiro is entirely wrong about natural genetic engineering. Indeed, there may be no need for it if the tissue itself can intelligently learn without genes. So, in fact, I have heavy suspicions that natural genetic engineering, NGE, is a dead end. I think Shapiro is probably wrong about it. Okay, with maybe a few specific counter examples that he raises, I suspect that in general it plays very little role in how evolution works. I could be wrong about that, by the way, but I'm just saying nothing in Levin's experiments really involved it. But it's hard to miss that Levin has spiritually made the very same point that Shapiro did. Shapiro noticed an overlooked problem and proposed maybe a wrong solution to it. But he was right about the problem and that people were overlooking it. I think Hancock proved correct that neo-Darwinian evolution has evolved into the Price equations which are gene neutral. Because of this, Hancock is technically correct that this is a more universal understanding of neo-Darwinism and really isn't threatened by any of the third way or anything Michael Levin is doing. Yet Hancock got the single most important thing wrong.

[01:45:54]  Bruce: There is room for a whole new paradigm in evolutionary biology for both thinking about cells as agents and to stop thinking every aspect of biology in terms of genes controlling chemicals. Ironically, it was this intense desire to quote not mislead the public that in fact misled the entire field of biology to miss this new paradigm. So even if Shapiro is wrong about natural genetic engineering, which he may well be, and even if Noble is wrong about Lamarckism being back, which I think surely he is, they were right about this point that the field is stuck in a reductionist mindset that has undermined progress in really important things like even cancer research. Once you realize this, it really seems to me like Game Set Match. It almost doesn't matter how many more points Hancock scores here or how he was for the most part correct and Noble was for the most part off base. This war was really about if biological evolutionists had fallen into rigid thinking that was causing them to miss important aspects of biology or evolution. Zach Hancock or Jerry Coyne was arguing, no, that isn't the case. And all it takes to refute their view is a single counter example. In this case, we've got one, Michael Levin's research program. It's a case of literally Hancock winning every battle, every single battle, and then losing the war. That is exactly what has happened, in my opinion. That's why I'm saying I'm going to take a definitive stance on this. You might still be wondering what any of this has to do with the problem of open-endedness or even AGI.

[01:47:39]  Unidentified speaker: Well, maybe nothing.

[01:47:41]  Bruce: Okay, now my interest is AGI. I'm exploring all sorts of different paths, most of which are probably going to be dead ends. Until we have a falsifiable theory of open-endedness or AGI, it is uncertain what connects to what. But think about Levin's theory from a 30,000 foot level. He's claiming that the reason evolution works as well as it does is because there are levels of intelligence, learning algorithms, all the way down. When evolution makes a mutation, the next level up can decide how to deal with that problem. This is one of the reasons that plasticity works.

[01:48:14]  Unidentified speaker: Plasticity is well accepted, okay?

[01:48:15]  Bruce: But this is why it works. There is, as Donald Campbell imagined, a hierarchy or maybe more like a web of blind variation and selective retention algorithms that interact. This is Campbell created the idea of downward causation based on this, okay? He's the one who coined that term. So biological evolution is thus evolution of evolution of evolution, so to speak. So have you ever actually seen an AI or ML algorithm try to model genetic evolution as anything but the simplistic and reductionist genes eye view of evolution?

[01:48:52]  Unidentified speaker: That's the way it's always done.

[01:48:54]  Bruce: So at a minimum, this shows that we've missed something important about evolution, and thus can't really replicate—and that may be one of the reasons why we can't really replicate it with an algorithm due to this. Of course, this may not turn out to be true. This does not turn out to be the thing we missed about evolution. It might be a thing we missed about evolution.

[01:49:15]  Unidentified speaker: There could be other things.

[01:49:17]  Bruce: But this is definitely a case where it's definitely worth a look, right? At a minimum, this is an interesting new direction in biological evolution, where we start to think of cells and tissue as being learning algorithms, and start to really think about how learning algorithms, just like Karl Popper said, are key to understanding how evolution works. That's not something I think I've seen anywhere else, other than basically Karl Popper, and then maybe arguably Denis Noble, and then Michael Levin, right?

[01:49:53]  Bruce: So I think this is really why I feel like I can actually take a pretty strong stance here. Levin's work really does show that there's something wrong, and what a new interesting direction is. Whether that then turns into solving the problem of open-endedness or AGI, not as sure.

[01:50:10]  Unidentified speaker: But you could see how maybe it could, right?

[01:50:12]  Bruce: You could see how, because there's this giant hierarchy of learning algorithms, maybe open-endedness comes out of not a single learning algorithm, but like a group of learning algorithms. That's not an impossibility, right? Maybe that is something that needs to be in our heads as we start to think about how to solve the problem of open-endedness, something that we could learn from how evolution actually works.

[01:50:38]  Unidentified speaker: All right.

[01:50:39]  Unidentified speaker: That's my thoughts.

[01:50:42]  Peter: Well, Bruce, you're on your own unique path in life, and I love listening to you, and it's a trip.

[01:50:51]  Unidentified speaker: I'll say that.

[01:50:52]  Peter: But thank you for your thoughts here, and I'll look forward to listening again while I edit.

[01:51:00]  Unidentified speaker: All right.

[01:51:01]  Unidentified speaker: Thank you very much.

[01:51:10]  Unidentified speaker: Hello again.

[01:51:11]  Peter: If you've made it this far, please consider giving us a nice rating on whatever platform you use, or even making a financial contribution through the link provided in the show notes. As you probably know, we are a podcast loosely tied together by the Popper Deutsch Theory of Knowledge. We believe David Deutsch's Four Strands tie everything together. So we discuss science, knowledge, computation, politics, art, and especially the search for artificial general intelligence. Also, please consider connecting with Bruce on X at BNielsen01. Also, please consider joining the Facebook group The Many Worlds of David Deutsch, where Bruce and I first started connecting.

[01:51:54]  Unidentified speaker: Thank you.